Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Feb 23.
Published in final edited form as: Subst Use Misuse. 2013 Aug 23;49(1-2):116–123. doi: 10.3109/10826084.2013.824468

Substance use among adolescents in California: A latent class analysis

Tamika D Gilreath a,*, Ron A Astor a, Joey N Estrada Jr b, Renee M Johnson c, Rami Benbenishty d, Jennifer B Unger e
PMCID: PMC3842372  NIHMSID: NIHMS491141  PMID: 23971853

Abstract

Data from the California Healthy Kids Survey of 7th, 9th, and 11th graders were used to identify latent classes/clusters of alcohol, tobacco, and marijuana use (N=418,702). Analyses revealed four latent classes of substance use which included non-users (61.1%), alcohol experimenters (some recent alcohol use; 22.8%), mild poly-substance users (lifetime use of all substances with less than three days of recent use; 9.2%), and frequent poly-substance users (used all substances three or more times in the past month; 6.9%). The results revealed that alcohol and marijuana use are salient to California adolescents. This information can be used to target and tailor school-based prevention efforts.

Keywords: adolescents, alcohol, tobacco, marijuana

Introduction

Adolescence is a crucial period for the initiation of substance use, and use can negatively impact health and well-being in adulthood.(Eaton et al., 2010; Grunbaum et al., 2004; Irwin, Burg, & Cart, 2002; Jessor, 1991) The health and financial costs associated with substance abuse are excessive, making prevention an important goal for policymakers and practitioners.(Ettner et al., 2006) For instance, tobacco use is responsible for the majority of preventable diseases and death in the United States.(USDHHS, 2004) It has been estimated that underage drinking results in $62 billion annually. This figure includes medical and societal expenses, costs associated with diminished quality of life related to motor vehicle crashes, and other consequences of drinking behavior.(Miller, Levy, Spicer, & Taylor, 2006) These realities underscore the importance of prevention research on adolescent substance use. To move the field forward, we examine patterns of substance use among youth in the state of California.

National data indicate that substantial proportions of youth are using alcohol, tobacco, and marijuana. National Youth Risk Behavior Surveillance Survey (YRBS) data from 2009 show that 73% of all high school-attending youth reported lifetime use of alcohol, whereas 42% reported past 30-day use.(Centers for Disease Control and Prevention, 2011) The lifetime and past 30-day prevalence of cigarette use was lower than for alcohol, but still substantial (respectively, 46% and 20%). More than one-third of adolescents reported lifetime marijuana use, and 21% reported past 30-day use.(Centers for Disease Control and Prevention, 2011) Although early adolescents are less likely than high school-attending youth to use alcohol, tobacco, and marijuana (Johnston, O’Malley, Bachman, & Schulenberg, 2012), the adverse consequences of use for this age group are substantial with earlier initiation having serious developmental and social implications (Newcomb, Scheier, & Bentler, 1997; Squeglia, Jacobus, & Tapert, 2009).

Although informative, binary measures of lifetime and past 30-day use of alcohol, tobacco and marijuana provide only part of the picture of patterns of adolescent substance use. Comprehensive information about how frequently youth use substances, as well as about joint use of substances (i.e., “polysubstance use”) is also needed. Although substance use researchers emphasize that it is rare for youth to use only one substance, it is uncommon to see nuanced analyses of polysubstance use, and such analyses rarely account for recency and frequency of use (e.g., number of times used). Importantly, the number and types of substances used by the same person have been shown to predict other behavioral health risks.(Gilreath, Connell, & Leventhal, in press) Connell, Gilreath, and Hansen (2009) found that – along with frequency and recency of substance use – the use of multiple substances was strongly associated with sexual risk behavior.(Connell, Gilreath, & Hansen, 2009) Therefore, more detailed data about patterns of adolescent substance use could be used to inform and enhance prevention programs to address substance use as well as other risk behaviors. Latent class analysis (LCA) is one of the best methodological tools available to understand nuanced patterns of risk and risk behaviors (Stephanie T. Lanza, Rhoades, Greenberg, & Cox, 2011; Sullivan, Childs, & O’Connell, 2010).

There is significant variation in patterns of adolescent substance use by geographic location.(Connell, Gilreath, Aklin, & Brex, 2010) As an illustration, several key differences are apparent when comparing two articles that include similar LCAs of adolescent substance use. The first article included non-metropolitan youth in New England, and the second used a nationally representative sample (Connell, et al., 2010; Connell, et al., 2009). Among those classified as “frequent polysubstance users (PSUs)”, the response probabilities (i.e., the likelihood of having used any one drug or a combination of drugs) were distinct for the two populations. Nationally, frequent PSUs had a 52% chance of drinking alcohol on more than 6 days in the past month, compared to a 66% chance among youth in the New England study. Additionally, in the national study frequent PSUs had a 47% chance of having used marijuana on >6 days in the past month, compared to a 75% chance in the New England sample.

Patterns of substance use among California youth are likely to differ from national estimates for several reasons. First, California is the most populous and demographically diverse state in the U.S., and 25% of its population is younger than 18 years.(U.S. Census Bureau, 2012) Additionally, California was the first state to enact a comprehensive statewide tobacco control policy in 1989 with the overarching goal of reducing tobacco use among its population.(Bal, Kizer, Felton, Mozar, & Niemeyer, 1990) Finally, California passed the first medical marijuana proposition in 1996, and established dispensaries in 2003.(California Department of Public Health, 2012) Medical marijuana laws in California are less stringent compared to other states. Recent data indicate that the national prevalence of adolescent marijuana use has increased since 2008; it has been suggested that the increase is due to in youths’ declining perceptions of marijuana as dangerous.(Johnston, et al., 2012; Kuehn, 2011) Because California youth have grown up in an era of medical marijuana and comprehensive tobacco control, their patterns of substance use may differ from youth in the rest of the country.

The current study explicates patterns of substance use that take account for both frequency and recency of use of alcohol, tobacco and marijuana using LCA. We also examine how demographic factors are associated with substance use. Few studies have employed LCAs on large-scale or representative databases of youth; (Cleveland, Collins, Lanza, Greenberg, & Feinberg, 2010; Connell, et al., 2010; Connell, et al., 2009; Gilreath, et al., in press) and none have focused on California youth specifically. What is learned will be used to improve our understanding of the etiology of substance use and to inform prevention strategies in California and beyond.

Methods

The data used in this study are from the California Healthy Kids Survey (CHKS), conducted by WestEd, a nonprofit research, development, and service agency in collaboration with the California Department of Education. The CHKS is a biennial survey that consists of a core survey module that gathers demographic background data (e.g., grade, sex, and race/ethnicity) and inquires about students’ health-related behaviors (e.g., tobacco use, alcohol use, drug use, violence behaviors), and school safety. The CHKS was required to be administered biennially by all schools that received Title IV funding under the federal Safe and Drug Free Schools and Communities Act or the State’s Tobacco Use Prevention Education program (~85% of districts statewide). Under such mandates, schools must survey a representative district wide grade-level sample of students in the 5th, 7th, 9th, and 11th grades according to standards set forth by the California Department of Education (CDE). The CDE sampling procedure requires that 1) 100% of all district schools participate; or 100% of all selected schools from an approved sampling plan; 2) An appropriate class subject or class period was identified and used; 3) 100% of selected classrooms participated; AND 4) The number of completed, usable answer forms obtained per grade was 60% or more of the selected sample; OR 5) If active parental consent is used, 70% or more parents within each grade’s selected sample returned signed permission forms, either consenting or not consenting to their child’s participation. Additional details of the recommended sampling procedure is described in detail elsewhere (Austin & Duerr, 2004).

Prior to the survey being administered at a school site, parental consent was gathered by each school district through the CDE and WestEd for each participant. The core survey was administered by school staff members familiar with questionnaire administration or by WestEd employees if a school site chose to hire professionally trained survey administrators. Proctoring instructions were given to all survey administrators and an introductory script was read to the student participants. Participants were encouraged to answer questions honestly and assured their responses would remain anonymous. Participants were allowed to withdraw from the survey at any time. The survey took approximately 50 minutes to complete.

CHKS data collected for the 2005–2007 academic school years from students who self-reported as Hispanic ethnicity, African American, or White will be used in the present study (n=418,702). A weighting procedure was used to adjust the total of grade level respondents to represent the total district enrollment for the particular grade levels of interest and district level clustering was also accounted for. District-level consent procedures were followed and the present study has appropriate IRB approval from the University of Southern California.

Alcohol, tobacco, and marijuana use were each assessed using a single item that gauged lifetime use and frequency of past 30-day use. Specifically, the response categories were: never used, lifetime use without any past month use, used 1–2 days in past month, used 3–9 days in past month, or used 10 or more days in the past month.

Latent class analysis (LCA) was conducted using Mplus 6.1 (Lubke & Muthén, 2005; McCutcheon, 1987). LCA is used to identify homogeneous subgroups within a heterogeneous population(Auerbach & Collins, 2006; S.T. Lanza, Collins, Lemmon, & Schafer, 2007; Magidson & Vermunt, 2002). Multinomial logistic regression analyses were completed simultaneous with class estimation to account for measurement error related to class assignment. Gender, race/ethnicity (White, African American, and Hispanic/a), and educational level (7th, 9th, 11th) were included as demographic covariates in that regression in that regression.

A series of models was run to determine the appropriate number of classes for substance use starting with a 1-class (no covariates) model followed by a series of models with covariates specifying increased number of classes (e.g., 2-class, 3-class, etc.) representing different patterns of substance use behavior. Optimal model selection was based upon recommended indices including low Adjusted Bayesian Information Criterion (BIC) relative to other models, significant Lo-Mendell-Rubin Likelihood Ratio Test (LMR LRT), and acceptable quality of classification.(Nylund, Asparouhov, & Muthén, 2007)

Results

The sample was 47.5% male. Thirty six percent of the students were in 7th; 34.1% were in 9th grade. Hispanics comprised 58.8% of the sample, followed by Whites (33.1%), and African Americans (8.1%). Prevalence of lifetime use was 44.1% for alcohol, 28.7% for tobacco, and 22.5% for marijuana (Table 1). As shown in Table 2, a four-class model provided the best overall fit to the data for substance use behavior. This is exemplified by the non-significant p-value for the 5 class model which indicates that the (K-1)-class model should not be rejected in favor of a model with at least K-classes. The four classes were termed: Frequent Polysubstance Users (PSUs), Moderate PSUs, Polysubstance Experimenters, and Non-Users. Sixty-one percent of the respondents were in the non-user group. These youth reported little or no history of substance use.

Table 1.

Overall Demographic and Substance use characteristics, CHKS, 2005–2007

Demographic Characteristics Risk Behaviors
Weighted % Unweighted n Weighted % Unweighted n
Sex Alcohol Use
 Male 47.5 201913  Never Used 55.8 234199
 Female 52.5 219862  No Recent Use 20.0 80112
Grade  1 or 2 Days Recent Use 13.7 55813
 7th grade 36.2 152023  3 to 9 Days Recent Use 6.8 29362
 9th grade 35.1 148929  10 to 30 Days Recent Use 3.6 15475
 11th grade 28.7 123934 Tobacco Use
Race/Ethnicity  Never Used 71.3 299988
 Hispanic 58.8 223072  No Recent Use 20.4 81322
 White 33.1 172242  1 or 2 Days Recent Use 4.2 17330
 African American 8.1 29572  3 to 9 Days Recent Use 1.9 8343
 10 to 30 Days Recent Use 2.2 10304
Marijuana Use
 Never Used 77.6 325374
 No Recent Use 11.9 47336
 1 or 2 Days Recent Use 4.4 17822
 3 to 9 Days Recent Use 2.8 11230
 10 to 30 Days Recent Use 3.4 14809
Open in a new tab

Table 2.

Fit statistic comparisons of latent class analysis models of substance use in California

Model Description Substance Use in California
Adjusted BIC LMR LRT p-value Entropy
1 One-class (no co-variates) 5323905.862 0.0000 –
2 Two-class 2055183.168 0.0008 0.80
3 Three-class 2009106.900 0.0000 0.756
4 Four-class 2002007.651 0.000 0.738
5 Five-class 1992189.895 0.7602 0.712
Open in a new tab

Notes: BIC – Bayesian Information Criterion. LMR LRT – Lo-Mendell-Rubin Likelihood Ratio Test p-value for (K-1)-classes. A significant p-value indicates that the (K-1)-class model should be rejected in favor of a model with at least K-classes.

Best fitting models identified in bold.

Conditional probabilities for substance use are summarized in Table 3. Polysubstance experimenters accounted for 22.8% of the sample. The members of this class reported lifetime use, but low likelihood of recent use of alcohol, tobacco, or marijuana. Moderate PSUs accounted for 9.2% of the sample; they had at least a 30% chance of reporting use of alcohol, tobacco, and marijuana on at least 1 day in the past month. The frequent PSUs comprised 6.9% of the sample and had at least a 40% chance of indicating that they used alcohol, tobacco, and marijuana on three or more days in the past month. They had over a 40% chance of reporting using marijuana ten or more days in the past month.

Table 3.

Conditional probabilities of substance use (n=418, 702)

Class Prevalence Frequent Polysubstance Users Moderate Polysubstance Users Polysubstance Experimenters Non-Users
6.9% 9.2% 22.8% 61.1%
Tobacco
 Never Used 0.107 0.233 0.458 0.948
 No Recent Use 0.247 0.448 0.518 0.047
 1 or 2 Days Recent Use 0.164 0.260 0.017 0.004
 3 to 9 Days Recent Use 0.179 0.057 0.003 0.001
 10 to 30 Days Recent Use 0.302 0.002 0.004 0.001
Alcohol
 Never Used 0.028 0.024 0.165 0.844
 No Recent Use 0.084 0.115 0.554 0.095
 1 or 2 Days Recent Use 0.170 0.520 0.213 0.048
 3 to 9 Days Recent Use 0.364 0.308 0.043 0.008
 10 to 30 Days Recent Use 0.355 0.034 0.025 0.005
Marijuana
 Never Used 0.080 0.310 0.591 0.992
 No Recent Use 0.184 0.289 0.341 0.003
 1 or 2 Days Recent Use 0.139 0.258 0.039 0.003
 3 to 9 Days Recent Use 0.180 0.121 0.016 0.001
 10 to 30 Days Recent Use 0.417 0.022 0.014 0.001
Open in a new tab

Multinomial logistic regression analyses showed that demographic factors influenced class membership. Not surprisingly, being in a higher grade was associated with membership in any substance use class (i.e., Frequent PSU, Moderate PSU, or Polysubstance Experimenter) compared to the Non-user class. Compared to males, females were significantly more likely to be moderate PSUs (OR=1.31, 95% CI=1.12–1.54), but were significantly less likely to be frequent PSUs (OR=0.60, CI=0.56–0.63). Compared to Whites, African Americans were twice as likely to be polysubstance experimenters (OR=2.13, CI=1.29–3.53), and were significantly less likely to be either a moderate or frequent PSU. Hispanic/as were more likely than Whites to be polysubstance experimenters (OR=2.30, CI=1.94– 2.73) and moderate polysubstance users (OR=1.60, CI=1.33–1.93).

Discussion

This study sought to characterize patterns of frequency and recency of alcohol, tobacco, and marijuana use among a large sample of adolescents in California. The results of the present study indicate that there is considerable variation in substance use among youth and that this use is qualitatively and quantitatively different from national findings. Quantitatively, the majority of youth were classified as non-users (~61%) compared to only 27.7% nationally (Connell, Gilreath, Hansen, 2009). A majority of the present sample consisted of Hispanic youth (58.8%). Thus, the overall higher prevalence of non-users found may be attributable to the large proportion of youth who are likely members of immigrant families in California and may not be fully acculturated. Research has shown that adolescents who are less acculturated and recent immigrants are less likely to use drugs in adolescence than those who report greater acculturation (Almeida, Johnson, Godette, & Atsusi, in press; De la Rosa, 2002).

Qualitatively, the response probabilities that defined classes were also divergent. In each of the substance use classes (excluding non-users) the likelihood of recent tobacco use was lower in the present study compared to national estimates. This was likely driven by the fact that, overall, tobacco use was lower in this population as compared to the nation and underscore the importance of geographic differences substance use. The rates of tobacco use were also consistently lower than rates of marijuana use making alcohol and marijuana the primary drugs of use across the alcohol experimenters, moderate PSU, and frequent PSU classes. This is different from other large-scale findings of adolescent substance use, which generally show alcohol and tobacco use being predominant (Cleveland, et al., 2010; Connell, et al., 2009). Studies of adult tobacco use and mortality indicate that tobacco use has decreased in California at higher rates than national estimates (Cowling & Yang, 2010; Siegel et al., 2000). This study lends support to similar evidence for adolescents. However, in general rates of marijuana use were comparable to national estimates.

Approximately 7% of the youth in California are likely to report that they are using multiple substances frequently in the month prior to completing the survey (Frequent PSUs). Given that substance use in adolescence predisposes youth to numerous negative health and social outcomes (Newcomb, et al., 1997; Squeglia, et al., 2009). This class of frequent polysubstance users is of great concern because early initiation increases the likelihood of addiction and substance abuse in adulthood.

Finally, African American, Hispanic/a, and White secondary students in California schools vary in likelihood of the conjoint usage of these three substances combined with the frequency and recency of usage. For example, African American youth were significantly less likely to be moderate or frequent polysubstance users compared to their White counterparts. Hispanic/as were as likely as Whites to be classified as frequent polysubstance users. It may be that cultural differences in the norms of the peer group and popularity of multiple substances could contribute to these very specific patterns. For example higher levels of immigrant status among Hispanic youth likely contribute to lower rates while research focused on differences in substance use between African American and White samples show support for religiosity and cultural proscriptions against substance use as protective factors among African American adolescents (Bachman, Brown, Laveist, & Wallace, 2003; Clark, Scarisbrick-Hauser, Gautam, & Wirk, 1999).

The present study does have limitations. First the data are cross-sectional and no exploration of patterns of use over time can be assessed. As with most research on substance use, the data are based on self-report. Thirdly, this study only included students who were present at school and students who were absent or truant could potentially have different patterns of substance use. Finally, the instrument is limited in questions asked about frequency and recency of other drugs (including illicit and prescription medications). National data shows that consideration of other drug use including prescription and OTC drugs is important in understanding adolescent substance use and identifying targets for intervention (Connell, et al., 2009; Eaton, et al., 2010).

Currently, substance use tends to be assessed from a one-dimensional perspective, limited to whether or not respondents report any lifetime use or any recent use of each of the substances, most commonly use of alcohol, tobacco and marijuana (Eaton, et al., 2010; Johnston, O’Malley, Bachman, & Schulenberg, 2006). This is understandable, as it can be computationally difficult and require substantial sample sizes to account for the use of multiple substances at varying frequencies. However, detailed multi-dimensional understanding of substance use behaviors may be critical in: 1) triaging treatment protocols and interventions when resources are limited to target those who are engaged in the highest levels of polysubstance use; 2) identifying particular combinations of substances used for tailoring treatment; 3) understanding levels of addiction based upon frequency and recency of use.

These results suggest that, generally, primary intervention programs in California schools should target alcohol and marijuana use concurrently. More specifically, there is a subgroup of PSU youth who should be targeted for secondary interventions of tobacco, alcohol, and marijuana use. Important differences between frequent PSUs and moderate PSUs show that the former has a 30% chance of using tobacco, alcohol, and/or marijuana 10+ times in the past month. This level of usage places these youth well on their way to substance use disorders which will have substantial impacts on transitions into adulthood and beyond. Substance use and abuse in adolescence has long been a target of prevention and intervention science. This article provides empirical and evidence of the need to better understand the complexity of substance use behaviors and the need for different intervention strategies based upon severity of usage experimenters. It is important that public health and prevention scientists be responsive to the growing diversity of needs of the populations they serve and identify ways to maximize efficiency in intervention design and implementation by utilizing not only evidence-based but also data-driven person-centered models such as the one presented.

Table 4.

Multinomial Logistic Regression results of substance use in California (n=418, 702)

Covariates Polysubstance experimenters vs. Non-users Moderate polysubstance users vs. Non-users Frequent polysubstance users vs. Non-users
OR (95% CI) OR (95% CI) OR (95%)
Grade 1.77 (1.72 – 1.82) 1.77 (1.66 – 1.88) 1.87 (1.80 – 1.94)
Female 0.90 (0.80 – 1.01) 1.31 (1.12 – 1.54) 0.60 (0.56 – 0.63)
African American 2.13 (1.29 – 3.53) 0.55 (0.42 – 0.72) 0.71 (0.62 – 0.82)
Hispanic 2.30 (1.94 – 2.73) 1.60 (1.33 – 1.93) 0.94 (0.85 – 1.03)
Open in a new tab

References

  1. Almeida J, Johnson RM, Godette D, Atsusi M. Alcohol, tobacco, and marijuana use among immigrants: Results from the 2008 Boston Youth Survey. Social Science & Medicine in press. [Google Scholar]
  2. Auerbach KJ, Collins LM. A Multidimensional Developmental Model of Alcohol Use During Emerging Adulthood. Journal of Studies on Alcohol. 2006;67(6):917–925. doi: 10.15288/jsa.2006.67.917. [DOI] [PubMed] [Google Scholar]
  3. Austin G, Duerr M. Guidebook for the California Healthy Kids Survey. Part I: Administration. 2004–2005 Edition. San Francisco: WestEd; 2004. [Google Scholar]
  4. Bachman JG, Brown TN, Laveist TA, Wallace JM., Jr The influence of race and religion on abstinence from alcohol, cigarettes and marijuana among adolescents *. Journal of Studies on Alcohol. 2003 Nov;64:843. doi: 10.15288/jsa.2003.64.843. [DOI] [PubMed] [Google Scholar]
  5. Bal DG, Kizer KW, Felton PG, Mozar HN, Niemeyer D. Reducing tobacco consumption in California: Development of a statewide anti-tobacco use campaign. Journal of the American Medical Association. 1990;264:1570–1574. [PubMed] [Google Scholar]
  6. California Department of Public Health. Medical Marijuana Program. 2012 Retrieved March 2, 2012, 2012, from http://www.cdph.ca.gov/programs/mmp/Pages/Medical%20Marijuana%20Program.aspx.
  7. Centers for Disease Control and Prevention. 1991–2009 High School Youth Risk Behavior Survey Data. 2011 Retrieved May 1, 2011, from http://apps.nccd.cdc.gov/youthonline.
  8. Clark PI, Scarisbrick-Hauser A, Gautam SP, Wirk SJ. Anti-tobacco socialization in homes of African-American and white parents, and smoking and nonsmoking parents. Journal of Adolescent Health. 1999;24(5):329–339. doi: 10.1016/s1054-139x(98)00117-7. [DOI] [PubMed] [Google Scholar]
  9. Cleveland M, Collins L, Lanza ST, Greenberg M, Feinberg M. Does Individual Risk Moderate the Effect of Contextual-Level Protective Factors? A Latent Class Analysis of Substance Use. Journal of Prevention & Intervention in the Community. 2010;38(3):213–228. doi: 10.1080/10852352.2010.486299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Connell C, Gilreath T, Aklin WM, Brex RA. Social-ecological influences on patterns of substance use among non-metropolitan high school students. American Journal of Community Psychology. 2010;45(1–2):36–48. doi: 10.1007/s10464-009-9289-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Connell C, Gilreath T, Hansen N. A Multiprocess Latent Class Analysis of the Co-Occurrence of Substance Use and Sexual Risk Behavior Among Adolescents. Journal of Studies on Alcohol and Drugs. 2009;70(6):943–951. doi: 10.15288/jsad.2009.70.943. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cowling DW, Yang J. Smoking-attributable cancer mortality in California, 1979–2005. Tobacco Control. 2010;19(Suppl 1):i62–i67. doi: 10.1136/tc.2009.030791. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. De la Rosa MR. Acculturation and latino adolescents substance use: A research agenda for the future. Substance Use & Misuse. 2002;37(4):429–456. doi: 10.1081/ja-120002804. [DOI] [PubMed] [Google Scholar]
  14. Eaton DK, Kann L, Kinchen S, Shanklin S, Ross J, Hawkins J, et al. Youth risk behavior surveillance - United States, 2009. Morbidity and Mortality Weekly Report Surveillance Summary. 2010;59(5):1–142. [PubMed] [Google Scholar]
  15. Ettner S, Huang D, Evans E, Ash D, Hardy M, Jourabchi M, et al. Benefit-cost in the California Treatment Outcome Project: Does substance abuse treatment “pay for itself”? Health Services Research. 2006;41(1):192–213. doi: 10.1111/j.1475-6773.2005.00466.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gilreath T, Connell C, Leventhal A. Tobacco use and suicidality: Latent patterns of co-occurrence among African American adolescents. Nicotine & Tobacco Research. doi: 10.1093/ntr/ntr322. in press. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Grunbaum JA, Kann L, Kinchen S, Ross J, Hawkins J, Lowry R, et al. Youth Risk Behavior Surveillance-United States, 2003 (Abridged) Journal of School Health. 2004;74(8):307–324. doi: 10.1111/j.1746-1561.2004.tb06620.x. [DOI] [PubMed] [Google Scholar]
  18. Irwin CE, Jr, Burg SJ, Cart CU. America’s adolescents: Where have we been, where are we going? Journal of Adolescent Health. 2002;31(6, Suppl 1):91–121. doi: 10.1016/s1054-139x(02)00489-5. [DOI] [PubMed] [Google Scholar]
  19. Jessor R. Risk behavior in adolescence: a psychosocial framework for understanding and action. Journal of Adolescent Health. 1991;12(8):597–605. doi: 10.1016/1054-139x(91)90007-k. [DOI] [PubMed] [Google Scholar]
  20. Johnston LD, O’Malley PM, Bachman JG, Schulenberg JE. Monitoring the Future national survey results on drug use, 1975–2005: Volume I, Secondary school students (NIH Publication No. 06-5883) Bethesda, MD: National Institute on Drug Abuse; 2006. [Google Scholar]
  21. Johnston LD, O’Malley PM, Bachman JG, Schulenberg JE. Monitoring the Future national results on adolescent drug use: Overview of key findings, 2011. The University of Michigan; Ann Arbor: 2012. [Google Scholar]
  22. Kuehn BM. Teen marijuana use on the rise. Journal of the American Medical Association. 2011;305(3):242. doi: 10.1001/jama.2010.1927. [DOI] [PubMed] [Google Scholar]
  23. Lanza ST, Collins LM, Lemmon DR, Schafer JL. PROC LCA: A SAS procedure for latent class analysis. Structural Equation Modeling. 2007;14(4):671–694. doi: 10.1080/10705510701575602. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lanza ST, Rhoades BL, Greenberg MT, Cox M. Modeling multiple risks during infancy to predict quality of the caregiving environment: Contributions of a person-centered approach. Infant Behavior and Development. 2011;34(3):390–406. doi: 10.1016/j.infbeh.2011.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lubke GH, Muthén B. Investigating Population Heterogeneity with Factor Mixture Models. Psychological Methods. 2005;10(1):21–39. doi: 10.1037/1082-989X.10.1.21. [DOI] [PubMed] [Google Scholar]
  26. Magidson J, Vermunt JK. Latent Class Models for Clustering: A Comparison with K-Means. Canadian Journal of Marketing Research. 2002;20:37–44. [Google Scholar]
  27. McCutcheon A. Latent Class Analysis. Beverly Hills, CA: Sage Publications; 1987. [Google Scholar]
  28. Miller TR, Levy DT, Spicer RS, Taylor DM. Societal Costs of Underage Drinking. Journal of Studies on Alcohol. 2006;67(4):519–528. doi: 10.15288/jsa.2006.67.519. [DOI] [PubMed] [Google Scholar]
  29. Newcomb MD, Scheier LM, Bentler PM. Effects of adolescent drug use on adult mental health: A prospective study of a community sample. In: Marlatt GA, VandenBos GR, editors. Addictive behaviors: Readings on etiology, prevention, and treatment. Washington, DC, US: American Psychological Association; 1997. pp. 169–211. [Google Scholar]
  30. Nylund KL, Asparouhov T, Muthén BO. Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carol Simultation Study. Structural Equation Modeling. 2007;14(4):535–569. [Google Scholar]
  31. Siegel M, Mowery PD, Pechacek TP, Strauss WJ, Schooley MW, Merritt RK, et al. Trends in adult cigarette smoking in California compared with the rest of the United States, 1978–1994. American Journal of Public Health. 2000;90(3):372–379. doi: 10.2105/ajph.90.3.372. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Squeglia L, Jacobus J, Tapert S. The influence of substance use on adolescent brain development. Clinical Electroencephalography and Neuroscience. 2009;40(1):31–38. doi: 10.1177/155005940904000110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Sullivan C, Childs K, O’Connell D. Adolescent Risk Behavior Subgroups: An Empirical Assessment. Journal of Youth and Adolescence. 2010;39(5):541–562. doi: 10.1007/s10964-009-9445-5. [DOI] [PubMed] [Google Scholar]
  34. U.S. Census Bureau. State and County QuickFacts. 2012 Retrieved 02-06-2012 from. [Google Scholar]
  35. USDHHS. The health consequences of smoking: A report of the Surgeon General. Rockville: Centers for Disease Control, Office on Smoking and Health; 2004. [PubMed] [Google Scholar]

RESOURCES