NCJ Number
152549
Journal
Journal of Research in Crime and Delinquency Volume: 31 Issue: 4 Dated: (November 1994) Pages: 434-463
Date Published
1994
Length
30 pages
Annotation
Most of the methods used in criminology to infer relationships are based on mean values of distributions; this essay explores the historical origins of this issue and some counterproductive consequences.
Abstract
This essay suggests that we can and should do a great deal to improve the state of research in criminology and criminal justice. Collecting a large data set, pouring it into a computer, and turning an analytic crank may provide results that are statistically significant and, perhaps, even worthy of publication, but more can be done with the data we collect to improve the state of our knowledge about crime and its correction. Part of the reason for this, the author argues, is historical. Criminologists have been conditioned to assume that a single data set will produce a single pattern that can be characterized by mean values. Data sets should be analyzed to see if more than one type of individual is contained within the data, if more than one behavioral mode is manifested, and if more than one outcome might result from a treatment. Also, the increased availability of high-speed, high-capacity computers and excellent data analysis and graphics programs means that researchers can let the data speak for themselves. Also, researchers should consider the possibility that the data researchers currently collect and the categories currently used may be lacking. Attempts should be made to collect different kinds of data, using more relevant categories, and recognizing that this may require the application of judgment by data collectors. 23 notes and 91 references