Generic Areas and Small Area Estimates for Subnational Estimation on the National Crime Victimization Survey

• Activity Number: 181 - Contributed Poster Presentations: Government Statistics Section
• Type: Contributed
• Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Philip Lee* and Marcus Berzofsky and Lynn Langton and Michael Planty
• Companies: RTI and RTI International and Bureau of Justice Statistics and RTI International
• Keywords: Generic Areas ; Small Area Estimates ; NCVS ; National Crime Victimization Survey

Abstract

The National Crime Victimization Survey has historically produced nationally-representative criminal victimization estimates for major types of crime in the US. However, there is interest in understanding local level crime problems and examining state and local variations in crime rates. To address this demand, BJS is developing a portfolio of both direct and indirect estimation procedures. Two broad methods that can be used to produce subnational estimates from the existing sample when sample is insufficient for reliable estimates: (1) Generic Areas and (2) Small Area Estimates (SAE). Generic areas can be constructed using variables available on the NCVS public use file (region, population size, and urbanicity) allowing users to dynamically produce subnational estimates; however, estimates for specific areas are unknown. SAEs use existing data to model subnational estimates for specific areas. SAEs offer victimization rates for a specific state or MSA but are static estimates. We discuss the advantages and disadvantages of each method. In addition, through graphics, we demonstrate and show examples of the types of estimates produced under each approach.