Abstract:
|
Repeated cross-sectional surveys try to maintain continuity in their instruments to allow for trend analysis. However, changes are necessary over time to adapt to improved methodology and to update to current circumstances. Changes may include certain questions being phrased differently across time, the addition or removal of particular response options, or the reorganization of the questions themselves. When conducting trend analysis in repeated cross-sectional surveys, it is necessary to account for potential measurement error spawning from such changes.
The Survey of Prison Inmates (SPI) sponsored by the Bureau of Justice Statistics has been conducted seven times since 1974. The survey provides information about U.S. prisoners on a wide range of topics, such as demographics, current offenses and sentences, criminal history, gun possession and use, drug use and treatment, and mental health history and treatment. In this paper, we assess how similar constructs have changed in the SPI instrument over time and then discuss how those changes impact survey estimates and their comparability. Our analysis will specifically focus on comparing the most recent survey year, 2016, to the most recent prior version, 2004.
|