![IconGems-Print](images/IconGems-Print.png)
295 – SPEED: Big Data, Small Area Estimation, and Methodological Innovations Under Development, Part 1
Modifying State Sample Sizes for the National Crime Victimization Survey
David Hornick
U.S. Census Bureau
Sandra Peterson
U.S. Census Bureau
Samantha Spiers
U.S. Census Bureau
The National Crime Victimization Survey historically was designed to produce an annual national estimate of crime victimization in the United States. Beginning in 2016, the sample was redesigned and increased to support state-level estimates for the largest 22 states (called “boost states”) using three years of data. The national and state-level sample sizes were calculated before data collection began, based on the critical assumption that each state-level estimate would match the national estimate. State-level sample sizes also were calculated assuming a national design effect based on historical survey results. We examine the impact of these assumptions on state-level sample sizes by comparing current sample sizes to ones calculated using state-specific victimization estimates and design effects. We also discuss the various formulas considered when calculating design effects, and the effect of the victimization estimate on them.