Addictive Behaviors 50 (2015) 149–156
Contents lists available at ScienceDirect
Addictive Behaviors
Risk and protective factors for recreational and hard drug use among Malaysian adolescents and young adults Muzafar Mohd Razali a, Wendy Kliewer b,⁎ a b
No. 25, Jalan Segar 7 Taman Segar 35900 Tanjong Malim, Perak Malaysia Virginia Commonwealth University, United States
H I G H L I G H T S • • • • •
Risk and protective factors for substance use occurred in multiple domains. Risk and protective factors for recreational and hard drug use were similar. Risk and protective factors for substance use differed by age group. Risk factors in the individual and peer domains were particularly salient. Religious practices and prosocial school involvement were protective factors.
a r t i c l e
i n f o
Available online 12 June 2015 Keywords: Malaysia Substance use Risk factors Protective factors Adolescents Young adults
a b s t r a c t Introduction: This study investigated risk and protective factors for recreational and hard drug use in Malaysian adolescents and young adults. Methods: Participants (n = 859; M age = 17.24 years, SD = 2.75 years, range = 13–25 years; 59% male) were recruited from secondary schools, technical colleges, a juvenile detention center and a national training center in Malaysia. A version of the Communities That Care survey validated for use in Malaysia (Razali & Kliewer, 2015) was used to assess study constructs. Results: One in 6 adolescents and 1 in 3 young adults reported lifetime recreational and hard drug use, with greater use reported by males across all drug categories. Structural equation modeling was used to determine the strongest risk and protective factors for recreational and hard drug use. The overall pattern of findings was similar for recreational and hard drug use. Shared risk factors for lifetime recreational and hard drug use included early initiation of antisocial behavior, peer antisocial behavior, and peer reinforcement for engaging in antisocial behavior; shared protective factors included religious practices and opportunities for prosocial school involvement. Multiple group analyses comparing adolescents and young adults indicated that patterns of risk and protective factors predicting drug use differed across these age groups. There were fewer significant predictors of either recreational or hard drug use for young adults relative to adolescents. Conclusions: Results suggest that interventions should target multiple microsystems (e.g., peer groups, family systems, school environments) and be tailored to the developmental stage of the individual. © 2015 Elsevier Ltd. All rights reserved.
1. Introduction Drug use in adolescence and young adulthood is a social and health problem worldwide. It is estimated that in 2012, between 162 million and 324 million people, corresponding to between 3.5% and 7.0% of the world's population aged 15–64 had used an illicit drug – mainly a ⁎ Corresponding author at: Department of Psychology, Virginia Commonwealth University, PO Box 842018, Richmond VA 23284-2018, United States. Tel.: +1 804 828 8089; fax: +804 828 2237. E-mail addresses: [email protected] (M.M. Razali), [email protected] (W. Kliewer).
http://dx.doi.org/10.1016/j.addbeh.2015.06.022 0306-4603/© 2015 Elsevier Ltd. All rights reserved.
substance belonging to the cannabis, opioid, cocaine or amphetaminetype stimulants group – at least once in the previous year (United Nations, 2014). Malaysia is concerned with the social and economic impact of drug abuse on young people (Ministry of Women, Family and Community Development, & UNICEF Malaysia, 2013). The National AntiDrug Agency (2013) reported out of the total 11,108 new addicts in 2012, 4.14% were ages 13–19 and 76.54% were ages 20–34. The percentage of new addicts who were adolescents and young adults increased from 2008, when half of the new addicts were in that age range. Malaysia has observed a shift from recreational to hard drug use among young people in the last decade (National AntiDrug Agency, 2013).
150
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
The trends in drug use observed in Malaysia – particularly among those aged 18 to 25 – are consistent with data elsewhere. Substance use is extremely high in this developmental period, and consequences of drug use and misuse can be severe. Further, risk and protective factors for substance use in young adulthood are important to identify as they may be distinct from factors that predict adolescent substance use (Stone, Becker, Huber, & Catalano, 2012). Malaysia is interested in addressing the issue of substance abuse among adolescents and young adults, but has not been successful, perhaps because prevention and intervention programs have not been evidence-based (Mohd Muzafar Shah Mohd Razali, 2007). There is a dearth of research in Malaysia on risk and protective factors for substance use or abuse among adolescents and young adults. The purpose of the present study was to address this knowledge gap and extend research on risk and protective factors for substance use to the Malaysian context. Building on work on risk and protective factors conducted both within (Arthur, Hawkins, Pollard, Catalano, & Baglioini, 2002, Arthur et al., 2007) and outside (Kliewer & Murelle, 2007; Wongtongkam, Ward, Day, & Winefield, 2014) of the United States, we extend prior research in several ways. First, we use an assessment tool that underwent an extensive adaptation and validation process for use in the Malaysian context. Second, we focus on both adolescents and young adults (through age 25) in our sample. Third, we assess both recreational (alcohol, marijuana, inhalants) and hard (pills, stimulants, heroin, morphine) drug use. 2. Method This cross-sectional survey included participants from secondary schools (n = 498), technical colleges (n = 267), a juvenile detention center (n = 38) and a national training center (n = 56). Because of the focus on adolescents and young adults, participants over the age of 25 were excluded from the sample. Participants ranged in age from 13 to 25 (M = 17.24 years, SD = 2.75 years), and 59% were male. Most of the sample were Malay (92%), but also included Indian (3.8%), Chinese (2.7), and other (1.5%) ethnicities. Prior to conducting the study, special permission was obtained from the educational and correctional authorities respectively. This permission is required by the Malaysian Ministry of Education before conducting research in schools and colleges under their authority. Permission to conduct the study in religious schools was obtained from the State Islamic Affairs Department. Additionally, permission was obtained from the Department of Welfare for the juvenile center participants. Surveys in the secondary schools were administered by a group of Sultan Idris Education University student counselors undergoing the final semester of their internship training. All counselors underwent training in the responsible conduct of research prior to administering the surveys. The ethics of participant confidentiality were emphasized in the training. The counselors were familiar to the students and a high level of trust existed between the counselors and the study participants. Questionnaires for the participants in the juvenile center were administered by their counselors. All participants were guaranteed confidentiality of their responses and also their right to withdraw from participation at any time. The schools and juvenile center also were given the right to withdraw from this study if they wished to do so. 2.1. Assessment of risk and protective factors for drug use Risk and protective factors were assessed with a version of the Communities That Care (CTC; Arthur et al., 2002, 2007) survey adapted and validated for use in Malaysia (Razali & Kliewer). The Malaysian version of the CTC included 137 items assessing risk and protection in a total of 27 individual, peer, family, school, and community domains. See Table 1 for a list of the risk and protective factors included in the study.
Table 1 Descriptive information on risk and protective factor scales included in the study. Scale
Alpha Mean SD
Risk factors – individual domain Rebelliousness Sensation seeking Favorable attitudes towards drug use Favorable attitudes towards antisocial behavior Early initiation of antisocial behavior
.72a .87 .94 .92 .85
3.85 6.85 8.58 14.04 18.49
Risk factors – family domain Poor parental monitoring Family conflict Family antisocial behavior Favorable attitudes towards drug use Favorable attitudes towards antisocial behavior
.85 .83 .97 .97 .92
17.39 6.42 4.54 2.24 28.80 11.03 7.93 5.00 9.77 5.56
Risk factors – peer domain Peer antisocial behavior Peer drug use Peer rewards for antisocial behavior
.92 .96 .87
28.07 14.92 6.89
8.83 5.56 4.05
Risk factors – school domain Low school commitment Poor academic performance
.86 .61a
13.39 5.10
5.35 1.91
.83a .72 .86
4.92 8.29 6.36
2.19 3.43 3.57
.94
16.77
9.73
Protective factors – individual domain Religious practices Belief in a moral order
.87a .88
8.20 17.43
2.09 3.69
Protective factors – family domain Attachment to family Opportunities for prosocial family involvement Rewards for prosocial family involvement
.77 .85 .63a
15.70 11.62 7.78
3.76 3.04 1.79
Protective factors – school domain Opportunities for prosocial school involvement Rewards for prosocial school involvement
.75 .81
18.09 14.02
3.91 3.41
Protective factors – community domain Rewards for prosocial community involvement
.65a
6.92
2.18
Risk factors – neighborhood domain Low neighborhood attachment Community disorganization Favorable community attitudes towards tobacco, alcohol, and cannabis use Perceived availability of drugs
1.81 3.52 5.17 8.35 4.67
a
Note. reflects the correlation between two items. For all risk factors, higher values indicate greater level of risk; for all protective factors, higher values indicate greater level of protection.
2.2. Assessment of recreational and hard drug use Past year and lifetime use of all substances were assessed, but for the purposes of this report only lifetime use is reported. Lifetime use was assessed on a 5-point scale where 0 = never used; 1 = used 1–2 times; 2 = used sometimes but not regularly; 3 = regular user previously but not currently; and 4 = current regular user. Recreational drug use included beer, wine and liquor, inhalants, and marijuana. Hard drug use included psychotropic pills, stimulants, heroin, and morphine. Psychotropic pill use in Malaysia is analogous to non-medical use of prescription drugs (NMUPD) in the United States. That is, these pills are medications that are prescribed to treat mental (e.g., depression, anxiety) and physical health (e.g., pain) problems, but are used in a manner not prescribed, or used with a prescription, often for the purpose of getting high. Table 2 presents frequency distributions of recreational and hard drug use for the adolescents and young adults in the sample.
2.3. Data analysis Descriptive statistics on the frequency of lifetime recreational and hard drug use are presented first. In order to evaluate the contribution
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156 Table 2 Frequency of lifetime recreational and hard drug use by age group (percentage endorsing). Adolescents
Young adults
Beer Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
83.2 12.4 2.4 1.5 0.5
64.7 28.9 2.4 0.8 3.2
Wine/Liquor Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
83.8 12.5 2.0 1.0 0.7
64.2 30.9 2.1 1.2 1.6
Inhalants Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
83.7 13.6 1.5 0.7 0.5
64.4 31.2 2.8 0.4 1.2
Marijuana Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.0 13.3 1.9 0.5 0.3
63.2 30.4 2.4 2.4 1.6
Psychotropic pills Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.5 12.5 0.8 1.3 0.8
64.7 28.9 2.4 0.8 3.2
Stimulants Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.8 13.8 0.5 0.5 0.5
65.2 30.0 0.8 1.6 2.4
Heroin Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.8 14.2 0.8 0.2 0
64.9 32.7 0.4 1.2 0.8
Morphine Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.5 13.8 1.3 0.2 0.2
65.9 31.3 0.8 1.2 0.8
of risk factors to drug use, two sets of Structural Equation Model (SEM) were run using Mplus 7.2 (Muthen & Muthen, 2014). One set included recreational drugs as the outcome, and the four recreational drugs were manifest indicators of a latent factor of recreational drug use in these models. The second set included hard drugs as the outcome, and the four hard drugs were manifest indicators of a latent factor of hard drug use in these models. All drug outcomes were transformed in the analyses due to skewness and kurtosis. Within each set of analyses, a model was run within each risk domain (e.g., individual, family, peer, school, and community) followed by a model with protective factors in order to identify significant domain-specific risk and protective factors that were associated with drug use. A final comprehensive model included the significant predictors from all prior models. Age and gender were controlled in all analyses. Once a comprehensive model was specified, multiple group analyses were conducted to determine if the pattern of associations between risk and protective factors and drug use varied by age group (adolescents
151
[n = 600] versus young adults [n = 259]). An unconstrained model where path coefficients were allowed to vary by age group was compared to a constrained model where path coefficients were set to be equal across age groups. Model fit was assessed with the Root Mean Square Error of Approximation (RMSEA; Browne & Cudeck, 1993) and the Comparative Fit Index (CFI; Bentler, 1992), and was deemed adequate if the RMSEA was b.08 and/or the CFI was N.90. Additionally, fits for the unconstrained and constrained models were evaluated by examining the χ2 difference test and differences in the models based on the CFI, RMSEA, and the Bayesian Information Criterion (BIC). Lower BIC scores suggested a better model fit. 3. Results 3.1. Lifetime prevalence of recreational and hard drug use As seen in Table 2, lifetime substance use varied by age, as was expected. Approximately 1 in 6 adolescents and 1 in 3 young adults reported lifetime use of both recreational and hard drugs. However, current, regular use of alcohol or other drugs was very low. In contrast, a quarter of the adolescents and half of the young adults reported trying tobacco (not shown in Table 2); 9.2% of the adolescents and 21.2% of the young adults reported current, regular tobacco use. In order to assess overall gender differences in substance use, a multivariate analysis was conducted. The multivariate effect of gender, controlling for age, was significant, F(8, 807) = 7.93, p b .001, with males reporting higher prevalence of substance use than females on all substances. As reflected in Table 2, there also was an overall effect of age, F(8, 807) = 2.99, p b .01. 3.2. Results of models predicting recreational drug use 3.2.1. Individual risk model The model that included the five risk factors from the individual domain was an adequate fit to the data, (n = 859, X2(23) = 194.92, p b .001, RMSEA = .093, CFI = .954). Sensation seeking (b = .11, p b .01) and early initiation of antisocial behavior (b = .28, p b .001) were the significant risk factors for recreational drug use in the model. 3.2.2. Family risk model The model that included the five risk factors from the family domain also was an adequate fit to the data, (n = 859, X2(23) = 190.93, p b .001, RMSEA = .092, CFI = .955). Family involvement in antisocial behavior (b = .18, p b .01) and parental favorable attitudes towards antisocial behavior (b = .25, p b .001) were significant risk factors for recreational drug use in the model. Parental favorable attitudes towards drug use also were associated with participant recreational drug use but in a direction opposite of prediction (b = −.17, p b .01). 3.2.3. Peer risk model The model that included the three risk factors from the peer domain also was an adequate fit to the data (n = 859, X2 (17) = 190.74, p b .001, RMSEA = .109, CFI = .954). Peer antisocial behavior (b = .17, p b .01) and peer rewards for antisocial behavior (b = .23, p b .001) were significant risk factors for recreational drug use in the model. 3.2.4. School risk model The school risk model was tested next. The model was an adequate fit to the data (n = 859, X2(14) = 192.17, p b .001, RMSEA = .122, CFI = .952). Low school commitment was a significant risk factor for recreational drug use in the model (b = .19, p b .001).
152
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
3.2.5. Community risk model The community risk model also was an adequate fit to the data (n = 859, X2(20) = 179.08, p b .001, RMSEA = .096, CFI = .957). Community disorganization (b = .14, p b .001), community attitudes favorable to drug use (b = .11, p b .001), and perceived availability of drugs (b = .14, p b .001) were significant risk factors for recreational drug use in the model. 3.2.6. Protective factor model The protective factor model included all eight protective factors and was an adequate fit to the data (n = 859, X2(32) = 207.88, p b .001, RMSEA = .080, CFI = .953). Religious practices (b = − .15, p b .001), belief in a moral order (b = − .07, p b .05), and opportunities for prosocial school involvement (b = −.13, p b .01) each were significantly associated with lower lifetime recreational drug use; rewards for prosocial school involvement (b = .10, p b .05) and rewards for prosocial community involvement (b = .12, p b .001) were significantly associated with higher lifetime recreational drug use, which was unexpected.
3.2.7. Final model predicting recreational drug use with significant predictors from previous models A final model that included the significant predictors of recreational drug use from previous models was run in order to determine the strongest correlates of recreational drug use. This model was an adequate fit to the data (n = 859, X2(56) = 245.89, p b .001, RMSEA = .063, CFI = .951). Sensation seeking (b = .07, p b .05), early initiation of antisocial behavior (b = .16, p b .001), peer antisocial behavior (b = .10, p b .05), and peer rewards for antisocial behavior (b = .13, p b .001) were uniquely associated with increased risk of lifetime recreational drug use, as was rewards for prosocial community involvement (b = .09, p b .001), which was an unexpected effect. Religious practices (b = −.13, p b .001) and opportunities for prosocial school involvement (b = − .15, p b .001) were uniquely associated with the likelihood of lower lifetime recreational drug use. 3.2.8. Multiple group analyses: age group effects Comparison of the constrained and unconstrained models across age group slightly favored the unconstrained model based on the BIC (constrained BIC = 7002.484; unconstrained BIC 6919.560), although other indicators of model fit were similar (X2(136) = 459.47, p b .001, RMSEA = .074, CFI = .921 for the constrained model; X2(118) = 420.81, p b .001, RMSEA = .077, CFI = .926 for the unconstrained model). Inspection of the paths (see Fig. 1 and Table 3) revealed two paths that were significant for both age groups and eight paths were associated with recreational drug use for one age group but not the other. For both adolescents (b = .18, p b .001) and young adults (b = .24, p b .001) early initiation of antisocial behavior was a risk factor for lifetime recreational drug use, and opportunities for prosocial school involvement reduced the likelihood of lifetime recreational drug use (b = −.10, p b .05 for adolescents; b = −.23, p b .05 for young adults). Predictors that were uniquely associated with adolescents' risk for recreational drug use included peer reinforcement for antisocial behavior (b = .13, p b .01), low school commitment (b = .11, p b .01), community disorganization (b = .09, p b .05), community attitudes favorable to drug use (b = .13, p b .01), and low levels of religious practices (b = −.14, p b .001). Counter to prediction, rewards for prosocial community involvement were positively associated with risk for recreational drug use among adolescents (b = .15, p b .001). For young adults, in addition to the predictors noted above, peer engagement in antisocial behavior (b = .24, p b .01) was a risk factor for recreational drug use. Counter to prediction, rewards for prosocial school involvement were positively associated with risk for recreational drug use young adults (b = .25, p b .001).
3.3. Results of models predicting hard drug use The models with recreational drug use as the outcome were repeated with hard drug use as the outcome. In general the findings mirrored those with recreational drug use. 3.3.1. Individual risk model The model that included the five risk factors from the individual domain was an adequate fit to the data (n = 859, X2(23) = 53.54, p b .001, RMSEA = .039, CFI = .994). Sensation seeking (b = .10, p b .01) and early initiation of antisocial behavior (b = .24, p b .001) were the significant predictors of hard drug use in the model. 3.3.2. Family risk model The model that included the five risk factors from the family domain also was an adequate fit to the data, (n = 859, X2(23) = 47.59, p b .001, RMSEA = .035, CFI = .995). Family involvement in antisocial behavior (b = .16, p b .001) and parental favorable attitudes towards antisocial behavior (b = .20, p b .01) were significant risk factors for hard drug use in the model. Parental favorable attitudes towards drug use also were associated with participant hard drug use but in a direction opposite of prediction (b = −.13, p b .05). 3.3.3. Peer risk model The model that included the three risk factors from the peer domain also was an adequate fit to the data (n = 859, X2(17) = 50.90, p b .001, RMSEA = .048, CFI = .994). As before, peer antisocial behavior (b = .18, p b .01) and peer rewards for antisocial behavior (b = .21, p b .001) were significant risk factors for hard drug use in the model. 3.3.4. School risk model The school risk model was tested next. The model was an adequate fit to the data (n = 859, X2(14) = 25.37, p b .05, RMSEA = .030, CFI = .998). Low school commitment was a significant risk factor for hard drug use in the model (b = .19, p b .001). 3.3.5. Community risk model The community risk model was tested last. This model also was an adequate fit to the data (n = 859, X2(20) = 42.21, p b .001, RMSEA = .036, CFI = .996). Community disorganization (b = .13, p b .001), community attitudes favorable to drug use (b = .10, p b .01), and perceived availability of drugs (b = .12, p b .001) each heightened risk for hard drug use in the model. 3.3.6. Protective factor model The protective factor model included all eight protective factors and was an adequate fit to the data, (n = 859, X2(32) = 54.84, p b .01, RMSEA = .029, CFI = .996). Religious practices (b = − .13, p b .001), belief in a moral order (b = −.08, p b .05), rewards for prosocial family involvement (b = − .09, p b .05), and opportunities for prosocial school involvement (b = −.13, p b .01) each were significantly associated with lower lifetime hard drug use. Counter to expectations, rewards for prosocial school involvement (b = .12, p b .01) and rewards for prosocial community involvement (b = .12, p b .001) were significantly associated with higher lifetime hard drug use. 3.3.7. Final model predicting hard drug use with significant predictors from previous models A final model that included the significant predictors of hard drug use from previous models was run in order to determine the strongest correlates of hard drug use. This model was an adequate fit to the data (n = 859, X2(59) = 110.99, p b .001, RMSEA = .032, CFI = .991). Early initiation of antisocial behavior (b = .12, p b .01), peer antisocial behavior (b = .10, p b .05), peer rewards for antisocial behavior (b = .13, p b .001), and low school commitment (b = .08, p b .05) were uniquely associated with increased risk of lifetime hard drug use, as
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
153
Early Init of Anti-Social Behavior A .18***/ YA .24***
Peer Anti-Social Behavior
A .07/ YA .24***
Peer Rewards Anti-Social Behavior
A .13** / YA.13
Beer
A .11** / YA -.01 Low School Commitment
Community Disorganiz.
Recreational Drug Use
Wine/Liquor
A .09* / YA -.02 Inhalants A .13** / YA -.07
Community Favorable Attitudes Drugs
Marijuana A -.14***/ YA -.12 Sex / Age
Religious Practices
Opportunities for Prosocial School Involvement
A -.10* / YA -.23*
A -.01 / YA .25*** Rewards for Prosocial School Involvement
Rewards for Prosocial Community Involvement
A .15*** / YA -.03
Fig. 1. Final model reflecting best risk and protective factor predictors of lifetime recreational drug use in Malaysian adolescents and young adults. Only predictors that were significant for either adolescents or young adults are presented. Significant multiple group estimates for adolescents (A) and young adults (YA) representing the moderation by age group are presented. Standardized parameter estimates are depicted. * p b .05; ** p b .01; *** p b .001.
Table 3 Results of multiple group analyses (standardized beta weights) comparing adolescents and young adults in models predicting recreational and hard drug use. Recreational drug use
Hard drug use
Risk or protective factor in the model
Adolescents
Young adults
Adolescents
Young adults
Sensation seeking Early initiation of drug use Family involvement in antisocial behavior Parental favorable attitudes towards drug use Parental favorable attitudes towards antisocial behavior Peer antisocial behavior Peer rewards for antisocial behavior Low school commitment Community disengagement Community favorable attitudes towards drug use Community perceived availability of drugs Religious practices Belief in a moral order Rewards for prosocial family involvement Opportunities for prosocial school involvement Rewards for prosocial school involvement Rewards for prosocial community involvement
.06 .18*** −.04 −.11 .11 .07 .13** .11** .09* .13** 0 −.14*** 0 −− −.10* −.01 .15***
.08 .24*** −.07 −.04 −.07 .24*** .13 −.01 −.02 −.07 −.02 −.12 −.04 −− −.23* .25*** −.03
.07 .10 −.02 −.06 .06 .06 .12** .13** .07 .10* −.03 −.10** −.04 −.03 −.09* .05 .18***
.06 .21** −.10 −.04 −.13 .27*** .14* .03 .01 −.02 0 −.08 −.05 −.07 −.15 .26** 0
Note. *p b .05; **p b .01; ***p b .001.
154
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
were rewards for prosocial school involvement (b = .12, p b .01) and rewards for prosocial community involvement (b = .12, p b .01), which were unexpected effects. Religious practices (b = − .10, p b .01), rewards for prosocial family involvement (b = −.08, p b .05), and opportunities for prosocial school involvement (b = − .11, p b .05) were uniquely associated with the likelihood of lower lifetime hard drug use. 3.3.8. Multiple group analyses: age group effects As with the model predicting recreational drug use, comparison of the constrained and unconstrained models across age group slightly favored the unconstrained model based on the BIC (constrained BIC = 9139.649; unconstrained BIC 9052.963), although other indicators of model fit were similar (X2(143) = 323.10, p b .001, RMSEA = .054, CFI = .967 for the constrained model; X2(124) = 281.447, p b .001, RMSEA = .054, CFI = .972 for the unconstrained model). Inspection of the paths (see Fig. 2 and Table 3) revealed only one path that was significant for both age groups and eight paths were associated with hard drug use for one age group but not the other. For both adolescents (b = .12, p b .01) and young adults (b = .14, p b .01) peer reinforcement for antisocial behavior increased the likelihood of lifetime hard drug use. Predictors that were uniquely associated with adolescents' risk for hard drug use included low school commitment (b = .13, p b .01), community attitudes favorable to drug use (b = .10,
p b .05), low levels of religious practices (b = −.10, p b .01), and few opportunities for prosocial school involvement (b = −.09, p b .05). Counter to prediction, rewards for prosocial community involvement were positively associated with risk for hard drug use among adolescents (b = .18, p b .001). For young adults, in addition to peer reinforcement, early initiation of antisocial behavior (b = .21, p b .01) and peer engagement in antisocial behavior (b = .27, p b .001) were risk factors for hard drug use. Counter to prediction, rewards for prosocial school involvement was positively associated with risk for hard drug use young adults (b = .26, p b .01). 4. Discussion Consistent with prior research, early initiation of antisocial behavior in the individual risk factor domain (White et al., 2006; Wills & Dishion, 2004), peer antisocial behavior and peer reinforcement of antisocial behavior in the peer risk factor domain (Pandina, Johnson, & White, 2010; Tam, 2014) each were associated with enhanced risk of substance use in multivariate analyses that simultaneously accounted for other risk and protective factors. Further, religious practices (Chitwood, Weiss, & Leukefeld, 2008), and opportunities for prosocial school involvement, which potentially increase bonding to school, an institution that promotes conventional values (Kliewer & Zaharakis, 2014), were associated with reduced risk of substance use in these analyses. Additionally,
Early Init of Anti-Social Behavior A .10/ YA .21**
Peer Anti-Social Behavior
Peer Rewards for Anti-Social Behavior
A .06/ YA .27***
Psychotropic Pills
A .12**/ YA .14*
Community Favorable Attitudes – Drug Use
A .10* / YA -.02
Religious Practices
A-.10** / YA-.08
Stimulants
Hard Drug Use
Heroin Opportunities for Prosocial School Involvement
A -.09* / YA -.15
Morphine Sex/ Age Rewards for Prosocial School Involvement
Rewards for Prosocial Community Involvement
A .05 / YA.26**
A .18***/ YA 0
Fig. 2. Final model reflecting best risk and protective factor predictors of lifetime hard drug use in Malaysian adolescents and young adults. Only predictors that were significant for either adolescents or young adults are presented. Significant multiple group estimates for adolescents (A) and young adults (YA) representing the moderation by age group are presented. Standardized parameter estimates are depicted. * p b .05; ** p b .01; *** p b .001.
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
as support by past research, sensation seeking (Thatcher & Clark, 2010) was a unique predictor of recreational drug use, and low school commitment and (few) rewards for prosocial family involvement (Kliewer & Zaharakis, 2014) were unique predictors of hard drug use. Notably, in the final models that incorporated significant terms from previous models, both risk and protective factors made unique contributions to lifetime risk for recreational and hard drug use. Further, factors emerged across multiple domains of functioning, pointing to the importance of assessing risk and protective factors across the various contexts that impinge on adolescents' and young adults' risk for substance use. In addition to expecting findings, our data revealed several findings that were opposite of prediction. These findings emerged when the sample was considered together, prior to examining differential patterns by age group. First, the rewards for prosocial community involvement items were associated with increased risk of both recreational and hard drug use. This finding is difficult to explain, but may be linked to unique aspects of the Malaysian community context. Future studies might explore whether there are coercive aspects to rewards for prosocial community involvement, and how these might be associated with the iatrogenic effects observed here. Second, the rewards for prosocial school involvement subscale were associated with increased risk of hard drug use. Prosocial school involvement was operationalized in the survey as positive relationships with teachers. Although unexpected, this finding mirrors a study from another culture outside of the United States. In a study with over 17,000 Central American youth, Kliewer and Murelle (2007) found that positive student–teacher communication was associated with lower risk of marijuana use, but not with lower risk of cigarette use, alcohol use, drunkenness, other drug use, or problems with drugs and alcohol. The multiple group analyses suggested some developmental differences in patterns of risk. For recreational drug use, there were more unique predictors of drug use for adolescents versus young adults. Aside from shared risk and protective factors for recreational drug use, adolescents appeared to be more strongly influenced by peer reinforcement, by low school commitment, by community risk factors such as disorganization and favorable attitudes towards drug use, and by religious practices. Conversely, early initiation of antisocial behavior emerged as a strong risk factor for young adults but not for adolescents. Most of these findings were replicated for hard drug use. Although young adults are influenced by their environmental contexts, these data are consistent with research demonstrating the strong influence of peers and neighborhoods on adolescents (Pandina et al., 2010). The positive influence of religious practices on adolescent risk for substance use has been observed worldwide and may reflect a strong familial as well as cultural influence (Chitwood et al., 2008; Jennings, Kliewer, Ray, & Worthington, 2014). The patterns observed in the data may reflect underlying biological processes as well, as brain development associated with risk-taking is still occurring through middle adolescence and into young adulthood (Steinberg, 2007).
4.1. Study strengths and limitations Study strengths included a comprehensive assessment of risk and protective factors across multiple life domains; assessment of both recreational and hard drug use; inclusion of adolescents and young adults in the sample; and a multivariate approach to the analyses which assumes that risk factors cluster and protective factors do not operate in isolation. This approach is particularly important because if risk and protective factors are considered in isolation their influence is overestimated. Despite these strengths, there were a number of limitations to the study. These include the cross-sectional design and the use of a single reporter, although this is a common practice in risk and protective factor research using surveys like the Communities That Care instrument.
155
4.2. Implications for prevention and intervention These data suggest that prevention approaches should target multiple domains, with a specific focus on the individual and peer domains in this time frame. Within the individual domain, interventionists might provide opportunities for adolescents to engage in safe risk-taking, particularly for those youth with high sensation seeking tendencies. Finding ways to reduce peer influence might necessitate working through the individual, family, or school domains, perhaps by enhancing adolescent connections to conventional contexts such as school or family. The peer risk factors associated with adolescent substance use were correlated with risk factors in other domains, thus such a strategy would seem to make sense. Evidence-based approaches addressing risk and protective factors might be implemented in school or after-school contexts, or in other locations that appeal to youth. These data also suggest that prevention and intervention efforts should not stop at the secondary school level, as rates of substance use and abuse are high among students attending college and among young adults aged 18–25 residing in the community. Role of funding sources This study was not externally funded. Contributors Dr. Razali and Dr. Kliewer co-designed the study. Dr. Razali wrote portions of the introduction and discussion sections. Dr. Kliewer conducted the statistical analyses, wrote the methods and results sections and edited the manuscript. Both authors contributed to and have approved the final manuscript. Conflict of interest None of the authors have any conflicts of interests to declare.
References Arthur, M. W., Briney, J. S., Hawkins, J. D., Abbott, R. D., Brooke-Weiss, B. L., & Catalano, R. F. (2007). Measuring risk and protection in communities using the Communities That Care Youth Survey. Evaluation and Program Planning, 30, 197–211. http://dx.doi.org/ 10.1016/j.evalprogplan.2007.01.009. Arthur, M. W., Hawkins, J. D., Pollard, J. A., Catalano, R. F., & Baglioini, A. J., Jr. (2002). Measuring risk and protective factors for substance use, delinquency, and other adolescent problems behaviors. The Communities That Care Youth Survey. Evaluation Review, 26, 575–601. http://dx.doi.org/10.1177/019384102237850. Bentler, P. M. (1992). On the fit of models to covariances and methodology to the Bulletin. Psychological Bulletin, 112, 400–404. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen, & J. S. Long (Eds.), Testing structural models. Newbury Park, CA: American Book. Chitwood, D. D., Weiss, M. L., & Leukefeld, C. G. (2008). A systematic review of recent literature on religiosity and substance use. Journal of Drug Issues, 38, 653–688. http://dx.doi.org/10.1177/002204260803800302. Jennings, J., Kliewer, W., Ray, M., & Worthington, E. (2014). Religion and spirituality during adolescence. In T. P. Gullotta, & M. Bloom (Eds.), Encyclopedia of primary prevention and health promotion (pp. 1424–1433) (2nd ed.). New York, NY: Kluwer. Kliewer, W., & Murelle, L. (2007). Risk and protective factors for adolescent substance use: Findings from a study in selected Central American countries. Journal of Adolescent Health, 40, 448–455. http://dx.doi.org/10.1016/j.jadohealth.2006.11.148. Kliewer, W., & Zaharakis, N. (2014). Family-based models of drug etiology. In L. M. Scheier, & W. B. Hansen (Eds.), Parenting and teen drug use (pp. 37–61). New York, NY: Oxford University Press. Ministry of Women, Family and Community Development, & UNICEF Malaysia (2013). The Malaysian juvenile justice system. A study of mechanisms for handling children in conflict with the law. Putrajaya, Malaysia: Ministry of Women, Family and Community Development. Mohd Muzafar Shah Mohd Razali (2007). The relationship between the age of onset for delinquent behavior and chronic drug abuse among adolescents. Jurnal Antidadah Malaysia, 2, 97–110. Muthen, L. K., & Muthen, B. O. (2014). Mplus user's guide. Los Angeles, CA: Muthen & Muthen. National AntiDrug Agency (2013). Drug information 2012. Putrajaya, Malaysia: National Anti-Drug Agency, Ministry of Home Affairs. Pandina, R. J., Johnson, V. L., & White, H. R. (2010). Peer influences on substance use during adolescence and emerging adulthood. In L. M. Scheier (Ed.), Handbook of drug use etiology: Theory, methods, and empirical findings (pp. 383–401). Washington, DC: American Psychological Association. Razali, M. M., & Kliewer, W. Validation of the Communities That Care measure adapted for use in Malaysia. International perspectives in psychology: Research, practice, consultation
156
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
Steinberg, L. (2007). Risk taking in adolescence: New perspectives from brain and behavioral science. Current Directions in Psychological Science, 16, 55–59. http://dx. doi.org/10.1111/j.1467-8721.2007.00475.x. Stone, A. L., Becker, L. G., Huber, A. M., & Catalano, R. F. (2012). Review of risk and protective factors of substance use and problem use in emerging adulthood. Addictive Behaviors, 37, 747–775. http://dx.doi.org/10.1016/j.addbeh.2012.02.014. Tam, F. W. M. (2014). A Structural Equation Model of drug abuse among secondary students in Hong Kong. International Journal of Mental Health Addiction, 12, 255–269. http://dx.doi.org/10.1007/s11469-013-9450-5. Thatcher, D. L., & Clark, D. B. (2010). Neurodevelopmental liability for adolescent substance use disorders. In L. M. Scheier (Ed.), Handbook of drug use etiology: Theory, methods, and empirical findings (pp. 209–224). Washington, DC: American Psychological Association. United Nations (2014). World drug report 2014. UN Publication No. E.14.XI.7. New York, NY: United Nations Office on Drugs and Crime.
White, H. R., McMorris, B. J., Catalano, R. F., Fleming, C. B., Haggerty, K. P., & Abbott, R. D. (2006). Increases in alcohol and marijuana use during the transition out of high school into emerging adulthood: The effects of leaving home, going to college, and high school protective factors. Journal of Studies on Alcohol, 67, 810–822. Wills, T. A., & Dishion, T. J. (2004). Temperament and adolescent substance use: A transactional analysis of emerging self-control. Journal of Clinical Child and Adolescent Psychology, 33, 69–81. Wongtongkam, N., Ward, P. R., Day, A., & Winefield, A. H. (2014). The influence of protective and risk factors in individual, peer and school domains on Thai adolescents alcohol and illicit drug use: A survey. Addictive Behaviors, 39, 1447–1451. http://dx. doi.org/10.1016/j.addbeh.2014.05.026.
Contents lists available at ScienceDirect
Addictive Behaviors
Risk and protective factors for recreational and hard drug use among Malaysian adolescents and young adults Muzafar Mohd Razali a, Wendy Kliewer b,⁎ a b
No. 25, Jalan Segar 7 Taman Segar 35900 Tanjong Malim, Perak Malaysia Virginia Commonwealth University, United States
H I G H L I G H T S • • • • •
Risk and protective factors for substance use occurred in multiple domains. Risk and protective factors for recreational and hard drug use were similar. Risk and protective factors for substance use differed by age group. Risk factors in the individual and peer domains were particularly salient. Religious practices and prosocial school involvement were protective factors.
a r t i c l e
i n f o
Available online 12 June 2015 Keywords: Malaysia Substance use Risk factors Protective factors Adolescents Young adults
a b s t r a c t Introduction: This study investigated risk and protective factors for recreational and hard drug use in Malaysian adolescents and young adults. Methods: Participants (n = 859; M age = 17.24 years, SD = 2.75 years, range = 13–25 years; 59% male) were recruited from secondary schools, technical colleges, a juvenile detention center and a national training center in Malaysia. A version of the Communities That Care survey validated for use in Malaysia (Razali & Kliewer, 2015) was used to assess study constructs. Results: One in 6 adolescents and 1 in 3 young adults reported lifetime recreational and hard drug use, with greater use reported by males across all drug categories. Structural equation modeling was used to determine the strongest risk and protective factors for recreational and hard drug use. The overall pattern of findings was similar for recreational and hard drug use. Shared risk factors for lifetime recreational and hard drug use included early initiation of antisocial behavior, peer antisocial behavior, and peer reinforcement for engaging in antisocial behavior; shared protective factors included religious practices and opportunities for prosocial school involvement. Multiple group analyses comparing adolescents and young adults indicated that patterns of risk and protective factors predicting drug use differed across these age groups. There were fewer significant predictors of either recreational or hard drug use for young adults relative to adolescents. Conclusions: Results suggest that interventions should target multiple microsystems (e.g., peer groups, family systems, school environments) and be tailored to the developmental stage of the individual. © 2015 Elsevier Ltd. All rights reserved.
1. Introduction Drug use in adolescence and young adulthood is a social and health problem worldwide. It is estimated that in 2012, between 162 million and 324 million people, corresponding to between 3.5% and 7.0% of the world's population aged 15–64 had used an illicit drug – mainly a ⁎ Corresponding author at: Department of Psychology, Virginia Commonwealth University, PO Box 842018, Richmond VA 23284-2018, United States. Tel.: +1 804 828 8089; fax: +804 828 2237. E-mail addresses: [email protected] (M.M. Razali), [email protected] (W. Kliewer).
http://dx.doi.org/10.1016/j.addbeh.2015.06.022 0306-4603/© 2015 Elsevier Ltd. All rights reserved.
substance belonging to the cannabis, opioid, cocaine or amphetaminetype stimulants group – at least once in the previous year (United Nations, 2014). Malaysia is concerned with the social and economic impact of drug abuse on young people (Ministry of Women, Family and Community Development, & UNICEF Malaysia, 2013). The National AntiDrug Agency (2013) reported out of the total 11,108 new addicts in 2012, 4.14% were ages 13–19 and 76.54% were ages 20–34. The percentage of new addicts who were adolescents and young adults increased from 2008, when half of the new addicts were in that age range. Malaysia has observed a shift from recreational to hard drug use among young people in the last decade (National AntiDrug Agency, 2013).
150
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
The trends in drug use observed in Malaysia – particularly among those aged 18 to 25 – are consistent with data elsewhere. Substance use is extremely high in this developmental period, and consequences of drug use and misuse can be severe. Further, risk and protective factors for substance use in young adulthood are important to identify as they may be distinct from factors that predict adolescent substance use (Stone, Becker, Huber, & Catalano, 2012). Malaysia is interested in addressing the issue of substance abuse among adolescents and young adults, but has not been successful, perhaps because prevention and intervention programs have not been evidence-based (Mohd Muzafar Shah Mohd Razali, 2007). There is a dearth of research in Malaysia on risk and protective factors for substance use or abuse among adolescents and young adults. The purpose of the present study was to address this knowledge gap and extend research on risk and protective factors for substance use to the Malaysian context. Building on work on risk and protective factors conducted both within (Arthur, Hawkins, Pollard, Catalano, & Baglioini, 2002, Arthur et al., 2007) and outside (Kliewer & Murelle, 2007; Wongtongkam, Ward, Day, & Winefield, 2014) of the United States, we extend prior research in several ways. First, we use an assessment tool that underwent an extensive adaptation and validation process for use in the Malaysian context. Second, we focus on both adolescents and young adults (through age 25) in our sample. Third, we assess both recreational (alcohol, marijuana, inhalants) and hard (pills, stimulants, heroin, morphine) drug use. 2. Method This cross-sectional survey included participants from secondary schools (n = 498), technical colleges (n = 267), a juvenile detention center (n = 38) and a national training center (n = 56). Because of the focus on adolescents and young adults, participants over the age of 25 were excluded from the sample. Participants ranged in age from 13 to 25 (M = 17.24 years, SD = 2.75 years), and 59% were male. Most of the sample were Malay (92%), but also included Indian (3.8%), Chinese (2.7), and other (1.5%) ethnicities. Prior to conducting the study, special permission was obtained from the educational and correctional authorities respectively. This permission is required by the Malaysian Ministry of Education before conducting research in schools and colleges under their authority. Permission to conduct the study in religious schools was obtained from the State Islamic Affairs Department. Additionally, permission was obtained from the Department of Welfare for the juvenile center participants. Surveys in the secondary schools were administered by a group of Sultan Idris Education University student counselors undergoing the final semester of their internship training. All counselors underwent training in the responsible conduct of research prior to administering the surveys. The ethics of participant confidentiality were emphasized in the training. The counselors were familiar to the students and a high level of trust existed between the counselors and the study participants. Questionnaires for the participants in the juvenile center were administered by their counselors. All participants were guaranteed confidentiality of their responses and also their right to withdraw from participation at any time. The schools and juvenile center also were given the right to withdraw from this study if they wished to do so. 2.1. Assessment of risk and protective factors for drug use Risk and protective factors were assessed with a version of the Communities That Care (CTC; Arthur et al., 2002, 2007) survey adapted and validated for use in Malaysia (Razali & Kliewer). The Malaysian version of the CTC included 137 items assessing risk and protection in a total of 27 individual, peer, family, school, and community domains. See Table 1 for a list of the risk and protective factors included in the study.
Table 1 Descriptive information on risk and protective factor scales included in the study. Scale
Alpha Mean SD
Risk factors – individual domain Rebelliousness Sensation seeking Favorable attitudes towards drug use Favorable attitudes towards antisocial behavior Early initiation of antisocial behavior
.72a .87 .94 .92 .85
3.85 6.85 8.58 14.04 18.49
Risk factors – family domain Poor parental monitoring Family conflict Family antisocial behavior Favorable attitudes towards drug use Favorable attitudes towards antisocial behavior
.85 .83 .97 .97 .92
17.39 6.42 4.54 2.24 28.80 11.03 7.93 5.00 9.77 5.56
Risk factors – peer domain Peer antisocial behavior Peer drug use Peer rewards for antisocial behavior
.92 .96 .87
28.07 14.92 6.89
8.83 5.56 4.05
Risk factors – school domain Low school commitment Poor academic performance
.86 .61a
13.39 5.10
5.35 1.91
.83a .72 .86
4.92 8.29 6.36
2.19 3.43 3.57
.94
16.77
9.73
Protective factors – individual domain Religious practices Belief in a moral order
.87a .88
8.20 17.43
2.09 3.69
Protective factors – family domain Attachment to family Opportunities for prosocial family involvement Rewards for prosocial family involvement
.77 .85 .63a
15.70 11.62 7.78
3.76 3.04 1.79
Protective factors – school domain Opportunities for prosocial school involvement Rewards for prosocial school involvement
.75 .81
18.09 14.02
3.91 3.41
Protective factors – community domain Rewards for prosocial community involvement
.65a
6.92
2.18
Risk factors – neighborhood domain Low neighborhood attachment Community disorganization Favorable community attitudes towards tobacco, alcohol, and cannabis use Perceived availability of drugs
1.81 3.52 5.17 8.35 4.67
a
Note. reflects the correlation between two items. For all risk factors, higher values indicate greater level of risk; for all protective factors, higher values indicate greater level of protection.
2.2. Assessment of recreational and hard drug use Past year and lifetime use of all substances were assessed, but for the purposes of this report only lifetime use is reported. Lifetime use was assessed on a 5-point scale where 0 = never used; 1 = used 1–2 times; 2 = used sometimes but not regularly; 3 = regular user previously but not currently; and 4 = current regular user. Recreational drug use included beer, wine and liquor, inhalants, and marijuana. Hard drug use included psychotropic pills, stimulants, heroin, and morphine. Psychotropic pill use in Malaysia is analogous to non-medical use of prescription drugs (NMUPD) in the United States. That is, these pills are medications that are prescribed to treat mental (e.g., depression, anxiety) and physical health (e.g., pain) problems, but are used in a manner not prescribed, or used with a prescription, often for the purpose of getting high. Table 2 presents frequency distributions of recreational and hard drug use for the adolescents and young adults in the sample.
2.3. Data analysis Descriptive statistics on the frequency of lifetime recreational and hard drug use are presented first. In order to evaluate the contribution
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156 Table 2 Frequency of lifetime recreational and hard drug use by age group (percentage endorsing). Adolescents
Young adults
Beer Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
83.2 12.4 2.4 1.5 0.5
64.7 28.9 2.4 0.8 3.2
Wine/Liquor Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
83.8 12.5 2.0 1.0 0.7
64.2 30.9 2.1 1.2 1.6
Inhalants Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
83.7 13.6 1.5 0.7 0.5
64.4 31.2 2.8 0.4 1.2
Marijuana Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.0 13.3 1.9 0.5 0.3
63.2 30.4 2.4 2.4 1.6
Psychotropic pills Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.5 12.5 0.8 1.3 0.8
64.7 28.9 2.4 0.8 3.2
Stimulants Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.8 13.8 0.5 0.5 0.5
65.2 30.0 0.8 1.6 2.4
Heroin Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.8 14.2 0.8 0.2 0
64.9 32.7 0.4 1.2 0.8
Morphine Never used Used 1–2 times Used sometimes but not regularly Regular user previously but not currently Current regular user
84.5 13.8 1.3 0.2 0.2
65.9 31.3 0.8 1.2 0.8
of risk factors to drug use, two sets of Structural Equation Model (SEM) were run using Mplus 7.2 (Muthen & Muthen, 2014). One set included recreational drugs as the outcome, and the four recreational drugs were manifest indicators of a latent factor of recreational drug use in these models. The second set included hard drugs as the outcome, and the four hard drugs were manifest indicators of a latent factor of hard drug use in these models. All drug outcomes were transformed in the analyses due to skewness and kurtosis. Within each set of analyses, a model was run within each risk domain (e.g., individual, family, peer, school, and community) followed by a model with protective factors in order to identify significant domain-specific risk and protective factors that were associated with drug use. A final comprehensive model included the significant predictors from all prior models. Age and gender were controlled in all analyses. Once a comprehensive model was specified, multiple group analyses were conducted to determine if the pattern of associations between risk and protective factors and drug use varied by age group (adolescents
151
[n = 600] versus young adults [n = 259]). An unconstrained model where path coefficients were allowed to vary by age group was compared to a constrained model where path coefficients were set to be equal across age groups. Model fit was assessed with the Root Mean Square Error of Approximation (RMSEA; Browne & Cudeck, 1993) and the Comparative Fit Index (CFI; Bentler, 1992), and was deemed adequate if the RMSEA was b.08 and/or the CFI was N.90. Additionally, fits for the unconstrained and constrained models were evaluated by examining the χ2 difference test and differences in the models based on the CFI, RMSEA, and the Bayesian Information Criterion (BIC). Lower BIC scores suggested a better model fit. 3. Results 3.1. Lifetime prevalence of recreational and hard drug use As seen in Table 2, lifetime substance use varied by age, as was expected. Approximately 1 in 6 adolescents and 1 in 3 young adults reported lifetime use of both recreational and hard drugs. However, current, regular use of alcohol or other drugs was very low. In contrast, a quarter of the adolescents and half of the young adults reported trying tobacco (not shown in Table 2); 9.2% of the adolescents and 21.2% of the young adults reported current, regular tobacco use. In order to assess overall gender differences in substance use, a multivariate analysis was conducted. The multivariate effect of gender, controlling for age, was significant, F(8, 807) = 7.93, p b .001, with males reporting higher prevalence of substance use than females on all substances. As reflected in Table 2, there also was an overall effect of age, F(8, 807) = 2.99, p b .01. 3.2. Results of models predicting recreational drug use 3.2.1. Individual risk model The model that included the five risk factors from the individual domain was an adequate fit to the data, (n = 859, X2(23) = 194.92, p b .001, RMSEA = .093, CFI = .954). Sensation seeking (b = .11, p b .01) and early initiation of antisocial behavior (b = .28, p b .001) were the significant risk factors for recreational drug use in the model. 3.2.2. Family risk model The model that included the five risk factors from the family domain also was an adequate fit to the data, (n = 859, X2(23) = 190.93, p b .001, RMSEA = .092, CFI = .955). Family involvement in antisocial behavior (b = .18, p b .01) and parental favorable attitudes towards antisocial behavior (b = .25, p b .001) were significant risk factors for recreational drug use in the model. Parental favorable attitudes towards drug use also were associated with participant recreational drug use but in a direction opposite of prediction (b = −.17, p b .01). 3.2.3. Peer risk model The model that included the three risk factors from the peer domain also was an adequate fit to the data (n = 859, X2 (17) = 190.74, p b .001, RMSEA = .109, CFI = .954). Peer antisocial behavior (b = .17, p b .01) and peer rewards for antisocial behavior (b = .23, p b .001) were significant risk factors for recreational drug use in the model. 3.2.4. School risk model The school risk model was tested next. The model was an adequate fit to the data (n = 859, X2(14) = 192.17, p b .001, RMSEA = .122, CFI = .952). Low school commitment was a significant risk factor for recreational drug use in the model (b = .19, p b .001).
152
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
3.2.5. Community risk model The community risk model also was an adequate fit to the data (n = 859, X2(20) = 179.08, p b .001, RMSEA = .096, CFI = .957). Community disorganization (b = .14, p b .001), community attitudes favorable to drug use (b = .11, p b .001), and perceived availability of drugs (b = .14, p b .001) were significant risk factors for recreational drug use in the model. 3.2.6. Protective factor model The protective factor model included all eight protective factors and was an adequate fit to the data (n = 859, X2(32) = 207.88, p b .001, RMSEA = .080, CFI = .953). Religious practices (b = − .15, p b .001), belief in a moral order (b = − .07, p b .05), and opportunities for prosocial school involvement (b = −.13, p b .01) each were significantly associated with lower lifetime recreational drug use; rewards for prosocial school involvement (b = .10, p b .05) and rewards for prosocial community involvement (b = .12, p b .001) were significantly associated with higher lifetime recreational drug use, which was unexpected.
3.2.7. Final model predicting recreational drug use with significant predictors from previous models A final model that included the significant predictors of recreational drug use from previous models was run in order to determine the strongest correlates of recreational drug use. This model was an adequate fit to the data (n = 859, X2(56) = 245.89, p b .001, RMSEA = .063, CFI = .951). Sensation seeking (b = .07, p b .05), early initiation of antisocial behavior (b = .16, p b .001), peer antisocial behavior (b = .10, p b .05), and peer rewards for antisocial behavior (b = .13, p b .001) were uniquely associated with increased risk of lifetime recreational drug use, as was rewards for prosocial community involvement (b = .09, p b .001), which was an unexpected effect. Religious practices (b = −.13, p b .001) and opportunities for prosocial school involvement (b = − .15, p b .001) were uniquely associated with the likelihood of lower lifetime recreational drug use. 3.2.8. Multiple group analyses: age group effects Comparison of the constrained and unconstrained models across age group slightly favored the unconstrained model based on the BIC (constrained BIC = 7002.484; unconstrained BIC 6919.560), although other indicators of model fit were similar (X2(136) = 459.47, p b .001, RMSEA = .074, CFI = .921 for the constrained model; X2(118) = 420.81, p b .001, RMSEA = .077, CFI = .926 for the unconstrained model). Inspection of the paths (see Fig. 1 and Table 3) revealed two paths that were significant for both age groups and eight paths were associated with recreational drug use for one age group but not the other. For both adolescents (b = .18, p b .001) and young adults (b = .24, p b .001) early initiation of antisocial behavior was a risk factor for lifetime recreational drug use, and opportunities for prosocial school involvement reduced the likelihood of lifetime recreational drug use (b = −.10, p b .05 for adolescents; b = −.23, p b .05 for young adults). Predictors that were uniquely associated with adolescents' risk for recreational drug use included peer reinforcement for antisocial behavior (b = .13, p b .01), low school commitment (b = .11, p b .01), community disorganization (b = .09, p b .05), community attitudes favorable to drug use (b = .13, p b .01), and low levels of religious practices (b = −.14, p b .001). Counter to prediction, rewards for prosocial community involvement were positively associated with risk for recreational drug use among adolescents (b = .15, p b .001). For young adults, in addition to the predictors noted above, peer engagement in antisocial behavior (b = .24, p b .01) was a risk factor for recreational drug use. Counter to prediction, rewards for prosocial school involvement were positively associated with risk for recreational drug use young adults (b = .25, p b .001).
3.3. Results of models predicting hard drug use The models with recreational drug use as the outcome were repeated with hard drug use as the outcome. In general the findings mirrored those with recreational drug use. 3.3.1. Individual risk model The model that included the five risk factors from the individual domain was an adequate fit to the data (n = 859, X2(23) = 53.54, p b .001, RMSEA = .039, CFI = .994). Sensation seeking (b = .10, p b .01) and early initiation of antisocial behavior (b = .24, p b .001) were the significant predictors of hard drug use in the model. 3.3.2. Family risk model The model that included the five risk factors from the family domain also was an adequate fit to the data, (n = 859, X2(23) = 47.59, p b .001, RMSEA = .035, CFI = .995). Family involvement in antisocial behavior (b = .16, p b .001) and parental favorable attitudes towards antisocial behavior (b = .20, p b .01) were significant risk factors for hard drug use in the model. Parental favorable attitudes towards drug use also were associated with participant hard drug use but in a direction opposite of prediction (b = −.13, p b .05). 3.3.3. Peer risk model The model that included the three risk factors from the peer domain also was an adequate fit to the data (n = 859, X2(17) = 50.90, p b .001, RMSEA = .048, CFI = .994). As before, peer antisocial behavior (b = .18, p b .01) and peer rewards for antisocial behavior (b = .21, p b .001) were significant risk factors for hard drug use in the model. 3.3.4. School risk model The school risk model was tested next. The model was an adequate fit to the data (n = 859, X2(14) = 25.37, p b .05, RMSEA = .030, CFI = .998). Low school commitment was a significant risk factor for hard drug use in the model (b = .19, p b .001). 3.3.5. Community risk model The community risk model was tested last. This model also was an adequate fit to the data (n = 859, X2(20) = 42.21, p b .001, RMSEA = .036, CFI = .996). Community disorganization (b = .13, p b .001), community attitudes favorable to drug use (b = .10, p b .01), and perceived availability of drugs (b = .12, p b .001) each heightened risk for hard drug use in the model. 3.3.6. Protective factor model The protective factor model included all eight protective factors and was an adequate fit to the data, (n = 859, X2(32) = 54.84, p b .01, RMSEA = .029, CFI = .996). Religious practices (b = − .13, p b .001), belief in a moral order (b = −.08, p b .05), rewards for prosocial family involvement (b = − .09, p b .05), and opportunities for prosocial school involvement (b = −.13, p b .01) each were significantly associated with lower lifetime hard drug use. Counter to expectations, rewards for prosocial school involvement (b = .12, p b .01) and rewards for prosocial community involvement (b = .12, p b .001) were significantly associated with higher lifetime hard drug use. 3.3.7. Final model predicting hard drug use with significant predictors from previous models A final model that included the significant predictors of hard drug use from previous models was run in order to determine the strongest correlates of hard drug use. This model was an adequate fit to the data (n = 859, X2(59) = 110.99, p b .001, RMSEA = .032, CFI = .991). Early initiation of antisocial behavior (b = .12, p b .01), peer antisocial behavior (b = .10, p b .05), peer rewards for antisocial behavior (b = .13, p b .001), and low school commitment (b = .08, p b .05) were uniquely associated with increased risk of lifetime hard drug use, as
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
153
Early Init of Anti-Social Behavior A .18***/ YA .24***
Peer Anti-Social Behavior
A .07/ YA .24***
Peer Rewards Anti-Social Behavior
A .13** / YA.13
Beer
A .11** / YA -.01 Low School Commitment
Community Disorganiz.
Recreational Drug Use
Wine/Liquor
A .09* / YA -.02 Inhalants A .13** / YA -.07
Community Favorable Attitudes Drugs
Marijuana A -.14***/ YA -.12 Sex / Age
Religious Practices
Opportunities for Prosocial School Involvement
A -.10* / YA -.23*
A -.01 / YA .25*** Rewards for Prosocial School Involvement
Rewards for Prosocial Community Involvement
A .15*** / YA -.03
Fig. 1. Final model reflecting best risk and protective factor predictors of lifetime recreational drug use in Malaysian adolescents and young adults. Only predictors that were significant for either adolescents or young adults are presented. Significant multiple group estimates for adolescents (A) and young adults (YA) representing the moderation by age group are presented. Standardized parameter estimates are depicted. * p b .05; ** p b .01; *** p b .001.
Table 3 Results of multiple group analyses (standardized beta weights) comparing adolescents and young adults in models predicting recreational and hard drug use. Recreational drug use
Hard drug use
Risk or protective factor in the model
Adolescents
Young adults
Adolescents
Young adults
Sensation seeking Early initiation of drug use Family involvement in antisocial behavior Parental favorable attitudes towards drug use Parental favorable attitudes towards antisocial behavior Peer antisocial behavior Peer rewards for antisocial behavior Low school commitment Community disengagement Community favorable attitudes towards drug use Community perceived availability of drugs Religious practices Belief in a moral order Rewards for prosocial family involvement Opportunities for prosocial school involvement Rewards for prosocial school involvement Rewards for prosocial community involvement
.06 .18*** −.04 −.11 .11 .07 .13** .11** .09* .13** 0 −.14*** 0 −− −.10* −.01 .15***
.08 .24*** −.07 −.04 −.07 .24*** .13 −.01 −.02 −.07 −.02 −.12 −.04 −− −.23* .25*** −.03
.07 .10 −.02 −.06 .06 .06 .12** .13** .07 .10* −.03 −.10** −.04 −.03 −.09* .05 .18***
.06 .21** −.10 −.04 −.13 .27*** .14* .03 .01 −.02 0 −.08 −.05 −.07 −.15 .26** 0
Note. *p b .05; **p b .01; ***p b .001.
154
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
were rewards for prosocial school involvement (b = .12, p b .01) and rewards for prosocial community involvement (b = .12, p b .01), which were unexpected effects. Religious practices (b = − .10, p b .01), rewards for prosocial family involvement (b = −.08, p b .05), and opportunities for prosocial school involvement (b = − .11, p b .05) were uniquely associated with the likelihood of lower lifetime hard drug use. 3.3.8. Multiple group analyses: age group effects As with the model predicting recreational drug use, comparison of the constrained and unconstrained models across age group slightly favored the unconstrained model based on the BIC (constrained BIC = 9139.649; unconstrained BIC 9052.963), although other indicators of model fit were similar (X2(143) = 323.10, p b .001, RMSEA = .054, CFI = .967 for the constrained model; X2(124) = 281.447, p b .001, RMSEA = .054, CFI = .972 for the unconstrained model). Inspection of the paths (see Fig. 2 and Table 3) revealed only one path that was significant for both age groups and eight paths were associated with hard drug use for one age group but not the other. For both adolescents (b = .12, p b .01) and young adults (b = .14, p b .01) peer reinforcement for antisocial behavior increased the likelihood of lifetime hard drug use. Predictors that were uniquely associated with adolescents' risk for hard drug use included low school commitment (b = .13, p b .01), community attitudes favorable to drug use (b = .10,
p b .05), low levels of religious practices (b = −.10, p b .01), and few opportunities for prosocial school involvement (b = −.09, p b .05). Counter to prediction, rewards for prosocial community involvement were positively associated with risk for hard drug use among adolescents (b = .18, p b .001). For young adults, in addition to peer reinforcement, early initiation of antisocial behavior (b = .21, p b .01) and peer engagement in antisocial behavior (b = .27, p b .001) were risk factors for hard drug use. Counter to prediction, rewards for prosocial school involvement was positively associated with risk for hard drug use young adults (b = .26, p b .01). 4. Discussion Consistent with prior research, early initiation of antisocial behavior in the individual risk factor domain (White et al., 2006; Wills & Dishion, 2004), peer antisocial behavior and peer reinforcement of antisocial behavior in the peer risk factor domain (Pandina, Johnson, & White, 2010; Tam, 2014) each were associated with enhanced risk of substance use in multivariate analyses that simultaneously accounted for other risk and protective factors. Further, religious practices (Chitwood, Weiss, & Leukefeld, 2008), and opportunities for prosocial school involvement, which potentially increase bonding to school, an institution that promotes conventional values (Kliewer & Zaharakis, 2014), were associated with reduced risk of substance use in these analyses. Additionally,
Early Init of Anti-Social Behavior A .10/ YA .21**
Peer Anti-Social Behavior
Peer Rewards for Anti-Social Behavior
A .06/ YA .27***
Psychotropic Pills
A .12**/ YA .14*
Community Favorable Attitudes – Drug Use
A .10* / YA -.02
Religious Practices
A-.10** / YA-.08
Stimulants
Hard Drug Use
Heroin Opportunities for Prosocial School Involvement
A -.09* / YA -.15
Morphine Sex/ Age Rewards for Prosocial School Involvement
Rewards for Prosocial Community Involvement
A .05 / YA.26**
A .18***/ YA 0
Fig. 2. Final model reflecting best risk and protective factor predictors of lifetime hard drug use in Malaysian adolescents and young adults. Only predictors that were significant for either adolescents or young adults are presented. Significant multiple group estimates for adolescents (A) and young adults (YA) representing the moderation by age group are presented. Standardized parameter estimates are depicted. * p b .05; ** p b .01; *** p b .001.
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
as support by past research, sensation seeking (Thatcher & Clark, 2010) was a unique predictor of recreational drug use, and low school commitment and (few) rewards for prosocial family involvement (Kliewer & Zaharakis, 2014) were unique predictors of hard drug use. Notably, in the final models that incorporated significant terms from previous models, both risk and protective factors made unique contributions to lifetime risk for recreational and hard drug use. Further, factors emerged across multiple domains of functioning, pointing to the importance of assessing risk and protective factors across the various contexts that impinge on adolescents' and young adults' risk for substance use. In addition to expecting findings, our data revealed several findings that were opposite of prediction. These findings emerged when the sample was considered together, prior to examining differential patterns by age group. First, the rewards for prosocial community involvement items were associated with increased risk of both recreational and hard drug use. This finding is difficult to explain, but may be linked to unique aspects of the Malaysian community context. Future studies might explore whether there are coercive aspects to rewards for prosocial community involvement, and how these might be associated with the iatrogenic effects observed here. Second, the rewards for prosocial school involvement subscale were associated with increased risk of hard drug use. Prosocial school involvement was operationalized in the survey as positive relationships with teachers. Although unexpected, this finding mirrors a study from another culture outside of the United States. In a study with over 17,000 Central American youth, Kliewer and Murelle (2007) found that positive student–teacher communication was associated with lower risk of marijuana use, but not with lower risk of cigarette use, alcohol use, drunkenness, other drug use, or problems with drugs and alcohol. The multiple group analyses suggested some developmental differences in patterns of risk. For recreational drug use, there were more unique predictors of drug use for adolescents versus young adults. Aside from shared risk and protective factors for recreational drug use, adolescents appeared to be more strongly influenced by peer reinforcement, by low school commitment, by community risk factors such as disorganization and favorable attitudes towards drug use, and by religious practices. Conversely, early initiation of antisocial behavior emerged as a strong risk factor for young adults but not for adolescents. Most of these findings were replicated for hard drug use. Although young adults are influenced by their environmental contexts, these data are consistent with research demonstrating the strong influence of peers and neighborhoods on adolescents (Pandina et al., 2010). The positive influence of religious practices on adolescent risk for substance use has been observed worldwide and may reflect a strong familial as well as cultural influence (Chitwood et al., 2008; Jennings, Kliewer, Ray, & Worthington, 2014). The patterns observed in the data may reflect underlying biological processes as well, as brain development associated with risk-taking is still occurring through middle adolescence and into young adulthood (Steinberg, 2007).
4.1. Study strengths and limitations Study strengths included a comprehensive assessment of risk and protective factors across multiple life domains; assessment of both recreational and hard drug use; inclusion of adolescents and young adults in the sample; and a multivariate approach to the analyses which assumes that risk factors cluster and protective factors do not operate in isolation. This approach is particularly important because if risk and protective factors are considered in isolation their influence is overestimated. Despite these strengths, there were a number of limitations to the study. These include the cross-sectional design and the use of a single reporter, although this is a common practice in risk and protective factor research using surveys like the Communities That Care instrument.
155
4.2. Implications for prevention and intervention These data suggest that prevention approaches should target multiple domains, with a specific focus on the individual and peer domains in this time frame. Within the individual domain, interventionists might provide opportunities for adolescents to engage in safe risk-taking, particularly for those youth with high sensation seeking tendencies. Finding ways to reduce peer influence might necessitate working through the individual, family, or school domains, perhaps by enhancing adolescent connections to conventional contexts such as school or family. The peer risk factors associated with adolescent substance use were correlated with risk factors in other domains, thus such a strategy would seem to make sense. Evidence-based approaches addressing risk and protective factors might be implemented in school or after-school contexts, or in other locations that appeal to youth. These data also suggest that prevention and intervention efforts should not stop at the secondary school level, as rates of substance use and abuse are high among students attending college and among young adults aged 18–25 residing in the community. Role of funding sources This study was not externally funded. Contributors Dr. Razali and Dr. Kliewer co-designed the study. Dr. Razali wrote portions of the introduction and discussion sections. Dr. Kliewer conducted the statistical analyses, wrote the methods and results sections and edited the manuscript. Both authors contributed to and have approved the final manuscript. Conflict of interest None of the authors have any conflicts of interests to declare.
References Arthur, M. W., Briney, J. S., Hawkins, J. D., Abbott, R. D., Brooke-Weiss, B. L., & Catalano, R. F. (2007). Measuring risk and protection in communities using the Communities That Care Youth Survey. Evaluation and Program Planning, 30, 197–211. http://dx.doi.org/ 10.1016/j.evalprogplan.2007.01.009. Arthur, M. W., Hawkins, J. D., Pollard, J. A., Catalano, R. F., & Baglioini, A. J., Jr. (2002). Measuring risk and protective factors for substance use, delinquency, and other adolescent problems behaviors. The Communities That Care Youth Survey. Evaluation Review, 26, 575–601. http://dx.doi.org/10.1177/019384102237850. Bentler, P. M. (1992). On the fit of models to covariances and methodology to the Bulletin. Psychological Bulletin, 112, 400–404. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen, & J. S. Long (Eds.), Testing structural models. Newbury Park, CA: American Book. Chitwood, D. D., Weiss, M. L., & Leukefeld, C. G. (2008). A systematic review of recent literature on religiosity and substance use. Journal of Drug Issues, 38, 653–688. http://dx.doi.org/10.1177/002204260803800302. Jennings, J., Kliewer, W., Ray, M., & Worthington, E. (2014). Religion and spirituality during adolescence. In T. P. Gullotta, & M. Bloom (Eds.), Encyclopedia of primary prevention and health promotion (pp. 1424–1433) (2nd ed.). New York, NY: Kluwer. Kliewer, W., & Murelle, L. (2007). Risk and protective factors for adolescent substance use: Findings from a study in selected Central American countries. Journal of Adolescent Health, 40, 448–455. http://dx.doi.org/10.1016/j.jadohealth.2006.11.148. Kliewer, W., & Zaharakis, N. (2014). Family-based models of drug etiology. In L. M. Scheier, & W. B. Hansen (Eds.), Parenting and teen drug use (pp. 37–61). New York, NY: Oxford University Press. Ministry of Women, Family and Community Development, & UNICEF Malaysia (2013). The Malaysian juvenile justice system. A study of mechanisms for handling children in conflict with the law. Putrajaya, Malaysia: Ministry of Women, Family and Community Development. Mohd Muzafar Shah Mohd Razali (2007). The relationship between the age of onset for delinquent behavior and chronic drug abuse among adolescents. Jurnal Antidadah Malaysia, 2, 97–110. Muthen, L. K., & Muthen, B. O. (2014). Mplus user's guide. Los Angeles, CA: Muthen & Muthen. National AntiDrug Agency (2013). Drug information 2012. Putrajaya, Malaysia: National Anti-Drug Agency, Ministry of Home Affairs. Pandina, R. J., Johnson, V. L., & White, H. R. (2010). Peer influences on substance use during adolescence and emerging adulthood. In L. M. Scheier (Ed.), Handbook of drug use etiology: Theory, methods, and empirical findings (pp. 383–401). Washington, DC: American Psychological Association. Razali, M. M., & Kliewer, W. Validation of the Communities That Care measure adapted for use in Malaysia. International perspectives in psychology: Research, practice, consultation
156
M.M. Razali, W. Kliewer / Addictive Behaviors 50 (2015) 149–156
Steinberg, L. (2007). Risk taking in adolescence: New perspectives from brain and behavioral science. Current Directions in Psychological Science, 16, 55–59. http://dx. doi.org/10.1111/j.1467-8721.2007.00475.x. Stone, A. L., Becker, L. G., Huber, A. M., & Catalano, R. F. (2012). Review of risk and protective factors of substance use and problem use in emerging adulthood. Addictive Behaviors, 37, 747–775. http://dx.doi.org/10.1016/j.addbeh.2012.02.014. Tam, F. W. M. (2014). A Structural Equation Model of drug abuse among secondary students in Hong Kong. International Journal of Mental Health Addiction, 12, 255–269. http://dx.doi.org/10.1007/s11469-013-9450-5. Thatcher, D. L., & Clark, D. B. (2010). Neurodevelopmental liability for adolescent substance use disorders. In L. M. Scheier (Ed.), Handbook of drug use etiology: Theory, methods, and empirical findings (pp. 209–224). Washington, DC: American Psychological Association. United Nations (2014). World drug report 2014. UN Publication No. E.14.XI.7. New York, NY: United Nations Office on Drugs and Crime.
White, H. R., McMorris, B. J., Catalano, R. F., Fleming, C. B., Haggerty, K. P., & Abbott, R. D. (2006). Increases in alcohol and marijuana use during the transition out of high school into emerging adulthood: The effects of leaving home, going to college, and high school protective factors. Journal of Studies on Alcohol, 67, 810–822. Wills, T. A., & Dishion, T. J. (2004). Temperament and adolescent substance use: A transactional analysis of emerging self-control. Journal of Clinical Child and Adolescent Psychology, 33, 69–81. Wongtongkam, N., Ward, P. R., Day, A., & Winefield, A. H. (2014). The influence of protective and risk factors in individual, peer and school domains on Thai adolescents alcohol and illicit drug use: A survey. Addictive Behaviors, 39, 1447–1451. http://dx. doi.org/10.1016/j.addbeh.2014.05.026.
