Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 60


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 59
Procedures and Measures for Further Research 59 There are generally five approaches one can take to the problem of under-representation of rare behaviors in a random household survey. The first is to over-sample in certain sub- areas of the region. The second and similar approach is to target a portion of the sample into such areas as would have been over-sampled. A third approach is to use a secondary sampling procedure, such as an intercept survey, to find transit riders (or other relevant rare behaviors), and to obtain telephone numbers for the households of those encountered in the intercept survey. The fourth approach is to organize an independent survey, such as an on-board tran- sit survey, and obtain the augment sample from this source. The fifth approach is to stratify the population into the groups of interest, and then use screening to fill the samples for each stratum. While benefits and problems associated with each of these methods are generally well under- stood, there is a need to review recent practice and productivity of the different methods of aug- mentation. There is a need to look at other possible ways to augment household and personal travel surveys for rare travel behaviors and specific rare socio-demographic characteristics. Future research would need to examine the costs of the different approaches and determine some mea- sure of cost-effectiveness for them. 4.2.13 S-4: Stratification Options for Samples Although the usual aim of stratification in household and personal travel surveys is to ensure coverage of household characteristics, it will generally have the effect of reducing the sampling error. This aspect of stratification has been largely ignored in travel survey sample designs. While the aim of stratification is to ensure that the sample contains households in specific geographic subdivisions of the region and that each household size and vehicle ownership combination of significance is represented in the final sample, there does not appear to have been any investiga- tion of the effects of this on model estimation. It would appear to be useful and valuable to provide guidance on the stratification designs. As far as the literature reveals, little attention has been paid to the effects of stratification on the error properties of modeling steps beyond trip generation. Second, it has not been established that strat- ifying on the variables of trip generation necessarily produces more efficient samples and samples with desirable error properties. Third, there has been little or no investigation of whether there may be good alternative stratification schemes that can be used. Fourth, there is little guidance on what sample sizes to choose for each cell of the stratification matrix. In the absence of informa- tion on the variances in trip rates for each cell, there is no guidance on whether choosing equal samples in each cell is appropriate or whether there is some other possible method of determin- ing an appropriate sample size for each cell. Finally, the relative advantages and disadvantages of stratified sampling versus simple random sampling have not been investigated for household travel surveys. Because there is a cost to stratified sampling--which cannot generally be done based on prior identification of households as to the stratum to which each belongs--the use of stratified sampling generally requires contacting households to determine membership in a stra- tum and then qualifying or disqualifying the household on the basis of the required sample in a cell. The costs of this method over increasing the sample size for a simple random sample are not known for household travel surveys. Standardized procedures are probably not appropriate in this area. Rather, what appears to be needed is guidance. As discussed above, it is recommended that future research examine the impacts of stratification by the variables of trip generation modeling on both subsequent model- ing steps and on the sample properties. Recent surveys should be reviewed to determine whether other stratification schemes have been used and to determine what effect these have had on sam- ple properties. Research should develop guidance as to how to choose sampling strategies and how to choose the sample sizes in the cells of a stratification matrix.