A Practical Measure of Offence Seriousness: Sentence Severity, continued
The problem of distortion of measures becomes magnified when, assuming interval level scaling, the attempt is made to add Crime Severity weights and to compare summed scores across offence types. According to the National Survey of Crime Severity (NSCS) values, minor injury inflicted in four or five offenses (seriousness score is 7.29 to 8.50), or three perjury, or four false statements, would be equivalent to one murder (seriousness score is 35.71). It seems hard to justify these equivalences. In fact, Sellin and Wolfgang (1964) did not ask their subjects to rate the seriousness of the components of offenses described. Wellford and Wiatrowski (1975), it is true, found a near-perfect correlation (r = .969) between scores obtained from their study sample of ratings of complex events and the sums of ratings of these events’ components. Correlation, in this instance however, does not reveal whether the scores obtained were equal or merely in serial concordance with each other. Pease, Ireson and Thorpe (1973) and Wagner and Pease (1978) found that only between 18% and 32% of their respondents actually rated two offenses as twice as serious as one offence. Their results suggest that some people are inclined to under- estimate a multiplicity of offence counts, while others assign incremental importance to a plurality of counts as compared to single or isolated instances of crimes.
It would appear worthwhile, therefore, to examine the additivity of weightings from the National Survey of Crime Severity (NSCS). Some of the weightings assigned for combined events are almost identical with the sums of the weightings for their components. Murder and robbery combined are weighted at 43.24 and sum from their components to 43.69. Weapons dangerous and robbery yield a combined figure of 9.72 and a summed figure of 10.42. Some other values show considerable variations between their ratings when combined and the sum of ratings for component events. Rape combined with murder is weighted at 52.80, but summed from the components yields a weight of 61.50. Robbery and assault in a single crime is rated at a value of 14.60,but when summed from the components yields a weighting of 19.93. Break, enter and theft is valued at 9.60, but when handled as components sums to 6.81. It seems likely that internal inconsistencies in the scale arise from the absence of a means by which to anchor the subjective ratio scaling in a physically distributed measure.