U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Recidivism and Research Design - Limitations of Experimental-Control Research (From First National Conference on Criminal Justice Evaluation - Selected Papers, P 279-291, 1981, Joel H Garner and Victoria Jaycox, ed. - See NCJ-82918)

NCJ Number
82925
Author(s)
R Martinson; J Wilks
Date Published
1981
Length
13 pages
Annotation
Preliminary procedures and findings are presented for a study that is examining research methods and outcomes of studies focusing on recidivism rates, with a view toward developing an analytical procedure which cuts across projects while making use of the differences among studies.
Abstract
The present report is based on 7,341 recidivism rates taken from slightly over 470 documents. Some preliminary comparisons of experimental and nonexperimental research have raised a host of issues which will be explored. Two key concepts being used in the study are the 'batch' and the 'computable' recidivism rate. A batch is any number of persons at some specifiable status in the criminal justice system for whom a proper recidivism rate is computable. The primary unit of analysis in the survey is the computable recidivism rate, which specifies what proportion of any batch shall be identified as recidivists according to some operational definition of recidivism used by the researcher. The mean of all the recidivism rates of all the studies examined is 23.3 percent over studies conducted over the last 40 years or so. The research designs of the studies have been placed in three categories: matched or random allocation of subjects ('random'), the remainder of the experimental or quasi-experimental research ('nonrandom'), and the after-only studies without the use of control groups ('ex post facto'). Tabular data present the distribution of recidivism rates by type of research design for varying periods of followup. Other tables show the distribution of recidivism rates of research design categories for various definitions of recidivism and the distribution of rates by design and subject status in the criminal justice system. Questions remaining to be answered by the study are identified.