|
Abstract:
|
The Omnibus Household Survey is a random-digit-dial survey of slightly more than 1,000 non-institutionalized residents in telephone households in the U.S. Information is collected to monitor expectations of and satisfaction with the transportation system. Since the response rate for the survey is approximately 50%, there has been considerable interest in gauging the nonresponse bias in this survey. There are mainly three options for assessing nonresponse bias. The first option, used frequently, and hence widely accepted, is an intensive follow-up of nonrespondents. This was not an option here because of cost. The second option is external validation -- benchmarking against some other data, and this was done here for commonly used demographic variables. Another option is to use data on survey operations. This presentation discusses results from using call history data to estimate nonresp
|
