Gender Differences in the Correlates of Adolescents' Cannabis Use

Adolescents' gender-specific cannabis use rates and their correlates were examined. Data were obtained via a cross-sectional survey conducted in 2004 in British Columbia, Canada, funded by the Canadian Institutes of Health Research. School districts were invited to participate, and schools within consenting districts were recruited. In total, 8,225 students (50% male)from Grades 7 to 12 participated. About 73% were “White” and 47% had used cannabis in their lifetime. Cannabis users were grouped according to their frequency of use: “never users” “frequent users” or “heavy users” Male heavy cannabis users (14.3% of boys) were more likely to be in Grade 9 or higher; be Aboriginal; report poorer economic status; never feel like an outsider; frequently use alcohol and tobacco; and have lower satisfaction with family, friends, and school compared with boys that never used. Female heavy users (8.7% of girls) were more likely to be in a higher grade; report poorer economic status, mental health, and academic performance; frequently use alcohol and tobacco; and have lower satisfaction with their school compared with female never users. Three important gender differences in the multivariate analysis of the correlates of cannabis use were noted: school grade (for boys only), Aboriginal status (for boys only), and mental health (for girls only). Despite the limitations of relying on self-reports, a subset of youth appears to be at risk for excessive cannabis use that may impair life opportunities and health. The gender differences may be important in the design and implementation of prevention or treatment programs for adolescents.

Cannabis use is on the rise in Canada; recent data from the Canadian Community Health Survey (CCHS) indicate that cannabis use by Canadians aged 15 years and older almost doubled in just over a decade (Tjepkema, 2004). In 2002, 12.2% of Canadians reported using cannabis in the previous year, compared to 6.5% in 1989. There is regional variation in cannabis use rates, with the highest prevalence of 16% found in British Columbia (BC), Canada's westernmost province. Cannabis use is most prevalent among youth: 29% of 15-to 17-year-olds and 38% of 18-to 19-year-olds, in the CCHS, reported having tried it. In the 1998/99 Canadian National Longitudinal Study of Children and Youth, 9% of 13-year-olds, 25% of 14-year-olds, and 38% of 15-year-olds reported that they had used cannabis in the preceding year (Hotton and Haans, 2003). There is emerging evidence that the average age of initiation of cannabis use has declined in recent years, although the quality of these data is limited (Hall, Johnston, and Donnelly, 1999).
Although most youth who use cannabis do so only occasionally, there is a subset of youth that smoke it very frequently. According to the Ontario Student Drug Use Survey (OSDUS), between 1999 and 2003, there was a significant increase in the percentage of students in Grades 7 through 12 that reported daily cannabis smoking. In 2003, 4.2% of OSDUS participants reported daily cannabis use compared with 2.5% of participants in 1999 (Adlaf and Paglia, 2003). Daily use appears to be more common among boys rather than girls (6.2% of boys reported daily use compared to 2.2% of girls; Adlaf and Paglia). Higher rates of frequent use have been documented in BC, and they appear to be increasing. The McCreary Centre Adolescent Health Surveys of BC high-school students, which have been conducted about every 6 years, found that the percentages of 10-to 15+-year-old boys that smoked cannabis 20 or more times each month were 9% in 1992, 13% in 1998, and 18% in 2003. The percentages of female cannabis users who smoked 20 or more times each month were 4% in 1992, 6% in 1998, and 8% in 2003(McCreary Centre Society, 1998. Population-based surveys of adolescents in Belgium have also found a gender difference in the prevalence rate and increasing rates over time (Kohn, Dramaix, Favresse, Kittel, and Piette, 2005;Kohn, Kittel, and Piette, 2004).
Several researchers have examined the factors associated with adolescents' cannabis use. Coffey, Lynskey, Wolfe, and Patton (2000) followed an Australian cohort of adolescents to obtain repeated measures of cannabis use. They found that correlates of cannabis use among students of about 15 years of age included having divorced or separated parents, peers that used cannabis, concomitant tobacco use, relatively heavy alcohol use, and antisocial behavior including property damage, interpersonal aggression, and stealing. Gender, which had a bivariate relationship with cannabis use (odds ratio = 1.4; 95% CI: 1.2-1.8), was eliminated from the multivariate model when these correlates were included. In a German cohort, von Sydow, Lieb, Pfister, Hofler, and Wittchen (2002) found that the significant predictors of "higher frequency" use included social-contextual factors such as parental death, deprived economic status, and the use of other illicit drugs. In a survey of Belgian adolescents, notable predictors of monthly use were being male; less educated; of non-Belgian nationality; tobacco, alcohol, or other illicit drug use; and moderate family integration (measured with items that assessed ease of talking with parents, whether the youths engaged in activities with their parents and in activities that had parental approval; Kohn, Kittel, and Piette, 2004). Predictors of ever use were being older; tobacco, alcohol, or illicit drug use; and stronger peer integration assessed as having close friends, being able to make new friends, and spending time with friends. In the UK, Miller and Plant examined the characteristics of 201 15 to 16-year-old "frequent users" of cannabis (Miller and Plant, 2002). Using cluster analysis of the youths' responses to questions about their demographics, substance use, families (e.g., parents' knowledge of the youth's whereabouts, rules, warmth, and support), friends (e.g., number of good friends, warmth, and mental support), leisure activities, and psychological status (e.g., self-esteem, aggression, and delinquency), three clusters emerged: a small group of mostly boys that was characterized by antisocial behaviour; a group that appeared to be "unhappy" (they had lower self-esteem, depression, and poorer parental and peer support); and a group that was characterized as "ordinary" (i.e., had good relationships with their family and friends, were obedient to society's rules, and displayed little antisocial behavior). Other observed associations with cannabis use include being Aboriginal (Novins and Mitchell, 1998), being peer integrated (Grunbaum, Tortolero, Weller, and Gingiss, 2000), truancy (Kohn, Dramaix, Favresse, Kittel, and Piette, 2005), lower academic performance (Resnicow, Smith, Harrison, and Drucker, 1999;Windle and Wiesner, 2004), poor physical health (Tims et al., 2002), and poor mental health (Patton et al., 2002;Rey, Martin, and Krabman, 2004), including depression (Degenhardt, Hall, and Lynskey, 2003;Patton et al.).
A gender difference in cannabis use has been identified although the nature of the difference has not been well explored. Previous studies that have noted gender differences have either done so within a specific population (Novins and Mitchell, 1998), without examining many correlates (Pape, Hammer, and Vaglum, 1994;Rodham, Hawton, Evans, and Weatherall, 2005), without stratifying their analysis by gender (Butters, 2005;Hofler et al., 1999;Kohn, Kittel, and Piette, 2004;Resnicow, Smith, Harrison, and Drucker, 1999;Swift, Hall, and Teesson, 2001), and without specifically focusing their analysis on cannabis use (Challier, Chau, Predine, Choquet, and Legras, 2000;Poulin, Hand, Boudreau, and Santor, 2005). The most consistent finding has been that boys are more likely to be "heavy users" than are girls (Kohn, Kittel, and Piette; Novins and Mitchell; Resnicow, Smith, Harrison, and Drucker). The factors associated with boys' and girls' use are not well described, and it is not known whether the risk factors differ by gender. This study aimed (a) to determine the gender-specific cannabis use rates of a large sample of adolescents from the province of BC, Canada, and (b) to determine factors associated with cannabis use and how those factors differ by frequency of use and gender. Based on the literature and what was possible with the available data set, we explored six factors of interest: sociodemographics, health status (physical and mental), life satisfaction, peer integration, academic performance, and other substance use.

Methods
The data analyzed in this study were obtained via a cross-sectional survey focused on tobacco use, the British Columbia Youth Survey on Smoking and Health 2 (BCYSOSH2), conducted between March and June 2004. As seen from the title, the survey questionnaire was constructed primarily to investigate youths' tobacco use and emerging tobacco dependence. In the past, youths advised our study team to ask about their cannabis use. The tobacco field has demonstrated that factors such as socioeconomic status, ethnicity, gender, and school performance are predictors of tobacco use, hence the inclusion of relevant indicators in the questionnaire.
To collect data from a large and diverse number of adolescent tobacco smokers, the majority of data was collected in regional school districts located outside the Greater Vancouver area (the largest city in the province). Of the 60 school districts in the province, 19 were contacted. Of the 19 school districts, 14 gave their schools permission to participate at their own discretion. Of the 86 schools in the 14 consenting school districts, 49 (57%) agreed to participate: 42 secondary schools, 5 alternative schools (designed for students not able to be successful in regular school environments), and 2 middle schools (with Grade 8 students).
The selection of students for inclusion varied across the 49 schools. The entire student body or all students in a particular grade were recruited in 22 schools, and the remaining 27 schools selectively recruited students. The nonrandom selection of students was typically carried out by attempting to include classes taken by most students (e.g., all students in the Grade 9 course, Career and Personal Planning). Two modes of questionnaire administration were offered to the schools, paper-based and Web-based formats. A passive consent procedure was employed; parents were informed about the study through letters carried home by the students. The recruitment and consent procedures and the questionnaire were approved by the relevant university and school district ethics review boards. To ensure that the participants benefited in some small way for their contribution to this research, we employed high school students to prepare school-level reports, which were posted on a website, so that the youths could see how their immediate peer groups compared with others in the Province.
Trained research personnel were present in every classroom to describe the purpose of the survey, to answer questions, and to take attendance. All research personnel were relatively young women with graduate level preparation in the social or health sciences, who had previous experience working with youth. Students were informed about the confidentiality of the survey and their rights as research participants, including their rights not to participate or to refrain from answering particular questions. The average response rate within each school was 84% with student absenteeism accounting for most of the non-response (the average refusal rate was less than 1%). About 3% of the surveys had to be destroyed mostly due to implausible responses. In total, 8,225 students completed the survey (6,544 paper-based and 1,681 web-based). Table 1 provides the measures used in the study. The responses to two questions on cannabis use were used to create the dependent variable: "number of times used cannabis in entire life" (categorized as 0, 1-2, 3-9, 10-19, 20-39, 40-99, and 100+ times), and "number of times used cannabis in past 30 days" (0, 1-2, 3-9, 10-19, 20-39, 40+ times). We initially classified the participants into five groups: those who had never used cannabis ("never users"), those who had used cannabis but not in the last 30 days ("experimenters"), those who had used cannabis 1-2 times in the past 30 days ("occasional users"), those who had used cannabis 3-9 times in the past 30 days ("frequent users"), and those who had used cannabis 10+ times in the past 30 days ("heavy users"). The "experimenters" were found to be extremely heterogeneous in their amount of lifetime use and were omitted from these analyses. The "occasional" and "frequent" users were found to have similar associations and were therefore combined. Thus, the categories reported here are "never users," "frequent users" (used cannabis 1-9 times in past 30 days), and "heavy users" (10+ times in past 30 days).

Explanatory Variables
Sociodemographic factors included gender, grade, ethnicity, and economic status. Because age and grade were highly correlated, only grade was entered into the model. Both had  Asian (18.5%) youth comprised the two largest ethnic groups within the "other" category. Economic status was determined by asking the participants how they compared their family's financial situation with other families in BC ("much better," "somewhat better," "about the same," "somewhat worse," or "much worse").
Health status was measured with two items and a scale. Each student rated his or her physical health and mental health on a 5-point ordinal scale from "excellent" to "poor." Twelve of the 20 items of the Center for Epidemiological Studies Depression (CES-D) scale (Radloff, 1977) were used to assess depressive symptoms. Due to concerns about questionnaire length, only the 12 highest loading items from factor analysis were included in the survey. To develop cut points comparable to the full CES-D we created a possible total score of 60 by dividing the sum of the items by 12 and multiplying by 20. We categorized the severity of depressive symptoms by considering the common adult measure of ≥16 and the adolescent cut-off score of ≥24 (Roberts, Lewinsohn, and Seeley, 1991). We classified those with scores <16 as having minimal to no depressive symptoms, between 16 and 23 as having mild depressive symptoms, and ≥24 as having moderate to severe depressive symptoms. The CES-D has been shown to have good internal consistency and test-retest reliability with adolescent students, with results being comparable to those found in adult samples (Roberts, Andrews, Lewinsohn, and Hops, 1990).
Life satisfaction was measured with four facets of the Multidimensional Students' Life Satisfaction Scale (MSLSS): satisfaction with family, friends, school, and self (Huebner and Gilman, 2002). Previous studies using the MSLSS have reported high reliability and provided support for the dimensionality of the scale (Huebner and Gilman).
To assess academic performance, students were given a 7-point scale ("far above average," "above average," "slightly above average," "average," "slightly below average," "below average," or "far below average") to rate their school work, compared with other students in their school.
Peer integration was determined by asking how often the participant felt like an outsider at school ("all the time," "most of the time," "some of the time," "rarely," or "never").
Other substance use was determined from two questions: one about lifetime cigarette use ("none," "one or a few puffs," "1-5 cigarettes," "6-15 cigarettes," "16-99 cigarettes," "more than 100 cigarettes") and the other on alcohol consumption in past year ("none or one sip," "only at special events," "once a month or less," "2 or 3 times a month," "once a week," or "more than once a week").

Data Analysis
Those with missing gender data were omitted from the analysis (n = 62). Differences between gender, sociodemographic, and cannabis use frequencies, were tested using chisquare or Mann-Whitney tests, where appropriate. Frequencies of the type of cannabis user ("never," "frequent," or "heavy") stratified by the explanatory variables were produced by cross-tabulation for both genders; odds ratios were estimated with univariate multinomial logistic regression analysis. Multivariate multinomial logistic regression analysis was then performed to determine the associations between the levels of cannabis use with all significant explanatory variables from the crude analyses. Listwise deletion of participants with missing data led to a large percentage of students being deleted; therefore, multiple imputation was used to impute data. Those participants with missing data for more than 5 variables (n = 955) were omitted from the imputation procedure. The others had their missing information imputed using Schafer's NORM program (Schafer, 1999). Only the multivariate analysis was performed with the imputed data included. A p < 0.05 was considered statistically significant. SPSS version 12.0 (SPSS, Chicago, IL) was used for all analyses. Table 2 displays the sociodemographic characteristics of the participants, the amount of cannabis they used, and the age of initiation by gender. Our sample consists of equal numbers of boys and girls and relatively equal grade distributions from Grades 8 through 11, with a slight decline in numbers in Grade 12. The participants were predominantly white, which was expected given the sampling of the geographic areas and is consistent with census data. Many students did not report their family's economic status as being any worse than those in the rest of the Province; the boys more frequently reported that their economic status was better than did the girls. About one-half of the sample (47.6%) indicated that they had tried cannabis in their lifetime and 28.9% reported that they used cannabis in the past 30 days. About the same percentage of boys and girls had used cannabis, but the girls tended to use it less frequently. Seventy-eight percent of the participants indicated that they first used cannabis between the ages of 12-15 years, which corresponds to Grades 7-10. The boys tended to initiate their use earlier than did the girls.

Findings
Tables 3 and 4 display the odds ratios obtained from the univariate multinomial logistic regression analyses for boys and girls, respectively. Stratifying our analysis by recruitment method (non-random vs. random) and data collection method (paper based vs. Web based), we found similar associations between cannabis use and all explanatory variables.
Male frequent cannabis users tended to be in the higher grades, be Aboriginal, have poorer mental health, have moderate to severe depressive symptoms, report poorer academic performance, drink alcohol and smoke tobacco more frequently, and have lower satisfaction with their family, friends, and school compared with the boys who never used cannabis. In addition to these characteristics, male heavy users were more likely to have self-reported poorer economic status, poorer physical health, and lower satisfaction with themselves compared with the boys who never used cannabis. The "heavy users" were less likely to report rarely or sometimes feeling like an outsider at school compared with the boys who never used cannabis.
Female frequent cannabis users tended to be in the higher grades, be Aboriginal, have poorer economic status, be in poorer physical and mental health, have mild to severe depressive symptoms, report poorer academic performance, drink alcohol and smoke tobacco more frequently, and have lower satisfaction with their self, family, friends, and school compared with the girls who never used cannabis. Other ethnicity had a protective effect on cannabis use compared with self-identified "white" students. Female heavy users had the same associations as frequent female users, except that they were less likely to "rarely feel" like an outsider, compared with girls that never used cannabis.
The multivariate analysis is displayed in Table 5. After adjusting for all the explanatory variables, male frequent users were more likely to be in the higher grades, be Aboriginal, have poorer academic performance, drink alcohol and smoke tobacco more frequently, have higher self-satisfaction, and lower satisfaction with their family compared with the boys who never used cannabis. Male heavy users were more likely to be in the higher grades, be Aboriginal, report poorer economic status, never feel like an outsider at school, drink alcohol and smoke tobacco more frequently, and have lower satisfaction with their family, friends, and school compared with the boys who never used cannabis.
Female frequent users were more likely to report poorer economic status, poorer mental health, more frequent alcohol and tobacco consumption, being more satisfied with themselves, and being less satisfied with their family compared with the girls that never used cannabis. Girls who were categorized in the "other" ethnic category were less likely to be frequent cannabis users than were self-identified "white" girls. Female heavy users were more likely to report being in a higher grade, poorer economic status, poorer mental health, poorer academic performance, more frequent alcohol and tobacco consumption, and lower satisfaction with their school than did girls who never used cannabis.

Conclusions
The prevalence of cannabis use found in this survey was much higher than that reported by the BC Adolescent Health Survey III (47.6% vs. 37.0%), which was carried out in 2003 by the McCreary Centre Society (2004), a nonprofit organization concerned with adolescent health. This was expected to some extent because we purposely sampled regions with higher than average rates of adolescent tobacco use. Nonetheless, the rate of use is higher than that reported in most, if not all, other countries (Korf, 2001;Vega et al., 2002). And, the rate of use in the previous month surpassed the rate of tobacco use over the same time period in this population (44.7% vs. 17.5%; Richardson et al., 2007). Of particular concern is the high number of "heavy users" (i.e., those who reported smoking cannabis 10+ times in the past month), particularly among boys (14.3%); a rate that has increased over the past 10 years (Tonkin, 2005). The multivariate analysis revealed some striking gender differences in the correlates of both frequent and heavy cannabis use. In particular, grade is predictive of boys' frequent cannabis use, but not of girls'. This might suggest that girls are less influenced by the cannabis use of their peers or the social milieu established in school. Second, reporting relatively poor mental health placed girls at risk for frequent and heavy cannabis use, a factor not observed in the boys' model. Similarly, being Aboriginal placed boys at greater risk of frequent and heavy use of cannabis relative to White boys, a risk factor not observed among the girls. Aboriginal boys appear to be at "double risk" due to their Aboriginality and gender. Many researchers have observed that Aboriginal people in Canada live in circumstances that place them at risk for substance misuse. These conditions have arisen from the adverse sequelae of colonization, forced assimilation through residential schools where many were exposed to emotional, physical, and sexual abuse, and poverty. There is little work reported concerning ethnic and gender interactions in cannabis use, however a similar pattern has been noted among the Aboriginal communities of northern Australia (Clough et al., 2004).
A substance abuse 1 problem among Aboriginals has been well documented in Canada (Health Canada, 1998Statistics Canada, 1993). In 1997, the BC First Nations' Regional Health Survey (BCFNRH) reported that drug and alcohol issues were the primary health concerns among Aboriginal people in BC (First Nations Chiefs' Health Committee, 2000). The majority of participants of the BCFNRH reported that there had been no progress made in efforts to reduce the problem in the previous two years. From their Adolescent Health Survey III data, the McCreary Centre described the health status of Aboriginal youth (McCreary Centre Society, 2005). They reported that 53% of BC Aboriginal youth had ever used cannabis (compared with 36% of non-Aboriginal students). Among those that had tried cannabis, 36% used it 1-9 times in the past month and 25% used it 10+ times in the past month. Among the Aboriginal students that we surveyed, 40.8% had ever tried cannabis with 32.2% and 29.3% having tried it 1-9 times or 10+ times in the past month, respectively.
Self-reported economic status was found to be associated with cannabis use. Those who believed that their economic status was poorer than other families in BC were more likely to be frequent or "heavy users" (although not significant in frequent male users) compared with those who believed that their economic status was better than other families. The association between social-economic status (SES) and cannabis use has changed over time. Older prospective studies found that adolescents from families of higher SES were more at risk to use cannabis (Baumrind, 1985;Kandel, Kessler, and Margulies, 1978;Kaplan, Martin, and Robbins, 1985). The results of recent studies have been conflicting with some researchers reporting no association and others an inverse association between SES and cannabis use (Challier, Chau, Predine, Choquet, and Legras, 2000;Geckova, van Dijk, Groothoff, and Post, 2002;Kandel, Chen, Warner, Kessler, and Grant, 1997;Miller and Miller, 1997;von Sydow et al., 2002). This trend may be related to the decreasing cost of cannabis or the increasing ease of access.
Physical health was not associated with cannabis use. There was no discernable association between depressive symptoms and cannabis use. There have been conflicting reports about the association between depression and cannabis use (de Irala, Ruiz-Canela, and Martinez-Gonzalez, 2005). We found an association in the bivariate analyses, which may have been confounded with the more global assessment of mental health and gender.
Poor academic performance was associated with frequent use by boys and heavy use by girls, although the trend was inconsistent for the girls. Several studies have found that cannabis use increases the risk of poor academic performance and early school leaving (Fergusson, Horwood, and Beautrais, 2003;Lynskey and Hall, 2000;Miller and Miller, 1997). Fergusson, Horwood, and Beautrais suggested that cannabis use influences academic performance through social networks that produce attitudes or values that promote school failure or dropout. Our finding that those adolescents who reported feeling like outsiders at their schools were less likely to be cannabis users supports that theory (i.e., cannabis users develop "normative" social networks that share common values).
Other substance use proved to have the strongest association with cannabis use. The greater an adolescent's reported consumption of alcohol or tobacco the more likely he or she was to report being a "frequent" or "heavy user" of cannabis. This confirms the report from the McCreary Centre Society (2004) adolescent health survey that very few students use cannabis without using other substances.
In terms of life satisfaction, "frequent users" were more likely to be more satisfied with themselves and less satisfied with their families than were those who never used cannabis, regardless of gender. This may come about because cannabis use itself can lead to misperceptions about social cues, to protection from the social anxieties experienced by most youth, and to relatively strong group attachments with those engaged in similar behavior. Male "heavy users" are more likely to be less satisfied with their family and school than are boys who never use cannabis. These results support the findings of Butters who found that poorer family and school relations were associated with cannabis use (Butters, 2005).
There are some potential limitations to this study. First, the study relied on self-reported data; therefore, the responses may be biased. However, the survey was anonymous and in most cases, a large percentage of the school took part, so students should not have feared being identified by their responses. Second, we sampled from a population of adolescents with higher than average tobacco-smoking rates; therefore, comparisons can only be made with similar populations. Third, the small percentage of students that missed class the day the survey was administered was mostly truant. It has been reported that those who are truant have a higher risk of substance use (Hallfors, Cho, Brodish, Flewelling, and Khatapoush, 2006); therefore, the prevalence reports may be underestimated. However, our sample size and participation rate were large enough so that the effects would be minimal. Fourth, economic status and academic performance were assessed by the students' perceptions. In light of the limitations associated with self-reported academic performance, these findings should be viewed with caution. The association that we reported, however, may be underestimated because the validity of self-reported grades interacts with actual levels of school performance such that the poorer performers are likely to overestimate their performance (Kuncel, Crede, and Thomas, 2005). Less is known about the validity of adolescents' self-reported economic status. Frojd, Marttunen, Pelkonen, der Pahlen, and Kaltiala-Heino (2006) suggested that, although adolescents' perceptions of their family's economic status is likely associated with the actual status, they also may be indicative of the "psychological meaning" attached to the family's situation (i.e., financial problems) and could be considered a potential measure of risk independent of the actual status. Finally, we categorized the respondents into three categories: never, "frequent," and "heavy users." Although we combined various levels of use that had similar strengths of association with the hypothesized correlates, it may be that different levels of use or cut points yield different results.
The findings show important gender differences in predictors of "frequent" and "heavy cannabis use." These differences may be important in the design and implementation of prevention or treatment programs for adolescents. For example, programs focussed on girls may need to address specifically mental health issues. Current adolescent substance abuse treatment programs have an average sustained abstinence rate of 32% at 12 months after treatment (Williams and Chang, 2000). Although there is evidence that treatment is better than no treatment, improvements in treatment can still be made. The effectiveness of prevention programs has been inconclusive. Two separate reviews have provided differing conclusions about the effect of school-based substance use prevention (Wiehe, Garrison, Christakis, Ebel, and Rivara, 2005;Skara and Sussman, 2003), which might be best suited for boys. Programs should be designed for each specific target population and not be seen as an all-encompassing tool.
In conclusion, the rates of cannabis use reported here, particularly excessive use, are troubling. Increasingly, youth appear to be turning to cannabis use while tobacco use is diminishing. Important gender differences in the prevalence of cannabis use, and its correlates, were noted. For example, mental health status was associated with girls' cannabis use and not boys'. The findings suggest that the uptake and use of cannabis may be very different for girls and boys. In addition, the correlates of "frequent" and "heavy use" differ, which suggests that excessive use may result for reasons other than ease of access and behavioral normalization. Further research is required to uncover the nature of these differences and to explicate their implications for intervention. Cannabis use has often been disregarded as a benign, youth-specific form of recreation. This may be the case for many if not most users; however, a subset of youth appears to be at risk for excessive use that may impair their life opportunities and health.
No conflicts of interest.