The Importance of Nonuniformities in Measurement for the Appearance of Divergence

Abstract

Once we have applied the adjustments that appear useful for dealing with procedural anomalies and definitional incongruities between the NCS and UCR, the two series correspond much more closely. Nonetheless, major discrepancies between the series remain even when we restrict their scope to those crimes that are most comparable definitionally and procedurally. As social indicators, the two systems would sometimes have sent conflicting signals about crime during the period studied even had their reporting embodied the kinds of adjustments we have been able to make for these definitional and procedural differences. Undoubtedly, more precise and complete information on incidents that are out of scope in one series, but present in the other would further our understanding of these divergences. Some of their lack of accord may be due to the nonuniform application of definitions and procedures, to changes in procedures or to expedients that were adopted to facilitate data collection but which introduced series correlated measurement effects. For instance, UCR participating agencies that progressively report more multiple burglaries in a motel as separate incidents have misapplied the “hotel” rule and introduced nonuniformity of the first type. The substantial increase in telephone interviewing in the NCS since 1979 is an illustration of a procedural change that could affect series comparability. The decision to include data from “unbounded” interviews with residents of sampled units in making NCS estimates is illustrative of the third type of non-uniformity.