Home

CHANGING VALUES OVER THE LAST HALF-CENTURY:
THE STORY OF THURSTONE’S CRIME SCALES

 David J. Krus, Jay L. Sherman, and Patricia H. Kennedy
Arizona State University 

 Summary.—The present article is a follow-up of Thurstone’s study, published 50 yr. ago, concerning the perceived seriousness of crimes. Comparisons were also made with Coombs follow-up study from the mid-sixties. The continuing trend toward the judgment of sex offenses as less serious and offenses against the person as more serious was observed. The obtained scale also showed heightened sensi­tivity of subjects to offenses such as burglary, libel, counterfeiting, and forgery, re­flecting the prevailing public opinion during the political trials of the mid-seventies.

In the late twenties, Thurstone (192 7b) had subjects choose between all possible pairs of 19 crimes and offenses. Using the method of pair comparisons, he built a scale of crimes from the more serious as rape and homicide to the less serious as bootlegging and vagrancy. This study was one of the earliest quantitative studies of attitudes and was replicated 40 yr. later by Coombs (1967), who found changing patterns of attitudes reflecting changes in societal structure as evolved during the intervening years. The present study is a 50-yr-later attempt to continue Thurstone’s legacy. The analysis of contemporary data was complemented by reanalysis of the preceding studies using a more powerful computational model. The reanalysis of the data collected earlier also permitted the direct comparison between scales obtained from all three studies.

 METHOD

The list of 171 pairs of nineteen offenses was administered to 209 undergraduate students at the University of Southern California. The instructions were identical to those used by Thurstone (1927b, p. 386) with two exceptions, i.e., first, instructions pertaining to the use of optically scannable answer sheets were given and second, the list of crime definitions was changed to conform with the contemporary legal code. The number of omissions did not exceed 4.8% for any pair of offenses. The slight elevation of omissions (1.9% for Thurstone’s study and 1.7% for Coombs study) was possibly introduced by the use of mechanically scored answer sheets. Although the sample of students at the Univer­sity of Southern California was only superficially matched with Coombs University of Michigan subjects and with Thurstones University of Chicago sample, the home background of USC students was varied enough to compensate for possible regional fluctuations. The student responses were tabulated (Table 1) and analyzed by a computer program for normal scaling of unidimensional dominance matrices in the pair-comparisons mode (Krus, 1976).

 TABLE 1

Proportion of Subjects Who Judged Column Offense More Serious Than Row Offense

Thurstone’s (1927b, p. 399) original data were also included in the analysis, as were Coombs’ (1967, p. 86) data from the earlier replication of the experiment. All three sets of data were analyzed using Case III of the law of comparative judgment (Thurstone, 1927a). Obtained scales were area transformed to T scales (McCall, 1922) to permit inter-distributional comparisons (Krus & Krus, 1977).

 RESULTS AND DISCUSSION

The obtained scale values of judged seriousness of 19 offenses and their corresponding discriminal dispersions are listed in Table 2. The Case III of the law of comparative judgment was used in construction of all three scales.

TABLE 2

Obtained Case III Scale Values of Judged Seriousness of Offenses

and Their Corresponding Discriminal Dispersions

Since it is not possible to compare the scale values at this point, it is instructive to observe the changes in the discriminal dispersions across time as the criminality of an offense became more controversial, the variability of its severity judgments increased as exemplified by increased variability in the contemporary judgments of the seriousness of abortion and seduction, and by the variability in judgments of bootlegging at the height of the prohibition years. The exception from this general pattern was the case of homicide, where the variability of judgments probably reflected the uncertainty of judges with respect to the intent, caused by the failure to include murder as one of the rated crimes. This omission can be explained post hoc by the difficulty Thurstone had with handling the indeterminate cases of 1.00—0.00 proportions (cf. Edwards, 1957, pp. 40-46; Krus, 1976). The T scale values for all three studies are listed in Table 3 and graphically plotted in Fig. 1.

 TABLE 3
Normalized Case III Values of Judged Offenses

Thurstone’s original classification of the offenses into sex offenses, injury to the person, and property offenses was preserved by listing the first two groups at the left of each scale and the last group of property-related offenses to the right.

The trend, originally observed by Coombs, continued. Sexual offenses perpetuated their descent and crimes against the person continued to be perceived as more serious. Homicide topped the list in 1966 as well as in 1976, and it appears that the Lindbergh case elevated kidnapping to the third place to stay.

 FIG. 1. Comparison of scale values of judged offenses for the 1926, 1966, and 1976 studies

 The rise of smuggling possibly reflected the increased perception of this of­fense as related to drug traffic. Also, the general elevation of crimes such as burglary, counterfeiting, libel, and forgery reflected heightened public sensi­tivity to these offenses in the aftermath of Watergate.

 Footnote
Computer reanalysis of Thurstone’s (1927 b, p. 399) original matrix of proportions revealed errors in the homicide-arson and homicide-assault and battery cells. Reported values of .083 and .030 were changed to .983 and .930 since the complementary values reported in the table were .017 and .070, respectively. However, we are not certain that these were the actual obtained values.

 REFERENCES

Coombs, C. H. Thurstone’s measurement of social values revisited forty years later. Journal of Personality and Social Psychology, 1967, 6, 85-91.

Edwards, A. L. Techniques of attitude scale construction. New York: Appleton-Century-Crofts, 1957.

Krus, D. J. Fortran program for the domain-referenced scaling of the unidimensional dominance matrices. Arizona State University, University Testing Services: Tech­nical Reports, 1976, 8, 1-35.

Krus, D. J., & Kennedy, P. H. Lost: McCall’s T scores: Why? Educational and Psychological Measurement, 1977, 37, in press.

McCall, W. A. How to measure in education. New York: Macmillan, 1922.

Thurstone, L. L. A law of comparative judgment. Psychological Review, 1927, 34, 273-286. (a)

Thurstone, L. L. The method of paired comparisons for social values. Journal of Abnormal and Social Psychology, 1927, 21, 384-400. (b)