Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Aug 1.
Published in final edited form as: J Child Fam Stud. 2023 Aug 25;32(10):3187–3199. doi: 10.1007/s10826-023-02615-y

Associations between Family Factors and Youth Substance Use Across the Rural-Urban Continuum: A Person-/Variable-Centered Approach

Melissa Pearman Fenton 1, Larry F Forthun 2, Nicolette Corley Grajo 2
PMCID: PMC11293492  NIHMSID: NIHMS1958742  PMID: 39092004

Abstract

Few studies have evaluated the influence of both family factors and geographic location on youth substance use. To address this gap, a person-/variable-centered approach was used to: (1) identify latent profiles of family risk and protective factors for substance use, (2) test profile membership as a predictor of lifetime and 30-day substance use, (3) test rurality, as measured by school geographic location, as a predictor, and (4) explore interaction effects between profile membership and rurality. Youth (N=9,104; 53% female) residing in a state in the southeastern U.S. completed a statewide substance abuse and risk behavior survey including questions about family risk and protective factors and substance use behaviors. Using latent profile analysis to identify subgroups of participants with similar means and variances on the family factors, four latent profiles emerged. Risk of 30-day and lifetime substance use varied across profiles, with the profile characterized by high family-level protective factors and low family-level risk factors indicating the lowest risk for substance use. Urban youth had increased odds of reporting lifetime marijuana use compared to suburban youth; however, geographic location did not appear to confer significantly increased or decreased risk across other substances. No significant interaction results were found. These results emphasize the importance of family functioning on substance use regardless of geographic location, and that evidence-based prevention programming that reduces family risk, strengthens family protection, and is accessible to all types of communities is important to reducing or delaying substance use among youth.

Keywords: Youth, Substance Use, Risk Factors, Protective Factors, Rurality

Introduction

The communities where youth live influence risk of substance use (Cleveland et al., 2008). Evidence suggests that rates of alcohol use and drunk driving are higher among rural youth compared to urban youth, and youth from rural areas are more likely to engage in extreme binge drinking (15 or more drinks in a row) than those from large metropolitan areas (Jiang et al., 2016; Patrick et al., 2013). In contrast, recent research suggests no significant rural and urban differences in lifetime use of alcohol, binge drinking, or heavy drinking among a nationally representative sample of youth (Lenardson et al., 2020). There is evidence of marked differences between rural and urban youth in family risk and protective factors influencing youth substance use indicate that rurality may influence parenting and family management practices (Clark et al., 2011; Jiang et al., 2016; Zhen-Duan & Taylor, 2014). To inform researchers, practitioners, and policymakers of youth populations at risk for substance use that would benefit from evidence-based prevention, more research is needed incorporating granular measures of the rural-urban continuum with a focus on malleable risk and protective factors to inform these efforts. Consistent with Bioecological Systems Theory (Bronfenbrenner & Ceci, 1994), this study explores the interaction of personal-individual factors (youth substance use) with proximal microsystemic processes (interactions with family members) and distal macrosystemic conditions (rurality of school location).

Defining Rurality

Rurality is represented in this study as school locale code. Schools were assigned to their corresponding National Center of Education Statistics (NCES) Locale Codes (Geverdt, 2015). These codes classify the type of area where a school is located and uses population size (for City and Suburban locations) and proximity to metro areas (for Town and Rural locations) to differentiate locales using standard urban and rural definitions developed by the U.S. Census Bureau, core-based statistical areas, and places. Table 1 provides descriptions for each of the NCES Locale Codes and how they were combined to create the four categories of ‘Urban’, ‘Suburban’, ‘Small Town’, and ‘Rural’ that will be used in this study (see also Manly et al., 2019). Most students appear to live geographically near their schools where on average adolescents spent 18 minutes traveling to school, with the most common commute duration being between 5 and 10 minutes (Voulgaris et al., 2019). Thus, these locale codes indicate the general geographic location of the youth attending those schools and provide a more granular measure of the urban-rural continuum compared to an urban-rural dichotomy based on school county.

Table 1.

Location variable creation using NCES Locale Codes

Location NCES Designation (Geverdt, 2015) NCES Locale Codes (Geverdt, 2015) Coded
Urban NCES City designation applies to territory within principal cities of urbanized areas. ‘City, Large’=population of 250,000 or more
‘City, Midsize’=population less than 250,000 and greater than or equal to 100,000
‘City, Small’=population less than 100,000		 1
Suburban NCES Suburban designation applies to territory inside an urbanized area but located outside of the boundary of a principal city of a metropolitan area. ‘Suburban, Large’=population of 250,000 or more
‘Suburban, Midsize’=population less than 250,000 and greater than or equal to 100,000
‘Suburban, Small’=population less than 100,000		 0
Small Town NCES Town designation classifies urban clusters (UC) based on the town’s proximity to an urbanized area (UA). ‘Town, Fringe’=less than or equal to 10 miles from a UA
‘Town, Distant’=more than 10 miles or less than or equal to 35 miles from a UA
‘Town, Remote’=more than 35 miles from a UA		 2
Rural The NCES Rural designation requires the location is in a Census-defined rural territory and differentiates these rural locations based on their proximity to both a UA and a UC. ‘Rural, Fringe’=less than or equal to 5 miles from a UA and is less than or equal to 2.5 miles from a UC
‘Rural, Distant’=more than 5 miles but less than or equal to 25 miles from a UA and more than 2.5 miles but less than or equal to 10 miles from a UC
‘Rural, Remote’=more than 25 miles from a UA and more than 10 miles from a UC		 3
Open in a new tab

Note: There were no ‘Town, Remote’ schools and no ‘Rural, Remote’ schools in this sample, thus the two remaining designations were collapsed into their respective category.

Rurality and Youth Substance Use

Conflicting evidence exists about the prevalence of youth substance use across the rural-urban continuum. Youth living in remote areas may have a higher likelihood of alcohol use than those living in medium-rural or small-urban areas (Swaim & Stanley, 2011). On the other hand, a recent study demonstrated that youth living in rural areas had an increased risk of prescription opioid misuse compared to their urban peers but found no significant differences between the likelihood of youth in small urban and rural areas misusing prescription opioids (Monnat & Rigg, 2016). High school youth living on farms were more likely to report alcohol use, smokeless tobacco use, inhalant use, and other illicit drug use in the past 30-days and binge drinking within the past two weeks compared to youth living in towns (Rhew et al., 2011). Rural youth used substances at the same or lower rates when compared to their urban peers, except lifetime opioid pain reliever misuse, which was higher among rural young adults compared to urban young adults (Lenardson et al., 2020). While the literature supports that rates of alcohol use and prescription opioid misuse generally tended to be higher among youth living in rural areas than youth living in urban areas, there is less agreement on differences in the prevalence of other substance use among rural youth. These mixed findings support the need to continue to examine youth substance use in communities that are representative of the many types of communities in the U.S.

Family Factors, Rurality, and Youth Substance Use

Prior studies comparing family-specific risk and protective factors for substance use between rural and urban youth produced evidence of significant differences. Among African American youth, family and community risk and protective factors were associated with alcohol and marijuana use for those living in rural areas, while individual and peer factors were related to substance use for those in urban settings (Clark et al., 2011). In addition, family structure, supervision, discipline, and attachment mediated the relationship between rural location and youth cigarette and marijuana use (Jiang et al., 2016), with parental monitoring more predictive of lower marijuana use among rural than urban Latino/a youth (Zhen-Duan & Taylor. 2014). In the same study, parental emotional involvement predicted lower alcohol use among rural males, while parental school involvement predicted lower marijuana use among rural females (Zhen-Duan & Taylor. 2014). This literature suggests that, in general, family supervision, disciplinary practices, and family involvement appear to be more strongly predictive of substance use among rural youth than among urban youth (Clark et al., 2011; Jiang et al., 2016; Zhen-Duan & Taylor, 2014).

According to the Social Development Model, which synthesizes concepts from social learning and social bonding theories, family risk emerges from poor social management practices and increased family conflict (Cambron, et al., 2019). Poor family management and conflict discourage youths’ close relationships with parents and other family members, impacting socialization in prosocial commitments, beliefs, and behaviors. Family conflict weakens family bonds, disrupts prosocial socialization processes, and encourages bonding to antisocial peers, while poor family discipline can interfere with prosocial socialization by failing to reinforce behavioral standards within the family. In contrast, protective processes include providing opportunities for youth to engage in prosocial activities and rewarding them for prosocial behaviors, which strengthens bonding to prosocial groups and encourages prosocial beliefs and commitments (Choi et al., 2005; Lonczak et al., 2001). The opportunity to engage in prosocial activities is a socialization pathway through which youth bond to conventional persons and institutions. Likewise, when families offer rewards for prosocial involvement, they strengthen family bonding and reinforce moral behavior.

While the Social Development Model provides a strong theoretical description and explanation of the association between family risk and protective factors and substance use among youth, it remains a linear explanatory model. Linear (variable-centered) models are useful when research seeks to explain causal processes associated with concepts and theoretical propositions (Howard & Hoffman, 2017). However, variable-centered models are limited in that they fail to acknowledge the complex interrelationships between variables at the microsystemic level. A person-centered analysis can alleviate some of these concerns (Howard & Hoffman, 2017; Lanza & Cooper, 2016). More specifically, a typology of family risk and protection along the dimensions of family management, conflict/hostility, and prosocial engagement may provide greater insight into family system-level processes that may be expressed in unique ways among youth in different geographic contexts. Like research into parenting styles (Baumrind, 2013; Maccoby & Martin, 1983), individual differences in family risk and protection may help identify configurations of family characteristics most associated with substance use among adolescents while, at the same time, differentiate configurations of family risk and protection that may benefit sub-populations based on rurality and demographic differences.

Current Study

While person-centered analytical approaches have become increasingly popular (Howard & Hoffman, 2017; Lanza & Cooper, 2016; Tomczyk et al., 2016), most contemporary research drawing concepts from the Social Development Model have examined substance use profiles rather than family risk and protective profiles (Cleveland et al., 2010; Kulis et al., 2016; Lamont et al., 2014; Swaim & Stanley, 2021; Wu et al., 2020). This study aims to take a person-centered approach to examining the family microsystem and answer the following research questions: (1) Within a statewide sample of middle and high school students, what family profiles emerge from measures of family risk and protective factors? (2) Do the family profiles predict odds of current and lifetime substance use among youth? (3) Does rurality predict odds of current and lifetime youth substance use? (4) Do family risk and protective factor profiles interact with rurality in predicting odds of substance use among youth? In general, results were expected to match previous studies where youth living in rural areas reported increased odds of alcohol use and prescription drug misuse. Due to the nature of latent profile analysis, predictions were not made about the number of groups or the pattern of results. However, it was anticipated that youth in profiles with more risk factors and fewer protective factors would have increased odds of reporting substance use. Likewise, family profiles characterized by high protection (prosocial engagement) and low risk (poor supervision, poor discipline, and conflict) were hypothesized to be more strongly related to rural youth’s alcohol use and prescription drug misuse compared to family profiles characterized by high risk and low protection.

Method

Data are from the 2019 Florida Youth Substance Abuse Survey (FYSAS), an anonymous, in-school survey administered annually by the Florida Department of Children and Families (DCF). The FYSAS is based on the Communities that Care (CTC) Youth Survey (Arthur et al., 2002). It assesses the prevalence of substance use and risk behaviors and identifies risk and protective factors that exist within the community, family, school, peer, and individual environments. A stratified, two-stage cluster sampling strategy was used to obtain a representative sample of youth attending public middle and high schools in Florida. In the first stage, a random sample of 93 middle schools and 82 high schools were selected from all public schools in Florida to take part in the survey. In the second stage, classrooms in each school were randomly selected to participate in the survey. Overall, within the schools that agreed to participate, 74.6% of middle school students and 67.8% of high school students participated (Florida DCF, 2019). The University Institutional Review Board determined this secondary analysis was exempt from review.

Participants

Among the 9,819 Florida youth who participated in the survey administration, 715 were removed for missing data for a final sample size of 9,104 Florida youth between the ages of 12 and 17 (88%). This study had slightly more female (n=4,834; 53%) than male (n=4,270; 47%) youth. Florida youth identified themselves as White/Caucasian (43%), Spanish/Hispanic/Latinx (38%), Black/African American (21%), Other (6%), Native American (5%), Asian (4%), and Native Hawaiian/Pacific Islander (1%). Over half of the participants (53%) were in high school (grades 9–12). Participants mainly were from suburban areas (54%), with participants also hailing from urban areas (21%), rural areas (17%), and towns (8%).

CTC Youth Survey Protective Factors

Family Opportunities for Prosocial Involvement

This family-level protective factor was assessed using three statements asking about opportunities to a) participate in family decision-making, b) do fun activities with parent(s), and c) ask parent(s) for help (Arthur et al., 2002). For example, “If I had a personal problem, I could ask my mom or dad for help.” Items were scored on a 4-point Likert-type scale (0–3), with higher scores indicating more opportunities for prosocial involvement. The scale had acceptable reliability (α=.77).

Family Rewards for Prosocial Involvement

This family-level protective factor was assessed using four statements asking about receiving praise and recognition from their parent(s) and enjoying time spent with their parent(s) (Arthur et al., 2002). For example, “How often do your parents tell you they’re proud of you for something you’ve done?” Items were scored on a 4-point Likert-type scale (0–3), with higher scores indicating more rewards for prosocial involvement. The scale had acceptable reliability (α=.77).

CTC Youth Survey Risk Factors

Family Conflict

Family conflict was assessed using three statements about the frequency and severity of arguments in the family (Arthur et al., 2002). For example, “People in my family often insult or yell at each other.” Items were scored on a 4-point Likert-type scale (0–3), with higher scores indicating higher levels of family conflict. The scale had acceptable reliability (α=.78).

Poor Family Supervision

Poor family supervision was measured by five items on youth’s perceptions of “parent supervision of homework, curfew, and whereabouts, and family rules” (Guttmannova et al., 2017, p. 352). For example, “Would your parents know if you did not come home on time?” Items were scored on a 4-point Likert-type scale (0–3) and then reverse scored so that higher scores indicate lower levels of family supervision. The scale had acceptable reliability (α=.74).

Poor Family Discipline

This family-level risk factor was measured by three items on youth’s perceptions of “consequences for their alcohol use, skipping school, or carrying a handgun” (Guttmannova et al., 2017, p. 352). For example, “If you drank some beer, wine, or liquor without your parents’ permission, would you be caught by your parents?” Items were scored on a 4-point Likert-type scale (0–3) and then reverse scored so that higher scores indicate lower levels of family discipline. The scale had high reliability (α=.81).

Substance Use Outcomes

Alcohol use, marijuana use, prescription drug misuse, and illegal drug use were each measured as lifetime and past 30-day use. Illegal drug use included the use of methamphetamine, Ecstasy, GHB, Rohypnol, ketamine, LSD, PCP, hallucinogenic mushrooms, cocaine, crack, and heroin. Prescription drug misuse included using depressants, tranquilizers, prescription pain relievers, and amphetamines without a doctor’s orders and using over-the-counter medications to get high. Participants were asked to report the number of occasions they had used each substance in their lifetime and in the past 30 days with response options in intervals ranging from 0 to 40 or more occasions. Each substance use variable was dichotomously coded where 0 occasions were coded as ‘0’ and between 1 and 40 occasions were coded as ‘1’. Dichotomization was employed due to research questions seeking to predict odds, not frequency, of youth substance use (MacCallum et al., 2002).

Demographic Items

Age was a categorical control variable with ten possible response options ranging from ‘10’ to ‘19 or older’. Participants were asked to report their grade level ranging from ‘6th’ through ‘12th’. Grades 6–8 were coded as middle school (0) and grades 9–12 were coded as high school (1). Sex was a control variable coded female (0) and male (1). Participants were also asked to describe themselves by choosing one or more of the following responses: ‘American Indian/Native American or Alaska Native’, ‘Asian’, Black/African American’, ‘Spanish/Hispanic/Latino’, ‘Native Hawaiian or other Pacific Islander’, ‘White/Caucasian’, or ‘Other’. For inclusion in logistic regression analysis, race was dichotomously coded so that ‘1’ indicated the identification of each racial and ethnic group in relation to a ‘0’ reference group composed of all other ethnic/racial categories.

Rurality

Rurality is measured by NCES Locale Code for the school location participants attended as ‘Urban’, ‘Suburban’, ‘Small Town’, and ‘Rural’.

Data Analysis

A latent profile analysis (LPA) was used to describe groups of individuals according to their patterns of family risk and protective factors. LPA identifies these latent groups using conditional means and variances of unobservable characteristics represented by continuous variables (Nylund-Gibson & Choi, 2018). Mplus v8 (Muthén & Muthén, 2017) was used for the LPA analysis. A series of seven models were estimated to determine the appropriate number of profiles to describe the subgroups of family risk and protective factors experienced by youth, beginning with a one-profile model with no covariates followed by models specifying an increased number of profiles (e.g., two, three, and four). The MIXTURE COMPLEX syntax was used to account for data being nested within schools, and school identification number was used as the cluster variable (Muthén & Muthén, 2017). Model selection was based on recommended fit indices, including a reduction in the Akaike Information Criterion (AIC) and Adjusted Bayesian Information Criterion (BIC) compared to less complex (K-1) models and significant Lo-Mendell-Rubin Likelihood Ratio Test (LMR LRT) (Masyn, 2013). The Bootstrap Likelihood Ratio Test (BLRT) is unavailable when using complex survey data (Muthén, 2016). Additionally, the researchers considered models that (a) produced the most parsimonious solution without exceeding the smallest profile size of less than 5% of cases and (b) held theoretical support (Masyn, 2013). Missing data were handled with maximum likelihood in the LPA.

After selecting the optimal model, profile membership was exported as a separate variable from Mplus using posterior probabilities. Per Nagin’s (2005) diagnostic guidelines, the selected model had higher than the recommended 0.7 average posterior probabilities (ranging from 0.889 to 0.946). Further, the average posterior and classification probabilities were almost identical, indicating an appropriate model fit. Entropy of the final model was 0.865, above the 0.8 recommended cut-off for using profile membership as a variable for further analysis (Muthén, 2008). Crosstabulations and chi-square tests of association were conducted to describe participants in each profile. Descriptive statistics for all variables included in the statistical models, bivariate correlations, and point-biserial correlations of family-level risk, protective factors, and substance use were calculated. Rates of missingness for independent, control, and dependent variables ranged from 1–10%. To address higher missingness in some variables, 10 multiple imputations were conducted in IBM SPSS Statistics 27 for all missing independent, control, and dependent variables to reduce bias in the logistic regression analysis (Schafer & Graham, 2002).

Profile membership was used as an independent variable in a set of hierarchical logistic regressions to analyze the association between covariates, profiles, and rurality on each substance use outcome. The final model included interaction terms (latent profile of family risk and protective factors*rurality) to investigate if rurality moderated the relationship between family risk and protection profiles and substance use. To control for Type I errors, p values of <0.001 for both main and interaction effects were considered statistically significant (Forster et al., 2020). Pooled results from analyses using the multiple imputations in IBM SPSS Statistics 27 are reported.

Results

Descriptive Statistics

Most youth did not report using substances. The most prevalent substance used was alcohol, with 37% of youth reporting having used alcohol in their lifetimes and 15% reporting having used alcohol in the last 30 days. Second in prevalence was marijuana use, with 19% of participants reporting lifetime use and 10% reporting past 30-day use. Ten percent of youth reported lifetime prescription drug misuse, and 4% reported 30-day prescription drug misuse. Four percent of youth reported lifetime illegal drug use, and less than 2% reported using illegal drugs in the last 30 days. Mean scores on continuous variables and bivariate and point-biserial correlations of study variables are presented in Table 2. Of note, higher levels of family opportunities for prosocial involvement and rewards for prosocial involvement were negatively related to poor discipline, poor supervision, family conflict, and all substance use outcomes. Higher levels of poor discipline, poor supervision, and family conflict were positively related to all substance use outcomes. Poor discipline and family conflict were uncorrelated.

Table 2.

Descriptive statistics and bivariate and point-biserial correlations for family risk and protective factors and substance use (N=9104)

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
1. Opportunities 1.00
2. Rewards 0.77* 1.00
3. Discipline −0.35* −0.36* 1.00
4. Supervision −0.58* −0.58* 0.67* 1.00
5. Conflict −0.33* −0.31* −0.00 0.06* 1.00
6. LT Marijuana −0.15* −0.15* 0.23* 0.17* 0.16* 1.00
7. 30-day Marijuana −0.11* −0.11* 0.19* 0.14* 0.11* 0.67* 1.00
8. LT Alcohol −0.15* −0.15* 0.17* 0.13* 0.23* 0.47* 0.33* 1.00
9. 30-Day Alcohol −0.12* −0.12* 0.17* 0.14* 0.16* 0.38* 0.36* 0.50* 1.00
10. LT Prescription −0.15* −0.14* 0.13* 0.13* 0.16* 0.32* 0.31* 0.26* 0.26* 1.00
11. 30-Day Prescription −0.11* −0.10* 0.10* 0.12* 0.10* 0.19* 0.21* 0.15* 0.20* 0.58* 1.00
12. LT Illegal −0.08* −0.10* 0.14* 0.12* 0.07* 0.32* 0.36* 0.19* 0.23* 0.35* 0.27* 1.00
13. 30-Day Illegal −0.06* −0.07* 0.10* 0.10* 0.03 0.16* 0.20* 0.09* 0.14* 0.20* 0.23* 0.58* 1.00
Mean (SD) 1.97 (0.80) 2.00 (0.76) 0.94 (0.93) 0.72 (0.64) 1.14 (0.82)
Range 0–3 0–3 0–3 0–3 0–3
Open in a new tab

Opportunities=Family Opportunities for Prosocial Involvement, Rewards=Family Rewards for Prosocial Involvement, Discipline=Poor Family Discipline, Supervision=Poor Family Supervision, LT=lifetime for all substance use outcomes listed

*

p<0.001

Results of the LPA

The results of the LPA are shown in Table 3. A four-profile model was the best overall fit for the five-family level risk and protective factor indicators. The AIC, BIC, and SABIC continued to decline as the number of profiles increased; however, diminishing gains in model fit were observed after the four-profile solution. The four-profile model had a significantly better LMR than a three-profile solution. Models with increasing solutions each had significantly better LMR; however, these models included a profile with less than 5% of the participants. The four-profile solution’s smallest class size was 5% and showed an additional distinct profile that was masked in the three-profile solution (the permissive family management profile). The profiles also mapped onto the same number of profiles as Baumrind’s (2013) parenting typology, indicating theoretical support for this model (Kuppens & Ceulemans, 2019). Thus, the preponderance of statistical and theoretical evidence supported a four-profile solution as the most parsimonious model.

Table 3.

Fit statistic comparisons of latent profile analysis models of family-level risk and protective factors among

Model Log Likelihood AIC BIC SABIC Entropy Smallest Class % LMR p-value LMR Meaning
1 −52432.82 104885.64 104956.80 104925.03
2 −47252.78 94537.55 94651.42 94600.57 0.79 32.74% <.001 2>1
3 −44714.62 89473.24 89629.81 89559.89 0.85 5.19% <.001 3>2
4 −42891.03 85838.06 86037.32 85948.34 0.87 5.00% <.001 4>3
5 −41605.39 83278.77 83520.73 83412.69 0.85 3.52% <.001 5>4
6 −41112.72 82305.44 82590.10 82462.98 0.85 3.36% 0.004 6>5
7 −40667.37 81426.75 81754.10 81607.92 0.82 3.21% 0.024 7>6
Open in a new tab

Florida youth

Note. N=9104; AIC=Akaike’s Information Criterion; BIC=Bayesian Information Criterion; SABIC=Sample-Adjusted BIC; LMR=Lo-Mendell Ruben.

Profiles were conceptualized as: prosocial family management, permissive family management, uninvolved family management, and high family conflict. Figure 1 provides the mean scores for each profile on the five-family risk and protective factor scales. The largest profile was prosocial family management (n=4,495; 49%), including youth who reported the highest mean scores in prosocial opportunities and rewards (protective factors) and low mean scores in poor supervision, poor discipline, and family conflict (risk factors). This profile appears similar to the authoritative parenting style characterized by providing firm behavioral control, being responsive to children, and supporting autonomy in decision-making (Baumrind, 1991, 2013). The high family conflict profile (n=3,268; 36%) reported limited prosocial opportunities and rewards, the second lowest mean score in poor discipline and poor supervision, and the highest mean score in family conflict. This profile seems most closely aligned with the authoritarian profile, with the high levels of behavioral control but limited responsiveness, which may manifest in conflict (Baumrind, 1991, 2013). The permissive family management profile (n=885; 10%) reported consistent prosocial opportunities and rewards at the second highest mean scores, high mean scores in poor supervision and poor discipline, and a low mean score in family conflict. The permissive family management profile shared similarities with the permissive prototype, characterized by a lack of direction and behavioral control from parents but responsive, accepting, and supporting autonomy (Baumrind, 1991, 2013). The uninvolved family management profile (n=456; 5%) reported the lowest mean scores in prosocial opportunities and rewards, the highest mean scores in poor discipline and poor supervision, and a low mean score in family conflict. This profile most closely mirrored the disengaged parents who provided limited behavioral control and were nonresponsive to their children (Baumrind, 1991, 2013).

Figure 1.

Figure 1.

Open in a new tab

Mean scores on family-level risk and protective factors for the four-profile model

Crosstabulations and chi-square tests of association revealed that youth members within each profile were significantly different in rurality, sex, and grade. Table 4 presents the counts and frequencies of profile membership for rurality, sex, and grade. More females were in the high family conflict profile than males, while more males were in the uninvolved and permissive family management profiles than females. A smaller proportion of youth from towns were in the high family conflict profile compared to youth from other locations, and a larger proportion of youth from towns were in the prosocial family management profile than youth from other locales. Nearly equivalent proportions of urban, suburban, and rural youth were in the prosocial family management profile. High school youth were in the prosocial family management and high family conflict profiles in almost equal proportions.

Table 4.

Characteristics of family risk and protective factor profiles

Variable Permissive Family Management (n=885) Uninvolved Family Management (n=456) Prosocial Family Management (n=4495) High Family Conflict (n=3268) Chi square tests of association
Rurality
Urban 201 (10%) 109 (6%) 903 (47%) 725 (37%) χ2(9)=31.82*
φ=0.06*
Suburban 476 (10%) 247 (5%) 2414 (49%) 1783 (36%)
Towns 75 (10%) 37 (5%) 405 (57%) 201 (28%)
Rural Areas 133 (9%) 63 (4%) 774 (51%) 559 (36%)
Sex
Female 357 (7%) 180 (4%) 2394 (50%) 1904 (39%) χ2(3)=126.78*
φ=0.12*
Male 528 (12%) 276 (7%) 2101 (49%) 1364 (32%)
Grade
Middle School 385 (9%) 172 (4%) 2444 (57%) 1274 (30%) χ2(3)=201.87*
φ=0.15*
High School 500 (10%) 284 (6%) 2052 (43%) 1994 (41%)
Open in a new tab
*

p<0.001.

φ=effect size (phi coefficient).

Associations between Family Risk and Protection Profiles and Substance Use Outcomes

Results of hierarchical logistic regression models for each of the substance use outcomes are reported in Table 5. The modal category of prosocial family management was used as the reference category in all logistic regression models. When compared to the prosocial family management profile, youth in the permissive family management profile and high family conflict profile had significantly increased odds of reporting lifetime marijuana use, past 30-day marijuana use, lifetime alcohol use, past 30-day alcohol use, lifetime prescription drug misuse, past 30-day prescription drug misuse, lifetime illegal drug use, and past 30-day illegal drug use. For example, youth in the permissive family management profile had 133% increased odds of reporting using alcohol in the past 30 days and 153% increased odds of reporting lifetime prescription drug misuse compared to the youth in the prosocial family management profile. Youth in the high family conflict profile had 158% increased odds of reporting using alcohol in the past 30 days and 206% increased odds of reporting lifetime prescription drug misuse. These findings are strong and consistent across substances. Youth in the uninvolved family management profile reported significantly increased odds of using all substances in their lifetimes and within the past 30 days except lifetime alcohol use. For instance, youth in the uninvolved family management profile had the highest increased odds of reporting misusing prescription drugs in the last 30 days and 211% increased odds of reporting using marijuana in the last 30 days.

Table 5.

Odds ratios for the association between family risk and protection profile membership and rurality and substance use

Lifetime Marijuana Use 30-Day Marijuana Use Lifetime Alcohol Use 30-Day Alcohol Use Lifetime Prescription Drug Misuse 30-Day Prescription Drug Misuse Lifetime Illegal Drug Use 30-Day Illegal Drug Use
Family Profile
 Permissive 2.65(2.18–3.22)* 3.05(2.40–3.89)* 1.53(1.30–1.80)* 2.33(1.89–2.87)* 2.53(1.94–3.30)* 2.89(1.92–4.36)* 3.88(2.71–5.55)* 5.27(3.20–8.70)*
 Uninvolved 2.21(1.71–2.87)* 3.11(2.28–4.23)* 1.16(0.92–1.44) 2.35(1.78–3.08)* 3.62(2.62–5.00)* 6.39(4.20–9.70)* 4.84(3.09–7.59)* 5.78(2.87–11.61)*
 Conflict 2.83(2.49–3.22)* 2.93(2.46–3.49)* 2.40(2.18–2.66)* 2.58(2.25–2.96)* 3.06(2.57–3.06)* 2.98(2.25–3.95)* 2.92(2.19–3.90)* 2.72(1.75–4.26)*
Rurality
 Urban 1.28(1.12–1.48)* 1.23(1.03–1.48) 1.05(0.94–1.18) 0.98(0.83–1.15) 0.97(0.80–1.18) 0.87(0.84–1.17) 0.89(0.66–1.18) 0.93(0.58–1.48)
 Rural Areas 1.20(1.03–1.41) 1.04(0.85–1.28) 1.04(0.92–1.18) 1.05(0.89–1.25) 1.30(1.07–1.57) 1.11(0.81–1.53) 1.21(0.91–21.61) 1.27(1.80–2.02)
 Town 1.19(0.92–1.53) 1.39(1.03–1.88) 1.17(0.98–1.41) 1.34(1.06–1.71) 1.29(0.97–1.72) 0.97(0.61–1.53) 1.61(1.06–2.44) 1.93(1.09–3.43)
Open in a new tab

Cells are odds ratios and 95% confidence intervals (CI). All models adjusted for age, gender, and race.

Reference category for family profile is Prosocial Family Management. Reference category for location is Suburban.

*

p<.001

Associations between Rurality and Substance Use Outcomes

Results for associations between rurality and substance use are reported in Table 5. Suburban school location was used as the reference category as this was the largest school locale for participants. Only one association between rurality and substance use outcomes was statistically significant. Youth in urban areas had significantly increased odds (28%) of reporting using marijuana in their lifetimes compared to suburban youth. In an association that approached significance, youth in rural areas had increased odds of reporting lifetime prescription drug misuse compared to suburban youth (OR: 1.30, CI: 1.07–1.57, p=0.007). All other associations between rurality, as measured by school geographic location, and substance use were non-significant.

Family-Risk and Protection Profiles Interaction with Rurality to Predict Substance Use Outcomes

As planned, potential interaction effects between family risk and protection latent profiles and rurality were tested to determine if they predicted current and past youth substance use. No significant interaction effects were found (see supplementary materials for odds ratios and confidence intervals for all interactions tested).

Discussion

This study explored several research questions to better understand the role of family risk and protective factors in substance use among youth living in an urban, suburban, small town, and rural location in the southeastern U.S. While research examining family risk/protection, substance use, and other problem behaviors among youth is well-documented (Arthur et al., 2007; Cleveland et al., 2008), there remains a need to examine how family factors may interact with geographic location to promote or protect against youth substance use. Using statewide data from an annual youth substance use survey (Florida DCF, 2019), family risk and protective factor profiles were created using a latent profile analytic procedure. Results suggested that a four-profile solution was the most parsimonious, with profile characteristics representing (1) high protection and low risk (prosocial family management), (2) low protection and high risk, except for family conflict (uninvolved family management), (3) high protection and high risk, except for family conflict (permissive family management), and (4) moderate protection, moderate risk, and high conflict (high family conflict). Higher levels of opportunities and rewards for prosocial involvement characterize greater protection, while higher levels of conflict reflect greater risk. Since poor supervision and poor discipline were measured as risk factors, lower levels indicate greater parental supervision and discipline within the family and lower risk.

Accepting the theoretical justification of the influence of family risk and protective factors on substance use as contained in the Social Development Model while incorporating a person-centered analysis of the family microsystem consistent with the bioecological model, analysis of the associations among family profiles and substance use across all youth in the sample resulted in several novel findings. Compared to youth from families that offered opportunities and rewards for prosocial involvement, supervision and discipline, and limited family conflict (prosocial family management profile), youth within all other family risk and protection profiles had significantly higher odds of 30-day alcohol use and both current and lifetime marijuana, prescription drug, and illegal drug use. Thus, the combination of high protection and low risk in the prosocial family management profile represents the ideal combination of factors that have been shown in past research to be associated with lower rates of reported substance use (Baumrind, 1991; Cleveland et al., 2008; Hawkins et al., 1992). However, youth within the uninvolved family management profile did not have significantly increased odds of lifetime alcohol use, the only nonsignificant odds ratio found between family management profile membership and substance use. Despite reporting the highest levels of poor supervision and poor discipline and the lowest levels of prosocial opportunities and rewards, these youth are not significantly more likely to report using alcohol in their lifetime than youth in the profile representing greatest protection and lowest risk (prosocial family management). However, they have higher odds of reporting using all other substances, including past 30-day alcohol use. This finding contradicts previous research (Baumrind, 1991) and calls into question what, if any, protection the unique combination of risk and protective factors represented by the uninvolved family management profile may offer youth against lifetime alcohol use.

An additional question for this study was whether the family management profiles may interact with rurality, operationalized as school geographic location, to predict substance use among youth. When exploring the association, a more fine-grained measure of location was used (e.g., urban, suburban, small town, and rural), compared to an urban-rural dichotomy, to increase sensitivity to geographic differences. For substance use, urban youth had significantly increased odds of lifetime marijuana use compared to suburban youth. This trend is consistent with the results of the most recent Monitoring the Future survey, which found that rates of marijuana use were slightly lower in non-metropolitan areas compared to metropolitan areas of the U.S. (Johnston et al., 2020). However, no significant associations existed between other geographic locations and substance use. This is surprising given previous research that found relatively consistent urban-rural differences for some substances. For example, youth from small towns and/or rural areas in the U.S. have been shown to be at particularly high risk for prescription drug misuse given the ease of access to prescription medications, paucity of behavioral health specialists, and few opportunities for prosocial activities (Andrilla et al., 2018; Dew et al., 2007; Monnat & Rigg, 2016; Pulver et al., 2015). It may be that the more sensitive measurement of geographic location employed in this study dispersed the more extreme scores more evenly across groups, lowering sensitivity to group differences. Dichotomizing urban-rural environments could draw sharper distinctions between social phenomena that may be more nuanced (Bennett et al., 2019). Other research that has documented differences in substance use across the rural-urban continuum used geographic measures (e.g. county) of their home instead of their school location (Gfroerer et al., 2007), which may be another explanation of these results. On the other hand, the results may indicate a genuine lack of variation across the urban-rural continuum in the U.S. state under investigation, implying that youth substance use may no longer differ as significantly as it once did across geographic locations. A recent study using nationally representative data and an urban-rural dichotomy found no significant differences among urban and rural youth’s substance use and misuse supporting this conclusion (Lenardson et al., 2020). While one of the tests for geographic differences approached significance, it does not support an overarching conclusion that youth in this state’s small towns and rural areas are inherently more at risk for substance use.

Implications

The person-centered analysis employed in this study identified family profiles with indicators drawn from the Social Development Model. However, like other person-centered parenting models (Baumrind, 1991, 2013), the unique indicators clustered into four profiles differentiated primarily by family prosocial behaviors and family conflict. While it is beyond the scope of this study to empirically test similarities between the parenting styles proposed by Baumrind (2013) and the family risk/protective profiles drawn from the Social Development Model (Cambron et al., 2019), the statistical solution points to a latent theoretical structure that is worth exploring. When evaluating the family microsystem as a constellation of uniform characteristics, the universal benefit of the prosocial family management profile, which would be labeled as authoritative parenting by Baumrind (2013), suggests that all types of communities would benefit from family-based substance use prevention programs that improve prosocial involvement, discipline, and supervision and lower family conflict. Indeed, these findings suggest that the parenting profiles could be viewed as a universal protective influence, supporting a need to improve the reach of and access to family-based prevention efforts in all communities.

This study did not find the expected interactions between the family management profiles and rurality, again suggesting some universal latent family protective structure. However, it does not change the research evidence that rural areas experience disparities in access to evidence-based prevention (EBP) programs (Substance Abuse & Mental Health Services Administration, 2007). Indeed, youth self-reported participation in school-based substance use prevention programs declined from 2002 to 2016 among youth in rural areas by a 22% proportional decline (Salas-Wright et al., 2019). A lack of practitioners, funding, infrastructure, and facilities to support EBP delivery and reduced cost-effectiveness due to lower population numbers contribute to these disparities (Gregg, 2012). Investments in online delivery of family-based prevention, increasing the number of trained practitioners, and providing adequate funding to offer these programs to geographically dispersed families would help prevent youth substance use in small towns and rural areas. Further, parents would benefit from increased community opportunities for children to be involved in structured and supervised activities, as well as information that teaches parents how to involve their children in family decisions and activities and reward and recognize children when they are involved to encourage prosocial involvement.

Strengths and Limitations

The strengths of this study include the use of a large, representative sample of youth between the ages of 10 and 19 from a state in the southeastern U.S. The state youth survey was adapted from the Communities That Care (CTC) Survey, which supports the validity of the constructs and accessibility of findings to other users of the CTC framework (Arthur et al., 2002; 2007). In addition, participants were coded with a more precise measure of rurality to better understand the interaction between family management profiles and location. However, the findings of this study should be evaluated with the following limitations. While the sample size was large and youth were distributed across family management profiles, the prevalence of use was small for some substances (e.g., illegal drugs), which may have limited the power to detect interactions between binary variables. As a result, confidence intervals for low prevalence substance use variables should be interpreted with caution. In addition, substance use tends to be underreported due to social desirability bias (Latkin et al., 2017). Substance use variables were dichotomized to compensate for skewed distributions and ease interpretation of interaction terms; however, understanding the frequency of use is also essential for prevention and intervention. Additionally, the cross-sectional nature of the study prevents causal inference. Following the Social Development Model, family factors, including conflict, poor supervision, and poor discipline, were conceptualized as promoting the risk of substance use; however, it is also possible that these factors are consequences of youth substance use. Models could not account for potential shared variance with siblings due to limitations in the data about familial relationships between participants. Moreover, school geographic location could differ from home geographic location, an additional limitation of this study’s findings. The exploration of demographic differences in type and frequency of substance use and profile membership beyond geographic location exceeds the scope of this study; however, future studies would benefit from investigating other possible differences based on characteristics such as age, race, and ethnicity.

Conclusion

To address the persistent public health problem of youth substance use, research must continue to explore how family risk and protection interact with all ecological systems to contribute to the use of substances by youth. In this study, a family profile with high protection and low risk consistently had decreased odds of substance use compared to those with moderate to high risk. These findings suggest that improving the family’s ability to provide opportunities and rewards for prosocial involvement, adequate supervision, appropriate discipline, and reducing family conflict are all necessary components of family-based substance use prevention efforts. The lack of significant associations between rurality and substance use indicates the prevention of youth substance use should be prioritized in all types of communities. Increasing investments in family-based prevention programs that specifically address the challenges of reaching diverse and geographically dispersed families are needed as part of a holistic approach to reducing youth substance use.

Supplementary Material

Supplementary Table

Highlights:

  • Families supporting prosocial behavior, appropriately supervising, and limiting conflict protected youth from substance use

  • School geographic location was not a significant predictor of odds of substance use among youth

  • Families and youth in all communities would benefit from access to family-based prevention to reduce substance use

References

  1. Andrilla CHA, Patterson DG, Garberson LA, Coulthard C, & Larson EH (2018). Geographic variation in the supply of selected behavioral health providers. American Journal of Preventive Medicine, 54(6), S3, S199–S207. 10.1016/j.amepre.2018.01.004 [DOI] [PubMed] [Google Scholar]
  2. Arthur MW, Riney JS, Hawkins JD, Abbott RD, Brooke-Weiss BL, & Catalano RF (2007). Measuring risk and protection in communities using the Communities That Care Youth Survey. Evaluation and Program Planning, 30(2), 197–211. 10.1016/j.evalprogplan.2007.01.009 [DOI] [PubMed] [Google Scholar]
  3. Arthur MW, Hawkins JD, Pollard JA, Catalano RF, & Baglioni AJ (2002). Measuring risk and protective factors for substance use, delinquency, and other adolescent problem behaviors. The Communities That Care Youth Survey. Evaluation Review, 26(6), 575–601. 10.1177/0193841X0202600601 [DOI] [PubMed] [Google Scholar]
  4. Baumrind D (1991). The influence of parenting style on adolescent competence and substance use. Journal of Early Adolescence, 11(1), 56–95. [Google Scholar]
  5. Baumrind D (2013). Authoritative parenting revisited: History and current status. In Larzelere RE, Morris AS, & Harrist AW (Eds.), Authoritative parenting: Synthesizing nurturance and discipline for optimal child development. (pp. 11–34). American Psychological Association. 10.1037/13948-002 [DOI] [Google Scholar]
  6. Bennett KJ, Borders TF, Holmes GM, Kozhimannil KB, & Ziller E (2019). What is rural? Challenges and implications of definitions that inadequately encompass rural people and places. Health Affairs, 38(12), 1985–1992. 10.1377/hlthaff.2019.00910 [DOI] [PubMed] [Google Scholar]
  7. Bronfenbrenner U, & Ceci SJ (1994). Nature-nurture reconceptualized in developmental perspective: A bioecological model. Psychological Review, 101(4), 568–586. 10.1037/0033-295X.101.4.568 [DOI] [PubMed] [Google Scholar]
  8. Cambron C, Catalano RF, & Hawkins JD (2019). The social development model. In Farrington DP, Kazemian L, & Piquero AR (Eds.), The Oxford handbook of developmental and life-course criminology (pp. 1–30). Oxford University Press. 10.1093/oxfordhb/9780190201371.001.0001 [DOI] [Google Scholar]
  9. Choi Y, Harachi TW, Gillmore MR, & Catalano RF (2005). Applicability of the social development model to urban ethnic minority youth: Examining the relationship between external constraints, family socialization, and problem behaviors. Journal of Research on Adolescence, pp. 15,–505–534. 10.1111/j.1532-7795.2005.00109.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Clark TT, Nguyen AB, & Belgrave FZ (2011). Risk and protective factors for alcohol and marijuana use among African-American rural and urban adolescents. Journal of Child and Adolescent Substance Abuse, 20(3), 205–220. 10.1080/1067828X.2011.581898 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cleveland MJ, Collins LM, Lanza ST, Greenberg MT, & Feinberg ME (2010). 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, 38(3), 213–228. 10.1080/10852352.2010.486299 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cleveland MJ, Feinberg ME, Bontempo DE, & Greenberg MT (2008). The role of risk and protective factors in substance use across adolescence. Journal of Adolescent Health, 43(2), 157–164. 10.1016/j.jadohealth.2008.01.015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dew B, Elifson K, & Dozier M (2007). Social and environmental factors and their influence on drug use vulnerability and resiliency in rural populations. Journal of Rural Health, 23, 16–21. 10.1111/j.1748-0361.2007.00119.x [DOI] [PubMed] [Google Scholar]
  14. Florida Department of Children and Families. (2019). 2019 Florida Youth Substance Abuse Survey state report. https://www.myflfamilies.com/service-programs/samh/prevention/fysas/2019/docs/2019%20FYSAS%20State%20Report%20(Final).pdf
  15. Forster M, Grigsby TJ, Gower AL, Mehus CJ, & McMorris BJ (2020). The role of social support in the association between childhood adversity and adolescent self-injury and suicide: Findings from a statewide sample of high school students. Journal of Youth and Adolescence, 49(6), 1195–1208. 10.1007/s10964-020-01235-9 [DOI] [PubMed] [Google Scholar]
  16. Geverdt D (2015). Education demographic and geographic estimates program (EDGE): Locale boundaries user’s manual. National Center for Education Statistics. https://nces.ed.gov/programs/EDGE/docs/NCES_LOCALE_USERSMANUAL_2016012.pdf [Google Scholar]
  17. Gfroerer JC, Larson SL, & Colliver JD (2007). Drug use patterns and trends in rural communities. The Journal of Rural Health, 23(s1), 10–15. 10.1111/j.1748-0361.2007.00118.x [DOI] [PubMed] [Google Scholar]
  18. Gregg JA (2012). Rural adolescent substance abuse: Prevention implications from the evidence. Online Journal of Rural Nursing and Health Care, 12(2), 41–50. 10.14574/ojrnhc.v12i2.21 [DOI] [Google Scholar]
  19. Guttmannova K, Wheeler MJ, Hill KG, Evans-Campbell TA, Hartigan LA, Jones TM, Hawkins JD, & Catalano RF (2017). Assessment of risk and protection in Native American youth: Steps toward conducting culturally relevant, sustainable prevention in Indian Country. Journal of Community Psychology, 45(3), 346–362. 10.1002/jcop.21852 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hawkins JD, Catalano RF, & Miller JY (1992). Risk and protective factors for alcohol and other drug problems in adolescence and early adulthood: Implications for substance abuse prevention. Psychological Bulletin, 112(1), 64–105. 10.1037/0033-2909.112.1.64 [DOI] [PubMed] [Google Scholar]
  21. Howard MC, & Hoffman ME (2018). Variable-centered, person-centered, and person-specific approaches: Where theory meets the method. Organizational Research Methods, 21(4), 846–876. 10.1177/1094428117744021 [DOI] [Google Scholar]
  22. Jiang G, Sun F, & Marsiglia FF (2016). Rural-urban disparities in adolescent risky behaviors: A family social capital perspective. Journal of Community Psychology, 44(8), 1027–1039. 10.1002/jcop.21825 [DOI] [Google Scholar]
  23. Johnston LD, Miech RA, O’Malley PM, Bachman JG, Schulenberg JE, & Patrick ME (2020). Demographic subgroup trends among adolescents in the use of various licit and illicit drugs, 1975–2019 (Monitoring the Future Occasional Paper No. 94). Institute for Social Research, The University of Michigan. https://deepblue.lib.umich.edu/bitstream/handle/2027.42/162581/mtf-occ94.pdf [Google Scholar]
  24. Kulis SS, Jager J, Ayers SL, Lateef H, & Kiehne E (2016). Substance use profiles of urban American Indian adolescents: A latent class analysis. Substance Use and Misuse, 51(9), 1159–1173. 10.3109/10826084.2016.1160125 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kuppens S, & Ceulemans E (2019). Parenting styles: A closer look at a well-known concept. Journal of Child and Family Studies, 28(1), 168–18. 10.1007/s10826-018-1242-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lamont AE, Woodlief D, & Malone PS (2014). Predicting high-risk versus higher-risk substance use during late adolescence from early adolescent risk factors using latent class analysis. Addiction Research & Theory, 22(1), 78–89. 10.3109/16066359.2013.772587 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lanza ST, & Cooper BR (2016). Latent class analysis for developmental research. Child Development Perspectives, 10(1), 59–64. 10.1111/CDEP.12163 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Latkin CA, Edwards C, Davey-Rothwell MA, & Tobin KE (2017). The relationship between social desirability bias and self-reports of health, substance use, and social network factors among urban substance users in Baltimore, Maryland. Addictive Behaviors, pp. 73, 133–136. 10.1016/J.ADDBEH.2017.05.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lenardson JD, Paluso N, & Ziller EC (2020). Substance use among rural and urban youth and young adults. Maine Rural Health Research Center. https://digitalcommons.usm.maine.edu/cgi/viewcontent.cgi?article=1072&context=behavioral_health [Google Scholar]
  30. Lonczak HS, Huang B, Catalano RF, Hawkins JD, Hill KG, Abbott RD, Ryan JA, & Kosterman R (2001). The social predictors of adolescent alcohol misuse: A test of the social development model. Journal of Studies on Alcohol, 62(2), 179–189. 10.15288/jsa.2001.62.179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Maccoby EE, & Martin JA (1983). Socialization in the context of the family: Parent-child interaction. In Mussen PH (Series Ed.) & Hetherington EM (Vol. Ed.), Handbook of Child Psychology: Vol. IV. Socialization, Personality and Social Development (4th Ed., pp. 1–101). Wiley. [Google Scholar]
  32. Manly CA, Wells RS, & Kommers S (2019). Who are rural students? How definitions of rurality affect research on college completion. Research in Higher Education, 61, 764–779. 10.1007/s11162-019-09556-w [DOI] [Google Scholar]
  33. Masyn KE (2013). Latent class analysis and finite mixture modeling. In Little TD (Ed.), The oxford handbook of qualitative methods: Statistical analysis (pp. 551–611). Oxford University Press. [Google Scholar]
  34. MacCallum RC, Zhang S, Preacher KJ, & Rucker DD (2002). On the practice of dichotomization of quantitative variables. Psychological Methods, 7(1), 19–40. 10.1037/1082-989X.7.1.19 [DOI] [PubMed] [Google Scholar]
  35. Monnat SM, & Rigg KK (2016). Examining rural/urban differences in prescription opioid misuse among US adolescents. Journal of Rural Health, 32(2), 204–218. 10.1111/jrh.12141 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Muthén BO (2008, November 21). What is a good value of entropy. http://www.statmodel.com/discussion/messages/13/2562.html?1237580237
  37. Muthén BO (2016, January 29). LCA and sampling weights. http://www.statmodel.com/discussion/messages/13/1202.html?1511279738
  38. Muthén LK, & Muthén BO (2017). Mplus user’s guide (8th ed.). Muthén & Muthén. https://www.statmodel.com/download/usersguide/MplusUserGuideVer_8.pdf [Google Scholar]
  39. Nagin DS (2005). Group-based modeling of development. Harvard University Press. [Google Scholar]
  40. Nylund-Gibson K, & Choi AY (2018). Ten frequently asked questions about latent class analysis. Translational Issues in Psychological Science, 4(4), 440–461. 10.1037/tps0000176 [DOI] [Google Scholar]
  41. Patrick ME, Schulenberg JE, Martz ME, Maggs JL, O’Malley PM, & Johnston LD (2013). Extreme binge drinking among 12th-grade students in the United States: Prevalence and predictors. JAMA Pediatrics, 167(11), 1019–1025. 10.1001/jamapediatrics.2013.2392 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Pulver A, Davison C, & Pickett W (2015). Time-use patterns and the recreational use of prescription medications among rural and small-town youth. The Journal of Rural Health, 31, 217–228. 10.1111/jrh.12103. [DOI] [PubMed] [Google Scholar]
  43. Rhew IC, Hawkins JD, & Oesterle S (2011). Drug use and risk among youth in different rural contexts. Health and Place, 17(3), 775–783. 10.1016/j.healthplace.2011.02.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Salas-Wright CP, AbiNader MA, Vaughn MG, Schwartz SJ, Oh S, Delva J, & Marsiglia FF (2019). Trends in substance use prevention program participation among Adolescents in the U.S. Journal of Adolescent Health, 65(3), 426–429. 10.1016/j.jadohealth.2019.04.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Schafer JL, & Graham JW (2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147–177. 10.1037/1082-989X.7.2.147 [DOI] [PubMed] [Google Scholar]
  46. Substance Abuse and Mental Health Services Administration, Center for Mental Health Services. (2007). Promotion and prevention in mental health: Strengthening parenting and enhancing child resilience.https://store.samhsa.gov/sites/default/files/d7/priv/svp07-0186.pdf
  47. Swaim RC, & Stanley LR (2011). Rurality, region, ethnic community make-up and alcohol use among rural youth. Journal of Rural Health, 27(1), 91–102. 10.1111/j.1748-0361.2010.00324.x [DOI] [PubMed] [Google Scholar]
  48. Swaim RC, & Stanley LR (2021). Latent class analysis and predictors of marijuana use among reservation-based American Indian high school students. Journal of Psychoactive Drugs. 10.1080/02791072.2021.1918806 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Tomczyk S, Isensee B, & Hanewinkel R (2016). Latent classes of polysubstance use among adolescents—a systematic review. Drug and Alcohol Dependence, 160, 12–29. 10.1016/j.drugalcdep.2015.11.035 [DOI] [PubMed] [Google Scholar]
  50. Voulgaris CT, Smart MJ, & Taylor BD (2019). Tired of Commuting? Relationships among journeys to school, sleep, and exercise among American teenagers. Journal of Planning Education and Research, 39(2), 142–154. 10.1177/0739456X17725148 [DOI] [Google Scholar]
  51. Wu S, Yan S, Marsiglia FF, & Perron B (2020). Patterns and social determinants of substance use among Arizona youth: A latent class analysis approach. Children and Youth Services Review, 110. 10.1016/j.childyouth.2020.104769 [DOI] [Google Scholar]
  52. Zhen-Duan J, & Taylor MJ (2014). The use of an ecodevelopmental approach to examining substance use among rural and urban Latino/a youth: Peer, parental, and school influences. Journal of Ethnicity in Substance Abuse, 13(2), 104–125. 10.1080/15332640.2013.873006 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Table
NIHMS1958742-supplement-Supplementary_Table.docx (17.5KB, docx)

RESOURCES