Panel Conditioning Effects in Longitudinal Social Science Surveys

点击翻译按钮获取中文摘要

Longitudinal (or panel) surveys provide great methodological leverage in establishing causality and in understanding processes that unfold over time. Despite their tremendous value for research in innumerable academic and applied fields, longitudinal surveys also present a variety of unique methodological problems. Among the least well understood of these problems is known as "panel conditioning," or bias introduced when participation in one wave of a longitudinal study alters respondents' subsequent reports of their attitudes and/or behaviors. If partaking in a longitudinal survey alters participants' responses to subsequent attitudinal and/or behavioral survey questions, then this calls into question the validity of information derived from any number of widely used data resources. The investigator will use an experimental research design with corrections for panel attrition and data on a wide range of attitudinal and behavioral measures to test a series of theoretically derived hypotheses about the circumstances under which conditioning effects should be most severe. The basic research design for assessing the magnitude of panel conditioning effects involves conducting a mail survey of individuals who were assigned, at random, to either a treatment group that has been interviewed repeatedly since 1988 or to a control group that has never been interviewed. The precise nature of the "treatment" is complicated by differences across measures in how frequently they have appeared on the annual longitudinal surveys and by heterogeneity across respondents in how frequently they have participated in those surveys.Major federal, local, and private funding agencies have spent many millions of dollars in recent decades to invest in longitudinal panel surveys. This investment is motivated by the desire to better understand causal processes and to learn more about processes that unfold over time. Nearly all social science disciplines rely on such data. Although researchers have carefully developed research methods and statistical techniques to account for some problems associated with longitudinal research (such as panel attrition), very little is known about panel conditioning. The results of this project will vastly improve our understanding of the magnitude of panel conditioning effects as well as our knowledge about the circumstances under which users of longitudinal survey data should be concerned about this form of bias.