Adjusting Survey Estimates for Response Bias: An Application to Trends in Alcohol and Marijuana Use

In: Public opinion quarterly: journal of the American Association for Public Opinion Research, Band 62, Heft 3, S. 354-377

Abstract

Draws on retrospective self-report data on ages at first use of alcohol & marijuana from nine National Household Surveys of Drug Abuse, 1982-1995 (N = approximately 160,000), to show that estimates of alcohol & marijuana incidence (initiation) during early adolescence decline with increases in time interval between data collection & reference periods. The consistency of this finding by gender & across eight birth cohorts, interviewed at different ages & lengths of retention, supports an interpretation in terms of retrospective reporting bias. An exponential decay model is applied to adjust estimates for response bias & use the model to show how bias distorts trends in alcohol & marijuana incidence, 1961-1990. An analysis of changes in lifetime-incidence & age-at-first-use reports of birth cohorts as they age suggests that forward telescoping accounts for most underreporting of early alcohol use, & intentional concealment accounts for most underreporting of marijuana use. 4 Tables, 2 Figures, 1 Appendix, 57 References. Adapted from the source document.