U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Estimating Crime Victimization in Large States and MSAs Through Reweighting: Evaluation and Methodology

NCJ Number
252701
Author(s)
Andrew Moore; G. Lance Couzens; Marcus Berzofsky
Date Published
April 2021
Length
138 pages
Annotation

This report examines the extent to which weighting for National Crime Victimization Survey (NCVS) data can be adjusted to produce reliable subnational violent- victimization statistics, and it also provides guidelines for aggregating NCVS data over multiple years to produce subnational estimates of interest.

Abstract

The NCVS was originally designed to obtain national-level estimates of criminal victimization, using a nationally representative sample of U.S. households. The estimates in the current report use NCVS data collected before the NCVS sample was expanded in 2016 to facilitate direct estimation in the largest 32 states. Without directly estimated NCVS data, this report develops a proof-of-concept that describes re-weighting methods that can be applied to produce subnational estimates for the years before the sample increase and reallocation. In some cases, the resulting estimates for subnational areas may not be as robust or accurate as estimates produced after the sample boost. The three subnational area types considered for the estimate are states, metropolitan statistical areas (MSAs), and cities. City estimates are not included because most cities were determined insufficient to produce estimates with reasonable precision. This report considers three types of analyses that can be conducted for subnational areas: 1) comparison of areas to the national average within a single time period; 2) comparison within a specific area during a single time period; and 3) comparisons within a specific area across time; however, the latter comparisons are not recommended and should be viewed with caution. 27 tables, 38 figures, and 16 references