426 – Data Quality
Quality and Measurement Error Assessment of Juvenile Interviews in the NCVS
Caroline W. Blanton
RTI International
Marcus E. Berzofsky
RTI International
Kimberly C. Peterson
RTI International
Philip Lee
RTI International
This paper presents a quality assessment of a critical subpopulation within a national survey. The National Crime Victimization Survey (NCVS), sponsored by the Bureau of Justice Statistics, is a survey of U.S. households which measures non-fatal crime victimization counts and rates as well as characteristics of victimizations. The survey interviews all persons 12 years and older living in sampled households and uses the same survey instrument for all respondents. Juveniles (those 12 - 17 years old) have a high potential for various types of non-sampling error including nonresponse and measurement error. Sources of these non-sampling errors may include parental monitoring affecting the interview, cognitive ability to understand the instrument, and lack of availability to participate in the survey. Moreover, if participation rates are low among juveniles, the precision of the estimates may be inferior to adult estimates. This paper conducts a secondary analysis of the 2006 - 2012 NCVS to assess the quality of juvenile estimates and discusses the reasons for potentially lower quality estimates. Descriptive analyses of paradata and survey data are presented and discussed.