Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 7
Compressed Work Week. CWW allows employees to work a greater-than-standard number of hours each day so as to reduce the total number of days worked, and hence, the number of times it is necessary to commute to the work site. A popular arrangement is the 9/80 schedule in which employees work 9 hours per day versus the standard 8 hours, and then get the 10th day off. Generally, employees arrange with the employer which day of the week will be their day off, typ- ically choosing a Monday or Friday in order to extend their weekend. Telecommuting. Telecommuting (or "Telework") is an arrangement whereby an employee is per- mitted to work at a remote location one or more days a week rather than commute to the work site. The availability of telecommunications technology which allows the worker to remain in "virtual" contact with the work site via a networked home or other remote computer is the basis for the nam- ing of this strategy. However, not all employees who telecommute are necessarily working on a computer, and hence the real characteristic of the strategy is that the employee is allowed to work offsite. The offsite location is generally the home or sometimes at a remote "telework center" which has the necessary equipment. As with CWW or staggered work hours, employees who telecom- mute generally do so on a fixed schedule that they negotiate with their employer. Analytical Considerations Attempting to quantify the impact of TDM strategies on traveler behavior brings several critical analytical considerations to light. · It is almost never the case that a given TDM strategy is implemented (or evaluated) in isolation, as a unique action. Rather, strategies are normally implemented in combinations, or "packages," such that ascertaining the effectiveness of an individual action is statistically very challenging. · Added to the statistical challenge is the fact that the available data for conducting these analy- ses are seriously limited. If travel data have been collected, they are almost always for a post- implementation period and do not offer a comparative "before" measure. The data collection methods themselves are also often suspect, as is the aggregate format (total mode split or "aver- age vehicle ridership") in which they are typically presented. · Accounting for setting and context is very important. The effects of some strategies may be much more significant in an environment where there is good transit service, where parking is limited and priced, or where other activities are reachable from the work site without a per- sonal vehicle. · Finally, the details on program strategies themselves are often incomplete or inconsistent. It is often the case that neither the magnitude of incentives/services nor the time a strategy has been in place is reported. Moreover, it is usually not recorded whether individual strategies were poorly implemented or well run. As is evident from the list in the preceding subsection, there are a large number of TDM strategies. Given the many ways in which they can be bundled and applied, attempting to present estimates of travel impact for individual strategies is simply not always possible with the data and analyses from existing research studies and databases. Many TDM studies are both topical and qualitative, focusing on a particular type of programmatic strategy or approach, but offering little numerical information as to impacts. These studies tend to 19-7
OCR for page 8
be focused more on reporting "who" has used a particular strategy or approach, "how" they imple- mented it, and the circumstances underlying "why" that approach was taken. In many cases, the impetus is a legal or regulatory requirement, and the issue is in assessing the reasonableness or overall effectiveness of the requirement. It might be expected that significant TDM impact information would be available from the vari- ous state- or regionally-enacted employer trip reduction programs introduced in the late 1980s/ early 1990s, largely in response to concerns about worsening air quality in major metropolitan areas. California was the vanguard in these efforts, with passage of Regulation XV in 1987. This action legally required employers in the Los Angeles region with 100 or more employees to introduce mea- sures to discourage driving and to reduce employee vehicle trips by 15 percent. This state initiative presaged a similar national initiative written into the 1990 Clean Air Act Amendments, requiring metropolitan areas in "severe" non-attainment of National Ambient Air Quality Standards (NAAQS) for ozone to implement similarly-structured Employee Commute Options (ECO) programs. The California program, and many of those patterned after it (e.g., Washington State's Commute Trip Reduction Law and Portland, Oregon's Regional Transportation Options program) formal- ized employer participation by requiring travel surveys of employees and the development of a plan detailing the trip reduction goal, the current gap, and the specific set of TDM strategies that would be used to reach the goal. These travel data and employer TDM plans were computerized and stored in massive databases to be used for subsequent tracking of performance and evaluation of effectiveness. Employers would be evaluated on a biennial basis, necessitating a new employer survey and a revised plan with additional strategies as necessary to achieve their respective trip reduction target. While these data archives would appear to offer great opportunity to obtain measurements of exactly the type of strategy-impact relationships which are the subject of this chapter, a number of factors cause these data to be of limited use: · The travel survey data collected by employers were reduced to aggregate measures before data release. Five-day-week average mode shares for the employer were computed. Also computed was the Average Vehicle Ridership (AVR), an amalgam of employee commute trips by all modes and means into an effective commutes-per-vehicle rate (covered employees reporting to work divided by the vehicle round trips thus produced, per average day, over a 5-day week). To the extent that some employees telecommuted or participated in a CWW schedule, that too was embedded into the AVR for an average day.1 · The processed data lack any record of the characteristics of individual trips, and there is no tracking of individuals between surveys. The analyst must try to explain changes in the aggre- gate travel measures--between consecutive plans--with nothing more than information about the composition of the TDM program. The processed data provide no clue as to individual 1 The inverse of the AVR is average vehicle trips per employee, one of two common vehicle trip rate expres- sions found in the TDM literature, though not always implying a rigorous 5-day average. The other common vehicle trip rate expression is average vehicle trips per 100 employees. Both are used within this chapter. The former is always less than one, e.g., 0.85, while the latter is nearly always a two-digit number, e.g., 85. Both are expressed in terms of complete round trips, in contrast to the usual transportation demand analysis prac- tice of counting individual one-way trips or trip ends. Either trip rate expression (and also AVR) may be used to compute vehicle trip reduction (VTR), a dimensionless rate, so long as consistency is maintained. 19-8
OCR for page 9
worker socio-demographics, trip origins/destinations, trip length, quality of travel alterna- tives, or changes over time in travel alternatives available. · There is considerable uncertainty in determining exactly what strategies (or incentives) were being applied in a given employer program, based on the information supplied in the required Trip Reduction Plans. Details were often found to be missing in terms of specific strategies, when they were invoked, to whom they were available, or what monetary value they represented. For example, it might be recorded that a transit subsidy was provided, but with no informa- tion on the dollar value of that subsidy. · The reliability of the data, as entered into the massive databases involved, has often been found wanting. In some areas, such as in Washington State and Portland, Oregon, the mas- sive amounts of information collected tended to overwhelm budgeted data processing capa- bilities, resulting in large amounts of these data waiting to be processed or purged of questionable data records. Numerous entities have attempted to take on the challenge of analyzing these data, but with lim- ited success (Giuliano, 1992; Comsis, 1993b; CUTR, 2004). Simply put, the precision of the data was found to be too coarse to permit use of the necessary multivariate statistical tools. Two exceptions are noted here, largely because of an important departure from the standard analytic approach: · The first is a study sponsored by the California Air Resources Board (CARB) which, recognizing the limitations of the Regulation XV database, commissioned an original survey of 45 employers in Los Angeles and Sacramento. This enabled the researchers to control the quality of the response data on employee travel and specific strategies in operation, and to analyze travel response at an individual traveler level, rather than an aggregate outcome for the employer. Unfortunately, the research did not have access to pre-program baseline data for the surveyed employees, which seriously limited its ability to analyze behavioral changes in response to the given employer pro- gram (Comsis, 1993a). Nevertheless, the findings from this study are somewhat unique and are reported later in the "Related Information and Impacts" section. · The second is a more recent effort by the University of South Florida's Center for Urban Transportation Research (CUTR), done under its National Center for Transit Research, which has attempted to develop a Worksite Trip Reduction Model using data from the California (Regulation XV), Washington State Commute Trip Reduction (CTR), and Tucson, Arizona, employer databases. While CUTR experienced the same types of data problems as identified by Comsis in its 1993 study for CARB, it opted to use a different statistical approach, one incor- porating "neural networks," to extract relationships (CUTR, 2004). This research is also sum- marized in the "Related Information and Impacts" section. (The model itself has subsequently been updated, as covered in Footnotes 14 and 20 of the "Related Information and Impacts" and "Additional Resources" sections, respectively.) The primary focus of this chapter on quantifying the link between TDM strategies and travel behavior ultimately led to a small number of studies that specifically focused on that aspect of TDM programs without attempting to rely exclusively on the large databases critiqued above. The three studies most extensively utilized are as follows: · Comsis Corporation and Institute of Transportation Engineers, "Implementing Effective Travel Demand Management Measures: Inventory of Measures and Synthesis of Experience." Prepared for Federal Highway Administration and Federal Transit Administration, Washington, DC (1993). 19-9
OCR for page 10
· Rutherford, G. S., Badgett, S. I., Ishimaru, J. M., and MacLachlan, S., "Transportation Demand Management: Case Studies of Medium-Sized Employers." Transportation Research Record 1459 (1994). · Comsis Corporation. "Task 2 Working Paper: An Examination of Cost/Benefit and Other Decision Factors Used in Design of Employer-Based TDM Programs." TCRP Project B-4. Unpublished research findings and associated data files, Transportation Research Board, Washington, DC (1994).2 While somewhat dated, these studies nevertheless provide the most comprehensive set of employer and institutional TDM project examples available in which a travel change has been calculated. The calculation is based either on before and after data, or on comparison with an objective control site or measure against which to gauge a travel behavior difference.3 The primary travel change measure utilized is the vehicle trip reduction for the site. Vehicle trip reduction (VTR), a term commonly used in both analytical studies of TDM and travel mitigation legislation, is the percentage of vehicles removed (actual, presumed, or estimated) from a site's com- mute traffic load. More specifically, it is the incremental reduction achieved in the vehicle trip rate, expressed as a percentage of the starting-point trip rate. Planned and apparently successful trip reductions are reported as positive VTRs. A negative VTR indicates that the travel change appears to have been an increase in the vehicle trip rate. A VTR may be associated with TDM, land-use and site-design modifications, or related actions, although TDM per se is the subject of this chapter. VTR may be computed from mode shift data if one knows the before-and-after carpool occupancy rates and the amount of trip suppression attained from CWW and telecommuting strategies. 2 TCRP Project B-4, "Cost-Effectiveness of TDM Strategies," is an important TDM research project whose detailed findings were never formally published. This project was centered on a national survey of 50 employers (49 officially accepted) covering the details of their TDM programs. The sample was designed to include employers of all sizes, basic industry groups including institutions, locations within or outside urban areas, presence or lack of a regulatory requirement, and various types of strategies within programs. Extensive analyses were done with the survey data, including the effect of both program strategies and environment on vehicle trip reduction, reasons for engaging in TDM, sources of information for program design, changes made over time and compelling circumstances for the change, and costs and benefits. The initial analysis was documented in the identified "working memo" and technical appendices, but the review panel opted to frame the final report in a less technical, more executive format. The authors of Chapter 19 have had access to this unpublished technical information, and the TCRP Report 95 team received support for using these Project B-4 data and findings in developing this third edition Handbook. 3 Slightly different methods were used by the three studies to establish a baseline against which to gauge the given program's presumed vehicle trip reduction (VTR). The Rutherford et al. (1994) case studies compared the given project with area-wide mode split averages. TCRP Project B-4 selected as its datum the area modal split from the 1990 Census Transportation Planning Package (CTPP). The earliest study--Comsis and ITE (1993)-- used a variety of comparative methods depending on data availability. These included before-and-after survey information, comparison against a local peer site of similar character but without a formal TDM program, or transportation mode split taken from regional model estimates for the subarea. Occasionally one of these methods gives an individual trip reduction value that is open to question (key instances are highlighted in chapter footnotes). Overall, however, the approaches used are a reasonable response to difficult data constraints. For simplicity, travel differences relative to the baseline are referred to in this chapter as vehicle trip reduction (VTR) although in fact most are inferred by comparison against the selected baseline. 19-10
OCR for page 11
Successful vehicle trip reduction does not necessarily infer less traffic than before. VTR is a dimen- sionless rate, referring to the reduction in vehicle trips from what they would have been without the TDM or other travel mitigation strategy. There are a combined 80 different project examples provided by the three primary studies listed above. A fourth study (Comsis et al., 1996) provided two of the examples. These examples cover a wide range of situations in terms of employer type and size, setting, and composition of TDM pro- gram. Using this collection of cases as an analysis platform, the approach taken in this chapter is along the lines of a descriptive analysis. Namely, for this reasonably large and diverse sample of employer and institutional TDM projects--for which there is fairly reliable and complete data-- an analysis has been structured that partitions the examples into cases that do versus cases that do not have a particular TDM element or other characteristic. The evaluation then attempts to infer the relative importance of particular categories of TDM strategies (e.g., support versus incentives), and even to some degree of particular strategies (e.g., a transit subsidy versus an HOV parking dis- count), through pair-wise comparisons from the sample. Unweighted project averages within cat- egories are used in these comparisons. This approach is used to examine each of the four basic TDM strategy groupings: employer support, transportation services, financial incentives, and alternative work arrangements. This 82-program sample becomes the basis or starting point for the analysis in each of the program areas. To the extent that there is sufficient clarity about an individual TDM strategy--either from the 82 examples or, more often, a supplemental study out of the literature that focuses on that strategy--additional detail about the nature and impact of that strategy is provided. There are four major caveats in the primary approach of this analysis, which uses the sample of 82 programs as a basis. First and foremost, in no way should these 82 examples be construed as a random sample of TDM programs. While there are a number of examples that are fairly average in their impact, many of the examples were originally selected because they were distinct in some way--often because they were regarded as success stories. Hence, the performance of this group of programs should not be looked upon as being "typical." As a group, they are probably consid- erably above average in effectiveness relative to what one would expect to find in a large random sample of programs. Second, even though there can be reasonable confidence about the quality of the data for this sam- ple, the "pairwise" approach used in evaluation is a very simplistic means for trying to quantify the impact of individual strategies. The only way that one can begin to discern the relative impact of any given strategy is through a multivariate statistical approach, such as regression. Unfortunately, the data in this sample are too coarse and too aggregate to enable such a statistical approach, so one is resigned to trying to learn as much as possible from the data available, acknowledging the many shortcomings. Later in the report, in the "Related Information and Impacts" section, the results of two research studies that have attempted to apply more advanced statistical methods to this task are presented, though both were ultimately limited by the same data issues outlined here. Third, except where before and after data is available and exogenous influences are known, causal- ity is only inferred--not demonstrated. It is presumed that observed travel differences are largely caused by the TDM measures in place, but there is no analytical proof. In discussion, the terms "implied" or "inferred" are occasionally inserted as a reminder that the effect is presumed, not proven. Indeed, a majority of the VTR outcomes presented are not direct observations but rather estimates/inferences based on comparison with control sites or areas. 19-11