Effects of Unbounded Interviews, Time-in-Sample, and Recency on Reported Crimes in the National Crime Victimization Survey

• Activity Number: 140
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Robert E. Fay*+ and Jianzhu Li
• Companies: Westat and Westat
• Keywords: Response error ; recall error ; telescoping

Abstract

The overall design of the National Crime Victimization Survey (NCVS) has been largely stable for over 30 years. Households in sampled housing units are interviewed for 7 waves, each collecting data for the previous 6 months. Until recently, the 1st wave has been omitted from the published estimates to exclude reports outside of the intended 6-month reference period. Using publicly available data for 1998-2004, we report on a series of analyses to investigate more current effects of the bounding and find evidence of more general time-in-sample effects. We will also report on the effect of recency in the observed incident reports, where more crimes are reported in the 1st month preceding the interview date than each of the previous 5 months. These phenomena are important in considering a range of design options for the NCVS that would alter the reference period or panel design.