Gender and Delinquenvency

8/3/2019 Gender and Delinquenvency

1/29

The Class Structure of Gender and Delinquency: Toward a Power-Control Theory

of Common Delinquent Behavior

John Hagan; A. R. Gillis; John Simpson

The American Journal of Sociology, Vol. 90, No. 6. (May, 1985), pp. 1151-1178.

Stable URL:

http://links.jstor.org/sici?sici=0002-9602%28198505%2990%3A6%3C1151%3ATCSOGA%3E2.0.CO%3B2-L

The American Journal of Sociology is currently published by The University of Chicago Press.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtainedprior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content inthe JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/journals/ucpress.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academicjournals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community takeadvantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.orgThu Nov 29 04:03:51 2007
http://links.jstor.org/sici?sici=0002-9602%28198505%2990%3A6%3C1151%3ATCSOGA%3E2.0.CO%3B2-Lhttp://www.jstor.org/about/terms.htmlhttp://www.jstor.org/journals/ucpress.htmlhttp://www.jstor.org/journals/ucpress.htmlhttp://www.jstor.org/about/terms.htmlhttp://links.jstor.org/sici?sici=0002-9602%28198505%2990%3A6%3C1151%3ATCSOGA%3E2.0.CO%3B2-L


2/29

The Class Structure of Gender andDelinquency: Toward a Power-ControlTheory of Common Delinquent Behavior1John Hagan, A. R. Gillis, and John SimpsonUn ive rsity of Toron to

Though seldom considered together, class and gender are amongthe most frequently analyzed correlates of delinquency today. Thispaper formulates and tests a neo-Marxian, class-based, power-control theory of gender and delinquency. Using this theory and aprediction made by Bonger more than a half-century ago, the articledemonstrates that the relationship between gender and commonforms of delinquency declines with each step down the class struc-ture. Furthermore, where this relationship is strongest, it can bestatistically removed by taking theoretically predicted variables intoaccount. A power-control theory does much to specify and explainthe class structure of gender and delinquency, and in doing so itdemonstrates the social bases of this relationship.

Class and gender are among the most frequently analyzed correlates ofdelinquency today. Gender is a strong and consistent correlate (e.g., seeSimon 1975; Harris 1977), whereas class is weak and uncertain (Hin-delang, Hirschi, and Weis 1981). The situation is an embarrassment tosociological theories of delinquency. Although most such theories attachgreat importance to class, there is doubt about the correlation on whichthey rest. Furthermore, although it is generally assumed that the effect ofgender is socially based, there is no clear evidence that the gender-delinquency relationship can be removed when social variables are takeninto account. Class apparent ly accounts for too little delinquency (Hirschi1972); gender stubbornly accounts for too much (Steffensmier 1980).Curiously, the issues of class and gender have not been joined in delin-quency research. We believe that this is a crucial omission, for a com-

This study was made possible by funding from the Social Sciences and HumanitiesResearch Council of Canada and the Ministry of the Solicitor General of Canada. Weassume full responsibility for the results and interpretations presented here. Requestsfor reprints should be sent to John Hagan, Faculty of Law, University of Toronto,Toronto, Ontario, Canada M5S 1 A l .0 1985 by The University of Chicago. All rights reserved0002-960218519006-0001$01.50

AJS Volume 90 Number 6 1151


3/29

American Journal of Sociologybined consideration of class structure and gender is the key to a sociolog-ical understanding of the effect of gender on delinquency.The failure to link class and gender in delinquency research betrays aneglect of classical criminological theory. The father figure of Marxiancriminology, William Bonger, offered one of the first statistical demon-strations of the strong correlation between gender and criminality. Hethen pointed specifically to the importance of class structure for a theoret-ical understanding of the social basis of this relationship: "A very conclu-sive proof of the thesis that the social position of woman is what explainsher lower criminality, is as follows. T he difference in the manner of life ofthe two sexes decreases as we descend the social scale. If the social posi-tion of woman is then an important determinant of her lower criminality,the figures ought to show that the criminality of men differs more fromthat of women in the well-to-do classes than in classes less privileged"(1916, p. 477).

Of course, Bonger had neither the data nor the analytic techniques totest adequately his prediction of the interactive effect of class and genderon delinquency. Today we do; but such testing requires that we first thinkcarefully about several issues of theory and measurement.

THE STUDY OF CLASS AND DELINQUENCYTwo fascinating empirical regularities involve the level of agreement thatsurvey respondents demonstrate in ranking the prestige of occupationsand the seriousness of crimes (Rossi et al. 1974, p. 224). These regularitiesmay explain why sociologists who have sought to link class with delin-quent behavior have substituted the measurement of socioeconomicstatus for the operationalization of class and have worried so much aboutthe seriousness of the illegal acts they have studied. By carefully measur-ing socioeconomic status and/or focusing on serious offenses, they mayhave hoped to salvage the theoretically expected class-delinquency rela-tionship. The magnitude of the results is a questionable match for theefforts expended: when found, the relationship is modest (Elliot and Age-ton 1980; Thornberry and Farnsworth 1982), uncertain (Braithwaite1981), and possibly in decline (Tittle, Villemez, and Smith 1978).

The paucity of prior results alone might encourage a rethinking of theclass-delinquency issue, but there are also theoretical and methodologicalreasons to pursue alternative formulations. To begin, the substitution ofsocioeconomic status for class is inappropriate. Such measures are aWeberian offshoot of the Marxian conceptualization of class (Bendix1974). An attractive feature of these measures is that they provide pre-cise, continuous scores that can be used to rank individuals in terms ofstatus. Delinquency theories, however, rarely focus on such fine grada-


4/29

Gender and Delinquencytions of status. Hirschi makes this point with regard to the lower end ofthe class structure, noting that "the class model implicit in most theoriesof delinquency is a peculiarly top-heavy, two-class model made up of theoverwhelming majority of respectable people on the one hand and thelumpenproletariat on the other" (1969, p. 71). Conflict and Marxian theo-ries extend attention to the top of the class structure. Still, there is notheoretically informed basis for dividing gradational status measures intodiscrete class groupings. Alternatively, neo-Marxian scholars have devel-oped survey measures tha t operationalize the classes in relational, that is,structural, rather than gradational terms (e.g., Wright 1980, p. 198).Within this framework, classes are conceived as not merely "above" or"below" one another. Instead, they are defined in terms of their socialrelation to one another, with each class located in a discrete structuralposition within the social organization of the relations of production.

This kind of relational class measure seems especially well suited to thejuncture we have reached in the empirical study of class and delinquency.Gradational status measures have led researchers to look for an uncondi-tional linear relationship between class and delinquency. But the effect ofclass on delinquency may be conditioned, indeed suppressed, by othervariables (Hirschi 1969, p. 73). Or, to put the matter the other wayaround, as in the discussion of Bonger above, class may condition theinfluence of other important variables (e.g., gender) on delinquency. Gra-dational measures of status do not lend themselves to the exploration ofthese kinds of discrete conditional relationships.The measurement of delinquency is an equally important issue. Self-report surveys were an important innovation in the measurement ofdelinquency. They made it possible to collect extensive information onsuspected causes of delinquency, along with first-person reports ofdelinquent behavior. Early self-report surveys concentrated on minor butfrequent forms of juvenile misconduct that were only weakly, if a t all,related to socioeconomic status. Recent surveys have concentrated onmeasures of more serious delinquency. The move to more serious self-report items has important methodological and theoretical implications.Methodologically, as Hindelang et al. (1979) note, very serious forms ofdelinquency (such as murder, forcible rape, and armed robbery) aresufficiently rare to make survey designs problematic. Equally important,however, is the fact that conflict, Marxian, and other theories of crimeregard conceptions of seriousness as a matter to be explained rather thanassumed (e.g. , Black 1979). Th e recent emphasis on "serious" forms ofdelinquency may therefore mistake an issue of theory for an issue ofmethod. Certainly, it is a mistake to equate what is serious with what isimportant. Consider the following:

Theories of adult criminality are often extensions of theories of delin-


5/29

American Journal of Sociologyquency. This is partly because of the large theoretical importance at-tached to childhood and adolescent socialization experiences, but alsobecause adolescents are more easily studied. We have few etiologicaltheories of white-collar crime (Wheeler 1976; Hagan, Nagel, and Al-bonetti 1980) and no theories of white-collar delinquency. If we restrictour attention to very serious forms of delinquency, there will be little onwhich to build such theories: the theoretical problem of white-collar de-linquency will have been defined away by our measures. Because therewill also be very few serious female delinquents, it may be similarlydifficult to study gender and delinquency. We are arguing, then, for themoderation of a trend, for the study of more common as well as moreserious forms of delinquency. Hindelang et al. defend this position wellwhen they note that "self-report measures of delinquency must reflect thedefinition of delinquency implicit in the theory at issue. . . . Restrictingresearch to a single measuring device would inhibit the growth of newtheories and would lock the field into a rigid pattern of social reporting oraccounting" (1981, pp. 88-89). We now turn to the theory of delinquencywe wish to address.

TOWARD A POWER-CONTROL THEORY OF COMMONDELINQUENT BEHAVIORTwo concepts organize the classical theories of delinquency: power andcontrol. The empirical distinction between these concepts is partly one oflevel of analysis. Power theories tend to be macrostructural and controltheories microstructural, but they share a structural interest in relationsof dominance. Power theories focus on relations of dominance that derivefrom control over the means of production; control theories focus onrelations of dominance established within the family. In this article wejoin parts of these two theoretical traditions to form a power-controltheory of common delinquent behavior.Our discussion will focus on what we have noted to be one of thestrongest and most consistent correlations in delinquency research: thatbetween gender and delinquency. Power-control theory specifies wherethis correlation is strongest and most difficult to remove, as well as whereit is weakest and most easily explained. Relations of dominance empha-sized in the power tradition, and defined in terms of class, are used tospecify the conditions under which the gender-delinquency relationshiprises and falls. Relations of dominance emphasized in the control tradi-tion and explored in terms of the family are used to explain gender-delinquency relationships within specific classes.

Both the power and control traditions lead us to consider the conditionsunder which adolescents are free to deviate from social norms. Both the


6/29

Gender and Delinquencypresence of power and the absence of control contribute to these condi-tions. It is assumed that freedom to deviate is directly related to classposition, that males are freer to deviate than are females, and that malesare freest to deviate in the higher classes. Note that this set of assump-tions forms a basis for a prediction such as Bonger's: the relationshipbetween gender and delinquency will increase with movement up theclass structure, and it will decrease with movement down.Our reversal of the theoretically expected, negative class-delinquencyrelationship is unconventional, but not unprecedented. Indeed, the pro-posed positive relationship is as durable as the observation that powercorrupts and has found one place in sociological theory through Sorokinand Lunden's (1959) Powe r and Moral i ty . They propose that power hasan "intoxifying" effect, such that holders of power come to see themselvesas above ( i.e ., free of) the moral and legal precepts that control ordinarypersons. The expected result is that "the moral behavior of ruling groupstends to be more criminal and submoral than that of the ruled strata ofthe same society" (1959, p. 37). Because "ruling groups" have not beenmeaningfully distinguished in self-report research, this proposition re-mains untested for delinquents as well as adults. Instead, it thrives onstories of "rich kids" and tales such as those about the young Kennedys(Collier and Horowitz 1984).A similar theme was suggested in Veblen's Th e Theory of the LeisureClass. In a passage that stimulated Matza and Sykes's (1961) theory ofsubterranean values, Veblen wrote that "the ideal pecuniary man is likethe ideal delinquent in his unscrupulous conversion of goods and personsto his own ends, and in a callous disregard of [i.e., freedom from] thefeelings and wishes of others or the remoter effects of his actions" (1934,p. 237). Matza and Sykes argue that this similarity reflects a dispersion of

leisure class values-the search for adventure, excitement, and thrills, orwhat we call a "taste for riskn-throughout society, causing delinquencyat all class levels. They call special attention to common forms of "whitecollar delinquency" (p. 718) but stop short of asserting a positive class-delinquency relationship. The dispersion they emphasize has a democ-ratizing, and therefore diminishing, effect. Still, a small positive relation-ship between class position and common forms of delinquency is fullyconsistent with Matza and Sykes's theory: the dispersion they propose isdownward through the class structure. A power-control theory of delin-quent behavior proposes a class-delinquency relationship of similar sizeand direction.

If it is indeed relational position in the social structure, rather than typeof individual, that explains delinquent behavior, it should be possible tospecify and explain the gender-delinquency correlation by taking rela-tional position fully into account. In the data analysis that follows, we


7/29

American Journal of Sociologyconsider these two kinds of relations of dominance: the controls exercisedor experienced by the head of household in relation to others in theworkplace and the controls exercised by parents in relation to their chil-dren. We have already discussed the role of class relations in specifyingthe gender-delinquency relationship. We turn now to the role of familialcontrols in transmitting the effects of gender on delinquency within classcategories.A fundamental instrument-object relationship structures family-basedrelations of dominance (Hagan, Simpson, and Gillis 1979). The two sidesof this relationship are that mothers more than fathers are the instrumentsof familial controls and that daughters more than sons are the objects offamilial controls. This relationship is the core of what Rosabeth Kanter(1974) calls the "intimate oppression" of informal social control. This isthe kind of relationship that a Marxist-feminist theory suggests is centralto the "reproduction of order ." There is evidence (Cummings 1977) thatsuch an instrument-object cycle even persists among working womenwho come to believe in "Horatia Alger as a feminist role modeln-awoman who makes time to be both the primary socializer of her childrenand the architect of a career. In other words, even among more liberatedwomen, the instrument-object relationship may be perpetuated. How,then, does this relationship mediate the effects of gender on delinquency?

The answer to this question ties family relations of dominance to issuesof deterrence and legal sanctions. Gibbs provocatively observes, "Thesecret scandal of the Marxist theory of criminal law is that it tacitlyattributes validity to the deterrence doctrine" (1978, p. 106). Gibbs makeshis point by posing a rhetorical question: "How can legal punishment beused as a repressive instrument by a dominant class if the threat ofpunishment does not deter?" A power-control theory of delinquent behav-ior asserts that threat of punishment, or at least the perception of such athreat , does deter delinquency. This much is not new. What is new is ourargument that the bases of this repressive effect are the relations of domi-nance established in the family. That is, adolescents, especially femaleadolescents, are taught to avoid risks generally and the risk of legalsanctions specifically. The testable implications of this part of a power-control theory of delinquency are that females will be deterred more bythe threat of legal sanctions than males and that this effect will be pro-duced more through maternal than paternal controls. The causal modelof gender and delinquency that we have described is summarized in figure1. The class categories within which this model will be explored are setout in table 1 (discussed in greater detail below).The class structure of the gender-delinquency relationship should bereemphasized. What a power-control theory of common delinquent be-havior is saying is tha t in all classes males are freer to be delinquent than


8/29

Gender and Delinquency-

Gender Parental Taste for Psrcsivsd Self-ReportedControl Rish Sanction Delinquency

MaternalPatsrnal+ RiskGettingCaughtGetting -b

PunishedA A A

FIG. 1.-Causal model of gend er an d delinqu ency

females but that it is in the most powerful classes that males are freest tobe delinquent. The presence of power and the absence of control play ajoint role in specifying and mediating this gender-delinquency relation-ship.

TH E DATABefore we introduce measures of theoretical concepts, it is necessary todescribe our data, which come from a survey conducted in the Torontometropolitan area. Past surveys have not included the employment infor-mation necessary to form the relational class measures that are central toa power-control theory of delinquent behavior. This is understandable, inthat adolescents might not have been assumed to have accurate knowl-edge of the necessary information. Parents of students included in oursurvey were followed up by telephone, and we collected from them theinformation needed for our measure of class position, which is discussedin detail below. Our survey brings together this indicator of class positionwith measures of other etiological variables for the first time in a singlestudy.The survey was conducted during the first four months of 1979. Thepopulation consisted of the students and parents from seven secondaryschools serving widely varied neighborhoods. The sampling frame con-sisted of school board lists of the names and addresses of students, fromwhich a stratified random sample was drawn. Addresses with apartmentand unit numbers were used to distinguish respondents living in multiple-family and single detached housing, allowing us to select equal numbersof respondents from each type of residence. Our purpose was to assurethat our sample varied widely in class composition (see also Simpson andHagan 1983).We paid each student $5.00 to participate at school in the survey. By


9/29

American Journal of Sociologypaying the students and assuring them of the confidentiality of theirresponses, we communicated the seriousness of the study. We believe thatthis increased the quality as well as the quantity of participation. Thequestionnaire was administered to groups of students and the responserate was 7 2%. One of the investigators read the questionnaire aloud, withrespondents following along and filling out their own questionnaires. Weused this procedure to increase comprehension and the reliability of re-sponses. Parents subsequently were contacted by telephone to obtain theemployment data that we discuss next. Analyses reported in this paperinvolve 458 adolescents. For all of them, a head of household was knownto be either employed or unemployed (not retired, a student, or a house-wife). Additional occupational information was available concerning em-ployed heads of household. Within-class means and standard deviationsfor variables described below are presented in the Appendix.

A NEO-MARXIAN MEASURE OF CLASSAlthough the concept of class is central to Marxian and conflict theories ofcrime (Colvin and Pauly 1983; Spitzer 1975; Greenberg 1977; Chamblissand Seidman 197 1; Quinney 197 7; Taylor, Walton, and Young 1973), thisconcept has never before been operationalized in self-report research (cf.Hagan and Albonetti 1982). Our approach to the measurement of class isinformed by Marxian ideas. Most important, we proceed from the as-sumption that classes are to be measured in relational rather than grada-tional terms. We use three conceptual criteria to distinguish four classpositions. The three criteria consider control over the means of produc-tion, control over the work of others, and relationship to labor power.The four class positions include employers, managers, workers, and un-employed workers, whom, following Marx, we call the surplus popula-tion. The relationships between the conceptual criteria and class catego-ries are presented in table 1. Respondents are located in class categorieson the basis of four survey questions:

1. "Is the head of your household currently working full-time?"2 . "Does the head of your household currently work for himlherself orfor someone else?"3. "Are there any people who work for him or her or are paid by him orher?"4. "Does the head of your household supervise anybody as part of hisor her job?"Employers are first of all o wn e r s of the means of production. Thiscriterion is reflected in the second item above: those who work for them-selves own some means of production. Second, employers are in control

of employees. An affirmative response to question 3 above captures this


10/29

Gender a nd DelinquencyT A B L E 1

C R ITER IAFO R C L A S SCATEGORIES

ClassControl overmeans ofProduction

Control overWork ofOthers

Relation toLabor Power Distr ibution

EmployersManagersWorkers .Surplus population

. . . . . . . . .. . . . . . . . . .. . . . . . . . . .

. .

O w n e rN onow nerN onow nerN onow ner

Controls em-ployees

Controls sub-ordinates

No controlof others

No controlof others

Buys laborSells laborSells laborUnable to

sell laborNOTE.-IV'S in parentheses

second dimension. This second criterion is important as a means ofeliminating from the employer class persons who are self-employed butemploy no others, that is, members of the petty bourgeoisie. (We do notconsider the latter class in our analysis because Marxists typically arguethat "the petty bourgeoisie represents a remnant from an earlier era ofcapitalist development and, as a class, . . . is progressively becoming lessimportant" [Wright and Perrone 1977, p. 431.) Third, employers arebuyers of labor power . This criterion is satisfied by affirmative responsesto questions 1 and 2 on current work status. Defined in this way, employ-ers constitute 8.95% of our sample. We refer to the occupants of this classposition as "employers" rather than "capitalists" because most in thiscategory employ fewer than 10 workers (Wright 1978, p. 1370). Ideally,we would distinguish between small employers and actual capitalists. AsWright et al. (1982, p. 712) have recently noted, however, to do so is torestrict empirical analyses to an extraordinarily small part of the popula-tion that is difficult to reach with a survey design. Our strategy followsWright in merging small with large employers into a more diffuse "em-ployer" class-category. Our employer class is within about 1% of theestimate given for such a class in Wright's work (see Wright 1978; Wrightand Perrone 1977; Wright et al. 1982).Our second class-category consists of managers. Members areidentified first as n o n o w n e r s by a negative answer to the question aboutself-employment. They are also identified as being i n con trol o f subord i -nates by an affirmative answer to the fourth question. Finally, managersare sellers of labor. Heads of household are classified thus on the basis of apositive answer to the first question, about current work status, and anegative answer to the second question, about self-employment. Defined


11/29

American Journal of Sociologyin this way , the m anagerial class const itutes 43.89% of ou r sample. T hisfigure corresponds closely to the "minimum" estimate given to a com-bined grouping of managers and supervisors in Wright 's most recentAm erican survey (W right et al . 1982), a grouping th at corresponds tow ha t we a re call ing the managerial class . Thi s grouping can be reducedby adding considerat ions of sanct ion or task authori ty and hierarchy tothe cr i ter ia applied. As no ted in a n earlier sect ion, how ever, del inquencytheory has focused on the top a n d bottom of the class s tructu re, an d weha ve therefore no t introduc ed this detail into the m iddle levels of ou r classanalysis.O u r third class-category is mad e u p of workers . M em bers of this classare obviously nonowners of the m ean s of produ ction; this fact is cap tur edby the self-employm ent item . Also, they exercise no control over the workof others. Negative answers to quest ions 3 and 4 place heads of house-holds in this category. Finally, they are sellers of labor; this criterion issatisfied by the foregoing information and a positive answer to the ques-t ion about current work status. Defined in this way, the working classconst i tutes 37.99% of our sample. Wright et al . observe, "If . . . weexclude from the wo rking class those w ho ind icate in their que st ionnairestha t they in an y way supervise o ther people or th a t they have even verymodest levels of au ton om y, th e size of the w orking class is redu ced to 35percent" and that "what these figures suggest is that 35 percent of allposi tions are unamb iguously w orking class . . ." (1982, p. 718). Altern a-tive criteria increase the size of th e w orkin g class an d decrease th e size ofthe managerial class. If our criteria are in error, i t is probably in thedirection of pro tecting the integrity of the w orking-class category.

T h e last category in ou r classification consists of unemp loyed wo rkers,whom we cal l the surplus populat ion. Although Marxian cr iminologistshave emphasized the importance of this class (Colvin and Pauly 1983;Spitzer 1975; Clel land an d C arter 1979), recent Marx ian o perat ional iza-t ions of the class s truc ture (e .g. , Wrigh t 1978; Wright an d Perrone 1977;Robinson an d Kelly 1979; W right et al . 1982) hav e no t included i t . This isi ronic because Marx at tached great importance to this "surplus popula-tion" in deve loping his "general la w of capitalist accu mu lation" (1912, p.7) . Because del inquency theories in general , and Marxian theories ofcrime in particular, attach great significance to the bottom of the classstructure, we include the surplus populat ion in our operat ional izat ion.T he su rp lus popu la tion a re nonowners of the means of production, asindicated by negative responses to the self-employment quest ion; theyhave no control over other workers, as indicated by responses to quest ions3 and 4 ; and they a re unable to sell their labor, as indicated by theirresponses to the quest ion o n cu rrent wo rk s tatus. Defined in this way, th e


12/29

Gende r a nd De l inquencysurplus population consti tutes 9.17% of our sample, a f igure that corre-sponds wel l to curren t unem ploymen t s ta t is t ics .Overal l , our measure appears to represent the c lass s t ructure wel l . I tmay s l ight ly underrepresent the working class an d s l ightly overrepresentthe managerial class, but even so, the representation of these classes iswithin the range of reasonable estimates established by Wright 's (1982)American su rvey. At m inim um , w e hav e achieved our goal of secur ing asample of broad class composit ion. For purposes of comparison, we alsoinclude in ou r analyses D un can 's (1961) Am erican socioeconomic index ofoccupat ions and Bl ishen 's (1961) Canadian occupat ional c lass scale .These belong to the type of gradational status measures used in conven-tional self-report surveys.

A THEORETICALLY RELEVANT SCALE OF DELINQUENCYT h e theory we have proposed assumes that the presence of power an d theabsence of control exercise their influence, a t least in p ar t , throu gh acognitive process in which actors evaluate courses of action. We havetherefore included variables representing actors ' socially acquired tastesfor r isk an d their perception s of the risks of gett ing cau gh t an d punish ed,for del inquent beh avior , as crucial med iat ing factors in the causal modeldepicted in figure 1. Th ese variab les represen t cognitive states. F or themto be operat ive , there must be some calculat ion that leads to the del in-qu en t beha vior considered. O u r premise is th at this will be truer of minorforms of theft and aggression than i t will be of more serious criminalbehavior, particularly the crimes of violence emphasized in indices of"serious" cr im e an d del inquency. O urs is a theory of com mon d el inquentbehavior .Past s tudies of common del inquency have included "many i tems thatare not sufficiently specified to justify the assumption that the reportedbehavior w ould be reasonably def inable as del inquent" (Hindelang et a l .1981, p . 45) . We have addressed this problem by using an adapted ver-sion of Hirschi 's (1969) self-report delinqu ency scale. O ur only alterationwas to restrict the period of coverage to the preceding year, a decisionenco urage d by Hirschi 's sub seq ue nt analysis (see 1969, p. 62), by reph ras-ing the i tems an d using the following as response categories: never, once,two to three t imes, of ten, an d m any t imes. T h e s ix- item scale asked howoften in th e las t year the respondents had : taken things (worth less than$2.00, between $2.00 an d $50, over $50) tha t did not belong to them;taken a car for a r ide without the owner 's permiss ion; in tent ional lybanged u p som ething th at d id not belong to them ; an d, excluding fightswith a brother or s is ter , in tent ional ly beaten up or hur t anyone. Hirschi


13/29

American Journal of Sociologycontrasts this scale with others, noting that "items included in our delin-quency scale have logical validity, since they measure petty theft andgrand larceny, auto theft, vandalism (malicious mischief), and battery-all offenses tha t are commonly thought to result in punishment by agentsof the larger society, if detected" (1969, pp. 55-56).

Some argue that common delinquencies are not reported as accuratelyas serious ones (Kleck 1982). The recent work of Hindelang, Hirschi, andWeis suggests that the reverse is true, "apparently because the latter aretypically more complex and ambiguous than the former" (Hirschi et al.1982, p. 434). A second concern involves class bias in recall. Hindelang etal. find no evidence of class bias and instead conclude that "self-reportdelinquency measures are as valid among lower-class as among middle-class white males" (1981, p. 196). If there were class bias, recent theoriz-ing suggests that it would make self-reports of common delinquency aconservative test of a positive class-delinquency relationship, because"middle- and upper-class people have more to lose by self-disclosure andtherefore would have more incentive to lie" (Tittle, Villemez, and Smith1982, p. 437; see also Hirschi 1969, p. 60). Scores on the six items areadded together to form the delinquency scale used below. The a reliabil-ity coefficient is .78.

Measurement of Intervening VariablesThree kinds of intervening variables are included in our causal model ofgender and delinquency: parental controls, taste for risk, and perceivedsanction risk. Our measures of parental controls are drawn from the workof Hirschi, who regards them as indicators of "attachment" (1969, pp.88-89). The four questions ask, "Does your (father) (mother) know(where you are) (who you are with) when you are away from home?" Theresponse categories-"always," "usually," "sometimes," and "nevern-are used to form separate two-item additive measures of maternal (a =.66) and paternal ( a = .78) controls. We use these items in conjunctionwith sex of respondent to explore the instrument-object relationship be-tween mothers and daughters postulated above. We believe that thesemeasures reflect the means by which important gender relations are es-tablished, with implications for intervening attitudes and consequent ac-tions, including delinquent behavior.Among the attitudes that we expect to be influenced by the instrument-object relationship we have emphasized is a socially acquired taste forrisk (a= .67). This is measured by adding Likert-scaled responses to twostatements: "I like to take risks" and "The things I like to do best aredangerous." Power-control theory predicts that taste for risk is sexuallystratified, with males more than females taught to value risk taking. In


14/29

Gender a nd Del inquencyturn, the taste for risk is expected to sexually stratify perceived risks ofgett ing caught and punished for del inquent behavior .There is now solid evidence that perceived risks of sanctioning deterdelinquent behavior ( e.g . , Jensen, Erickson, an d G ibbs 1978; Erickson,Gibb s, an d Jensen 1977). O u r causal model predicts tha t a higher taste forrisk leads to a lower perceived risk of sanctioning; in other words, tastefor risk and perception of risk are inversely related. Perceived risks ofgetting caught an d punished a re measured separately . T he three "r isk ofgetting caught" items are derived from the work of Jensen et al. (1978)and involve estimations of the likelihood of getting caught for specificdelinquent acts . R espondents were asked: "Could you (break in to a p lace)(steal from a store) (write graffiti) and not get caugh t?" Liker t responsesranging from "definitely yes" to "definitely no" were summed to scaleperceived risks of getting cau ght (a = .76). Perceived risk of punishm entwas m easured by asking respon dents to agree or d isagree in Likert fash-ion to more generally formulated statements such as, "If you break thelaw , you w ill w ind up being (charged by police) (sent to cou rt) (sent to aninstitution)." Scores on these items were sum me d to scale perceived risksof punish m ent (a = .74). O ur interes t is in the deterrentirepressive effectsof these perceptions and their mediating role in explaining the gender-delinquency relationship within the class categories whose effects we con-sider next.Eq u a t io n sIn th e first pa rt of ou r analysis we estimate the following equations :

self-reported delinquency = a + b l American SES, (1 )self-reported delinquency = a + b l Can ad ian SES , ( 2 )self-reported delinquency = a + b l Marxian four-class measure, (3)

an dself-reported delinquency = a + b l Marxian two-class measure (4)

+ b2 gender.This set of equations is estimated for each of the six self-report delin-quency items and then the scale (see table 2). The first two equationsreplicate traditional analyses of socioeconomic status and delinquency.T he third substitute s a neo-Marxian four-class me asure for socioeconomicstatus. O ur fou rth eq uation regresses self-reported delinquency on a moretradit ional , Marxian two-class measure that merges the surplus popula-tion, workers, and managers into a single working class, which is con-trasted with employers. This equation also includes gender. Both the


15/29


16/29

G ende r a nd D e l inquencyM arx ian two-c lass measure an d gender a re dum m y variab les , wi th em -ployers and men each coded one and workers and women coded zero.Gen der is unrelated to class or S E S , so inclusion of gender in the firstthree equat ions would provide no addi t ional information. As we will see ,the first three equations are of interest for comparative purposes only.

So far w e ha ve considered only the main effects of class an d gender o nself-reported delinque ncy; these effects represent only a small p ar t of ou rtheoretical interests. The remainder of our analysis explores the interac-tion of class an d gender on th e delinqu ency scale by estim ating within th efour neo-M arxian classes the effects of gender o n delinquen cy. T h e causalmodel of gen der an d delinq uenc y presented in figure 1 form s the basis forthis analysis. We explore this model with a set of reduced-form andstructura l equat ions es t imated w ithin each class . Tab les 3-6 present thereduced-form and structural coefficients for the four classes. The equa-tions are:

self-reported de linquen cy scale = a + b l gender , ( 51self-reported delinquency scale = a + b1 gender + b2 mate rna l (6)controls + b3 patern al controls ,self-reported delinq uenc y scale = a + b1 gender + b2 maternal ( 7 )

controls + b3 paternal controls+ b4 taste for r isk,a n d

self-reported delinq uenc y scale = a + b l gender + b2 maternal (8)controls + b3 patern al controls+ b4 taste for risk + bS riskof getting caught + b6 risk ofgetting punished.

T h e first colum ns of tables 3-6 presen t the regression of delinqu ency ongender (eq. [S]), providing a measure of the total effect of gender withinthe specified classes. T h e second column ad ds the tw o sources of paren talinfluence, matern al a nd patern al controls (eq. [6]). T h e third column ad dstaste for r isk (eq. [7]) . T h e last column ad ds the perceived risks of gett ingcaug ht a n d getting punish ed (eq. [8]) . T h e differences in the coefficients inan y two adjac ent columns represent the indirect effects of var iables in thepr ior column by way of var iables added in the subsequent column.Coefficients in the fourth column are the direct effects of the variables(Alwin and Hauser 1975). Comparison of the gender coefficients in thefirst column of each table will allow us to test the declining effect ofgender on del inquency predicted by Bonger to occur with each s tep downthe class structure. Consideration of the effects estimated in the remain-


17/29

American Journal of SociologyTABLE 3

REGRESSIONS DELINQUENCY VARIABLESF SELF-REPORTED O N INDEPENDENTIN R E D U C E DFORM EQUATIONS CLASSND STRUCTURAL WITHIN EMPLOYER

Gender: b . . . . . . . . . . . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . . . . . . . . Cons tan t

Parental controls: Mate rna l : b . . . . . . . . . . . . . .

B . . . . . . . . . . . . . . . . . . . . . . Pate rna l : b . . . . . . . . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . . . . . . . .

ConstantTaste for risk: b . . . . . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . . . . . . . .

Cons tan tPerception of risk:

Getting caught: b . . . . . . . . . . . . .p . . . . . . . . . . . . . . . . . . . . . . . . . .Getting punished: b . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . . . . . . .

Cons tan t . . . . . . . . . . . . . . . . . . . Indirect effect . . . . . . . . . . . . . . . . . . . % of total effect . . . . . . . . . . . . . . . . .

NOTE.-Standard errors in parentheses.* Significant at 10 level * * Significant at .05 level *** Significant a t 01 level.

ing columns will allow us to determine whether gender exercises its in-fluence within classes in the way our causal model and earlier theoreticaldiscussion predict.

RESULTSThe regression coefficients that result from estimating equations (1)-(4)are presented in table 2 . Recall that the main effects estimated in this


18/29

Gender a nd De l inquencyTABLE 4

Gender: b . . . . . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . Cons tan t . . . . . . . . . . . . .

Parental controls: Materna l : b . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . Paternal: b . . . . p . . . . . . . . . . .

Cons tan t . . . . . . . . . . . . . Taste for risk: b . . . . . . . . . p . . . . . . . . . . . . . . . . . . .


Getting caught: b . .p . . . . . . . . . . . . . . .Getting punished: b . . . . p . . . . . . . . . . . . . . . . . . .

Cons tan t . . . . . . . . . . . . . Indirect effect . . . . . . . . . . . . % of total effect . . . . . . . . . . .

NOTE.-Standard errors in parentheses* Significant a t .10 level. ** Significant at .05 level. *** Significant at .O1 level. **** Significant at 001 level

table ar e not o ur pr im ary interest; ou r focus is the interact ion of c lass andgender on delinquency, which we consider below. Because this researchexplores a new theoretical per spe ctiv e, we use a .1 0 criterion of signifi-cance.As in most prior studies focusing on common forms of self-reporteddel inquency, wh ether a n Am erican (Du ncan 1961) or Can adian (Blishen1961) stat us me asure is used, there is no evidence of a significant relation-


19/29

American Journal of SociologyTABLE 5

Gender: b . . . . . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . . Cons tan t

Parental controls:Mate rna l : b . . . . . . . . . . .p . . . . . . . . . . . . . . . . . . .Paternal: b . . . . . . . . . . . . p . . . . . . . . . . . . . . . . . . .

Cons tan tTaste for risk: b . . . . . . . . . p . . . . . . . . . . . . . . . . . . .


Getting caught: b . . . . . .p . . . . . . . . . . . . . . . . . . .Getting punished: bp . . . . . . . . . . . . . .

Cons tan t . . . . . . . . . . . . . Indirect effect . . . . . . . . . . . . % of total effect

- ~NOTE. -S tandard e r rors in parentheses* Signif icant a t .1 0 level. ** Signif icant a t .0 5 level. *** Signif icant a t . O 1 level. **** Signif icant a t 001 level.

ship between socioeconomic status and common delinquency. The stan-dardized coefficients for equations (1) and ( 2 ) hover around zero anddepart in either direction about equally often, without ever reachingstatistical significance. When a neo-Marxian four-class measure is sub-stituted for socioeconomic status in equation ( 3 ) , the coefficients becomemore consistently positive. But only the coefficient for fighting is sig-nificant (P < .l o) , and it is weak (p = .08). When a more traditionally


20/29

Gender and DelinquencyT A B L E 6

REGRESSIONS O F SELF-REPORTED DELINQUENCY ON INDEPENDENTARIABLESIN REDUCED FORMAND STRUCTURAL EQUATIONSITHIN SURPLUS POPULATION

Gender : b . . . . . . . . . . . p . . . . . . . . . . . . . . . . Cons tan t . . . . . . . . .

Parental controls: Maternal : b . . . . . . . .

p . . . . . . . . . . . . . . . . Paternal: bp . . . . . .

Cons tan t . . . . . . . . . . Taste for r isk: b . . . . . . p . . . . . . . . . . . . . . . .

Cons tan tPercep tion of risk:

Gett ing caught: b p . . . . . . . . . . . . Getting punished: b . p . . . . . . . . . . . . . . . .

Cons tan t . . . . . . . . . . Indirect effect . . . . . . . . . % of total effect . . . . . . .

NOTE.-Standard errors in parentheses* Significant a t . I 0 level. ** Signihcant at 05 level. *** Signihcant at .O1 level. **** Significant at ,001 level

Ma rxian two-class m easure is in troduced in equa tion (4), a m ore consis-tent pattern of significance and slightly greater strength emerges. Five ofthe six item coefficients (the three levels of th eft, ca r thef t, a n d fighting)and the combined scale are significant and positive in direction. Thismeans that children of the employer class are slightly more likely thanthose of other classes to engage in these common forms of delinquency.The standardized coefficient for the scale is .11, with children of the


21/29

American Journal of Sociologyemployer class scoring on average 1.43 points higher than children ofother classes (P < .05) on the delinquency scale.Equation (4) also includes gender, the effect of which is stronger thanclass for every item and the scale. On average, males score 2.28 pointshigher on the delinquency scale than females (P = .31, P < ,001). Thestrongest gender effects are for taking cars (P = .25, P < ,001) andfighting (P = .35, P < ,001). All seven gender coefficients are significant,and the strength of the relationships is consistent with those found inprevious American and Canadian studies (see Linden and Fillmore 1980).The foregoing main effects of class and gender are consistent with thepower-control theory we have proposed. The relative power that derivesfrom being male and located in the employer class produces the expectedeffects in terms of somewhat higher rates of common delinquency. Themore challenging prediction of the theory is that class and gender interactin such a way that males from the employer class are the most likely to bedelinquent. We have argued that it is these respondents who are in thehighest positions of power and therefore are most free to deviate. As wewill see, the interaction that tests this set of assumptions is indeedsignificant. Of course, this interaction can be examined in two ways: howgender conditions the effect of class on delinquency or how class condi-tions the effect of gender on delinquency. We will emphasize the latterbecause this is the way that Bonger sets up the problem and the way inwhich we have developed our theory. Nonetheless, before proceeding tothese results, we can first summarize the way in which gender conditionsthe effect of class: males from the employer class score on average 2.39points higher than males from other classes on the delinquency scale (P =.16, P < .001), whereas females from the employer class score on averageonly .41 points higher than females from other classes on this scale (P =.04, P > . l o ) . A between-class t-test for these coefficients indicates thestatistical significance of this difference (t = 2 . 2 2 , P < .05). Gendermodifies and suppresses the effect of class on delinquency in the expectedway.

Bonger, however, implicitly proposed an even more stringent test ofinteraction: he argued that the effect of gender on delinquency woulddecline with each step down the class structure. We have identified fourclasses that are consistent with Bonger's Marxian perspective. By com-paring the results of estimating equation (5) in each of the four classes,that is, the coefficients in column 1 of tables 3-6, we are provided with atest of his thesis. The results follow the steplike sequence predicted: onaverage, males score 4.08 points higher than females on the delinquencyscale in the employer class (P = .41, P < .01), 2.29 points higher thanfemales in the managerial class (6 = .34, P < .001), 2.05 points higher inthe working class (P = .29, P < .001), and only 1.42 points higher in the


22/29

Gender and Del inquencyT A B L E 7

C O MPA R ISO N G EN D ERCOEFFICIENTSF CLASSCATEGORYEmployers1 Employers1 Employers1 Surplus1 Surplus1Equation Managers Workers Surplus Workers Managers

(5):Difference . . . . 1.79t-value . . . . . . . 2.29**

(6):Difference . . . . .66t -value . . . . . . .87

(7):Difference . . . . .95t -value . . . . . . 1 . 28(8):Difference . . . . .4 7t -value . . . . . . . .44* Signif icant at .10 level** Significant at . O j level.*** Significant at .01 level.

surplus populat ion (P = .17 , P > . lo ) .N ote th at th e gender coefficient inthe s urp lus population is reduce d to insignificance. Th ese results stronglysupp ort Bonger 's arg um en t th at the effect of gender can be shown to be aproduct of social posit ion by specifying where in the social structure therelationship is observed.

T a b l e 7 shows the results of t- tests for the interclass comparison ofgender coefficients. The first row corresponds to the estimated equationsth at we ha ve considered w ithin each class to this point. T hese significancetests confirm w ha t is ap pa ren t f rom the gender coeff ic ients : i t i s wi thin theemployer class tha t the effect of gen der differs most from the rest , becausethis coefficient is significantly differe nt from t ha t of gen der in ea ch of theother three classes. A power-control theory of delinquent behavior nowtr ies to address whether we can fur ther specify how this occurs . Thepremise of the theory is that the presence of power and the absence ofcontrol give freedo m to dev iate th at results amo ng adolescents, especiallyemployer c lass males , in com mon forms of del inquency. Eq uat ion s (6)-(8)tes t th is premise by br inging into our analysis the three remaining con-cepts emphasized in our theory: parental controls , tas te for r isk , andperception of sanction risk.

Equ at ion (6) in troduces parental controls , m aterna l and patern al , in toour analysis. Within each class, consideration of parental controls sub-stantially reduces the effect of gen der on delinqu ency (tables 3-6). O naverage , males a re now 2 .44 points higher than females on the del in-


23/29

American Journal of Sociologyquency scale in the employer class (P = . 25 , P < . l o ) , score 1 .78 pointshigher than females in the managerial class (P = . 26 , P < .001), 1.33points higher in the w orking class (P = . 1 9 , P < .05), an d no higher in thesurplus populat ion (b = - . 03 , P = . 00 , P > . lo ) . Note tha t the gendercoefficient in the surplus population is now not only nonsignificant butnonexistent . In the higher classes the gender coefficients are reduced insize but are stil l individually significant. However, the interclass com-parisons in table 7 reveal t h at th e gender coefficient in the employer classis no longer significantly differen t from th at of gende r in the man agerialor working class. Consequently, differences in parental control accountfor the higher del inquency of males in the employer class than in themanag erial or w orking class; the former a ppa rently a re lef t f reer by theirparents to pursue common forms of de l inquency, and they do so . In a l lthree of the higher classes, how ever , males are s t il l more d el inquent th anfemales, even w hen pare ntal controls are taken into account . T his is notthe case in the s urplu s populat ion: w hen paren tal controls are taken intoacco unt here, th ere is no evidence of an y gender-delinquency relation-ship. O veral l , these resul ts confirm th at freedom from paren tal controlshas m uch to d o with differences within a nd between classes in th e effectof gender on del inquency . T hese are the m ost powerful intervening vari-ables in our analysis . We n ow examine w hy this is so.

T h e power-control theory of del inquent beha vior postulates an instru-ment-object relat ionship between mothers and daughters as a key factorin explaining gender differences in del inquency. The instrumental roleassigned to mothers is clear in our data. Within each class, the effect ofm ater na l contro l in equatio n (6) is statistically significant, where as theeffect of pa terna l control is not . I n oth er words, mothers are more instru-menta l th an fa thers in contro ll ing the i r chi ldren . T h a t daughters morethan sons are the objects of this control is reflected in the substantialreduction in the effect of gender on delinquency that results from in-troducing mate rnal control into equation (6). Pate rnal control comes veryclose to exercising significant effects in th e em ployer (b = - . 7 2 , P < .20)and m anager ia l (b = - . 28 , P < .20) classes, an d this effect is largest inthe employer class. Nonetheless, in these classes as in the others, theinstrument-object roles assigned to m others a nd d aug hters is clearly dem-onstrated. The other s ide of this relat ionship is the freedom to deviateth at i t gra nts to adolescent males, part icular ly in the employer class .

We turn next to the sexually s trat i f ied at t i tudes and perceptions thatare postulated in o ur mo del to resul t f rom th e instrument-object relat ion-ship we ha ve identified. T h e first of these , taste for risk, is introdu ced inequation (7) . This variable has a relat ively s trong (P = .31 to .48) an dsignificant effect on de linq ue nt beh avior in all fou r classes (tables 3-6).T he im po rtan t mediat ing effect of this variable in explaining del inquen cy


24/29

Gender and Delinquencyis also reflected in the fact that it substantially reduces the effect ofmaternal controls in all four classes. Nevertheless, although this variablereduces the effect of gender on delinquency in the managerial and work-ing classes (as noted above, there is no remaining significant effect in thesurplus population), it has absolutely no mediating role in the employerclass. The gender coefficients for the employer class in equations (6) and(7) are identical (2.44). The potential importance of this finding is dis-cussed further below.The last intervening variables introduced into equation (8) involveperceptions of sanction risk. These variables reverse the pattern for tastefor risk: only in the employer class do perceptions of sanction risk sub-stantially reduce the effect of gender on delinquency (b = 1.88, P = .19);this effect is no longer statistically significant. More specifically, it is theperceived risk of getting punished (P = - .53, b = - .28, P < . l o ) thatfinally reduces the effect of gender to nonsignificance in this class. Malesare apparently more delinquent than females in the employer class be-cause they are less controlled by their mothers and less likely to perceivethe risks of getting punished as threatening. Under these conditions,males in the employer class are the most delinquent respondents in oursample.

Finally, we call attention to the bottom row of table 7 . This rowpresents the interclass significance tests for the gender coefficients inequation (8), which includes all the variables in our model. When thezero-order gender coefficients were compared between classes in row 1,the employer class coefficient was significantly different from all others.Now the surplus population stands out. That is, with parental controlsand the risk variables taken into account in equation (8), there are nosignificant differences in the size of the gender effects between the upperthree classes. But controlling for these variables in the surplus populationresults in a reversal of sign for gender (b = - .61). Although this effectindividually is not statistically significant, it is significantly different fromthe gender effects in the other three classes. In this sense, at least, theeffect of gender remains unique in the surplus population.

DISCUSSION AN D CO NCLUSIONSOurs is not the first theory of delinquent behavior to be informed byMarxian ideas (see Spitzer 1975; Greenberg 1979; Colvin and Pauly1983). It is, however, the first such formulation to be empirically tested;and it is the first Marxian-based theory to address one of the most impor-tant relationships in delinquency research: that between gender and de-linquency. We have demonstrated, using a power-control theory of delin-quency and a prediction made by Bonger more than a half-century ago,


25/29

American Journal of Sociologythat the relat ionship between gender and common forms of del inquencydeclines with each ste p dow n the class s tructu re. Furtherm ore, w here thisrelationship is strong est, i t can be statistically removed by tak ing theoreti-cally predicted variables into account . A pow er-control theory does muchto specify and explain th e class s tructu re of gender a nd del inquency, an din doing so it demonstrates the social bases of this relationship.

T h e core assumption of ou r theory is th at the presence of power an d theabsence of control create cond itions of freedom th at perm it commo nforms of del inquency. I t is impo rtant to em phasize tha t this formulat ion isdifferent from existing Marxian theories of delinquency. This theoreticaldifference may derive from the different pheno me na selected for explana-tory at tent ion. Recently, Marxian theoris ts have fol lowed del inquencyresearchers in focusing on serious forms of del inquency. For example,Colvin an d Pauly a nno unc e in the f irst sentence of their s tatem ent of a"structural-Marxist theory of del inquency production" t ha t their interestis in "serious pat tern ed d el inq uen t behavior , defined as repeated engage-me nt of a juvenile in the FB I 's Pa rt O ne Index cr imes" (1983, p. 513).

W e hav e noted previously th at persons high in the class s tructure rarelycom mit suc h offenses even on ce, mu ch less repeatedly . Th is is, of course,the class premise of a Ma rxia n theory of serious delinq uenc y. T h e theoret-ical quest ion is , Wh at is i t ab ou t condit ions a t the bottom of the classstructure th at pressures persons to commit such acts?Our focus, however, has been on common forms of del inquency. Wesee no reason to assume th at class, measured in M arxia n terms, is relatedto common forms of del inquency in the same way that i t is to ser iousdelinquency. Inde ed, w e hav e arg ued th at posi tions of pow er, defined interms of class and gender, are conducive to higher rates of commondelinquency. T h e theoretical quest ion we ask is , W ha t is i t ab ou t condi-t ions a t t he top of the social struc ture tha t allows persons to com mit suchacts?T h e f irs t kind of theory assum es tha t people are driven to del inquency;the second, th at they are more del inquent because they are free to be so(see Hirschi 1969, ch ap . 1) . T h at one theory receives empirical sup portneed not m ean t ha t the o ther is wrong. T he tw o kinds of de l inquency maydem and di f ferent explanat ions .

I t is important to clar ify further why we have focused on commonforms of de l inquency. H indelang e t a l . hav e es t imated tha t the ann ualprevalence of serious delinquency in the population is less than 2%-3%(1979, p. 1010). Given the small class fract ions that interest M arx ianscholars ( the surplus populat ion is smaller than 10% and the capital is tclass smaller stil l) and the infrequency of such delinquencies, i t will beextremely difficult to test Marxian theories of serious delinquency. Theskewness of the two distributions will require either exceptionally large


26/29

Gender and De l inquencysamples or samples that are heavi ly s t rat i f ied on the basis of knowncorrelates of ser ious del inquency (e .g . , kno wn cou rt records) . Th e formerapproach wil l require f inancing on a grander scale than contemporaryvictimization survey s (wh ich involv e more th an 100,000 screening inter-views); the latter approach raises serious problems of generalization, in-volving a use of off ic ial da ta sources tha t has th us far proved an ath em a toradical cr iminologis ts. O n the basis of this knowledge, w e fear th at em -pirical tests of Marxian theories of serious delinquency are unlikely toemerge soon.

Meanwhi le , there i s much to be learned by both Marxis t s and non-M arxis ts abou t the class s t ructu re of gender an d more com mon form s ofde l inquency. F or exam ple , our d a t a a re in some ways qui te congenial to aconventional functionalist understanding of stratification and its conse-quences. Consider ou r f indings regarding one variable , " taste for r isk."Th is varia ble plays a significant role in med iating th e gender-delinquencyrelat ionship in th e manag erial a nd w orking class bu t not in the em ployerc lass. T o the ex tent tha t r isk tak ing i s va lued , a t l east among m en inW estern capitalist societies, the form er findings are consistent with tradi-tional und erstandin gs of th e role of socialization in fostering att i tud es thatare conducive to upward mobil i ty . Del inquency can be understood hereas an u ninten ded consequence. Nonetheless , a t the top of the class s truc-ture , males in the em ployer class are more d el inquent than females , notbecause they ha ve a higher taste for risk bu t simply because they are lesscontrolled by their parents and believe that they are less l ikely to bepunished for their del inquencies . This is the s tuff f rom wh ich more cr i ti -cal theories are ma de.O u r findings therefore ar e not all of a piece. T he y do affirm one simpleconclusion, however: Marxists and non-Marxists alike will benefit fromoperat ional iz ing the concepts of c lass , gender , and del inquency as wehave d one in this pap er . Inde ed, our f indings indicate tha t to do otherwiseis to obscure the class structure of gender and common forms of delin-quency.


27/29

Am erican Journal of SociologyA P P E N D I X

M E A N SAND STANDARD OF VARIABLES EQUATIONSEVIATIONS USEDIN REGRESSION

FullSample Employers

ProfessionalManager ia l

Class WorkersSurplus

Population. . . . . . . . . . . . . . . . . . . .ender

Parental control:Maternal . . .Paternal . . . . . . . . . . . . . . . . .

Taste for riskPerceived sanction risk:Getting caught . . . . . . . .

Getting punished . . . . . . . . .Self-reporteddelinquency

NOTE.-SDS in parentheses.

R E F E R E N C E SAlwin, Duane, and Robert Hauser . 1975. "The Decomposi t ion of Effec ts in PathAnalysis ." American Sociological Review 40:37-47.Bendix, Re inha rd. 1974. " Inequal i ty and Socia l Struc ture : A Com parison of Marx a ndWeber . " American Sociological Review 39: 149-61.Black, D ona ld . 1979. "Com mo n Sense in the Sociology of La w ." American Sociolog-ical Revie w 44:18-26.Bl ishen, B. 1961. "The Cons truct ion and Use of an Occupat ional Class Scale ." InCanad ian Socie t y , edi ted by B. Bl ishen e t a l . Toronto: Macmil lan.Bonger , Wil l iam. 1916. Crime and Economic Condi t ions . Book 2. Boston: Litt le ,B r o w n .Brai thwai te , Jo hn. 1981. "Th e M yth of Socia l Class and C riminal i ty Recons idered."American Sociological Review 46:36-57.Chambl i s s , Wi l l i am, and Rober t Se idman . 1971 . Law, Order and Power. Reading ,Mass . : Addison-Wesley.C le l l and , Don a ld , and T im othy Car te r . 1980 . "The N ew M yth of C lass and C r ime ."Criminology 18:319-36.Col l ie r , Pe ter , and David Horowitz . 1984. Kennedys: An American Drama. N e wYork : Summi t .Colvin , Ma rk , an d John Pa uly . 1983. "A Cri t ique of Criminology: Tow ard an Inte-gra ted Structura l -M arxis t Theo ry of Del inquency Pro duct ion." American Journal ofSociology 89:513-52.C u m m i n g s , L. D. 1977. "Value Stretch in Definit ions of Career among CollegeWom en: Hora t io Alge r a s Femin i s t Mode l . " Social Problems 25:65-74.


28/29

Gender and DelinquencyD u n c a n , 0 . D. 1961. "A Socioeconomic Index for All Occupations." In Occupat ionsand Soc ia l S ta tus , edi ted by Albert Reiss . New York: Free Press.Elliot , Delbert , a nd Sw an Ageton. 1980. "Reconciling Race a nd Class Differences inSelf-Reported and Official Estimates of Delinquency." Ame rican Soc io logical R e-

v i e w 4 5 9 - 1 10.E r ickson , M. L . , J . P . G ibbs , and G . F . J ensen . 1977 . "De te r rence and the Pe rce ivedCerta inty of Legal Punishment ." Amer ican Soc io logical Rev i ew 42:305-17.Gib bs, Jack. 1978. "D eterren ce, Penal Policy and the Sociology of La w ." In Researchi n Law and Soc io logy , ed i ted by R i ta S imon . Greenwich , Co nn . : JAI .Greenberg, David. 1977. "Delinquency and the Age Structure of Society." Contempo-rary Crises 1:189-223.H ag an , John , an d Celes ta Albonett i . 1982. "Race , Class an d the Percept ion of Crimi-nal In jus t ice in America ." Am eric an Journal of Sociology 88:329-55.H ag an , John, I lene Nagel , a nd Celesta Albonet ti . 1980. "The Different ia l Sentencingof W hite Col lar Offenders in Ten F edera l Dis t r ic t Co urts ." American SociologicalR e v i e w 45:802-20.H ag an , John , John Simp son, and A. R. Gi ll is . 1979. "The Sexual Stra ti f ica t ion ofSocial Control." Bri t i sh Journal of Soc io logy 30:25-38.Ha rr is , Antho ny. 1977. "Sex an d Theories of Deviance: Tow ard a Funct ional Theoryof Dev iant Type-Scripts ." Amer ican Soc io logical Rev i ew 42:3-15.Hindelang , M ichael , Trav is Hirschi , a nd Joseph Weis . 1979. "Corre la tes of Del in-quency: T h e Il lusion of Discrepancy betw een Self-Report an d Official M easure s."Amer ican Soc io logical Rev iew 44:995-1014.. 1981. Measuring Del inquency. Beverly Hills , Calif . : Sage.Hirschi , Travis . 1969. Causes of Delinquency. Berkeley and Los Angeles: Universityof California Press.. 1972. "Socia l Class an d Crim e." In I s sue s i n Socia l Ine qua l i t y , edited by D.The i lba r and S . Fe ld man . Bos ton : L i tt l e , Brow n.Hirschi , Travis , Michael Hin delan g, an d Joseph W eis. 1982. "Reply to 'On the Use ofSelf-Report D ata to Determine the Class Dis t r ibut ion of Criminal and Del inquentBehavior . ' " Amer ican Soc iological R ev ie w 47:433-35.Jensen, G . F. , M . L . Erickson, an d J . P . Gibbs . 1978. "Perceived Risk of Punishm entand Self-reported Del inquency." Social Forces 57:57-78.Ka nter , Rosabeth . 1974. " Int imate Repression." Sociological Quarterly 15 (2 ): 320-24.Kleck, Gar y. 1982. "On the Use of Self-Report Da ta to Determine the Class Dis t r ibu-t ion of Crim inal and Del inquent Beh avior ." Ame rican Socio logical Rev iew 47:427-33.Linden, R ick, and Cath y Fi l lmore . 1980. "A C omp ara t ive Stu dy of Del inquencyInvo lvement . " In C r i m e in Canad ian Soc ie t y , edi ted by R . Si lverman an d J .Teevan . Toron to : Bu t te rwor ths .Marx, Karl . 1912. Capital . Vol. 1 . Chicago: Ke rr .Matza , David, and Gresham Sykes . 1961. "Juveni le Del inquency and SubterraneanValues." Amer ican Soc iological R ev ie w 26:7 12-1 7.Quinney, Richard. 1977. Class , S ta te and Crime . N e w Y o r k : L o n gm a n .Robinson, Robert , an d Jonatha n Kel ly . 1979. "Class as Conceived by Marx an dDahrendorf ." Ame rican Soc io logical Rev i ew 44:38-58.Ross i, P . , E . Wai te , C . Base , and R . B erk . 1974 . "The Se r iousness of Cr ime : Norm a-t ive Structure an d Individual Differences." Amer ican Soc io logical Re v iew 39:224-37.Simo n, Ri ta . 1975. W o m e n a nd C r i m e. Ne w York : Lex ing ton .Simpson, John , an d John H ag an . 1983. "Evaluat ion of a Vandal ism Prevention Proj-ect." Report to the Research Division. Ottawa: Ministry of the Solicitor General ofC a n a d a .Sorok in , P i t ir im, an d Wal te r Lun den . 1959 . Power and 1Morality. Boston: Sargent .


29/29

Am erican Journal of SociologySpitzer , Steven. 1975. "To wa rd a Marx ian Th eory of Deviance ." Social Problems22:638-5 1.Steffensmier, D. 1980. "Sex Differences in Patterns of Adult Crimes, 1965-77: AReview an d Assessment ." Social Forces 58: 1080-1 108.Tay lo r , I an , Pau l Wal ton , an d Jock You ng . 1973 . Th e Ne w Cr im ino logy. London:Routledge & Kegan Paul .Tho rnb erry , Terrence , an d Ma rgar e t Far nsw orth . 1982. "Social Corre la tes of Crimi-nal Involvemen t : Furth er Evidence on the Rela t ionship be tween Socia l Sta tus andCr imina l B ehav io r ." Amer ican Soc io logical Rev iew 47505-17 .Ti t t le , Charles , W ayne Villemez, and Douglas S m ith . 1978. "The My th of Social Classand Criminal i ty : An Empir ica l Assessment of the Empir ica l Evidence ." Amer icanSociological Review 43:643-56.. 1982 . "One S tep Forwa rd , Tw o S teps Back : More on the C lass ICrimina l ityControversy." Amer ican Soc io logical Rev iew 47:435-37.Veblen, Thorstein. 1934. Th e Theory o f the Le isure Class . N e w Y o r k : M e n to r .Wheeler , Stan ton. 1976. "Trends an d Pro blems in the Sociologica l Study of C rime."Social Problems 23:525-33.Wright , Erik Olin . 1978. "Race , Class and Income Inequal i ty ." Amer ican Journal ofSoc io logy 83:1368-97.. 1980. "Varieties of Marxist Conceptions of Class Structure." Politics andSoc ie t y 9:299-322. \Wright , Erik Olin , and Luca Perrone . 1977. "Marxis t Class Categories and IncomeInequal i ty ." Amer ican Soc io logical R ev ie w 42:32-55.Wrigh t , Eri k Olin , Cynthia Cos te llo , Da vid Hac hen , an d Joey Sp rague. 1982. "TheAmerican Class Structure ." Amer ican Soc io logical Rev iew 47:709-26.

Gender and Delinquenvency

Documents

Transcript of Gender and Delinquenvency