NCJ Number
85331
Editor(s)
J A Fox
Date Published
1981
Length
206 pages
Annotation
This series of essays presents some of the newer methodological mathematical models appearing in the criminology literature, indicates how regression analysis can be useful in criminological studies if correctly employed, and exhibits the use of time series data for forecasting and evaluation.
Abstract
The first essay discusses panel models, an approach that uses data gathered both longitudinally and cross sectionally to permit casual interpretations that neither of the two dimensions alone can provide, while the second presentation considers the long-linear model. One particular log-linear approach to categorical data is described and illustrated. Through a reanalysis of homicide data, the array of possible hypotheses that become testable with this approach is demonstrated. The next contribution describes a category of survey designs, termed randomized response, that offers survey respondents the protection essential in sensitive inquiries while still providing sufficient data analytic capabilities for the researcher. Another essay addresses the problems of response errors and response bias in surveys. The next three presentations demonstrate how one basic tool, regression analysis, can be most fruitful if used with restrained imagination. The final two essays which demonstrate the use of time series data for forecasting and evaluation, compare several methods of forecasting crime data and demonstrated the use of a model that measures the impact of policy interventions. Tabular data, mathematical equations, and references accompany each presentation, and a subject index is provided. For individual entries, see NCJ 85332-40.