Revealing Victimization: The Impact of Methodological Features in the National Crime Victimization Survey.

This study examines the impact of methodological features of the National Crime Victimization Survey (NCVS) on respondent willingness to report violent, serious violent, and property victimizations to the NCVS. Bounded and unbounded data from the 1999-2005 NCVS are used to create a longitudinal file of respondents, and survey-weighted logistic regression models are used to assess the factors associated with the reporting of victimization. Net of sociodemographic control variables, unbounded interviews produced higher estimates of serious violence (72%), violence (66%), and property victimization (67%). Mobile respondents reported higher estimates than nonmobile respondents of serious violence (48%), violence (35%), and property victimization (15%). Compared with in-person interviews, interviewing by telephone increased reporting for serious violence (7%), violence (12%), and property victimization (17%). This study highlights the importance of controlling for these factors in both longitudinal and cross-sectional analyses to estimate victimization risk.