NCJ Number
211569
Journal
Journal of Quantitative Criminology Volume: 21 Issue: 3 Dated: September 2005 Pages: 321-343
Date Published
September 2005
Length
23 pages
Annotation
This paper discusses the "bounding" procedures used by the National Crime Victimization Survey (NCVS), explains the NCVS and School Crime Supplement (SCS) data and analytical methods, and examines the effects of bounding and mobility on reports of victimization.
Abstract
Although criminologists are primarily interested in the substantive differences in victimization, they must also contend with response errors in these data. This becomes more problematic when the substantive and methodological causes are confounded and cannot be easily separated. One common example of this problem is the inclusion of replacement respondents in NCVS victimization estimates. These replacement respondents are individuals who move into an NCVS household after the initial "bounding" interview has been conducted. Although the use of replacement respondents is necessary to prevent a substantial loss in sample size, it is a source of measurement error in the survey. This is because initial interviews are more likely to elicit victimization reporting than subsequent interviews; however, "movers" are more likely than nonmovers to experience victimization. Thus, victimization reports by replacement respondents may be biased due to their unbounded interview status (response error), mobility (actual experiences), or some combination of the two. The nature of the error introduced by these factors is unknown and is the focus of this study. Three unique data features enable this study to disentangle the effects of mobility from those of unbounded interview status: a source of incoming respondents' data, a measure of mobility status for the incoming respondents, and the ability to study individual respondents longitudinally. Using data from the 1995, 1999, and 2001 NCVS and their corresponding SCSs, this study determined that both bounding and mobility influenced reporting; however, this influence was inconsistent; bounding only had significant effects on reports of property victimization; and moving only significantly affected reports of violent victimization. Further research is recommended. 5 tables and 25 references