Abstract:
|
Much of the research into item nonresponse for income questions has focused on in-person, mail, and phone surveys, which often have missing rates of 20% or even higher (Moore, Stinson & Welniack, 1999). We explore factors related to income nonresponse in online panel surveys, in which respondents' willingness to answer sensitive questions may be affected by their long-term relationship with the panel. For this research, respondents were randomly assigned to one of three versions of a household income question to test various ways of presenting the response options to a close-ended question. We examine correlates of missing income data and report the results of the experiment, identifying the version with the lowest rate of missingness. The results demonstrate that the inclusion of a "don't know" response and how it is presented have significant implications for the rate of missing. In addition to reporting the nonresponse rate for each of the experimental conditions, we investigate the validity of the data collected under each condition.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.