Sensitivity of Differences in Survey Conditions on Estimates of Criminal Victimization

Brian Wiersema, University of Maryland at College Park

ABSTRACT
Periodically, changes to the way the National Crime Victimization Survey (NCVS) is conducted are suggested, mostly to reduce cost and increase flexibility. Some recent suggestions have included eliminating the bounding interview, switching to an all-telephone model of administration, and lengthening the reference period. While one can hypothesize about the general effects of such changes, opportunities to measure them are rare. The availability of area-identified NCVS and the 12-city telephone surveys sponsored by the Justice Department's COPS office provide such an opportunity. The COPS surveys were based on the NCVS questionnaire and were fielded by the Census Bureau, but they differed from the standard NCVS design in terms of sampling design, initial contact, administration mode, reference period and founding procedure. While the effects of each of these differences cannot be separately measured, taken collectively, they offer an important chance to compare the effects of major design differences on recent victimization estimates.

(Return to Program Resources)

Updated 05/20/2006