Freight Transportation Surveys (2011)

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.

38 CHAPTER FOUR CASE STUDIES INTRODUCTION This chapter presents case studies for several types of sur- veys. The presentation complements the findings of the practitioners’ survey that were described in chapter three with details of actual surveys and their characteristics. The details are taken from material provided by practitioners and from the literature. In addition, the last section presents a comparison of surveys from the literature. It should be noted that the surveys are presented in two ways, depending on the type of survey and on the availability of source material. Case studies are presented individually for some survey types, and in others surveys are combined in order to exemplify specific points. In all cases, for a given survey type each of the surveys included presents a different aspect or variation. ROADSIDE/INTERCEPT SURVEYS Statewide Truck Lanes Needs Identification Study— Georgia The Georgia Department of Transportation (GDOT) con- ducted a series of roadside origin-destination surveys in 2006 as part of a statewide study to identify the need for truck lanes. The surveys were conducted at eight weigh stations along high-volume truck Interstate highways. The locations were also selected to provide a geographic distribu- tion across the state, and to avoid duplication with a similar survey that was conducted just prior to the GDOT surveys by the Atlanta Regional Council (ARC) in and around the Atlanta metropolitan area. Surveys also were conducted at two gates to the Port of Savannah, given the high volume of truck traffic to and from that area. Twenty-four hour counts were conducted at the same time as the surveys. These were used as the basis for expanding the samples at each site. A total of 3,636 trucks was sampled at the 10 sites: these represented between 4.0% and 15.8% of the daily truck volumes at the weigh station sites and 14.9% and 18.3% at the Port of Savannah gates. The GDOT survey was based on the ARC survey in order to further coordinate efforts and facilitate the sharing of sur- vey results. Some modifications were made to address the specific requirements of the GDOT truck lanes study, and to capture data on commodity type and category. The collected data comprised— Vehicle information: number of axles, number of units, • truck configuration, trailer style, hazardous materials status, and state of registration. Trip information: origin, destination, frequency, roads • use, purpose, origin facility type, destination facility type, load status, commodity type, routing decision- maker (who chooses the route), and load type/vehicle ownership. A shorter questionnaire was used for the two Port of Savannah surveys to expedite data collection and to allow for localized differences in truck origin-destination patterns. Some port-specific questions also were included. Of particular note is the use of a PDA for both surveys. The PDA allowed for direct entry of the data into an electronic database and the automatic data coding. This reduced costs by avoiding manual data entry and coding and increased accuracy by reducing the likelihood of key entry errors after the fact. The survey had four types of questions: Choose one (e.g., the location at which the survey was • conducted) Choose multiple (e.g., trailer style)• Text entry (e.g., description of the location at which the • trip originated) Numeric entry (e.g., number of axles).• Figure 2 presents example of these questions. Note that some of the questions require input from the driver and oth- ers are based only on the interviewer’s observations (4). Portland Freight Data Collection Program—Port of Portland et al. In 2005, a comprehensive freight data collection program was initiated in the metropolitan Portland, Oregon, region. The data were collected on behalf of the Port of Portland, the Oregon Department of Transportation, the Washington

39 State Department of Transportation, and Portland Metro (the region’s MPO). Roadside intercept surveys were conducted at nine external gateways (highways) to the region, rest areas, weigh stations, and truck inspection locations. Truck counts were conducted in order to allow for expansion of the data. FIGURE 2 PDA questions—Georgia DOT. [Source: Statewide Truck Lanes Needs Identification Study, Technical Memorandum 1: Data Collection (4).] The surveys collected data on truck type, cargo type, cargo weight, trip origin and destination, facility type at ori- gin and destination, location of the truck’s home base, activ- ity at the origin and destination (i.e., pick up, delivery, or return to base), carrier type, and whether the vehicle had a WIM transponder. The survey also asked about the highways used during the sampled trip, including the interchanges and access roads between the highways and the origin and desti- nation (5, 6). Figure 3 shows the survey form. Figure 4 shows a variation of the form that was used for roadside interviews at entrances to key freight terminals in the region. The form was administered to drivers at marine and rail terminals, tank farms, and pipeline terminals. Commercial Vehicle Survey—Ontario, Canada The Ministry of Transportation of Ontario, Canada, has conducted a large-scale truck roadside origin-destination survey at approximately 5-year intervals since 1978. The surveys are conducted on provincial highways and at inspec- tion stations, rest stops, and interprovincial and international (U.S.) border crossings. They focus on inter-urban trips. The surveys have collected information on carrier information; vehicle type, number of axles, and weight; trip origin and destination; and commodity type and value (9). The survey focuses on heavy trucks, so other commercial vehicles are not captured. Intraurban travel also is not captured. In 1999–2001, the National Roadside Survey (NRS) was conducted across Canada. This nationwide truck survey comprised the aforementioned Ontario survey, as well as surveys in other provinces and at U.S. border crossings. A selection of intermodal terminals also was surveyed. The NRS was a cooperative effort among the federal and pro- vincial ministries of transportation as well as the FHWA (with respect to the border crossings). Approximately 65,000 trucks were surveyed at 238 data collection sites. The NRS provides a proven example of a nationwide pro- gram for collecting inter-urban truck traffic, one that could be followed in the United States. A consistent survey form, a common general surveying procedure, and common classifi- cations and terms were developed and used across the coun- try. However, although driver interviews were conducted by local staff who were familiar with local travel and vehicle characteristics, some variations in the data collection were observed. Different groups having different objectives (e.g., enforcement, planning, and policy development) gathered the data. This reflected each province’s interest in participat- ing in the NRS. However, as a result of these different inter- ests, there was some variation in the focus of the interviews: some focused on vehicle weight and dimensions, which are important for enforcement, and others focused on trip details, which are important for planning. Although these differ- ences must be captured, local nuances must be captured in a “well-planned and consistent manner when national data for a wide range of uses are collected” (10). A further com- plication arose in the expansion of the 1999–2000 NRS data, whereby the provinces of Ontario and Québec and the fed- eral ministry of transportation each used a different method. This means that the resultant travel characteristics and trip tables may differ for the same location, depending on whose expansion method is used. Data from a 2005–2007 update to the NRS currently are being analyzed. Although direct data entry has been in place since the 1995 survey, this survey incorporated a GIS-based routing component that enabled the surveyor to confirm the route with the driver and, if necessary, to modify it in order to get an accurate profile of the highways used for the trip. The 1999–2001 NRS asked drivers to list the highways used in the trip: this information was used to confirm the route during the data processing stage (11). ROADSIDE/INTERCEPT SURVEYS—SPECIFIC TOPICS It is important to note that some roadside/intercept surveys, although following the general format of the aforementioned surveys, focus on specific geographical areas and issues. As a result, the types of questions may vary from the gen- eral format, with greater or lesser focus on specific types of data.

40 streets and freeways that were used for the trip. Traffic counts were collected at the same time. Approximately 10,000 surveys were distributed, and 3,300 of these (33%) were returned (13). A 2008 study of drayage at the Port of Houston gathered similar information to the aforementioned survey, but also asked about trip distance (average length of drayage haul), the frequency of trips on a typical day, location of “worst traffic” along the route, cause of delay, use of a toll facility, average wait time (idle time) entering the Port (Barbours Cut Terminal), and the actual wait time for the current trip. Demographic information and vehicle characteristics (age and total mileage) also were gathered (12). FIGURE 3 Roadside Intercept Truck Survey Form—Portland region. [Source: Roadside Intercept Survey (7).] For example, some studies have examined drayage activity at marine ports. Drayage refers to local movements between a port and its surrounding urban area (i.e., local transportation within the urban area). Because the dray trucks operate primar- ily in an urban environment, they can have a significant impact on congestion and air quality (12). A 2001 study surveyed truck drivers entering and exiting the Port of Los Angeles/Port of Long Beach. The survey solicited information about the truck type, the type of origin or destination (off-dock, intermodal facility, industrial facility or warehouse, another port terminal, other), the location of the origin or destination, and the specific

41 carriers to identify the amount of time late delivery is still considered to be “on time” (14). Figure 5 shows the results of this question. Ranking or identification of issues and prioritiza-• tion of solutions. The 2008 Austin Area Freight Transportation Study asked respondents to rank freight transportation challenges according to sever- ity on a scale of 1 to 5 (15). Bottlenecks and points of congestion in the freight • network. Challenges and opportunities concerning freight • transportation. Measure of the level of support for different types of • proposed initiatives. The 2008 Los Angeles SCAG (Southern California Association of Governments) Multi-County Goods Movement Action Plan services study asked respondents to indicate their level of sup- port for a long list of previously identified initiatives. Rankings ranged from 1 (no support) to 5 (highly sup- portive). The survey also provided a list of highway locations and asked respondents to identify which locations they think would benefit from a truck lane or an additional mixed flow lane. Respondents were then asked to rank their choices (16). Figure 6 is an excerpt from the question concerning potential initiatives. Figure 7 is an excerpt showing the question concern- ing truck lanes. Impacts of congestion and delays on business costs and • profit Freight infrastructure investment priorities• FIGURE 4 Gateway (Terminal) Intercept Truck Survey Form—Portland region. [Source: Task 4—Gate and Establishment Survey Plan Memorandum (8).] FOCUS AND STAKEHOLDER GROUP SURVEYS— FREIGHT STUDIES A number of organizations conducting freight transportation surveys or freight studies collect qualitative information alone, or in addition to, quantitative data. Qualitative surveys collect information about the “what,” the “why” or “why not,” and the “how do we make it better,” and they can be targeted to specific topics. The surveys may solicit factual information as well as attitudes. They are conducted by telephone, on paper, in person, or via the Internet. They may be conducted as part of focus or stakeholder group surveys. Although the results may be used in decision-making, they differ from “quantitative” surveys (e.g., origin-destination or commodity flow surveys) in that they are not necessarily statistically representative, commonly by design (as evidenced by the practice cited here); nor are they intended for modeling or forecasting. Several recent freight studies have collected a wide variety of qualitative information about freight transportation. Com- mon topics include: Type of business, primary products (for shippers), size, • location, modes of transportation used. Establishment operations questions (i.e., who makes • decisions about routing, mode, type of scheduling). A number of surveys collected information about opera- tions; the 2008 Washington Transportation Plan Update Freight Movement studied the nature and importance of on-time delivery. The survey asked shippers and

42 appreciated by stakeholders (17). The Kansas DOT Carrier interview guide included a question asking why some routes are chosen over others and whether routes are preplanned or chosen by the operator (18). Figure 8 shows sample open- ended questions from the Arizona Multimodal Freight Anal- ysis Study survey. There are three overall strategies for qualitative data gathering: FIGURE 5 “On-Time” delivery—Questionnaire tabulations—Washington State. [Source: Washington Transportation Plan Update Freight Movement (14).] The Arizona Multimodal Freight Analysis Study survey asks a number of questions about the importance of different factors to a shipper’s transportation decisions. The survey includes open-ended questions asking respondents to note what efforts would improve their experiences, or encour- age them to shift to another mode of transportation. These types of surveys attempt to gather anecdotal information about the strengths and weaknesses of the freight transpor- tation network and assess what investments would be most

43 a mix of question types. In addition to primarily qualita- tive surveys, some surveys were primarily quantitative with some qualitative questions included. Many surveys included an open-ended “comments” space where any feedback could be provided. More specifically than “surveys,” many agencies con- ducted specific stakeholder consultation interviews. These interviews tend to be more qualitative in nature and ask a FIGURE 6 Survey excerpt (Support for Potential Initiatives)—Southern California. [Source: Multi-County Goods Movement Action Plan: Technical Memorandum 2b: Public Outreach—Survey No. 2 Report (16).] Primarily qualitative surveys with little or no quantita-• tive data gathered, Primarily quantitative surveys with a small number of • qualitative questions, and Stakeholder interviews.• Many of the surveys reviewed were primarily qualitative; in these surveys, origin-destination, routing, and quantity of commodity are not addressed, although some surveys ask

44 Some agencies reported poor response rates. For the Ari- zona Multimodal Freight Analysis Study, the consultant team made more than 200 phone calls to shipper/receiver organi- zations but realized only 12 completed interviews (17). In addition to surveys of shippers, receivers, carriers, and other bodies directly involved in freight transportation, two other types of qualitative surveys are conducted: surveys of public agencies and public consultation surveys. For the Kansas DOT’s Statewide Freight Study, three surveys were administered: one for shipper/receivers, one for carriers, and one for public agencies. Questions were tailored to each group, but all asked about modes of choice and the perfor- mance of the freight transportation network (18). smaller number of respondents for their input on freight planning subjects, as discussed earlier. The 2008 Virginia Statewide Multimodal Freight Study included two types of interviews: scripted interviews with set questions and unscripted “free-ranging” interviews conducted by person- nel with experience in freight planning. Scripted interviews allowed for easy compilation of results, whereas unscripted interviews allowed for more detailed exploration of infor- mation that would not have been uncovered by the standard script (19). For the 2008 Atlanta Regional Freight Mobility Plan, a mix of surveys were used, with origin-destination surveys separate from stakeholder surveys and targeted interviews. The stakeholder survey was entirely quantita- tive, whereas targeted interviews were qualitative (20). FIGURE 7 Survey excerpt (Potential Truck Lane Locations)—Southern California. [Source: Multi-County Goods Movement Action Plan: Technical Memorandum 2b: Public Outreach— Survey No. 2 Report (16).]

45 survey, designed to be administered by the Metropolitan Wash- ington Council of Governments, solicited the type of quantita- tive information on a firm’s activities that can be gathered in an establishment survey (although, notably, vehicle surveys were not included) as well as “anecdotal” input regarding issues. The survey was intended to target “a sample subset” of area businesses that ship or receive a substantial amount of goods to, from, or within the region; trucking companies (including mail/package delivery firms); rail companies; air freight ser- vices; and shipping industry associations or representatives. However, it also was understood that this survey would not represent a “scientifically formulated survey” (22). FIGURE 8 Sample open-ended questions—Arizona. [Source: Arizona Multimodal Freight Analysis Study, Technical Memorandum #1, Analysis of Arizona’s Freight Dependent Industries (17).] Finally, several examples of surveys of “establishments” were found in the United States: however, it is important to dif- ferentiate the majority, which constituted attitudinal surveys on goods movement issues, from those that collected quanti- tative information using a statistically representative sample. The Establishment Surveys section presents case studies of the latter. An example of the former is a 2006 online and mail sur- vey of public and private establishments in Southern California (21). The survey solicited information regarding goods move- ment issues in the region, but it cannot be considered as either quantitative or as representing a statistically valid sample (nor was any intent to this effect claimed). A 2007 establishment

46 Figure 9 provides a sample question, which asks for driver perceptions regarding the quality of travel on freeway sec- tions, according to several factors. Questions on both the relative satisfaction of each factor and the importance of the factor are asked. A seven-point relative interval rating scale is used, where –3 is “least important/least satisfied,” 0 is “as important or as satisfied as others,” and +3 is “most important/ most satisfied.” The use of a seven-point scale “balances the level of detail in measurement of the respondents’ perceptions and the respondent burden and error.” The use of a “typical” interval-rating, ordinal or ranking scale was not considered to be appropriate. This was because a typical interval-rating scale question allows the respondent to give an equal impor- tance to all factors, thus making the distinction among factors difficult. An ordinal scale or ranking question does not allow mathematical interpretation of the survey responses, thereby restricting the applicability of various statistical analyses. The study’s authors noted the difficulty in gaining driver participation in the surveys. They first conducted the survey at the Florida Truck Driving Championship, where paper forms were distributed to drivers as they registered for the event. However, despite ongoing publicity during the event, only 11 of 148 surveys were returned. As a result, a second survey was conducted; this time, through the distribution by Florida DOT staff to truck drivers at four agricultural inspection stations. Four thousand mail-back questionnaires were distributed, of which 311 were returned (7.8% response rate). Another 300 mail-back truck manager surveys were sent to 60 Florida-based carriers (five surveys each): with follow-up telephone calls, 27 surveys were returned (9% response rate) (25). Finally, two 2008 studies in Washington are of interest. One study used online surveys of truck drivers who drive in that state and of trucking companies that operate in the state to gather information regarding the adequacy and availabil- ity of truck parking and services in the state, in light of fed- eral regulations that require minimum rest periods and of the need for short-term parking while drivers are waiting to make a delivery or pick up a load (26). The survey was publicized by e-mail among trucking associations and via the state DOT’s freight industry stakeholder e-mail list, in industry publications, and on trucking programs on satellite radio. No information is provided about the total population, and a sample was not drawn statistically. However, 473 driv- ers and 99 companies responded. The surveys generally were similar, and asked: Characteristics of the carrier/company type, including • normal areas of operation and types of vehicles used in the state, and frequency of trips Where trucks currently park, with a specific break-• down by rest and service area location FOCUS AND STAKEHOLDER GROUP SURVEYS— SPECIFIC TOPICS Another type of focus and stakeholder group surveys is not associated with statewide or regional freight studies but, rather, with specific topics. For example, a 2007 paper reported on an attitudinal survey that was conducted of 71 Georgia-based trucking firms regarding their opinions on a proposed truck-only toll lane scheme alongside Interstate highways in the Atlanta region, proposed to relieve conges- tion. The respondents included both small carriers (10 or fewer power units) and large carriers (more than 200 power units). The survey solicited information about the firm’s activities—notably, the geographic extent of its services, its frequency of use of the region’s highways, the most fre- quently used highways, the time of day of operations, loca- tions of the most severe congestion, use of alternate routes to avoid congestion, importance of delivery time, use of truck- only lanes, and the willingness to pay (23). A 2005 study of the acceptability by truck drivers of in- vehicle technological feedback systems to improve safety illustrates the use of focus groups and stakeholder surveys. (For example, an electronic sensor might monitor driver alertness.) Focus groups were used to collect qualitative information from 66 long-haul drivers in New England regarding attitudes toward technology and feedback. The results were used to develop a questionnaire, which then was administered to 198 long-haul truck drivers to provide quantitative information (24). More details on the efficacy of the focus group and stake- holder surveys are provided in a 2009 paper reported on the development of level of service (LOS) factors that are specific to the trucking community. This is in contrast to the com- monly used traffic engineering techniques, such as those of the Highway Capacity Manual, which develop a single LOS for all types of traffic combined. The development of specific LOS factors for the trucking community reflected the unique size and operating characteristics of large trucks: in turn, these factors could be used to assess how well the statewide road system meets the needs of freight transportation. The study sought input from two stakeholder groups: truck drivers (“the most important group concerning truck LOS, in that they are the ones who actually drive the trucks on the road”) and truck company managers (whose input was sought “so they could be compared with those of the truck drivers”). Perceptions on the “truck trip quality” were gathered from each group, first qualitatively through focus groups and then through the administration of stakeholder surveys to corroborate the perceptions. Two sets of forms were prepared in order to account for the differing back- ground characteristics; however, the analytical questions were identical.

47 (An initial sample frame for the latter included businesses that did not necessarily own commercial trucks; however, this resulted in a high number of ineligible [i.e., inapplicable] cases, and so the scope was narrowed in order to generate a usable response.) The survey was conducted as a telephone interview; however, respondents were offered the option of participating via an Internet survey. Of the 1,750 completed responses, 1,513 participated through a short telephone inter- view and 237 respondents completed a web survey for an overall response rate of 69.6%. Up to 10 contacts were made with each sampled organization, to encourage participation. The telephone interviews ranged between 5 and 50 minutes in length, with an average duration of 20.5 minutes. A com- puter-assisted telephone interview (CATI) was used to allow data validation to be conducted during the interview (e.g., by prompting the interviewer to ask about missing information as responses were processed) (27). FIGURE 9 Perception of truck trip quality on freeways—Florida. [Source: B. Ko et al. (2009) (25).] Parking requirements (size, width, height)• When and for how long they park• What services they use when they park• What improvements should be made at truck parking • areas Where it makes most sense to increase available truck • parking Where additional truck parking areas should be • developed. The second study solicited information on the economic impacts to Washington State shippers and carriers of weather- related closures to I-5 and I-90 (key north-south and east-west corridors, respectively) during the winter of 2007–2008. The survey contacted 2,758 establishments sampled from trucking industries and from businesses active in seven “freight depen- dent” industrial sectors that also owned commercial trucks.

48 “Form B” covered trips related to business. The form solicited information about the employee’s trip or “tour,” including the vehicle type, origin and destination location of each leg of the day’s activities, arrival and departure times, activity conducted at each stop, and value of service pro- vided. The list of activities accounted for both commercial services and the movement of goods. The list comprised business meeting, sales/marketing visit, provision of ser- vices, break/meal, vehicle service/refueling, pick up of material or equipment, drop off of material or equipment, return to work, return to home, and other. Respondents were asked to check all that applied (28). Figure 10 provides a sample of this form. FIGURE 10 Employee business trip log—Ohio. [Source: 2003–2004 Ohio Statewide General Establishment Survey, Technical Memorandum (28).] Commercial Vehicle Trip Diary—United Kingdom The United Kingdom’s Continuing Survey of Roads Goods Transport (CSRGT) is a long-standing survey of U.K.- registered heavy goods vehicles that operate on the country’s The survey solicited information about the December 2007 I-5 closure at Centralia, Washington, and the January 2008 I-90 closure at Snoqualmie Pass, including: Extent, if any, of “negative” economic impact to the • respondent’s business Types of response to the closure (cancel, postpone, or • reroute truck shipments; other) Number of truck shipments impacted by the closure• Characteristics of rerouted shipments:• Detour route used – Number of trucks using each detour route – Increases in journey time – Additional direct costs – Why this route was selected (available, less severe – weather, nearest alternative, safest, other) Source of information about detour routes (highway – posted radio channel, company dispatch or contact, truck weigh station, Washington DOT freight elec- tronic mailing list, media, State Patrol, other) Total additional costs incurred by the respondent dur-• ing the closure (ranges), broken down by type of cost [overtime hours and drivers’ expenses, additional fuel, acquisition of additional equipment, higher rates for expedited or guaranteed services, additional inventory and storage, costs of damaged goods (e.g., perishables), other] Duration of incurrence of additional costs• Total additional costs after the closure was reopened, • if any Whether any of additional costs were recovered, bro-• ken down by type of cost Whether the respondent planned to make any capital or • operating investments that year to prevent future busi- ness disruptions owing to potential closures at the same location; and the estimated value of these investments Value of lost sales because of the closure, and associ-• ated characteristics (how estimate was derived, per- centage of total annual revenues, duration of impact) Total annual sales (revenue) for the respondent’s busi-• ness in Washington. It should be noted that the study reports the survey results but does not analyze or interpret them. The report includes the CATI and web surveys. COMMERCIAL TRIP DIARY SURVEYS Employee Business Trip Log—Ohio The 2003–2004 Ohio Statewide General Establishment Sur- vey included a commercial trip diary survey. The survey was administered to employees of sampled establishments (see also Establishment Survey—Ohio for a discussion of these establishment surveys).

49 Survey results are tabulated each quarter. The activi- ties of each sampled vehicle are recorded for 1 week. If the vehicle has been sold or scrapped, then no further details are recorded on the form. The survey asks for information about the owner’s business; the characteristics of the vehicle, including the number and positioning of axles; and the type of work “normally” carried out by the vehicle. A question on “respondent burden” asks how many minutes it took for the respondent to complete the form. Finally, a 7-day trip log is to be completed: for each trip on each day, the origin and destination, type of goods carried, weight of goods car- ried at the beginning of the journey, distance traveled empty and loaded (miles), information on “split” delivery/pick-up journeys (i.e., tours in which both pick-ups and deliveries are made), and the frequency of similar journeys, with similar loads, made during the same day. Figures 12–14 present the three pages of the survey form. Multiday Urban Vehicle Trip Diary—United Kingdom An example of a multiday vehicle trip diary is provided by the 2001 Birmingham, Basingstoke, and Norwich, U.K. survey of freight distributors. That survey included a 3-day vehicle trip diary survey, which was administered to drivers of the vehicles of the seven surveyed freight distributors. The survey gathered four types of information (35): Vehicle activity: date of the delivery round (i.e., tour), • depot departure and return times, odometer readings at the start and end of the trip, vehicle age and type, and the commodity type on board the vehicle when it left the depot. Stop characteristics (for each stop): arrival and depar-• ture times, distance traveled between stops, activity type (delivery, pick-up, both, other), name and address of the stop, quantity of goods delivered or picked up, where the vehicle was parked, time taken for each delivery or pick-up, and any problems experienced en route. Vehicle characteristics: age, make and type, exter-• nal size, internal load space (volume), gross vehicle weight, maximum payload, fuel type, and vehicle fuel consumption rate. roads. The ongoing survey has been conducted for several decades in the United Kingdom and Northern Ireland. Sam- ples are drawn from driver and vehicle registries. The sampled driver or operator of each vehicle completes a log for each trip that is made over the course of one week. Participation, once selected for the sample, is mandatory. The CSRGT is one of several programs that collect infor- mation on goods movement. One result is that inconsisten- cies have been reported in some tabulations; notably, in vehicle activity (vehicle-kilometers traveled), and there was evidence of under-reporting from the CSRGT because of apparent record-keeping inaccuracies (paper forms are used by the drivers), the need for drivers to distinguish domestic legs of a multicountry trip, apparent difficulty in the ability of automatic traffic recorders (which are used to expand the data) to distinguish between some large and small vehicles (the CSRGT includes only heavy vehicles over 3.5 tonnes gross vehicle weight), the exclusion of certain types of vehi- cles from the CSRGT sample (including cranes, fire engines, and large army vehicles), and a lack of precision in tachno- graph data, which some respondents were using to record their mileage (29). Figure 11 provides a sample of the form. Goods Vehicle Trip Diary—Ireland The National Survey of Transport of Goods by Road is con- ducted by the Central Statistics Office of the Republic of Ireland. As with its U.K. counterpart, participation is man- datory. Approximately 30,000 goods vehicles are sampled per year. A random sample is drawn from official registries of all goods vehicles having an unladen weight of 2 tonnes or more. The sample is spread evenly throughout the year. Samples are selected weekly, and a questionnaire is mailed to the registered owner of the vehicle. Three strata are used for sampling, according to unladen weight (sampling rates are shown in parentheses): 2–5 tonnes (15% sample), 5–10 tonnes (50%), and 10 tonnes and over (90%). The strata sam- pling rates remain constant for each weekly draw. Because the rates have remained constant over several years, and because the fleet size has increased over time, the weekly sample size increases gradually (31). FIGURE 11 Commercial vehicle trip diary (extract)—United Kingdom. [Source: Department for Transport (30).]

50 classification counts. The surveys comprised telephone and in-person interviews of the proprietors of key freight-gener- ating facilities in the region, such as warehouses and manu- facturing facilities. The surveys gathered data on the facility type and its characteristics: number of employees, hours of operation, facility size, number of loading/unloading bays, whether the facility supported cross-docking capabilities, type (commodity) of products handled, volume of inbound/ outbound shipments, distribution of the shipments by time of day and by season, the most common origins and destina- tions of shipments, average payload weight, empty truck trip fractions, and primary highway access routes to and from the facility (and whether drivers were given routing instruc- tions) (8, 36). The in-person survey is illustrated in Figure 15. Note that the questions solicit information regarding “typical” freight activity. Respondents also were given the opportunity to comment on selected issues. FIGURE 12 Goods vehicle trip diary, first page—Ireland. [Source: National Survey of Transport of Goods by Roads (32).] Vehicle utilization (over the 3-day period): time idle • (empty) at the depot, time vehicle was out on pick-ups or deliveries, and time vehicle was loading/unloading or waiting at the depot, or during a driver rest period. The large number of stops made by some firms required the assistance of the researchers to accompany the drivers and fill out the survey forms themselves. ESTABLISHMENT SURVEYS Portland Freight Data Collection Program—Port of Portland et al. The Portland freight data program conducted establishment surveys in addition to roadside intercept surveys and vehicle

51 advance letter (which provided more information) was sent by facsimile or by e-mail. A survey date was assigned. The sur- vey forms, including instructions, then were sent by courier. “Form A” verified information regarding the establishment’s characteristics that had been collected during the recruitment. The survey also collected information regarding the types of goods shipped and/or services provided off-site, fleet vehicle information, and the composition of the company’s labor force (number of employees by industry and occupation clas- sification). “Form B” was an employee business trip log (see Employee Business Trip Log—Ohio earlier in this chapter) (28). “Form C” gathered data about outgoing goods shipments on the assigned survey date. The data included type of goods shipped; the quantity shipped; shipment destination; whether a courier, shipper, or common carrier was used (and the firm’s identity); type of vehicle used, company vehicle used (if any); activity/trip purpose; and vehicle occupancy. Figure 16 pro- vides a sample of Form C. FIGURE 13 Goods vehicle trip diary, second page—Ireland. [Source: National Survey of Transport of Goods by Roads (33).] Establishment Survey—Ohio An establishment survey was conducted as part of the Ohio statewide freight data collection initiative in 2003. This sur- vey covered both goods movement and commercial activ- ity. The survey complemented existing data on freight flows from the TRANSEARCH commercial database and from the U.S. Commodity Flow Survey: to this end, it focused on the types of establishments that are not generally cov- ered in these other sources. (See Commodity Flow Survey section later in this chapter.) Of an initial sample of 9,231 establishments, 593 returned completed survey forms: these represented a response rate of 13% of the sample and a 33% response rate of the 1,781 establishments that actually were recruited for the survey. The study comprised three separate surveys. Initial recruitment was conducted by telephone, following which an

52 tinations of the trucks, pick-up and drop-off activity at the warehouse, and seasonal and temporal variations in truck activity (4). Figure 17 presents the survey form. Establishment Surveys—Edmonton/Calgary, Canada The cities of Edmonton and Calgary, Canada, collaborated on the development of an urban goods model and on the under- lying data collection. The two cities, which are 180 miles apart, are the largest cities in the Province of Alberta and have approximately the same population (900,000–950,000 at the time of the surveys described here). The surveys were used in the development of micro-simulation goods and ser- vices models for the two cities (37). The surveys were applied in Calgary in 2000–2001 and in Edmonton in 2001–2002. Each combined an establishment survey with an origin-destination survey of truck drivers. FIGURE 14 Goods vehicle trip diary, third page—Ireland. [Source: National Survey of Transport of Goods by Roads (34).] Establishment Survey—Georgia The Georgia Statewide Truck Lanes Needs Identification Study (described in Statewide Truck Lanes Needs Identifi- cation Study—Georgia) conducted surveys at seven Savan- nah-area warehouses. The surveys were used to confirm that these facilities operated primarily as intermediate freight destinations between the Port of Savannah and points fur- ther inland, and secondarily as distribution centers for goods that are consumed locally. A mix of public and private ware- houses participated in the survey. Together, they accounted for approximately 200,000 trucks moving in and out of their facilities annually. The focus was on the movement of con- tainers (although allowances were made for other trucks); and the data were differentiated according to whether or not the containers were loaded or were empty. The warehouse operators provided information on their hours of operation, the numbers of trucks using the facility, the origins and des-

53 were deemed to be eligible. Of the contacted establish- ments, 13,792 were determined to be eligible, and of these 4,324 agreed to participate in the survey. The survey was expanded according to three independent variables: number of employees, industry category, and geographic location, using the total number of employees within each variable to determine the individual establishment expansion factors. The resultant average expansion factor for all establishments was 2.36 (38). The Calgary experience, which surveyed 3,411 establish- ments, is instructive insofar as developing the establishment sample is concerned. Difficulties with finding a workable sample database caused delays in the process and required that the sample be verified before the actual survey could begin. Samples ultimately were drawn from the Provincial Treasury ministry’s registry of businesses and from the city of Calgary’s City Business Tax database. For the city of Cal- FIGURE 15 Establishment survey—Portland. [Source: Task 4—Gate and Establishment Survey Plan Memorandum (8).] The establishment survey captured the activities of a signifi- cant sample of all business establishments in the respective region. Drivers of commercial vehicles leaving the estab- lishment then were surveyed regarding the specifics of their goods movement over one weekday. The surveys covered all establishments involved in the shipment of both goods and services, including transportation depots. These urban sur- veys were complemented by a roadside survey of trucks at an external cordon surrounding each city, to capture inter- urban goods movement to, from, and through each city. The surveys attempted to sample all types of businesses. In the Edmonton survey, 27,478 business establishments were contacted to ascertain their eligibility for participating in the establishment survey. Information also was collected regarding the number of employees, location, and industry category of the establishment. Establishments that produced either a product or a service that required transportation

54 delivered to the establishment. Even with the availability of complete or partial support (i.e., the interviewing contrac- tor was available to complete all or part of the survey for the respondent), 92% of the surveys required amendments, including correction of addresses and compilation of the required information from original sources. Respondents had operational differences that impacted the establishment and driver surveys: in particular, the inability to record mon- etary values of the shipments because product owners did not share this information with them (and the goods were insured by the producing company), value information was kept elsewhere (at a head office), or details could not be bro- ken out. More reporting units of measure than were expected also were received, and had to be reconciled. Consistency for service vehicle reporting was difficult depending, for example, on whether or not the service vehicle was parked at home (i.e., in which case, the first trip of the day may be that from home to the service site, or to work to pick up the service vehicle). Once an establishment had been recruited by telephone, a face-to-face visit followed, in order to obtain management buy-in and personalize the project (39). FIGURE 17 Establishment survey of warehouses—Georgia. [Source: Statewide Truck Lanes Needs Identification Study, Technical Memorandum 1: Data Collection (4).] gary, a sample of 49,354 companies, approximately 3.4% of the sample could not be reached because of incorrect tele- phone numbers (for initial contact and recruitment), the provision only of a facsimile number, or that the company no longer was in business. Another 25.1% did not qualify because they did not ship or used only the postal service or personal couriers, and a further 7.2% were duplicates. Of the remainder, 2.4% declined to participate. In total, 3,791 establishments, or 7.7% of the initial sample, pre-qualified. Of these, 3,150 or 6.4% were recruited, and 3,107 establish- ments ultimately provided usable data. Another 304 sur- veys were completed in the surrounding region, for a total of 3,411 surveyed establishments (39). As the city of Cal- gary’s experience indicates, the coverage and representation of the establishment survey depends on the availability of appropriate databases; much of the available data were erro- neous, duplicates, or otherwise unusable. Moreover, the use of telephone recruitment ensured that sampling quotas were met and that, once recruited, respondents’ participation was maximized. FIGURE 16 Outgoing goods shipment survey—Ohio. [Source: 2003–2004 Ohio Statewide General Establishment Survey, Technical Memorandum (28).] The challenges of reporting were noted in the Calgary survey. Following the telephone recruitment, packages were

55 Establishment Survey—San Bernardino Valley Following on the experience of the Calgary survey, the importance of considering both the sampling base and the recruitment method is illustrated further by an establish- ment survey that was conducted of local trucking companies and shipper/receiver businesses in the western San Bernar- dino Valley region (California). Questionnaires were sent to 405 businesses, which were identified through the California Trucking Association and the cities of Ontario and Rancho Cucamonga. Of these, 37 responses, or 9%, were returned: the low response rate was attributed as being “typical” of this type of survey and also because of “other recent sur- veys” (on unspecified topics) that might have led respondents to view this survey as redundant. It is not clear whether or not direct follow-up contacts were made or if assistance was provided. Moreover, the survey was shorter and less detailed and required less specificity (in terms of shipment type and origin and destination addresses) than, by comparison, the Calgary and Edmonton surveys (40). Online Establishment Survey—Atlanta The 2008 Atlanta Regional Freight Mobility Plan provides an example of the use of online surveys for conducting an estab- lishment survey. This survey was part of a series of data col- lection activities, and sampled shippers and carriers that were considered to be “major freight generators” in the Atlanta region. The survey captured information regarding business type; the number of deliveries to warehouses, distribution centers, and businesses; the day of week and (typical) time of day of deliveries; the number of inbound and outbound loads and shipments; and the destination of loads (used to determine routings within the region). A commercial online survey tool was used, which allowed variations of the survey to be distributed to different groups: shippers and receivers, transportation carriers and operators, and the general public. There were 74, 24, and 31 responses respectively. The results were not used for detailed modeling and were combined with other data—including commercial (TRANSEARCH) com- modity flow data and roadside interviews—for analysis (41). As noted, the sample was limited to “major” generators; and no indication is given that a statistically viable sample was drawn for the “general public.” Nonetheless, the study dem- onstrates the potential of online surveys as a data gathering tool and also as a means of engaging the “public” regarding its freight-generating activities. COMMODITY FLOW SURVEYS The Bureau of Transportation Statistics (BTS) and the Bureau of the Census conduct jointly nationwide Com- modity Flow Surveys (CFS) at approximately 5-year inter- vals (1993, 1997, 2002, and most recently in 2007). Freight characteristics were captured in earlier surveys, from 1963 through 1977. However, the CFS represented an important improvement in method, sample size, and scope over the ear- lier surveys (42). The CFS is a nationwide survey of business establish- ments in selected industries, specifically in mining, manu- facturing, wholesale trade, and selected retail and services establishments (auxiliary establishments). An establishment is defined as “a single physical location where business trans- actions take place” (42). The CFS supplies data on the flow of goods generated by the sampled establishments by mode of transportation in the United States. Data are provided on tons, miles, ton-miles, value, shipment distance, commodity type, and weight. All major modes of freight transportation are captured (43). A sample of establishments is drawn across all 50 states and the District of Columbia. Participation by sampled estab- lishments is mandatory because it is linked to the 5-year Economic Census. The sampling frame is drawn from the Census Bureau’s Business Register of approximately 6 mil- lion establishments, of which approximately 754,000 estab- lishments (in 2007) were in the industry categories covered by the CFS (44). The sample dropped steadily from 200,000 establishments in 1993 to 100,000 in 1997 and 50,000 in 2002 (43). However, it increased back to approximately 100,000 establishments in 2007 (44) (i.e., in effect doubling the sample from 2002, as the total number of candidate establishments was approximately the same in both years). A stratified three-stage sampling process was used in the 2007 CFS (and also previously), as follows (44): Establishment selection. The sampling frame was first • stratified by geography [accounting for the 50 states, the District of Columbia, and 65 metropolitan areas (the last according to population and importance as transportation hubs)]. Within each geographic strata, 48 industry groups were defined (i.e., within the can- didate industry types) according to the 2002 North American Industry Classification System). Separate strata of hazardous materials shippers also were cre- ated to gain more information on these shipments. The combined geography-by-industry stratification resulted in 2,745 primary strata. Based on these strata, a sample size of 102,369 establishments ensured a minimum of 2 samples and a maximum of 100 samples per stratum. Reporting week selection. The sampled establishments • were asked to report on 4 weeks—one in each calendar quarter for 2007 (i.e., January 6, 2007, to January 4, 2008). Because different establishments were assigned different times, the sample covered all 52 weeks of the year. Shipment selection. If respondents made more than 40 • shipments per week, they were asked to conduct a sys- tematic sample to report a minimum of 20 shipments and a maximum of 40 shipments. If respondents made

56 40 or fewer shipments per week, they reported all ship- ments (42). Each of the four surveys used a mail-back document, with online assistance provided in 2002 and 2007. Respondents were asked to record the total numbers of their outbound shipments and, for a sample of these shipments, information on value, weight, commodity, domestic destination or port of exit (from the United States), and mode(s) of transportation. Instructions were provided on how to sample the shipments (43). The CFS has the benefit of being the only nationwide source of goods movement data. However, several concerns have been identified (43): The CFS covers only some industry sectors. This • appears to represent less than three-quarters of all goods moved within the United States. Government, farms, construction, oil and gas, and household [which also generate goods] are not sampled. The CFS misses the rapidly growing service sector and most retail establishments (45). The CFS also does not cover all modes well—in par-• ticular, air cargo is not captured well because many of the industries that depend on air are not included in the sampling frame. Also, not all truck activity is cap- tured: only that associated with the industrial sectors covered in the CFS is included (10). There is a lack of geographic and commodity detail • at the state and local levels. This constraint reflects both the stratification of the sample to ensure broad industry and geographic coverage and the need to pro- tect the confidentiality of individual establishments (some of which could be identified easily by their size and location). In addition, the CFS breaks down metropolitan areas along state lines, thereby making it impossible to distinguish intraregional flows from inter-regional flows in multistate urban regions (9). The varying CFS sample sizes contribute to the lack of geographic detail (10). There is no coverage of the external leg outside the • United States beyond the ultimate destination. That is, only the mode to the port of exit is identified. Through flows that traverse the United States (e.g., Canada to Mexico) also are not covered. No information is cap- tured regarding imports to the United States, except where they arrive in the country for shipment else- where in the United States (10). Routing information is not collected (• 9). Rather, the BTS synthesizes routes as part of the post-survey anal- ysis (44). Although CFS participation is mandatory, establish-• ment response rates consistently have been on the order of 70% (45). Respondent burden has been cited as one reason, and the CFS form invites suggestions on ways to reduce this burden. The turnaround time for processing the data—on the • order of 2 years—limits the information’s timeliness and effectiveness. Moreover, the 5-year cycle can- not capture rapid changes in economic cycles or the impacts of new technologies or policies that might take place in the intervening years. The cost of the CFS is “substantial”—approximately • $15 million in 1993, $19 million in 1997, $13 million in 2002 (45), and $14 million in 2007. INTELLIGENT TRANSPORTATION SYSTEMS TECHNOLOGIES GPS have been used in recent years to provide detailed and accurate data on vehicle location and speeds. A 2008 paper cited several potential benefits of GPS “to supplement, and eventually replace, data collected from roadside surveys,” as follows (11): GPS are “nonintrusive,” which get around “emerg-• ing privacy concerns [that] are making the conduct of roadside interviews more difficult.” Availability of the technology is less of a concern than • it once was, as the number of trucks equipped with GPS receivers has been increasing steadily in recent years. “Coverage of urban freight movement with detailed • route origins and destinations and performance indica- tors [such as, travel time and delay]. “Link-level congestion analysis, including travel time • and speed. “Near real-time international border transit time • monitoring [i.e., at approaches to international border crossings]. “Tools and reporting systems to measure economic • impacts of delays because of incidents. “Fuel consumption and pollution analysis using GPS • units that include engine data retrievers. “Impacts of high-occupancy-vehicle lanes on general-• purpose-lane traffic.” A 2007 paper reported on the use of GPS on 18,000 trucks traveling in Ontario, Canada. The monitored fleet includes 4,000 that were domiciled in the United States and Canadian trucks traveling extensively in the United States, thereby extending the geographic coverage of the monitoring. The low-cost GPS units were supplied by a third-party vendor on behalf of the provincial and federal transportation minis- tries at “low” costs to industry. This “low-cost technological solution with high resolution (polling every 700 feet or 20 to 30 seconds depending on speed and variance)” avoided “costly communication fees, compared to some satellite ser- vices.” Data collected by each truck were stored on its GPS

57 and were downloaded daily to a receiver. Only the vendor had access to the raw GPS, thereby maintaining confidenti- ality and privacy. Selected data were provided to the trans- portation ministries, including time, location, instantaneous and average travel speeds over selected road and highway segments, “hardbrake” (sudden) deceleration, and fuel con- sumption; and only samples of these data were provided. No address information was provided. Of particular interest were the border crossings with the United States, where the monitored trucks provided “near real-time border wait times for trucks via the web, with a 15-minute delay in processing these data” (46). Enquiries also have been made about the use of GPS as a means of guiding drivers. A 2005 freight operator survey in a London, U.K., suburb asked whether the operator’s vehi- cles were equipped with automatic vehicle location equip- ment (that is, a GPS) (47). Several studies have assessed the capabilities of GPS in providing freight data: A 2003 study in the state of Washington evaluated • the performance of wireless transponders and GPS in providing “accurate” measurement of truck location and travel times, “to support regional and state trans- portation data collection efforts” in the Puget Sound region. The evaluation was based on field tests on vehi- cles at two local trucking companies. The evaluation found that the GPS units “showed promise.” However, the high costs of collecting the data across the entire region, the high variability of the GPS data, and the difficulties of integrating the data with a geographic information system (GIS; that is, to display data graph- ically) were cited as limitations. Because the GPS data were vehicle-specific, other sources, such as freeway loop data, provided a more accurate depiction of actual operating conditions. By comparison, two truck tran- sponder networks already had been in place for several years, and so the acquisition and processing of a large, robust sample of data was straightforward. The study recommended that GPS should not be deployed until the technologies and data processing software matured and costs came down, at which time a large-scale test should be conducted. In the meantime, the available transponder data should be incorporated into freight planning efforts to generate appropriate freight travel time statistics that could be used for modeling (48). A 2005 study reviewed five data collection technolo-• gies for potential use in measuring travel times along several Interstate corridors. The purpose of the review was to identify the technologies that could provide “the best combination of geographic and temporal coverage, density of observations, usability of data formats, and cost-effectiveness for collecting information on truck movements along the corridors.” Although the study did not assess the technologies, it pointed out the fol- lowing respective advantages and disadvantages (49): Satellite GPS transmits a continuous or periodic – signal to an earth orbit satellite. The GPS receiver is used to determine the vehicle’s location based upon algorithms or a signal triangulation. The review found that this technology offered “superior ser- vice” in rural areas. Terrestrial tracking is based upon analog or digi- – tal cellular technology, which uses multiple cover- age base station areas. The benefit is the coverage provided within urban regions where satellites are not as effective. Disadvantages include limited coverage in rural areas and on Interstate highways, less precise locating capabilities than satellite GPS (50–150 meters), and multisystem interoperability problems. Hybrid tracking systems combine elements of the – two preceding technologies. New systems were combining terrestrial coverage in urban areas sup- ported by satellite service in those areas where ter- restrial coverage is not available. This minimizes overall system cost compared with a satellite track- ing system while providing nationwide satellite coverage that is not available with the disparate ter- restrial systems. On-board computer tracking is the least advanced – technology. The simplest and most labor-intensive form of vehicle tracking, it consists of cabin-mounted measuring/sensing processors that electronically or mechanically record such data as speed, idle time, and mileage. Benefits include being the least costly tracking option, its ubiquity (various forms of the technology exist in most vehicles), and its low cost for supplying simple statistics. Disadvantages include its relative lack of sophistication, limited data availability for the user, and in some systems the need for substantial manual processing to extract and analyze data. Fixed site systems such as electronic toll collection – systems (e.g., Ez Pass systems). Although these systems are in place along several Interstate sec- tions, the number of sites is limited and off-corridor coverage is not feasible. The review found that these systems could not be used as a primary means of freight data collection but could be used to augment other systems. Two 2006 papers reported on a “benchmarking” • study in Washington State. The study compared GPS with Commercial Vehicle Information Systems and Networks’ automated vehicle identification (CVISN- AVI). One paper cited the high accuracy of GPS on travel routes and on individual road segments: “the advantage of the GPS devices is that they can monitor the actual route taken by instrumented vehicles. This makes the GPS [data] far more robust than the transpon-

58 der data.” However, the primary disadvantage of GPS is the relatively small number of GPS-equipped trucks and, accordingly, “in a large metropolitan region insuf- ficient data may be collected on many routes unless a fairly large sample of trucks actively participates in the data collection effort.” Also, the GPS data collection requires “considerable staff effort” to coordinate, and a “mechanism for recruiting trucking company partici- pation” is required (50). The second paper noted that CVISN-AVI-equipped vehicles were much more com- mon and little labor was required to retrieve the data (compared with GPS), but data could be collected only at routes equipped with CVISN-AVI readers at weigh stations. Moreover, the data cover several trucks across long stretches of road, and specific points of congestion can be difficult to locate (51). A 2007 research survey in Peel Region, Canada (a sub-• urb of Toronto), included a tour information form that was to be completed by the responding establishment’s drivers. The form began with information at the start location; that is, the work location where the driver’s “work day” began. Some of the vehicles were fitted with GPS units to enable comparison with the paper driver surveys. The combined data also were used to describe tours. The paper surveys faced significant stop non- reporting issues, such as truncated surveys, stops that were missing in the middle of a tour, and incorrectly or inaccurately recorded stop location information. At the same time, the GPS units provided a useful level of pre- cision but did not capture well stops of short duration (of less than 5 minutes, which comprised a significant portion of the stops). Also, small differences in the identified (compared) stop locations resulted in some misinterpretations of the actual tour (52). Finally, a 2006 study proposed the use of optical char-• acter recognition (OCR) technology to collect vehicle license plates from trucks accessing selected terminals at the ports of Los Angeles and Long Beach. The pur- pose of the data collection was to develop an initial estimate of heavy-heavy-duty diesel truck activity and population, for application to the San Pedro bay Ports Clean Air Action Plan. To develop the estimate, data were collected for a continuous 37-day sample at five container terminals. The sample was considered rep- resentative of the Ports’ 14 terminals. The OCR data were first analyzed to identify unique license plates and then to obtain the frequency of access to the Ports. The analysis yielded 15,700 unique license plates that comprised more than 253,000 trips. Of these, only vehicles registered in California were retained, result- ing in approximately 12,000 vehicles and 244,000 trips that were used in the analysis. The remaining truck and trip records were categorized into three groups: Frequent callers—trucks that visited the terminals – one or more times daily (50.4%) Semifrequent callers—trucks that visited at least – every other day (30.3%) Nonfrequent callers—trucks that visited less than – once every other day (19.3%) (53). COMPARISON OF TECHNIQUES Some of the case studies described earlier compare alter- nate survey techniques, whereas others combine several techniques. A definitive description of the circumstances under which a particular type of survey should be used does not exist. However, three recent comparisons provide some guidance. They also illustrate the complexity of the topic and demonstrate how the definition varies according to the perspective. A 2001 study identified two survey types among three types of truck data collection methods: roadside intercept surveys and travel diary surveys. The third method is vehicle classification counts. The three methods and the associated characteristics are listed in Table 28 (54). The roadside inter- cept and the travel diary surveys have long been used as an analytical basis for truck trip generation rates and model- ing, and they often are accompanied by vehicle classification counts (in part to expand survey samples). This categoriza- tion also has been cited and expanded by other sources [e.g., (9)], but the general basis is common. However, the perspec- tive is that of quantitative analysis and modeling which, although important, represents only part of the picture. A 2004 study in Portland, Oregon, compared differ- ent truck trip data collection methods (9). One test exam- ined roadside surveys along a highway, at a marine port, and at a private transportation depot/distribution center. Another series of tests examined different combinations of self-completion surveys (i.e., different combinations of mail-back or facsimile surveys with telephone contact and follow-up), with surveys sent to “known” samples (that is, establishments with which the sponsoring authorities had an established working relationship regarding other freight initiatives) and “unknown” samples (freight establishments without a prior contact). The roadside surveys experienced high response and com- pletion rates at all three locations, with some qualifications: the preponderance of container traffic at the port limited the driver’s knowledge of payload information, obtaining detailed origin-destination addresses was difficult for all interviews (though less so for the distribution center interviewees), and finding large numbers of private firms to participate in this type of survey could prove challenging (9). The mail-back and facsimile surveys presented more challenges, although they provided valuable information. There was no measurable difference in the response rate

59 or in the quality of the information between the known and unknown samples (i.e., familiarity did not make a differ- ence). Both contained a “very high” percentage of incor- rect and invalid information. The pre-survey and follow-up telephone contacts improved response rates overall, though not of completed questionnaires (in other words, increased contact was associated with increased refusal rates). Nonre- spondents comprised almost half the known and unknown samples; follow-up contact might improve response rates. Respondents provided several reasons for not participating, TABLE 29 METHODS BY WHICH URBAN FREIGHT SURVEY TECHNIQUES CAN BE CONDUCTED Establishment surveys (surveys of the 1. shipments made by businesses—the shippers and/or receivers of goods and services, with specific origin-destination information) Self-completion (post, fax, or e-mail)• Self-completion (post, fax, or e-mail with initial and reminder phone call)• Self-completion (left and collected in person)• Telephone interview• Face-to-face interview• Commodity flow survey (surveys of businesses 2. on the quantities of goods shipped. Can include some origin-destination information) Self-completion (post, fax, or e-mail)• Self-completion (post, fax, or e-mail with initial and reminder phone call)• Self-completion (left and collected in person)• Telephone interview• Face-to-face interview• Freight operator survey (surveys of logistics 3. managers of businesses or of carriers regarding the fleet’s activities, including origin- destination) Self-completion (post, fax, or e-mail)• Self-completion (post, fax, or e-mail with initial and reminder phone call)• Self-completion (left and collected in person)• Telephone interview• Face-to-face interview• Driver survey (surveys of a driver’s activities 4. on his/her rounds for a given period) Self-completion (left in person)• Face-to-face interview• Roadside interview survey (surveys of the 5. vehicle’s activities for the trip being made when the vehicle is stopped for the interview) Face-to-face interview• Vehicle observation survey [observations by 6. others of a vehicle’s activities at a given site(s). Does not necessarily involve the vehicle driver] In-person observation• Observation using film/camera• Parking survey (observations by others of a 7. vehicle’s activities while it is parked or being loaded or unloaded at a stop) In-person observation• Observation using film/camera• Vehicle trip diaries (surveys of a vehicle’s 8. activities on its rounds for a given period. Similar to the driver survey but specific to the vehicle) Self-completion (post, fax, or e-mail)• Self-completion (post, fax, or e-mail with initial and reminder phone call)• Self-completion (left and collected in person)• GPS survey (electronic surveys of a vehicle’s 9. exact location. Also captures travel times) Equipment/transmitter fitted in vehicle• Suppliers survey (surveys of supplier 10. businesses—suppliers to the supply chain—on the goods being shipped and on the supporting vehicle activity) Self-completion (post, fax, or e-mail)• Self-completion (post, fax, or e-mail with initial and reminder phone call)• Self-completion (left and collected in person)• Telephone interview• Face-to-face interview• Service providers survey (surveys of services-11. generating businesses regarding the characteristics of its employees’ trips. Similar to freight operator survey but specific to services) Self-completion (post, fax, or e-mail)• Self-completion (post, fax, or e-mail with initial and reminder phone call)• Self-completion (left and collected in person)• Telephone interview• Face-to-face interview• Vehicle traffic counts12. Manual (in-person) counts• Automated counts (using sensors, film, cameras, or other technology)• Source: Allen and Browne (2008) (2).

60 TABLE 30 ADVANTAGES AND DISADVANTAGES OF METHODS FOR CONDUCTING URBAN FREIGHT SURVEYS Survey Method Advantages Disadvantages Face-to-face interviews and telephone surveys (for wide range of survey techniques including establishment, commodity flow, vehicle operator, shipper, and service provider surveys) High response rate compared with self-• completion due to personal contact Can provide better quality, more detailed • information than self-completion method Provides opportunity to query responses• Good for open-ended questions and • in-depth discussion about responses Easier to make follow-up contacts• Telephone surveys offer better • opportunity to survey over large geographical area than face-to-face interviews Face-to-face interview allows more • in-depth discussion and use of other techniques (such as supply chain mapping etc.) More expensive and time consuming per • respondent than self-completion (especially face-to-face interviews) Can prove too expensive for a large sample • size (especially face-to-face interviews) Often difficult to obtain initial and • participation and requires call backs Self-completion surveys (for wide range of survey techniques including establishment, commodity flow, vehicle operator, shipper, and service provider surveys) Lower cost method than interviews of • self-completion with initial contact Permits larger and more representative • samples than interviews Offers better opportunity to survey over • large geographical area than face-to-face interviews Generally lower response rates than with • interviews or self-completion with initial contact. Difficult to ensure that right person in • organization will respond No way of knowing whether respondent • understood question in way intended No opportunity to check/clarify or discuss • responses Difficult to interpret nonresponses to • questions Not good for open-ended questions• Self-completion with initial contact and reminder by phone call or in person (for wide range of survey techniques including establishment, commodity flow, vehicle operator, shipper, and service provider surveys) Lower cost method than interviews—• effective method Can provide better response rate than • basic self completion method Phone/in-person follow-up can allow • opportunity to clarify/discuss responses (but difficult to achieve in practice) Offers better opportunity to survey over • large geographical area than face-to-face interviews More expensive than basic self completion • method Other disadvantages same as basic self-• completion method Roadside (face-to-face) interviews instead of vehicle trip diaries (self-completion) (for obtaining vehicle journey data) High response rate• Can provide information on trip purpose, • goods carried and origin/destination, and route Disruption to traffic flow• Staffing requirements are high, making it • expensive No opportunity for follow-up with • respondents Requires involvement of police and/or other • bodies Does not provide details about entire journey • and stops In-person observation instead of using film/ camera (for vehicle observation/parking surveys) Reduced potential to cause traffic/• delivery disruption No risk of equipment/ recording failure• Provides actual data about number and • timing of deliveries and collections, unlike establishment survey Staffing requirements are high, making it • expensive Limited to hours/days of observation, so does • not capture all activity Neither in-person nor film observation can • capture all delivery and collection activity, especially not of vehicles stopping off-street or in side roads Manual traffic counts instead of automated traffic counts Reduced potential to cause traffic • disruption Complete disaggregation of vehicle type • possible if trained surveyors used Vehicles not wrongly identified• No risk of equipment failure• Staffing requirements are high making it • expensive Difficult to collect traffic count data at many • locations without it being very expensive Source: Allen and Browne (2008) (2).

61 including privacy, inapplicability, lack of time, or unavail- ability of an appropriate person at that location. Information related to trip detail for inbound and outbound shipments was the most difficult to obtain through the mail survey (9). Given these findings, and noting that these were pilot tests that were based on small samples, the study made sev- eral recommendations (9): Establish an updated freight contact list to improve the • response rates and quality. Use roadside interviews as the basis for data collection • for “inter- and intra-regional freight movements” at the three types of locations (highway, terminals, and pri- vate warehouse/distribution centers), and ensure that the survey sites are geographically distributed in the metropolitan (Portland) area. Use the mail-back/facsimile surveys of warehouses/• distribution centers to complement the roadside inter- views and gather selected pieces of information (nota- bly, about the activity at the facility’s location, and not about the trip detail). Use improved survey design techniques to improve the • response rate. Employ all possible follow-up techniques, subject to • budget. Finally, a 2008 review of urban freight surveys compared survey techniques. Table 29 lists the techniques that are most commonly used for 11 survey types (and also includes vehicle traffic counts, because these commonly complement survey activities). The following are six basic survey tech- niques (2): Face-to-face interviews (i.e., at the establishment loca-• tion) and telephone surveys. Self-completion surveys (mail-back or on the • Internet). Self-completion survey, with initial contact and fol-• low-up reminder by telephone or in person. Roadside (face-to-face) interviews, as distinct from • self-completed driver surveys or vehicle trip diaries. In-person observation, instead of using film or camera • (for vehicle observation or parking surveys; also has been used to capture driver or vehicle activity, through people accompanying the driver on his/her rounds). Manual traffic counts instead of automated traffic • counts (i.e., to distinguish vehicle types in the counted traffic). This technique is identified in the source docu- ment as being relevant to urban freight surveys; how- ever, it is not a survey type in itself. Table 30 compares the survey types’ relative advantages and disadvantages (2): Face-to-face interviews enable a high response rate, • can provide better quality information and details, enable open-ended discussion and probing, and allow for both quantitative and qualitative input. However, they are expensive and time consuming. Telephone surveys have the same advantages as face-• to-face interviews, and though they are also time con- suming they can provide a less expensive means of capturing a larger sample size (per unit cost). Self-completion surveys have low unit costs and • can reach a larger number of respondents. However, response rates typically are lower, and there is no con- trol over the actual response, its level of detail, accu- racy, or quality (i.e., questions can be misinterpreted or omitted), or whether the appropriate person is responding. Self-completion surveys with telephone or in-person • follow-up provide the opportunity to improve response rate, clarify the questions, and otherwise guide respon- dents in completing the survey properly and com- pletely. However, this approach is more expensive than the stand-alone self-completion surveys. Roadside interviews have a high response rate and • allow for solicitation of detailed information of the trip at hand. However, they disrupt traffic flow, are expen- sive (because of staffing requirements), and are able to gather only limited information beyond the immediate trip. In-person observations (compared with filming or • camera) provide data on actual conditions. However, they are expensive and are limited to the actual times of observation (i.e., no supplementary data). Manual traffic counts (compared with automated traf-• fic counts) provide accurate data on vehicle types. However, staffing requirements make them expensive.