BLOW UP: EXPANDING A COMPLEX RANDOM SAMPLE TRAVEL SURVEY

Abstract

In April 1991 the Southern California Association of Governments contracted with the Applied Management and Planning Group to conduct an origin-destination survey of 15,700 households in five Southern California counties. The survey sample was stratified for each county by regional statistical areas (RSAs). Within each county, the sample was stratified also on housing type, vehicle ownership, and household size, yielding a 30-cell sampling matrix for each of the five counties but needing to be reported by the 49 RSAs in the study area. The overall sampling frame was thus a combination of 49 cells (RSAs) and 150 cells (stratification within county). The survey data were collected between April and June 1991. The subsequent file creation and analyses extended far enough into 1992 to enable the use of the 1990 census data to expand the sample to the population. The available census data for expansion consisted of the one-way frequencies of each of the three sociodemographic variables but did not provide any cross tabulations of these variables. The expansion and weighting procedure used to generate population estimates that match as closely as possible the characteristics measured in the 1990 census is described. It is shown that the procedure is relatively simple, even though the sampling procedure was complex. In addition, the sampling errors are reported and are compared with those that would have been obtained from a simple random sample of the entire region.