Assessing the Coverage and Reliability of Subnational Geographic Identifiers in the NCVS Public-Use File

• NCJ Number: 249467
• Author(s): Bonnie E. Shook-Sa; Philip Lee; Marcus Berzofsky
• Date Published: December 2015
• Length: 57 pages
• Agencies: BJS-Sponsored
• Publication Type: Research (Applied/Empirical); Report (Study/Research); Report (Grant Sponsored)

Annotation

The primary purpose of this study was to assess the coverage and reliability of the National Crime Victimization Survey (NCVS) sample in the sub-national geographic areas created from the public-use files (PUFs), i.e., census region, population size, and urbanicity.

Abstract

Ideally, the Bureau of Justice Statistics (BJS), which administers the NCVS, wants to create generic areas based on the aforementioned three variables and examine patterns and trends in victimization rates across these areas. Before releasing generic-area estimates, however, it was necessary to first examine the coverage and reliability of the samples in these areas, so as to determine the appropriate survey weights to use in estimation and also assess the best variance estimation method for producing reliable estimates. The evaluation consisted of three parts: an assessment of coverage, an assessment of re-weighting methods, and an assessment of the reliability of the generalized variance function (GVF) estimates for generic areas. Section 2 of this report describes the assessment of coverage within generic areas formed on the basis of two- and three-variable crosses of census region, population size, and urbanicity. Section 3 discusses the approach for re-weighting the NCVS sample within generic areas and the analysis of the effects of re-weighting on key NCVS estimates and precision. Also discussed in this section are recommendations on the collapsing of levels within generic areas and pooling data across multiple years to achieve adequate precision. Section 4 assesses the reliability of GVF estimates within generic areas. Section 5 summarizes key findings across the analysis and presents recommendations for calculating estimates within the generic areas formed with the three sub-national variables. 25 tables, 8 figures, and 2 references