Published online by Cambridge University Press: 27 December 2018
The treatment of white-collar offenders by the criminal justice system has been a central concern since the concept of white-collar crime was first introduced In general, it has been assumed that those higher up the social hierarchy have an advantage in every part of the legal process, including the punishment they receive as white-collar criminals. In a controversial study of white-collar crime sentencing in the federal district courts, Wheeler, Weisburd, and Bode contradicted this assumption when they found that those of higher status were more likely to be imprisoned and, when sentenced to prison, were likely to receive longer prison terms than comparable offenders of lower status. While they argued that results were consistent with “what those who do the sentencing often say about it,” their analyses failed to control for the role of social class in the sentencing process. In this article we reanalyze the Wheeler et al sentencing data, including both measures of socioeconomic status and class position. Our findings show that class position does have an independent influence on judicial sentencing behavior. But this effect does not demand revision in the major findings reported in the earlier study.
1 E. Sutherland, White Collar Crime: The Uncut Version (New Haven, Conn.: Yale University Press, 1983).Google Scholar
2 D. Black, The Behavior of Law (New York: Academic Press, 1976); A. Lizotte, “Extra-legal Factors in Chicago's Criminal Courts: Testing the Conflict Model of Criminal Justice,” 25 Soc. Probs. 564 (1978).Google Scholar
3 Wheeler, S. et al., “Sentencing the White-Collar Offender: Rhetoric and Reality,” 47 Am. Soc. Rev. 641, 658 (1982).CrossRefGoogle Scholar
4 Id.Google Scholar
5 S. Wheeler, K. Mann, & A. Sarat, Sitting in Judgment: The Sentencing of White Collar offenders (New Haven, Conn.: Yale University Press, 1988).Google Scholar
6 Wheeler, et al., 47 Am Soc. Rev. at 657.Google Scholar
7 Hagan, J. & Parker, P., “White Collar Crime and Punishment: The Class Structure and Legal Sanctioning of Securities Violations,” 50 Am. Soc. Rev. 302 (1985).CrossRefGoogle Scholar
8 Benson, M. & Walker, E., “Sentencing the White Collar Offender,” 53 Am. Soc. Rev. 295 (1988).CrossRefGoogle Scholar
9 Wheeler, et al., 47 Am. Soc. Rev. at 657. See also Hagan, & Parker, , 50 Am. Soc. Rev. at 302; D. Weisburd & E. Chayet, “Good Time: An Agenda for Research,” 16 Crim Just. & Behav. 183 (1989).Google Scholar
10 Benson, & Walker, , 50 Am. Soc. Rev. at 295.Google Scholar
11 Hagan, & Parker, , 50 Am. Soc. Rev. at 302.CrossRefGoogle Scholar
12 Id. at 312.Google Scholar
13 Treatments of this debate are available in the following sources. R. Collins, Conflict Sociology 49–64 (New York: Academic Press, 1975); N. C. Mullins, Theories and Theory Groups in American Sociology (New York: Harper & Row, 1973); J. H. Turner, The Structure of Sociological Theory (Homewood, Ill.: Dorsey, 1973); L. Coser, “Sociological Theory from the Chicago Dominance to 1965,” 2 Ann. Rev. Soc. 145 (1976); M. R. Haung, “Measurement in Social Stratification,” 3 Ann. Rev. Soc. 551 (1977).Google Scholar
14 M. Abramson, E. Mizruchi, & C. Hornung, Stratification and Mobility (New York: Macmillan, 1976).Google Scholar
15 We should note that education might also be seen as a status measure and in fact was included by Wheeler et al. in their sentencing analyses (but with no significant effects on judicial behavior). Overall education did not capture the broad status concerns addressed by SEI.Google Scholar
16 A. Reiss, Occupations and Social Status (New York: Free Press, 1961).Google Scholar
17 Hagan, & Parker, , 50 Am. Soc. Rev. at 302.CrossRefGoogle Scholar
18 E. O. Wright, “Varieties of Marxist Conceptions of Class Structure,” 9 Pol & Soc'y 229 (1980); E. O. Wright, Classes (London: Verso, 1985); E. O. Wright & Luca Perrone, “Marxist Class Categories and Income Inequality,” 47 Am. Soc. Rev. 32 (1985); E. O. Wright, C. Costello, D. Hachen, & J. Sprague, “The American Class Structure,” 47 Am. Soc. Rev. 709 (1982).Google Scholar
19 Following the scheme of the US. Census Bureau, class of worker had four categories in this data—employee of a private company, government employee, self-employed in own business, and unemployed.Google Scholar
20 The details of the coding of this variable are given in the code book for the Wheeler et al. project (available from the Inter University Consortium for Social and Political Research, Ann Arbor, Michigan).Google Scholar
21 US. Census Bureau codes were used.Google Scholar
22 If either indicated that an individual was a manager, he or she was so classed.Google Scholar
23 However, partners in companies were included in the employer category. We believe that the presentence investigations provide thorough information on matters such as work and financial history and that individuals who clearly own controlling interest in the companies for which they work are so identified and thus accurately coded.Google Scholar
24 E. O. Wright, Classes (London: Verso, 1985).Google Scholar
25 For a discussion of the literature on this issue, see Haung, 3 Ann. Rev. Soc. 51.Google Scholar
26 For details on the nature of the sample, see Wheeler, S., Weisburd, D., Waring, E., & Bode, N., “White Collar Crimes and Criminals,” 25 Am. Crim. L Rev. 331 (1988).Google Scholar
27 For a more detailed review of the study design, see Wheeler, et al., 47 Am. Soc. Rev. at 642 (cited in note 3). Hagan and Parker include only securities violators in their sample.Google Scholar
28 Those districts are southern New York, western Washington, central California, northern Georgia, northern Illinois, Maryland, and northern Texas.Google Scholar
29 See Wheeler, et al., 47 Am. Soc. Rev. at 644.Google Scholar
30 The Cronbach's alpha reliability coefficient of the new impeccability measure is 68. The zero order correlation between the two indicators is 78.Google Scholar
31 As noted earlier, however, we include the revised impeccability measure here. The Duncan scores for unemployed members of the sample are randomly allocated. See Wheeler, et al., 47 Am. Soc. Rev. at 649.Google Scholar
32 The logic response function for the probability of imprisonment (p of Imprisonment= 1/1 +e-xb) is approximated here by setting the value of xb at 0 and then allowing the variable being estimated (bi) to vary (xb=0+bixi).Google Scholar
33 Because of the skewness of sentences meted out by judges, the natural log of prison length was used as the response variable rather than the actual distribution of sentences. This served to pull the longest sentences closer to those of 6 or 12 months, better approximating the actual intervals of the decision the judges make (see Wheeler, et al., 47 Am. Soc. Rev. at 653 n.25).Google Scholar
34 Table 6 in id. at 654 is based on SEI scores from 60 to 960 (rounded to .001). The coefficient for the Socio Economic Index, when based on the range 6.0 to 96.0 is close to that which we report. The difference in significance between our model I results for status and those reported by Wheeler et al. derive primarily from changes in the impeccability index and the addition of lambda. Because the t coefficient for status in the original analysis was 1.97, these changes bring the significance of the variable below p >. 05..+05.>Google Scholar
35 The percentage increase in the months of imprisonment imposed caused by a one-unit change in an independent a variable may be calculated with the formula (ebi - 1) where bi, is the coefficient of the variable. See E. Tufte, Data Analysis for Politics and Policy 124 (Englewood Cliffs, N.J: Prentice-Hall, 1974).Google Scholar
36 A. Blumstein, J. Cohen, S. Martin, & M. Tonroy, eds., National Research Council Panel on Sentencing Research, 1 Research on Sentencing: The Search for Reform 103 (Washington, D.C.: National Academy Press, 1983).Google Scholar
37 This correction factor, known as lambda, is created by using the prediction model for the probability of imprisonment to calculate each individual's hazard of being excluded from the smaller group on which the length of imprisonment analysis is performed. This approach follows that presented in R. Berk, “An Introduction to Sample Selection Bias in Sociological Data,” 48 Am. Soc. Rev. 386 (1983), and J. Heckman, “Sample Selection Bias as a Specification Error,” 47 Econometrica 153 (1979).CrossRefGoogle Scholar
38 Hagan, & Parker, , 50 Am. Soc. Rev. at 309 (cited in note 7).Google Scholar
39 Comparing those prosecuted under statutes that allow for imprisonment of more than a year and those allowing only imprisonment up to one year, they find that the type of statutory charge is the only significant influence upon sentence severity.Google Scholar
40 As Berk observes, “In principle … there exists an almost infinite regress for any data set in which at some point sample selection bias becomes a potential problem. As for traditional specification errors and measurement errors, the question is not typically whether one has biased (or even consistent) estimates. The question is whether the bias in small enough to be safely ignored.”Berk, , 48 Am. Soc. Rev. at 392.Google Scholar
41 Max Weber, Economy and Society 926–39, ed., G. Roth & C. Wittich (Berkeley: University of California Press, 1978).Google Scholar
42 Wheeler et al., 47 Am. Soc. Rev. 657. See also Wheeler, Mann, & Sarat, Sitting in Judgment (New Haven, Conn.: Yale University Press, 1988).Google Scholar
43 Hagan, & Parker, , 50 Am. Soc Rev. at 313.Google Scholar
44 D. Weisburd, S. Wheeler, E. Waring, & N. Bode, Crimes of the Middle Ck: White Collar offenders in the Federal Courts (New Haven, Conn.: Yale University Press, in press).Google Scholar