A Practical Measure of Offence Seriousness: Sentence Severity, continued
The Offence Seriousness Index has been investigated extensively (Blumstein, 1974; Bridges & Lisagor, 1975; Figlio, 1975; Gottfredson, Young & Laufer, 1980; Hindelang, 1974; Kelly & Winslow, 1973; Lesieur & Lehman, 1975; Rose, 1966; Wagner & Pease, 1978; Walker, 1978; Wellford & Wiatrowski, 1975), replicated internationally (Akman & Normandeau, 1968; Hsu, 1973; Normandeau, 1966; Rossi, Waite, Bose & Berk, 1974; Velez-Diaz & Megargee, 1971) and has gained wide recognition as an index of crime in society. It is probably the most sophisticated method available for measuring offence seriousness. Its use of behaviour descriptions means that, in principle, transferral across jurisdictions should be possible. However, it has some noteworthy limitations for purposes such as programme evaluation. Its utility following sentencing is limited because of the difficulty in transforming categories used in legal offence records into the behaviour descriptions employed as scoring criteria. When an offender is undergoing a programme to be evaluated, and is directly in contact with researchers, descriptions concerning his current or prior offenses are likely to be too sketchy to permit adequate scoring of pre-programme offence seriousness. Also, the offender’s own descriptions of offenses frequently do not correspond, even in gross detail, to those records which are available. By the time follow-up is undertaken, the only records available are file records which tend to contain only the category of offence, the number of counts and the sentence imposed. Even if police occurrence reports are available, they rarely contain the kind of detail needed to score offenses on the Offence Seriousness Index. Thus, for most evaluation research, the usefulness and accuracy of this scale in measuring offence seriousness is seriously limited by the quality and quantity of descriptive information available.
In addition to the impracticality of the Offence Seriousness Index for standard use in programme evaluation, the scale has another difficulty. Although it, and particularly its next generation development – The National Survey of Crime Severity (Wolfgang, Figlio, Tracy & Singer, 1985) – appear to approximate interval level scaling along which various types of offenses are distributed in a reasonable ordinal sequence, their base data are nominal data derived from different criminal acts or classes of behaviour. This creates some doubt that the scales could really qualify as interval scales to justify the addition of scores across various offence types. The researchers’ use of the psychophysical Power Law to establish a subjective scale of seriousness along which offenses might be measured relative to one another creates the impression that a physical scale of equal units underlies the subjective ratio judgements made by their subjects. The nature of this implicit underlying physical scale is conceived as the social harm consequences of various offensive acts. However, this idea really begs the question since it does not provide a physical scale against which to verify ratios in judgements made.
At the same time, Wolfgang, et al. (1985) have provided some information concerning the correspondence of seriousness judgements to a physical scale in incremental dollar loss associated with theft, robbery and arson crimes. At the lower seriousness values, there is some evidence of a fit between the ratios of subjective values and dollar figures (a 1:2:3 ratio expected for dollar amounts of $100, $1,000 and $10,000 or an equivalent series). The subjective weightings for breaking into a school and stealing $10 or $1,000, for example, were 3.08 and 9.72, respectively, roughly in the 1:3 ratio suggested by the psychophysical Power Law. By interpolation, stealing $100 in that setting should receive a weighting of 6.31. Similarly, the values for stealing property worth $100, $1,000 and $10,000 from outside a building were 3.59, 6.86 and 10.94, respectively, approximately in the ratio 1:2:3. However, the seriousness weights for breaking into a home and stealing $100 or $1,000 were 3.14 and 9.60, respectively, varying from the 1:2 ratio which would be predicted. Also, values for arson losses of $10,000, $100,000 and $500,000 received seriousness weights of 12.75, 24.86 and 22.90, respectively. It appears, therefore, as Wolfgang, et al. (1985) mentioned, that there is an attenuation in subjective seriousness values in the upper scale ranges.