Survey of Projection Techniques

• NCJ Number: 79728
• Author(s): J C Warmbrodt
• Date Published: 1980
• Length: 27 pages
• Corporate Author: Kentucky Dept of Justice
• Sale Source: National Institute of Justice/; Kentucky Dept of Justice
• Publication Type: Survey
• Language: English
• Country: United States of America

Annotation

Based on questionnaires completed by 49 States and the District of Columbia in 1980, this report provides an overview of methodologies currently being used to project prison populations and compares these findings with a similar survey conducted in 1977.

Abstract

The Kentucky Bureau of Corrections' Research and Evaluation Unit mailed questionnaires regarding prison population projections to corrections agencies in 50 States and the District of Columbia and received replies from all except Nebraska. Analysis of these data revealed that no single methodology has been adopted by a majority nor has any technique proved most reliable. The projection methodologies fall into these groups: linear regression, ratio, multiple regression/regression analysis, and simulation model. The numbers of States using each method are listed, along with a description of the advantages and limitations of the respective techniques. Uses of the projections were fairly standard and included budgeting, determining capital development needs, and aiding in policy and program planning. A discussion of the information used to generate projections and their reliability notes that most States rated their projections' reliability as fair to good, but that these ratings improved as the sophistication of the projection techniques increased. No consistent relationship between the size of the inmate population and the methodology used or reliability of projections was found. Comparisons with a 1977 survey of projection techniques conducted by the Florida Department of Offender Rehabilitation showed that the use of increasingly sophisticated projection methodologies had expanded. However, the general dissatisfaction with reliability indicates that the science of prediction actually made little progress in 3 years. The most promising approach is the simulation model, largely because it considers so many variables. Tables, 8 footnotes, and a chart summarizing each State's projection methods are included.