The consequences of respondent attrition in panel studies: A simulation based on the Rochester youth development study

Show other versions (2)

Subject attrition is a potentially serious threat to the validity of inferences drawn from panel studies of delinquency and drug use. Prior assessments of this issue produce somewhat conflicting results, with some finding that respondents who leave the panel have higher rates of delinquency and drug...

Full description

Saved in:  
Bibliographic Details
Main Author: Thornberry, Terence P. (Author)
Contributors: Bjerregaard, Beth ; Miles, William
Format: Electronic Article
Language:English
Published: 1993
In: Journal of quantitative criminology
Year: 1993, Volume: 9, Issue: 2, Pages: 127-158
Online Access: Volltext (lizenzpflichtig)
Volltext (lizenzpflichtig)
Journals Online & Print:
Drawer...
Check availability: HBZ Gateway
Keywords:
longitudinal research
Delinquency
panel study
subject attrition
Description
Summary:Subject attrition is a potentially serious threat to the validity of inferences drawn from panel studies of delinquency and drug use. Prior assessments of this issue produce somewhat conflicting results, with some finding that respondents who leave the panel have higher rates of delinquency and drug use than those who remain, while other studies report little or no differential attrition. Despite these findings, there has been virtually no examination of the extent to which respondent attrition can bias substantive findings in panel studies of delinquency and drug use. The present article addresses this issue by simulating higher levels of attrition in an ongoing panel study that has a low rate of attrition and little differential attrition. It finds that failure to include more elusive respondents (those who are more mobile) would bias estimates of prevalence and frequency of delinquency and drug use as well as results from basic regression analyses. Failure to include less cooperative respondents (those who require more contacts) produces similar, but somewhat smaller, differences. The methodological implications of the results are discussed.
ISSN:1573-7799
DOI:10.1007/BF01071165