NCJ Number
99718
Journal
Journal of Quanitative Criminology Volume: 1 Issue: 3 Dated: (September 1985) Pages: 227-240
Date Published
1985
Length
14 pages
Annotation
Logistic regressions of age-period-cohort models for city arrest rates were estimated with data from seven cities for the years 1970-80 to examine the dependence of officially designated criminality in selected offense categories on age.
Abstract
Analyses of goodness of fit for one-, two-, and three-factor models indicate that the three-factor model provides the best fit. Age parameters show a decline in arrests from age 15-19 to age 20-24, followed by an increase at age 25-29 and a slow decline thereafter. Age parameters were different for different offense categories. Property crime arrests decline steadily from the age 15-19 maximum. Drunken driving rises to age 35-39, then declines; violent crime shows a similar pattern but declines more slowly with age. Murder and rape peak at age 15-19 then decline, while age parameters for aggravated assault rise steadily to age 35-39. Auto theft, burglary, and robbery show patterns similar to those for murder and rape, while larceny declines to age 20-24 and then remains flat to age 55-59. While these results appear striking at face value, a sensitivity analysis indicates that the parameter estimates are highly sensitive to model specification. Implications of this sensitivity are discussed. Included are 33 references.