A Practical Measure of Offence Seriousness: Sentence Severity , continued
Study 1 suggests that sentence severity is both a valid and a reliable measure of offence seriousness. A correlation coefficient can indicate two things: (a) the degree of functional relationship between two variables, or the extent to which one variable predicts the other variable (validity); and, by squaring its value, (b) the accuracy of the prediction (reliability). The high correlation obtained (r=.84) represents high predictive validity attenuated partly by sampling bias and partly by measurement bias — the two sets of variables were selected to match each other on a face valid basis by one expert, and the data were drawn from two different jurisdictions (U.S.A. and Canada). Given these two sources of error in the data, the observed degree of concomitance between the two scales is quite good. The hypothesis of a valid relationship between sentence severity and offence seriousness is supported, and each of the measures is a relatively reliable predictor of the other. Study 2 demonstrated that the strong mutual relationship between sentence severity and offence seriousness is only moderately attenuated (r=.68) by the methodological changes in the construction of the scale of measurements, i.e., the extended range of offenses, the inclusion of offenses regarded as jurisdictional variants, and the use of the single most representative crime description in determining offence seriousness. However, the lower correlation obtained in Study 2 suggests that jurisdictional differences in judgments of the seriousness of particular offenses (in combination with other factors) may contribute error to the measurement of seriousness. Two corollaries would follow. Firstly, judges in a local jurisdiction probably better reflect the local community’s concepts of seriousness than judges from other jurisdictions, and local estimates of offence seriousness based on the sentencing practices of local judges should be obtained whenever possible. Secondly, a replication of the present study in each jurisdiction is advisable before assuming that offence seriousness estimates obtained from other jurisdictions, e.g., NSCS estimates, would in all cases correspond closely to estimates obtained from data collected locally.
Comparison of Sentence Severity and Offence Seriousness
It might appropriately be said that, in seeking to find a measure of offence seriousness which is convenient to use, the present study has merely “re-invented the scales of justice”. Perhaps a convenient to use, simple, and valid psychometric scale of offence seriousness, in the form of sentence length, has been available all along to workers in correctional programme evaluation. We contend that the sentences imposed by judges amount to an application of the same method of psychophysical judgment used in the most popular and sophisticated method for scaling offence seriousness which is currently available (Wolfgang, et al., 1985). Indeed, in many ways, using sentences imposed by court judges has methodological advantages over the procedure of asking samples of lay raters to perform judgments equivalent to those made by court judges. These advantages include the possibility that raters, as compared to court judges, may undertake their ratings more capriciously, and may not be as familiar with the range and characteristics of criminal acts. Moreover, court judges have access to all of the standardizing and correcting means available in the justice system. Finally, unlike the raters in the NSCS study who were required to make psychophysical judgements in psychological space anchored only by a standard modulus, court judges have the advantage of having a socially and economically meaningful physical scale divided into equal units (days) along which to distribute their judgements of offence seriousness.