Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 103


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
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 102
U S I N G G L O B A L P O S I T I O N I N G S Y S T E M D ATA T O I N F O R M T R AV E L S U RV E Y M E T H O D S 91 A comparison of the CATI-reported trips with the between the two equations. That is, it accounts for the GPS-detected trips identified 280 GPS-detected trips that potential presence of unobserved individual factors (such were not reported by the drivers in the CATI travel sur- as, say, an overall disinclination to respond to surveys or vey. [The extraction of trips from the GPS traces was substantial time constraints) that influences both based on a multilevel trip detection algorithm developed whether an individual underreports and the level of by GeoStats, with several built-in checks to avoid ghost underreporting. For a more detailed discussion of the trips (such as due to starts and stops at street lights). The modeling portion of this research, see Bricka and Bhat. full details of the GPS processing are available in the study The fundamental hypothesis underlying our empirical report by NuStats.] A descriptive analysis of trip underre- analysis was that trip underreporting is largely due to porting by driver demographics, driver travel characteris- three areas of influence: who the driver is (driver demo- tics, and driver adherence to survey protocols was graphics such as household type, age, number of house- undertaken. The results related to demographic charac- hold vehicles, employment status, etc.), the characteristics teristics suggested that drivers between the ages of 50 and of trips made (total number of trips, average distance of 69 who were male, with low education levels, were not trips, and level of trip chaining), and how well the driver employed or were employed in salesclerical occupations, adhered to the survey protocol (whether driver used the working at locations characterized as residential, from travel diary to record all travel and whether driver talked single-adult or retired households, from one- or three- directly with interviewer). All exogenous inputs to the person households, and from 3+ vehicle households were model were classified according to these broad categories. the most likely to underreport trips. The driver travel The final variable specifications for the binary model of characteristics indicated that drivers who made a rela- underreporting and the ordered-response model for level tively large number of total trips during the survey day, of underreporting among underreporting individuals pursued long distance trips, and undertook trip chaining were developed by adopting a systematic procedure of on the survey day were overrepresented in the pool of eliminating statistically insignificant variables. Of course, those who underreport. Finally, in the category of driver as indicated earlier, the entire specification effort was also adherence to survey protocols, the results suggested that informed by the results of earlier studies and intuitive drivers who did not use their travel diaries for recording considerations. travel and who had their travel details reported by proxy were more likely to underreport. These descriptive statistics provide suggestive evi- RESULTS AND IMPLICATIONS FOR dence of the effect of various driver attributes on the SURVEY METHODS propensity to underreport trips. However, these are uni- dimensional statistics in that they do not control for the The results from our modeling effort provide important influence of other variables when the impact of any sin- insights about underreporting tendencies in traditional gle variable is being examined. For instance, the gender household travel surveys. First, the underlying mecha- difference in underreporting may be a manifestation of nism that represents whether an individual underreports different travel patterns of men and women. Further, the or not is different from the mechanism that determines descriptive analysis does not focus on the characteristics the level of underreporting. At the same time, there are affecting the level of trip underreporting. To obtain a common unobserved factors that influence both the comprehensive picture of the factors affecting whether underreporting propensity and the propensity associated an individual underreports and the level of underreport- with the level of underreporting. Consequently, it is ing, it is necessary to pursue a multidimensional and important to use the joint binary-unordered response comprehensive analysis that examines the effects of all framework of the current study to analyze trip underre- potential determinants of both underreporting propen- porting and its magnitude. Second, the effect of driver sity and the level of underreporting propensity. The next demographics indicates that young adults (less than 30 section presents the model structure and empirical analy- years of age); men; individuals with less than high school sis for such a methodology. education; unemployed individuals; individuals working in clerical and manufacturing professions; workers employed at residential, industrial, and medical land MODEL SUMMARY uses; and individuals in nuclear families are all more likely to underreport trips in household travel surveys The approach adopted in this study used two equations: than other respondents. Third, driver travel characteris- a binary model for whether an individual underreports tics that affect the tendency to underreport include mak- or not and an ordered-response model for the number of ing a high number of trips on the survey day, traveling trips underreported if there is underreporting at all. The long distances per trip, and trip chaining. Fourth, drivers methodology accounts for the correlation in error terms who do not use the travel diary to record their travel are