1 Sociologists are the only social scientists to have focused on criminal achievement. Delinquency trajectories have been analyzed as life projects in which subjects seek to realize themselves in various ways: through skill, for Edwin Sutherland (1937); wealth for Robert Merton (1965); popularity or prestige for David Greenberg (Greenberg 1977; Rebellon and Manasse 2004); strength of character for Gresham Sykes (Sykes 1958; Preble and Casey 1969) and freedom for Jack Katz (1988). Patterns in criminal achievement are viewed as an outcome of social mobility mechanisms (Cuin 1993), and it is this understanding that gives classic (primarily American) criminal sociology its specificity and narrative unity.

2 The working hypothesis in American sociology from Sutherland to Daniel Bell was that under certain conditions, delinquency trajectories could account for upward social and occupational mobility patterns (Bell 1965; Cloward 1959). This hypothesis has not been demonstrated. On the one hand, professional delinquent paths show little in the way of “mobility.” Adler and Adler’s 1983 field study of 65 accomplished smugglers and dealers operating in a small American city on the border with Mexico (see also Adler 1993) suggests that delinquency trajectories show primarily lateral mobility. They note that with few exceptions, the split between high and low echelons of illegal substance selling reproduces delinquents’ original occupational positions. A variant of this thesis may be found in Holtzman (1983). The author examined the record of legal jobs held by a cohort of thieves and burglars. Surprised to see that the vast majority had conventional jobs, he developed the hypothesis that their delinquent activities offered all the features of a moonlighter’s job, in this case unskilled workers who could only “make ends meet” by working on the black market.

3 On the other hand, most research using self-report surveys has denied any direct influence of social class background on crime involvement (Tittle, Villemez and Smith 1978; Tittle and Meier 1990; Reiss and Rhodes 1961; Roché 2001), strengthening the thesis that social labeling mechanisms suffice to account for the “social composition bias” found in police files and prison populations. However, social scientists may have dismissed the effect of social class background prematurely. Wright et al. (1999) took a different approach, claiming that the absence of correlation observed in Tittle and Meier’s meta-analysis (1990) did not necessarily mean an absence of causation. They argued that social class and crime are not necessarily linked in linear fashion. Both ends of the social class spectrum could exert a positive influence on crime involvement through different causal sequences. Citing self-report surveys of delinquent teenagers and young adults, they first observed no apparent relationship between respondent’s social background and criminal participation (proportion involved in delinquent acts), offense frequency (number of crimes commited) or offense seriousness. But they then discovered that feeling of impunity and taste for non-conformity increased with each rung on the social ladder, a finding that explains how it is that delinquent behavior may actually increase with social position.

4 These results, counter to American criminologists’ standard opinion, are not the only ones supporting the class-crime relationship. Hagan, Gillis and Simpson (1985), taking inspiration from William Bonger’s Marxist study (1916), confirmed that social class background influences the relation between delinquent activities and differential family supervision of teenagers of both sexes: differences in delinquent behavior by sex are sharp at the higher end of the social scale but decrease in the mid-level classes and are very slight at the lower end. Furthermore, Grasmick, Jacobs and McCollum’s self-report survey of delinquency (1983) showed that perceived risk of arrest and assessment of its consequences varied inversely with social class background for the teenagers interviewed. Wright et al. (1999) found that delinquent behavior also varied as a direct function of relative deprivation and degree of family disorganization, these in turn varying inversely with social class background. These results corroborated those of McCarthy and Hagan (2004), who added questions on personal income and expenses in their self-report delinquency survey. They found that social background interacts with criminal propensity: every increase in personal income lowers the delinquent conduct of teenagers at the bottom of the social scale while having no such effect on teenagers who enjoy the resources of a more privileged social milieu. The analytic solution put forward by Wright and his colleagues is elegant but has two drawbacks: “psychologizing” social class turns it into a construct that can explain both A and its opposite; the conceptual scenario was developed a posteriori to account for an expected result—i.e., social background’s apparent non-impact on self-reported delinquency.

5 In this article we examine offending outcomes rather than processes (i.e., offense frequency or seriousness). Past research has not tried to investigate empirically why, with level of criminal activity held constant, some delinquents manage to obtain considerable gains whereas others do not. This is surprising since it was a central issue for Cloward (1959), who argued that attractive criminal opportunities were themselves socially stratified. This major flaw is particularly clear in Hobbs’ assessment (1995) of published research on “professional” delinquency from 1937 to 1993. Despite the importance that Sutherland and Cressey (1966) attribute to training in the professionalization model of criminal trajectories (see Dubar and Tripier 2005), none of the studies reviewed by Hobbs addresses this point. Only recently, in connection with a survey of the criminal activities of an inmate sample from three Canadian federal prisons, did Morselli, Tremblay and McCarthy (2006) actually put the question to respondents directly: “Among the persons that have influenced you in your life, is there one who introduced you into the criminal milieu and whom you think of as your mentor?” Their results suggest that mentors do play an important role (39% of inmates interviewed said they had had a mentor), and that criminal mentors exhibit the same characteristics as mentors on conventional labor markets. The age difference between mentors and protegés is considerable (11 years on average), and no respondent said he had had “more than one mentor,” a point that brings to light the special, privileged nature of that relationship. In support of the thesis that mentor training is helpful, the study shows that delinquents who had the “luck” to meet a mentor reported median criminal incomes ten times higher than those who had not, and had spent less time in prison than mentorless delinquents in the three years preceding their incarceration. Meeting mentors (which is likely to occur when the delinquent-to-be is between 18 and 20) probably does not explain why individuals get started in crime but rather how their trajectories develop or “mature.” The mentor role is not incentive (i.e., it does not encourage non-delinquents to become delinquents) but formative, ensuring that a motivated delinquent will become a “good”—successful—one. Mentors “profess” crime as a vocation and seem to play a crucial role in professionalizing delinquents.

6 It is hard to analyze delinquents’ professional mobility without measuring their criminal performance. Contemporary criminology has shown little eagerness to do so. With the exception of a few field studies (McCarthy and Hagan 2001), the only self-report delinquency surveys to question delinquents on their individual performance (in this case, the income they get from their illegal activities) were done in the 1970s on representative inmate samples. Respondents were asked to detail their delinquency trajectories, the nature of the illicit activities they had engaged in the 36 months prior to their incarceration, the episodes of freedom and detention and the pace or tempo of their criminal activities by type (Peterson and Braiker 1981; Chaiken and Chaiken 1982; Blumstein et al. 1986). It was not until 15 years later that researchers began listening to what prisoners themselves had to say about their criminal performance (Tremblay and Morselli [2000] analyzed a 1976 survey of 672 detainees; Robitaille [2004] examined a 1978 survey of 2,670 detainees). These new analyses suggest that the sentences being served by “less skilled” offenders were as harsh as those of their “more skilled” neighbors; that the “more skilled” were, understandably, more inclined to quit the conventional job market than the “less skilled”; that the highest-performing delinquents were also the ones most likely to reoffend after release from prison. Criminal achievement, then, is not fortuitous but “deserved.” Success increases in proportion to the care delinquents take in planning and organizing their crimes. Heavy alcohol use compromises criminal performance, whereas cocaine or heroin addiction does not. Lastly, offenders who abstain from violence obtain less from their criminal activities than those who show no such scruples.

7 This article examines the hypothesis that a delinquent’s social class background influences his pathway in crime. The delinquent practices studied here (car-theft, illegal substance trafficking, aggravated or violent theft, “ordinary” credit card or check fraud) may be described as “common crimes” accessible to individuals from any social class background. This hypothesis differs from Cloward’s (1959), which argued that chances of succeeding in crime were independent of chances of succeeding on the conventional labor market. It also differs from the thesis that access to conventional jobs is the main or only mechanism stratifying crime opportunities (i.e., business crime versus “ordinary” crime). On the other hand, our hypothesis has the virtue of combining two research traditions that have always ignored each other: specialists interested in delinquents’ social backgrounds are necessarily less interested in their professional trajectories, while specialists interested in delinquents’ professional trajectories have not been interested in their social backgrounds.

8 The article proceeds through five stages. The first section describes the conditions in which inmates were interviewed and the strategy used to piece together their life trajectories during the three years preceding the prison term served at the time of the interview. The second section presents our two major analytic variables—upstream, delinquents’ social background; downstream, their criminal incomes during the “window period.” The third section analyzes how offenders’ criminal incomes vary with their social backgrounds. The fourth section presents three mechanisms likely to have conditioned the observed relations between social background and criminal achievement: delinquents’ personal ambition or desire to “get rich”; their access to criminal education (meeting a “criminal mentor”); their choice of criminal activity/ies. The fifth section examines the effects of social class on the relative severity of the criminal courts. Though the “class-based justice” notion has been largely overshadowed, particularly in the United States, by study of skin color or racial discrimination, we pursue here a tradition of research initiated by Herpin (1977), Pires, Landreville and Blankevoort (1981) and Aubusson de Cavarlay (1985, 1987; see also Aubusson de Cavarlay et al. 1995), though from a different perspective and with a more circumscribed focus. All inmates interviewed for this study shared the same penal adversity (serving a federal term of two years or more) and presented the same self-reported criminal profile (common offenses). Our aim is to determine whether, with criminal record and type of criminal practices held constant, bad boys from “good families” have an advantage in the criminal courts and if so, by what means they secure this advantage.

Respondent Sample and Interview Procedure

9 Two hundred and sixty-two inmates in five federal penitentiaries [1] in the Montreal area were interviewed during the summer months of 1999, 2000 and 2001. While inmate surveys are usually self-administered and filled out by groups of inmates (Peterson and Braiker 1981; Chaiken and Chaiken 1982; Horney, Osgood and Marshall 1995), our interviews were conducted face-to-face, allowing interviewers (three male and six female students) to make sure that each question was correctly understood and to obtain sufficiently precise answers. To minimize memory errors, respondents who had been in prison for more than seven years were excluded—a screening precaution of limited use since the majority of respondents (76%) had been in prison less than two years. Average interview time was an hour and thirty minutes. The interviews did not focus on the prison term inmates were serving but on the 36 months preceding the moment that prison term began, what we are calling the window period.

10 Inmates were randomly selected from a list of prisoners physically present on the given day. Half the interviews were made in medium-security prisons (n = 123, or 47%) and half in minimum-security prisons (n = 123, or 47%). [2] In the medium-security prisons, interviewers were provided with rooms where prisoners usually meet with their lawyers. Each room had a telephone connected to various sections of the penitentiary and it was up to the interviewers to contact the inmates. In minimum-security prisons, the call for respondents worked somewhat differently due to residents’ freer movement. The easiest way to reach them was over the intercom system from a control post near the interviewing room. The order to go to the interviewing post was heard in each unit of the establishment, but as in the other prisons, the reason for the call was not indicated. Though 262 prisoners were interviewed, this study is based on a 195-interview sample: 45 respondents were disqualified because their criminal activities were not motivated by financial gain (sexual offenses, for example, dangerous driving, etc.) and another 22 for incomplete questionnaires.

11 The interview began with the usual introductions and guarantees of anonymity and confidentiality. Inmates could refuse to be interviewed (20% took that option); [3] they could interrupt the interview at any time and refuse to answer any question. The title of the questionnaire was deliberately vague: “Survey on inmates’ financial situation before current prison term.” Nonetheless, it was clear that the questions would bear on respondents’ licit and illegal income. We wanted to offer symbolic compensation for participation but the prison authorities vetoed that proposition.

12 To facilitate reconstruction of respondents’ lives during the three-year period, answers were transposed onto three 12-month calendars (on this strategy for contextualizing life stories, see Horney, Osgood and Marshall 1995; Freedman et al. 1988; Bradburn, Lance and Shevell 1987). Served prison time was the first piece of information put on the calendars. Respondents had to reconstruct their “prison trajectories,” indicating months in which arrests occurred, periods spent in prison, and months under probation, conditional release or in a “transition house.” The vast majority of respondents knew the exact entry and departure dates of their various stays in prison. The interviewer then went through a list of life events sufficiently important for most inmates to be able to indicate the months in which they occurred (if they did): death (s) of close relative (s) or friend (s), hospitalization (s), therapy program (s), suicide attempt (s), separation (s) or divorce, childbirth (s), job loss (es), significant money loss (es) and housing move (s) were all marked on the calendars. Legitimate work experience was then added to this first set of path markers. After specifying the three most important jobs held during the period, respondents were asked to indicate what months they had held them and their income for each month worked.

13 Delinquent activities were taken up last. [4] Respondents were asked first to indicate if they had engaged in various property offenses (aggravated theft, burglary, car theft, theft, fraud, con games, swindling) or trafficking on illegal markets (drug dealing and distribution, contreband, fencing of illicit goods, illegal gambling, loansharking, pimping). [5] After specifying the parameters for each of these activities (degree of planning, pace of criminal acts, average earning per offense, number of co-offenders implicated, most frequent targets, etc.), respondents were asked to indicate an estimate of their illegal gains (for a detailed analysis see Charest 2007).

Analytic Strategy

14 Interviewed inmates’ positions on the social scale were derived from the nature of their fathers’ jobs (or former jobs). Specifically, we used Nakao and Treas’ occupational prestige scale (1994) to group employment spheres into six categories: 1) liberal professions and management positions (manager, teacher, engineer, doctor, judge, architect); 2) technical jobs, sales, administration (IT specialist, civil servant, buyer); 3) service jobs (clerk, police officer, firefighter, restaurant owner); 4) “skilled” manual jobs (plumber, machine operator, welder, electrician); 5) “unskilled” manual labor (cleaning, truckdriving, delivery, unskilled construction work) and 6) farming, fishing, forestry jobs. These employment categories and the order they appear in differ very little from other occupational prestige scales used in Canada (namely Blishen 1958). Nakao and Treas’ scale was chosen primarily because it includes a wider, more contemporary array of occupations. Given the limited number of respondents in some categories (forestry and farming), we restricted our scale to four groups: 1) professionals and managers; 2) technical, sales and services; 3) skilled manual labor; 4) unskilled manual labor. To these four categories we added “jobless,” comprising all respondents whose fathers have been unemployed for most of respondent’s life (see Appendix for a presentation of fathers’ occupations).

15 Social position was found to be correlated with a series of other indicators of family’s socio-economic level. For example, 91% of professional and manager fathers were homeowners, against 48% of manual workers (the contingency coefficient, C. C., is 0.26, p < 0.001). Frequency of recourse to social assistance rises as social background declines (C. C. = 0.34, p < 0.001). Only 4% of professional families (one in 23) had received welfare at one time or another in delinquent’s life, as opposed to 42% of unskilled worker families. Moreover, the higher the occupational prestige, the more likely respondents’ fathers were to have subordinates working for them (C. C. = 0.35, p < 0.001). Number of employees for which father is accountable also varies proportionately with social status (C. C. = 0.49, p < 0.001). Social position also tallies with respondent’s personal assessment of family’s financial situation (r = 0.396, p < 0.001). Most respondent families did not seem to have experienced notable financial difficulty when inmates were young. More than 30% of respondents thought of their family situation as “well-off”; 49% as “within the average”; only 21% of respondents considered their family’s income below average.

TABLE 1

Inmates’ Fathers’ Occupational Categories

Occupation category	N	%	% of Quebec population	Prestige rating	% owners	% employed	Jobless	% on welfare	Subjective class (3 categs.)
Liberal professions, manager	23	12	11	65	91	100	3.9	4	2.8
Technical, sales and service	32	16	30	48	59	76	1.9	26	2.1
Skilled manual	56	29	19	43	71	66	1.7	18	2.2
Unskilled manual	78	40	32	30	55	45	0.9	42	1.8
Jobless	6	3	9	0	0	0	0	83	1.5
Total-Average	195	100	100	41	63	64	1.7	29	2.1

Inmates’ Fathers’ Occupational Categories

16 Table 1 shows respondent number and proportion by father’s job category as well as occupational categories of male Quebeckers aged 55 and over (an age similar to respondents’ fathers’ ages). [6] We see that there are approximately as many sons of professionals, businessmen and managers in our sample as in the population at large: 12% versus 11%. At the bottom of the social scale (jobless fathers’ sons) the proportions are also similar: 3% versus 9%. Delinquents’ fathers are not proportionately more likely to be receiving unemployment or other welfare benefits than the Quebec population at large. However, poorer social backgrounds are overrepresented in our sample. While approximately 50% of male Quebeckers age 55 and over have jobs in the manual labor sectors (either skilled or unskilled), the figure is 69% for our inmate sample. Two-thirds are therefore of working-class background. The overrepresentation corresponds to underrepresentation of the middle classes, i.e., fathers working in technical, service or sales jobs. Nearly one-third (30%) of male Quebeckers work in these sectors whereas they account for only 16% of our sample. Despite these differences, there is significant social background diversity. The social composition of this inmate sample is similar to the one found in Zamble and Porporino’s 1988 study of federal prisoners in the Canadian province of Ontario, where 28% of prisoners were from middle-and upper-class backgrounds (professionals, managers, skilled white collar), 39% from the working class and 33% from the lowest classes (unskilled or unemployed).

17 To measure criminal incomes, respondents were first asked to indicate type (s) of criminal activity practiced in the window period, the frequency of activity for each type of crime committed and the average income per crime or transaction. Monthly and cumulative income during the 36 months was estimated by combining these three data. In the second section of the survey, we asked inmates for their own estimate of their illegal income for each month in which they were criminally “active” before incarceration (using three 12-month calendars). The two measures took into account months of respondent activity and inactivity, so that the “delinquent activity period” may be thought of as a constant. The correlation between these estimations is somewhat strong (0.52) though they require distinct intellectual operations (analytic for the first one and synthetic for the second). Discrepancies between estimations cannot be explained by interviewer’s sex or age, prison in which the interview took place, number of delinquent’s prison stays or time interval between interview date and start of the window period (Charest 2004). Reported incomes (the second measure) were generally more conservatives than estimated (or derived) incomes and were used as our main dependent variable. A logarithmic base transformation was ultimately applied for several distinct reasons: doing so normalizes income distribution (see, for example, Tremblay and Morselli 2000; Matsueda et al. 1992), corrects for differential perceptual discrimination across the income continuum (Tiberghien 1984), and reduces possible distorsion effects due to bragging or exaggeration. The correlation between logged reported and estimated incomes is almost perfect (r = 0.97).

The Effects of Social Class on Criminal Achievement

18 Social class background is positively related to criminal achievement measured by illegal earnings (r = 0.156, p = 0.026). What is true at the individual level is also true when individuals are grouped together (Table 2). The criminal revenues of offenders at the higher end of the social scale (sons of professionals and managers) are distinctively higher than those of the lower echelons; they average $170,000 over three years [2001 Canadian dollars], or $56,000 per year. As social position declines, so does illegal income. In the second echelon, average criminal income is only $42,000 ($14,000 per year). In the manual labor categories, sons of skilled workers (third echelon) made on average $30,907 from their criminal activities, whereas sons of unskilled workers (fourth echelon) made less than $17,000 ($5,600 a year). This gradual decline is reversed at the bottom of the social scale: jobless men’s and welfare recipients’ sons generally succeed better than delinquents at the center of the distribution (a result consistent with the hypothesis of Wright et al., 1999). Their criminal gains ($100,719, or $33,573 a year), while below those of delinquents from the privileged class, are far above those of “middle-” class delinquents.

TABLE 2

Real and Potential Licit Earnings and Actual Criminal Income (Geometric Mean) by Social Class Background

Father’s occupational category	N	Mos. of work (% of mos. of freedom)	Observed annual legitimate income ($)	Potential legitimate income ($)	Self-reported annual criminal income ($)
Professionals, managers	23	13 (45%)	10,253	30,584	56,399
Technical, sales and services	32	12 (34%)	12,638	30,640	13,939
Skilled manual	56	15 (46%)	8,524	29,903	10,302
Unskilled manual	78	13 (41%)	10,471	29,269	5,652
Jobless	6	10 (34%)	7,201	27,882	33,573
Total	195	13 (42%)	10,141	29,781	10,791

Real and Potential Licit Earnings and Actual Criminal Income (Geometric Mean) by Social Class Background

19 What explains these performance inequalities? One scenario is that middle and upper class “bad boys” have relinquished advantageous legitimate employment income and therefore demand higher criminal compensation. As it turns out, legitimate income is fairly stable across social class backgrounds. Inmates’ hourly wages during the window period do not differ perceptibly from one class to another, suggesting that the quality of jobs held does not really differ with social background. [7] Most delinquents have jobs at the bottom of the occupational prestige scale. A possible alternative scenario: legitimate income of delinquents from the relatively privileged classes is just as low as others’ but they may be less motivated to remain in low-quality jobs. However, this scenario does not hold up since number of months at work does not vary across social class background. All respondents shared a pattern of sporadic participation in the conventional labor market (13 of the 36 months window period). Third scenario: while the actual performance of more socially advantaged offenders does not differ from other offenders’, they may believe that their earning prospects would be more enticing if they actually put their minds and social resources to criminal activity. This scenario too must be discarded: delinquents of middle- and upper-class background are just as likely to have dropped out of school. Though offenders from privileged social backgrounds do show higher average educational attainment (48% of them finished high school as opposed to 19% of the second echelon, 14% of the third and 13% of the fourth), few went beyond high school, meaning that educational differences are slight and unlikely to have an impact on the job market. [8] We estimated those “potential” legitimate incomes [9] (see Table 2, column 4) and found no correlation with respondent’s social status.

20 Offenders’ criminal achievement does not offer a reverse image of the licit earnings they have relinquished. It is not because members of privileged social classes have more to lose that they demand higher compensation from crime. There is yet another alternative scenario, however: offenders may not all have the same standards of achievement. A classic hypothesis has been that social class background affects delinquent’s aspirations or degree of ambition (Hyman 1965). Delinquents from relatively poor backgrounds seem ready to settle for more “modest” licit or illicit income. We included in our survey a question on delinquents’ monetary goals: “Do you agree or disagree with the following statement: ‘I want to make a lot of money’?” The results show a positive correlation between social class and desire for high financial returns (r = 0.157, p = 0.03), suggesting that aspirations are indeed influenced by social background.

Ambition, Social Capital and Criminal Occupations

21 We now examine the relevance of this last scenario. The aim is to model the series of processes by which delinquents from relatively privileged backgrounds obtain more successful trajectories than others. In this analysis, criminal income is the dependent variable, social class background the antecedent variable, respondent’s age the control variable, and ambition an “intervening” variable since social class may influence delinquents’ degree of criminal ambition. This study also explores the impact of social capital and criminal mentoring on offender performance (Morselli and Tremblay 2004; Morselli, Tremblay and McCarthy 2006) and incorporates a “professionalization” indicator into the model: presence or absence of a criminal mentor. Though these intervening variables both precede criminal performance, ambition is conceived as a relatively stable characteristic that precedes the encounter with a mentor. As previously mentioned, offenders are typically coopted by mentors at age 18 or 19. While it is true that social class influences mobility opportunities, we also need to account for delinquents’ criminal choices, especially the decision to pursue “entrepreneurial” or market activities [10]. Charest (2007) has shown that participation in such activities enables offenders to bring in attractive, stable incomes while incurring lower detection risk.

TABLE 3

Regressions of Combined Inmate Sample Criminal Trajectory Parameters (Performance, Training, Professional Choice) on Delinquents’ Social Class Background and Age

M1 Criminal income b (beta)		M2 Ambitions	M3 Criminal income	M4 Mentor	M5 Criminal income	M6 Entrepre neurial crime	M7 Entrepre neurial crime	M8 Criminal income
		b (beta)	b (beta)	Exp (B) (Wald)	b (beta)	Exp (B) (Wald)	Exp (B) (Wald)	b (beta)
Age –.067** –.040** –.058** .969+ –.055** .990 .996 –.054** (–.360) (–.288) (–.309) (2.76) (–.290) (.361) (.045) (–.288)
Social class .228* (.133)		.150+ (.117)	.172 (.099)	1.48** (6.49)	.121 (.070)	1.31+ (3.44)	1.21 (1.59)	.097 (.056)
Ambitions --		- -	.240* (.179)	1.35* (4.34)	.209* (.155)	1.16 (1.46)	1.10 (.546)	.197* (.147)
Mentor --		- -	- -	- -	.593* (.157)	- -	2.91** (10.14)	.450+ (.119)
Entrepreneurial -- crime		- -	- -	- -	- -	- -	- -	.556* (.155)
Constant	5.98**	5.75**	4.63**	– 2.06	4.58**	– 1.08	– 1.21	4.46**
R2 or pseudo R2	.153	.091	.165	.131	.184	.048	.118	.202
N	195	186	186	186	186	186	186	186

Regressions of Combined Inmate Sample Criminal Trajectory Parameters (Performance, Training, Professional Choice) on Delinquents’ Social Class Background and Age

Note: **? < .01; *? < .05; +? < .10.

22 Model 1 indicates that offenders’ social class background improves their criminal performance (b = 0.228, p < .05). Criminal earnings also decline with age (b = –0.067, p < .01). This may be due to weariness (older delinquents are more “tired” or less active) or a self-selection effect (dynamic and successful delinquents may abandon their criminal careers more quickly). As hypothetized, three mechanisms may contribute to this class effect: aspirations or desire to obtain personal wealth (Model 2), training by a criminal mentor (Model 4) and participation in entrepreneurial crime (Model 6). Model 2 shows that ambition level is indeed modulated by social class background (b = 0.150, p < .10), and Model 3 that with individual desire to get rich controlled for, social class background effect on criminal performance vanishes (b = 0.172). Model 4 shows that delinquents from privileged social backgrounds are more inclined to seek out and find criminal mentors when they become young adults (b = 1.48, p < .01) and that the most ambitious delinquents are the ones most likely to attract mentors’ attention (b = 1.35, p < .01). Model 5 shows that when the effects of training (mentors) and motivation (desire to get rich) are taken into account, the direct effect of social class background on criminal performance likewise disappears (b = 0.121). Delinquents’ occupational choices also play a role. Model 6 shows that bad boys from the upper classes are more likely to opt for entrepreneurial crime (b = 0.131, p < .10). Mentors seem to be what tip the scales: when presence or absence of a mentor is controlled for, the social class effect on decision to engage in entrepreneurial crime disappears (Model 7). The last model provides a synthesis of the preceding analyses. Delinquents’ trajectories reproduce the social inequalities to be found among their fathers, and this reproduction is framed by three factors: delinquents from relatively privileged families manifest, from the outset, a relatively strong desire or will to “get a good return” on their illicit activities (b = 0.197, p < .05); they are more inclined to seek out and find criminal mentors who will provide them with useful training (b = 0.45, p < .10) and they are more inclined to choose entrepreneurial crime (b = 0.56, p < .01). These three factors have cumulative effects on performance but are also interdependent: ambition attracts the attention of mentors, who then guide ambitious delinquents into entrepreneurial crime.

Social Class, Sentence Leniency and Criminal Achievement

23 We might be tempted to believe that the most prolific, lucrative criminal careers are also likely to attract police attention and incur judicial severity. Preliminary inspection of our survey data indicate that prison sentences meted out inside the three-year study period were not influenced by delinquent’s degree of success. The most successful delinquents were not the ones most often sent to prison (r = 0.103, p = 0.15, n = 195) or for the longest time (r = 0.031, p = 0.67). The results are the same for sentences that inmates were serving at the time of the interview (see Table 4 for distribution of sentences being served by social class background). Offenders at the top of the income scale did not get longer federal sentences (r = –0.024, p = 0.77). If we isolate income for the last year of the calendar (assuming the current prison sentence derives from offenses committed that year), the relation does not improve (r = 0.004, p = 0.96). Sentence severity, then, is not directly proportionate to the gains delinquents obtain from their offenses.

24 However, labeling theory suggests that offenders with a middle and upper class background involved in “common” delinquency are more likely to obtain lenient rulings from criminal courts. For offenders with this social profile, the “lawyer’s role as a neutralization mechanism capable of extracting additional advantages for the accused is remarkable. We are not referring directly to the lawyer’s function but rather the structural, organizational, and ideological possibility of requesting or seeking out exceptional measures without discrediting oneself in the eyes of the other players (police, prosecuting attorney, judge, etc.)” (Pires 1983; Pires, Landreville and Blankevoort 1981: 333). This hypothesis, derived from a small number of interviews, [11] deserves a closer look.

TABLE 4

Inmate Respondent Profile: Offense Convicted of, Legal History, Lawyer’s Fees Paid out in most Recent Trial

	Age	Length (severity) of sentence (mos.)	Sentence severity relative to offense convicted of	Delin quent’s incarce ration history	Use of private law firm
	Aver.	Aver.	Violent crime	Proportion	% who paid	$ (aver.)	$ (total aver.)	% receiving financial support	% not called upon to reimburse anything
	(standard deviation)	(Median)	(Homicides) %	(average number)
Liberal profs, management	31 9.5	51	34 (4.3)	74% (2.3)	70	43,825	33,390	31	40
		(36)
Technical, sales and service	31 8.6	63	50 (9.4)	59% (2.5)	56	47,040	27,314	41	75
		(48)
Skilled manual	31 9.4	57	43 (8.9)	82% (2.1)	54	36,643	19,630	31	67
		(38)
Unskilled manual	32 10.1	62	42 (3.6)	77% (2.5)	55	18,015	9,865	35	59
		(48)
Total	31 9.5	59.6	42 (6.2)	75% (2.3)	56	31,599	18,104	34	62

Inmate Respondent Profile: Offense Convicted of, Legal History, Lawyer’s Fees Paid out in most Recent Trial

25 Table 4 suggests a social class background effect on prison sentence length at both distribution extremes: an average sentence of 51 months (median 36) for wayward sons from the most affluent social class, and an average sentence of 62 months (median 48) for sons from the most disadvantaged milieus. Note that prior record indicators, likely to influence criminal courts (prior incarcerations, length of prior prison terms), do not vary by social class background. Average age of respondents is also similar from one group to the next, indicating comparable periods of exposure to the risk of incarceration.

26 Nonetheless, our data do confirm Pires, Landreville and Blankevoort’s hypothesis (1981): the higher a delinquent’s social class background, the more likely he (or his family) will hire a lawyer to defend him (column 5) and the more they are willing to pay for that service (columns 6 and 7). One third of delinquents receive family financial support for their legal expenses. This is often unconditional support: two-thirds of delinquents receiving it are not expected to pay it back. Interestingly, defendants from upper-class backgrounds are more likely than others to repay the money lent to them (60% as against 38% for the sample as a whole). Note that delinquents with the highest criminal incomes are also those who pay the most in lawyers’ fees—around $33,000, compared to the average of $18,000. These findings are perfectly consistent with the general argument that mobilization of legal resources increases with defendant’s social status (Black 1976: 11-36; 1989: 20-40). These findings also challenge conventional labeling research’s one-sided emphasis on rejection and exclusion (Cusson 1983).

27 Some findings reported in Table 4 are not easy to interpret because a number of factors have to be controlled for to isolate possible effect of social class background on court sentencing severity. First, we would expect that delinquents who spent the most in legal fees were those at risk of being incarcerated for long periods of time. Seriousness of charges should therefore be controlled for. Second, it is reasonable to assume that offenders will be more motivated to pay lawyers when the evidence against them is circumstancial or relatively weak (see Albonetti 1987 on the decisive effects of evidence “solidity” on trial results). We do not have a direct measure of the strength or weakness of respondents’ legal cases, but it is reasonable to assume that the money spent on lawyer’s fees measures the intensity of lawyer’s efforts to negotiate a sentence that will benefit their clients. The presence or absence of a specific social class background effect on sentencing severity is more likely to be revealed if money spent on lawyer’s fees is kept constant. Our working hypothesis, then, is that delinquent’s social class background only becomes an advantage when an authorized third party (i.e., the defense lawyer) is in a position to use that background “without discrediting himself in the eyes of the other court players.” To isolate the effect of social class on sentence severity, then, we conducted an analysis that took into account the parallel effects of charge seriousness, lawyer’s fees, age, prior incarceration history, criminal earnings and other characteristics (being head of household, having a conventional job).

28 Table 5 presents three analytic models. The first two conceptualize the effect of social class background on sentence severity. The third examines social class effect on a mechanism that plays a crucial role in the final determination of sentence severity, the use of a private defense lawyer.

TABLE 5

Regressions of Sentencing Severity (Months in Prison) and Lawyer’s Fees on Social Class Background, Offense Convicted of, Criminal Performance, Age and Other Characteristics

Sentence severity									Lawyer’s fees
									b beta sig b beta sig Age 0.13 0.025 0.713 0.13 0.026 0.692 Social class –3.52 –0.075 0.200 –5.82 –0.123 0.033 Incarcerationhistory 1.67 0.015 0.808 1.17 0.011 0.859 Violence 19.58 0.198 0.001 20.03 0.205 0.001 Homicide 111.48 0.550 0.000 104.78 0.526 0.000 Lawyer’s fees (x $1,000) -- -- -- 0.22 0.194 0.002 Mos.ofwork 0.37 0.100 0.096 0.32 0.087 0.138 Dependentchildren 2.51 0.026 0.677 0.75 0.008 0.898 Criminalincome 0.06 0.002 0.973 –1.80 –0.066 0.308								b	beta	sig
																	185.0	0.041	0.609
5,629.6	0.134	0.053
– 4,496.8	– 0.046	0.527
– 11,616.4	– 0.135	0.063
36,846.5	0.209	0.004
- -	- -	- -
– 24.0	– 0.007	0.917
– 697.8	– 0.008	0.911
6,767.2	0.282	0.000
Constant	39.95	- -	0.018		50.36	- -	0.002		- 21,949.2	- -	0.210
R2	0.399				0.441				0.163
N	191				191				191

Regressions of Sentencing Severity (Months in Prison) and Lawyer’s Fees on Social Class Background, Offense Convicted of, Criminal Performance, Age and Other Characteristics

29 As predicted, the more serious the offense, the harsher the sentence. While the average sentence is 40 months (the constant), offenders convicted of homicide get 111 additional months; those convicted of violent crime (aggravated theft, assault and battery, etc.) get approximately 20 additional months. Do criminal courts take criminal achievement (or lack thereof) into account in their decisions? They do not seem to (p = 0.97). A “bungled” or unsuccessful criminal career is not a mitigating circumstance, and a more successful career is not an aggravating one. Having dependent children is not a mitigating circumstance (p = 0.68), nor is delinquent’s age (p = 0.71). History of imprisonment is not a significantly aggravating circumstantance (1.67 additional months in prison per prior prison term). Prior record is an important criterion in criminal court decisions to sentence the defendant to prison. But all the subjects in our study were active, persistent offenders and most had been in prison before.

30 For otherwise equivalent legal cases, class background effect on sentence severity seems slight: each next-higher echelon of the social scale (four in this study) decreases sentence by 3.5 months. Given an average prison term of 60 months, this means a marginal decrease of 5% for each echelon of the social scale. However, the cumulative impact amounts to a 20% disparity in sentencing severity for cases located at the bottom and top of our social scale. This gap amounts to a sizeable “discount” of minus one year behind bars.

31 What happens if we take into account offender’s ability to hire a private lawyer and spend “what it takes”? Those who can afford a private lawyer spend $5,000 in lawyer fees for each additional month of prison sentence (b = 0.22, p < .002). Obviously defense attorneys are unlikely to be the cause of sharper court severity. We therefore interpret this positive relationship as indicating that the stronger the evidence against the defendant, the more costly the defense. Is it rational for offenders to invest in such conspicuous expenditure? It would seem to be. For those in a position to take this option, each echelon of their social class background reduces that sentence by nearly six months (b = –5.8, p < 0.03). Sentence disparity between offenders located at the bottom of the social scale and those at the top is 28 months (5.8 months x 4). Since average sentence is 60 months, delinquents from the most privileged social classes seem to get their sentences reduced by a third (5.8 x 4/60 months) as a result of being represented by private lawyers. Since they have spent on average $33,000 in attorneys’ fees (see Table 4), we can assume that a year less in prison is worth $16,500.

32 Which offenders can hire a private defense lawyer? The third model in Table 5 analyzes what factors predict the amount of money defendants invest in legal representation. The finding is that the more successful offenders are in their criminal activities, the more money they spend on lawyers committed to their interests (p < .01). Criminal earning is the strongest predictor of legal ressources invested (beta = .28). Social class background also has an effect: offenders pay approximately $5,600 more in defense lawyer fees at each echelon of the social ladder (b = 5,630, p = 0.02). This is a remarkable effect. On the one hand, social class background influences delinquent performance and it is precisely this monetary “success” that enables them to hire a private lawyer. On the other hand, the indirect effect is accompanied by a direct one: offenders from relatively privileged social backgrounds are more inclined to pay whatever it takes to successfully “negotiate” a lower sentence, regardless of their criminal income. Family financial support operates in all social classes, but with each echelon, family wealth obviously increases, as well as money available for legal defense.

33 When charged with murder or manslaughter, all defendants, regardless of class background, tend to hire a private defense lawyer, even if this means going into debt. Since our sample includes offenders convicted of homicide at all social scale echelons, the fact that they all find money for legal defense partially scrambles or undermines disparity effects imputable to delinquent’s background. This explains why that disparity is much less pronounced in Table 4’s bivariate analyses than in the multivariate analyses in Table 5.

34 Cuin (1993: 44) argued that American sociologists’ academic interest in measuring social mobility rates between 1945 and 1970 was primarily driven by the three main illusions of the “American dream”; namely the “classless society,” “equal opportunity,” and “social success.” It seemed natural in this context to formulate the hypothesis that criminal trajectories would fit the dynamic of real or hoped for “upward social mobility.” As belief in the American dream waned, the focus of contemporary criminal sociology (Shapiro 1985; Albonetti 1999; Weisburd, Waring and Wheeler 1990 in the United States; Hagan and Parker 1995 in Canada; Lascoumes 1997 in France) was restricted to white-collar crime. Sociologists lost interest in the questions of unequal crime opportunities (mentoring, for example; Morselli, Tremblay and McCarthy 2006) and distribution of episodic or cumulative success in criminal trajectories (Tremblay and Morselli 2000). The analytic strategy adopted in this article steers clear of vague categorizations (e.g., “crime propensity”) and fascination with first causes ( “explaining causes of juvenile delinquency”). Our focus here is not delinquency but criminal achievement. Inequalities in criminal performance are likely to be somewhat irrelevant where occasional offenders are involved but they are a key concern for persistent offenders willing to “pay the price.”

35 English-language criminology, primarily American, has, with a few notable exceptions (Hagan 1992), banished “class effects” from its vocabulary. Self-report surveys of crime (most of which focus on juvenile delinquency) suggest that social class has no direct impact on participation frequency or even the seriousness of criminal behavior (Tittle, Villemez and Smith 1978; Tittle and Meier 1990). Any class effect is explained away as exclusion or discrimination practiced by public institutions and the ampliflying effects of social regulation mechanisms (Bernburg and Krohn 2003). This understanding seems incomplete. The fact is that offenders’ social class background may determine their chances of succeeding in their criminal activities as well as their ability to negotiate the penal repercussions of those activities.

36 Findings reported in this study are not consistent with prior research. Social class has a direct and positive effect on criminal achievement among offenders engaged in common street crimes. Sons of well-off fathers are more likely to succeed in crime, more ambitious, more likely to seek and find criminal mentors; those mentors are more likely to improve their protegés’ chances of succeeding by facilitating access to entrepreneurial crime (rather than “employee” or “appropriation” crime). These conclusions do not exclude the scenario that successful delinquent trajectories may coincide with an upward social mobility pattern (as suggested in Merton’s theory). But this eventuality should be theorized rather than derived a priori from an implicit social mobility assumption. Social class origin also impacts on offenders’ ability to negotiate legal adversity advantageously. Privileged social background increases their motivation to fight back, to hire private lawyers to defend their interests; second, it indirectly increases (i.e., through higher criminal earnings) their ability to invest the considerable sums required to get “full” legal defense. We have seen that the services offered by private lawyers to their most wealthy clients are significant: the disparity in sentence lengths for defendants at the bottom and top of the social ladder, for equivalent offenses and other characteristics, amounts to a sentence length reduction of one-third. Past sentencing research suggest that lawyers objectively work against their clients’ interests (Blumberg 1967) or that they offer no particular advantage (Nardulli 1986). But no such study has simultaneously taken into account defendants’ legal expenses (for a counter example see Feeley 1979), social class backgrounds and relative success in criminal pursuits. It is reasonable to conclude that studying the micro-dynamics of penal or legal inequalities requires examining the aims of delinquent behavior and the results delinquents manage to obtain (Cusson 1981). Still, the immediate or remote causes of the successes and failures that punctuate delinquency trajectories also have to be explicitly examined.