NCJ Number
58154
Date Published
Unknown
Length
5 pages
Annotation
INTERVAL ESTIMATION IS INTRODUCED IN THIS PROBABILITY MODEL FOR CRIME ANALYSIS. IT IS ASSUMED THAT INCIDENTS OF CRIME HAVE A POISSON DISTRIBUTION, WITH SOME UNKNOWN PARAMETER VIEWED AS A MEASURE OF CRIMINALITY.
Abstract
IN TERMS OF PERCENTAGE CHANGES, CRIME VOLUMES FOR LARGER AGENCIES TEND TO BE MORE STABLE THAN CRIME VOLUMES FOR SMALLER AGENCIES. THIS MEANS THAT AN AGENCY'S CRIME CONDITIONS CAN BE BETTER DESCRIBED IF CONVENTIONAL PERCENTAGE CHANGE ANALYSIS IS AUGMENTED BY ADDITIONAL INFORMATION. USING A BAYESIAN APPROACH, THE PROBABILITY MODEL REFINES CONVENTIONAL PERCENTAGE-CHANGE ANALYSIS BY INTRODUCING INTERVAL ESTIMATION. CRIMINALITY IS EXPRESSED BY A GAMMA DISTRIBUTION, WITH THE OBSERVED VOLUME OF CRIME AS A PARAMETER. ASSUMING THAT THE LEVEL OF CRIMINALITY IN TWO CONSECUTIVE PERIODS IS INDEPENDENT, AN UPPER BOUND AND A LOWER BOUND FOR PERCENTAGE CHANGES ARE EXPRESSED USING A BETA FUNCTION. THESE UPPER AND LOWER BOUNDS, WHICH GIVE RISE TO AN INTERVAL ESTIMATION FOR PERCENTAGE CHANGES, ARE FUNCTIONS OF CRIME VOLUME, PERCENTAGE CHANGE, AND THE PRESCRIBED LEVEL OF CONFIDENCE. AN ASYMPTOTIC APPROXIMATION IS INTRODUCED FOR THE PURPOSE OF COMPUTATIONAL SIMPLICITY. AN EXAMPLE OF THE MODEL'S APPLICATION IS DEMONSTRATED IN UNIFORM CRIME REPORTS OF SMALL AGENCIES WHERE INSTABILITY OF THE PERCENTAGE-CHANGE VARIABLE OFTEN PRESENTS PROBLEMS RELATED TO DATA INTERPRETATION. THE PROBABILITY MODEL HAS ALSO BEEN APPLIED TO THE DEVELOPMENT OF A DATA QUALITY CONTROL METHOD FOR UNIFORM CRIME REPORTS. EQUATIONS AND DATA ON THE MODEL'S APPLICATION ARE INCLUDED. (AUTHOR ABSTRACT MODIFIED--DEP)