Risky behavior among Chilean youths

Novella, Rafael; Repetto, Andrea Article Risky behavior among Chilean youths Estudios de Economía Provided in Cooperation with: Department of Economics, University of Chile Suggested Citation: Novella, Rafael; Repetto, Andrea (2024) : Risky behavior among Chilean youths, Estudios de Economía, ISSN 0718-5286, Universidad de Chile, Departamento de Economía, Santiago de Chile, Vol. 51, Iss. 2, pp. 579-606 This Version is available at: https://hdl.handle.net/10419/314468 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-sa/4.0/ 579 Estudios de Economía, Vol.51 - Nº 2, Diciembre 2024. Págs 579-606 Risky Behavior among Chilean Youths* Conductas de riesgo entre los jóvenes en Chile RAFAEL NOVELLA** ANDREA REPETTO*** Abstract This paper examines the connection between risky behaviors and various socioeconomic factors, including skills, preferences, aspirations, expectations, and exposure to shocks. Using a representative sample of Chilean youths aged 15 to 19 years old, our analysis identifies self-esteem, risk aversion, and educational aspirations as relevant factors associated with participation in risky activities. Remarkably, even after accounting for socio-demographic factors, skills, expectations, aspirations, and preferences, we uncover a significant correlation between exposure to shocks at both individual and family levels and engagement in risky behavior. Particularly striking is the association between experiencing job loss and family illness and the prevalence of risky behaviors. Additionally, we observe positive correlations among the unexplained variability of these behaviors, suggesting a complementary relationship between these activities. While these patterns are correlational rather than causal, they offer valuable insights into the determinants of risky decision-making among youths. Key words: Smoking, alcohol, violence, unemployment, health, self-esteem JEL Classification: D91, I12, I31 * ** *** We would like to thank Manuel Agosín, Paola Bordón, Rómulo Chumacero, José De Gregorio, Mauricio Duce, Sebastián Edwards, Claudia Martínez, Francisco Pino, Cristina Riquelme, and “50 Years of Estudios de Economía” seminar participants for their comments and suggestions. The authors thank Joy Batthacharya for excellent research assistance. University College London, London, UK. https://orcid.org/0000-0002-2139-3291. E-mail: r[email protected] Pontificia Universidad Católica de Chile. https://orcid.org/0000-0003-2378-9684. Corresponding author: [email protected], Avda Vicuña Mackenna 4860, Edificio Mide UC, piso 3, Macul, Santiago, Chile. Received: August, 2023 Accepted: August, 2024 580 Estudios de Economía, Vol.51 - Nº 2 Resumen Este trabajo examina la relación entre conductas de riesgo y diversos factores socioeconómicos, incluyendo habilidades, preferencias, aspiraciones, expectativas y exposición a shocks. Utilizando una muestra representativa de jóvenes chilenos entre los 15 y 19 años de edad, nuestro análisis identifica la autoestima, la aversión al riesgo y las aspiraciones educativas como factores relevantes asociados a la participación en actividades riesgosas. Es interesante notar que, incluso después de controlar por factores sociodemográficos, habilidades, expectativas, aspiraciones y preferencias, observamos una correlación significativa entre la exposición a shocks tanto a nivel individual como familiar y la participación en conductas de riesgo. Resulta especialmente llamativa la relación entre la pérdida de empleo y las enfermedades familiares y la prevalencia de conductas de riesgo. Asimismo, observamos correlaciones positivas entre la variabilidad no explicada de los distintos comportamientos, lo que sugiere una relación complementaria entre estas actividades. Aunque los patrones descritos son correlacionales y no causales, ofrecen una pespectiva valiosa sobre los determinantes de las decisiones de riesgo de los jóvenes. Palabras clave: Tabaquismo, alcohol, violencia, desempleo, salud, autoestima Clasificación JEL: D91, I12, I31 1. INTRODUCTION Adolescence is a critical phase in life characterized by physiological, psychological, and social changes that have lifelong impacts. Many risky behaviors, such as smoking, consuming alcohol, engaging in sex, and committing crimes, occur for the first time during this period. These behaviors may have consequences for youths’ well-being, as they are associated with health, education, productivity, and labor market outcomes. Furthermore, these behaviors can also impact others through their relationship with crime, accidents, the cost of insurance, and potential dependence on public resources. The economics literature on risky behaviors describes individuals as trading off present and future costs and benefits. Individuals compare the expected present satisfaction associated with smoking or drinking alcohol to the expected (discounted) cost of future health problems or low productivity (Becker & Murphy, 1988). This approach has been complemented by behavioral economics literature, which describes circumstances where individuals do not necessarily act in their own best interests, such as engagement in risky activities 581 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto (O’Donoghue & Rabin, 2001). When it comes to teenagers and risky behaviors, developmental psychologists also concern themselves with how cognitive, affective, and social development affect decision-making (Fischhoff, 1992). The aim of this study is to analyze risky behaviors among teenagers in Chile. Our analysis has three goals. First, we aim to document the prevalence of specific risky behaviors among adolescents in the country and the patterns of association between these activities. Second, we seek to describe the correlations between risky behaviors and economic variables, including preferences, expectations, aspirations, and skills. Finally, we intend to describe the relationship between participation in risky behaviors and shocks, including unemployment and illness. Using a representative sample of Chilean youths aged 15-19, we find correlations between engaging in risky behavior and socioemotional skills, risk aversion, and educational aspirations. Interestingly, we find a significant correlation between exposure to shocks at both individual and family levels. Notably, experiencing job loss and health problems in the family are associated with a higher prevalence of risky behaviors. Furthermore, we document positive correlations between the unexplained variation of these behaviors, with the strongest associations observed among drugs and alcohol use. While these patterns are correlational rather than causal, we believe they offer valuable insights into the determinants of risky decision-making among youths. Figures 1, 2, and 3 reveal a decline in participation in several risky behaviors among adolescents in Chile, although some rates remain higher than in advanced countries. Figure 1 illustrates the prevalence of alcohol, tobacco, marijuana, and cocaine consumption among students in grades 8 to 12. Remarkably, the rate of alcohol and tobacco use dropped by 38% and 71% in the past two decades, respectively. However, according to PAHO statistics, the rates of tobacco use are still much higher than those in Canada (1%) and the United States (4.6%), and even higher than in other Latin American and Caribbean (LAC) countries, such as Brazil (6.9%), Peru (7.2%), and Uruguay (11.5%). Cocaine use also fell, from a prevalence of 1.5% in 2003 to 1% in 2021. However, the prevalence of marijuana use shows a different dynamic: it grew from 6.8% in 2003 to 20.1% in 2015, and then declined steadily to 11.2% in 2021. Figure 2 demonstrates that, after experiencing a plateau in the early 1990s, teen fertility in Chile steadily declined throughout the following decade. It rose again in the mid-2000s, then declined in the early 2010s. The downward trend is also observed in other regions, although it is less pronounced in LAC. Currently, Chile’s teenage fertility rate is about half of LAC’s rate but almost twice the rate observed in Europe and North America. 582 Estudios de Economía, Vol.51 - Nº 2 Furthermore, there has been a decrease in the number of adolescent offenders entering the justice system. Figure 3 indicates that the rate of adolescents in conflict with the law has steadily decreased, and at a much faster rate than that of adults. Although not strictly comparable, juvenile crime statistics in the United States also show a decreasing trend over time (Hockenberry and Puzzanchera, 2023).1 This study aims to make contributions to at least three distinct bodies of literature. The first relates to protective factors that mitigate the prevalence of risky behaviors among the youth population. These factors encompass socioemotional skills and educational aspirations (Donnellan et al., 2005; Chiteji, 2010; Favara and Sánchez, 2017), as well as peer and friend behaviors (Clark and Lohéac, 2006; Card and Giuliano, 2013; Einsberg et al., 2014). Moreover, the literature has evidenced that youth do respond to economic incentives such as tobacco and alcohol prices and taxes (Cook and Moore, 2001; Carpenter and Cook, 2008; Paraje et al., 2021; Assael, 2023), along with access restrictions (Cook and Moore, 2001; DiNardo and Lemieux, 2001; Wagenaar and Toomey, 2002). Similarly, contextual factors like the length of the school day and the type of educational institution attended can also influence the prevalence of certain behaviors (Berthelon and Kruger, 2011; Figlio and Ludwig, 2012). The second literature refers to the consequences of risky behavior among youths, encompassing effects on educational outcomes (Renna, 2007; Fletcher and Lehrer, 2009; Parkes et al., 2010; Lye and Hirschberg, 2010; Carrell et al., 2011), health (Patton et al., 2016), fertility (Kearny and Levine, 2012), and incarceration (Levitt and Lochner, 2001). Finally, our study contributes to the literature on the social consequences of economic shocks. Beyond the existing evidence on persistent earnings losses (Jacobson et al., 1993; von Wachter et al., 2009: Albagli et al., 2020), the literature has highlighted a connection with stress-related health issues and reduced life expectancy (Burgard et al., 2007; Sullivan and von Wachter, 2009; Eliason and Storrie, 2009), decreased school performance of children (Oreopoulos et al., 2008; Stevens and Schaller, 2011), and a higher incidence of divorce (Charles and Stephens, 2004). Furthermore, economic shocks are linked to lower happiness and life satisfaction (Frey and Stutzer, 2002). The remainder of the paper proceeds as follows. Section 2 presents the survey design and data, while Section 3 describes and discusses the results. Section 4 concludes. 1 In the United States, the overall delinquency rate among youths aged 10-16 in 2020 was 65%, below the 2005 rate. 583 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto 2. SURVEY DESIGN AND DATA 2.1 The Millennials in Latin America and the Caribbean Survey Our analysis is based on the Millennials in LAC survey, a cross-sectional survey conducted in Chile and six other countries in Latin America and the Caribbean.2 The survey was designed to study the schooling and labor market decisions of youths. The Chilean survey was administered between July and October 2017 and included information on 3,560 individuals aged 15 to 24 years living in the urban areas of the Metropolitan, Biobío, and Valparaíso regions. Households were selected based on previous censuses using a stratified multistage sampling method. The final stage of the sampling method consisted of randomly choosing a young person within the household. The survey consists of two questionnaires. The first contains standard demographic and socioeconomic questions. It also gathers information on cognitive and socioemotional skills, expectations, and aspirations, among other variables. The second questionnaire collected information about risky behaviors and was self-administered to improve the data’s response rate and quality (Tourangeau et al., 1997; Krumpal, 2013). Written consent was obtained from the participants if they were 18 or older or their parents otherwise. 2.2 Study Measures3 The survey gathers information on several measures of risky behavior. First, we measure whether the individual engaged in unprotected sex during the last sexual intercourse. Second, we measure violent behavior: whether the respondent or someone in his/her group of friends carried a weapon in the previous thirty days or committed a robbery in the last 12 months. Finally, we measure the consumption of alcohol and drugs: tobacco smoking, binge drinking, and marijuana and other drugs consumption in the last 12 months. We created dummy variables indicating whether the individual has engaged in each of these seven behaviors and a summary variable adding up the dummies. The survey also gathers information on risk and intertemporal preferences. To measure risk tolerance, subjects were asked about their willingness to pay 5% of a monthly minimum wage to participate in hypothetical lotteries with 2 The data used in this paper were collected as part of the “Millennials in Latin America and the Caribbean: to work or study?” project. The project, including the data collection, was funded by the International Development Research Centre (IDRC-Canada) and the Inter-American Development Bank (IDB). The countries in the study are Brazil, Chile, Colombia, El Salvador, Haiti, Mexico, and Paraguay. See Novella et al. (2018) and Alvarado et al. (2020) for more information. 3 This section is based on Alvarado et al. (2020) and Gantier et al. (2023). 584 Estudios de Economía, Vol.51 - Nº 2 prizes between 1% and 5% of the same minimum wage. We created a risk averse dummy variable equal to 1 when the individual is unwilling to participate in any of these lotteries and 0 otherwise. We expect a lower likelihood of risky behaviors among risk averse individuals. We include traditional sociodemographic control variables such as age and gender. Education is measured by the number of years of schooling achieved. We also include a teenage parenthood dummy that equals one if the youth had a child when younger than 20 years old or is pregnant and younger than 20, and zero otherwise. Finally, we include measures of dependency in the family, that is, the number of household members below five and above 65 years of age. We also include the household’s monthly income per capita. In addition, we include measures of economic shocks experienced by the youth (job loss and illness) and the family (divorce, illness or death, and crime) in the last 12 months to capture sources of unexpected variation in household resources. We created a set of dummy variables indicating separately the experience of each shock and an aggregated variable that adds up these variables. The survey considers a basic numeracy test that poses simple problems in which respondents must divide and multiply to obtain the correct answers. To evaluate cognitive achievement, we use the standardized percentage of correct answers such that the measure has a zero mean and unit variance (z-scores). The survey also measures socioemotional skills. In particular, it includes the Rosenberg self-esteem test that measures people’s image of themselves (Rosenberg, 1965). We normalize this measure to have a mean of zero and a standard deviation of one. A higher score reflects higher self-esteem. Therefore, we expect socioemotional skills to be negatively correlated with risky behavior. The questionnaire also gathers perceptions about wages. To measure the perceived returns to schooling, we include the difference between the monthly salary youths believe college graduates earn in their local area in logs, and the earnings of local secondary education graduates (also in logs); i.e., the expected return to a college education. We hypothesize that individuals who expect higher returns are less likely to engage in risky behavior. Finally, to measure educational aspirations, the survey asks individuals about the highest academic degree they would like to complete assuming no constraints. Our aspirational measure is the number of additional years of education the respondents would like to meet beyond what they have achieved. We hypothesize a negative correlation between aspirations and risky behavior. 585 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto 2.3 Sample Our sample consists of teenage youths (i.e., those aged between 15 and 19) with complete information on the relevant variables. The full sample contains 1,916 individuals in the relevant age group. We lose 489 observations due to missing relevant data. Appendix Table 1 compares our sample and the sample of individuals missing information. Table 1 contains the main statistics of our sample and according to the youth’s engagement in at least one of our risky behavior measures. On average, youths in the sample are 17 years old and have completed almost ten years of education. The sample is evenly distributed by gender. Six percent have already had a child or were expecting one at the time of the interview.4 3. RISKY BEHAVIOR AMONG YOUTHS 3.1 Correlates of Risky Behavior In Table 1 we present the main statistics of the final sample, categorized by youths’ engagement in at least one of our risky behavior measures. On average, individuals in the sample have participated in 1.4 risky behaviors out of a potential of 7 over the last 12 months. The most prevalent activities are binge drinking and marijuana consumption, reported by 42% and 34% of the sample, respectively. Among those who have engaged in at least one risky behavior, the mean number of risky activities rises to 2.4. Figures 4 and 5 illustrate simple correlations between engagement in each risky activity and various characteristics and environmental factors of youths. The statistical significance of difference-in-means tests for aggregate risky behavior is reported in the final columns of Table 1. The figures and tests reveal differences between individuals who engage in risky behaviors and those who do not across relevant aspects. Figure 4 demonstrates that men (Panel a) and older individuals (Panel b) are more likely to engage in risky behavior. Panels c and d also indicate differences based on numerical and socioemotional skills, with a lower prevalence 4 To explore the external validity of our results, we compared our final sample to the nationally representative 2017 CASEN survey sample. We find many demographic similarities when limiting the analysis to youths aged 15 to 19 living in the urban areas of Santiago, Biobío, and Valparaíso. In the CASEN sample, 50% of individuals are men, the average age is 17, and the average years of education is 10.7. However, we find a significant difference in income per capita: including subsidies, households’ income in CASEN is twice that in our sample. Possibly, youths are unaware of the monetary resources available in their households, as in CASEN, the primary respondent is an adult. 586 Estudios de Economía, Vol.51 - Nº 2 of risky activities among individuals with higher skills.5 Furthermore, Panels e and f suggest that individuals who engage in risky behaviors have lower educational aspirations and expect lower returns from a college education, respectively. Panel g shows that they are also less likely to be risk averse. Notably, youths who engage in risky behaviors have experienced a larger number of shocks and a higher prevalence of all types of shocks except for parental separation. Particularly, they are more likely to have lost their jobs, experienced an illness of death in the family, or had someone in the household become a crime victim. These differences are statistically significant, as shown in Table 1. Figure 5 depicts the prevalence of each risky activity by shock experience, highlighting the noteworthy differences. For instance, those who experienced a shock are 62% more likely to consume marijuana and to smoke tobacco than those who have not. 3.2 Co-Occurrence of Risky Behaviors In this section, we investigate the relationships between different risky behaviors, exploring whether individuals who engage in one behavior are also likely to engage in others. To mitigate the risk of spurious correlations stemming from youth characteristics and environmental factors, we initially regress each risky behavior dummy variable on the observable covariates presented in Table 1. Subsequently, we estimate pairwise correlations between the residuals. Specifically, we estimate seven linear probability models, each corresponding to a relevant behavior (unprotected sex, smoking, etc.). Our models control for various sociodemographic variables (including gender, age, years of education, household’s monthly income per capita, teenage parenthood status, and the number of household members younger than five and older than 65), measures of cognitive and non cognitive skills (numeracy and self-esteem), as well as variables capturing expected returns, educational aspirations, risk averse, and indicator variables for shocks. Additionally, we incorporate dummy variables to account for regional differences that may influence youths’ decisions, encompassing factors like labor market conditions, the educational environment, and social preferences. Appendix Table 2 presents the estimation results, while Table 2 summarizes the residuals correlations. Our findings reveal relevant correlation among risky activities. Particularly noteworthy is the close connection observed among binge drinking, tobacco use, and consumption of marijuana and other drugs. These results may reflect 5 We divide the sample into halves when we plot behavior by numeracy skills, self-es- teem, aspirations, and expectations. 593 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto Wagenaar, A. C., & Toomey, T. L. (2002). Effects of minimum drinking age laws: review and analyses of the literature from 1960 to 2000. Journal of Studies on Alcohol, supplement, (14), 206-225. 594 Estudios de Economía, Vol.51 - Nº 2 FIGURE 1 PREVALENCE OF TOBACCO,ALCOHOL, MARIJUANA, AND COCAINE USE, 8TH-12TH GRADE (% USE IN PREVIOUS MONTH) Note: SENDA (2023). FIGURE 2 FERTILITY RATES, WOMEN AGED 15-19 (BIRTHS PER 1000 WOMEN) Note: United Nations, World Population Prospects 2022. 595 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto FIGURE 3 ADMISSIONS OF CRIMINAL CASES BY AGE OF DEFENDANT (PER 100.000 INHABITANTS IN THE RESPECTIVE AGE GROUP) Note: UNICEF and Defensoría Penal Pública (2020). FIGURE 4 PREVALENCE OF RISKY BEHAVIORS BY INDIVIDUAL CHARACTERISTICS A. GENDER 596 Estudios de Economía, Vol.51 - Nº 2 B. AGE C. NUMERACY SKILLS 597 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto D. SELF-ESTEEM E. EDUCATIONAL ASPIRATIONS 598 Estudios de Economía, Vol.51 - Nº 2 F. EXPECTED RETURNS TO COLLEGE G. RISK AVERSION 599 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto FIGURE 5 PREVALENCE OF RISKY BEHAVIORS BY EXPERIENCE OF SHOCKS FIGURE 6 CONDITIONAL CORRELATION BETWEEN RISKY BEHAVIORS AND INDIVIDUAL SHOCKS 600 Estudios de Economía, Vol.51 - Nº 2 TABLE 1 DESCRIPTIVE STATISTICS All No risky behavior At least one risky behavior Means test Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Difference p-value Any risky behavior 0.60 0.49 0.00 0.00 1.00 0.00 Number of risky behaviors 1.43 1.55 0.00 0.00 2.40 1.31 Unprotected sex 0.09 0.29 0.00 0.00 0.16 0.37 Carries a weapon 0.07 0.25 0.00 0.00 0.11 0.32 Robbery 0.19 0.39 0.00 0.00 0.31 0.46 Tobacco 0.22 0.41 0.00 0.00 0.36 0.48 Binge drinking 0.42 0.49 0.00 0.00 0.70 0.46 Marijuana 0.34 0.48 0.00 0.00 0.58 0.50 Other drugs 0.11 0.31 0.00 0.00 0.18 0.39 Age 17.03 1.39 16.61 1.37 17.31 1.33 -0.70 0.27 Male 0.51 0.50 0.49 0.50 0.52 0.50 -0.03 0.00 Years of education 9.95 1.86 9.57 1.94 10.21 1.76 -0.64 0.06 Teenage parenthood 0.06 0.24 0.05 0.21 0.07 0.25 -0.02 0.34 Household members under 5 years 0.25 0.53 0.24 0.55 0.26 0.52 -0.03 0.60 Household members over 65 years 0.22 0.50 0.23 0.50 0.21 0.50 0.01 0.47 601 Risky Behavior among Chilean Youths / Rafael Novella, Andrea Repetto Note: This table presents averages for the complete sample, and according to youths’ engagement in risky behavior. The table also presents the p-value for the test of differences of means between youths engaging and those not engaging in risky behavior. Any risky behavior is a dummy variable indicating whether the youth engaged in at least one of the behaviors listed below. Unprotected sex, carries a weapon, robbery, tobacco, binge drinking, marijuana, other drugs, teenage pregnancy, risk averse, and the shock variables are all dummy variables. Numeracy skills and Rosenberg score are normalized variables. Income per capita (thousand pesos) 121.20 87.50 118.75 86.69 122.93 88.04 -4.17 0.60 Numeracy skills 0.03 1.00 0.05 0.99 0.02 1.01 0.03 0.38 Rosenberg score 0.01 1.00 0.08 0.99 -0.04 1.00 0.12 0.03 Expected return to a college education 0.63 0.50 0.63 0.49 0.63 0.51 0.01 0.75 Educational aspirations 6.38 2.65 6.78 2.66 6.11 2.61 0.67 0.00 Risk averse 0.54 0.50 0.61 0.49 0.49 0.50 0.12 0.00 Number of economic shocks (0-5) 0.70 0.87 0.56 0.77 0.80 0.92 -0.24 0.00 Job loss youth 0.09 0.28 0.04 0.20 0.12 0.32 -0.07 0.00 Parental separation 0.06 0.23 0.05 0.22 0.06 0.24 -0.01 0.34 Health shock youth 0.11 0.33 0.09 0.28 0.12 0.33 -0.04 0.02 Health or death in family 0.27 0.44 0.24 0.43 0.29 0.45 -0.05 0.05 Crime victim 0.18 0.39 0.14 0.34 0.21 0.41 -0.08 0.00 Number of observations 1427 1427 577 577 850 850 1427 1427 602 Estudios de Economía, Vol.51 - Nº 2 TABLE 2 RESIDUAL CORRELATIONS Note: This table presents the correlations between the residuals of linear probability regression models of each risky behavior dummy variable on the set of observable covariates in Table 1. The first entry is the correlation, while the second entry is its significance. Unprotected sex Carries a weapon Robbery Tobaco Binge drinking Marijuana Other drugs Unprotected sex 1.000 0.000 Carries a weapon 0.030 1.000 0.260 0.000 Robbery 0.075 0.071 1.000 0.004 0.008 0.000 Tobaco 0.037 0.056 0.096 1.000 0.168 0.034 0.000 0.000 Binge drinking 0.102 0.064 0.239 0.338 1.000 0.000 0.015 0.000 0.000 0.000 Marijuana 0.121 0.177 0.176 0.435 0.484 1.000 0.000 0.000 0.000 0.000 0.000 0.000 Other drugs 0.046 0.071 0.140 0.194 0.251 0.405 1.000 0.084 0.008 0.000 0.000 0.000 0.000 0.000