Guidebook on Conducting Airport User Surveys and Other Customer Research (2021)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

P A R T   I I Specific Survey Guidelines Introduction User surveys are still the most common way for airports to obtain information about their customers and other users. Although they are undertaken for a wide variety of reasons, they can be grouped into two broad categories: those intended to measure customer or user satisfac- tion or related customer research aspects, and those intended to obtain information to support airport planning and development activities. Each of these categories is discussed in the follow- ing sections. In planning an airport user survey, or even deciding whether one is necessary, managers and planners need to consider what information is already available, what additional informa- tion is needed, and how accurate that information needs to be, and then balance the cost of the survey against the value of the resulting information. This is not an easy task. One goal of this guidebook is to help airport managers and planners address this issue. An important consideration is how frequently to conduct such surveys. If the information being collected does not change appreciably over time, then it may be appropriate to rely on data from past surveys. On the other hand, if the desired information is believed to be changing over time, or it is simply not known whether this is the case, more frequent surveys are needed. Many airports undertake relatively small-sample surveys on a regular basis, such as quarterly, as a way to monitor any changes over time, identify emerging problems or issues in a timely way, and help determine whether a more detailed survey is needed. Air passenger customer satisfaction surveys are commonly undertaken on a regular basis. One reason for this is to identify particular issues that may have arisen that adversely affect customer satisfaction so that they can be addressed in a timely way. Another reason is to monitor general levels of customer satisfaction over time. Air passenger surveys to gather information to support airport planning and development activities are typically performed less frequently. Survey Role in Airport Customer Research A broad interpretation of airport customer research is that it is any effort to better understand the needs, opinions, and behavior of airport users, employees, or other stakeholders. More specifically, this kind of research addresses questions related to: • Customer satisfaction with airport facilities and services, • Air passenger and airport employee satisfaction with existing concession services and likely use of additional services,

122 Guidebook for Conducting Airport User Surveys and Other Customer Research • How air passengers spend their time in the airport and which facilities and services they use, • Spending patterns by air passengers and airport employees in airport concessions and how these vary by air passenger or employee characteristics, and • Attitudes toward the airport by users and external stakeholders. Since many of these questions involve attitudes or user characteristics that cannot be observed directly, the only way to answer them is to ask the users themselves. Thus, surveys form an essential tool in airport customer research. In addition to directly answering questions of interest, airport user surveys can also provide useful information to better interpret findings from other customer research methods, such as analysis of social media posts or use of smart- phone tracking technology. Survey Role in Airport Planning and Development Airport managers and planners require a wide range of information to support their plan- ning activities and decision making for facility development and day-to-day operations. The quality of plans and decisions depends on the quality of the information on which they are based. In particular, in any customer-serving organization, two considerations are critical to effective planning and management: • Understanding the market • Understanding the customer Airports function within the air transportation system and indeed within the larger trans- portation system. Therefore, airport managers and planners must understand the nature of the market being served. This includes such issues as the trip purposes and travel patterns of air passengers using the airport as well as the nature of goods being shipped through the airport. Where the local travel market is served by more than one airport, it is helpful to understand relative market share as well as the factors that influence it. Depending on the length of the trip, travelers and shippers may also have a choice between air and surface transportation (such as car, bus, truck, rail, or ship). Effective planning and development of airport facilities and services also requires an under- standing of the characteristics, needs, and behavior of the customers using the airport, principally the air passengers and shippers but also the intermediate service providers, such as the airlines, freight forwarders, and ground transportation providers. This customer information includes such considerations as air travel party size, the amount of baggage checked, where passengers began or ended their trips to and from the airport, and the ground transportation modes that they used. While some of the information needed can be obtained by direct observation or from statistics that are routinely collected, much cannot. Some airport user attributes are not directly observ- able, and some statistical data are not routinely collected. Airport user surveys are the only way to obtain information of this type, which includes the following: • Air passenger characteristics for facility planning, such as arrival time before flight departure or the number of well-wishers accompanying passengers into the terminal • Ground transportation used by air passengers and airport employees to travel to and from the airport, particularly via public transport, which cannot be inferred from the service frequency • Trip origins and destinations of airport users within the region served by the airport • Use of competing airports and the factors affecting the choice of airport • Economic impact of the airport on the surrounding region

Specific Survey Guidelines 123   Having good information on the wide range of issues typically addressed by airport user surveys improves the quality of planning and decision making. At the same time, obtaining good information costs money. In deciding how much effort to devote to collecting this infor- mation, airports should consider how the information will be used and the likely costs of poor planning, design, or operating decisions. In general, where information will be used in planning new facilities, the cost of a survey will be small compared to the cost of the facilities. The ability to appropriately size the facilities and phase future expansion could significantly reduce the cost of a project or forestall the need for expensive modifications after construction has commenced or the facilities have been opened. Survey Concepts For guidebook users who are considering performing a survey for the first time or have limited experience with surveys, this section provides an overview of the basic concepts and some of the common terminology used in survey practice. The purpose of an airport user survey is to gather information about the characteristics or opinions of a defined group of airport users, such as air passengers or those employed at the airport. This group is referred to as the “target population” (often shortened to “population”). In general, it is not practical to gather the desired information from every member of the target population, so a subgroup of the target population is selected. This subgroup is referred to as the “sample.” The process of performing a survey involves selecting an appropriate sample, requesting the desired information from each member of the sample, analyzing the responses, and reporting the resulting findings. The process of performing a survey starts by planning each of these steps. When planning a survey, consideration needs to be given to clearly defining the target popu- lation and to deciding how large a sample is required and how to select the sample of respon- dents. These are not always obvious decisions. For example, in performing an air passenger survey, is the target population all passengers using the airport in a given period, including those connecting between flights, or only those starting their air trips at the airport? Is the target population only departing passengers, or does it include arriving passengers as well? The answers to these questions will affect the survey methodology as well as the approach to selecting the sample. Guidance on addressing these issues is provided in Chapter 5. Ideally, a survey sample would be randomly selected individuals from the target population, where random selection means that any individual in the target population has an equal likeli- hood of being selected. Random selection of survey respondents allows inferences to be made from the sample about the characteristics of the target population with known confidence in the accuracy of those inferences, for reasons explained in more detail in Chapter 4. However, in practice, such random selection is often difficult to achieve. The methods used to select survey respondents inevitably constrain the selection so that it is rarely completely random. For example, if an air passenger survey of selected flights is conducted in airline gate lounges, the sample is necessarily restricted to those passengers in the lounges in question, who are generally passengers on the flight about to depart from that gate. The respondents on any given flight will generally have a different distribution of characteristics from the target popula- tion as a result of the specific market served by the flight, the time at which the flight departs, and possibly other factors, such as the airline in question. Even if the flights to be surveyed have been randomly selected, it is unlikely that the selected flights will cover all possible combinations of market, airline, time of day, and day of week

124 Guidebook for Conducting Airport User Surveys and Other Customer Research because of budgetary limitations on the number of flights that can be included in the survey. Therefore, the smaller the number of flights sampled, the more such combinations will be excluded, and the less likely it is that the characteristics of the passengers on the sampled flights will correspond to those of the target population as a whole. A sample in which the distribution of characteristics in the sample differ from the popula- tion is referred to as a “biased sample.” An important aspect of planning a survey is to design a sampling strategy that reduces potential bias. After the survey has been performed, techniques can be applied to control for any bias that remains. These techniques involve assigning weights to each response so that the weighted results more closely correspond to the expected distribution of the characteristics of the target population. Of course, this implies some knowledge of that expected distribution and a method to calculate the response weights. These issues are discussed in more detail later in this guidebook, in Chapter 4, and in the chapters specific to each type of survey. After the sample has been defined, the survey respondents in the sample are asked to provide the desired information through the use of a survey instrument or questionnaire. [“Survey instrument” is broader than “questionnaire” because it includes, for example, data recording sheets in observational studies. The airport user surveys discussed in this guidebook exclude observational studies (which are considered qualitative methods and are discussed separately from airport user surveys), and the term “questionnaire” is used throughout this guidebook when referring to airport user surveys.] The design of the questionnaire is critical because the wording and sequence of questions will affect the responses that are obtained. Consideration needs to be given to how the survey will be performed, including how respondents will be selected and whether they will be interviewed by survey staff or will complete the questionnaire themselves. Once the initial design of the questionnaire has been completed, it should be pretested on potential respondents to make sure that the wording of the questions is clear, the answer options provided are appropriate, the question sequence is logical and easy to follow, and the survey generates the desired information. The results of the pretest may call for some redesign of the questionnaire and possibly another pretest if the changes are substantive. When the question- naire has been finalized, a pilot test is usually performed to test the survey procedures in the field. Typically, the pilot test will involve a representative group of the survey interviewers, or field staff, and should be performed well enough in advance of the full survey that any logistical issues can be resolved before the survey gets underway. If the survey questionnaire has been used before, or changes from a previous questionnaire are not substantial, a separate pretest may not be necessary, and any testing of the questionnaire can be done as part of the pilot test. After the survey data collection has been completed, the data will generally need to be checked and cleaned to correct identifiable errors (e.g., illogical, incomplete, or inconsistent answers) and determine whether to eliminate incomplete or invalid responses before the results are tabulated or analyzed. Depending on the length and nature of the questionnaire, data cleaning can involve a significant amount of work because, in order to determine what an apparently incorrect answer should have said, it is likely to require a detailed examination of the answers to other questions or research into which of conflicting answers is likely to be correct. However, this step is crucial to the quality of the results. For example, a transposition of digits in a postal zip code could place the reported location of the trip ground origin many miles from where it actually was. In summary, the basic survey concepts are: • Target population—a defined group of airport users for which information is required; • Survey sample—a subgroup of the target population selected to provide the desired information;

Specific Survey Guidelines 125   • Sampling strategy—the strategy for selecting survey respondents to reduce and control potential bias in the survey sample (i.e., the extent to which the characteristics or opinions of the sample differ from the target population); • Questionnaire—the questions used to collect information from the survey sample (or a printed document containing those questions); • Surveying method—the approach, such as an in-person interview or an online survey, used to collect information from the sampled respondents; • Pretest—testing of a questionnaire with a small number of potential respondents; • Pilot test—testing of the entire survey process; and • Data cleaning—correction of identifiable errors and elimination of incomplete or invalid responses before the survey results are tabulated and analyzed. Main Survey Types and Methods Airport user surveys come in many types, each with its own set of goals and objectives. The main types considered in this guidebook are air passenger surveys (Chapter 10), employee surveys (Chapter  11), concessionaire and other tenant surveys (Chapter  12), surveys of area residents (Chapter 13), surveys of area businesses (Chapter 14), and air cargo surveys (Chapter 15). Air passenger surveys are the most common type of airport user survey and tend to focus on passenger characteristics, demands on facilities, or satisfaction with the airport. Air passenger surveys are widely used to generate information for airport planning and management. Employee surveys are typically conducted to measure satisfaction with airport facilities and services as well as the general work environment, obtain information for transportation or conces- sion planning, and address issues such as communications and knowledge of airport procedures. Concessionaire and other tenant surveys tend to focus on concessionaire and tenants’ satisfaction with the airport as a landlord and on gathering information to help determine the economic impacts of the airport. Surveys of the general public can be undertaken for a variety of reasons. Perhaps the most common purposes are determining the public’s perceptions of an airport, investigating the factors that influence airport choice, and assessing the extent to which area residents are choosing to fly from other airports. Surveys of area businesses are frequently done as part of studies to determine the economic impact of the airport and its value to the community. Surveys may also be conducted to assist in airport planning and air service development studies by collecting information on travel by local businesspeople and their use of all the airports in the region. Air cargo surveys can address many aspects of the movement of air cargo at an airport, including cargo carried on all-cargo aircraft or in the belly holds of passenger aircraft, as well as transit cargo that arrives and departs by road. Among the issues that such surveys can address are the mix of commodities transported, the proportion of enplaned and deplaned cargo that is connecting between flights, and the number of truck trips generated by the cargo that arrive at and depart from the airport by road. Survey Methods Airport user surveys can be conducted by a variety of methods: • Intercept interviews • Self-administered surveys

126 Guidebook for Conducting Airport User Surveys and Other Customer Research • Online surveys • Telephone surveys • Mail surveys Intercept interviews by survey staff are still perhaps the most common type of airport user survey. Survey interviewers approach potential respondents at an appropriate location where the respondents can be expected to have time to answer the survey questions, such as an air- port gate lounge, and solicit the potential respondents’ participation in the survey. The survey questions are printed on a paper questionnaire or programmed on a tablet computer or similar electronic device. The interviewers ask the questions on the questionnaire and enter the responses on the questionnaire or tablet. A self-administered survey follows a similar procedure, but the paper questionnaire or tablet is handed to the survey respondent, who completes it and returns it to the survey field staff. This has the advantage that several respondents can be completing the survey at the same time. Paper questionnaires can be distributed to a large number of potential respondents, such as everyone in an airport gate lounge, and collected when they have been completed. However, this method has the disadvantage that there is no opportunity for survey field staff to identify questionable responses and ask clarifying questions, although potentially such checks can be programmed into self-administered surveys being performed by providing respondents with tablet computers or having them complete an online survey using their own devices, as discussed in more detail in Chapter 5. Online surveys are programmed into a website. Since they are performed online, the respon- dents can be anywhere that they have an Internet connection. Potential locations for online surveys of air passengers are discussed on more detail in Chapter 10. Respondents are provided the link to access the survey and complete it online. This obviously requires respondents to have a device that can access the survey, although the widespread use of smartphones make this less of an issue than in the past. The principal challenge with online surveys is how to recruit representative survey respondents and provide them with the link to the survey. Potential survey respondents may be less willing to spend the time completing an online survey than to respond to a survey being administered by someone standing in front of them. Telephone and mail surveys are more appropriate when potential survey respondents are geographically dispersed. Telephone surveys are typically conducted using computer-assisted telephone interviewing, where the questions are programmed in a computer and an interviewer asks the questions and enters the responses into the computer. Potential respondents can be identified by random digit dialing or from some appropriate list of telephone numbers. Both approaches have their advantages and disadvantages and are discussed in more detail in Chapter 13: Surveys of Area Residents and Chapter 14: Surveys of Area Businesses. Mail surveys obviously require mailing addresses of potential survey respondents and, therefore, are more appropriate when conducting surveys of a defined group of desired survey respondents for which the airport has mailing addresses. Potential survey respondents can either be mailed a paper questionnaire, which they complete and mail back, or provided a link to an online survey. These different survey methods are discussed in more detail in the following chapters that cover each of the different types of airport user survey. Adapting to a Pandemic Environment The COVID-19 pandemic that swept the world in 2020 has required airports to adapt their survey methods and techniques to reflect the need to reduce the risk of virus transmission and reassure potential survey respondents that it is safe to participate. These measures include

Specific Survey Guidelines 127   greater use of online and telephone surveys as well as steps to reduce concerns about touching questionnaires, pens, or tablets in interview surveys. The recommended 6-foot social distance still allows for interview surveys so long as interviewers take care to maintain this. Printed questionnaires and pens or pencils can be wrapped in plastic and opened by the respondent. Completed questionnaires and pens or pencils that the respondent does not wish to keep can be returned by the respondent to a pouch or container and not handled by the interviewers. Where tablets are handed to respondents for a self-administered survey, they can be wiped with a disinfectant before and after use. Conspicuous use of protocols to eliminate the risk of virus transmission can help reassure potential respondents and enhance willingness to participate in the survey. How long such procedures will need to be maintained is unknown at the time of writing. However, even after the airport environment returns to pre-pandemic conditions, if it ever does, there is always the possibility of a future pandemic, so airports may need to apply similar measures in response.

128 10.1 Introduction A survey of air passengers involves a number of difficult challenges because of the wide range of information that may be required and the limited opportunity to perform the survey. Some air passengers are anxious to reach the boarding gate in good time to board their flight, while others spend time in airport concessions and arrive at their boarding gate shortly before board- ing commences. Thus, air passenger surveys require careful attention to survey methodology and development of an effective sampling plan and questionnaire. Related aspects that are also addressed in this chapter include gathering information on greeters and well-wishers as well as information on ground vehicle trips; this information can be used in planning airport ground- side facilities. Information on greeters and well-wishers is included in this chapter because of their relationship to air passenger activity and the fact that this information is often obtained from air passenger surveys. 10.2 Purpose of the Survey Much of the information about air passengers that is needed for planning or operational decisions cannot be directly observed or is not readily available from statistics that are routinely collected; it can only be obtained by asking the passengers themselves. Surveys of air passengers are the most common type of airport user survey and are performed for a variety of reasons, including for data collection related to the following: • Air passenger satisfaction with airport facilities or services • Air party characteristics for airport terminal planning • Air passenger use of ground transportation for airport groundside planning and regional transportation planning • Air travelers’ choice of airport in a multi-airport region In the following discussion, surveys addressing these different purposes are broadly divided in two categories: • Customer satisfaction surveys • Surveys to determine air passenger characteristics for airport planning and development The former is typically conducted on a regular basis, often quarterly, and primarily includes questions asking the respondents to indicate their satisfaction with their experiences in using the airport facilities and services, although these surveys usually collect other information about the respondents as well. The latter are generally conducted less frequently and contain more detailed questions about the respondents’ characteristics and use of ground transportation for their access and egress trips to and from the airport. C H A P T E R   1 0 Air Passenger Surveys

Air Passenger Surveys 129   An air passenger survey may be initiated to gather information on a specific issue, such as the use of different ground transportation modes in order to perform an air quality emissions analysis for environmental impact documentation of an airport project. However, given the cost and effort involved in performing a survey, consideration should be given to whether there are other information needs that can be met at the same time by expanding the scope of the planned survey. Such scope expansion will require a careful trade-off between the need for the additional information and the potential impact on the cost and complexity of the survey. The information obtained from surveys of air passenger characteristics and use of ground transportation for their trips to and from the airport is so important to airport planning and management that many airports perform such surveys on a regular basis, such as every year or two. In order to provide a consistent time series of information, these surveys often contain the same core set of questions, but questions may be added to a specific survey to address particular issues of interest. However, care is needed to preserve the continuity of the information from the core questions. In particular, rewording or omitting particular questions can compromise the ability to track changes over time. One important role of air passenger surveys is to provide information on air traveler char- acteristics and decisions in order to develop models for air travel demand forecasting, airport ground access mode choice, and airport choice. These models play an important technical role in airport planning, regional transportation planning, and airport system planning studies and generally require detailed data on a large sample of individual air travel parties. Although air passenger surveys are generally undertaken by airport operators, in some cases they may be undertaken by regional transportation planning or state aviation agencies to gather information on air travel characteristics and airport user ground transportation mode use patterns for regional or state transportation planning purposes or preparation of regional or state airport system plans. Such surveys are typically undertaken in regions with multiple air-carrier airports, where surveys performed by the different airport operators may not provide consistent information on air traveler use of the airports in the region and associated ground transportation activity, whether due to being performed at different points in time or not includ- ing the same questions. The purpose for which the survey is being undertaken influences a wide range of survey planning decisions, including how and where the survey will be performed, the questions that will be asked, the sample size required, and the sample strategy adopted. Therefore, the purpose needs careful thought at the start of the survey planning process. Tables 10-1 to 10-3 illustrate the range of information obtained in a representative sample of recent air passenger surveys and are discussed further in the following subsections. It can be seen that the information obtained by the different surveys varies considerably, both in terms of the issues addressed by each survey and the level of detail implied by the number of attributes covered by the questions. 10.2.1 Customer Satisfaction Surveys The most widely used airport customer satisfaction survey is the Airport Service Quality (ASQ) Survey program undertaken by Air- ports Council International, in which many of its member airports participate (http://www.aci.aero/Airport-Service-Quality/ASQ-Home). Best Practices “The ASQ surveys are done on tablets, and [we] conduct 400 to 500 [surveys] per month. By looking at passenger rankings, we have determined the six factors that are most important to our guests, including seating in the holding rooms, restroom cleanliness, Wi-Fi access, wayfinding, checkpoint wait times, and staff courtesy and helpfulness.” —Research participant

130 Guidebook for Conducting Airport User Surveys and Other Customer Research This program includes several different surveys covering air passengers and airport employees, but the most widely used is the survey of departing air passengers. These surveys are generally undertaken four times a year using a standardized survey questionnaire and a minimum of 350 respondents for each survey. As of March 2020, 36 U.S. airports participated in this program. The ASQ departure survey asks respondents to rate 30 aspects of their experience at the airport where they were surveyed as well as their overall satisfaction with the airport. These aspects include access, check-in, passport/identification inspection, security, wayfinding, airport facilities, and the airport environment. In addition, the survey includes several questions about the current air trip being taken as well as traveler characteristics. In addition, many airports conduct their own customer satisfaction surveys. Three recent such surveys are: • John Wayne Orange County Airport (SNA) 2019 Passenger Survey (Phoenix Marketing International 2019), • San Francisco International Airport (SFO) 2018 Customer Survey (San Francisco International Airport 2020), and • Tulsa International Airport (TUL) 2016 Customer Survey (survey details provided by airport staff). Table 10-1 shows the information collected by each survey, the number of facilities and services evaluated, and the number of evaluation criteria that respondents were asked to assess. An “x” in Table 10-1 indicates that the attribute or question was included in the survey. It can be seen that the number of respondent trip details, travel party details, and personal characteristics reported in the surveys varies significantly. Trip details refer to such aspects as the destination, trip purpose, and airline flown on. Travel party details refer to such characteristics as the ground access mode used to travel to the airport, the travel party size, and the number of checked bags. Personal characteristics refer to such aspects as age, household income, and gender. It should be noted that, in the case of multi-person travel parties, the personal charac- teristics will typically be reported for the member of the travel party responding to the survey and may be different from those of other members of the travel party. This needs to be borne in mind when analyzing such characteristics as the number of flights taken from the airport by the respondent in the previous year. Not all surveys asked how many people were in the travel party. This needs to be taken into account in expanding survey results to the total population of air passengers using the airport, since the characteristics and use of airport facilities and services may be quite different for single-person travel parties and multi-person travel parties, particularly as they relate to use of airport ground access modes or such facilities as airline club lounges. The distinction between the number of facilities and services evaluated and the number of evaluation criteria arises because the surveys often had several questions about specific facilities or services, such as security screening or airport food and beverage services. In some cases, the distinction between facilities and services is somewhat arbitrary, but facilities were defined as physical facilities that would be used by all or most passengers (or at least passengers using the airport in a given direction, such as departing passengers), whereas services were defined as those that would be used selectively, such as retail or food and beverage services. Ground transportation services and facilities, such as parking facilities or taxi services, were classified as services, although it is recognized that some of these involve major airport facilities as well. General airport attributes and amenities include such aspects as signage, feelings of safety and security, and cleanliness of the airport terminal generally. Cleanliness of restrooms was considered a facility attribute.

Air Passenger Surveys 131   Three of the surveys asked questions about facilities or services (such as baggage claim) for arriving passengers, although the surveys were only administered to departing passengers. These surveys asked the respondents to answer based on their previous use of the airport if they had used it before (as generally visitors to the region would have on their arrival). TUL speci- fically surveyed some arriving passengers as well as greeters and well-wishers in the terminal, although the arriving passengers were not asked to evaluate any facilities (such as baggage claim) dedicated to their use. 10.2.2 Surveys of Air Passenger Characteristics In order to illustrate the range of information obtained in surveys of air passenger charac- teristics as well as the wide variation in sample size across different surveys, Table 10-2 shows the information collected by five such surveys conducted at airports of varying sizes over the 9 years prior to the time of this writing: • Boston Logan International Airport (BOS) 2019 Air Passenger Ground Access Survey (Mark Keifer Consulting and Desautels Consulting, LLC 2019) • Los Angeles International Airport (LAX) 2019 Passenger Survey (Unison Consulting 2019) • Minneapolis–St. Paul International Airport (MSP) 2011 Special Generator Study (Cambridge Systematics, Inc. 2012) Survey Attribute or Question ACIASQ SFO 2018 SNA 2019 TUL 2016 Responses (most recent year or year shown) (a) 2,809 613 252 Departing passengers Yes Yes Yes 100 Arriving passengers 102 Meeters and well-wishers 50 Respondent trip details 5 3 1 5 Respondent travel party details 3 9 7 1 Respondent personal characteristics 6 10 18 9 Satisfaction rating scale 1–5 1–5 1–5 1–10 Facilities/services evaluated Airport facilities 8 4 4 1 Concessions and services 7 8 9 2 Value for money 3 General airport attributes and amenities 8 8 8 2 Number of evaluation criteria Departing passenger facilities/services 18 17 18 5 Arriving passenger facilities/services 3 2 Greeter/well-wisher facilities/services 2 General airport attributes 11 8 14 3 Questions with free-form comments 1 5 4 Three most important aspects x Best and worst experience at airport x One thing to improve airport experience x Products or services would like to see at airport x x Statements about the airport you agree with x Likely to recommend airport to friend/colleague 0–10 0–10 Overall evaluation 1–5 1–5 1–5 1–10 Reasons for overall evaluation x Notes: (a) Typically about 1,400 (350/quarter). An “x” indicates that the attribute or question was included in the survey, and a blank cell indicates that the attribute or question was not included in the survey. Table 10-1. Airport customer satisfaction surveys.

132 Guidebook for Conducting Airport User Surveys and Other Customer Research • Oakland International Airport (OAK) 2014/15 Ground Access Study (Corey, Canapary & Galanis Research 2016) • Orlando International Airport (MCO) 6-Question Survey (Orlando International Airport 2019) The BOS and LAX surveys were the latest in a continuing series of surveys conducted by the two airport authorities. The previous ground access survey at BOS was performed in 2016, while the previous passenger survey at LAX was performed in 2015. The MSP Special Generator Study survey was undertaken for the Metropolitan Council (the regional metropolitan planning Survey Attribute or Question BOS2019 LAX 2019 MSP 2011 OAK 2014/15 MCO 2019 Responses 8,763 14,760 1,009 9,340 n/a Number of questions 44 90 57 52 6 Number of survey periods 1 2 1 Continuous Continuous Survey duration (days) 14 11 + 10 8 371 Date and time of survey x x x x Airline x x (g) x Flight number x (g) Originating/returning/connecting x x x x Trip purpose 2 4 8 7 7 Travel party size x 6 (b) 6 (h) (j) (l) Final trip destination x (c) x x Air trip duration (nights away or in area) x (d) x x Number of checked bags x 6 (e) x x Number of carry-on bags x Trip origin type 8 7 9 8 Trip origin address or zip code x x x x Ground access mode (and sub-modes) 19 (8) 20 (10) 11 (5) 15 (8) How transit was accessed 10 7 10 Departure time from ground origin x x Arrival time at airport x x Arrival time before scheduled flight departure 9 5 Number of well-wishers coming into terminal 6 Time before scheduled departure reached security 5 Security screening lane used 5 Ground egress mode (a) 17 20 11 14 Return or arrival time x x Access vehicle occupancy x 6 x Travel cost reimbursed x 4 x Residence zip code or country (foreign) x (f) x x Household income 10 11 8 Household size x 5 (i) Number of autos available to household x x Age x 7 x 7 Gender 3 4 2 3 Disability 6 Employment status 6 7 Air trips in past year 6 x (k) Notes: (a) on arrival for visitors to region or on return for residents of region; (b) by four age ranges; (c) four regions; (d) Southern California visitors only; (e) respondent only; (f) Southern California residents only; (g) inferred from gate number and date/time; (h) plus number age under 18; (i) plus number age 18 and under; (j) plus number age 12 and under; (k) at each of the four largest Bay Area/Sacramento region airports; (l) by six age ranges. An “x” indicates that the attribute or question was included in the survey, and a blank cell indicates that the attribute or question was not included in the survey; n/a = not available Table 10-2. Air passenger surveys.

Air Passenger Surveys 133   organization) rather than the Metropolitan Airports Commission in order to gather data on ground access travel to and from MSP for use in developing a stand-alone airport travel behavior model as part of the Metropolitan Council’s regional travel behavior inventory efforts. In conse- quence, the survey was only conducted on weekdays, since the regional travel demand modeling only considered weekday travel. The Oakland International Airport 2014/15 Ground Access Study survey was one part of a study conducted jointly at OAK and SFO using the same methodology and essentially the same questions and differing only in question wording and response options that were airport- specific. Although this survey was conducted at the two airports, there was no intent to use the survey results to examine regional air travel patterns, and two other commercial service airports in the region were not included in the survey. Rather, the objective of conducting the survey at both airports at the same time and using the same approach was simply to share the costs of designing and conducting the survey. Therefore, this survey has been included in the discussion of single-airport surveys. A particularly interesting feature of the survey is that it was conducted over a 13-month period from May 2014 to May 2015 and thus provided data on seasonal changes in air passenger characteristics. The Orlando International Airport 6-Question Survey was located on the airport’s website with an invitation on the survey web page asking website visitors to participate. However, at the time of writing, there were no links to the survey on the website, so it is unclear how airport users became aware of the invitation to complete the survey or find the web page. Table 10-2 does not attempt to show all of the questions asked in each of these surveys because some of the surveys were quite long (as indicated by the number of questions). However, it should be noted that, in most cases, not all respondents were asked every question. The LAX 2019 survey also asked multiple questions that were specific to respondents who started their trip to the airport from different types of origins (such as hotels or regional attractions) or specific to users of particular ground transportation modes, of which a large number were defined in the survey. An “x” in Table 10-2 indicates that the attribute or question was included in the survey in question. A numeric value for a travel party attribute indicates the number of predefined options that were included in the survey instrument. Where these included “other,” this was counted as one option, even if the respondents were allowed to specify details that were sub- sequently coded. Ground access and egress sub-modes primarily relate to whether air passengers using private vehicles or rental cars were dropped off at the terminal curb before the vehicle was parked or returned to the rental car company and where the vehicle was parked and for how long. Specific ground transportation service providers, such as different shared-ride van companies or different public bus operators, are not considered sub-modes. The distinction between different transit operators and different modes is often fuzzy because different transit operators may use different technologies. Thus, in some surveys, light rail and heavy rail may be considered different modes, while in other surveys they may be treated as sub-modes of “rail” (or even of “public transit”). The large number of different ground trans- portation modes and services at most airports, and particularly at large airports, presents a challenge in asking survey respondents how they got to the airport, since respondents may use different terminology for essentially the same mode. Some of the surveys documented in Table 10-2 asked respondents for the name of the transportation provider or transit operator that they used. This can be helpful in classifying responses if a respondent is unsure about the terminology of the different modes or misstated the mode used.

134 Guidebook for Conducting Airport User Surveys and Other Customer Research It can be seen from Table 10-2 that, although some characteristics of air travel parties were collected by all four of the longer surveys, other characteristics were only collected in some of the surveys. In addition to presenting different options for the same question, the surveys also varied in the level of detail for a particular question. The 2019 LAX survey only asked how many trips the respondent had made from LAX in the previous year (using six ranges), whereas the 2014/15 OAK survey asked for the actual number of trips made from each of the three largest airports serving the nine-county San Francisco Bay Area as well as from Sacramento Inter national Airport. 10.2.3 Multi-Airport Regional Surveys Two representative air passenger surveys undertaken by regional planning agencies in multi-airport regions are: • Originating passenger surveys conducted in 2015 at the two commercial service passenger airports in the Dallas/Fort Worth metropolitan region for the North Central Texas Council of Governments (Unison Consulting 2016a, 2016b), and • The Washington–Baltimore 2017 Regional Air Passenger Survey (National Capital Region Transportation Planning Board 2018). The surveys for the North Central Texas Council of Governments were performed at Dallas Love Field Airport (DAL) and Dallas/Fort Worth International Airport (DFW). The surveys for the Washington–Baltimore region were performed at Baltimore/Washington International Thurgood Marshall Airport (BWI), Washington National Airport (DCA), and Washington Dulles International Airport (IAD). The Washington–Baltimore Regional Air Passenger Survey is the most recent of a long series of such surveys that have been conducted every 2 years since 1998, representing the most consistent and frequent sequence of air passenger surveys for any multi-airport region in the country. The surveys were performed at each airport in the respective region using essentially the same survey questions (apart from minor differences in question wording and response options that were airport-specific), although the surveys differed between the two regions. The infor- mation collected in the two sets of surveys is shown in Table 10-3, although the table does not represent every question because some were asked to determine which options to offer the respondents, to obtain details for “other” answers, or to probe for more detailed information on specific responses, such as the parking facility used. The DAL survey asked some questions that were not included in the DFW survey. As with earlier tables, an “x” indicates that the attribute or question was included in the survey, and a numeric value for a travel party attribute indicates the number of predefined options that were included in the survey instrument, with “other” counted as one option even if the respondents were allowed to specify a detailed answer. Values separated by a semicolon are for surveys at DAL and DFW, respectively. The surveys in the Dallas/Fort Worth region obtained 2,138 responses at DAL and 8,379 responses at DFW. These surveys were only performed on weekdays. The surveys in the Washington–Baltimore region obtained 8,808 responses at BWI, 6,362 responses at DCA, and 6,517 responses at IAD. Although the primary purpose of both sets of surveys was to support regional travel demand analysis for trips to and from the airport, neither set of surveys asked about egress travel from the airport, unlike four of the five air passenger surveys described in Table 10-2. Furthermore, the surveys for the Dallas–Fort Worth region were only conducted on weekdays. While metro politan planning organizations typically only model travel on weekdays, it is quite likely that airport ground access and egress travel patterns are significantly different on weekends than on weekdays.

Air Passenger Surveys 135   10.3 Survey Methodology The circumstances under which air passengers spend time at an airport have an important influence on the choice of survey methodology. Passengers are available to be surveyed for only a relatively short time and may have activities they need or wish to undertake during this time. Departing passengers are concerned that they not miss their flight, while arriving passengers may have people waiting to meet them or be anxious to claim checked bags and be on their way. These constraints determine how passenger surveys can be performed. The three principal decisions on survey methodology are as follows: • Whether to interview passengers or use self-administered questionnaires • Where to perform the survey • When to perform the survey There are a number of issues specific to air passenger surveys that need to be considered in planning the survey and designing the questionnaire (as discussed in the following subsection). Survey Attribute or Question Dallas/Ft. Worth2015 Washington/Baltimore 2017 Responses 10,517 21,687 Number of questions 56; 50 24 Number of survey periods (each airport) 1 1 Survey duration (days) 25; 56 14 Date and time of survey x x Airline x (b) Flight number x (b) Originating/returning/connecting x x Trip purpose 6 7 Travel party size x x Final trip destination x x Air trip duration (nights away or in area) 4 x Number of checked bags (c) Trip origin type 6 5 Trip origin address or zip code x x Ground access mode (and sub-modes) 9 (6; 7) 11 (3) Reason for choice of mode 10 6 Departure time from ground origin x Arrival time at airport x Travel time to airport from ground origin x Number of well-wishers coming to airport x Number of vehicles used by travel party x Travel cost reimbursed x Residence zip code or country (foreign) (a) x Household income 6 8 Household size x x Number of autos available to household x Age 6 6 Gender 2 Ethnicity/race 7 Employment status 6 Notes: (a) zip code for Texas residents; state or country for others; (b) inferred from the surveyed flight and date/time; (c) respondent only. An “x” indicates that the attribute or question was included in the survey, and a blank cell indicates that the attribute or question was not included in the survey. Table 10-3. Multi-airport region air passenger surveys.

136 Guidebook for Conducting Airport User Surveys and Other Customer Research These include how to account for air passengers traveling together and the variation in air travel characteristics by time of the day and day of the week. 10.3.1 Issues Specific to Air Passenger Surveys The first issue to consider is that air passengers often travel in groups (air travel parties), so it is important for surveys to collect information on the composition of the air travel party. Whether it is best to collect data on the basis of the air travel party or from air passengers as individuals is really a function of the survey methodology. Because it may not be obvious who is in the same party when the questionnaires are distributed, self-administered surveys are typically given to every adult passenger. Handing out questionnaires to every passenger rather than to a selected sample is fairly low in cost and helps improve the overall response because some passengers may not complete the survey. If the survey involves survey staff interviewing respondents, typically only one passenger from each party is interviewed. In general, it makes no sense to ask the same questions of other members of the same party, although there are a couple of caveats to this. The first caveat is that with large air travel parties, such as tour groups, an interview survey may get multiple responses from the same party because the members of the party may not be standing or sitting together, and it may not be obvious who is in the party. Indeed, it may be desirable to get multiple responses from the same party because their characteristics may be different (e.g., they may have traveled to the airport separately). This situation leads to the second caveat. There may be a difference between the air travel party (often shortened to “air party”) and the ground access party (e.g., two colleagues going on a business trip together who travel to the airport independently from their homes). These issues have to be addressed in the question wording (see Figure 10-1). Including yourself, how many people are in your air travel party today? By air travel party, I mean all of the people who are traveling together with you on the same flight. If more than one person in air travel party, ask: And how many of these are children under the age of 18? If air travel party is more than six people, ask: Are you traveling as part of an organized group, such as a tour group, school party, or sports team? Yes/No If Yes: What is the name of this group? (Ask the following questions after asking about airport access mode used) If more than one person in air travel party, ask: Including yourself, how many of the people in your air travel party came to the airport together in the same vehicle as you? For transportation network companies (TNCs) (Lyft, Uber, etc.), ask: Did you share the ride to the airport with anyone from a different air travel party? For shared-ride modes only (shared-ride van, courtesy shuttles, scheduled airport bus, charter bus or van, or shared TNC), ask: How many other passengers (not including your air travel party or the driver) were in the vehicle when it arrived at the airport? Figure 10-1. Sample question wording to address travel party characteristics.

Air Passenger Surveys 137   A related issue is the extension from air party travel patterns to vehicle trips, which are typi- cally required for groundside and ground access planning. In some cases (e.g., rental cars or private vehicles parked at the airport), there is usually a one-to-one correspondence between the air party trip and the associated vehicle trip. In other cases (e.g., passengers dropped off at the airport by private vehicles), there will be two one-way vehicle trips for each one-way air party trip. In the case of taxis, hired limousines, or transportation network companies (TNCs; Lyft, Uber, etc.), there may be additional vehicle trips, depending on whether the operator is able to obtain a fare in the other direction. In the case of shared-ride modes, the number of air parties in each vehicle trip can vary widely. Therefore, for some modes, it is useful to determine how many of the air party were in the vehicle that the survey respondent traveled in to the airport and how many other passengers (not from the air party) were also in the vehicle. Figure 10-1 shows some possible question wording to address this issue. Since air passenger surveys generally only obtain responses from adult passengers, and respondents will report their own individual characteristics, such as age, some care is needed in reporting survey results. The average age of survey respondents will not be the same as the average age of air passengers. In order to account for the proportion of children in the air passenger population, questions about the number of passengers in each air party should ask how many of these are children younger than a specified age (typically 18). The differences in air and ground access party composition and characteristics raise the question of whether to present survey results in terms of air parties, ground access parties, or air passengers. Depending on how the results are going to be used, it may be desirable to present the results more than one way. From the perspective of ground access planning, it may be best to present results in terms of ground access party or air parties because, generally, each air party represents a single ground access decision (with the caveats noted previously). However, for shared-ride modes, one may want to know what percentage of air passengers use the mode because, generally, ridership statistics on such modes count people rather than parties. The bottom line is that to interpret the survey results properly, it is important to understand the distinction and be clear what is being shown. Knowing the average air party size for each mode from the survey, the data can always be re-expressed on whatever basis makes the most sense for a given issue. Another issue that has to be addressed in planning an air passenger survey is the variation in passenger characteristics over the time of day and the days of the week. Typically, a higher proportion of business travel occurs at the start and end of the day and on weekdays rather than weekends, although a significant amount of originating business travel may occur on Sundays. The pattern of travel is also different for residents of the area and visitors: residents tend to leave earlier in the day, while visitors are more likely to travel later in the day. Example of Variation Over Day of Week and Time of Day To illustrate these effects, Figures 10-2 and 10-3 show the variation in the composition of air parties for two airports in the San Francisco Bay Area: OAK and SFO. The composition varies both by day of the week (Figure 10-2) and time of day (Figure 10-3), although the variation is more pronounced by day of the week, particularly for OAK. There are also clear differences in traffic composition between the two airports. At OAK [Figure 10-2(a)], the proportion of total traffic accounted for by business travel is highest on Wednesdays, although this is only about 43%. The split between business travel by residents and visitors is quite different, with the highest proportion of resident business trips early in the week and the highest proportion of visitor business trips from the middle to the end of the week. Not surprisingly, business travel on the weekend is significantly lower than on weekdays.

138 Guidebook for Conducting Airport User Surveys and Other Customer Research (a) Oakland International Airport (b) San Francisco International Airport Source: Corey, Canapary & Galanis Research 2016 (project team analysis of survey response data). Figure 10-2. Variation in air party composition by day of week – departing air passengers.

Air Passenger Surveys 139   (a) Oakland International Airport (b) San Francisco International Airport Source: Corey, Canapary & Galanis Research 2016 (project team analysis of survey response data). Figure 10-3. Variation in air party composition by hour of the day – departing air passengers.

140 Guidebook for Conducting Airport User Surveys and Other Customer Research Personal travel shows the reverse pattern, with resident personal trips increasing toward the end of the week, with the highest proportion on Fridays. Visitor personal trips are at their highest proportion on Sundays and decline steadily during the following weekdays, reaching their lowest proportion on Thursdays and increasing again on Fridays and Saturdays. The variation of traffic composition by day of the week at SFO [Figure 10-2(b)] shows a similar pattern, although business travel remains strong through the end of the work week, with the highest proportion of visitor business trips on Thursdays. The highest proportion of resident personal trips occurs on Fridays and Saturdays. The traffic composition by hour of the day at OAK [Figure 10-3(a)] varies fairly steadily throughout the day, with the proportion of visitor trips increasing until about 5 p.m. (i.e., hour 17 on the 24-hour clock) and declining thereafter, with visitor personal trips reaching their highest proportion of total traffic at about 12 p.m. (hour 12). The traffic composition at SFO [Figure 10-3(b)] shows somewhat less variation by hour of the day, with a higher proportion of business trips than at OAK in the late evening after about 7 p.m. (hour 19). The survey results for OAK appear to show almost no business trips after 10 p.m. (hour 22 on), although the small sample sizes at these hours make these proportions statistically unreliable. At both airports, the proportion of business trips made by residents is highest in the early morning and declines progressively through the day with a sharp increase from 10 p.m. on. The traffic patterns at OAK and SFO illustrate the importance of ensuring adequate survey coverage for each day of the week and hour of the day. Performing a survey on only some days or during only some hours would bias the results. 10.3.2 Approaches to Surveying Air Passengers There are two different approaches to performing air passenger surveys in airport terminals: intercept interview surveys, in which survey staff select potential respondents and record their answers to the survey questions, and self-administered surveys, in which questionnaires are distributed to potential recipients to complete and return. The advantages and disadvantages of each, together with some practical considerations, are discussed in the following paragraphs. Intercept Interview Surveys Intercept interview surveys are more costly to perform than self-administered surveys because of the staff time required to perform the interviews, although there are data-entry costs with self-administered surveys using printed questionnaires that can be avoided with interview surveys using EDCDs. However, it is generally believed that the resulting data are of better quality because the interviewers can ensure that questions are not skipped, clarify questions for the respondents, resolve ambiguous or unclear responses, and attempt to obtain responses to open-ended questions—such as the trip origin address—in the level of detail required. Interview surveys can also include more-complex branching and follow-up questions because the respon- dent does not generally see the questionnaire. The two principal planning issues with intercept interview surveys are: • The protocol for selecting the potential respondents to approach and ask to participate in the survey, and • Whether to use printed questionnaires or EDCDs. In the case of interviews in airline gate lounges, because of time and staffing constraints, it is generally not feasible to interview every air party in the lounge, even if it is desired to maximize the number of completed interviews for that flight. Interviews can only take place over a limited

Air Passenger Surveys 141   period (typically 30 to 40 minutes) after enough passengers are in the lounge to provide a representative sample of air parties and before boarding commences. If the interview takes an average of 5 minutes, each interviewer would be able to complete six to eight interviews. A typical domestic flight with a narrow-body aircraft might have 70 or more air parties (105 passengers), depending on the size of the aircraft and the load factor. At least 10 interviewers would therefore be needed to survey every air party. Quite apart from the logistical difficulty of trying to conduct that many interviews simultaneously, passengers do not arrive in the gate lounge uniformly, so it would be necessary to interview a larger proportion of air parties closer to boarding time, further increasing the number of interviewers required. As a result, intercept interview surveyors usually attempt to survey only a sample of air parties. In the foregoing example, a team of three interviewers might be able to survey every fourth air party. Two issues arise in sampling passengers in an airline gate lounge. The first is that passengers will arrive in the lounge as the interviews are in progress. Some of these passengers will sit in areas of the lounge that have already been surveyed and therefore will be missed in the sampling process. Thus, passengers who arrive in the lounge well before boarding commences will have a higher probability of being sampled than those arriving closer to boarding time. This occurrence needs to be considered in weighting the survey results. The second issue is that, as the seating in the lounge fills up, some passengers may choose to stand. Members of large air parties, in particular, may not sit down, or some members of the group may stand while others sit. Passengers arriving in the lounge shortly before boarding is scheduled to commence may choose not to sit down, even if there are empty seats. Therefore, the sampling protocol needs to include standing passengers as well as those seated. Some passengers may also change seats or stand in a different part of the lounge as the inter- views are in progress. In particular, many passengers will get up and stand closer to the boarding point as the time to board approaches. Many airlines have formalized this process by establish- ing separate queues for passengers in different boarding groups or installing numbered markers in the queuing area corresponding to a sequence number on the boarding pass. In principle, when passengers leave an area that has not yet been surveyed and move to an area that has been surveyed or join a queue waiting to board the flight, it is the same as if they arrived in the lounge at the time they changed positions. Obviously, there is a small chance that passengers who change position in the lounge could be sampled twice. However, interviewers will usually recognize people they have just interviewed, and passengers will generally indicate that they have already been surveyed if they were inter- viewed by another interviewer. A sampling protocol should be defined for interview surveys that the survey staff will follow in order to avoid respondent selection bias (discussed in Section 5.3.2). The following protocols are examples: • For an airline gate lounge interview with two interviewers, one interviewer should start with the right-most passenger (as viewed by the interviewer) seated in the row of seats furthest from the airline podium, then select every fifth passenger, seated or standing, counting to the left and proceeding around the rows of seats, from the outermost seats toward the center of the lounge. The second interviewer should start with the right-most passenger seated in the Best Practices “From a sampling approach, we spend a lot of time training our surveyors to get a good cross-section of customer types. [We make] sure our folks are trained to survey all the way through [to] when the flight is boarding. You can have a quota and get to your flight early and [complete] your 10 surveys, but you’re going to miss folks who . . . get there closer to the time when the flight is boarding. If you’ve [got] all your responses from the same type of traveler (those who arrive early), then you’re missing an important subgroup.” —Research participant

142 Guidebook for Conducting Airport User Surveys and Other Customer Research row of seats closest to the podium, then select every fifth passenger, seated or standing, also counting to the left and proceeding outward from the seats closest to the podium in a general counterclockwise direction. • For a survey of passengers exiting security screening, the interviewer should select the next passenger to exit the screening area after completing each interview. In the first example, the logic behind sampling every fifth passenger (when two interviewers might expect to survey about one-seventh of the passengers on the flight) rather than every seventh passenger is to increase the sampling rate to allow for passengers who arrive after the surveying has started and sit in areas of the lounge that have already been surveyed, as well as those passengers who choose not to sit in the lounge at all or cannot find a seat and either stand outside the lounge or sit in adjacent gate lounges. Because the layout of seating in airline gate lounges varies widely, sampling rules should be flexible enough to accommodate the different layouts that the interviewers are likely to encounter. Interpreting the sampling rules in different situations should be part of interviewer training. Examples of respondent sampling sequences in two different gate lounge layouts are shown in Figure 10-4. Since the number of interviews that each interviewer can perform in a given period is fairly constant, the protocols described previously will result in a higher sampling rate during periods when the flow of passengers is reduced or when the flights have fewer passengers. Therefore, staffing levels may need to be increased during busy periods or when surveying flights that use larger aircraft in order to achieve a fairly consistent sampling rate. The trade-offs between the use of printed questionnaires and EDCDs have been discussed in Section 5.10. For use in air passenger surveys, EDCDs have a number of attractive features. Figure 10-4. Typical respondent sampling sequences in airline gate lounges.

Air Passenger Surveys 143   They can be programmed to ask questions that probe for detailed information on specific issues from certain respondents in a way that is transparent to the respondents and the interviewers. They can also be programmed to perform consistency checks on the response data and generate clarifying questions to resolve apparently inconsistent or implausible responses. Certain infor- mation, such as the time the survey was performed, can be entered automatically. Information that repeats from one interview to the next, such as the location where the survey is being per- formed, can be entered once and then automatically recorded for subsequent interviews until the interviewer indicates that this has changed. Figure 10-5 provides examples of consistency checks that could be programmed into EDCDs, and Table 10-4 provides detailed follow-up questions that could be asked to resolve unclear responses or provide additional information. Another useful feature of EDCDs is the ability to vary the questions asked of different respondents. For example, a question that asks whether air passengers might have changed their ground access mode if a proposed new service had been available could vary the price or other attributes of the service in the wording of the question in order to explore how these factors affect the responses. Self-Administered Surveys Self-administered surveys should be conducted in a location where respondents are able to fill out the questionnaire, which limits the use of this approach to the airline gate lounge or similar location, such as a common seating area serving multiple gates, where passengers are seated and have time to do it. The advantage of self-administered surveys is that a large number of survey questionnaires can be distributed and collected by one or two survey personnel. A common procedure is to distribute the questionnaires to passengers as they enter the gate lounge area and collect them as the passengers board the aircraft, although some passengers may hand the completed questionnaires back before boarding commences. An important consideration is whether to attempt to survey passengers who arrive at the gate after boarding has begun. These are likely to include passengers who have been spending time in airline club lounges as well as connecting passengers whose inbound flights were delayed. The following consistency checks assume that EDCD programs can access relevant ground access service and other information. • Check that reported airline and flight number is reasonable for start time of interview and reported final destination of trip. • Check that reported parking facility and parking duration are reasonable for reported air trip duration and city of residence. • For reported use of ground transportation services with limited geographical availability (e.g., a shared-ride van operator serving only part of the region), check that reported trip origin is within service area. • For reported use of fixed-route ground transportation service from a given stop, check that the stop is a reasonable choice from reported trip origin. • For reported use of hotel/motel courtesy shuttle, check that reported trip origin hotel/motel offers courtesy shuttle service. • Check that number of air travel party members reported coming to airport in same ground access vehicle is not greater than reported size of air travel party. • Check that reported number of air travel party members coming to airport in same ground access vehicle is reasonable for reported vehicle. • Check that reported arrival time at airport is earlier than start time of interview by a reasonable amount. • Check that reported ground access time from trip origin to airport is reasonable for reported trip origin and ground access mode. • Check that reported number of checked bags is reasonable for reported size of air travel party. Figure 10-5. Sample consistency checks.

144 Guidebook for Conducting Airport User Surveys and Other Customer Research Excluding these passengers may bias the sample if those arriving close to flight departure time have different characteristics from those arriving earlier, as is quite likely. In general, passengers arriving at the gate once boarding has started will not have time to complete the questionnaire before boarding and, therefore, if survey questionnaires are to be distributed to these passengers, they will need to be designed so that they can be completed later (e.g., on the flight) and returned by mail. In this case, it will be desirable to have a second staff member collect the questionnaires as passengers board the aircraft so that the person distributing the questionnaires to the arriving passengers can continue to do so. The designers of the survey questionnaires need to consider that respondents may not have a flat surface to write on. Printing the questionnaires on thin cards with a three-part fold pro- duces a questionnaire that is convenient to handle and sufficiently rigid to write on while holding in one hand. Some examples are shown in Appendix E. Survey staff should have an adequate supply of pencils or pens to give to respondents who do not have them. Purpose Follow-Up Questions Resolve Inconsistencies Just to confirm that I have this correct, you stated that you began your trip in (city) and boarded the (airport bus) at (stop). Did I record that correctly? Yes/No If Yes: Did you travel directly from your trip origin to the stop, or did you stop somewhere along the way for some other purpose? (Record location of intermediate stop, if applicable, using trip origin questions.) If No: Return to relevant question and revise response. Just to confirm that I have this correct, you stated that (number) of your air travel party came to the airport in the same vehicle. Did I record that correctly? Yes/No If No: Revise relevant response. Just to confirm that I have this correct, you stated that you began your trip in (city) at (time) and arrived at the airport at (time). Did I record that correctly? Yes/No If Yes: Did you travel directly from your trip origin to the airport, or did you stop somewhere along the way for some other purpose? (Record location of intermediate stop, if applicable, using trip origin questions.) If No: Return to relevant question and revise response. Just to confirm that I have this correct, you stated that you arrived at the airport (number) hours ago at (time). Did I record that correctly? Yes/No If No: Revise relevant response. Just to confirm that I have this correct, you stated that the (number) members of your air travel party checked a total of (number) bags. Did I record that correctly? Yes/No If No: Revise relevant response. Obtain Additional Information for Specific Access Modes Rental car Scheduled airport bus Rail system Hotel/motel courtesy shuttle Were any passengers dropped off at the curb in front of the terminal before returning the rental car? Yes/No. At what stop did you begin your trip on the (airport bus)? (Check response option or write in.) How did you get to that (airport bus) stop? (Check response option or write in.) At what station did you begin your trip on (the train)? (Check response option or write in.) How did you get to that (train) station? (Check response option or write in.) Did you stay overnight at that hotel, or did you visit the hotel only for the purpose of getting to the airport? (Check response option.) Did you park at that hotel, or did you get to the hotel some other way? (Check response option or write in.) Table 10-4. Representative follow-up questions.

Air Passenger Surveys 145   A number of issues may arise with self-administered surveys and need to be carefully considered in the wording of the survey questions, design of the questionnaire, and analysis of the results: • When distributing the questionnaires, it will not always be possible to determine which passengers are in the same air party, so it is common practice to distribute questionnaires to every adult passenger. In some air parties, more than one passenger will provide responses, while other passengers may decide not to complete the questionnaire if they see someone else in their air party doing so, or they may complete it together. So that these responses are not double-counted, it will be necessary to ask about the size of the air party and identify multiple responses from the same party before analyzing the results. Some surveys ask how many respondents from the air party have responded to the survey. However, experience has shown that these answers are often unreliable, possibly due to respondents misunderstanding the term used to describe the air party or to other members of their party starting to fill out the questionnaire but not finishing. Therefore, it may be safer to identify multiple responses on the basis of the information provided, or validate the answers to a question on multiple responses from the same air party by analyzing other information provided in the survey responses from each respondent. • Respondents may not be familiar with local terminology, particularly for ground trans- portation services. It is important to describe response options in terms that can be generally understood, rather than use the names of particular local services or facilities. Having respon- dents provide the name of the transportation provider, service, or facility they used, where appropriate, can help resolve misunderstandings over terminology such as “airport bus” or “off-airport parking.” • Skip or branch patterns, when used, should be clearly shown on printed questionnaires, with bold arrows directing the respondent to the next relevant question. Where questions only apply to some respondents (e.g., which parking lot was used), it is better to explicitly direct those respondents to the question with a branch from a previous question rather than ask the question in a way that requires each respondent to decide whether to answer it. • Checkboxes for response options should not be too small or too close together. If the check- boxes are too close together, it may be difficult to determine which one a respondent intended to check. • Fonts used on printed questionnaires for text that the respondent is expected to read should be no smaller than 10 point. Many respondents may have poor eyesight, particularly in the prevailing lighting conditions in a gate lounge area. Survey procedures should also address what to do if a potential respondent cannot read or write. One solution is to have a member of the survey staff conduct an interview survey after the questionnaires have been distributed. This should be coordinated with the rest of the survey field team so that at least one team member continues to distribute questionnaires as passengers arrive at the gate. To identify responses from the same air party in self-administered surveys, the analyst can make use of such information as the following: • Trip origin address (or other location information) • Final destination for air trip • Home zip or postal code • Air party size • Air trip duration or duration of visit • Departure time from trip origin • Ground access mode used

146 Guidebook for Conducting Airport User Surveys and Other Customer Research While this process can be partly automated, there will be questionable cases where the analyst will not be able to resolve whether two responses are from the same air party, and inspection of the detailed response data will be necessary to make a decision. This process adds to the data-cleaning workload and constitutes one of the trade-offs with the lower cost of self-administered surveys. One issue that arises with self-administered surveys is what to do if members of what is obviously the same party (e.g., they started from the same address and are on the same flight to the same destination) give conflicting answers to other questions. This happens more often than might be expected. It may be naive to assume that every survey respondent understands the questions and gives the correct answer. There is also the possibility that apparently conflicting answers are in fact correct—for example, different members of the same party may have different trip purposes or may be returning at different times—or that the apparent inconsistency is a coding error or the result of an unclear response. In cases where both conflicting responses are unlikely to be correct (e.g., passengers traveling on the same ground access vehicle giving different departure times from the same trip origin), it will be necessary to define a rule for which response to accept. In some cases, examining the responses to other questions may resolve the issue. For example, one departure time may be physically impossible, given the arrival time at the airport, or be an obvious error, such as recording the time as a.m. rather than p.m. It should also be borne in mind that for some questions (such as the time of departure from the trip origin), respondents will be giving their answer to the best of their recollection. Minor differences in answers from members of the same air party are to be expected. Hand-Out/Mail-Back Surveys Because of the very low response rate that is typically experienced with mail-back surveys, this approach should only be used in situations when there is not enough time for the respon- dent to complete the survey and it is not practical to collect the completed questionnaires later. Mail-back questionnaires should include a prepaid return envelope or be designed to be folded so that the return address with the prepaid mail information forms one side. This approach is sometimes used for passengers who arrive at a boarding gate too close to flight departure time to complete an interview survey. Since this involves additional logistical measures, such as arrangements for prepaid return mail, and generally results in a fairly low response, survey sponsors may decide that these measures are not worth the effort and accept that the survey results will exclude passengers who arrive at the gate close to flight departure. This approach can also be used for surveys of those parking in an on-airport parking facility or boarding ground transportation modes where there is insufficient time to conduct an inter- view survey. 10.3.3 Survey Locations The choice of location to perform an air passenger survey has significant implications for the logistics involved as well as the ability to obtain a representative sample of the target population. This section discusses the advantages and disadvantages of the different options in meeting the objectives of an air passenger survey. Nonsecure Locations Inside the Terminal Building. Within the terminal building there are typically two possible nonsecure areas for conducting interviews: one area serving arriving passengers and one primarily serving departing passengers.

Air Passenger Surveys 147   The nonsecure arrivals area of the terminal can be either open or enclosed; the former is typical of smaller domestic operations, and the latter is more usual at larger domestic or inter- national operations. The domestic arrivals hall usually provides an opportunity to interview arriving air passengers with any greeters (if present) while passengers are waiting for baggage. However, passengers without checked baggage would not be included in such a survey pro- cess. This is a significant limitation, since passengers without checked bags differ from those with checked bags in many ways that are likely to affect the survey results. For some surveys, passengers have been intercepted as they exited the secured area of the terminal and have been interviewed as they walked to baggage claim or the terminal curb, although this limits the number of questions that it is practical to ask. For international arrivals, the greeters will typi- cally be waiting—en masse, which can make interviewing difficult—in front of the exit from the customs and immigration hall. While it may be possible to interview the greeters as they wait, once the passengers have joined them, the greeters and passengers will leave, and the opportu- nity to interview them will have been lost, although it may be possible to intercept them as they leave the waiting area. The most suitable nonsecure departures area is the lobby in front of the ticket and check-in counters. This area may include other services, such as food concessions and convenience stores. Well-wishers could still be with the air passengers in this area, and members of the entire group can be interviewed. While the lobby is a possible survey location, it is not recommended for surveying air passengers, who are likely to be more willing to take the time to be surveyed once they are past security screening, but should be considered as a location to obtain informa- tion from well-wishers. Although it is possible to perform an intercept survey of air passengers before they join the security screening line, passengers are usually anxious to complete the check-in and security screening process and may be reluctant to take the time. Passengers may spend some time waiting in line, either for check-in or security screening, but this environment is not particularly good for interviews. The passengers need to move with the line, which interrupts their attention to the interview, and they are usually near other passengers, which may make them reluctant to answer some types of questions. During less busy periods, the line may be too short to complete the survey before the passengers reach the check-in counter or screening location. Also, inter- viewing passengers in the check-in lobby will miss any passengers who already have a boarding pass, which is becoming more common now with Internet and cell-phone check-in, and are not checking bags. The best time to survey passengers before security is as they join the line for security screen- ing, because this line will include all passengers. If the line is fairly short, passengers may not mind being asked to step aside to complete the survey. It is advisable to determine their flight departure time to ensure that there is enough time for them to complete the survey without missing their flight. If the line is long, their willingness to participate may be increased if they can be offered the opportunity to go to the head of the line afterwards. This of course will require the agreement of the local TSA staff, but for passengers who may miss their flight, moving them to the head of the security line is sometimes done anyway when the line is long. The problem of potential bias from not surveying late-arriving passengers is no different from any other survey location, although it may involve a higher proportion of passengers, and it can be addressed by having mail-back questionnaires to distribute to such passengers. Groundside Locations. Intercept surveys of air passengers can also be performed in nonsecure locations on the groundside of the airport, such as the terminal curb front, parking lots or pay- ment machines, transit stations or boarding areas, rental car facilities, and inter-terminal shuttle bus stops or people-mover stations. (Section 10.12 discusses groundside surveys in more detail.)

148 Guidebook for Conducting Airport User Surveys and Other Customer Research With the exception of the terminal curb front, these locations will only allow a subset of air passengers to be surveyed. However, they may allow a larger sample of this subset to be obtained to supplement a more general sample of air passengers obtained at other locations, or the subset may be the target of the survey. For example, the proportion of air passengers using transit is typically fairly small. Thus, a survey in airline gate lounges will obtain relatively few responses from passengers who used transit to access the airport. If the survey sponsor is particularly interested in collecting data on transit use, it would be helpful to obtain a larger sample of transit users by surveying passengers at transit stops or stations. If the survey sponsor wishes to obtain information on air passengers’ airport egress travel rather than their access travel, surveys could be conducted in the baggage claim areas, at the terminal curb front, or as passengers exit the secure area of the terminal. Such surveys need to be fairly short because respondents will generally want to quickly complete their journey or may be anxious to meet some scheduled or pre-arranged ground transportation. A survey of passengers exiting the secure area of the terminal will intercept a sample of all arriving passengers, but some may not have made ground transportation arrangements and will be anxious to arrange travel. Also, this is the point at which arriving passengers are often met by greeters, which is not an ideal situation in which to perform a survey. As with many aspects of conducting air passenger surveys, there is no ideal solution. The choice of location to survey arriving air passengers involves a trade-off between the ability to sample all arriving passengers, their willingness to be surveyed, and the extent to which respon- dents know how they will reach their final destination, as well as such considerations as whether there is adequate space to perform the survey without obstructing the flow of other passengers and how many locations can be staffed at a given time. Experience with airport user surveys shows that people are generally cooperative and may go out of their way to assist with the survey. This cooperativeness is no different in the nonsecure area than in the secure area, although the time that respondents may be willing to spend being interviewed will often be much less. Other Considerations: Interview Time. Intercepting airport users in the nonsecure area is subject to a significant time constraint. Departing passengers may still have many steps to complete before boarding, including check-in, baggage drop, security screening, and perhaps shopping or obtaining a meal. It is therefore imperative that the survey be short. A good guideline is one single-sided questionnaire. If all the questions cannot fit into this space, the questionnaire is likely too long. It is possible to conduct a meaningful interview— obtaining many key characteristics or opinions of the air passengers, greeters, and well-wishers traveling in a group—in under a minute, although this is highly dependent on the skill of the interviewer. A maximum interview time of 2 minutes is recommended. For longer surveys, alterna- tive locations on the secure side of the terminal—or different methods, such as mail-back questionnaires—should be considered. Security Clearance. In the current airport security environment, personnel with access to the secure side of the terminal must undergo security clearance and be issued identification badges. Conducting a survey on the nonsecure side may remove some of this constraint, depending on the particular requirements of the airport. Even so, it may be worthwhile to issue survey field staff with identification badges and authorization letters in case they are challenged on their right to conduct the survey, as well as to reassure potential respondents that this activity is officially sanctioned.

Air Passenger Surveys 149   After Security Screening An alternative approach is to intercept passengers as they exit security screening. This approach has the advantage that air parties are generally still together, and the survey will intercept all passengers clearing security, whether they go directly to their gate or not. Passengers who do not go to their gate until the flight boarding time have an equal chance of being included in the sample as those that spend the time before the flight waiting in the gate lounge, thereby reducing any potential bias that can occur if passengers are interviewed in the gate lounge, as discussed in the next subsection. Intercepting passengers as they exit security screening also samples passengers on all flights departing from the gates served by the security screening channels, not just those on flights from selected gates. However, surveying passengers as they exit security screening is only suitable where informa- tion is not being collected on connecting passengers. Most connecting passengers do not pass through security unless they are changing terminals that are not connected beyond security, and the characteristics of the small number of connecting passengers that do go through security will likely be different from those of other connecting passengers. There are two other potential disadvantages with this location: • Passengers with limited time before their flight is due to depart may be anxious to reach their gate. The airlines may also be concerned that the survey will delay passengers. • There is often no seating in this area, so the interview may have to be conducted with the passengers standing. The first issue can be addressed by asking the passengers what time they were told to be at the gate for boarding and the gate number (for large terminals). This information is usually printed on their boarding pass. Asking them to check their boarding pass ensures that they give the correct information and provides a way of verifying their flight number and gate. If there is insufficient time to conduct the survey, the interviewers can ask if they can accompany the passengers to the gate and perform the interview there. Generally, there will be enough time to complete the survey in the gate area while other passengers are boarding. If passengers appear elderly or uncomfortable standing for the interview, the interviewers can ask if they would like to go to a nearby gate area to be interviewed so that they can sit down. Alternatively, it may be possible for the airport to provide some temporary seating in the area where the survey is being performed. Because surveying passengers as they exit security screening will obtain a sample of all originating passengers departing on flights from the gates served by that security checkpoint, but the sampling rate will most likely vary between the different checkpoints, it will be neces- sary to weight the results to reflect the total number of originating passengers on those flights. Passengers should be asked if they are connecting between flights so that they can be excluded from the weighting of responses in the analysis. Because the sampling rate will vary with the flow through the security checkpoint, and because the interview rate will be relatively con- stant for a given staffing level, it will be necessary for the weight assigned to a given response to reflect this varying sampling rate. If possible, passenger throughput counts should be obtained from the TSA in 10- or 15-minute intervals to allow for short-term fluctuation. Where these counts are not available, it will be necessary to estimate the flow from the flight schedule and the estimated passenger load on each flight. The distribution of the time before scheduled flight departure that passengers on a particular flight clear security can be estimated from the survey results and then applied to the estimated passenger loads to esti- mate the flow rate through security, with appropriate adjustments for connecting and through passengers.

150 Guidebook for Conducting Airport User Surveys and Other Customer Research Airline Gate Lounges One of the most common locations for an air passenger survey is the airline gate lounge. Passengers in the lounges are generally seated and are usually willing to participate in a survey. They tend to remain in the same seats, and the seats are typically grouped in rows, which in the case of an interview survey facilitates a consistent approach to selecting passengers to survey. However, there are a number of disadvantages to performing a survey in this location and some aspects that need careful planning. The most obvious consideration is that passengers in a given gate lounge are generally waiting for the next flight to depart from that gate and thus are all traveling to the same destination airport, although this may not be their final destination if it is an airline hub or international gateway. There must therefore be careful selection of flights to survey in order to obtain a representative sample of air passengers in all markets. Some passengers in the lounge may be waiting for a later flight or for a flight from a nearby gate. Should such passengers be included in the sample? On the one hand, including these passengers in the survey may result in obtaining fewer responses from the other passengers than anticipated in the flight sampling plan. On the other hand, including them may provide survey responses from a broader sample of flights than those included in the flight sampling plan. In cases where the survey is attempting to obtain responses from every adult passenger on a flight, including a few responses from passengers on other flights may unduly complicate the response weighting process or lead to biased results. It may be preferable to ask passengers which flight they are taking when handing out the questionnaires and only survey those on the sampled flight. However, with interview surveys, because of time and staffing constraints, responses will be obtained from a relatively small number of passengers on the sampled flights anyway. Use of weighting factors that expand the responses to reflect the variation in passenger traffic by time of day and day of the week in fairly broad groups of markets will generally be preferred to applying large weights to the few responses from a given flight based on the estimated passenger load for that flight. This approach avoids potential bias from over-weighting survey responses from flights for which a smaller proportion of air parties was interviewed. A major disadvantage of performing an interview survey in airline gate lounges is the small opportunity that interviewers will have to survey passengers who arrive at the gate just before or after boarding begins. If those arriving at the gate close to flight departure time have different characteristics from those arriving much earlier, as is quite likely, the survey will give biased results for those characteristics. This bias can partly be addressed by providing such passengers with a mail-back survey and through the process of weighting survey responses, discussed in Section 10.8. Another practical difficulty with airline gate lounge surveys arises from the limited time window for any given flight during which passengers can be surveyed. If interviews are started too soon, there will be nobody in the lounge to survey. Once boarding has commenced, it will generally be difficult to get passengers to agree to participate. This time window will generally begin about an hour before the scheduled flight departure time and end when boarding starts, which can be as much as 45 minutes before flight departure. In the case of flights using wide-body aircraft, passengers will often begin arriving in the gate area somewhat earlier, and there will usually be enough passengers in the lounge to begin interviewing an hour and a half before the scheduled flight departure time. However, boarding often begins even earlier than for flights using narrow-body aircraft. Some international flights start boarding as much as 60 minutes before flight departure, and there may be enough passengers in the gate lounge 2 hours before departure to allow interviews to begin.

Air Passenger Surveys 151   Once a survey team has finished surveying a particular flight, it will need to move to the next flight to be surveyed (unless the team is scheduled to take a break). The next flight should therefore have a scheduled departure time at least an hour later (an hour and a half in the case of a wide-body aircraft and 2 hours for international wide-body flights). This requirement limits the flight sampling plan because there may be few flights departing around that time. If those that are departing around that time are leaving from gates some distance away, additional time will be required for the survey team to travel between gates, particularly if this travel involves going through security screening again. As discussed in more detail in Section 10.6, develop- ing a flight sampling plan that provides reasonable coverage of different flight destinations and airlines—while utilizing survey teams efficiently and allowing staff to take required breaks at appropriate intervals—is a major challenge. 10.4 Issues with Interview Surveys The two principal issues with interview surveys are: • Where to conduct the survey, and • How to select a representative sample of air passengers to interview. Possible survey locations, as well as their respective advantages and disadvantages, are dis- cussed in Section 10.3.3. In addition to the issues discussed there, a major challenge with interview surveys is ensuring adequate coverage of different days of the week and times of day, particularly if the surveys are performed in airline gate lounges. Even if the flights selected for the survey are reasonably well distributed across airlines and flight destinations, and survey interviews take place throughout the day and on every day of the week, the limited number of flights that can be included in the survey will inevitably mean that flights to a given destination will only be surveyed at particular times of day or on some days of the week. Therefore, in selecting the flights to be surveyed, it is important to ensure that the selected flights include some flights to broad market regions at different times of day and each day of the week, even if it is not possible to survey flights to a specific destination at different times of day and days of the week. This issue is less of a concern with surveys that intercept air passengers at a common location, such as the exit from security screening, since this will obtain responses from passengers with destinations in all markets served by that location. As long as the survey interviews are conducted throughout the day and on all days of the week, the survey should obtain responses from passengers with a given destination at all times of day and days of the week. However, where there is more than one such location, such as separate security screening channels for different concourses, boarding areas, or passenger terminals, then it is important to assign the interviewers so that each location is surveyed at different times throughout the day and on each day of the week. 10.5 Issues with Online Surveys Online surveys have a number of attractive features. Typically, respondents can complete them at a convenient time and at their own pace. If the software allows, they can leave the survey website and return later to continue with the survey. It is possible to display images as part of questions. As with interview surveys programmed on to tablets or other EDCDs, it is possible to tailor or skip questions based on responses to earlier questions. With the near universal use of smartphones, most air passengers will easily be able to access a survey web page once they have been given a link to the page and have connected to the airport Wi-Fi. Alternatively,

152 Guidebook for Conducting Airport User Surveys and Other Customer Research or in addition, a tablet or other EDCD can be mounted in a kiosk and potential survey respondents directed to the kiosk and invited to com- plete the survey. Some airports have invited air passengers to complete a survey when they connect to the airport Wi-Fi system or even required a short survey to be completed as part of connecting to the Wi-Fi system. The latter has proven to be unpopular with those accessing the Wi-Fi system since they are usually doing so in order to get something else done and resent the interruption. Such surveys also necessarily have to be fairly short, which limits the amount of information that can be obtained. Perhaps the greatest challenge with online surveys is ensuring that responses are obtained from a representative sample of air passengers. This will generally require field staff to invite air passengers to partici- pate in the survey. Potential participants can be given a card with the link to the survey. While this incurs the cost of the field staff, issuing the invitations typically takes much less time than conducting interview surveys, so this may still be a less costly approach than an interview survey. A key factor is the percentage of invited participants who actually complete the survey. If this percentage is fairly low, this approach may not in fact be less costly than an interview survey, and there would be the concern that those completing the survey may not be represen- tative of the air passenger population. Where an airport has a reasonably extensive list of email addresses for air passengers (such as through requiring those accessing the airport Wi-Fi system to provide their email address), invitations can be sent by email. However, depending on the source of the list, it may not be fully representative of the air passenger population. For example, those arriving at the airport fairly close to their flight departure time or spending time in the concessions before their flight may not have the time or need to do anything that would require access to the Wi-Fi system. One potential application of online surveys is to obtain survey responses from passengers who arrive at a location where an interview survey is being conducted with insufficient time to complete the survey, such as reaching the airline gate lounge after boarding has commenced. Such passengers can be handed a card with the link to the online survey and invited to participate in the survey that way. In principle, this is no different from handing such passengers a printed questionnaire to complete and mail back, but an online survey can contain more complex skips and include consistency checks, and of course avoids the need for data entry and verification. Some survey respondents may find an online survey more convenient than having to fill out a printed questionnaire and remember to mail it. 10.5.1 Panel Surveys One approach to conducting an online survey is to make use of an existing panel of potential respondents. This approach is typically used by those firms conducting syndicated research studies, which usually recruit and maintain their own panels of people who have agreed to participate in online surveys. Some survey research firms recruit and maintain panels of potential survey participants that can be used for a specific survey, either performed by the firm or with the firm acting as an intermediary between the client and the panel members. Typically, some screening will be undertaken to ensure that the panel members participating in the survey conform to a desired target population (for example, people who have made an air trip in the past 3 months that used a particular airport, or residents of a particular region). Best Practices “I created [a Wi-Fi survey] program at [the] airport, and I think we were actually the first to build a formal research program around it. [We were] collecting about 2,500 completed surveys a day to a five-question survey of people that hit our Wi-Fi program. The nice thing about it is that we were able to do ongoing tracking research and do some ad-hoc surveys based on needs.” —Research participant

Air Passenger Surveys 153   A major concern with panel surveys is ensuring that the panel members participating in it are reasonably representative of the target population. Therefore, analysis should be under- taken to compare the relevant characteristics of the survey respondents to those of the target population. Minor differences in the distributions of given characteristics between the survey respondents and the target population can be corrected through the use of weighting (see Section 10.8), although no weighting can correct for segments of the target population that are missing entirely or are so under-represented that the reliability of the survey results for those segments is doubtful. Obviously, this presumes that information is available on the relevant characteristics of the target population. Some thought should be given to defining relevant control statistics for which reasonably accurate data are available, such as the distribution of trips by air travel market, day of week and time of day of travel, and parking duration of vehicles parked at the airport. Other control statistics may be obtained from the results of intercept surveys conducted at the airport in question, although these are inherently less reliable than statistics for which accurate counts are available since they may have a sampling bias themselves or may have been collected at a different point in time. 10.6 Sampling Approach Once the details of the survey methodology have been determined, the next steps are to decide on the required sample size, develop a sampling plan to provide adequate coverage of the variation in air passenger characteristics over time, and decide when to perform the survey. 10.6.1 Sample Size The details of calculating sample sizes for different sampling methods are discussed in Chapter 4. In air passenger surveys, the most common type of sampling method uses cluster sampling to select a random sample of flights as clusters and then attempts to survey either all the passengers on each selected flight (with a self-administered survey) or a sequential sample of passengers on each flight (with an interview survey). The accuracy of a cluster sample depends on both the variance of the characteristic of interest within each cluster and the variance between clusters. Where each cluster is a flight, whether the mean of a given characteristic varies signifi- cantly between flights will depend on the characteristic. However, many characteristics—such as airfares, ground access mode use, air trip duration, and air travel party size—are likely to vary significantly by destination, and hence their sample mean will vary between different flights. Of course, these characteristics vary widely within a given flight as well. Therefore, it will generally be necessary to use a larger sample size with flight-based cluster sampling than with random or sequential sampling of the air passenger population in order to achieve a similar level of accuracy. As a practical matter, the only way to perform random or sequential sampling of the air pas- senger population without resorting to a flight-based cluster sample is to interview passengers at a location where all passengers can be intercepted, such as the exit from security screening. The extent to which a flight-based cluster sample should increase the sample size to pro- vide an equivalent level of accuracy to that calculated for a random sample is dependent on the specific traffic composition at each airport, the characteristics of interest, and how these vary between passengers on particular flights and on average between flights. However, as a practical matter, information on how the relevant characteristics of the respondents vary

154 Guidebook for Conducting Airport User Surveys and Other Customer Research between flights is not typically known before conducting the survey. In the absence of more specific analysis, it would be prudent to increase the sample size by a factor that depends on the number of flights sampled and the average number of passengers interviewed per flight, as discussed in Appendix A. 10.6.2 Estimating the Population of Airport Users on a Given Day To develop a sampling plan for a given day, it is desirable to have an estimate of the number of air passengers (and possibly the associated greeters and well-wishers) using the airport on that day. If the sampling plan is being developed a fairly short time before the survey, it may be possible to obtain expected passenger loads on each flight from the airlines. If the sampling plan is being prepared well before the start of data collection or the airlines are not willing to provide this information, the population of these categories of airport users per day can be estimated as follows: • Passengers. Obtain a schedule of departing flights (and arriving flights if interviewing arriving passengers) for the survey period from the airport operator or sources such as the OAG (see bibliography). The schedule should indicate the flight departure (or arrival) time, airline, flight destination (or origin), terminal, aircraft type and seats, and whether it is a through flight or originating (or terminating) at the airport. This information can be deter- mined from the full flight itinerary or routing. For each flight, estimate the load factor using past data for the airline and city pair. This estimate could be further refined, if necessary, to allow for variation in load factor by month, day of week, and time of day where data for this level of detail are available. For through flights, estimate the proportion of passengers who will continue on the flight and not leave the aircraft. (The difference between arriving/departing and enplaned/deplaned passengers.) Some airports require airlines to report passenger loads by flight. Monthly data on passenger loads by airline and flight segment are available from the U.S. Department of Transportation Bureau of Transportation Statistics T-100 database. The proportion will vary by routing, airline, and airport and may need to be estimated based on limited qualitative inputs (e.g., comments by gate staff or other knowledgeable people). Multiply the seats, load factor, and proportion of passengers not continuing on the flight to estimate the enplaning (or deplaning) passengers. The daily or hourly passengers are then found by summing the estimated passengers over the flights in that time period. Connecting passengers do not make use of the groundside facilities or transportation system and can be excluded, where necessary, by multiplying by the ratio of origin–destination (O-D) passengers to enplaned/deplaned passengers for that airport. This approach can be extended to esti- mate the proportion of the passenger population in different market segments (e.g., domestic versus international passengers, or passengers using a specific terminal), based on the flight destination (or origin in the case of arriving flights). • Greeters and well-wishers. Numbers vary greatly with airport size, types of service available, proportions of business and visiting (nonlocal) passengers, and public transportation services to the airport, among other factors. The numbers of well-wishers and greeters that come into the terminal can be estimated based on the observed (or assumed) ratio of greeters and well-wishers to O-D passengers. While these vary significantly by airport, month, day of week, and time of day, a rough guide to the average number of well-wishers per originating passenger based on a number of airport surveys conducted during the period from 2003 to 2007 is shown in Table 10-5. While these surveys are now somewhat dated, there is no reason to expect that at any given airport the ratio of well-wishers to originating passengers has changed significantly. However, Table 10-5 makes it clear that the ratio appears likely to vary widely across different airports and markets, so airports should consider collecting their

Air Passenger Surveys 155   own data on the ratio of well-wishers to originating passengers in the course of performing air passenger surveys. The number of well-wishers can be estimated by multiplying the number of originating passengers by an appropriate factor based on the data in Table 10-5. Note that many of these well-wishers will be in groups and may be seeing off several passengers. Numbers of greeters are often assumed to be similar to the numbers of well-wishers, although this may not be the case in a given situation. Therefore, airports would ideally collect, from air passenger surveys, separate data on the numbers of well-wishers and greeters. Section 10.11 provides additional discussion of estimating the numbers of greeters and well-wishers. 10.6.3 Sampling Strategy The design of the sampling strategy is key to obtaining reliable results in an air passenger survey. A poorly designed sampling strategy can exclude certain subgroups from the sample entirely and lead to biased results. As discussed in Chapter 4, a controlled sample attempts to design the sampling strategy so that the composition of the sample reflects the underlying distribution of the population characteristics fairly accurately. In the case of air passenger surveys, this means that the sample has the same (or very similar) proportions as the population for such characteristics as the following: • Airline • Flight departure time • Day of the week • Destination • Originating versus connecting traffic • Domestic versus international trips In practice, this is difficult to achieve. Each of these characteristics varies independently of the others, resulting in a huge number of potential combinations of characteristics, and the logistics of performing the survey limit the ability to vary the sampling rate to match the underlying distribution of characteristics. In particular, intercept interview surveys tend to under-sample during busy periods and over-sample during less busy periods because it is difficult to schedule the number of interviewers to match the changes in passenger flow. When the sample is not controlled to ensure that the composition of the sample corresponds to that of the population, the results of the survey can be weighted so that the reported results reflect the composition of the population rather than that of the sample. The process for determining the weights to be used is discussed in Section 10.8. Airport Size Terminal Type Airport Year Well-Wishers per Originating Passenger Large hub International New York JFK Terminal 4 2003 0.29 International San Francisco, California 2006 0.17 Domestic San Francisco, California 2006 0.14 Medium hub Domestic San Jose, California 2003 0.16 Domestic Oakland, California 2006 0.18 Domestic Winnipeg, Manitoba 2007 0.25 Small hub Domestic Birmingham, Alabama 2005 0.09 Domestic Quebec City, Quebec 2006 0.19 Domestic Victoria, British Columbia 2006 0.35 Source: Biggs et al. 2009. Table 10-5. Representative ratios of well-wishers to originating passengers.

156 Guidebook for Conducting Airport User Surveys and Other Customer Research Sampling Passengers The first consideration in sampling air passengers is to decide whether to sample individual passengers or air travel parties. As discussed earlier in this chapter, this decision depends on the purpose of the survey and is influenced by the survey method adopted. Self-administered questionnaires are typically distributed to every adult passenger in the group of interest, while intercept interview surveys are usually designed to collect information on air parties, although in the case of customer satisfaction surveys, they typically attempt to obtain a sample of air passengers. There is obviously no difference between air parties and air passengers in the case of one-person parties. The objective of defining a sampling strategy to survey air passengers is to obtain a random sample of respondents. Where the flow of passengers or the sequence of passengers in a queue is already random with respect to the characteristics of interest, selecting every nth passenger will generate a random sample of respondents. When a respondent is in an air travel party of more than one person, the count to determine the next nth passenger in a sequential sampling strategy should start again if the next nth passenger is a member of the air party that has just been interviewed. This method will avoid interviewing multiple passengers from the same air party. Members of the same air party will typically be standing or seated together, and it will usually be obvious who is in the party. Where it is not obvious, it may be necessary for the interviewer to ask the respondent who has just been interviewed to indicate the members of the air party. In the case of large air parties (such as tour groups), members of the party may not remain together in the terminal, and thus the sampling strategy may happen to interview several members of the same party. Large air parties call for particular consideration in analyzing survey results (as discussed in Section 10.3.1) to account for the possibility that the survey may inter- view several members of the same party or that different members of the party may have traveled to the airport separately. With large air parties (more than about six people), it may be helpful to record the name of the group organizing the trip in order to identify other respondents from the same group when analyzing the results. Where passengers are moving, such as at an exit from security screening or an entrance to the terminal building, the duration of each interview will generally be sufficiently long that by the time the interview is completed, the next passenger in the stream will be effectively chosen at random. Also, the variation in the duration of each interview will help ensure the randomness of the sample. Where passengers are grouped by some characteristic, such as passengers in a check-in queue at an airline counter, the sampling strategy should attempt to distribute the interviews across the different groups in proportion to the passengers in each group, such as the number checking in on each airline. However, this is difficult to achieve in practice, particularly because those passengers in short queues may not be in the queue long enough to complete an interview. For these reasons, queues are generally not a good location to perform intercept interview surveys. In locations where passengers are not in an ordered sequence, such as in airline gate lounges or baggage claim areas, it will be necessary to define a starting point for the sequence of passengers in the sample and a rule for the direction to move to determine the next passenger to interview. One such rule for airline gate lounges was suggested in Section 10.3.2. A similar rule for baggage claim areas might be to start with the person standing closest to the airline baggage office door, then select every sixth person, proceeding clockwise around the baggage claim device. The exact starting point is not particularly important, and in ambiguous cases, interviewers can use their judgment as to which person is closest to a defined point. The point is to ensure that the interviewers follow a defined rule to select the passengers to interview rather than selecting them on the basis of whether they appear likely to be cooperative or some other criterion such as age or gender that might bias the sample. Similarly, the interval between sampled passengers is

Air Passenger Surveys 157   not particularly critical. The objective is to ensure that interviews take place throughout the area in question, and so the interval should be chosen in light of the number of interviews anticipated to be performed and the number of passengers expected to be present in the area. In the case of baggage claim areas, allowance should be made for greeters waiting with the passengers. When there is more than one interviewer performing interviews in a given area, the sampling rule will need to ensure that they each start at a different location and that the sequences of selected passengers do not overlap. Because the number of interviews that can be performed while the passengers are present in an airline gate lounge or baggage claim area is fairly limited, overlap is not usually a problem. For example, if each interview takes an average of 4 minutes between starting each successive interview, and the passengers in a gate lounge are surveyed over a 40-minute period, each interviewer will only be able to complete 10 interviews. One issue that arises with airline gate lounge surveys is that the seats may all be taken before boarding of the flight begins, and thus passengers arriving after this point will either stand in the gate area or find a seat in nearby lounges. A sampling protocol that includes only passengers seated in the gate lounge will systematically exclude passengers arriving close to flight depar- ture time. It is therefore desirable that the sampling rule include standing passengers in addition to seated passengers, as discussed in Section 10.3.2. In practice, where passengers sit (or stand) in a gate lounge or stand in a baggage claim area will be influenced by characteristics such as the time they reach the gate lounge or the order in which they arrive at the baggage claim device (which is influenced by the order in which they deplane and their walking speed through the terminal). However, as long as the sampling protocol includes all occupants of the gate lounge or claim area and samples people in propor- tion to the number occupying each part of the area, these differences will be reflected in the resulting sample. Sampling Flights In the case of airline gate lounge surveys (whether interview or self-administered), it is neces- sary to select a sample of flights to survey. This selection is a particularly challenging problem because of the need to ensure appropriate coverage of airlines, destinations, and times of day, while utilizing the survey personnel efficiently. Flights depart at irregular intervals throughout the day. There will be periods when a large number of flights will depart around the same time and periods with relatively few (or even no) flight departures. To complicate matters further, different airlines serving the same market will often have their flights depart around the same time, partly for competitive reasons and partly due to time-zone differences. Although the characteristics of passengers on different airlines in the same market may not necessarily be significantly different, this is not always the case. Factors that can cause passenger characteristics to vary by airline include the following: • Nationality of airline (flag of carrier) in international markets • Type of carrier: ultra-low cost, low cost, or network • Market share and flight frequency • If the destination is a hub for one of the airlines • Flight departure times Business travelers will tend to favor airlines that provide flight departures that better match the business day and those that offer higher frequency in case their travel plans change, while leisure travelers may be more willing to use flights at less convenient times and with fewer alternative departure times in order to obtain a lower fare. Passengers who are residents of cities that are airline hubs are more likely to be in the frequent flier program of the hub airline, while residents of spoke cities are more likely to be in the frequent flier program of another airline, particularly if that airline has a stronger presence in the spoke city.

158 Guidebook for Conducting Airport User Surveys and Other Customer Research Therefore, a sampling plan that fully reflects these factors would group flights by the following: • Airline • Destination (possibly grouped by region) • Time of day In practice, at many airports this approach would generate a very large number of separate groups, and it would be impossible to develop a viable sampling plan by selecting a sample of flights from each combination of these factors. Instead, the usual approach is to define a sampling strategy that ensures that the selected flights provide a reasonable sample of each of these three factors. A widely used approach is to list flights throughout the period of the survey in order of scheduled flight departure time and calculate the cumulative number of departing seats for each flight. If information on average load factor by market and time of day is known at the level of individual flights, then the cumulative number of expected passengers can be calculated in place of seats. Flights are then selected by finding every nth seat in the cumulative seat count, where n is the ratio of the total number of seats to the number of flights to be sampled, considering the expected number of completed interviews per sampled flight. This process ensures that larger aircraft have a higher chance of being selected than smaller aircraft, which offsets the smaller likelihood of a passenger on a large aircraft being interviewed if approximately the same number of interviews is performed for each sampled flight. This approach is referred to in the statistics literature as probability-proportionate-to-size sampling. If more interviews are performed for flights using larger aircraft, as could occur if more interviewers are assigned to those flights or more time is available for the interviews, or if all adult passengers are sampled using self-administered questionnaires, then either the flight sampling process needs to be modified or the results weighted accordingly. Where all or a constant proportion of passengers are sampled, the procedure is simpler. After arranging all flights in order of scheduled departure time, select every mth flight, where m is the ratio of total flights to the number of flights to be surveyed. The latter is calculated by the total number of passengers to be surveyed divided by the average number of passengers to be surveyed on each flight (the average passengers per flight times the proportion of passengers on each flight to be sampled). If the survey will only be conducted for certain periods each day, the cumulative list of seats is only determined for flights scheduled to depart during the times when the survey will be performed. After identifying an initial list of flights to sample, the characteristics of those flights in terms of the previous criteria can be compared to the corresponding proportions across all flights, and any necessary adjustments made to the sample by dropping some of the flights with characteristics that are over-sampled and replacing each of them with the next flight in the list of flights with the characteristics that are under-sampled. Further adjustments may be necessary to ensure that the survey field teams have a fairly steady workload throughout their shift. A flight that has a scheduled departure time too close to that of other flights in the sample could be dropped and replaced with another flight with the same characteristics but with a scheduled departure time during a period when the number of selected flights is not enough to keep the survey field teams occupied. 10.6.4 Survey Timing At most airports, air passenger characteristics vary seasonally, and it is therefore important to determine whether information on these characteristics is required for a specific period

Air Passenger Surveys 159   (e.g., the peak month) or on an annual basis. If information on air passenger characteristics is required on an annual basis, it will be necessary to perform the survey in two or more periods reflecting the seasonal pattern of traffic. Depending on the information of interest, it would be desirable to select a peak and off-peak period that represent the highest and lowest levels of the relevant characteristics (e.g., private vehicle trips to and from the airport, or proportion of business travel). This method should allow the corresponding characteristics to be estimated for other months that are not surveyed, although this may not always be possible, depending on how the different characteristics vary during the year. The survey planning team may need to undertake some analysis of monthly variation in those characteristics in order to select appropriate months for the survey. Where seasonal variation in travel characteristics is more complex than can be expressed in terms of peak and off-peak conditions (e.g., winter travel patterns are different from summer travel patterns as well as from travel patterns in the spring and fall), it may be necessary to divide the survey into three or four phases or conduct the surveys continuously throughout the year. In addition to seasonal variation, at almost all airports, air passenger characteristics vary by day of the week and time of day, as discussed in Section 10.3.1. This variation can be addressed by ensuring that the survey provides reasonable coverage of different times of day and days of the week. Ideally, the survey data collection would take place throughout the day for at least a full week. At many airports, passengers begin arriving for early morning flights by 5 a.m., and the last flight does not depart until midnight or later; therefore, the survey would need to cover about a 19-hour day. However, the staffing levels required to achieve this coverage are often impractical, and spreading the desired sample size across every time period may unduly constrain the sam- pling in each period. Therefore, it is common to survey for a limited period each day (typically at least 8 hours) but vary the timing of these periods from day to day. Where traffic peaks occur at certain times on particular days (e.g., Monday morning and Friday and Sunday evenings), the survey shifts should be scheduled to provide coverage for these periods. While some flights may depart early in the morning or late in the evening, typically these are fairly few, and the sampling strategy should reflect the relatively small proportion of passengers at these times. It may be adequate to have only one or two survey staff during these times and only survey on a few days. However, at airports with international service, departures for certain markets may take place late at night because of time-zone differences or to allow early morning arrivals at the destination. Similarly, some eastbound flights from the West Coast depart late at night in order to arrive at mid-continent hubs or East Coast destinations early the next morning. This may require additional survey staffing at these times to make sure that those markets are adequately surveyed. At other times of day, there may be few departures for periods of several hours at a time. Such flight schedules can present a challenge to survey staffing because, in general, staff will want to work a full shift. It may be possible to schedule staff so that meal breaks occur during periods of low activity, although there are constraints on how long staff can work between breaks, so meal breaks generally need to occur around the middle of the shift, although they are also influenced by time of day. With careful scheduling, it may be possible to provide increased staffing levels during busy periods by scheduling shifts to overlap during these periods. Figures 10-6 and 10-7 illustrate how traffic levels and the composition of the traffic can vary over the days of the week and time of day in the case of one particular airport: Boston Logan International Airport. The figures show the distribution of departing seats from BOS for different markets for the week of October 20–26, 2019, and hour of the day for Thursday, October 24. Although the variation in the traffic composition by day of the week and time of day will be

160 Guidebook for Conducting Airport User Surveys and Other Customer Research Source: OAG for week of October 20–26, 2019 (Official Airline Guide n.d.). Source: OAG for October 24, 2019 (no flights to Africa) (Official Airline Guide n.d.). Figure 10-6. Number of departing seats from BOS by day of week and destination region. Figure 10-7. Number of departing seats from BOS by time of day and destination region.

Air Passenger Surveys 161   different for other airports (and indeed may be different at BOS at other times of the year), the point of the example is to show the potential variation in traffic composition that typically occurs at major airports. It can be seen from Figure 10-6 that the overall traffic level and market composition does not vary that much from Sunday through Friday, although flights to Africa only occur on three days of the week. There are significantly fewer flights on Saturday, and the market composition is also different from the rest of the week. Figure 10-7 shows that the traffic level varies widely over the day, with early morning and late afternoon/evening peaks and a sharp drop in long-haul domestic departures (stage length of 1,000 miles or more) between about noon and 2:00 p.m. (shown as hours 12 and 13). Few flights depart after 11:00 p.m. (hour 23), and most of them are to Europe. As shown in Figure 10-7, departures for different international markets vary considerably through the day, with those to Central America and the Caribbean in the morning and those to Europe mostly in the evening. To adequately represent these markets in the sample, the sampling plan would need to ensure that the flights selected to be surveyed at those times include an appropriate number of inter- national flights. One approach to ensuring that the survey responses are appropriately distributed by time of day is to schedule interviewers so that the distribution of total interviewer time by time of day and day of week is similar to that of the departing seat capacity of the airport. For smaller total sample sizes (e.g., under 1,000 responses), interviewers are assigned to cover a larger number of flights by sampling only a small number of passengers from each flight (e.g., four to twelve, depending on the size of the aircraft). Because connecting passengers tend to arrive in airline gate lounges earlier than originating passengers, starting to survey flights too long before flight departure time will tend to over-sample connecting passengers. The interviews for a given flight should generally start no earlier than an hour before flight departure (90 minutes in the case of international flights) and be evenly distributed up to the time that boarding begins. This approach can work well at small to medium-sized airports. At larger airports, interviewer times would be determined for each terminal in proportion to the numbers of passengers or departing seats from those terminals. Responses are then weighted to match departing passengers by time of day, flight destination, and airline using flight schedules and estimated load factors. As a practical matter, it is difficult to schedule interviewers efficiently and match the inter- viewer time to the variation in departing seats on a daily basis. Instead, an attempt can be made to match interviewer time to departing seats in each hour separately for weekdays in total and weekend days in total. The start times and lengths of shifts and break times can be varied so that there is more coverage in the peak periods, with interviewers typically working between 4 and 8 hours per shift. Scheduling interviewers to match the distribution of departing seats (or passengers) has staffing and cost implications. Having interviewers work for varying times per day and starting their shifts at different times can be done, but it may be necessary to pay more for people to work short shifts on some days or the earliest and latest shifts. Field supervision also becomes significantly more complicated with interviewers starting their shifts at different times. These factors need to be considered carefully in survey planning. 10.7 Questionnaire Wording and Length The general principles of questionnaire design and length are discussed in Section 5.4, and sample passenger questionnaires are provided in Appendix E. The sequence and wording of survey questions can significantly affect the reliability of the responses that are obtained if

162 Guidebook for Conducting Airport User Surveys and Other Customer Research respondents misunderstand a question being asked. As a general rule, industry terminology and jargon (such as “queue”) should be avoided in questions since they may not be understood by respondents. Examples of problems that can arise include the following: • Terminology. For ground access modes, particularly modes such as scheduled airport bus, limousine, hotel courtesy shuttle, and charter van. In some areas, the term “limousine” is used for scheduled airport bus service and may appear in the name of the service, while a vehicle hired for the exclusive use of an air party may be referred to as a “black car.” The term “shuttle” is often used to refer to shared-ride van service (and again sometimes appears in the name of the operator). An air party taking a shared-ride van service from a hotel to the airport may consider this a “hotel courtesy shuttle,” although that is not what is intended by the term. The distinction between shared-ride van service and charter van may be a matter of whether the operator is licensed to carry multiple travel parties in a single trip, which may not be known to the travelers, particularly if the trip to the airport was arranged by someone else. • Trip purpose. Questions should generally provide more options than just business and personal. The term “leisure” would usually be considered to exclude a wide range of personal trip purposes (e.g., attending a funeral) and should be avoided. Response options should allow for trips that combine business and personal purposes, such as combining a business meeting or conference with vacation time or leisure activities or visiting family or friends. Asking the “main purpose” or “primary purpose” of the trip may be meaningless in cases where the trip only occurred because it allowed multiple purposes to be satisfied. The key issue to consider is why information on trip purpose is needed. If trip purpose is sought to distinguish between trips where the respondent is paying the travel costs and those where the costs are paid by their employer or other organization, it may be better to ask this question directly rather than assume that respondents reporting business trips are not paying for their travel costs themselves. • Trip origin. Typically, the response sought is the origin of the ground access trip to the airport where the survey is being performed. However, if not carefully worded, this question could be misunderstood by visitors as the origin of their entire trip from their home region. The expression “your trip to the airport today” is ambiguous to travelers making a 1-day return trip. In some cases, even the term “this airport” can be ambiguous. In a recent survey, respon- dents were asked to state the final destination airport of the air trip that they were about to begin. This was followed by a question that asked: “Is this airport the home end of your trip?” The second question was intended to mean the airport where the survey was being performed but could easily have been misunderstood to mean the destination airport referred to in the previous question. • Air party. This term may not be understood by many respondents and needs to be expressed in other words (e.g., traveling together on the same flight). For ground transportation planning purposes, it may be desirable to distinguish between the air travel party and the ground access travel party because different members of the air travel party may have come to the airport separately. 10.7.1 Trip Origin Information Sponsors of air passenger surveys often wish to obtain information on the ground access trip origin location (or ground egress trip destination location). This information is typically needed at a fairly detailed level to permit the trip origins or destinations to be coded to the system of transportation analysis zones (defined geographical areas for transportation modeling) used by the local regional transportation planning agency. This coding will allow information on highway and transit travel times and transit fares to be readily obtained from the data files maintained for regional transportation modeling. It will also allow the results of the air passenger

Air Passenger Surveys 163   survey to be integrated with other transportation planning studies. It is not uncommon for these zones to be significantly smaller than zip codes or postal codes. The usual approach is to request the street address and city of the trip origin. For obvious reasons, many survey respondents are reluctant to provide the actual address, although they may be willing to provide the block number or a nearby street intersection, which is sufficiently accurate. However, visitors to the area may have started or ended their access or egress trip at a hotel, business, or other discrete location (such as cruise ship terminal or convention center) for which they do not know the address. In such situations, respondents should be asked to provide the name of the hotel or other location, and these will have to be coded later so that the correct address can be assigned to the survey response. In such cases, to resolve situations in which there are several locations with similar or identical names, the trip origin or destination city should also be obtained. Hotel names can be particularly problematic. In a large city, there may be several hotels in the same chain, and while they will typically each have a unique name, respondents may not use the formal names but simply refer to them by the name of the chain. A related problem can arise when ownership of a hotel has recently changed. Respondents who previously stayed in the hotel before the change may refer to it by its former name. Because respondents—visitors in particular—may give partial or even incorrect names for hotels and other locations, it will be helpful to also obtain a nearby street intersection or the name of the street if this is known. While this may be redundant information in many cases, it can be invaluable in resolving ambiguous or unclear responses. In the case of printed questionnaires, it will be necessary to obtain the redundant information for trip origin or destination locations from all respondents because it would be too complicated to explain which respondents should provide it and which not. However, in the case of surveys using EDCDs, it may be possible to program the devices so that the street name or intersection question is skipped for responses that give location names that are clearly unambiguous. The issue of the ground egress trip destination is often ignored, and it is implicitly assumed that the ground access trip and egress trip are symmetrical. However, this is not always the case, and this may deserve explicit attention in the survey. In particular, situations where they are not the same may be an important factor in access and egress mode choice. Common examples are visitors who travel from the airport to a hotel on arrival in the area but return to the airport from another location, such as a business they are visiting, and residents who travel to the airport from their workplace but return home from their return flight. Collecting trip egress information requires some thought in questionnaire wording because the egress trip has not yet occurred for residents who are surveyed on the outbound leg of their travel. 10.7.2 Public Transportation Modes Another aspect that requires careful attention is the treatment of public transportation modes in questions about access and egress trips. The use of public transportation is often an important policy issue in situations where airports are trying to reduce or mitigate ground transporta- tion vehicle trips and may be a consideration in airport user satisfaction with airport ground- transportation services. It can also be a key consideration if one objective of the survey is to support the development of models of airport ground access mode choice. There are two aspects to the use of public transportation modes that may need to be con- sidered in the design of air passenger survey questions. The first is the appropriate definition of the different public transportation services available at the airport. Because different services will have different service areas and may have significantly different levels of service—such as frequencies, fares, and hours of service—it will often be necessary to distinguish between the services and not simply classify all such trips into broad categories. Self-administered

164 Guidebook for Conducting Airport User Surveys and Other Customer Research questionnaires should avoid the use of terms such as “public transit,” which can mean different things to different respondents. The design of the questionnaire (and, if EDCDs are used, the EDCD program) should provide the ability to distinguish between different modes and services as necessary. In the case of surveys at groundside access points (discussed in Section 10.12), it may be necessary to develop unique questionnaires or forms for different modes. Because of the potential for misreporting the actual public transportation service used, it is advisable to specify (or ask for) the name of the agency or firm operating the service and, in the case of local bus services, the number of the route. For other fixed-route services, it may be useful to request the name of the station or stop where the passenger boarded the service. While this level of detail may not be necessary for analyzing the results of the survey, it can be invaluable during data cleaning to correct mis- reported or misclassified services. In the case of fixed-route services, it may also be helpful to ask how the respondents got to the station or stop where they boarded the service because this can have a significant impact on the time and cost involved in using the service (e.g., a taxi trip to a transit station could easily cost more than the transit trip itself). While obtaining such information will increase the number of questions to be asked, these generally affect only a small proportion of respondents and so do not have a significant impact on the cost of the survey or the time required to complete it. The second aspect that the survey planning team may wish to explore is the familiarity of the survey respondents with the public transpor- tation system serving the airport. Do air passengers use public transit on a routine basis for other types of trips? This other usage will affect the familiarity of air passengers with the local transit system and may affect whether public transit is even considered as an option for getting to the airport. In the case of other forms of public transportation, such as privately operated scheduled airport bus services, air passengers (particularly visitors to the area) may not know anything about the services. 10.7.3 Parking Issues Private vehicles are the most widely used means of traveling to or from most airports, and parking revenues make up a major component of airport revenues. Therefore, information on the use of airport parking by air passengers—and those dropping them off or picking them up— is an important aspect of most air passenger surveys. However, survey questions addressing the use of airport or off-airport parking should be carefully worded to prevent misunderstanding by respondents. Careful wording is particularly necessary with self-administered surveys. The most important distinction to make is between vehicles parked by air passengers for the duration of their air trip and those parked by well-wishers or greeters. This distinction is complicated by the frequent practice of airports calling different parking lots or facilities short- term or long-term parking based on pricing and distance to the terminal rather than on the amount of time drivers are allowed to park. Long-term parking facilities may be located some distance from the terminal, and users may think of these as being off-airport, even though they are operated by the airport authority. Then there are privately operated parking lots in the vicinity, which are usually considered off-airport parking. In addition, some airport area hotels may offer parking at competitive rates to those of the airport. Although public transportation is often referred to as public transit, at many airports, public transportation includes a wide range of services, many operated by the private sector, in addition to the services provided by local transit agencies. Two reports produced under TCRP provide a good overview of the issues involved in public transportation access to airports: TCRP Report 62: Improving Public Transportation Access to Large Airports (Leigh Fisher Associates, M. A. Coogan, and MarketSense 2000) and TCRP Report 83: Strategies for Improving Public Transportation Access to Large Airports (Leigh Fisher Associates, M. A. Coogan, and MarketSense 2002). Furthermore, transit services may be provided by several different agencies, such as in cases where rail transit service is provided by a different agency from local bus services or where transit agencies that serve different areas each serve the airport.

Air Passenger Surveys 165   Although most passengers parking a vehicle for the duration of their air trip will use the parking facilities designated for daily or longer parking rather than those for hourly or short- term parking, some passengers making a 1-day or overnight trip may choose to use the closest parking to the terminal and pay the higher rate. Finally, some survey respondents may think of a private vehicle standing at the terminal curb for a few minutes while the passengers and their baggage are unloaded as being “parked” for a short while. It is therefore desirable to ask survey respondents to identify the parking facility where the vehicle is parked and the duration that it was (or will be) parked, rather than rely on vague categories such as “parked short term.” These questions should provide response options that use the formal designation of different parking facilities (e.g., hourly parking, economy lot, terminal garage), but allow respondents to write in or state other locations if they do not recognize the correct names of the facilities. This will often be necessary anyway in the case of off-airport parking, where there may be a large number of different providers. 10.8 Weighting Survey Responses As discussed in Section 5.3.4, there are two quite different reasons for weighting survey responses: • Correcting for any bias in the survey results arising from the sampling approach used in the survey • Expanding the survey results so that they are expressed in terms of some measure of airport activity, such as average daily air passengers In spite of the survey team’s best efforts to design the sampling plan to obtain a representative sample of air passenger trips, because of unavoidable consequences of the survey methodology as well as varying response rates by different categories of travelers, it is unlikely that the responses will fully reflect the composition of the target population. Therefore, it will usually be necessary to weight the survey responses in order to improve the accuracy of the resulting data. As noted in Section 5.3.4, in some cases, the survey responses may be close enough to a random sample, or the respondent characteristics may not vary significantly across the aspects where the sample differs from the air passenger population, so weighted survey results will not offer a significant improvement in accuracy over the use of unweighted response data. In order to determine whether either of these situations is the case, some analysis of the survey data and comparison to estimated or observed control totals for the air passenger population will generally be required. Because the exact composition of the air passenger population is generally unknown (this is one reason why the survey is being performed), a variety of other types of data are needed to calculate the survey response weights. These data should be assembled at the time the survey is performed—or as soon as possible afterward—and should include the following: • Enplaned passengers on each flight (if available) or enplaned passengers by flight destination for the month in question. These data are available from the U.S. Department of Transpor- tation Bureau of Transportation Statistics T-100 database. Some airports collect these data routinely. Where this is not the case, the airport may be able to obtain this information from the airlines for the period of the survey, with an assurance that it will not be made public and will only be used to help analyze the survey results. • Number of connecting passengers on each flight (if available) or estimate from available data. Quarterly data on connecting passengers by airline and flight sector can be estimated from the

166 Guidebook for Conducting Airport User Surveys and Other Customer Research U.S. Department of Transportation Bureau of Transportation Statistics Airline Origin and Destination Survey database at https://transtats.bts.gov/ (see bibliography). • Number or cars exiting from each airport parking facility by parking duration and hour for each day of the survey period. • Automated vehicle information system counts (where available) by class of vehicle (taxis, limousines, shared-ride vans, shuttle buses, etc.) and by hour for each day of the survey period. • Pick-up and drop-off trips by TNCs by hour for each day of the survey period. Airports increasingly have established a geofence around the airport and obtain reports of each trip by vehicles that cross the geofence. • Ridership statistics on airport-operated shuttle buses to remote rental car facilities or rail stations by hour for each day of the survey period. Where these are not routinely recorded, it will be necessary to arrange for their collection for the survey period. • Ridership statistics on scheduled airport bus services by run (or hour) for each day of the survey period. Where these are not routinely reported to the airport authority, the airport may be able to obtain them from the operators with an assurance that they will not be made public and will only be used to help analyze the survey results. • Terminal roadway traffic counts (where available) by hour for each day of the survey period. Where these data are not routinely collected, consideration should be given to placing traffic counters on the terminal roadways for the duration of the survey. The process of calculating survey response weights consists of two steps: 1. Calculation of weights to correct for known bias in the survey sampling methodology 2. Calculation of weights to correct for differences between the survey results and external data on traffic composition Each survey response should include the size of the air party. If the survey responses reflect air passengers (i.e., there are multiple survey responses for parties with more than one passenger), then counts obtained from the survey responses should be divided by the air party size in order to express the traffic composition in terms of air parties. Conversely, if the survey responses reflect air parties (i.e., there is only one survey response for each air travel party), then counts obtained from the survey responses should be multiplied by the air party size in order to express the traffic composition in terms of air passengers. Statistical computer software packages can perform these adjustments easily when tabulating survey results. The difference between expressing survey results in terms of air passengers or air parties is critically important to the correct interpretation of the survey results and should be clearly understood. 10.8.1 Proportional Weighting Proportional weighting uses weights that adjust the proportions of the survey response data to reflect the proportions of the control data without changing the total number of responses. In general, this will result in non-integer counts for many reported survey responses when expressed using weighted data. Proportional weights can only adjust survey response data to correspond to the proportions of a single characteristic of the control data. Separate weights can be determined for different characteristics, but in general, it is not possible to determine response weights that adjust survey response data to correspond to the proportions of multiple characteristics of the control data.

Air Passenger Surveys 167   If there are N total survey responses, and ni of those responses reported some characteristic i that composes a proportion pi of the population in the control data, then the proportional weight wi that should be assigned to each of the n responses is given by: =w p N n i i i Since all N of the survey responses must have reported some value for characteristic i (even if this was only “don’t know” or “refused”), a weight wi will be assigned to each survey response. Statistical analysis software can generally be set to optionally exclude missing data cases (such as “don’t know”) from the tabulated results. However, some users of the results may be interested in knowing the extent of missing data in the survey responses. Therefore, it is better to set the weight for missing data responses to 1 rather than 0 and adjust the other weights so that the weighted total of the non-missing cases corresponds to the unweighted total of the non-missing cases (i.e., replace N by N minus the number of missing data responses in the equation). 10.8.2 Correcting for Known Bias in the Sampling Methodology The sampling methodology adopted for the survey may introduce some bias into the response data that can be calculated and corrected. The most obvious example occurs with self-administered questionnaires handed out to all adult passengers in an airline gate lounge where fewer responses are received from a given air party than the number of passengers in the party. This case will always occur where there are children in the air party (who do not complete the survey). Some passenger surveys have asked the respondents to indicate how many members of their air party have completed a survey questionnaire and have then used this information to weight the results. However, experience indicates that these statements are often unreliable. Some respondents may misunderstand the meaning of the term used to describe the air party, while others may not realize that another member of their party is also completing a questionnaire. Or they may think that another member of their air party is completing the questionnaire, but in fact that questionnaire is not turned in. It is therefore preferable (although more time consuming) to examine the survey responses; identify responses from the same party based on the party characteristics, such as their trip origin address or other information; and revise the reported survey completion information before calculating weights to correct for underreporting of air party members. It is also quite common to apply weights to self-administered survey responses so that the responses sum to the number of passengers boarding the flight. There are two problems with this approach: • It can give a misleading impression of the number of survey responses (as discussed in Section 10.8.4) unless the resulting weights are adjusted to ensure that the total of the weighted responses across all the flights equals the actual total number of survey responses. • It will over-weight responses from under-sampled flights. For example, if generally 50% of passengers on sampled flights are surveyed, but on a particular flight only 10% of passengers are surveyed for some reason, the responses from passengers on the under-sampled flight will be weighted by a factor of 10 rather than the factor of 2 used on other flights. Differences between the distributions of characteristics on a particular flight with only a few respondents compared to other flights in the same market are most likely due to the higher variability that occurs with small samples (as discussed in Chapter 4) not because the characteristics of all the passengers on that flight are different. Weighting the responses

168 Guidebook for Conducting Airport User Surveys and Other Customer Research to correspond to the total number of passengers on the flight implicitly assumes that all the passengers on the flight have the same distribution of characteristics as the respondents. For example, if only four respondents are surveyed on a particular flight, and one of these is leaving on a 10-week trip to Japan, it would be incorrect to infer from this that 25% of the passengers on the flight are leaving on 10-week trips to Japan, but that would be the effect of weighting the responses in this way. There is of course no way to tell from the results of a survey whether differences in the characteristics of the respondents on different flights are due to a true difference or are simply a result of the sampling variance. It is possible to perform statistical tests to determine whether the hypothesis that the results are drawn from the same distribution can be rejected at some level of confidence, but that is not the same thing as knowing that they are different. With a small sample size, the variance in any particular characteristic is likely to be so high that it is unlikely to be possible to reject the hypothesis that the results are from the same distribution as that for other flights in the market at any reasonable level of confidence. Therefore, one is left to make the not unreasonable assumption that the actual distributions of the characteristics of passengers in a particular market are the same, and that differences in the distributions observed in the survey responses across flights in that market are due to sampling variance. Of course, the distribution of a particular characteristic within each market may vary by other dimensions, such as the time of day or day of the week, further complicating the analysis and reducing the ability to determine whether any apparent differences across flights in the market are simply due to chance. Because the characteristics of passengers on under-sampled flights are likely to be less representative of the characteristics of passengers on other flights in that market, weighting responses to the total number of passengers on a flight will assign extra weight to those passenger responses that are less representative of the characteristics of the market in question, potentially biasing the results of the survey. Therefore, it is better to consider those passengers in a given market who did complete the survey questionnaires as a representative sample of air passen- gers in that market and make any required adjustment to correct for differences between the survey results and the distribution of air traffic across different markets (as discussed in the next subsection). In the case of interview surveys, there is likely to be sampling bias that results from the sampling protocol (as discussed in Section 10.6.3). If the distributions of characteristics of respondents who are over-sampled are the same as those who are under-sampled, the difference in sampling rate will not affect the survey results. However, if the distribution of some characteristic is different, the results will be biased. For example, it is likely that some passenger characteristics, such as trip purpose or air party size, will differ between those arriving in an airline gate lounge well before boarding begins and those arriving shortly before boarding begins. Similarly, if interviews of passengers exiting security screening are performed at approximately the same rate—as is likely with a survey team of a constant size—the result will be a lower sampling rate during busy periods. If passenger char- acteristics are different between busy periods and slow periods (as is quite likely), the results will be biased. Weights can be calculated to adjust for these sources of bias by examining the results for different periods or subgroups of respondents to see if there are any differences in the distri- bution of characteristics that might vary by period or subgroup. If such differences are found, the survey responses can be weighted by the ratio of the number of air parties or air passengers in each period or subgroup to the number of responses obtained for that period or subgroup. Such weights should be adjusted so that the total number of weighted responses is the same as the number of actual responses.

Air Passenger Surveys 169   10.8.3 Correcting for Differences Between the Survey Results and External Data Once a set of weights has been determined to correct for known bias in the sampling meth- odology, an additional set of weighting factors can be calculated, using the weighted results to correct for differences between the weighted results and external data on the composition of the passenger traffic using the airport. The most obvious potential difference between the survey results and external data on the composition of the passenger traffic at the airport is if the percentage of passengers in each flight destination market given by the survey responses does not agree with the passenger traffic reported by the airlines. Because connecting passengers may have been sampled at a different rate from that of originating passengers, it will generally be advisable to consider connecting passengers boarding a flight as a separate market from originating passengers and calculate separate weights for each. Other characteristics of the survey respondents for which it may be worth calculating weighting factors include the following: • Airline • Time of day and day of week of flight • Ground access mode use by originating passengers Where several different weighting factors have been calculated for different survey response characteristics, in order to determine the sensitivity of the survey results to the choice of weighting factor, it will generally be advisable to compare the weighted survey results for each characteristic using the appropriate weighting factor with the corresponding results using each of the other weighting factors. 10.8.4 Weighting for Total Traffic It is common for survey responses to be assigned weights that convert the total number of survey responses to the corresponding count of annual passenger traffic or average daily traffic (effectively the same thing). While this process allows the survey results to be directly expressed in terms of the corresponding annual passenger characteristics, two important caveats should be borne in mind before doing this: • The characteristics of the air passenger market at a given airport will vary throughout the year, while the survey data will generally have been obtained at one or two discrete points in time. Thus, the resulting data may be quite misleading. For example, if a survey is performed during August, it will reflect a high proportion of vacation travel. This result is unlikely to correspond to the characteristics of the air passenger population during the rest of the year. • Expressing the results of the survey in terms of annual passengers conceals the true size of the survey sample and may give a completely false impression of the accuracy of the results. For example, for a survey with 1,200 responses at an airport handling 12 million annual passengers, each survey response is equivalent to 10,000 annual passengers. Thus, if the estimated number of annual passengers with some characteristic was given as 22,400 (after weighting for other considerations), it might easily be overlooked that this represents only two survey responses and is likely to be highly inaccurate. This is not a criticism of weighting itself, but rather of using it in this way. Therefore, it is recommended that survey results not be expressed as annual traffic, but rather that weights be calculated so that the resulting totals of weighted responses equal the size of the actual survey sample. It is easy enough for users of the survey results to express the results in

170 Guidebook for Conducting Airport User Surveys and Other Customer Research terms of the level of annual traffic if they so desire, but they will then be fully aware that they have done this and should recognize the accuracy limitations that this implies. One advantage of expressing the survey results in terms of the sample size is that it allows users to easily distinguish whether the results reflect the distribution of air passenger character- istics or air party characteristics because the response totals will be quite different in each case. For example, if the average number of air passengers per air party is 1.4, a survey with 5,000 air party responses will show results summing to 5,000 when showing the distribution of air party characteristics and 7,000 when showing the distribution of air passenger characteristics. If the results are weighted so that the total is equal to the annual passenger traffic, it may be unclear whether they are showing the distribution of passenger characteristics or air party characteristics (which will typically be different). (If the survey results will be expressed in terms of annual traffic, it is important that results showing air party characteristics be weighted to give the total number of annual air parties, not air passengers.) 10.9 Measures to Obtain Adequate Response Clearly, getting people to respond to a survey is an important component of project success. If there are insufficient respondents, it is questionable whether the results can be generalized to represent the population of interest. Airline passengers are almost by definition in a hurry and stressed. If they are leaving home, they have all the emotions associated with that. If they are returning home, they may be rested and exhilarated from a wonderful vacation or exhausted from a difficult business trip. Regardless, departing passengers have to stand in what are often long lines, deal with security, remember the latest rules or instructions, worry about getting to the gate on time, and actually find the gate. Even so, departing passengers are generally inclined to participate in surveys despite the hurry and the stress (at least in part because they may not have anything better to do while standing in line or waiting for their boarding call), but there are still several things the survey planning team can do to maximize the response: • Limit the length of the survey to the number of questions necessary to obtain needed information. • Make sure that potential respondents understand that the survey is sponsored or sanctioned by the airport, both by the way the interviewers are dressed (perhaps in identifying clothing) and by the content of the introduction. • Emphasize the survey purpose in the introduction and explain why the information is needed. • Ensure that the questions are clear, comprehensible, and sensitive to concerns about personal information and confidentiality. • Hire interviewers (or people to hand out questionnaires) who are intelligent, personable, and not afraid to approach a wide variety of strangers. • Pay interviewers enough to attract capable people and to ensure that they stay for the duration of the project. Aside from the difficulty of replacing interviewers, the longer interviewers stay, the more competent they tend to become. • Provide a thorough and comprehensive training session. • Ensure that interviewers are supervised, monitored, and coached as needed. • Provide retraining as needed. • Use positive feedback and incentives to maximize interviewer retention. • Establish and enforce appropriate standards of dress, grooming, and conduct. • Make sure the interviewers smile as they approach people. • Make sure all respondents are sincerely thanked.

Air Passenger Surveys 171   One other issue that is often raised in this context is the use of incentives for survey respon- dents. Those who have tried respondent incentives are inclined to think they are not worth the cost or the challenge of hauling them around and accounting for them. As noted previously, most air passengers are inclined to participate in airport-sponsored surveys anyway; all airports really need to do to ensure high response rates is make it pleasant for them to do so. The trade-offs between keeping the survey short and gathering information needed to better understand issues of concern or provide information needed by a range of stakeholders within airport management require careful thought. If the information to be obtained in a survey is defined too narrowly in the interest of limiting the number of questions, opportunities to expand the value of the information generated by the survey may be lost. 10.10 Location-Specific Guidelines Every airport is different. This section discusses some of these differences and how they can affect air passenger surveys. 10.10.1 Multi-Airport Cities A metropolitan area served by a number of airports presents a problem in determining the characteristics of air passengers with respect to the entire metropolitan area. Analysis at this level requires a coordinated survey approach at all airports. This coordination does not necessarily require simultaneous surveys but does require that the survey method and questionnaire are common at each airport; otherwise, the results will not be comparable or applicable to the total population. In general, to minimize differences due to seasonal effects, it would be desirable for the different surveys to be performed within a few weeks of one another. A common approach is to conduct surveys at each airport over a period of several weeks, surveying at just one of the airports on any given day and scheduling the days at each airport to provide survey coverage of that airport on each day of the week at some point during the survey period. If comparisons are required, stratified sampling could be used to ensure a representative sample across airports and sample sizes chosen so that similar levels of accuracy are obtained for each airport. Such surveys are commonly undertaken by regional agencies in cooperation with the airports involved, although in principle they could be undertaken as a joint effort by the airports in a region. In regions where some or all of the airports are operated by the same airport authority, this is obviously easier to arrange. 10.10.2 Multi-Terminal Airports Because different terminals in a multi-terminal airport typically serve different airlines, and often different types of traffic (e.g., domestic or international), a survey that is designed to capture the characteristics of the air passenger population at the airport will need to survey passengers in every terminal. In this respect, a multi-terminal airport is no different from a multi-airport city. The survey design must account for multiple terminals, with consideration given to an appropriate distribution of the survey responses among the terminals over the course of the survey period. The sampling plan must also take into consideration the multi-terminal environment. Interviewers will require time to switch terminals, which could involve considerable walking or riding inter-terminal transportation. Switching terminals may require leaving the secure area and re-entering it, which takes time. These factors must be taken into consideration in design- ing the survey.

172 Guidebook for Conducting Airport User Surveys and Other Customer Research 10.10.3 Local Terminology In designing survey questions, it should be recognized that words and phrases may have different meanings, or subtle variations of meaning, in different parts of the country. While it is possible to write a questionnaire using local terminology, many of the passengers will be visitors from outside the region and may misinterpret the questions. It is therefore critical that questions be worded using clear terminology likely to be understood by all passengers, with additional explanations given if necessary. With intercept interviews, interviewers must be trained so that they fully understand the questions and, if necessary, can clarify a question using terminology that the respondent understands. 10.11 Information on Greeters and Well-Wishers A survey sponsor may wish to gather information on greeters and well-wishers as part of an air passenger survey. Greeters and well-wishers account for a significant proportion of all airport access and egress trips, and some use airport facilities. The distinction needs to be made between greeters and well-wishers who come into the airport terminal building to meet arriving passengers or see departing passengers off on their trip, and those who come to the airport only to drop off or pick up passengers but do not come into the terminal or do so only briefly and do not use any airport services or concessions. These trips are sometimes referred to by airport planners as “serve passenger” trips. Although greeters and well-wishers cannot access the secure part of the terminal and thus will not be intercepted by surveys in that area, departing passengers will be able to provide some information on the number of well-wishers who accompanied them to the airport. However, they may not know how long the well-wishers remained at the airport after the passengers went through security or whether the well-wishers made any use of airport services or concessions before leaving. Similarly, departing passengers who are visitors to the area may be able to recall information about greeters who met them on their arrival, but they may not know how long those greeters spent at the airport or whether they used any airport services or concessions while waiting. Departing passengers who are residents of the area are less likely to be able to provide infor- mation about greeters who will meet them on their return because this has yet to take place. However, they may be able to recall information about any greeters who met them on their previous return trip to the airport. Therefore, it may be desirable to survey greeters and well-wishers themselves in nonsecure parts of the airport. This is especially true if it would be useful to have information on their use and satisfaction with the facilities, services, and concessions, or to obtain their suggestions. Because it will not always be obvious who are passengers and who are greeters or well-wishers without asking them (which of course is easily done), information on greeters and well-wishers will generally be obtained as part of a survey that includes passengers as well but using separate questionnaires or separate questions on a common questionnaire. These surveys are typically performed in the terminal lobby or at the various groundside locations. The problems with sampling bias are more acute with greeter and well-wisher surveys than with passenger surveys. Well-wishers will often not want to take the time to participate in a survey during the short period they have with the passengers, and they will typically not remain in the terminal long after the passengers they are seeing off proceed though security, which reduces the opportunity to survey them. Greeters are usually easier to interview as they wait for the passenger to arrive, but the sample will be biased toward greeters who spend a longer time at the airport, especially for delayed flights. This can result in biased estimates of characteristics such as the time spent in the terminal, the time their vehicle is parked, and the amount spent at

Air Passenger Surveys 173   the concessions. Therefore, it is important to record such information as the time the greeters arrived in the terminal, the scheduled arrival time of the flight taken by the passengers they were meeting, and the time that the greeters were interviewed. Care must be taken to ensure that greeters and well-wishers are not double-counted when analyzing the results of surveys of air passengers in which multiple responses may be received from a given air travel party. Greeters and well-wishers will usually be at the airport to see off all members of the air travel party, which must be taken into account when tallying the numbers of greeters and well-wishers per passenger. There will often be more than one greeter or well-wisher meeting or seeing off an air travel party, but only one survey response will be obtained from each group. Thus, the number of greeters or well-wishers in the group should be collected, as well as the size of the air travel party they are meeting or seeing off. Care should also be taken in expressing the results to account for air travel parties that do not have greeters or well-wishers. In the case of well-wishers, this information can be obtained from air passenger surveys. However, estimating the number of arriving air parties that are not met is more difficult. The total number of greeters and well-wishers can be estimated by comparing the reported use of short-term parking by greeters and well-wishers with statistics on vehicle exits from short-term parking by duration. Those parking for less than 3 hours are mostly greeters or well-wishers since air passengers and airport employees would park for much longer, if indeed airport employees would use the short-term parking facilities at all. Knowing from surveys the percentage of greeters and well-wishers who use airport parking allows the number of greeters and well-wishers to be estimated. 10.12 Groundside Surveys Groundside surveys are a special type of air passenger survey that can be used to obtain detailed information for planning airport groundside facilities. Groundside surveys are characterized by two key factors: first, the survey sample is typically not a controlled random sample of the target population, and second, the interview process is directed at the occupants of a vehicle rather than individual travelers. In the absence of a structured sample plan from which to derive weight factors for each interview, it is necessary to obtain ancillary data in order to calculate appropriate weights for the survey responses. 10.12.1 Purpose Groundside surveys are used to gather information on vehicle use patterns by air passengers and associated greeters and well-wishers to plan future groundside facilities or to create and calibrate a ground transportation model that will be used in future planning studies. They are also used to gather information on airport user satisfaction with groundside facilities and airport ground-transportation services. The vehicle trips associated with the passengers on a single flight place loads in time and space on the groundside facilities. The sum of these loads for all flights in a planning period gives the varying load on all groundside facilities, which determines the resulting level of service provided by those facilities. Groundside surveys are designed to collect complete information on the vehicle trips serving arriving and departing air passengers, including the characteristics of the well-wishers and greeters that accompany those passengers. This information enables the development of four time curves associated with air passenger departure and arrival activity that can be used to estimate flows of vehicles, passengers, well- wishers, and greeters generated by each flight arrival and departure:

174 Guidebook for Conducting Airport User Surveys and Other Customer Research • The time before flight departure that air passengers and any accompanying well-wishers arrive at the airport • The time well-wishers leave the airport, either before or after the flight departure • The time that greeters arrive at the airport with respect to the arrival time of the flight they are meeting • The time that air passengers and any accompanying greeters leave the airport after the flight arrival 10.12.2 Ancillary Data To handle the sampling rates of this type of survey during the analysis, it is necessary to ensure that the survey population is well defined. Groundside interviews will yield air passenger information linked to specific flights. Therefore, to know how each interview should be factored so that the survey results represent the characteristics of all O-D air passengers, passenger counts should be obtained for each flight. In the case of through flights, it will be necessary to obtain counts of both the terminating and originating passengers. If passenger loads on each flight are not available from the airlines or the airport, they must be estimated using an alternative data source. The airline schedule for the airport can be obtained prior to the survey for the survey period from the OAG or similar sources. This dataset will provide scheduled arrival and departure times, along with aircraft type, from which the number of seats can be estimated. Recent data from the airlines can be used to apply load factors to each flight, by flight sector, to generate estimates of enplaned and deplaned passengers. Since connecting passengers do not use the groundside facilities, they must be subtracted to obtain estimates of O-D passengers. Quarterly data on connecting passengers by airline and flight sector can be estimated from the U.S. Department of Transportation Bureau of Trans- portation Statistics Airline Origin and Destination Survey database (see bibliography). A key aspect of groundside surveys is that the interview is directed at the occupants of vehicles, who may include air passengers, greeters, well-wishers, or airport employees. With this in mind, the survey planning team must be able to supplement the interview process with data from traf- fic counts, possibly in combination with other data sources. The expansion process, involving vehicles as the base unit, requires data from a variety of sources so that the interviews of the occupants of each vehicle can be made representative of the population. These data sources include the following: • Parking ticket data—to factor the interviews at the parking lots. • Curb activity—to factor the interviews at the curbside, including the public vehicle area, taxi queues, and TNC pick-up and drop-off areas. [Information on conducting curb activity surveys can be found in the documentation for the 2005 groundside survey at Toronto Pearson International Airport (Cripwell and Turpin 2006).] • Rental car activity—to factor interviews in the rental car area, both pick-up and drop-off. • Hotel courtesy vehicles, rapid transit, and other public modes—possibly obtained from AVI systems. • TNC vehicle trips—obtained from operator reports of vehicles crossing the airport geofence. 10.12.3 Staff Requirements Temporary staff requirements always drive the cost of an interview survey. With ground- side surveys, there is considerably more latitude in defining these requirements, although the required sample size remains a governing aspect.

Air Passenger Surveys 175   Whether the survey is covering the entire airport for a short period or covering well-defined segments at different times over a longer period (as discussed in Section 10.12.4), the require- ments will be about the same in terms of the total number of interviewer days. While the calculation of the survey sample size (discussed in Section 10.6) is a key consider- ation, interviewer requirements are also determined by the number of locations that need to be surveyed to ensure representative results. This need for interviewers to cover all locations may result in a higher number of survey responses from some locations than is required to meet the minimum requirements for desired statistical confidence since it is inefficient for interviewers to keep moving between locations and they may remain in one location longer than strictly necessary to perform the target number of interviews. The following interviewer staffing guidelines are based on an extensive groundside survey at Toronto Pearson International Airport in 2005 (Cripwell and Turpin 2006): • Curb area: – Three interviewers per terminal door for a single multi-use curb. – Two interviewers per terminal door for a public use (inner or outer) curb area. (Inner curb areas are adjacent to the terminal and, at Toronto Pearson International Airport, are reserved for taxis, limousines, and the like, while outer curb areas are dedicated to public use.) – One interviewer for each taxi queue (arrivals area). – One interviewer for each of four designated stops for local bus or other services (such as shared-ride taxi, shuttle services, and hotel courtesy vehicles) in arrivals area. • Parking areas: – One interviewer per ticket spitter entrance area for a parking lot, or – Two interviewers per pay-on-foot parking payment machine area. • Rental car and remote parking: – One interviewer per pick-up area, where the rental vehicle is picked up adjacent to the terminal. – One interviewer per drop-off area, where the rental vehicle is dropped off adjacent to the terminal. – One interviewer for each shuttle pick-up area for remote locations where air passengers are picked up adjacent to the terminal by a shuttle service that takes them to a remote or off-airport car rental agency or parking facility. (Where shuttle service for remote or off- airport vehicle drop-off or parking is provided, the interview coverage is typically provided by the curb interviewers.) • Supervision requirements vary with the airport configuration and experience of the inter- view staff and should be included in estimating staffing levels. The number of supervisors required for each shift depends on how frequently they need to check on each interviewer, how much time they spend with the interviewer each time, and how long it takes to move between interview locations. Although this survey was undertaken a number of years ago, the general principles for con- ducting such surveys have not changed. These guidelines should be used in conjunction with the requirements for sample size and the information in Table 10-6 showing the approximate number of responses per interviewer expected under different conditions. These estimates are based on the use of well-trained inter- viewers who can approach a vehicle, engage its occupants in an interview, and then complete a 20-question interview (including some observation entries) in less than 2 minutes.

176 Guidebook for Conducting Airport User Surveys and Other Customer Research 10.12.4 Time and Space Considerations Large international airports have multiple terminals and groundside facilities. Groundside surveys must be conducted at numerous locations and across the whole day of airport activity, from 6:00 a.m. (or earlier) to 11:00 p.m. (or later). This time span calls for at least two shifts per day of interviewers. Whether interviews earlier than 6:00 a.m. or after 11:00 p.m. are required depends how important it is to obtain groundside data for these periods. The groundside survey design, however, will not necessarily require that all terminals and all areas be covered at the same time. For medium- to large-sized airports, such a requirement would take an enormous and unwieldy number of interviewers for short periods of time. Instead, the airport or terminal can be subdivided in time and space to make the survey more manageable. For this subdivision process, the survey design must maintain the link to the ancillary data so that the weighting can be completed for the analysis. Air passenger volumes will naturally divide into departing and arriving segments, as well as by terminal, and these volumes can be determined independently. In the analysis, it will then be possible to generate a weighting factor for each interview with a departing passenger as a function of the number of originating passengers. Similarly, the groundside curb areas are often divided into separate arrival and departure areas. Under these conditions, it is possible to design a groundside survey that segments the population into distinct groups, each of which can be surveyed independently. As an example, consider the following survey plan for a single terminal with an upper-level departure area and a lower-level arrival area. The departure level has a single curb, while the arrival level has an inner and outer curb with taxis and buses at the inner curb and the general public at the outer curb. In addition, there are pay-on-foot parking payment machines at two locations in the terminal, a single lobby for the rental car agencies, no off-site remote service, Area Conditions Estimated Number of Interviews per 8-Hour Shift Curb—public and taxi drop-off 150 Curb—public and taxi drop-off, public pick-up 125 100 Curb—taxi waiting queue 150 Curb—shuttle stops 80 Parking—ticket spitter 125 Parking—pay on foot 150 Rental car—pick-up 100 Rental car—drop-off 100 Rental car or remote parking—shuttle service 50 Source: Adapted from Table 5-4 of ACRP Report 26 (Biggs et al. 2009). As long as there are passengers waiting in the queue Constant high volume of activity throughout the survey period Constant high volume of activity throughout the survey period One peak period of activity and lower volumes during the shift N/A As long as there are passengers waiting in the queue Primary entrance to the parking lot; when traffic is heavy, the proportion of vehicles interviewed will be lower. Common drop-off area for all agencies, or interviewers move between drop-off areas Single lobby area for multiple agencies, or interviewers move between agencies during shift Pick-up area for service to remote area; rate assumes a lower volume of activity than in rental car lobby area Table 10-6. Number of responses expected per interviewer by groundside location.

Air Passenger Surveys 177   and a common drop-off location for all rental car agencies. Table 10-6 shows the expected number of completed interviews per 8-hour shift for short, focused interviews conducted by efficient interviewers. Note that the number of completed interviews per shift depends on the length of the questionnaire, traffic volumes and flows, and the method used to conduct the interviews; in some cases, the number completed may be significantly below the values shown. The rate at which completed interviews are obtained will vary over the day since traffic volumes vary. Table 10-7 shows the corresponding interviewer staffing requirement for this example. Note that fewer interviewers are required for the departures area than for the arrivals area, which is quite common since there are typically fewer locations where interviews need to take place in the departures area than in the arrivals area. With the interviewer resources established for each area, it is then a matter of making up a schedule of at least two shifts per day so that the arrivals and departures areas are surveyed an appropriate number of times during the survey period. Based on the required number of interviews, the number of shifts and therefore the number of days required for the survey can be calculated. The shift plan will also have to cover each direction for each day of the week so that, for example, an early shift for arrivals on each day of the week and an early shift for departures on each day are both included in the survey design, and likewise for the late shift. The survey design should also ensure that staff assigned to a late-night shift are not assigned to the next early-morning shift. It is assumed that week-to-week differences in the traffic pattern will be negligible. With this plan in effect for a terminal with an adequate volume of traffic throughout the day from 6:00 a.m. to 11:00 p.m., each shift of interviewers will perform 1,500 to 2,000 interviews. This example demonstrates why the sample size is less of a concern than the coverage in time and space and the correct weighting of responses. When the survey is conducted over a week or more, the numbers of responses at each individual survey location are usually sufficient for detailed planning purposes. 10.12.5 Questionnaires To lay out the specific questions and establish the overall flow of the interview, a generic groundside interview questionnaire should be developed first. This questionnaire will then be modified to suit each of the interview locations for the survey. Experience has shown that interviewers have difficulties with a generic questionnaire that covers activities that have both already occurred and are yet to occur. Arriving passengers did some things in the past that Departures Area Departures curb—three doors into terminal 9 Bus and shuttle drop-off 3 Rental car drop-off 1 Arrivals Area Arrivals curb—inner area, 12 stops, plus two taxi queues 5 Arrivals curb—outer area, three doors 6 Parking payment machines 4 Rental car pick-up lobby 1 Source: Adapted from Table 5-5 of ACRP Report 26 (Biggs et al. 2009). Area Interviewers Table 10-7. Example of number of interviewers required.

178 Guidebook for Conducting Airport User Surveys and Other Customer Research departing passengers will do in the future. Keeping the past and future verb tenses in a logical order on a generic questionnaire is not easy. Therefore, it is preferable to design questionnaires for each survey location and air passenger direction. The first part of the form will contain obser- vation information that the interviewer can complete as the target vehicle is approached. This information includes the interviewer, interview number, time, location, vehicle type, number of occupants, and number of pieces of baggage (where baggage is visible). Standard rules for counting people and baggage must be defined, such as: • Any child over the age of two is counted, and • All baggage (except purses) is counted. After a short introduction, the first question should always be: Have you been interviewed at the airport today? This question has several purposes: to remove undue survey burden on the airport users by not interviewing more than once, to eliminate duplicate information, and to count the number of vehicles that go to more than one groundside facility in one airport trip (implied by a respondent who reports having been interviewed already). For departure trips, visits to two groundside facilities could involve a stop at the curb followed by entry to the parking lot. For arrival trips, it could involve the opposite, with exit from the parking lot followed by a stop at the curb to pick up passengers and baggage. One section of the questionnaire will deal with the purpose of the vehicle trip: dropping off air passengers, picking up air passengers, or other activities (e.g., buying tickets, using airport concessions, or checking baggage for a later flight). While there will be several response options on the generic questionnaire, the number of responses on the questionnaire for each survey location may be reduced to a few or only one. For example, at the taxi drop-off or taxi queue areas, there are usually only air passengers; therefore, response options that greeters or well- wishers could choose would not appear on the questionnaire for those locations. Another section will deal with the trip origin or destination within the local area. At depar- ture locations, this section typically appears early on the questionnaire, reflecting the sequence of activities, whereas with arrival locations, this section typically appears later. These questions will include the type of trip origin or destination and its geographical location (as discussed in Section 10.7). Another section will cover the air trip, which includes the number of air passengers (in this vehicle) as well as the baggage that was (or will be) checked. Experience has shown that air passengers usually do not know their flight number, but they do know the airline, where they are coming from or going to, and the approximate time of arrival or departure. A question about whether there will be a transfer en route to the final destination is normally required. This information is usually sufficient to identify the flight on which the air passengers are departing or arriving. A final question will ask whether the air passengers are residents of the area served by the airport or are visitors to the area. Asking air passengers for their city and zip code (postal code) of residence can allow responses to be classified appropriately without relying on whether the respondents consider themselves a resident or a visitor and can provide useful additional infor- mation to resolve ambiguous cases. Of course, most respondents will have no difficulty deciding whether they are a resident of the area served by the airport, but this may be less clear to those who came by ground transportation from some distance outside the local area. A sample set of groundside interview questionnaires that were used in the 2005 groundside survey at Toronto Pearson International Airport (Cripwell and Turpin 2006) is provided in Appendix F.

Air Passenger Surveys 179   10.12.6 Calculating Response Weights Each groundside survey interview will have a number of different response weights attached to it at the time of analysis. These weights will all be calculated from the proportion of inter- views to a given population. During the analysis, the appropriate weights can then be applied to the responses. Following are two sample weighting factors and the source data that are required to calculate the corresponding weights: • Vehicle counts, possibly by mode. Given the total number of private vehicles that stopped at the departures curb and the number of vehicle occupant interviews in a specific time period, a weight factor can be calculated. The sample percentage will naturally vary with fluctuating demand and a static interviewer resource. Periods such as 1-hour intervals or peak and off-peak times could be used. Thus, the weight assigned to each interview during the morning peak will be higher than the weight assigned to each interview during a slow mid-morning period. A similar approach can be followed for taxis, parking, and other modes. • Passenger counts. Given the total number of originating or terminating passengers, a weight can be calculated based on the number of passengers covered by the interviews compared with this total number. An hourly weight can be calculated using the same approach as that for vehicle counts by time of day. A further refinement on the passenger-based weight factor can be calculated when passenger loads by flight are available, although the caution given in Section 10.8.2 about the use of flight-specific weights should be noted. Other weighting factors can be specified and calculated, depending on the available data and the goals of the survey. 10.13 Checklists Conducting airport passenger surveys involves numerous people conducting many different tasks. Checklists are a good way of identifying tasks, monitoring progress, and ensuring that all tasks are done. Checklists are particularly useful for organizational tasks involving a number of people. Common types of checklists used in passenger surveys, and examples of items covered, are as follows: • Preparation of contract—defining contract, taking care of legal and administration details, setting the survey schedule, preparing the RFP, defining evaluation criteria, determining where to publicize the RFP or to whom to send it, holding a pre-bid meeting, conducting proposal evaluation and contractor selection (holding interviews if required), informing bidders, negotiating contract, obtaining legal approvals and a signed contract, and holding a project initiation meeting. • Questionnaire development—getting input from all parties, preparing the initial draft, reviewing and formatting the questionnaire, conducting pretest and pilot test (including printing or programming EDCDs), making final changes, and final printing or programming of EDCDs. • Training—dates, venue (including seating, display boards or projector, etc.), trainers, trainees, other project team members to be present, parking, airport tour (with escort if trainees are not badged), curriculum content, handouts (instructions, questionnaires, procedures, schedules, contingency plans, etc.), EDCDs, walkie-talkies, development of interviewing skills, practice interviews, testing, and retraining. • Forms and supplies—approvals for conducting survey (as required from airport operator, airlines, etc.); letter from sponsoring organization authorizing interviewers to conduct survey; printing of questionnaires (different colors for different versions); timesheets for staff; survey

180 Guidebook for Conducting Airport User Surveys and Other Customer Research logbooks and flight record log sheets; check-out/check-in sheets for EDCDs, radios, and other equipment; prepaid mail-back envelopes; and flight gate schedule sheet. • Logistics—badging of interviewers, organizing of survey field office, equipment for field office (printer, copier, computers, Internet access, telephone, fax machine, power extension cords, etc.), storage space in field office for supplies and interviewers’ personal items, locking and unlocking of field office, vests and name tags for interviewers, survey equipment (EDCDs and chargers, pens or pencils, clipboards, and bags), communication equipment (walkie-talkies and chargers), and parking for staff. • Contingency plans—interviewers absent or arriving late, cancelled or delayed flights, gate changes, equipment malfunction (EDCDs, copier, printer, computer, Internet, walkie-talkie, etc.), bad weather (for groundside surveys outside), abusive passengers, delays at security checkpoints, uncooperative airline staff, and disruptions due to a breach of security or emergency situation. Examples of checklists are provided in Appendix D.

Employee Surveys 11.1 Introduction Many of the issues related to planning and designing employee surveys are common to other types of airport user surveys (and the reader will be referred to those sections of the guidebook where applicable). Employees working at the airport are an important user group. Although their numbers may be relatively small, they use the airport facilities and services frequently, and their atti- tudes toward those facilities and services can affect other people’s uses and perceptions. For the purpose of this chapter, airport employees include employees of the airport operator as well as employees of airlines, government agencies, and other organizations providing services or support functions at the airport. Employees of the airport authority or other organizations operating at the airport (such as airlines and concessions) who have direct contact with airport customers can strongly influence these customers’ perception of the airport. Therefore, airports have a strong interest in under- standing those employees’ attitudes; satisfaction with their employment, airport facilities, and services; and familiarity with airport procedures and services. 11.2 Purpose of the Survey and Data to Collect The reasons for conducting employee surveys are often similar to the reasons for conducting air passenger surveys, and frequently the two are part of the same study. Employee surveys are also conducted to address employee-related issues. The types of issues addressed through employee surveys include the following: • Satisfaction with airport facilities and services—to identify areas where improvements are required, track trends, and assess whether specific actions have improved satisfaction levels. • Concession planning and performance—to collect information on use of concessions, length of employee meal or other breaks, problems with current concessions, and desired improve- ments. Employees can be significant users of the concessions, especially food and beverage concessions, and should be considered in their design. • Transportation planning—to obtain information on employees’ modes of transport to and from work, routes, travel times, work schedules, parking requirements, and desired improvements. • Employee issues—to obtain feedback on employee awareness of security or emergency pro- cedures and effectiveness of employee communications; employee satisfaction with their work environment, airport, or their own management; and their experience as an employee. Surveys of employee satisfaction with their work environment or experience will generally be restricted to employees of the airport authority since it would typically not be appropriate C H A P T E R   1 1 181  

182 Guidebook for Conducting Airport User Surveys and Other Customer Research for an airport to survey employees of other organizations regarding their job satisfaction. However, situations may arise where the airport may partner with other employers to under- take an airport-wide survey on job satisfaction and related issues. As with all airport user surveys, the first step in conducting an employee survey is to outline its goals and purpose (as discussed in Chapter 3). The choice of employees to be included will depend on the purpose of the survey and could include the following: • All people working on airport property. • All employees based at the airport; this group excludes visiting flight crews, taxi drivers, and so forth. • Selected employer or employee groups, such as airport authority staff or airline flight crew. • Employees working in a specific facility or geographical location, such as the passenger terminal, the aircraft apron, and an airline maintenance base. Confidentiality can be an issue. Employees may be reluctant to provide truthful responses if they think their responses could be used against them by their employer or the airport. Questions related to airport performance or communication could raise this concern. If such reluctance is an issue, the survey could be designed so that the responses are anonymous. Alternatively, the survey could be conducted by a third party with assurances that the individual responses will not be provided to the airport, and results will be presented in aggregate form only. 11.3 Survey Methodology The choice of survey method is governed by the target employee group and the available information that allows these employees to be identified, sampled, and contacted. The methods appropriate for use in employee surveys are either on-site intercept surveys or self-administered surveys distributed on the basis of lists of employees provided by employers. For surveys of employees based at the airport, ideally lists of employees should be obtained from each employer and used to select the employees to be surveyed. Some employers may not be willing to provide a list of employees but may be willing to distribute the questionnaires themselves. If neither of these methods is possible, on-site surveys could be used, provided access can be obtained to work locations or areas where the employees have their breaks and that this does not interrupt work activities unless the employers have agreed to allow some interruption. Surveys of all employees working on airport property (including flight crews, for example) will need to include an on-site intercept component or be conducted entirely as on-site inter- cept surveys since some employees may only be in any one location fairly briefly. In particular, airline flight and cabin crews are typically at a given airport for a limited time and may not even be based there, although they may start and end their duty day there. 11.3.1 Online Surveys With the widespread availability of smartphones and Wi-Fi coverage at many airports, online surveys offer an effective way to survey employees and have the advantage that respondents can complete the survey when they have time, or even at home. Since many of the surveys will be completed using a smartphone, the design of the survey website should take this into account. The survey website should allow respondents to complete part of the survey, exit from the site, and return later to continue the survey. As with all online surveys, consideration should be given to limiting access to the target individuals. Respondents can log in with their email address or a login code provided in the

Employee Surveys 183 invitation to participate in the survey. It should be stressed that this code is unique to each respondent, and they should not share it with colleagues (or anyone else). In the case of respon- dents accessing the survey with their email address, the login code can be used in place of a password. Each login code should be unique and meet the usual criteria for passwords (adequate length and a mix of uppercase and lowercase letters, numbers, and special characters). The invitation to participate in the survey should stress that respondents may need to retain their login code so that they can return to the survey later if they exit before completing the survey. Most current online survey software recognizes returning participants as long as they are using the same device, although returning respondents will need to log in each time if they are using a shared computer. Since that cannot usually be determined from the device itself, the survey will need to ask if the response is from a shared computer. Once a survey has been completed and submitted, the login code for that respondent should be disabled. 11.3.2 On-Site Surveys For on-site intercept surveys, each of the employee groups to be surveyed should be identi- fied, and optimal locations to conduct the surveys should be determined. Employees should not be surveyed when they are busy with work, and it may be difficult to contact them during breaks or immediately before or after work. In these situations, self-administered questionnaires that employees can fill out when not working are the best option. The questionnaires could be distributed to staff in the break rooms, at the check-in counters or work locations, or while they are entering or leaving food courts. Nobody should be approached entering or leaving a restaurant or cafeteria or while they are eating unless permission has been obtained from the proprietor beforehand. In some situations, it may be possible to conduct intercept interviews if the questionnaire is fairly short. There are various options for collecting self-administered questionnaires. The survey staff could return later to pick up the questionnaire and could clarify any questions or responses with the employee at that time. Otherwise, drop boxes could be provided in the employee break rooms or employer offices, or the questionnaires could be mailed back. In terms of logistics, employers at the airport should be notified of the survey and, where appropriate, access to employee break rooms and provision of drop boxes should be arranged. The advantage of using on-site intercept surveys is that it is not necessary to obtain employee lists, which can be difficult to compile, especially if transient staff—such as airline crews and shuttle bus and other ground transportation drivers—are to be included. The main difficulty with on-site employee surveys is obtaining a representative sample. However, if the number of employees in each employee group is known, at least approximately, weightings can be applied in the analysis phase so that the results better match the employee population. 11.3.3 Employer or List-Based Approach The first step in using a list-based approach is to develop a list of all employees working at the airport by asking employers for contact information for their employees for use in conducting a survey. Some employers may prefer to distribute the questionnaire directly to their employees and not provide contact information. The information available on the employees will determine which of the following methods can be used to contact them: • Email—generally the quickest and easiest method, but email addresses may not be avail- able for all employees. The message would ask them to participate in the survey and could provide a link to an online questionnaire (discussed in Chapter 3 and the following section).

184 Guidebook for Conducting Airport User Surveys and Other Customer Research Alternative methods, such as including the questionnaire as an attachment or in the body of the email, are not recommended because inexperienced users often have trouble dealing with these files and returning the completed questionnaires can be problematic. If email addresses are known for most but not all employees, those without email addresses could be surveyed by mail or telephone. • Mail—usually the most reliable method for contacting employees whose home or work addresses are available, but responses are usually slow and response rates are typically low. • Telephone—allows the use of interviews, rather than self-administered questionnaires, but is generally more costly to administer. As discussed in Chapter 3, interview surveys generally provide more reliable results than self-administered surveys. It may take a number of calls just to reach the employee on the telephone, and many employees may not have a telephone at their work location. Section 13.4 includes an extensive discussion on conducting telephone surveys. • Employer-distributed—recommended to be placed in an envelope, with the employee’s name on the envelope to help ensure that each employee to be surveyed receives a questionnaire. Completed questionnaires could be collected using a drop box at the employee’s work location or could be returned by mail. Another option with printed questionnaires distributed by mail or via employers is to include a web address on the questionnaire and invite recipients to fill in the questionnaire online if they prefer this method and have Internet access. If returning questionnaires by mail is an option, postage-paid envelopes should be provided. The advantage of a list-based method is that it is possible to use a structured approach to the sampling of employees to ensure that an unbiased representative sample is chosen (as discussed in Section 5.3). The main disadvantage is that it may not be possible to identify all employees working at the airport and include them on the list, and a biased sample may result. Compiling the lists can also be time consuming, and some employers may not cooperate, resulting in an incomplete list. The initial step should be to ask employers if they will provide a list of employees and, if some will not, then decide how to proceed. It may be necessary to switch to an on-site survey or a hybrid approach in which some employees are surveyed with a list-based method and others are surveyed using an on-site method. Switching approaches part way through the survey planning process can increase the time and cost involved. 11.4 Sampling Approach The sample size and the method used to select the sample of employees to survey will depend on the total number of employees (population) and any specific subgroups to be analyzed, as well as the desired accuracy. If using mail, email, or employer distribution of questionnaires, the low response rates associated with these methods, sometimes well below 50%, should be consid- ered. When using one of these methods at a small airport, it may be best to survey all people on the list (i.e., to conduct a census survey). The process for determining the required sample size is described in Section 5.3. An estimate of the number of employees at the airport will be required, broken out by any specific subgroups to be analyzed in the survey results, even if this is only approximate. On-site surveys should cover weekday and weekend periods and different shifts throughout the day (including night shifts) as well as a wide range of locations. The locations should include areas where the employee groups go during their breaks, such as break rooms and public food courts. Given the practical difficulties of obtaining a truly random sample of employees to inter- view, the resulting responses may not be truly representative of the employee population. It is therefore recommended that responses be weighted to match subgroup sizes, for example by employer category, shift schedules, or work location.

Employee Surveys 185 11.5 Questionnaire Wording and Length Questionnaire length, format, and clarity, and the use of pretests and pilot tests, are dis- cussed in Chapter 5. The questionnaire for employee surveys should be relatively short since employees are usually busy during work time and value their breaks. The time needed to complete the survey should be no more than 5 to 10 minutes. Chapter 5 provides examples of types of wording that can be confusing, many of which are applicable to employee surveys. Other examples of potentially confusing wording include the following: • Amounts spent at concessions. Confusion may occur as to whether the amount is per visit (in which case the number of visits per day is also required) or the amount spent per day. It is difficult for employees to estimate the average amount spent per day at concessions that they only use once a week or once a month. A better approach is to ask when they last visited a particular type of concession and how much they recall spending. The responses then provide a distribution of the frequency of visits and the amount spent. The question should make it clear whether taxes and tips should be included. • Work location. Some employees work in multiple locations. • Category of employer. Employees may not recognize categories of employers described using industry jargon such as “concessions.” It is better to provide a wide range of options so that few people will respond with “other.” Include volunteer positions in the response options if the airport uses them (e.g., in information booths). If the survey responses are confidential, it may be better to ask for the name of the employer and classify the responses after they are obtained. Two sample questionnaires for employee surveys are provided in Appendix G. 11.6 Measures to Obtain and Enhance Responses Rates of non-response can be a major problem with employee surveys, particularly for surveys distributed by email, mail, or hand. Employees are often busy, have job responsibilities to attend to, and may view a survey as just one more distraction to be avoided or ignored. The two main considerations in obtaining an adequate response rate are to clearly explain the purpose of the survey in a way that will make the survey of interest to employees and to ensure that it is easy for them to respond. The introduction should identify who is conducting the survey and who it is for. It should state how the results will be used, highlighting aspects that could benefit them as airport employees. Aspects such as the quality and friendliness of interviewers and the length, format, and ease of understanding and completing the questionnaire (discussed in Section 10.9) apply to employee surveys as well. If questionnaires are being handed to employees while they are at work, they should be approached when not busy, asked to fill out the questionnaire at their convenience, and informed what to do with the completed questionnaire. Options include returning later to collect the completed questionnaire, having a drop box, or using prepaid reply mail. It will be necessary to go to the same work locations over a range of times and days of the week to cover employees working different shifts. Interviews in employee break rooms can be successful. The topics can become points of conversation among employees, and they develop more interest in the issues being covered. One of the best ways to obtain an adequate response to an employee survey is to convince their managers that the survey is important and seek their assistance in strongly encouraging

186 Guidebook for Conducting Airport User Surveys and Other Customer Research the employees working for them to respond, including allowing them to complete the survey during working hours. Involving the appropriate managers during the survey planning can help obtain their support. 11.7 Survey Budget Section 3.6 provides a detailed discussion of the steps in developing a survey budget. Concession and satisfaction surveys of employees are often conducted in conjunction with passenger surveys, and this can significantly reduce the costs of the employee survey, especially an on-site survey. Contact with employees is best done during quiet periods at the airport, so employee interviews can often be conducted by the same set of interviewers in periods when the employees are not busy with passengers. Costs for developing the questionnaire may also be lower as there will often be overlap between the employee and passenger versions. Where the employee survey is being conducted in conjunction with a passenger survey, costs can be reduced by keeping differences in the questionnaire and data analysis to a minimum and by reporting results as a section in the report for the passenger survey. Reductions in the sample size and time spent setting up the survey will decrease the costs but will lead to reductions in the accuracy of the results. When determining the scope and budget for an employee survey conducted in conjunction with an air passenger survey, note the proportion of airport users represented by employees and scale the budget for the air passenger survey accordingly. This proportion is usually relatively small, but at large-hub airports with airline crew bases, maintenance bases, and extensive cargo facilities, the ratio of average daily employees to average daily enplaning passengers can range from 25% to over 50%. At airline connecting hubs, the ratio of daily employee ground access trips to originating passenger access trips frequently approaches and, in some cases, exceeds 100%, as discussed in ACRP Synthesis 5: Airport Ground Access Mode Choice Models (Gosling 2008).

12.1 Introduction Many of the issues related to planning and designing concessionaire and other tenant surveys are common to other types of airport user surveys, and the reader will be referred to those sections of the guidebook where applicable. Airport tenants include a wide range of organizations, including concession operators (retail, food and beverage, car rental, courier, entertainment, etc.), airlines, government agencies, organizations providing aircraft and aviation services, organizations handling cargo and mail, general aviation aircraft owners and fixed-base operators, and (often) non-aviation businesses. The particular tenants to survey will depend on the goals and purpose of the survey. Although concessionaires are one type of airport tenant, their concession agreements are usually somewhat different from those of other types of tenants and often include fees based on a percentage of their revenue. In addition, the experience of air travelers in food, beverage, and retail concessions can play a significant role in their satisfaction with the services provided by the airport. For these reasons, the issues of interest in surveys of concessionaires may be different from those in surveys of other types of tenants. This is reflected in the discussion in the rest of the chapter, where issues specific to surveys of concessionaires are identified and addressed. 12.2 Purpose of the Study and Data to Collect Concessionaire and tenant surveys are conducted for a variety of reasons. Typical examples, and the types of data collected, include the following: • To obtain information to determine the economic impact of an airport. The types of infor- mation collected typically include data on gross revenues, wages, and taxes; numbers of employees by part-time or full-time status and type of work; and the value of capital assets and capital expenditures. • To determine satisfaction with the services provided by the airport in its role as landlord. This survey could include tracking concessionaire and tenant satisfaction and requirements for airport-provided services, identifying key factors that influence overall concessionaire and tenant satisfaction, gathering feedback about the quality of the services provided by the airport and how well these meet concessionaire and tenant needs, and identifying oppor- tunities for enhancing concessionaire and tenant satisfaction and value. Because this type of survey addresses the relationship between the airport authority and the concessionaires and tenants, and touches on competitive issues between the concessionaires and tenants, it should be conducted by a third party, and responses should be kept confidential and released in aggregate form only. C H A P T E R   1 2 Concessionaire and Other Tenant Surveys 187  

188 Guidebook for Conducting Airport User Surveys and Other Customer Research • To obtain information on customer service through techniques such as mystery shop- ping or studies to obtain price comparisons with identical or similar off-airport shops and services (e.g., restaurant chains). The results are used to make specific improvements to concessions. As with all airport user surveys, the first step in conducting a concessionaire or tenant survey is to outline its goals and purpose. (See Chapter 3 for a discussion on specifying goals, defining the purpose of a survey, the importance of doing so, and who should be involved.) 12.3 Survey Methodology The options for conducting concessionaire or tenant surveys include using online surveys, mail surveys, or on-site visits. Because the number of tenants is generally relatively small, all tenants in the categories of interest are usually surveyed, especially if the results will be compiled and reported for tenant subgroups. Where there are many tenants, selection of a sample is appropriate. On the other hand, surveying all tenants will help ensure an adequate response. 12.3.1 Online Surveys Online surveys are ideal for conducting concessionaire and tenant feedback and economic impact surveys. The airport should have contact information (which will generally include email addresses) for all tenants. There may be exceptions in the case of general aviation aircraft owners, who may have to be surveyed using another method. All tenants to be surveyed are sent an email asking them to participate in the survey and providing a link to the questionnaire on the Internet. (See the introduction to Part II for a discussion on setting up an online survey.) It is often a good idea to notify tenants in advance of the survey by email or telephone and to follow up with non-responding tenants, again by email or preferably telephone. Although telephone calls are more time consuming, they are less easily ignored than an email message and, therefore, may be more effective. Using an online survey with an initial contact via email has the following advantages: • The survey is relatively easy to set up and administer. • The use of email allows the survey invitation to be easily forwarded to the appropriate person in the company. • It allows respondents to complete the survey at a time convenient to them and to do so in more than one session if they wish (depending on the software used). • It allows easy tracking of the number of responses and requires no data entry. • Response rates are usually much better than with mail surveys. The main disadvantages are that some effort may be required to compile a tenant email address list and that some tenants may not have email addresses or easy access to the Internet. A mail survey could be used for those tenants without email addresses. Many online survey tools allow for mailed replies to be entered manually so that they can easily be incorporated into the online survey results. However, response rates for online surveys are not as high as with on-site visits. However, while an airport will generally have an email list of tenant contacts, some effort may be required to update this list to account for staff turnover. Also, depending on the purpose of the survey, the most appropriate contact at a tenant may be different.

Concessionaire and Other Tenant Surveys 189 12.3.2 Mail Surveys Mail surveys are similar in approach to Internet surveys but require each tenant’s mailing address rather than email address. Questionnaires are mailed to each tenant with a postage- paid reply envelope, and responses are mailed back. The advantage of mail surveys is that all tenants will have a mailing address. The main disadvantages are that response returns are often slow, the response rate is typically low, and responses must be entered into a database for analysis. Non-responses are usually followed up by telephone, which is more costly than sending bulk reminders via email but is generally more effective, particularly if the emails have not been reaching the correct person. 12.3.3 On-Site Surveys On-site surveys can be used for concessionaire or tenant feedback and economic impact surveys, although online or mail surveys are generally preferable. If on-site surveys are con- ducted, concessionaire or tenant staff available on-site will often not be able to provide the required information and may be busy with customers, so on-site interviews should be done by appointment. Tenant staff may also have to check company records to respond to some questions; thus, some questions might go unanswered with only a single interview. Generally, tenants are contacted by mail or telephone to advise them about the survey and possibly schedule an appointment. Sending the questionnaire in advance gives the tenants a chance to assemble any information needed. The advantage of on-site surveys is that a deeper understanding of the issues is possible through in-person interviews. The major disadvantages are that scheduling all the required interviews is often difficult, and on-site surveys are more costly to conduct. 12.3.4 Confidentiality As confidentiality is an issue for concessionaire or tenant surveys, it is recommended that they be conducted by respected third parties. Assurances should be given to the concessionaires or tenants that the information will be kept in strict confidence and provided to the airport operator in aggregate form only. However, protecting confidentiality by aggregating survey responses can impose significant limits on the analysis that can be done with the data and the resulting ability to address important questions. Except at the largest airports, even broad classes of tenants will be represented by a small number of respondents, and any cross-tabulations against other variables may make the respondents identifiable. Therefore, before making such assurances, careful thought should be given to what the airport hopes to find out from a survey and how that can be determined without violating the confidentiality of the responses. 12.4 Sampling Approach In the case of online or mail surveys, questionnaires or survey requests will generally be sent to all concessionaires and tenants at the airport. However, including all concessionaires and tenants could be costly for on-site surveys, especially at large airports. If only a sample of tenants is required, a stratified sample of tenants should be surveyed where the tenants are grouped by category—for example, concessions, airlines, government and security agencies, individual

190 Guidebook for Conducting Airport User Surveys and Other Customer Research aircraft owners, and other organizations. The sample size will depend on the purpose of the survey, the desired level of accuracy, and the number of tenants in each category. The process for determining the required sample sizes is described in Section 5.3. Depending on the purpose of the survey, it may be appropriate to sample a greater propor- tion of some subgroups than others. If a similar level of accuracy is desired for each subgroup, a higher sampling fraction would be used for smaller subgroups or for subgroups where the variation in the characteristic of interest, say economic impact, is greater. (See Chapter 4 for an explanation of why this is so.) Details of how to determine the appropriate sampling frac- tion in a given situation are explained in Sections 5.3, 4.4.3, and 4.4.4. For many categories of tenant at all but the largest airports, the total number of tenants is likely to be small enough (under 100) that for any reasonable level of desired accuracy, the sample size for a random sample is so close to the total number of tenants that it would make sense to simply survey every tenant. For economic impact surveys, the treatment of nonrespondents can significantly affect the results. The economic impact of individual tenants varies greatly, particularly between groups of tenants. Use of only the data from responding tenants would lead to underestimation of the total impact, and expanding the sample results on the basis of the average response for each group of tenants could result in large errors, depending on the tenants that did not respond. Every effort should be made to obtain some response from all tenants, even if only basic information, such as the number of employees or gross revenues, is obtained from some organizations. This information can then be used to develop weights to expand the data received from each group of tenants to represent the total group. If no information can be obtained from some tenants, the airport will always know something about the size of each tenant. In many cases, they will know the number of employees from issuing security badges. They will also know the size of the area leased and, in many cases, will have traffic or revenue data from required reports or concession fees. This information can be used to estimate the relevant economic measures for the nonrespondents based on the data from the responding tenants. 12.5 Questionnaire Wording and Length Tenant feedback and economic impact surveys should ideally take less than about 20 minutes, including finding the data. There is always a temptation to ask for more detailed information than is required; however, this can significantly affect the response rate. The survey should be restricted to requests for information that a knowledgeable person within the organization will know immediately or be able to obtain fairly easily. As discussed in Section 5.4, the questions to be asked should be determined by exactly what information is required and how the response from each question is going to be used. A sample tenant questionnaire is provided in Appendix H. 12.6 Measures to Obtain and Enhance Responses A high response rate is important for obtaining accurate, unbiased results. The following measures could be taken to improve the response rate: • In the introduction to the survey, include the name of the company or organization con- ducting the survey and the survey sponsor. The introduction should clearly state the purpose and how the results will be used, highlighting aspects that could benefit the respondents as airport tenants.

Concessionaire and Other Tenant Surveys 191 • Make it easy for the tenants to respond. Use of email to distribute the questionnaire allows it to be easily forwarded to the appropriate person. Try to keep the requested data to things the respondent will know immediately or be able to find out easily. • Alleviate confidentiality concerns. The survey should be conducted by a third party with an assurance that all responses will be kept confidential and that information will only be released in aggregate form so that responses from individual tenants cannot be identified. It may be useful to include some information about the survey organization to show that it is an organization with an industry reputation for protecting confidential information. • If conducting an online survey, use software that allows respondents to save partially completed responses and complete them at a later time. • Follow up with non-responding tenants. Initially this can take the form of email reminders but should include a telephone call if nothing has been received after two or three reminders. For those declining to complete the survey, the telephone call would be a good opportunity to attempt to obtain at least an estimate of the number of employees or gross revenue. • Give recipients adequate time to respond. The deadline for responding should allow sufficient time for people on vacation, leave, or work-related trips to respond when they return. Two weeks is a reasonable period, increased to three in July and August or around major holidays (although it is preferable not to schedule tenant surveys during these periods if possible). Reminders should be sent weekly, with a final reminder within 48 hours of the survey’s closing time. In contacting concessionaires and other tenants to solicit their participation, it is critical to stress that the requested information is considered important by the airport management. Also, to the extent that there is any potential benefit to the concessionaire or tenant based on the results of the survey, these benefits should be emphasized. 12.7 Survey Budget Concessionaire and tenant surveys are relatively inexpensive to conduct online, but costs can increase significantly if initial response rates are low and many telephone follow-up calls are required. Costs will also be much higher if on-site interviews are required. Section 3.6 provides a detailed discussion of the steps in developing a survey budget.

192 Surveys of Area Residents 13.1 Introduction As a general rule, the target audience for resident surveys will be adult members of the general public who live in the region served by the airport or another pertinent community area, such as the area surrounding a competing airport, and the survey sample will be rep- resentative of the community as a whole. This is particularly true in cases where the support of the general public for a planned under- taking is to be solicited. One important consideration in this regard is how “community” will be defined. The geography to be covered may be obvious given common understandings of an airport’s service area, or it may be more complex if notions of the service area are in flux or if some geography other than the service area is to be studied. This can be a particularly significant issue in multi-airport areas or for studies in which the efforts of other competing airports are to be assessed. In all probability, there will also be clearly identified subsets of air passengers. The survey could also be restricted to recent passengers, to passengers with specific travel patterns or airline preferences, or to some other population segment. The key issue is that the survey pur- pose will drive the selection of the target audience, and the characteris- tics of that audience will need to be carefully considered. It is also important to note that the narrower the target audience, the more costly the survey, because those placing the calls will need to search for the appropriate type of respondent. Although there are lists that could minimize this search in at least some cases, there are none that truly represent any particular audience—all are biased in one way or another. Thus, as a general rule, some random approach is preferable. 13.2 Purpose of the Study and Data to Collect Most surveys of area residents are conducted to obtain information for marketing and airport planning purposes. Common topics of inquiry include perceptions of the airport, overall airport experiences, the extent to which residents are using an airport in another area, reasons residents choose one airport over another, what might make the local airport more C H A P T E R   1 3 Best Practice “Every two to three years, we conduct a market perception survey. We target residents in our primary market area who are air travelers in order to learn their perceptions about our airport. We look at drivers of airport utilization and drivers of airport satisfaction to find opportunities for broad marketing strategies, including advertisements, product development, and services. We gather information on our areas of strength and areas for potential improvement.” —Research participant Best Practice “We would like to start talking to Bay Area residents and businesses in three phases to understand what are people worried about, sources of anxiety around travel, [and] what do they want to hear the airport talk about and do when [COVID] restrictions [are] lowering and people [are] coming back.” —Research participant

Surveys of Area Residents 193 attractive to passengers, what messages about the airport would resonate with passengers, and what information sources passengers are using to make airport choices. In unusual circumstances, such as the recent pandemic or the events of 9/11, resident surveys can be used to determine what would make potential passengers feel safe flying. 13.3 Survey Methodology Surveys of members of the general public are usually conducted by telephone because this is by far the most cost-effective method. Although some contend that such surveys can now be effectively conducted online given the widespread use of the Internet, online surveys generally have the lowest response rates of any of the available survey strategies. This in turn can lead to results that are unrepresentative of the population of interest. Online surveys also tend have relatively low data quality with respect to open-ended questions because there is no interviewer to probe for clarity and specifics and because people tend to work with greater haste online than on paper (such as in mail surveys). 13.3.1 Call Sequence and Design Most telephone studies that use lists of numbers do not involve dialing the numbers dozens of times, primarily for reasons of cost. The question then becomes how many calls should be placed. This matters, because the more calls that are made, the more representative the sample becomes as additional hard-to-reach people are included. The general rule among public opinion researchers outside academia is to use a sequence of between four and six calls spread over different days of the week and different times of day. Most call centers dial from 5 p.m. to 9 p.m. local time Monday through Thursday or Friday and during some hours Saturday and Sunday. (Friday evening is generally the least productive time, and Sunday evening is generally the most productive time.) Calls past 9 p.m. are frowned upon, as are calls before 10 a.m. Saturday and Sunday. 13.3.2 Sources of Bias One relatively small source of bias in surveys based on samples of telephone numbers derives from the exclusion of households without telephones, and another emerges from households with multiple landlines. Both of these are quite trivial and are generally, in most cases, dismissed as inconsequential. A larger potential source of bias is how numbers that are not answered are handled. Reputable call centers will treat these as potentially viable numbers that merit follow-up much like numbers with answering machines or voice mail messages. Whether these numbers actually create a bias in any given study is generally unknown, because in most cases pursuing the numbers is fairly rapidly abandoned. A much more important source of bias is refusals because it is clear from many studies that people who refuse differ from those who do not. As a result, it is generally wise only to use call centers that focus on and keep refusals under control. For a medium-interest, relatively brief, and well-designed survey of the general public, a final refusal rate of more than 30% may be an indicator that refusals are not being well-managed. While hard refusals (those that are adamant) are usually best left alone, less strong, soft refusals should be followed up on with refusal conversion efforts. Many call centers now attempt such efforts, and the general consensus is that they are worthwhile. In conversions, particularly

194 Guidebook for Conducting Airport User Surveys and Other Customer Research persuasive or even specialized interviewers try numbers at which a refusal occurred another time. Only if they are refused a second time is the number abandoned. It is also worth noting in this regard that a conversion attempt often reaches a different but eligible individual who would not have been inclined to refuse in the first place. Finally, it is important to recognize that many people refuse not because they never do surveys, but because the interviewer called at an inconvenient time. If callbacks are placed on different days and at different times of day, they will often occur at a better time for the respondent, and consent is readily obtained. It is also possible that someone else who is eligible to participate in the survey and with a generally more favorable attitude will be reached. Hard refusals, however—those who say they never do surveys or who request to be placed on a do-not-call list—are never called back, because the outcome is predictable, and people could be offended. Although survey research is not subject to the do-not-call laws, many people are unaware of this and thus ask for do-not-call protection. Most legitimate call centers oblige them. Last but far from least, there is the issue of cell-phone-only households, which represent more than half of all adults. If an organization is not using address-based sampling (ABS), random cell-phone and landline random-digit dialing (RDD) samples are used in proportion to their presence in the population. As these data vary by state, call centers use published data about cell-phone ownership to determine how to represent a state or local area. The issue of cell- phone sampling is further complicated by the fact that it is illegal to use automated dialing equipment, which many call centers routinely employ, to call cell phones. By law, cell-phone numbers must be dialed manually, thereby adding substantially to the data collection cost. In short, although there is no perfect method for conducting a telephone survey of area residents, experience indicates that the telephone yields better results than other methods like online or mail. As a general rule, blended samples (landline and cell phone) or ABS are used to achieve the most representative result. 13.3.3 Dates to Avoid Although it may seem obvious that certain dates should be avoided in conducting a telephone survey, some organizations overlook this basic issue. Dates to avoid include: • Major holidays, including the day before, day of, and day after Thanksgiving; • Any date during the Christmas and New Year’s holidays, customarily from December 15 through January 2; • The annual income tax due date, generally April 15, unless it falls on a weekend; and • The day of any major sporting event. 13.4 Sampling Approach 13.4.1 Sample Size Generally, sample sizes for surveys of area residents are determined using a formula for data that are expressed as proportions or percentages because most of the results that are obtained in such surveys are expressed as percentages. It is also generally assumed that the distribution of the data will be the worst-case or least-definitive scenario (a 50/50 split), where it is not clear what the majority view is. True pilot tests in order to establish a different and more favorable benchmark are rarely conducted in telephone research, although they are certainly possible. Further parameters for the sample are usually fixed in advance based on the available budget,

Surveys of Area Residents 195 assumptions about the importance of the results, and the risk of making a wrong decision based on the findings. In opinion research, the confidence level is customarily fixed at 95%. The confidence interval or margin of error is then stipulated based on the factors outlined in the previous paragraph. For standard public opinion research with a moderate level of importance, the margin of error is most often fixed at ±5 percentage points, leading to a sample size of 400 (rounded up from 384). The term “importance” refers both to the priority of the issues considered in the survey to its sponsor (in this case the airport) and to the potential consequences of making an incorrect deci- sion on the basis of the results. A margin error of ±3 percentage points, requiring a sample size of 1,000, is also fairly common for studies that are considered of substantial importance. Larger sample sizes are used for high-risk projects and when extensive subgroup analysis will be conducted. Finally, commercial market researchers frequently use a ±6 percentage point margin of error and thus a sample size of 300. Key reasons for this include custom in the profession, the belief that the relative importance of the results does not necessitate a larger sample size, and cost. Either custom or cost often predominates in decision making. Table 13-1 summarizes the margins of error and sample sizes typically used in various circumstances. The sample size information in the table comes from the public opinion and marketing research profession and is particularly useful for surveys typically conducted by telephone among a sam- ple of residents of an area. Detailed discussions of sampling for a variety of situations can be found in Chapter 4, which discusses the desired level of accuracy in survey results. 13.4.2 Sample Selection Unfortunately, there are no lists of telephone numbers of all members of the general public from which an airport could select a list of people to call. Accordingly, less-than-optimal lists or some alternative approach must be used. Various types of lists do exist, but none of them represent randomly selected samples of all people in a given geographic area. Traditionally, the alternative, which is theoretically elegant but messy in practice, has been to use the random-digit dialing method. RDD samples are constructed by combining known pairs of area codes and prefixes (the first three digits of a telephone number) with a random four-digit suffix. The elegant aspect of an RDD sample is that it represents a truly random sample of every telephone-owning household in an area. Unfortunately, however, response rates to RDD surveys have become so low over time that there has been a need to develop alternatives. The situation has been further complicated by the rise in the use of cell phones. When ACRP Report 26 was published in 2009, fully 97% of households had landline telephones, a level of coverage that was far higher than when telephone interviewing originally became popular in the 1970s. As of 2019, the distribution of household telephone ownership in the United States was: • Cell phone only: 59% • Cell phone mostly: 19% Situation Margin of Error Sample Size Low importance ±6 Moderate importance ±5 High importance ±3 High priority of cost ±6 300 400 1,000 300 Table 13-1. Margins of error relative to sample size.

196 Guidebook for Conducting Airport User Surveys and Other Customer Research • Dual use: 18% • Landline only: 3% • No phone: 1% (Blumberg and Luke 2020) Cell-phone numbers tend not to change over time, so for sampling purposes, they follow the same individual. The geographic location of cell phones, on the other hand, changes when people move. Thus, a number that started out in San Francisco can wind up in New York. This makes random-digit dialing with cell phones problematic when geography is an issue, which is certainly the case for airports seeking to study their potential constituencies. Randomly sampling combinations of local cellular area codes and prefixes will exclude those who have moved to the area and include those who have left it. An emerging and increasingly recommended approach is to use ABS, which is based on address lists derived from the United States Postal Service database. A random sample of households is mailed an invitation to participate, with two options for response: a unique online access code and a dedicated 800 number. The invitation can be branded with the airport name, and a small incentive is typically enclosed. Invitations can even be issued in two languages to account for important linguistic subgroups, with one language on each side of the invitation. A key drawback of this approach is that it is likely to underrepresent residents who are not English-speaking and those with lower socioeconomic status, although it otherwise does a good job of generating a representative sample. This can be compensated for by including an RDD cell-phone sample in the survey. ABS can also be also time-consuming and costly. 13.5 Questionnaire Wording and Length Experience suggests that surveys of the general public will start to experience unacceptable rates of refusals and terminations (people who quit in the middle of the interview) when the inter- view goes past about 10 minutes. Cooperation rates are highest when interviews are 5 minutes or shorter, although it is admittedly difficult to craft such a short survey on many topics. A detailed discussion of question wording issues can be found in Section 5.4.3. There are nuances in how questions are worded across methods (for example, the phrase “the following list” works well for self-administered questionnaires but sounds nonsensical over the telephone), but the fundamental principles are the same for all methods. 13.6 Measures to Obtain and Enhance Responses As noted previously, obtaining an adequate response is one key to a survey’s success. A low response rate leads to questionable results. Important aspects in conducting a telephone survey include: • A centralized facility, also referred to as a call center or data collection organization, where interviewers are closely supervised and regularly monitored during every shift; • A thorough interviewer training program; • A comprehensive and proactive approach to identifying and coaching interviewers whose skills need improvement or who are doing something inappropriate during their interviews; and • A state-of-the-art computer-assisted telephone interviewing (CATI) system into which the survey questionnaire and appropriate skip logic can be programmed. There is also the issue of languages. There are some communities in which interviewing needs to be conducted in a language other than English. There are also some areas in which a number

Surveys of Area Residents 197 of languages are spoken by large enough proportions of the population that multiple languages should be considered, and there are airports that translate questionnaires accordingly. It is also important to note that even in areas with high percentages of people who speak other languages, many of these people also speak English, and some would prefer to be interviewed in English. While the interview may take longer as a result, interviewers should be trained to determine a potential respondent’s language preferences. It is also wise to ask the call cen- ter conducting the interviewing what their experience is when questionnaires are translated, because it may be the case that a translation is not worth the time and cost where preferences for speaking English are high. Finally, it is important to note that interviewing in other languages requires both bilingual interviewers and a fixed written translation of the questionnaire that is programmed into the CATI system. If interviewers are allowed simply to translate on the fly, which has been observed on some not particularly rigorous studies, they will come up with many different wordings, which will cause inconsistencies in the way questions are interpreted and in the responses obtained. This can in turn compromise the utility of the results. 13.7 Survey Budget The budget for a telephone survey is influenced by a wide variety of factors. These include: • Complexity of the questionnaire design, • Length of the interview, • Number of open-ended questions, • Languages to be used, particularly if locating fluent interviewers is an issue, • The proportion of residents who are eligible to complete the survey, • The type of sample being used, • The call sequence, • The level of data analysis required, and • The nature of the deliverables in terms of reports and presentations. Geography is also important, in two respects. In areas with a high cost of living, survey costs will be higher because rents, salaries, and wages are higher. And in urban areas, costs will be higher because cooperation is more difficult to achieve than it is in rural or suburban areas. It is also worth noting in this regard, however, that call centers do not need to be in the geographic area where the airport is located. Many researchers work with call centers at a considerable remove, even ones in other time zones. Surveys that are shorter, are targeted to more cooperative areas, or have fewer open-ended questions will generally cost less; those that have a narrower target audience (for example, only people who have made an air trip in the preceding year) will cost more because it will take more effort to locate respondents who meets the criteria. Given all of these factors, it is difficult to estimate a price for a telephone survey of area residents. Table 3.3 presents tasks and direct expenses likely to be involved in a telephone survey, which can be used by airports to estimate budgets.

198 Surveys of Area Businesses 14.1 Introduction In most cases, the target audience for a business survey will be all businesses within the geographic area the airport typically serves. A business may be narrowly defined as a for-profit company, or it may be more broadly defined to include government agencies and nonprofit organizations. Thoughtful and clear identification of the target audience is an important component of the survey planning process and should be driven by the purposes of the study. The major principles of survey design were covered in Chapter 5 and can be referenced when planning a survey of businesses. Generally, these principles do not vary from one target audience to another; they remain critical to every research effort. 14.2 Purpose of the Study and Data to Collect Information derived from surveys of area businesses can help airports understand and address a wide variety of issues, including: • Awareness of the airport as a contributor to the local economy, • Overall perceptions of the airport as a citizen of the community, • The role of the airport in contributing to the local and regional economies for use in eco- nomic impact studies, C H A P T E R   1 4 Best Practice “In the past, we have sent out some surveys to surrounding businesses to better understand business travel trends in the area. Then we developed a business traveler’s advisory group, so we could talk more in-depth about business travel.” —Research participant Best Practice “We did a . . . survey around local Bay Area business, large and small, travel coordinators, asking them prior travel behavior and changes in policy [due to COVID] and when they expect to start booking again and the dimensions around that.” —Research participant

Surveys of Area Businesses 199 • Demand for airport services for presentation to current and potential airline partners, • In areas where there is airport choice, reasons for choosing the airport over another nearby airport, • Perceptions of airport accessibility, • Attitudes toward various ground access modes, • Use of various ground access modes, • Awareness of the airport’s airside and groundside services, • Assessments of specific facilities and services, • Facilities and services that are particularly attractive to business travelers, • Facilities and services that business travelers view as needing improvement, • Facilities and services that business travelers would like to see added, • Current and prospective business air travel for demand-forecasting purposes, • In the COVID era, or in other circumstances where there is passenger reluctance to fly, conditions under which business travelers would or would not fly, and • The role of the airport in contributing to the local and regional economies for use in economic impact studies. 14.3 Survey Methodology Four methods are typically used for business surveys. As might be expected, each has advan- tages and disadvantages. 14.3.1 Mail The use of mail surveys is a traditional strategy that continues to be used despite significant limitations. Over time, it has generally been supplanted by various online approaches. While it does offer a tangible reminder of the airport’s presence on letterhead, it suffers from low response rates. It is also more expensive and time-consuming than even the most sophisticated online strategies. 14.3.2 Telephone Telephone research is relatively expensive, and it is often time-consuming because of the need to reach out to people multiple times in order to connect. It does, however, have higher response rates than mail or online research. It will also yield higher-quality data, particularly for open-ended questions. Vague or overly general responses to open-ended questions can be probed by the interviewer for clarity and specifics that will ensure that the results are useful. 14.3.3 In-Person In-person interviewing at respondents’ offices is the most expensive approach, but it offers several advantages. Cooperation rates tend to be high, data quality is usually superior, and answers to open-ended questions are most likely to be thorough. Generally, in-person interview- ing tends to be used more for mostly qualitative studies than for standard surveys. Chapter 6 provides a detailed discussion of qualitative methods, with Section 6.4.2 focusing on in-depth interviews. 14.3.4 Online For surveys that are not so complex that they require specialized software, online research is the easiest method to implement and also the least expensive. If the survey software permits, respondents can save a partially completed questionnaire and complete it when it is convenient.

200 Guidebook for Conducting Airport User Surveys and Other Customer Research Response rates tend to be low, but this can be mitigated to some degree by advance communication and follow-up. Data quality for closed-ended questions in online surveys can be controlled via the question- naire programming. Answers to open-ended questions, on the other hand, can be problematic if respondents provide incomplete, vague, or enigmatic responses. Written instructions requesting clarity and specificity can help, but they rarely solve the problem entirely. Most surveys of businesses, regardless of method, will suffer from non-response bias to some degree or another. This is particularly important when nonrespondents differ from respondents on key survey metrics. Techniques for minimizing non-response include pre-survey commu- nication, follow-up, and in some cases incentives. Using incentives with businesses can be problematic, however, as some may be offended, and some may be unable to accept the offer. 14.4 Sampling Approach Unfortunately, there is no uniform list of all businesses in a community from which a sample can be selected. Potential sources include chambers of commerce, economic development organizations, employment and training agencies, convention and visitors’ bureaus, and munic- ipal or other local governments. Airport staff who are familiar with the local community may be aware of other resources. Obtaining membership lists can be difficult and may require efforts at the highest levels of the airport organization. Communications from the airport in this regard will need to state the survey purpose clearly, emphasize the limited amount of time that will be required, outline any benefits that might accrue to the organization and its members, and express gratitude for any assistance provided. As appropriate, referrals to other organizations might also be requested. If the potential benefits of participation are notable, a business might be willing to serve as a survey co-sponsor. On the other hand, there might be resistance to getting involved in what might be perceived as yet another survey, which is a key reason the purposes and benefits need to be clearly and compellingly articulated. If multiple lists are acquired, de-duplicating them is an essential step. This will require more than machine matching; some organizations may be listed slightly differently on different lists. If participating in another survey can be bothersome, another survey times two or three could be both alienating and detrimental to the airport’s image. In the best of all possible worlds, business survey samples would be stratified by business size, with potential respondents selected proportionally from the strata. In reality, however, business size information is often absent or suspect. Thus, if a census is not possible, either because the population is large or the budget is limited, random sampling is probably the most advisable approach. Sampling concepts are discussed in more detail in Chapter 5. One other sampling issue that will arise is who within the organization to direct the survey. Often, this information will not be provided at all; when it is, it may vary from organization to organization. Ideally, the airport should establish the type of position within a business that would be best suited to respond to the survey’s questions, determine via telephone or email who that might be within the organization, and direct the questionnaire to that individual. From a practical point of view, however, this is likely to be prohibitively expensive and time-consuming in most communities.

Surveys of Area Businesses 201 It therefore may be most appropriate to address the survey to the head of the organization and ask that it be redirected to the appropriate individual. The position or positions that indi- vidual might hold and the reasons for requesting that this particular person reply should be clearly explained. Two other potential benefits may accrue from this strategy. First, it is often the case that the approval of the leader of the organization would be needed in any event for someone to complete the survey. This is particularly true when information that is viewed as being sensitive or proprietary is being requested. Second, a request from the organization’s head may serve as a motivator for the identified individual to respond. Finally, it is important to be clear that the survey is requesting the point of view of the organization, or at least the respondent’s understanding of that position. Surveys of businesses are designed to represent business perspectives, not individual ones. 14.5 Questionnaire Wording and Length Fundamental principles of questionnaire wording, which are discussed in Chapter 5, apply to business surveys as well. These principles do not change when the potential respondent is representing a commercial, nonprofit, or government enterprise. Briefly, questions need to be straightforward, specific, unambiguous, unbiased, and clearly stated. In terms of survey length, it is particularly important in the case of business surveys to focus on what the airport needs to know rather than what would be nice to know. Most individuals who have the authority to respond to a survey on behalf of an organization have substantial limitations on their time and will be disinclined to respond to a lengthy survey. A general rule of thumb is to limit business surveys to an approximate completion time of no more than 5 minutes. A survey that takes more than 10 minutes to complete is likely to suffer from a low response rate. 14.6 Measures to Obtain and Enhance Responses As noted previously, business survey response rates—actually most response rates—are to some degree a function of the survey method. In the specific case of business surveys, how- ever, there are a number of strategies that can be used to enhance the response rate regardless of method. Promote the survey. If at all possible, seek the support of local business organizations—perhaps those from which the lists of potential respondents were obtained—and ask them to be of assistance by promoting the survey. Possible strategies include short articles in newsletters, e-blasts, and mentions at organizational meetings. Provide complete background information. In the cover letter, email request, or the intro duction to the survey, clearly state the purpose of the research, the reason it is being conducted, the importance of the information being collected, and the use to which the information will be put. Also, try to address the “what’s in it for me” factor. This could be the benefit to the individuals responding to the survey, to the participating businesses, or to the community as a whole. Be sensitive to burden. Every question in a survey should serve a need-to-know purpose, and this purpose should be clear in introductory information as well as in the question itself. “It would be interesting to know . . .” or “It would be nice to know . . .” are statements that often arise during planning sessions and that always need to be challenged. The ethical concept

202 Guidebook for Conducting Airport User Surveys and Other Customer Research is called “respondent burden.” No respondent should be burdened with excessive or unnecessary requests for information. Unless it is absolutely necessary, it is also important to avoid asking for information that is difficult to obtain. To the extent possible, try to find substitutes for requesting that people search for data or resort to checking records in order to respond. Allow enough time to reply. As a general rule, business respondents should be given 2 weeks to respond to a relatively short survey and 3 weeks to respond to a longer one. This rule of thumb reflects sensitivity both to the fact that most people are busy and to absences due to vacations, work-related trips, or personal time. Avoid the Thanksgiving, Christmas, and New Year’s holidays. Conducting surveys during Thanksgiving week, in December after about the 15th of the month, or early in January is unlikely to be particularly successful. When assessing survey timing, make sure to factor in any planned follow-up efforts. Use follow-up or reminders. Assuming that people will reply or respond the first time they are contacted or receive a questionnaire is overly optimistic. For telephone surveys or to schedule an in-person interview, at least three and ideally four calls should be placed, generally at 1-week intervals, and more often in the rare event that potential participants request it. For mail and Internet surveys, three reminders appear to maximize the rate of return; more do not appear to have much of an effect. For best results, reminders should be worded differently. Use Internet software that allows people to save partially completed responses. Unless a survey is extremely short, it is often not possible for people to complete it in a single sitting. Software that allows them to save a partially completed questionnaire and return to it shows sensitivity to the value of time and will likely increase the response rate. Avoid forcing responses. The requirement to answer a question in order to advance in an online survey is annoying to many respondents and can even be perceived as being offensive. (Interestingly, this is really the only method where forcing a response is possible.) Allow respondents to skip a question if they wish. The two situations in which this issue commonly arises are when people do not want to answer a question and when they actually cannot answer it. In the first instance, if it is important to know whether the question was skipped accidentally or intentionally, designers can include answer choices such as “no opinion” or “decline to answer.” In the second, it is important to ensure either that all of the potential answer choices are listed or that the question is structured as an open-end. Although it is common practice to include “other (please specify)” as an answer choice, caution should be exercised in doing so. Typically, this occurs when a question that is essentially open ended is converted to a closed-ended format, often for financial reasons given the cost of coding responses to open-ended questions. If insufficient thought is given to potential answer choices, “other” is often added to cover what is left over. The challenge is that “other” usually understates the prevalence of whatever is written there because not everyone thinks of a given possibility, so the results can be misleading.

Air Cargo Surveys 15.1 Introduction Airports require information on the air cargo activity and the operations of the firms engaged in air cargo operations at the airport for a variety of reasons, including: • Forecasting future levels of air cargo activity at the airport; • Planning air cargo and related facilities and infrastructure to meet current and future needs; • Estimating the volume, composition, and travel patterns of truck activity generated by the air cargo activity at the airport (for planning on-airport and airport access roadways), as well as accounting for the airport contribution to regional truck traffic in environmental impact studies; and • Accounting for air cargo operations in economic impact studies. Although many airports routinely collect data on enplaned and deplaned air cargo from monthly reports submitted by the air carriers operating at the airport, these reports typically only provide the total weight of air cargo loaded or unloaded on or off aircraft and provide little or no information on other key aspects of air cargo activity at the airport, including: • The weight of air cargo that is transferred between flights and does not move on or off the airport by truck; • The amount of freight that is trucked to and from the airport; • The commodities shipped through the airport; • The local, regional, state, and national air cargo markets served by the airport; and • The number of truck or smaller vehicle movements generated by the air cargo activity, the sizes and types of those vehicles, and the origins and destinations of those movements. The second bullet brings up an important consideration. Some air cargo is moved by truck over a considerable distance to or from the airport where it is enplaned or deplaned. In the case of international air cargo, this could be hundreds or even thousands of miles. For example, a shipment from New York to Indonesia, say, might be trucked to Los Angeles International Airport to be put on a trans-Pacific flight, since the cost of trucking the shipment across the country is so much lower than the cost of flying it. While the truck trip takes a couple of days longer than moving it by air, this extra time may be considered a worthwhile trade-off for the cost savings involved. Road feeder services are regularly scheduled airport-to-airport truck service between North American city pairs. A related issue can arise where an integrated carrier or freight forwarder has a sorting facility located at an airport. Some freight may not only be trucked to the sorting facility but also trucked from there to its final destination. Thus, some freight moving to and from the airport never gets put on an aircraft at all. In many cases, these truck movements carry both true air cargo (that is enplaned or deplaned at the airport) and other shipments that never get put on C H A P T E R   1 5 203  

204 Guidebook for Conducting Airport User Surveys and Other Customer Research an aircraft. These latter shipments may even be (and often are) transported using an air waybill (the document used to provide the necessary information about an air cargo shipment, such as its weight, shipper, consignee, and destination). Therefore, in order to undertake these types of studies, airports will need access to additional sources of air cargo data. In some cases, these data can be obtained from public or commercial sources (as discussed further in the following section). In other cases, it will be necessary to undertake specific surveys or similar data-collection activities to generate the needed data. With the growth of e-commerce, the aviation industry has been witnessing the most signifi- cant change in cargo transport since the rise of the integrated carriers (such as FedEx and UPS) that provide door-to-door service for both express (small high-priority shipments) and heavier air freight. The complex air cargo system includes (in addition to airports and shippers) airlines, cargo consolidators and handling companies, trucking companies, freight forwarders, postal authorities, and regulatory agencies such as the TSA and customs services in each country. 15.2 Purpose of the Study and Data to Collect The purpose of collecting air cargo data includes supporting the development of forecasts of likely future air cargo activity and identifying future airport airside and landside facility and infrastructure requirements. Air cargo activity may produce a significant amount of truck and other vehicle traffic on airport roadways and local roads and highways in the airport vicinity. Therefore, another purpose is to identify the relationship between the weight of air cargo and commodities shipped and the potential effect the resulting truck traffic may have on local and regional traffic patterns and routes. Tracking the amount of air cargo moved by truck can help determine leakage of air cargo activity from local to more distant airports. Other reasons for collecting air cargo data include assessing service levels provided by air cargo firms and determining needs for cargo air services, including types of carrier and routes. An important consideration in collecting air cargo data is that, while air cargo activity is typically measured by weight, the density of the cargo can vary widely. Therefore, the volume occupied by a given weight of air cargo can also vary widely, resulting in aircraft holds filling up before they reach the limit of the weight that the aircraft can carry. Similarly, the number and type of trucks needed to move a given weight of air cargo can also vary with the mix of the com- modities transported. Certain commodities require specialized ground transportation, which affects the number of vehicle trips needed. High-value cargo (e.g., diamonds, jewelry, and artwork) involves significant security considerations, while live animals will need special handling. 15.2.1 Factors Affecting the Collection of Air Cargo Data While the need for detailed air cargo data is clear, the data collection can be challenging because much of the air cargo information is considered confidential. The air cargo industry is extremely competitive, and customer-specific information is considered sensitive. Identifying the commodities, weight, and value of air cargo handled at an airport is the starting point for facility planning purposes, but other information about the air cargo activity at an airport is also required, such as the volume of trucks and other vehicles moving cargo to and from the airport; the sectors of the local, regional, and wider economy that are dependent on air cargo; the locations of freight forwarders and consolidators using the airport as well as the proportions of the cargo they handle; and the distribution of shipment origins and destinations. There are two different business models for air cargo airlines: integrated carriers (also referred to as integrators) that handle shipments from shipper to consignee, and traditional

Air Cargo Surveys 205 airport-to-airport airlines that depend on freight forwarders to pick up and deliver the ship- ments. The integrated carriers obviously have complete information on the air cargo they are handling. However, the airport-to-airport airlines may not know the full details of what is in a consolidated shipment moving between a freight forwarder’s facility at the shipment origin and another at the destination. Many airports worldwide have begun to develop air cargo community systems, which establish an integrated information technology infrastructure to facilitate the handling and movement of air cargo to and from the airport (Phillips and Tian 2019). One such system was established at Atlanta Hartsfield–Jackson International Airport in 2019 (Harry 2019). In principle, such systems could provide a valuable source of data on air cargo at the airport in question, although difficult issues of access to the data and what they can be used for would need to be addressed. 15.2.2 System Components International—and to some extent domestic—air cargo flows are influenced by the size of the aircraft that are operated in different markets from a given airport. Air cargo is often trans- ported in the belly hold of passenger aircraft (belly cargo), and the passenger volume will affect both the size of aircraft used in a market and the amount of space that is available in the aircraft to transport cargo. Air cargo is often consolidated over a wide geographic area and trucked to an international gateway or major-hub airport to be flown to its destination. Freight forwarders and consolidators play a key role in the air cargo system, picking up shipments from the original shipper, sorting and consolidating the individual shipments for transfer to the air carrier, receiving and breaking down inbound shipments, and delivering them to the consignee. As a result, the weight and volume of air cargo moved by truck between air freight forwarders and consolidators and the airports often bear little relationship to the weight and volume of individual shipments. In recent years, the air cargo industry has been steadily moving from a paper-based system of air waybills to electronic air waybills, as documented by the International Air Transport Association (IATA) (International Air Transport Association n.d.). However, thus far, this has been of little use for understanding air cargo flows at a given airport for two reasons. The first is that the information on individual air waybills is regarded as confidential, and the industry has not yet developed ways to produce de-identified summary data in a form that meets airport needs for planning and operational information. (There is a proprietary database operated by IATA based on the Cargo Accounts Settlement System that can provide detailed data on air cargo goods movement, but it is expensive to use and can withhold data in small markets due to privacy concerns.) The second arises from the consolida- tion activities of the freight forwarders, consolidators, and integrated carriers, which can result in a single truck movement carrying many individual shipments. The situation is somewhat better in the case of international air cargo because the U.S. Customs and Border Protection agency collects detailed data on all international shipments for customs purposes, and summary data are available for exports and imports at the level of individual ports of entry by commodity, weight, value, mode of transportation, and shipment origins or destinations within the United States by broad geographic areas. Of course, the firms and establishments shipping and receiving air cargo are the end users of the air cargo system. It would be useful for airports to establish and maintain an inventory of industries, manufacturers, retailers, and other firms located within the service area of the airport that receive or ship significant amounts of air cargo to better understand the market served by the airport. However, assembling and maintaining such an inventory is easier said than done and would require a sustained effort to integrate information from multiple sources, some of which are discussed in the following.

206 Guidebook for Conducting Airport User Surveys and Other Customer Research 15.2.3 Data Sources To obtain a more detailed profile of the air cargo moving through an airport, the total weights of enplaned and deplaned cargo reported by air carriers can be supplemented by data obtained directly or indirectly from air cargo surveys or detailed import and export data, and in some cases expanded using other types of data. There are three broad categories of such data: • Publicly available data collected or assembled at a national level but allowing more detailed analysis at the level of individual airports • Data available on a subscription or fee basis from commercial data providers that are derived in part from publicly available data but enhanced using additional public and industry data and organized to be easier to use and analyze than if working directly with the public data • Local surveys undertaken as part of an airport-specific, regional, or state air cargo study Publicly Available Data The two primary public data sources are the import and export trade data collected by the U.S. Customs Service and available through the U.S. Census Bureau and the Commodity Flow Survey (CFS) undertaken every 5 years by the U.S. Census Bureau and the Bureau of Transpor- tation Statistics of the U.S. Department of Transportation. A third public data source, derived in part from the CFS, is the Freight Analysis Framework (FAF), which is produced and main- tained by the FHWA in partnership with the Bureau of Transportation Statistics. These three data sources can be accessed at no cost, although access to detailed import and export trade data requires registering for a free account. Import and export trade data. These data (https://www.census.gov/foreign-trade/data/ index.html) are monthly data on imports and exports by value, weight, commodity, mode of transportation for import or export, and port of import or export, as well as other details. Origins of exports are reported at the zip-code level and aggregated to generate export data at the metropolitan-area level, while destinations of imports are available at the state level. There is no information about the mode of transportation between the export origin and the port of export or between the port of import and the import destination. Commodity Flow Survey. These data provide shipment-level data derived from a national survey of shippers conducted every 5 years (https://www.census.gov/programs-surveys/cfs.html). At the time of writing, the most recent survey was undertaken in 2017, and the results became available between July and September 2020. The survey covers almost 6 million shipments by approximately 60,000 responding establishments. Each shipment record includes the shipment origin and destination at the state and metropolitan area level, the industry classification of the shipper, the type of commodity, the mode of transportation, the value of the shipment, the weight of the shipment, the routed distance between the shipment origin and U.S. destination, whether the shipment was an export and if so, the export destination, and whether the shipment used temperature control or was a hazardous material. For each shipment, the survey record includes a weighting factor to expand the data to all U.S. shipments in 2017. A public-use file is available on the Census Bureau website and provides each survey record. It should be noted that all survey respondents decided what they considered a shipment. This could vary from a single package to an entire truckload. There is no information in the survey on how many ship- ments were transported together on a single truck, rail car, vessel, or aircraft. Freight Analysis Framework. The FAF (https://ops.fhwa.dot.gov/freight/freight_analysis/faf/) integrates data from a variety of sources, including the CFS, to create a comprehensive profile of freight movement among states and between major metropolitan areas by all modes of transportation. The FAF data expands on the CFS by including shipments for industry sectors

Air Cargo Surveys 207 omitted from the CFS. Although the underlying data are primarily shipment data, the FAF includes procedures for estimating truck flows on the highway network. The data include tonnage, value, ton-miles, commodity type, and mode. At the time of writing, the most recent version of the FAF (FAF4) was based on the 2012 CFS, but an updated version based on the 2017 CFS was expected in 2020. However, the availability of public air cargo data is much more limited than data on air pas- senger travel, where detailed itineraries of 10% of all air trips (at least in the United States), including the fares paid and airlines used, are publicly available online from the U.S. Department of Transportation Bureau of Transportation Statistics. Air passenger data are also reported on a more consistent and timely basis than are some of the public air cargo data sources. Commercial Data Sources Two commercial data sources that are widely used for freight studies are the WISERTrade data produced by the World Institute for Strategic Economic Research and the Transearch tool available from IHS Markit. As the name implies, the WISERTrade data (http://www.wisertrade.org/home/portal/index.jsp) provide international trade data as well as online analysis capabilities. These capabilities include a number of different products that produce customized reports that address different aspects of trade data with extensive use of graphics. The Transearch tool (https://ihsmarkit.com/products/transearch-freight-transportation- research.html) is designed to predict U.S. freight flows over a 30-year time horizon by origin, destination, commodity, and transportation mode. It also includes cross-border freight flows. Data are available for outbound, inbound, intra, and through shipments at the county, Bureau of Economic Analysis region, and state level. The data provide weight, value, and units of shipment for over 340 commodities. A 2016 study for the Florida Department of Transportation compared the Transearch and FAF data (RS&H, Inc. 2016), although it should be noted that both data sources have been enhanced since the study, so details of the comparison may no longer be current. Local Air Cargo Surveys Where more specific information is needed than can be obtained from the foregoing data sources, or to corroborate information inferred from those data sources, it will be necessary to undertake local surveys or other data-collection efforts, such as structured interviews or observation counts. In particular, the foregoing data sources are primarily shipment data, so translating weights of air cargo to truck or other vehicle movement will require additional information or assumptions. Also, the geographic resolution of the foregoing data will not allow the identification of any intermediate sortation of shipments, much less the location where that occurred. Thus, the data are unable to distinguish between a shipment trucked directly from its origin to its final destination and one trucked to an airport sorting facility (or any other sorting facility for that matter), sorted, and then trucked to its final destination. Air cargo surveys are typically conducted as part of a larger study of future air cargo facility or infrastructure needs, environmental documentation of planned air cargo facilities, or airport economic impact studies. Surveys and stakeholder interviews are also undertaken to supplement publicly available data by providing details on air cargo business practices, market dynamics, and factors that drive or affect cargo-related decisions. They can also provide information on expected trends that cannot be identified from historical data or through secondary research using industry publications and literature. The organizations surveyed will generally vary depending on the scope of the study but could include the air carriers, particularly the integrated

208 Guidebook for Conducting Airport User Surveys and Other Customer Research air cargo carriers, freight forwarders, and local shippers. The scope of the study may extend beyond a specific airport to the larger region or even the state. An air cargo planning study for the Phoenix region (InterVISTAS Consulting Group 2014) included an online survey of ship- pers in Arizona and the Mexican state of Sonora, adjacent to the U.S. border between Arizona and Mexico. The questionnaire for the survey of Arizona shippers is included in Appendix J. Shippers in Sonora proved unwilling to participate in the survey, and information regarding their potential use of Phoenix Sky Harbor International Airport for air cargo shipments was obtained through structured interviews with experts involved in trade between Sonora and the United States. Surveys may also be required in order to identify the extent of existing on- and off-airport facilities, such as available warehouse space, refrigerated storage capacity, the area available to park aircraft or trucks, and the number of parking positions for nose-loading aircraft. Several ACRP reports and publications have included examples of air cargo surveys. ACRP WebResource 1: Aligning Community Expectations with Airport Roles (Ward et al. 2017) includes an Aviation Toolkit with a section on air cargo that includes a sample air cargo shipper question- naire based on a logistics survey that was developed for the Great Falls Regional Airport Air Cargo Study in 2011. ACRP Web-Only Document 20: Guidebook for Estimating the Economic Impact of Air Cargo Operations at Airports (Balducci et al. 2014) includes an appendix with sample surveys of air carriers, freight forwarders, and shippers that were used in the case studies conducted as part of the research. However, the discussion in the reports indicates that responses to the surveys were limited. ACRP Web-Only Document 24: Air Cargo Facility Planning and Development—Final Report. (Maynard et al. 2015) includes an appendix that contains questionnaires for surveys of airport planning department staff, air carriers, freight forwarders, and other businesses involved in air cargo. Although the primary purpose of the surveys was to gather information on the facilities used by the respondents, the surveys included questions about the amount and characteristics of the air cargo handled by the respondents. The wording of these questions may be helpful to those designing a survey to obtain similar information about the air cargo activity at a specific airport. It is worth noting that the survey distributed to 265 air freight forwarders received less than a 5% response. Because of the competitive environment and sensitivity of many of the issues involved, interviews of freight forwarders have been found to be more effective than surveys, although these can follow a similar structure to a survey. 15.3 Survey Methodology Surveys of firms involved in air cargo will generally use the same approaches as those of other airport tenants described in Chapter 12 or local area businesses described in Chapter 14. Although obtaining appropriate contact information for these firms should be relatively straight- forward, getting their agreement to participate in a survey is an entirely different matter, and experience suggests that response rates will be quite limited. It can also be quite challenging to identify the person in an organization who is able to provide the required information. Not only is there likely to be a reluctance to provide what is considered proprietary data, but there may also be a reluctance to spend time on something that is viewed as having no direct benefit to the firm. This is likely to be a particular concern if the survey requests information that may take some time to assemble. It can be helpful to have the local community organizations involved in air cargo participate in promoting the survey or interviews in advance of the initial outreach to individual firms (as discussed further in Section 15.6).

Air Cargo Surveys 209 An alternative method for obtaining information on air cargo activities is to undertake stake- holder interviews. Although air carriers, shippers, and forwarders may be reluctant to release detailed information on air cargo shipments or cargo activity in response to a formal survey, it is possible to address the issues of interest in the course of an interview. Using the survey purpose as a base, a series of questions can form a structured interview to be conducted with all or selected air cargo operators at an airport. The cost and duration of such a survey would depend on the number of interviews to be conducted. In order to gather information on truck activity at airport cargo facilities, it may be possible to conduct driver interviews while the drivers are waiting at the cargo facility. This survey method was adopted as part of an extensive groundside survey performed at Toronto Pearson International Airport in 2005 (Cripwell and Turpin 2006), and the questionnaire used is included in Appendix J. The numbers of vehicle movements by category can be collected through observational studies, such as a truck count. An example of a field data sheet for a truck count undertaken as part of the groundside survey at Toronto Pearson International Airport is included in Appendix J. Such counts will not provide the detailed data on loaded weight and origin and destination, but they could provide information on the volume of truck movements, which can then be combined with other data to develop relationships between the weight of air cargo moving through the airport and the volume of truck movements. A potential enhancement would be to use cell- phone tracking data (as discussed in Chapter 9) to identify the movement patterns of trucks to and from the airport since the drivers will all be carrying cell phones, and the air cargo terminals are usually sufficiently separated from other airport facilities to allow the identification of cell-phone users who visit the cargo terminals. This will of course also identify cell phones carried by employees at the cargo terminals. However, their trips to and from the airport will usually be separated by the duration of their shift, in contrast to the truck drivers, who spend much less time at the terminal. 15.4 Sampling Approach The types of organizations that may be included in a survey include integrated and traditional air carries, freight forwarders, and key shippers. Each of these will require a different survey approach and questionnaire. Since there are generally a relatively small number of air carriers and other firms involved in air cargo activities at an airport, surveys will usually be sent to all the relevant organizations. At major airports in large metropolitan areas, where there is likely to be a large number of freight forwarders and consolidators using the airport, it may be necessary to select a sample to survey. Where the airport has some idea of the relative size of the different firms, such as from local business data, the sample should be chosen so that half the sample includes the largest firms by size, with smaller firms being selected in proportion to their size. If the airport has no information about the relative size of each firm, they can be listed in some logical order, such as by distance from the airport, and then selected randomly from the list. A similar approach would be used in the case of in-depth interviews, although the size of the sample is likely to be much smaller due to the cost of conducting the interviews. 15.5 Questionnaire Wording and Length The questions asked in any survey of air cargo firms or shippers obviously need to be tailored to the information sought. Questions should be worded in a way that is clear to the potential respondents in the industry, who may know relatively little about airport facilities planning. It is therefore important to use terminology that they are familiar with. This not only helps

210 Guidebook for Conducting Airport User Surveys and Other Customer Research ensure that the questions are answered correctly but indicates to the respondents that the designer of the survey is knowledgeable about the industry, which may help encourage them to respond. A particular difficulty arises with asking questions about commodities that are handled. Although there are standard commodity codes, it is unreasonable to expect survey respondents to refer to an extensive list of commodity codes in order to answer a question. Although it is more work, it is better to let respondents express commodities in their own words and then code these responses later. Obviously, there are a huge number of different commodities that are shipped by air cargo. It is therefore common to simply ask what are the top five or ten commodi- ties handled by the respondent’s firm by weight, as well as the percentage of the total air cargo tonnage handled that each of these represent. These estimates are likely to be fairly approximate but should be useful to compare the reported share of these commodities to those from more comprehensive shipment data such as the CFS or trade data. There is no need to ask air carriers to provide the total weight of air cargo they handle at a given airport, since those data are available from regular traffic reports they provide to the airport. How- ever, in the case of freight forwarders or other cargo handling firms, it is important to ask about the total weight of air cargo handled each month (or in the previous year) to determine their relative share of the air cargo activity at the airport. Any questions about air cargo commodities or the weight handled should distinguish between inbound and outbound cargo since direction flows are often quite different, both in terms of the amount and the composition of the cargo. In the case of freight forwarders, it is important that questions clearly refer to air cargo, since those firms may also handle cargo that is transported by other modes. It is also impor- tant to emphasize that information provided will be kept confidential and that the data will be aggregated so that the sources of the data cannot be identified. At some airports, the cargo terminals are operated by the air carriers, but at other airports (or at other terminals at the same airport), they may be operated by ground handling firms that receive shipments from freight forwarders or directly from shippers and transfer them to the air carrier. It is therefore important to tailor the questions to the specific role performed by the respondent and to ensure that all the relevant parties in the air cargo logistics chain are included in the survey. Questions that may be useful to ask air carriers include: • About what percentage of your inbound air cargo by weight is transferred to outbound flights? • What percentage of this transferred air cargo is transferred directly tail-to-tail without passing through the cargo terminal or other sorting facilities? • What are the top five commodities by weight in your inbound air cargo, and what percentage of the total air cargo does each constitute? • What are the top five commodities by weight in your outbound air cargo, and what percentage of the total air cargo does each constitute? Questions that may be useful to ask integrated air carriers include: • About what percentage of the cargo that arrives at your cargo terminal or sorting facility by truck is transported to its next or final destination by truck? • About what percentage of your total outbound air cargo by weight is composed of small packages weighing under 5 pounds? • About what percentage of your total outbound air cargo by weight is for international final destinations, even if initially it is transported to a domestic sorting facility? • What is the trend of e-commerce transported by air at the airport?

Air Cargo Surveys 211 Questions that may be useful to ask freight forwarders include: • On average per week, what is the weight of your outbound shipments by all modes? • About what percentage of your outbound shipments by weight are sent by air cargo? • On average per week, what is the weight of your inbound shipments by air cargo? • On average per week, how many truck trips do you make to the airport? • What are your policies regarding which gateway airport to use for international shipments? Questions that may be useful to ask shippers include: • On average per week, what is the weight of your outbound shipments by all modes? • About what percentage of your outbound shipments by weight are sent by air cargo? • About what percentage of your outbound air cargo shipments by weight are shipped using integrated carriers? • About what percentage of your outbound air cargo shipments by weight are shipped using freight forwarders? • On average per week, what is the weight of your inbound shipments by all modes? • About what percentage of your inbound shipments by weight are transported by air for at least part of their route? As noted in the following, the survey should be as short as possible since the potential respon- dents will usually not have a lot of time to devote to the survey. If there are certain issues that only apply to some respondents, it is desirable to tailor the questionnaire to exclude those questions from respondents to which they do not apply. This is simple to do with an online survey. In the case of a printed survey, it is better to have several different questionnaires that are sent to the appropriate respondents rather than having a large number of question skips that make the survey appear longer than it is. 15.6 Measures to Obtain and Enhance Responses Air carriers and other firms involved in air cargo are likely to be reluctant to respond to a survey or take the time to participate in a structured interview, and even more reluctant to provide detailed information about their operations. Therefore, it is imperative to clearly explain why their participation is needed and to assure them that any information they provide will be treated in confidence and that the results of the survey or interviews will be presented in aggregate form that will not allow the information from any one respondent to be identified. Where studies are conducted by consultants for an airport or other local agency, it is important for the study sponsor to stress that the studies are being conducted on their behalf and to encour- age the participation of the local air cargo community. It is desirable that the initial contact include a cover letter from the airport director or similar official that stresses that the study is important for the future of the airport and requests the assistance of the organization. If possible, the letter should point out the benefits of the study to the organization being contacted. Where there are industry organizations, such as a local air freight forwarders association or economic development agencies, representing segments of the air cargo industry at a particular airport, it would be helpful to solicit their assistance in promoting a survey. Firms such as air freight forwarders may be more likely to respond to a request from an association of which they are a member than from the airport management. As with other surveys, keeping the survey questionnaire or interview short and to the point will help encourage a positive response. If fairly detailed information is needed that an inter- view respondent would not necessarily have readily to hand, it may be necessary to leave a

212 Guidebook for Conducting Airport User Surveys and Other Customer Research questionnaire that the respondent can complete at a convenient time after the interview and return by mail or email. It is usually helpful to state that the results of the survey will be shared with the participants (if the survey sponsor allows this). Experience suggests that it is generally most effective to employ a balanced combination of surveys and interviews, depending on the study budget and urgency. Interviews usually yield extremely valuable information. When a significant proportion of those interviewed provide consistent responses, it is likely that this reflects the essence of a particular issue or the true situation.