Could Nonresponse Be Biasing Trends of Health Estimates?

• Activity Number: 419
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: SSC
• Author(s): Beatrice Baribeau*+
• Companies: Statistics Canada
• Keywords: Nonresponse ; bias ; health survey ; telephone frame

Abstract

 Statistics Canada's household survey, the Canadian Community Health Survey (CCHS), has experienced steady declines in response rates. The response rate threshold has not been met the last two years, with telephone interviewing being particularly low. Nonresponse is on the rise and the effect of nonresponse bias is a concern. In this study, the impact of nonresponse bias on estimates is simulated through the 2007 CCHS telephone respondent data. The group representing nonrespondents (the late responders) are shown to have characteristics that are different than the remaining respondents, thus affecting health estimates. Furthermore, this study demonstrates these differences cannot fully be corrected through weighting. Under today's reduced response rates, if these late responders are no longer being captured, estimates and trends may be biased. These results are relevant for the survey's redesign as collection is shifting to modes with lower response rates.