Role of Statistical Models in Planning Juvenile Corrections Capacity

• NCJ Number: 196970
• Author(s): Daniel P. Mears
• Date Published: July 2002
• Length: 29 pages
• Grant Number(s): 98-JB-VX-K004
• Sponsoring Agency: Office of Juvenile Justice and Delinquency Prevention (OJJDP)
• Sale Source: The Urban Institute
• Publication Format: Document
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America
• Note: Downloaded September 24, 2002

Annotation

This document examines the usefulness of relying on explanatory statistical models to inform juvenile correctional bed-space planning.

Abstract

Policymakers want forecasts that can help them anticipate future capacity needs because building and operating juvenile correctional bed space is costly. One forecasting approach consists of statistically modeling factors that may account for differential rates of juvenile incarceration. There are several hypotheses about the potential role of certain factors in driving State juvenile correctional capacity. These factors are that States use different upper age boundaries for defining whom is or is not a juvenile. Economic conditions vary across States. Investments in education or levels of educational achievement may reflect the extent to which a State is youth-oriented. Political factors affect juvenile incarceration rates since State juvenile correctional policy is directly related to legislative funding decisions. Social factors that might contribute to juvenile incarceration rates are the number of single-parent families, teen birth or death rates, minority populations, or greater population density. To examine these hypotheses, data on State juvenile incarceration rates and measures of the different independent variables were collected. Ordinary least squares (OLS) regression was used to examine the relationship between State-level juvenile incarceration rates and a set of predictors. Results show that statistical approaches to forecasting can provide the best guidance but these approaches suffer from many problems. Statistical analyses rarely explain much of the variance in past incarceration trends. They rely on assumptions or projections about predictors of future incarceration trends, which may be in error. They cannot adequately incorporate information about changing social and political conditions. Statistical models may be most useful when they provide general guidance about how factors are potentially linked to juvenile correctional bed-space needs and capacity. If forecasts are to be more accurate, credible, and useful, statistical analyses must be part of a more general and ongoing adaptive forecasting process. 6 figures, 8 tables, 28 references