FORUM on Corrections Research

Correlates of delinquency: A look at gender differences

Contemporary research on delinquency is challenging fundamental assumptions regarding female delinquency made by early theorists and researchers. As a result, correctional theory and research appears to be at a crossroad in terms of gender issues.

One important issue is the identification of risk factors for female youth. The apparent social bias of early female-delinquency theories suggests that exclusive reliance on personal and familial problems in the assessment of female risk may be inadequate. Yet, the relevance of male-based risk factors has not been filly evaluated.

Our study attempted to fill this gap through a systematic review of research that has examined the same risk factors for male and female youths separately. It should be noted here that our research and its findings focused on youth criminality (delinquency) rather than on adult criminality.

Female delinquency has historically been perceived as relatively rare and less serious than male delinquency.⁽²⁾ Early research using police and court records showed that for each female delinquent, there were three to seven male delinquents.⁽³⁾ In terms of offence type, female youths were perceived as committing relatively minor offences, such as running away, truancy and sexual acting out, while male youths were perceived as committing a much wider and more serious range of offences.⁽⁴⁾

Early female-delinquency theories were also dramatically different from those developed for male youths. Personal maladjustment was viewed as a fundamental cause of problem behaviour in female youths. Psychological problems, inadequate performance of the proper sex role, and a problematic home life were some of the proposed explanations for female deviance.⁽⁵⁾ In contrast, early male-delinquency theories focused on more external risk factors, such as peer group, lower social class, and lack of educational or occupational resources.⁽⁶⁾

Within the past 20 years, research has challenged some of these traditional views. Contemporary self-report studies, for example, have suggested a more accurate gender ratio of no more than three male delinquents to each female delinquent.⁽⁷⁾ Recent studies have also found that female youth are involved in a broad range of criminal behaviours, not just minor offences. One exception to the similarity of offences is the use of physical aggression; here, female involvement remains lower than that of male youths.⁽⁸⁾

To explain and predict delinquency, contemporary research has expanded its scope to consider personal factors (such as behaviour, personality, and cognition), interpersonal factors (such as family and peers), and structural factors (such as school and church). Although the tendency to omit female youths continues, there has been an emerging interest in female involvement in delinquency. Increasingly, studies on risk are including male and female youths in their research samples. Until now, these studies were never examined as a group. Our study The primary goal of our research was to review this portion of the delinquency literature. Specifically, we looked at published and unpublished studies conducted over the past 30 years. This review gave us an opportunity to comprehensively examine female youths on a wide range of risk factors. It also allowed us to examine risk factors not typically investigated among male samples. We sought answers to two basic questions:

• What are the important risk factors for each gender? For example, does associating with criminal peers place male youths at risk for delinquency, and does this factor also place female youths at risk?
• Are specific risk factors more important for a particular gender? For example, are family problems more strongly related to female delinquency than to male delinquency, and are school difficulties more strongly associated with male delinquency than with female delinquency?

We used a technique called "meta-analysis" to conduct this review.⁽⁹⁾ One of the main advantages of meta-analysis over the traditional (narrative) form of literature review is that meta-analysis summarizes large bodies of literature and reaches more definitive conclusions. This quantitative approach combines numerical information (actual data) from selected studies and computes an average result for each risk factor for males and females separately. This information then indicates the degree of association between delinquency and a particular risk factor.⁽¹⁰⁾

---

Table 1

Risk Factors

Lower social class

Family structure or parental problems (broken home, marital problems)

Personal distress (anxiety, low self-esteem, apathy)

Minor personility variables (empathy, moral reasoning)

Poor parent-child relations (attachment, supervision)

Educational difficulties (poor grades, dropout)

Temperament or misconduct problems (psychopathy, implusivity, substance use)

Antisocial peers or attitudes

Each study included in this review met three criteria:

• Male and female youths were sampled.
• Each gender was examined on the same risk factor.
• The data for each gender were reported separately.

These criteria ensured that male-female comparisons were based on the same delinquency and risk measures, thus eliminating biases that could result from using different measures for each gender.

Sixty studies met these requirements and provided 464 correlations between delinquency and risk factors. We then grouped the risk factors examined within these correlations into eight general risk categories based on previous reviews and common themes in the delinquency literature.

Table 1 lists the eight risk factors and provides a brief indication of their content. Some factors will be familiar to FORUM readers, as a previous issue on delinquency highlighted some of these risk areas.⁽¹¹⁾

What risk factors are most important for delinquency in each gender?

---

Table 2

Average Correlations for Eight Risk Factors
Risk Factor	Female youths r(n)	Male youths r(n)
Lower social class	0.07 (19)	0.06 (19)
Family structure or parental problems	0.07 (17)	0.09 (17)
Personal distress	0.10 (14)	0.09 (14)
Minor personality variables	0.18 (9)	0.22 (9)
Poor parent-child relations	0.20 (41)	0.22 (41)
Educational difficulties	0.24 (34)	0.23 (34)
Temperament or misconduct problems	0.35 (45)	0.36 (45)
Antisocial pers or attitudes	0.39 (53)	0.40 (53)
r = average correlation; n = number of correlations that contributed to that average

Table 2 provides a clear indication of the combined results of these 60 studies. Correlations of 0.15 or larger would be of substantive interest.

For female youths, the most important risk factors in descending order were antisocial peers or attitudes, temperament or misconduct problems, educational difficulties, poor parent-child relations, and minor personality variables. Personal distress, family structure or parental problems, and lower social class did not appear to be strongly related to delinquency.

A similar pattern emerged for the male youths: that is, the first three risk factors in the table were not strongly related to male delinquency, and the last five factors were important.⁽¹²⁾ Are specific risk factors more important for a particular gender? A comparison of the two columns of correlations in Table 2 indicates there were no differences in the risk factors across gender. Statistical tests supported this seemingly obvious finding. In other words, the general risk factors that were important for male delinquency were also important for female delinquency. Practical importance In practical terms, these data suggest that knowing a youth's socioeconomic status or family structure would provide little information about his or her risk of delinquency.

However, information indicating difficulties in the area of family relations, conduct or peers would provide valuable information about that individual's risk of delinquency. These findings are consistent with social psychological models of criminal conduct that suggest that a variety of factors are associated with delinquency.⁽¹³⁾ Control variables The next step in our research process was to assess whether particular aspects of these 60 studies contained systematic bias. For example, would the numbers in Table 2 change dramatically if we took into account the source of the information - that is, whether delinquency was measured by self-reports from the delinquents or by officials from the justice system? Would it matter if the sample consisted of high-school students or a group of offenders?

We considered 15 aspects related to the studies and their samples.⁽¹⁴⁾ While certain aspects did influence the size of the correlations in Table 2, the overall ranking of factors did not change. That is, despite taking into account various study characteristics, the data on social class, family structure or parent problems and personal distress still provided little information about an individual's risk for delinquency. Parent-child difficulties and school problems remained important. Finally, temperament or misconduct problems and antisocial peers or attitudes remained the factors most significantly related to delinquency, regardless of study characteristics.

To summarize, this meta-analytical review yielded three conclusions with respect to eight general risk factors:

• The general risk factors that were important for female youths were also important for male youths. Further, no risk factor was more important for a particular gender.
• The most important risk factors for both genders, in descending order, were antisocial attitudes and peers, temperament or misconduct problems, educational difficulties, poor parent-child relations, and minor personality variables. In contrast, lower social class, family structure or parental problems, and personal distress were not strongly related to delinquency.
• When various sample and study characteristics were taken into account, the general pattern of results remained the same.

Other risk factors Although this review suggests that the same risk factors are important for men and women, some might argue that gender differences may exist for other factors not captured by the eight areas we examined.

To explore this possibility, we grouped 96 correlations not captured by the eight general risk factors into 10 additional risk areas and assessed them for possible gender differences. Table 3 lists these factors and the number of times each factor was found in the studies examined. Undoubtedly more research has been conducted within each area; however, recall that we selected the studies based on three criteria listed earlier.

Since only a total of 96 correlations pertained to these 10 categories, we treated this set of findings as exploratory. A few highlights will be mentioned.

Two very promising categories of risk were lack of attachment to convention and sexual The general risk factors that were important for female youths were also important for male youths. Further, behaviour. Lack of attachment to convention pertained to an individual's lack of affiliation with prosocial people (such as parents and teachers) or institutions (such as family, school and church). Because these measures assessed attachment to multiple individuals or institutions, they could not be incorporated in any of the earlier categories, such as education or parent-child relations.

The data indicated that lack of attachment to convention was associated with delinquency for each gender (male average correlation = 0.23; female average correlation = 0.22). This was to be expected given that our earlier analysis found that lack of attachment to specific institutions (such as educational difficulties) and people (poor parent-child relations) were important risk factors.

---

Table 3

Other Risk Factors

Victimization (1)

Illegitimate opportunity (2)

Lack of legitimate opportunity (3)

Sexual behaviour (3)

Lack of hobbies or involvement (4)

Accommodation problems (4)

Self -concept issues (6)

Race (7)

Sex-role orientation (8)

Lack of attachment to convention (10)

Although the data for sexual behaviour were obtained from only three studies, there was consensus about the importance of this factor. The lowest correlations for male and female youths were 0.35 and 0.26, respectively. What remains unclear is whether gender differences exist - one study found large gender differences, another found minor differences, and the third found none.

The significance of sexual behaviour as a correlate of delinquency may be somewhat surprising, given current rates of sexual activity among teenagers. However, these three studies were conducted between 1966 and 1971. Sexual behaviour in those days may have served as an indicator of an individual's tolerance of rules or norms. An interesting question is whether the same degree of association would be noted for high-school students in the 1990s.

Two less promising categories were race and sex-role orientation. The issue of race refers solely to Black versus Caucasian samples, as no other racial comparisons were reported. The information from seven studies suggests no link between race and delinquency (male average correlation = -0.02; female average correlation =0.03).

Sex-role orientation (masculinity or femininity) was also found to be unrelated to delinquency (male average correlation = 0.05; female average correlation = 0.07). There may be a minor association of delinquency with sex-role orientation when lack of feminine traits (communal traits, such as sharing and caring) are examined. A narrative literature review of research in this area found similar results and concluded this line of research should be abandoned.15

The jury is still out for the six remaining risk categories. For these factors (victimization, illegitimate opportunity, lack of legitimate opportunity, lack of hobbies or involvement, accommodation problems, and self-concept issues), either too few studies were obtained or various study characteristics made interpretation of the findings difficult. Victimization, for example, was only measured by one study, which asked students between the ages of 12 and 16 whether they personally were a victim of five types of crimes ranging from theft of unguarded possessions to attack and assault.16 The correlations, based on overall victimization, not on victimization specific to violence, yielded no gender difference (male correlation = 0.14; female correlation = 0.16). Given the growing clinical interest in sexual abuse and victimization, particularly for females,17 future gender research should examine this issue empirically.

Another category, accommodation problems (e.g., crowding, high-crime neighbourhood), was assessed by four studies, whose results varied across sample and design, thus making interpretation difficult. The two cross-sectional studies indicated no association between accommodation problems and delinquency in male youths, but a minor association between them for female youths. The longitudinal study reported a significant association for men but not for women. Finally, the offender versus non-offender sample found accommodation problems to be a significant risk factor for both genders, although more important for women.

More questions are raised than answered when variation occurs to.this extent in the data. Clearly, for these six categories, more research must be gathered before any conclusions can be drawn.

In summary, our conclusions about the 10 categories described in this section are limited by the small number of correlations and the sometimes varying results. Future research could examine some of these factors for their use as risk factors and for possible gender differences. Conclusions The results of this literature review are clear. The risk factors that are important for male delinquency are also important for female delinquency. Of the risk factors examined, the most important are antisocial peers or attitudes, temperament or misconduct problems, educational difficulties, poor parent-child relations, and minor personality variables. In contrast, lower social class, family structure or parental problems, and personal distress are not strongly related to delinquency for either gender.

These results support recent social psychological models of criminal conduct that suggest a variety of personal, interpersonal and structural factors are related to delinquent behaviour in males and females.

However, our results seriously challenge the value of early delinquency theories. Most importantly, notions of female delinquency as exclusively symptomatic of personal distress or familial difficulties have been shown to be inadequate. Early male theories, which focused on lower social class as a major route to criminal behaviour, can also be questioned. Others previously challenged this social-location perspective, and a contemporary understanding of male delinquency has moved beyond this limited view.⁽¹⁸⁾

Several challenges remain for researchers and scientist-practitioners. First, these results do not eliminate the possibility that other factors are associated with delinquency in one or both genders. Future research could investigate, for example, the role of some of the less frequently measured factors explored in this study. Weaker personality variables, sexual behaviour, and sexual abuse and victimization are a few areas worthy of further consideration.

Second, these findings should lead to reformulated ideas and directions about theory and research on female delinquency. Those interested in female delinquency should learn from male-delinquency literature that has progressed beyond the early reliance on social class as the major explanation of male deviance. It is time to set aside antiquated ideas and to consider a larger group of factors as causes of delinquency. Our review examined gender differences and similarities in the correlates of delinquency, so the next step is to examine causal factors for individual female and male youths in the context of their experiences. That research would inform us of the need for gender-specific or gender-neutral theories of delinquency.

In conclusion, for some, the results of this review may simply state the obvious. For years, a small portion of the literature on delinquency has examined the same risk factors for male and female youths and independently, yet repeatedly, found the same results across gender. This review of the literature has pulled those results together in a quantitative fashion, and the similarity across gender can no longer be ignored. The factors examined to date suggest a unique set of correlates may not be required for female delinquency.

Future research will inform us about the role gender plays with respect to predictors and theories of criminal conduct. Consistent with this approach is the need to incorporate gender into the research design and to present the male and female data separately. Only then can a clearer understanding of the similarities and differences across gender be obtained.

---

⁽¹⁾Linda Simourd and D.A. Andrews, Department of Psychology, Carleton University, Loeb Building B-542, Ottawa, Ontario K1S 5B6.
⁽²⁾R. Canter, "Family Correlates of Male and Female Delinquency," Criminology: An Interdisciplinary Journal, 20 (1982): 149-167. See also P. Richards, "Quantitative and Qualitative Sex Differences in Middle-class Delinquency," Criminology: An Interdisciplinary Journal, 18 (1981): 453470.
⁽³⁾R. Canter, "Sex Differences in Self-report Delinquency," Criminology: An Interdisciplinary Journal, 20 (1982): 373393.
⁽⁴⁾Canter, Sex Differences in Self-report Delinquency.
⁽⁵⁾P. Giordano, "Girls, Guys and Gangs: The Changing Social Context of Female Delinquency," The Journal of Criminal Law and Criminology, 69 (1978): 122-132.
⁽⁶⁾Giordano, Girls, Guys and Gangs: The Changing Sodal Context of Female Delinquency.)
⁽⁷⁾The precise ratios are 1.2:1 to 2.5:1 as reported in Canter, Sex Differences in Self-report Delinquency.
⁽⁸⁾I. Gomme, "Predictors of Status and Criminal Offences among Male and Female Adolescents in an Ontario Community," Canadian Journal of Criminology, 27(1985): 147-159.
⁽⁹⁾For a complete discussion of meta-analysis see R. Rosenthal, "Meta-analysis: A Review," Psychosomatic Medicine, 53 (1991):
⁽¹⁰⁾Measures of variability are available from Linda Simourd; however, they are not discussed in this paper. Research, 3, 3 (1991).
⁽¹¹⁾See T. Nouwens and F. Porporino, eds., Forum on Corrections Research, 3,3 (1991).
⁽¹²⁾The findings for minor personality variables should be viewed with caution, given the small number of correlations representing this factor and fluctuations noted within this category.)
⁽¹³⁾D.A. Andrews and J. Bonta, The Psychology of Criminal Conduct (Cincinnati: Anderson, in press). See also M. LeBlanc, M. Ouimet and R.E. Tremblay, "An Integrative Control Theory of Delinquent Behaviour: A Validation 1976-1985", Psychiatry, 51 (1988): 146-176.
⁽¹⁴⁾These were 1) sample, 2) design, 3) sample size, 4) period of data collection, 5) gender focus of study, 6) predominant ethnicity, 7) source of delinquency information, 8) source or risk information, 9) retrieval source, 10) gender of author, 11) psychometrics of risk measure, 13) social class, 14) type of delinquency, and 15) nature of delinquency.
⁽¹⁵⁾N. Naffin, "The Masculinity-Femininity Hypothesis: A Consideration of Gender-based Personality Theories of Female Crime," British Journal of Criminology, 25 (1985): 365-381.
⁽¹⁶⁾R. Mawby, "Sex and Crime: The Results of a Self-report Study," British Journal of Sociology, 31(1980): 525-543.
⁽¹⁷⁾J Darke and H. McLean, "Invisible Women: The Treatment and Research Needs of Female Offenders," The Treatment of Female Offenders, ed. A. Loucks, Symposium conducted at the annual meeting of the Canadian Psychological Association (Ottawa: May 1990). See also A. Finbender and W. Friedrich, "Psychological Functioning and Behaviour of Sexually Abused Girls," Journal of Consulting and Clinical Psychology, 57 (1989): 155-157.
⁽¹⁸⁾This conclusion is consistent with that of a study conducted by C. Tittle and R. Meir, "Specifying the SES/Delinquency Relationship," Criminology: An Interdisciplinary Journal, 28 (1990): 271-296.