This study used the Auto Regressive Integrated Moving Average (ARIMA) models developed by Box, Tiao, and Jenkins. Data for Case 1 were UCR burglaries in a major southwestern city (City A) between January 1975 and May 1981. Detectives investigated all complaints for a 21-month period, during which time the level of UCR burglary series dropped abruptly. The detectives were more adept than patrol officers in applying formal definitions so that coding errors and consequently the official crime rate declined. Case 2 used monthly UCR burglaries for the same period of another large southwestern city where the series followed a slight downward trend prior to July 1979, but subsequently rose substantially. This was caused by a breakdown in hierarchical authority during the summer of 1979 when the police chief retired and an administrative shakeup replaced other departmental administrators. Using data from City A, the third case illustrates a typical screening procedure where sergeants in one police department routinely screened out complaints that they understood to be marginal. Because the sergeants relied on complainant characteristics, particularly race, to screen out cases, complaints from ghetto areas were less likely to result in dispatched service calls. This in turn reduced the official crime rate. In conclusion, UCR statistics are noncomparable across jurisdictions and may be used validly only for studying organizational processes. Tables, formulas, 4 footnotes, and 21 references are included.