unit nonresponse , item nonresponse , missing data mechanism , longitudinal studies , response continuum , total survey error , interviewer effects
Abstract:
While survey data are commonly used in political and economic decision-making, their validity can be threatened by missing data. Missing data can reduce the power and efficiency of study results and lead to bias in study results. If the respondents and the nonrespondents to the entire survey or to single survey questions systematically differ, serious bias may occur. In the worst case, this missing data can lead to poor decision-making. This dissertation examines the origins of missing data in order to identify opportunities to prevent its occurrence in the future, thereby minimizing the risk of poor decision-making. I used data from a longitudinal interviewer-administered survey, the Survey of Health, Ageing and Retirement (SHARE), which is widely used in EU policymaking. My study examined two common sources of missing data that are likely to occur non-randomly: nonresponse to the entire survey and nonresponse to financial questions. This thesis argues that in SHARE, these two sources of missing data are connected to each other and to the interviewers. I identified opportunities to prevent missingness by addressing these connections, laying the ground for future researchers to explore this under-researched area.
Dieser Eintrag ist Teil der Universitätsbibliographie.
Das Dokument wird vom Publikationsserver der Universitätsbibliothek Mannheim bereitgestellt.