An Examination of the Patterns of Substance Use in Activity Spaces and Their Relationship to Problematic Use

Abstract

During adolescence, adolescents are given more freedom to independently interact with a variety of social contexts. The eco-developmental model suggests that the activity spaces where adolescents spend their time affect substance-use behaviors beyond peer influences, and that the relationships may differ based on the adolescent’s demographic characteristics. This study examines adolescent patterns of reported substance use across activity spaces to determine whether the patterns of use are related to problematic substance use, and whether the relationships differ based on the participants’ race. Cross-sectional survey data from the study, Drug Use Among Young American Indians: Epidemiology and Prediction, 1993–2006 and 2009–2013, were used. Five patterns of adolescent alcohol use and six patterns of adolescent drug use in activity spaces were identified. There were significant differences in the relationship between class membership and problematic substance use by race, suggesting that contexts may be interacting with an adolescent’s race to influence use.

Introduction

Adolescent brains develop rapidly during a life stage characterized by increasing independence (Fuhrmann, Knoll, & Blakemore, 2015). The brain plasticity of adolescents makes them acutely sensitive to their context just as they are being given more freedom to interact with the world without supervision (Byrnes et al., 2016). Adolescent brain plasticity also makes behaviors and/or experiences formative for adult behavior during this time (Steinberg, 2014). In some cases, habits formed during adolescence may be helpful in adulthood; however, in other cases, such as substance use, harmful habits may lead to poor health outcomes later in life (Fuhrmann et al., 2015). Adolescent brains are even more vulnerable to developing habitual substance-use patterns because substance use during this period permanently affects reward systems as they are being developed (Steinberg, 2014). The increased vulnerability to the development of problematic substance-use behaviors in adolescents makes the current rates of substance use a persistent public health concern (Chen, Yi, & Faden, 2013).

Historically, substance-use prevention interventions have focused on the role of peers in substance-use behaviors. Currently, however, more attention is being paid to the role that context and/or activity spaces (defined as the locations an adolescent interacts with directly as a result of his or her daily routine) are playing in substance-use behaviors (Mason et al., 2015). Although it has been frequently assumed that the association between activity spaces and problematic substance use can be explained by peers who inhabit that space (Knecht, Burk, Weesie, & Steglich, 2011; Light, Greenan, Rusby, Nies, & Snijders, 2013), the eco-developmental model suggests that other aspects of the space (such as perceived risk and/or perceived stress) may also be affecting substance misuse within the space (Mason et al., 2015). Several studies have found associations between using substances in certain places (such as on school grounds) and substance-use behaviors; however, little is known about adolescent patterns of use across activity spaces (Hussong, 2000; Jones-Webb et al., 1997). In this study, we aim to understand (a) the patterns of adolescent substance use across several activity spaces, (b) the relationship between these patterns and problematic substance use (binge drinking, marijuana use, problems due to drinking, and doing drugs), and (c) whether these patterns differ based on age, gender, and race.

Eco-Developmental Model

Ecological systems theory emphasizes the importance of context in mental and behavioral health outcomes (Bronfenbrenner, 1979). Accordingly, individuals are affected by the multiple contexts within which they interact. Ecological systems theory posits that individuals experience stress when they encounter a discrepancy between their needs and their environment—a situation that requires them to adapt to re-achieve equilibrium. Efforts made to re-achieve equilibrium may be protective, such as reaching out to a friend for support, or they may be problematic, as in excessive substance use. When the developmental perspective is considered within this theory, the importance of and the level of exposure to contexts vary as an adolescent ages (Szapocznik & Coatsworth, 1999). During the adolescent stage, adolescents are typically given more freedom to interact within new and unsupervised contexts that expose them to new risk factors for substance use. Some contexts are considered high risk for adolescents when the opportunities to use are coupled with a lack of opportunity to express their emerging identities (Mason, Cheung, & Walker, 2004).

Spencer, Dupree, and Hartmann (1997) argue that there is a dynamic interplay between cultural contexts and identity formation. They assert that the norms within a given context communicate to adolescents who they are and whether they fit in, creating a perception that shapes adolescent behaviors within that context, thus, consequently affecting the attributes and norms of that space (Kondrat, 2002; LaScala, Freisthler, & Gruenewald, 2005). More specifically, an adolescent’s perception of his or her self in a given space will affect the choice to “use or downplay certain abilities, emphasize or draw attention away from certain physical attributes, adopt or suppress certain behaviors, engage in or shy away from certain activities” (Spencer et al., 1997, p. 818). Understanding how the place-person fit affects behavior and outcomes is especially important for racial-minority adolescents who are developing their identity in spaces (such as school) that may be dominated by White adolescents. Within the context of this study, this theory would posit that racial-minority adolescents may use substances in different spaces to achieve a better fit with their environment, and this use may have a differential impact on problematic use. This is especially true in spaces that are perceived as stressful and high risk; spaces such as these require coping mechanisms (i.e., substance use) to mitigate the stress (Spencer et al., 1997). In fact, Stock and colleagues (2013) found that Black youths, when living in a mostly Black neighborhood, were less likely to drink alcohol regardless of their level of racial identity; those adolescents with a low racial identity, when living in a mostly White neighborhood, drank more than those adolescents having a high racial identity (who drank as they normally would when living in a mostly Black neighborhood).

The majority of current research, which takes an ecological perspective in the examination of substance use, has focused on neighborhood environments and the relationships between neighborhood factors and substance use (Wiehe, Kwan, Wilson, & Fortenberry, 2013), whereas, to date, minimal research has been focused on (a) the role that activity spaces might have on substance use, and (b) how substance use within activity spaces might differ, based on an adolescent’s age, gender, and race.

Activity Spaces and Adolescent Substance Use

Although there is an increasing recognition that the features of the spaces where adolescents spend their time (activity spaces) may affect substance-use behaviors, this field of research is still in its early stages. This is possibly due to an overreliance on the summaries of large geographical areas (typically using Census data) where adolescents live (Mennis & Mason, 2011). This approach of reliance does not account for within-neighborhood variation, nor the subjective experiences of adolescents who frequently occupy spaces outside their home neighborhood (Tuan, 1977). A place in which an adolescent spends his or her time is more than its physical characteristics; it also includes social interactions and behavioral norms and expectations (Cresswell, 2004). Activity spaces may also present risks or stressors that challenge an adolescent’s ability to cope. Consequently, the consideration of exposure to activity spaces and the resulting behaviors of the adolescents in those spaces will advance our understanding of the risk factors in adolescent substance use (Browning & Soller, 2014).

The activity spaces where adolescents consume alcohol can be broadly divided into the categories of private and public activity spaces (Jones-Webb et al., 1997). According to Byrnes et al. (2016), adolescents spend approximately 28% of their time away from their home in spaces that could be considered public. Although it is frequently assumed that public spaces (i.e., after-school programs and recreation centers) are protective places for adolescents to spend their time, proximity to these spaces has been associated with more substance use. Mennis and Mason (2011) found that adolescents reported these spaces as stressful because there are often many substance users exhibiting violent behavior within the space; subsequently, the space becomes a particular risk for adolescent substance use. Although public spaces have been identified as more stressful and, therefore, more risky, using substances in private spaces (such as at home or at a party) may facilitate more frequent use due to a lack of supervision. Although adolescents have reported higher incidences of drinking in private activity spaces, such as their own home or that of a friend’s (Hussong, 2000; Kingston, Rose, Cohen-Serrins, & Knight, 2017; Mason et al., 2015), problematic drinking seems to occur more frequently in public spaces. Consumption during outdoor school events and in open spaces has been associated with heavy- and binge-drinking behaviors (Hussong, 2000; Jones-Webb et al., 1997). Similarly, Lipperman-Kreda, Mair, Bersamin, Gruenewald, and Grube (2015) found that adolescents who reported more frequent drinking during the past year had an increased likelihood of drinking at parties and the homes of friends, whereas heavy drinkers were more likely to drink in public spaces, such as parking lots or street corners.

Subsequently, it is not surprising that studies have also found that in unsupervised contexts (i.e., bars, restaurants, and nightclubs), there are often negative consequences of drinking whereby adolescents reported missing work or feeling ashamed of their behavior (Huckle, Gruenewald, & Ponicki, 2016). According to Wells, Graham, Speechley, and Koval (2005), national samples of adolescent drinking patterns showed frequent drinking in public locations to be associated with increased aggression and fighting. Adolescents who drink heavily at restaurants, bars, nightclubs, and someone else’s home reported violence-related behaviors, such as fighting and insulting other individuals while drinking (Mair, Lipperman-Kreda, Gruenewald, Bersamin, & Grube, 2015). These studies suggest that some activity spaces/drinking contexts present a greater risk of alcohol-related problems for adolescents (Huckle et al., 2016).

The eco-developmental model suggests the relationship between activity spaces and substance use varies, based on an adolescent’s demographic characteristics (Spencer et al., 1997). The relationship between substance use and spending time in risky activity spaces is stronger for older adolescents (Mennis & Mason, 2011). Older girls’ substance use is associated with their proximity to risky spaces, whereas older boys’ substance use is associated with the presence of after-school programs and recreation centers near their home (Mennis & Mason, 2011). When rates of use in different activity spaces were examined by age and gender, significant differences were found (Goncy & Mrug, 2013). Older adolescents reported using substances more often at the homes of friends, at school, and in cars, compared with younger adolescents who reported drinking more often at home. It has also been found that age is associated with drinking within more contexts and an increased likelihood of drinking at parties and the homes of friends (Lipperman et al., 2015). Also significant, males were more likely to report alcohol use in cars, and alcohol and marijuana use at school and on weeknights; females were more likely to report alcohol and marijuana use on the weekends (Goncy & Mrug, 2013).

Adolescents’ substance use in activity spaces has also been found to differ by race (Goncy & Mrug, 2013); for example, White adolescents consumed more alcohol on weekends and at the homes of friends, whereas Black adolescents consumed more alcohol at home and in cars, before, during, and after school, and on weeknights. White adolescents also reported smoking more marijuana on weekends and at the homes of friends; Black adolescents reported smoking more marijuana at school and during school. Although little is known about American Indian adolescents’ substance-use patterns within specific activity spaces, a recent study, which used the same data as the current study, found that American Indian adolescents (in Grades 8–10) living on or near a reservation had rates of use that were substantially higher for all substances than those of the national average (Swaim & Stanley, 2018). Although the rates of substance use for American Indian adolescents are substantially higher than those of other racial minority groups, particularly African American adolescents (Wallace et al., 2002), the contextual determinants of their uses may be similar due to their shared experiences of discrimination (Benner et al., 2018). Studies that have examined the relationship between specific contexts, such as neighborhoods/schools and substance use among Native American adolescents, have found relationships similar to those among other adolescents of color (Friese, Grube, & Seninger, 2016; Nalls et al., 2009). In a qualitative study that assessed their motivations to drink, Tingey et al. (2017) found that those Native American adolescents who drank did not participate in structured activities to the degree of those who did not drink; rather, the adolescents who drank described drinking during social time as including “walking around,” “being everywhere,” and “going to friends’ houses.” In this same study, American Indian adolescents who drank did not speak positively about school—a behavior consistent with Spencer’s theory that articulated coping and conformity motivations for binge drinking. Although these findings are not directly related to substance use within activity spaces, we may be able to hypothesize similar patterns of use within these contexts, based on Spencer’s claim that behaviors, such as substance use, are related to a fit or the lack of a fit between an adolescent and his or her environment.

The Current Study

Although several studies have tested the relationship between substance use within a single activity space (varying by age, gender, and race), it remains that little is known about adolescent patterns of use across multiple activity spaces. The one exception is a study that sums up the number of activity spaces where an adolescent uses substances (Goncy & Mrug, 2013). Based on the eco-developmental theoretical model, and building on previous research, the present study aims to determine the patterns of adolescent substance use across a variety of activity spaces. Once these patterns are established, our investigation will determine whether there are age, gender, and racial differences in adolescent patterns of use across activity spaces. We also aim to understand the relationship between patterns of use in activity spaces and problematic substance use, and whether this relationship differs by race. Although the identification of substance-use patterns within activity spaces is primarily exploratory, the results of previous studies (Goncy & Mrug, 2013; Lipperman et al., 2015; Mennis & Mason, 2011) suggest the following hypotheses:

• Hypothesis 1: There will be a distinction in patterns of use in both public and private spaces;

• Hypothesis 2: Older adolescents will use more in spaces outside of the home, and younger adolescents will use more in the home;

• Hypothesis 3: Males will use more after school; females will use more on the weekend and within a party context;

• Hypothesis 4: Black adolescents will use more before, during, and after school; White adolescents will use more at home and at parties;

• Hypothesis 5: Native American adolescents will use more during school and after school in public places than their White counterparts and will display similar patterns of substance use in activity spaces as do their Black counterparts; and

• Hypothesis 6: Using in public spaces will be related to heavier and more problematic use, whereas using in more private spaces will be related to more frequent use; these relationships may differ based on the adolescent’s race.

Method

Sample

The data used in this research study were collected from the Drug Use Among Young American Indians: Epidemiology and Prediction, 1993–2006 and 2009–2013 report (Beauvais & Swaim, 2015). The goal of the original report was to assess patterns of substance use among American Indian adolescents attending school on or near reservations. Surveys were administered in schools annually between the years 1993–2006 and 2009–2013. During subsequent data collection rounds, principal investigators excluded grades that had participated in the survey before, a step which ensured that a student participated in a survey only once (Beauvais & Swaim, 2015). Schools were eligible to be sampled if they had students in Grades 7 through 12, and a student body comprised 20% or more American Indian students. Surveys were completed during school hours. A stratified sampling frame by regions was used to ensure national coverage. The overall study response rate was 76%. Data from all waves were used in this study; however, it was limited to respondents who identified as White (N = 9,932, 39%), Black (N = 826, 3%), and American Indian (N = 14,457, 57%); with a total sample size of 25,215.

Measures

Dependent variables.

Problematic drinking and drug use were each measured with two items: (a) how often the adolescent reported being drunk (or using marijuana) within a month and (b) the number of problems the adolescent reported experiencing due to drinking (or drug use). To assess how often the adolescents had used substances during the last month, they were asked: How often during the last month have you drank to the point of being drunk? How often in the past month have you used marijuana? Possible responses included (0) = none, (1) = 1–2 times, (2) = 3–9 times, (3) = 10–19 times, (4) = 20+ times. Monthly marijuana was used, rather than all illicit drugs, because it was by far the most frequently used illicit substance, with 38.15% of adolescents reporting using within the last year, and it was the only illicit drug for which a 30-day use was measured. This should not, however, be equated with monthly rates of other substances with only a moderate to weak correlation between the rates of annual marijuana use and the annual rates of other illicit drugs. Substance-use problems were measured with two separate sets of 12 items, one for alcohol and one for drug use: Has your drinking (or drug use) ever caused you any of the following problems? (1) passing out, (2) not remembering what happened while drinking, (3) having a car accident, (4) being arrested, (5) receiving a traffic ticket, (6) having money problems, (7) getting into trouble at school, (8) failing in school work, (9) fighting with other students, (10) fighting with parents, (11) damaging a friendship, and (12) breaking/damaging something. Possible responses on a 4-point scale included (0) = no, (2) = 1–2 times, (3) = 3–9 times, (4) = 10+ times. It should be noted that marijuana use was singularly assessed when asking about substance-use behavior, and all drug use was assessed when asking questions regarding the consequences of use in the adolescent’s life.

Independent variables.

Alcohol and drug use in activity spaces was measured by asking respondents: During the last 12 months, where have you consumed alcohol (or drugs/marijuana): (1) at parties, (2) at school events, (3) on the way to school, (4) during school hours at school, (5) during school hours away from school, (6) immediately after school, (7) while driving around, (8) at home with parental knowledge (alcohol only), and (9) at home without parental knowledge. Possible responses were (0) = never, (1) = 1–2 times, (2) = 3–9 times, (3) = 10+ times.

Control variables.

Peer drinking was measured by asking respondents the following three questions: (1) How often do your friends ask you to get drunk? (0) = not at all, (1) = not much, (2) = some, (3) = a lot. (2) How many of your friends get drunk once in a while? and (3) How many of your friends get drunk every weekend? (0) = none, (1) = a few, (2) = most of them, (3) = all of them. Peer drug use was measured with a similar set of questions: (1) How often do your friends ask you to use marijuana? and (2) How many of your friends use marijuana? The same response categories were used for both sets of questions. Substance availability was measured with a single item for each substance that asked, How easy do you think it would be for you to get alcohol (marijuana) if you wanted some? Possible responses included (1) = probably impossible, (2) = very hard, (3) = hard, (4) = fairly easy, (5) = very easy.

Analysis Strategy

After the descriptive statistics were examined, the latent profile analysis (LPA) was conducted separately for alcohol use and illicit substance use. LPA uses all observations of continuous dependent variables to define classes using maximum likelihood estimation (Little & Rubin, 2002). The LPA was conducted in Mplus 7, employing a robust full-information maximum likelihood estimation procedure for handling missing data. The model is estimated iteratively, adding an additional class each time it is estimated, until the optimal model is reached. To determine the optimal number of classes, model solutions were evaluated using Bayesian information criterion (BIC) and the Lo–Mendell–Rubin (LMR) test (Lo, Mendell, & Rubin, 2001; Muthén & Muthén, 2000). Lower BIC indicated a better fitting model (Nylund, Asparouhov, & Muthén, 2007). The LMR is a bootstrap test estimating the value of twice the likelihood difference between k (estimated classes based on actual data) and k – 1 (generated data; Asparouhov & Muthén, 2012). Entropy, an index for assessing the precision of assigning latent classes, was also considered with a higher probability value, indicating greater precision in classification (Klonsky & Olino, 2008). Each individual was then assigned to a category based on the probability of being in a given class. The term class is used throughout to refer to the different patterns of substance use that occur within a space, which were identified in the LPA.

The data were then exported, and all subsequent analysis was conducted in STATA 14.

Once classes were assigned, bivariate tests were used to assess for age, gender, and racial differences within the classes. Analyses of variance (ANOVAs) were used to assess the mean differences in age, and chi squares were used to test differences by gender and race. To further assess the likelihood of belonging to each class based on demographic characteristics, multinominal logit models were estimated using the largest class (in both cases, the abstainers) as the reference group. Ordinary least square (OLS) regressions were then estimated to examine the relationship between problematic substance use and patterns of substance use within activity spaces. Finally, interactions between patterns of use within activity spaces and race were tested to understand whether the impact of these patterns of use varied significantly based on the adolescent’s race. All regression models predicting problematic substance use controlled for peer use and availability.

Results

Demographics

On average, adolescents in the sample were 15 years (ranging from 10 to 21) of age. The sample was evenly split between males (49.7%) and females (51.3%). On average, adolescents reported drinking and using substance within all the activity spaces between 0 and 1 to 2 times during the last 12 months (tables of descriptive statistics and multinominal logit models are available upon request). Adolescents most frequently used at parties with 31.5% reporting alcohol use and 25.4% reporting drug/marijuana use at least once during the past 12 months. This was followed by using while driving (19.3% alcohol use and 20.5% drug/marijuana use), when at home without parental knowledge (27.4% alcohol use and 21.2% reporting drug/marijuana), and after school (12.6% alcohol use and 18.7% drug/marijuana use; N = 21,545). Adolescents reported that their peers used alcohol (M = 1.67, SD = 0.98) more than marijuana (M = 1.09, SD = 0.97) and that alcohol (M = 3.90, SD = 1.38) was more available than marijuana (M = 3.58, SD = 1.52).

Alcohol Classes and Demographic Differences

When the LPA was estimated to test the first hypothesis for alcohol use, the 5-class solution was the best fitting model. While the BIC continued to decrease in the 6-class solution, the LMR test did not show significant differences in fit between the 5-class and the 6-class solution. Class 5 could clearly be identified as high users; Class 3 as abstainers; Class 1 as adolescents who reported drinking a moderate amount across most of the contexts, including on the way to school (see Figure 1); Class 2 as adolescents who reported drinking more on school campus during school, but less than most of the other groups in all other contexts; Class 4 as adolescents who frequently drink at parties, while driving, and at home without parental knowledge.

Figure 1.

Latent profiles of substance use (alcohol and marijuana/drugs) in activity spaces.

Note. LPA = latent profile analysis.

^aClass refers to the different patterns of substance use in space that were identified in the LPA.

The differences in the average age of adolescents in each alcohol class were tested (Hypothesis 2), and significant differences were found. The average age of high users (Class 5) was M = 15.4, SD = 1.74; the average age of abstainers (Class 3) was M = 14.88, SD = 1.70; and of moderate users (Class 1) was M = 15.35, SD = 1.62, whereas the average age of school users (Class 2) was M = 15.32, SD = 1.61, and the average age for party drinkers (Class 4) was M = 15.77, SD = 1.16. A global ANOVA test indicated there are significance differences in the average age of participants that fall within each class, F(4, 21,497) = 155.43, p < .001.

Gender differences in class membership were tested (Hypothesis 3). Significant differences were also found in the gender (χ² = 93.93, p > .001) makeup of each of the groups globally. To further understand where significant gender differences occurred, a multinominal logit was estimated, with the reported abstainers (Class 3) as the reference group. Males were less likely than females to be in Class 2 (B (SE) = −.55 (.08), p < .001), the group that uses at school, and Class 4 (B (SE) = −.11 (.05), p < .05), the group that uses at parties; however, they were 1.99 times more likely to be in Class 5, the group that uses frequently in all spaces (B (SE) = .69 (.12)).

Racial differences in class membership were tested (Hypotheses 4 and 5). When a chi-square test was conducted, significant differences were also found in the racial makeup (χ² = 191.84, p < .001) of each group. In multinominal logit models, American Indian adolescents were more likely than White adolescents (B (SE) = .64 (.07), relative risk ratio [RRR] = 1.89, p < .001), and equally as likely as Black adolescents, to be in Class 1 (characterized as moderate users within all contexts) than in Class 3 (the reference group). American Indian adolescents were, however, more likely than both White and Black adolescents to be in Class 2, the group that drinks on campus at school (W: B (SE) = .99 (.10), RRR = 2.68, p < .001; B: B (SE) = .73 (.32), RRR = 2.07, p < .001), and Class 4, the group that drinks frequently at parties and in cars (W: B (SE) = .31 (.05), RRR = 1.37, p < .001; B: B (SE) = .72 (.20), RRR = 2.06, p < .001) than in the group of abstainers. American Indian adolescents were also 1.57 times more likely than White adolescents (B (SE) = .45 (.12), p < .001), and as likely as Black adolescents, to be in Class 5, the high-use group, than to be in Class 3, the group that primarily abstained from drinking. Black adolescents, on the contrary, were not more likely than White adolescents to be in Class 1, the moderate-user group, and Class 2, the school-user group, than being in the abstainers group. They were, however, less likely to be in Class 4 (B (SE) = −.41 (.20), p < .001), the group that frequently drinks at parties, and 2.60 times more likely than White adolescents to be in Class 5 (B (SE) = .45 (.12), p < .001), the group with high levels of use within all contexts, than to be in the group characterized as abstaining in all contexts.

Predicting Problematic Alcohol Use

The relationship between class membership and problematic drinking was tested using OLS regression (Hypothesis 6). Table 1 illustrates a significant association found between class membership and the number of times the adolescents reported having been drunk during the past month and their reported problems due to alcohol use. In Class 5, the group (characterized as frequent drinkers in all spaces) was associated with a 1.76 increase in reported days drunk during the past month and an increase of .90 reported problems due to drinking when compared with the abstainers (when controlling for age, gender, race, peer use, and availability). If adolescents reported drinking patterns that were identified as party drinkers (Class 4), we could expect a .94 increase in the reported times of being drunk during the past month and a .42 increase in reported problems due to drinking compared with the abstainers. Adolescents who reported drinking patterns consistent with Class 2 (primarily school drinkers) had a .52 increase in reported days drunk during the past month and a .32 increase in reported problems attributed to drinking compared with the abstainers. Finally, adolescents who reported drinking moderately across all contexts also reported a .89 increase in reported days drunk during the last month and reported .45 more problems due to drinking compared with the abstainers.

Table 1.

Multiple Regression Testing the Association Between Classes of Alcohol Use in Activity Spaces and Problematic Substance Use.

	Alcohol
	Monthly use			Problems
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)
Age	.07 (.01)***	.02 (.00)***	.02 (.00)***	.07 (.01)***	.00 (.01)*	.00 (.00)**
Gender	.05 (.01)***	.06 (.01)**	.06 (.01)**	.05 (.01)***	.01 (.00)**	.01 (.00)**
Black	.06 (.04)	.06 (.03)*	.02 (.01)	.06 (.01)	.06 (.01)***	.04 (.01)**
American Indian	.06 (.01)***	−.00 (.01)	.03 (.01)**	.06 (.04)***	.06 (.00)***	.05 (.00)***
Class 1		.89 (.02)***	.97 (.04)***		.45 (.01)***	.47 (.02)***
Class 2		.59 (.02)***	.62 (.05)***		.32 (.01)***	.30 (.02)***
Class 4		.94 (.01)***	1.03 (.02)***		.42 (.01)***	.36 (.01)***
Class 5		1.76 (.03)***	2.10 (.06)***		.90 (.02)***	1.01 (.03)***
Class 1 × American Indian			−.12 (.04)			−.03 (.02)
Class 1 × Black			.16 (.14)			.00 (.06)
Class 2 × American Indian			−.06 (.06)			.02 (.03)
Class 2 × Black			.92 (.20)***			.20 (.09)*
Class 4 × American Indian			−.16 (.03)***			.09 (.1)***
Class 4 × Black			.23 (.13)			.17 (.06)**
Class 5 × American Indian			−.53 (.07)***			−.16 (.03)***
Class 5 × Black			−.23 (.17)			.00 (.07)
Peer use		.04 (.00)***	.04 (.00)***		.04 (.00)***	.04 (.00)***
Availability		.02 (.00)***	.02 (.00)***		.02 (.00)***	.02 (.00)***

Note. LPA = latent profile analysis.

^***p < .001,

^**p < .01,

^*p < .05

Class refers to the different patterns of substance use in space that were identified in the LPA.

To test whether these relationships differed based on race, the interaction between drinking patterns and race was examined, and some significant differences were observed (see Table 1 and Figure 2). Black adolescents in Class 2 (primarily drinking at school, on campus, and at parties) reported being drunk significantly more times and being in trouble due to drinking significantly more often than their White and American Indian counterparts who had similar drinking patterns across activity spaces. Similarly, Black adolescents in Class 4 (characterized as drinking at parties and while driving) reported significantly more times drunk and more problems due to drinking than their White counterparts. American Indian adolescents in this group also reported significantly more problems due to drinking than their White counterparts in this class. American Indian adolescents in Class 5 (the high users across contexts) reported significantly less times drunk during the past month and less problems due to drinking than their White and Black counterparts in the same class.

Figure 2.

Racial difference in problematic substance use by the patterns of use in activity spaces.

Note. LPA = latent profile analysis.

^aClass refers to the different patterns of substance use in space that were identified in the LPA.

Marijuana/Drug Use Classes and Demographic Differences

To further test the first hypothesis, LPA was estimated for drug/marijuana use in activity spaces, and a 6-class solution was found to be the best fitting model. While the BIC continued to decrease in the 7-class solution, the LMR test showed no significant difference in the fit between a 6-class and a 7-class solution. Similar to the patterns of alcohol use in activity spaces, the LPA identified a class that could clearly be identified as high users (Class 3) and one that could be identified as abstainers (Class 6). Class 1 could be characterized as adolescents who used a moderate amount of drugs at parties and outside of school contexts. Class 2 could be characterized as adolescents who used drugs on school campus during school, but less than most of the other groups within all other contexts. Class 4 could be characterized as adolescents who frequently used at parties and used a moderate amount within all other contexts, and Class 5 could be characterized as high users outside of school contexts.

Differences in the average age of participants within each group were tested (Hypothesis 2), and significant differences were found. The average age of high users (Class 3) was M = 15.69, SD = 1.64; the average age of abstainers (Class 6) was M = 14.89, SD = 1.73; the average age of low users (Class 1) was M = 15.26, SD = 1.64; and the average age of moderate users (Class 4) was M = 15.62, SD = 1.59, whereas the average age of school users (Class 2) was M = 15.12, SD = 1.58, and the average age for outside of school users (Class 4) was M = 15.62, SD = 1.59. An ANOVA test indicated there were significant differences in the average age of participants in each class, F(5, 16,656) = 61.20, p < .001.

Significant gender differences in group membership were tested (Hypothesis 3). A chi-square test indicated significant differences were found in the gender composition of typology generally (χ² = 52.34, p > .001). To further understand where significant gender differences occurred, a multinominal logit was estimated, with the reported abstainers (Class 6) as the reference group. No gender differences were found in the likelihood of reporting patterns of use consistent with Class 1, Class 5, and Class 4 when compared with the abstainers. Males were less likely than females to be in the group that reported using at school (Class 2) (B (SE) = −.18 (.08), p < .05), but were 1.52 times more likely to be in Class 3 (B (SE) = .42 (.09), p < .001), the group that reported high use across all contexts, than to be in the group that abstained in all contexts.

Similar to the analysis of alcohol-use patterns, significant differences in the racial makeup of each group were tested (Hypotheses 4 and 5). A chi-square test revealed significant differences in the racial makeup of each drug-use pattern (χ² = 741.57, p > .001). In multinominal logit models (Class 6 as reference), American Indian adolescents were more likely than Black and White adolescents to be in Class 1 (moderate use within all contexts except school) (W: B (SE) = .98 (.07), RRR = 2.65, p < .001; B: B (SE) = .65 (.24), RRR = 1.91, p < .01); Class 2 (those who use at school on campus) (W: B (SE) = 1.58 (.11), RRR = 4.87; B: B (SE) = .68 (.33), RRR = 1.97, p < .05); and Class 5 (those who report being heavy users outside of school contexts) (W: B (SE) = 1.07 (.11), RRR = 2.91 p < .001; B: B (SE) = .82 (.42), RRR = 2.27, p < .05). American Indian adolescents were more likely than White adolescents (W: B (SE) = 1.46 (.11), RRR = 4.32, p < .001), but not more likely than Black adolescents, to be in Class 3 (those who reported heavy use within all contexts). Similarly, American Indian adolescents were 4.76 times more likely than White adolescents (B (SE) = 1.56 (.15), p < .001) and less likely than Black adolescents to be in Class 4 (those who reported moderate use). Black adolescents were 2.47 times more likely than White adolescents to be in Class 2 (B (SE) = .91 (.34), p < .01) (those who reported using on school campus), and 3.55 times more likely to be in Class 3 (B (SE) = 1.27 (.28), p < .001) (those who reported heavy use across all contexts); they were not, however, more likely than their White counterparts to report use consistent with the other groups.

Predicting Problematic Marijuana/Drug Use

To test Hypothesis 5, the relationship between class membership and reported problematic drug use was estimated. A significant association was found between class membership and both the reported number of times the adolescents smoked marijuana during the past month and their reports of problems due to drug use (see Table 2). Class 5 (those who reported heavy use off campus) was associated with a 2.42 increase in the reported frequency of marijuana use during the past month and a .26 increase in the reported problems due to drug use when controlling for age, gender, race, peer use, and availability. Adolescents in Class 4 (moderate users across all contexts) were associated with a 2.34 increase in the reported frequency of marijuana use during the past month and a .35 increase in reported problems due to drug use. Class 3 high users were associated with a 2.52 increase in the reported frequency of drug use during the past month and a .44 increase in reported problems attributed to drug use. Class 2 school users were associated with a .88 increase in the reported frequency of marijuana use and a .17 increase in reported problems due to drug use. Finally, adolescents with low levels of drug use across all contexts reported a 1.06 increase in the frequency of marijuana use during the past month and a .15 increase in the number of reported problems due to drug use.

Table 2.

Multiple Regression Testing the Association Between Classes of Marijuana/Drug Use in Activity Spaces and Problematic Marijuana/Drug Use.

	Drugs
	Monthly use			Problems
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)
Age	.10 (.01)***	−.01 (.00)*	−.01 (.00)*	.01 (.00)***	−.01 (.00)***	−.01 (.00)**
Gender	.14 (.02)***	.10 (.01)***	.11 (.01)***	.01 (.00)	.00 (.00)	.00 (.00)
Black	.30 (.06)***	.03 (.05)	.06 (.05)	.07 (.01)***	.06 (.01)***	.01 (.02)
American Indian	.61 (.02)***	.07 (.01)***	.03 (.01)	.07 (.00)***	−.01 (.00)	.00 (.00)
Class 1		1.06 (.02)***	1.00 (.04)***		.15 (.01)***	.17 (.01)***
Class 2		.88 (.03)***	.64 (.08)***		.17 (.01)***	.12 (.03)***
Class 3		2.52 (.04)***	1.91 (.08)***		.44 (.01)***	.57 (.03)***
Class 4		2.34 (.05)***	2.12 (.11)***		.35 (.02)***	.52 (.04)***
Class 5		2.42 (.04)***	2.21 (.07)***		.26 (.01)***	.32 (.02)***
Class 1 × American Indian			.09 (.05)			−.04 (.02)*
Class 1 × Black			.17 (.19)			.09 (.06)
Class 2 × American Indian			.29 (.09)***			.06 (.03)*
Class 2 × Black			.33 (.26)			.18 (.09)*
Class 3 × American Indian			.78 (.08)***			−.19 (.03)***
Class 3 × Black			−.13 (.22)			.81 (.08)**
Class 4 × American Indian			.29 (.12)*			−.21 (.04)***
Class 4 × Black			.62 (.41)			−.59 (.13)***
Class 5 × American Indian			.31 (.08)***			−.08 (.03)**
Class 5 × Black			−.81 (.33)*			−.12 (.11)
Peer use		.30 (.01)***	.30 (.01)***		.05 (.00)***	.05 (.00)***
Availability		.00 (.00)	.00 (.00)		.00 (.00)	.00 (.00)

Note. LPA = latent profile analysis.

^aClass refers to the different patterns of substance use in space that were identified in the LPA.

To test whether these relationships differed by race, the interactions between drug-use patterns and race were examined; some significant differences were observed (see Table 2 and Figure 2). American Indian adolescents reported significantly less problems attributed to drug use when they were in Class 1, as compared with the White and Black adolescents. American Indian adolescents also reported using marijuana significantly less days and reported less problems attributed to drug use when they were in Class 2 (those who used the most at school). Black adolescents reported significantly more problems due to drug use when they were in Class 2, but not necessarily more marijuana use. When in Class 3, American Indian adolescents reported significantly more marijuana use and significantly less problems due to drug use than any other group. This finding could be contrasted with Black adolescents who reported experiencing significantly more problems due to drug use when in Class 3, but did not report significantly higher levels of marijuana use. Similarly, in Class 4, American Indian adolescents reported significantly less marijuana use and problems due to drugs as compared with their White counterparts. In this class, Black adolescents reported significantly less problems due to drugs—this finding, however, must be interpreted with caution due to the small sample size. In Class 5, where adolescents reported using somewhat regularly in all spaces other than at school, American Indian adolescents reported significantly more marijuana use and significantly less problems due to use, as compared with their White counterparts. Black adolescents in this class reported significantly less marijuana use than their White and American Indian counterparts; however, to note again, there were very few Black adolescents in Class 5.

Discussion

The aim of this study was twofold: (a) to understand the patterns of substance use in different activity spaces to determine whether the patterns of use differed based on the adolescents’ age, gender, and/or race and (b) to ascertain whether some patterns of use are more strongly related to problematic use for some adolescents more than others. Our analyses reveal patterns of use that did not appear to be context-specific. For the majority of adolescents in the sample, they abstained in all contexts, used moderately in all contexts, or used heavily in all contexts. There were, however, interesting distinctions between a subset of adolescents who used in private spaces (such as at parties and at home) and no public spaces (such as at school), and a subset of adolescents who used more at school and less in other spaces. The two classes (three in the case of illicit drug use) identified in this study are partially consistent with the differences in private and public use previously identified (Jones-Webb et al., 1997). There are also interesting age, gender, and racial differences in class membership, and in the relationship of being in either of these classes and problematic substance use.

It is important to note that the patterns of use differed significantly by age as hypothesized, suggesting a developmental pattern in how adolescents use substances within activity spaces. More specifically, the mean age of adolescents in each class suggests a progression of use. Adolescents who abstained from both alcohol and drug use were, on average, the youngest. The second two youngest groups were moderate drinkers and drug users across all contexts; they were also the group that primarily used at school. The oldest adolescents were more likely to be in the group of heavy drinkers and drug users across all contexts and were those that drank and used drugs the most at parties, while driving, and at home without parental knowledge. These findings are consistent with previous studies which reported that spending time within certain activity spaces primarily affected older adolescents’ substance-use behaviors (Mennis & Mason, 2011). Although it is consistent that older adolescents (more likely to be heavy users) used more at parties, at school, and in cars, the results indicate that younger adolescents may be using at school in ways that are important to consider, even if the frequency is not more than that of older adolescents. Substance use at school may be a transition for some adolescents from the group that uses very little to the group that uses at parties. Adolescents may start to associate with substance-using peers at a younger age, while at school, by using with them and joining with them later in adolescence in using at parties or at home. Testing this developmental hypothesis is beyond the scope of this study, but it might be an important line of inquiry for future research.

Consistent with the previous assertion that there is a dynamic interplay between contexts and the adolescent’s characteristics that may affect behaviors, such as substance-use behaviors (Spencer et al., 1997), and partially in line with this study’s hypotheses, the results suggest that females were more likely than males to be in the group that drank at school and the group that used in private spaces, whereas males were more likely to be in the group that used in all contexts. Females were also more likely to be in the group that primarily used drugs at school. The study’s finding that males are more likely to be in the group that uses in all contexts is consistent with previous findings that males drink more after school (Goncy & Mrug, 2013); however, the finding that females are more likely to be in the class that primarily drinks at school is a distinction that would have been missed if single items were considered in isolation. These substance-use patterns may be due to a mismatch between expected gender roles and the environment they inhabit. Girls may use drugs and alcohol at school solely to fit in with peers or to cope with stressors within that space. It is important to note the males were more likely to be in the class that reported drinking and using drugs heavily within all contexts, including frequently after school; a finding consistent with the research of Mennis and Mason (2011), who reported that males are more likely to use substances in after-school programs and recreation centers. Based on the data available in the current study, it is unclear whether after-school drinking was occurring in those spaces.

Finally, some evidence regarding the impact of the interaction between context and the adolescents’ characteristics on substance behavior was also observed when examining patterns of use by race (Spencer et al., 1997). To interpret the theoretical implications of these interactions, it should be noted that (a) the data were collected to understand Native American adolescent experiences when living on or near a reservation and (b) the inclusion criteria for the study included attending a school that was at least 20% Native American. This sampling strategy has important implications for our understanding of the context in which the adolescents’ race is being expressed, and in which the adolescents were attempting to achieve a fit between their demographic characteristics and their environment. It should also be noted that while Spencer and colleagues (1997) theorized about the relationship between identity and context, we can only speculate that the identified racial category of the adolescents corresponds with their identity, as it was not measured in the study. In this sample, American Indian adolescents were more likely than White adolescents to exhibit all patterns of use in space when compared with both the alcohol-use and drug-use patterns of abstainers. There were, however, some interesting similarities and differences between the American Indian and Black adolescents’ use in activity spaces. American Indian adolescents were more likely than Black adolescents to be in the group that drinks and uses drugs on campus, the group that drinks more often at parties and in cars, and the group that uses drugs moderately across all contexts, except school. They were, however, equally likely to be in the group that reported high alcohol and drug use across all contexts and the group that uses drugs after school, at home, and at parties. This is consistent with our hypothesis that Native American adolescents would use more at school and in public places; however, it also demonstrates that this pattern of use is distinct from their Black counterparts.

These findings are consistent with both the national trends that posit Native American adolescents use more substances than all other groups, and the trends found in this data set in previous studies (Swaim & Stanley, 2018). However, we also found the overrepresentation of American Indian adolescents within each class was not related to more days being drunk or more reported problems with alcohol when compared with the abstainers. In fact, American Indian adolescents, who were heavy users across all contexts and within private spaces, were less likely to report getting drunk than the White and Black adolescents in the heavy-user class. Similarly, American Indian adolescents in all classes were less likely to report more problems due to alcohol use than their White and Black counterparts. These findings may be due to the larger context in which they were using. In this study, it is possible that Native American adolescents were using in contexts where the majority of adolescents were Native American. Therefore, their high level of use across contexts may reflect community norms around substance use, and their lower level of reported days getting drunk, plus their problems due to drug and alcohol use, might reflect a fit between the adolescent’s race and the context in which they were using. This potential fit, as argued by Spencer et al. (1997), may lower the amount of stress the adolescents experience in activity spaces, making it less necessary for them to use heavily as a way to cope. This is, to some extent, consistent with Stock and colleagues’ (2013) findings that Black adolescents drank less when they were in neighborhoods that had a higher percentage of Black people, but differed in that Native American adolescents used at higher rates, although their use was not necessarily associated with problematic substance use. Regarding the American Indian adolescents, it may also be the case that their use across all contexts is in response to both the stereotypes that Native Americans use more substances and the discrimination toward substance use that follows. There was one exception to this trend. Although American Indian adolescents used less marijuana when they were in the class that used at school and/or the class that used moderately across all contexts, they used more marijuana than their White and Black counterparts when they were in a group of heavy users and those users that used before and after school, but not during.

Patterns of Black adolescents’ substance use also present some evidence to the theory that the fit between an adolescent’s racial category and the context of his or her activity spaces affects problematic substance use. Trends of use for Black adolescents regarding both parameters need to be interpreted with caution due to the small number of Black adolescents whose patterns of use in activity spaces fall into the primary groups of interest: the group that uses at school and the group that uses in private spaces (n = 286). Despite these small cell sizes, some interesting patterns emerged that should be noted. As hypothesized, Black adolescents were less likely to be in the group that drank in private spaces; however, when they were in that group, they reported more problems due to alcohol than the American Indian and White adolescents within that group. Both the lesser likelihood of participating in drinking at parties, cars, and so on, and the greater likelihood of experiencing problems when they do drink may reflect the adolescents’ struggle to fit within an environment that is majority Native American or White. When in the minority, Black adolescents may be less inclined to engage in a social setting where substance use is taking place; however, when they do, they may engage in problematic drinking to relieve the stress that occurs when there is a mismatch in an environment where they feel they do not belong. Contrary to our hypothesis, however, Black adolescents were not more likely to be in the group that drank at school but when they were, they reported more days of getting drunk and having more problems due to drinking than their White and American Indian counterparts. Again, the relationship between drinking at school and problematic drinking suggests that Black adolescents may be drinking in this context to cope with stress and a lack of fit with the environment. Consistent with this explanation, Black adolescents were more likely than White adolescents to use drugs at school. Being in this group did not translate into more marijuana use for Black adolescents; however, it did translate into significantly more problems due to drug use than that of their White and American Indian counterparts. It is possible that these associations are due to the use of substances other than marijuana, as the items used to measure problems and use in activity spaces both refer to drugs more globally. These patterns suggest that Black adolescents are not necessarily using more when they are using at school, but that they are simply more likely to experience negative consequences for that use. Although it is outside the scope of this study, it is possible that the increase in problems experienced by Black adolescents (when in these groups) may also be due to racial bias in policing substance use within a school setting and any increased conflict between adolescents within social settings due to racial dynamics. This remains another aspect of substance use that warrants further investigation.

Limitations

The results of this study should be interpreted in light of several limitations. Black adolescents comprised only 3% of the total sample; therefore, all findings related to this group should be interpreted with caution. In addition, this sample was drawn from schools that were either on or around a Native American reservation and where at least 20% of the student body identified as Native American (Beauvais & Swaim, 2015). Subsequently, there is a high probability that these communities and schools had a higher proportion of Native American adolescents than the national average. The density of Native American adolescents may affect all adolescent experiences of the activity spaces they frequent, and, therefore, the findings of this study may not be generalized. Because this is a school-based sample drawn from schools with a high proportion of American Indian students, the potentially high rate of drop-outs within this population may be excluding the most high-risk adolescents, consequently, biasing the findings (National Center for Education Statistics, 2018). This study’s cross-sectional data limited our ability to test the causal relationship between drinking within activity spaces and problematic use. Similarly, some schools were resurveyed over the course of data collection, 2001–2006 and 2009–2013 (Beauvais & Swaim, 2015). Even though the authors of the data stated they dropped any grade that had been surveyed previously, it is possible that individuals were surveyed more than once, creating a statistical dependence within these data. Although random, stratified sampling was used at the school level, no weights were provided to ensure that the data were representative. In addition, some of the measures of activity spaces are rather vague; after school, for instance, does not give any information about where that adolescent is located. The peer drinking and the American Indian adolescent’s availability controls are not activity-space specific. Future research should work to delineate attributes of given spaces and account for all adolescent interactions within that space, and not simply whether they have chosen that space specifically to drink. In addition, future research interested in understanding the interactions between context and identity should include a measure of racial and gender identity rather than relying on racial and gender categories (Ashmore, Deaux, & McLaughlin-Volpe, 2004). It is also the case that identity not only interacts with context, but it may also change based upon the context (Stryker & Burke, 2000). Real-time evaluation of racial and gender identity, or their self-categorization of and/or psychological attachment to those categories, and how it may vary based upon context, may also be an important line of inquiry to pursue.

Conclusion

Age, gender, and race differences in patterns of reported substance use across activity spaces and the way these differences are associated with problematic use suggests that contexts and individual characteristics are interacting to affect substance use. Differential relationships seem to follow, at least in part, the idea that adolescents may participate in problematic use when they are unable to achieve a fit between their racial/gender category and their environment. If this is the case, then successful prevention efforts may need to consider either modifying aspects of the context or situating prevention content within realistic representations of adolescents’ lived experiences. Before this can be accomplished, more knowledge needs to be gained detailing how adolescents perceive and experience their racial/gender categories within different spaces and the characteristics of space in which adolescents’ substance use becomes problematic must be identified.

Author Biographies

Jaime M. Booth is an assistant professor in the School of Social Work at the University of Pittsburgh. Dr. Booth’s research examines the role of neighborhoods in adolescents’ substance use and delinquency, the impact of differential stress experiences on health disparities in minority populations and ultimately strives to identify protective factors that can be enhanced to mitigate these outcomes.

Jildyz Urbaeva is an assistant professor in the School of Social Welfare at the University at Albany SUNY. Her research focuses on health and social disparities in transitional economies, with a focus on Central Asian countries. Her current work focuses on access to health and social services for women and minority populations, substance use among youth, and risky sex behaviors in the context of HIV epidemics.

Daejun Park is a PhD candidate in social work at University at Albany, SUNY. His research interests focus on alcohol and other drug use, and health disparities. He has served as a research assistant at UAlbany.