Skip to main content
NCBI home page
Journal of Studies on Alcohol and Drugs logoLink to Journal of Studies on Alcohol and Drugs
. 2021 Sep 19;82(5):584–594. doi: 10.15288/jsad.2021.82.584

Social Role, Behavior, and Belief Changes Associated With Driving After Using Marijuana Among U.S. Young Adults, and Comparisons With Driving After 5+ Drinking

Yvonne M Terry-McElrath a,*, Patrick M O’Malley a
PMCID: PMC8819607  PMID: 34546904

Abstract

Objective:

This study examined past-2-week driving after marijuana use (DMU) and driving after having five or more drinks (D5D) during young adulthood, specifically focusing on associations between within-person change in social roles (living situation, marriage, parenthood, education, employment) and mediators (perceived risk, evenings out, and religiosity) from modal ages 19 to 30.

Method:

Multilevel analyses were conducted using survey data collected from 2013 to 2019 from 1,873 adults (1,060 women; total number of data collection waves = 7,037) participating in the longitudinal Monitoring the Future study.

Results:

Change across waves from not being married to married was associated with lower DMU likelihood at any particular wave both directly and via mediation through wave-level change in evenings out. Change in employment (not employed to employed full time) was associated with higher D5D likelihood at any particular wave both directly and via mediation through change in evenings out. Wave-level change in other social roles was indirectly associated with DMU/D5D likelihood via wave-level change in evenings out.

Conclusions:

Change in all social roles examined was associated with change in evenings out, which appears to be a primary, proximal predictor of young adult DMU/D5D. Improved understanding of how socialization change is associated with driving after substance use may strengthen efforts to reduce the harms associated with such driving behaviors.

In 2018, approximately 11.8 million people age 16 and older in the United States reported driving under the influence of marijuana at least once in the past year (Substance Abuse and Mental Health Services Administration, 2019). Increases have been observed in self-reported driving after marijuana use (O’Malley & Johnston, 2013) and the percentage of fatal crashes involving marijuana-positive drivers (National Highway Traffic Safety Administration, n.d.; Traffic Injury Research Foundation, 2019). Although alcohol’s impairment effects on driving are widely recognized, less clarity exists regarding marijuana’s effects on driving (Berning et al., 2015; Compton, 2017). Marijuana significantly impairs safe driving skills (Centers for Disease Control and Prevention, 2017; Compton, 2017; Ogourtsova et al., 2018; Ramaekers et al., 2004; Ronen et al., 2008), but the degree to which impairment increases crash risk is not yet fully understood (Asbridge et al., 2012; Compton, 2017; European Monitoring Centre for Drugs and Drug Addiction, 2018; Li et al., 2012; National Institute on Drug Abuse, 2020; Rogeberg & Elvik, 2016). When marijuana and alcohol are used simultaneously, impairment and accident risk increase markedly (Chihuri et al., 2017; Ramaekers et al., 2004; Sewell et al., 2009). Organizations supporting legalized adult marijuana use emphasize never operating motor vehicles under the influence (Armentano, n.d.).

Young adults are at highest risk for driving after marijuana and alcohol use (Azofeifa et al., 2019; Kelley-Baker et al., 2017; Lipari et al., 2016; National Highway Traffic Safety Administration, 2019). There is a strong, positive association between use frequency and driving after using marijuana (Borodovsky et al., 2020; Goodman et al., 2020; Sukhawathanakul et al., 2019) or alcohol (Escobedo et al., 1995; Moan et al., 2013). However, understanding factors beyond use frequency that underlie age-related change in driving after marijuana or alcohol use would help identify high-risk young adult population subgroups as well as risk factors that may be amenable to change via prevention or intervention (Bingham et al., 2007). Such factors may include social roles, related beliefs, and behaviors.

Marijuana (and alcohol) use changes as young adults assume different social roles, particularly living arrangements, marital status, parenthood, student status, and employment status (Bachman et al., 1997, 2002). Few studies have examined social role associations with driving after marijuana or alcohol use. Living with parents was associated with a lower likelihood of driving after using marijuana or alcohol before controlling for substance use and driving frequency (Greene et al., 2019). A lower likelihood of driving after alcohol use has been associated with marriage (Chang et al., 1996; Fan et al., 2019; Sloan et al., 2014) and unemployment (Impinen et al., 2011; Sloan et al., 2014; Vaez & Laflamme, 2005); results are mixed for college attendance/completion associations (Fan et al., 2019; Sloan et al., 2014).

Social role associations with driving after substance use may reflect mediating roles of drug use attitudes/beliefs (e.g., perceived risk), socializing (e.g., evenings out for fun/recreation), and religiosity (e.g., importance of religion/religious attendance). Similar associations have been observed for social roles and substance use. Bachman et al. (2002) found that attending college was associated with increased socializing and decreased perceived risk; marriage and parenthood were associated with decreased socializing and increased risk perceptions. Perceived risk has been associated with driving after alcohol use: among one cross-sectional sample of young adult drivers, driving after alcohol use was associated with lower perceived risk of getting stopped by police, losing one’s license, lower driving abilities, and likelihood of crash, injury, or fine (Bingham et al., 2007).

The studies referenced above were cross-sectional, considering all measures as person-level characteristics. However, the concept underlying young adult social role and substance use associations is that within-person role change is associated with within-person change in use (Arnett, 2000; Schulenberg & Maggs, 2002; Staff et al., 2010). Staff et al. (2010) found that changes in socializing and religiosity mediated young adult associations between social roles, marijuana use, and heavy drinking. Within-person analyses may identify (a) specific social role transitions during which individuals are at heightened risk for driving after marijuana and alcohol use, and (b) if such social role transitions are associated with corresponding changes in mediating factors that may be amenable to change via prevention and intervention messaging. The current study examines between- and within-person associations between social roles, related mediators, and driving after marijuana use and 5+ drinking via two research questions (RQs): (1) How does the prevalence of driving after use vary across young adulthood? (2) To what extent are between- and within-person social role differences (living situation, marriage, parenthood, education, employment) and mediators (perceived risk, religiosity, evenings out) associated with driving after use, with key covariates being controlled for?

Method

Sample

Analyses used data from the Monitoring the Future study; detailed methodology is available elsewhere (Bachman et al., 2015; Schulenberg et al., 2020). In brief, nationally representative samples of approximately 15,000 12th-graders (modal age [hereafter referred to simply as age] 18) from about 130 schools in the 48 contiguous U.S. states are surveyed annually, completing self-administered surveys. A subsample of about 2,400 12th-graders is selected from each annual sample for longitudinal follow-up (with oversampling of drug users). A randomly selected half of the follow-up sample begins biennial follow-up 1 year after 12th grade (age 19), whereas the other half begins biennial follow-up 2 years after 12th grade (age 20). For the current analysis, mailed and online questionnaires were used to collect data at six follow-up waves: ages 19/20, 21/22, 23/24, 25/26, 27/28, and 29/30. The resulting data include responses at all ages from 19 through 30 (individual respondents provided data at a maximum of six ages). A University of Michigan Institutional Review Board approved the study.

Analysis was limited to the most recent seven cohorts (12th-grade cohorts from 2001 to 2007) that had the opportunity to complete all waves through age 29/30. Age 29/30 data were collected from 2013 through 2019. Measures examining driving after substance use were asked on one of six randomly distributed Monitoring the Future questionnaire forms. A total of 2,860 individuals selected for participation filled out the relevant form during the 12th-grade survey. Cases were limited to 1,960 respondents (68.5% of 2,860) who participated in at least one of the six waves (see Supplemental Appendix A for sensitivity analyses [supplemental material appears as an online-only addendum to this article on the journal’s website]). Of these respondents, data on both outcomes of interest were provided on at least one wave by 1,957 (99.8% of 1,960). A total of 7,879 waves were completed by these 1,957 respondents. The analytic method used in the current article requires no item-level missing data at any specific wave; 842 (10.7% of 7,879) waves were removed because of missing data, leaving a total of 7,037 (89.3% of 7,879) waves from 1,873 respondents for analysis (see Supplemental Figure 1 for sample flowchart). Mean waves per respondent was 3.8 (range: 1–6); 473 (25.3%) of the 1,873 respondents provided data at all six waves; 1,087 (58.0%) provided data at four or more waves. Attrition adjustments are discussed below.

Measures

Driving measures.

At each wave, respondents were asked “During the last two weeks, how many times (if any) have you driven a car, truck, or motorcycle after … (a) smoking1 marijuana? (b) having 5 or more drinks in a row?” Two dichotomous any/none measures were coded for analysis: any driving after marijuana use (DMU), and any driving after 5+ drinking (D5D).

Social roles.

Social roles (asked at each wave) included live with parents (yes/no); marital status (married/other); parental status (any children/none); college attendance (not attending, attending part time, attending full time at any school/college); and employment (not employed [no outside job/laid off/no paid employment], employed part time [2+jobs/1 part-time job], employed full time [1 full-time job]).

Mediators.

At each wave, items on perceived risk of regular marijuana use and perceived risk of 5+ drinking were asked as, “How much do you think people risk harming themselves (physically or in other ways) if they … (a) … use marijuana regularly? (b) … have five or more drinks once or twice each weekend?” with response options of 1 = no risk, 2 = slight risk, 3 = moderate risk, 4 = great risk. To measure socialization, evenings out was asked as, “During a typical week, on how many evenings do you go out for fun and recreation?” with response options of 1 = less than one, 2 = one, 3 = two, 4 = three, 5 = four or five, 6 = six or seven. Religiosity ranged from 1 to 4 and was an average of two items (one missing response allowed) assessing the importance of religion (1 = not important, 2 = a little, 3 = pretty, 4 = very important) and frequency of attendance at religious services (1 = never, 2 = rarely, 3 = 1–2/month, 4 = 1+/week or more).

Covariates.

At age 18, respondents reported their gender (male/female) and race/ethnicity (Black, Hispanic, White, other). (Gender and racial/ethnic identity are risk factors for driving after marijuana and alcohol use; Azofeifa et al., 2019; Chang et al., 1996; Delcher et al., 2013; Fan et al., 2019; Kelley-Baker et al., 2017; Oh et al., 2020; Vaez & Laflamme, 2005.) Remaining covariates were measured at each wave: average weekly miles driven (1 = none, 2 = 1–10 miles, 3 = 11–50 miles, 4 = 51–100 miles, 5 = 101–200 miles, 6 = more than 200 miles); and past-30-day marijuana use and alcohol use frequency (each measured as 1 = 0 occasions, 2 = 1–2, 3 = 3–5, 4 = 6–9, 5 = 10–19, 6 = 20–39, 7 = ≥40 occasions). Age was initially coded as integers from 19 to 30. For regression analysis, the linear age term was centered at the age in which the specific driving measure of interest peaked in prevalence; an additional quadratic age term was created using the centered age measure.

Analysis

All analyses were weighted to adjust for sampling and nonresponse (based on extensive information available from 12th-grade measures including gender, race/ethnicity, region, number of parents in household, average parental education, religiosity, average high school grades, truancy, college plans, and substance use). For RQ1, SAS v.9.4 (SAS Institute Inc., Cary, NC) survey commands were used to obtain descriptive statistics and age-specific estimates of DMU and D5D prevalence. For RQ2, analyses followed recommended multilevel model procedures (Heeringa et al., 2017; Nezlek, 2012; Raudenbush & Bryk, 2002; Singer & Willett, 2003; Snijders & Bosker, 1999; Sommet & Morselli, 2017) to model up to six wave data collection points (Level 1) nested within persons (Level 2) using Mplus v.7.4, specifying type=twolevel and using the MLR estimator with Monte Carlo integration. Wave- and person-specific weights were used in estimation (Heeringa et al., 2017), and asymmetric confidence intervals for mediation tests were obtained using the Monte Carlo method (Selig & Preacher, 2008). Person-level predictors were gender and race/ethnicity, as well as grand–mean-centered social roles, mediators, and covariates (average miles driven, marijuana use, and alcohol use). Wave-level (within-person) predictors were the centered and quadratic age terms (the centered term was included as a random slope; the quadratic age term showed no significant person-level variance), as well as person-centered social roles, mediators, and covariates (average miles driven, marijuana use, and alcohol use). See Supplemental Appendix B for further description of multilevel model specification using grand-mean and person-mean centering.

RQ2 analyses were conducted in three steps. (1) Bivariate social role associations: Both person- and wave-level indicators were included for each social role. (2) Multivariable models without mediation: Simultaneous inclusion of age, gender, and race/ethnicity, as well as both person- and wave-level indicators for all social roles, average miles driven, marijuana use, and alcohol use. (3) Multivariable models with wave-level mediation: Simultaneous inclusion of all measures in Step 2 above, as well as both person- and wave-level indicators for one mediator per model. All social roles were included simultaneously, but mediation was examined with only one social role per model. Figure 1 presents the general mediation model used (Preacher et al., 2010); Table 1 provides detailed descriptions of parameter estimates.

Figure 1.

Figure 1.

Open in a new tab

Hypothesized mediation in associations between social roles and driving after any marijuana use and 5+ drinking among U.S. young adults

Table 1.

Key mediation model regression coefficients for models examining driving after marijuana use (DMU) and driving after 5+ drinking (D5D)

graphic file with name jsad.2021.82.584tbl1.jpg

Coefficient Association Interpretation
ad X→M Direct association between within-person change in the specified social rolea (X) and within-person change in the specified mediatorb (M) at any specific wave, controlling for within-person change in other social roles, age, and all between-person differences. Associations should be the same (or very close to the same) for each social role/mediator pair when comparing DMU and D5D. For example, the direct association between living with parents and evenings out should not vary substantially between DMU and D5D models.
be M→Y Direct association between within-person change in the specified mediator (M) and the likelihood of the specified driving outcomec (Y) at any specific wave, controlling for within-person change in the specified social role (X), within-person change in other social roles, age, and all between-person differences. Associations should be the same (or very close to the same) for all b estimates within each driving outcome. For example, the direct association between evenings out and DMU should not vary substantially between models testing different social role/evenings out mediation associations with DMU.
c'e X→Y Direct association between within-person change in the specified social role (X) and the likelihood of the specified driving outcome (Y) at any specific wave, controlling for within-person change in the specified mediator (M), within-person change in other social roles, age, and all between-person differences.
a*be X→M→Y Association sequence from within-person change in the specified social role (X) through within-person change in the specified mediator (M) to the likelihood of the specified driving outcome (Y) at any specific wave, controlling for within-person change in other social roles, age, and all between-person differences:
Mediation association: If there is a significant total association between within-person change in the specified social role (X) and the likelihood of the specified driving outcome (Y) at any specific wave, within-person change in the specified mediator (M) transmits at least part of the significant association.
Indirect association: If there is no significant total association between within-person change in the specified social role (X) and the likelihood of the specified driving outcome (Y) at any specific wave, within-person change in the specified social role is indirectly associated with the likelihood of the specified driving outcome (Y) by being significantly associated with within-person change in the specified mediator (M), which then is significantly associated with the likelihood of the specific driving outcome (Y) at any specific wave.
Open in a new tab

Notes:

a

Either living with parents, marriage, parenthood, college attendance, or employment;

b

either perceived risk (of regular marijuana use if the outcome is driving after marijuana use, or perceived risk of 5+ drinking if the outcome is driving after 5+ drinking), evenings out, or religiosity;

c

either any driving after marijuana use, or any driving after 5+ drinking;

d

expressed as nonstandardized multivariable regression coefficients;

e

expressed as adjusted odds ratios of the multivariable likelihood of reporting any driving (either driving after marijuana use, or driving after 5+ drinking).

Results

Sample characteristics (Table 2)

Table 2.

Descriptive statistics

graphic file with name jsad.2021.82.584tbl2.jpg

Variable Range Person levela %/M (SD) Wave levelb %/M (SE)
Driving after substance use, past 2 weeks, %
Marijuana 0,1 18.0 (38.41) 8.9 (0.55)
5+ Drinks 0,1 19.8 (39.84) 8.7 (0.53)
Social roles
Live with parents (vs. other), % 0,1 59.6 (49.07) 31.7 (0.88)
Married (vs. other), % 0,1 36.3 (48.10) 21.7 (0.78)
Parent (vs. other), % 0,1 32.9 (46.99) 19.2 (0.83)
College attendance, % 1–3
None 82.2 (38.28) 58.2 (0.75)
Part time 27.0 (44.40) 10.1 (0.49)
Full time 63.7 (48.09) 31.7 (0.67)
Employment, % 1–3
None 50.1 (50.00) 23.2 (0.74)
Part time 59.8 (49.03) 30.2 (0.73)
Full time 72.8 (44.49) 46.5 (0.80)
Mediators
Risk of regular marijuana usec, M 1–4 3.2 (0.80) 3.2 (0.02)
Risk of 5+ drinking 1–2 times on weekendsc, M 1–4 3.2 (0.72) 3.2 (0.02)
Evenings outd, M 1–6 2.8 (1.03) 2.7 (0.02)
Religiosity, M 1–4 2.5 (0.90) 2.5 (0.02)
Covariates
Average weekly miles drivene, M 1–6 3.9 (1.24) 3.9 (0.03)
30-day alcohol use frequencyf, M 1–7 2.7 (1.35) 2.7 (0.03)
30-day marijuana use frequencyf, M 1–7 1.6 (1.25) 1.5 (0.03)
Modal age, M 19–30 23.5 (2.12) 24.1 (0.03)
Male (vs. female), % 0,1 48.4 (49.97) 44.8 (1.36)
Race/ethnicity, %
Black 0,1 12.1 (32.56) 9.7 (0.91)
Hispanic 0,1 13.3 (34.01) 12.1 (0.98)
Other 0,1 9.0 (28.55) 8.0 (0.74)
White 0,1 65.6 (47.49) 70.1 (1.32)
Open in a new tab

Notes: n(unweighted) = 7,067 waves from 1,873 individuals. All estimates obtained from weighted models.

a

Person-level percentages indicate the percentage of individuals who ever reported the specified measure. Person-level means indicate the average across all individual averages. Because these estimates were measured at the person level, no adjustment for clustering was needed and variance is reported using standard deviations (SD).

b

Wave-level percentages indicate the percentage of all waves that involved the specified measure. Wave means indicate the average value across all waves. Estimates obtained from models accounting for clustering by respondent and thus report standard errors (SE).

c

Risk responses: 1 = no risk, 2 = slight risk, 3 = moderate risk, 4 = great risk.

d

Evenings out responses: 1 = less than one, 2 = one, 3 = two, 4 = three, 5 = four or five, 6 = six or seven.

e

Miles driven responses: 1 = none, 2 = 1–10 miles, 3 = 11–50 miles, 4 = 51–100 miles, 5 = 101–200 miles, 6 = more than 200 miles.

f

Use frequency responses: 1 = 0 occasions, 2 = 1–2, 3 = 3–5, 4 = 6–9, 5 = 10–19, 6 = 20–39, 7 = 40 or more occasions.

Approximately half (48.4%) of respondents were male; racial/ethnic distribution was 12.1% Black, 13.3% Hispanic, 65.6% White, and 9.0% other. At 1+ waves, 59.6% of respondents reported living with parents, 36.3% reported being married, and 32.9% reported being a parent. Full-time and part-time college attendance at 1+ waves were reported by 63.7% and 27.0% of respondents, respectively. The majority of respondents reported being employed at 1+ waves: 72.8% full time and 59.8% part time. Average perceived risk was 3.2 (moderate) for both smoking marijuana regularly and having 5+ drinks once or twice each weekend; religiosity averaged 2.5 out of a range of 1–4. Mean evenings out was 2.8, indicating just under two nights per week. DMU and D5D were reported at 1+ waves by 18.0% and 19.8% of respondents, respectively. At the wave level, DMU was reported on 8.9% of the 7,037 waves and D5D on 8.7% of waves. Additional analyses (not tabled) found that of the 1,175 waves in which respondents reported any driving after either marijuana or alcohol use, 43.2% [SE = 2.34] involved driving after marijuana only, 41.8% [2.28] driving after alcohol only, and 15.0% [1.48] driving after both marijuana and alcohol.

Age-specific prevalence of driving after substance use

Figure 2 presents age-specific estimates of DMU and D5D from ages 19 to 30 at the person level. DMU prevalence increased slightly from 10.2% [SE = 1.15] at age 19 to 10.9% [1.36] at age 21, and then decreased through age 30 to 6.4% [1.38]. D5D increased from 7.5% [0.96] at age 19 to 12.4% [1.70] at age 25, and then decreased to 4.7% [0.95] at age 30.

Figure 2.

Figure 2.

Open in a new tab

Trends by modal age in prevalence of past-2-week driving after any marijuana use and 5+ drinking among U.S. young adults: 2002–2019. Notes: N(unwtd) = 1,873 respondents.

Social roles and driving behaviors without mediation

The intraclass correlation for DMU was .740 (95% CI [.692, .788]), indicating 74.0% of variance was between respondents (person level), whereas 26.0% was at the wave level. The intraclass correlation for D5D was .570 (95% CI [.504, .635]), indicating that variance was roughly split between person (57.0%) and wave (43.0%) levels. Table 3 presents bivariate and multivariable (without mediators) social role associations. In bivariate models, wave-level associations for the specified covariate control for person-level differences in the specified covariate. Because of space limitations, multivariable results for covariates (age, average miles driven, marijuana use, alcohol use, gender, and race/ethnicity) are not presented.

Table 3.

Wave- and person-level associations between social roles and driving after marijuana use and 5+ drinking among U.S. young adults ages 19–30 (no mediation)

graphic file with name jsad.2021.82.584tbl3.jpg

Drive after marijuana use Drive after 5+ drinking
Variable Bivariatea OR [95% CI] Multivariable without mediatorsb AOR [95% CI] Bivariate OR [95% CI] Multivariable without mediators AOR [95% CI]
Level 1: Wave (within-person)
Person centered measures (wave specific):c
Live with parents (vs. other) 1.54 [1.05, 2.25] 1.41 [0.79, 2.53] 1.49 [1.04, 2.12] 1.35 [0.91, 2.02]
Married (vs. other) 0.36 [0.19, 0.71] 0.42 [0.21, 0.85] 0.37 [0.24, 0.58] 0.67 [0.41, 1.10]
Parent (vs. other) 0.53 [0.32, 0.90] 0.77 [0.37, 1.60] 0.47 [0.32, 0.69] 0.72 [0.46, 1.12]
Attendance (vs. not attending)
Part time 1.29 [0.75, 2.21] 1.00 [0.47, 2.12] 1.43 [0.86, 2.37] 1.72 [0.99, 3.01]
Full time 1.60 [1.12, 2.29] 1.0† [0.54, 1.89] 1.27 [0.94, 1.70] 1.39 [0.88, 2.19]
Employment (vs. not employed)
Part time 1.34 [0.85, 2.11] 0.56 [0.30, 1.04] 1.52 [1.00, 2.29] 1.18 [0.74, 1.90]
Full time 0.82 [0.54, 1.25] 0.51 [0.26, 1.00] 1.43 [0.96, 2.15] 1.65 [0.98, 2.77]
Level 2: Person (between-person)
Grand-mean centered measures (average proportion of waves):d
Live with parents (vs. other) 0.98 [0.44, 2.17] 1.79 [0.87, 3.67] 0.92 [0.52, 1.64] 1.13 [0.62, 2.03]
Married (vs. other) 0.04 [0.02, 0.12] 0.50 [0.21, 1.18] 0.24 [0.12, 0.48] 0.38 [0.17, 0.86]
Parent (vs. other) 0.65 [0.29, 1.48] 0.88 [0.43, 1.82] 0.81 [0.47, 1.42] 1.72 [0.91, 3.24]
Attendance (vs. not attending)
Part time 0.87 [0.23, 3.24] 1.24 [0.36, 4.23] 0.89 [0.35, 2.27] 0.88 [0.31, 2.55]
Full time 0.22 [0.09, 0.57] 0.84 [0.37, 1.87] 0.52 [0.26, 1.02] 0.70 [0.32, 1.55]
Employment (vs. not employed)
Part time 0.36 [0.13, 1.01] 0.59 [0.25, 1.40] 1.47 [0.66, 3.25] 1.31 [0.63, 2.71]
Full time 0.80 [0.34, 1.91] 0.88 [0.37, 2.07] 3.04 [1.51, 6.11] 1.48 [0.70, 3.13]
Open in a new tab

Notes: Ns(unwtd.) = 7,037 waves from 1,873 individuals. All estimates obtained from weighted models. OR = bivariate odds ratio; AOR = multivariable odds ratio; CI = confidence interval. Bold font indicates associations significant at p = .05 or stronger.

a

Bivariate models included only one predictor per model (both wave- and person-level indicators for that predictor).

b

Multivariable models simultaneously included all wave- and person-level indicators for all five social roles; wave- and person-level indicators for average miles driven, past-30-day marijuana use frequency, and past-30-day alcohol use frequency; age and age2 at the wave level, and gender and race/ethnicity at the person level.

c

Person-centered as wave-specific value minus personal mean.

d

Grand–mean centered as person mean minus grand motive mean.

Driving after marijuana use.

Person-level bivariate associations (Table 3, column 1) indicated that lower DMU likelihood was associated with more waves married (odds ratio [OR] = 0.04) and attending college full time (vs. not attending, OR = 0.22). In multivariable models (column 2), no significant person-level associations between social roles and DMU likelihood remained. When we controlled for person-level differences, wave-level bivariate associations indicated that DMU likelihood was lower at specific waves when respondents were married (vs. not, OR = 0.36) or parents (vs. not, OR = 0.53), but higher at specific waves when respondents were living with parents (vs. not, OR = 1.54) or attending college full time (vs. not attending, OR = 1.60). In multivariable models, the wave-level association for marriage remained significant (AOR = 0.42); also, DMU likelihood was lower at specific waves when respondents were employed full time (vs. not employed, AOR = 0.51).

Driving after 5+ drinking.

Person-level bivariate associations (Table 3, column 3) showed more waves married were associated with lower D5D likelihood (OR = 0.24); more waves employed full time (vs. not employed) were associated with higher D5D likelihood (OR = 3.04). In multivariable models (column 4), only marriage remained significant (AOR = 0.38). When we controlled for person-level differences, wave-level bivariate associations indicated that D5D likelihood was lower at waves when respondents were married (vs. not, OR = 0.37) or parents (vs. not, OR = 0.47); bivariate D5D likelihood was higher at waves when respondents lived with parents (vs. not, OR = 1.49) or were employed part time (vs. not employed, OR = 1.52). In multivariable models, no wave-level associations remained significant.

Social roles and driving behaviors with mediation

Multivariable models examined wave-level mediation of each social role via perceived risk (risk of regular marijuana use for DMU; perceived risk of 5+ drinking for D5D), evenings out, and religiosity. Wave-level change in some social roles was associated with wave-level change in perceived risk and religiosity, but wave-level change in perceived risk and religiosity were not associated with DMU or D5D likelihood (see Supplemental Tables 1 and 2). Wave-level results for DMU and D5D models examining mediation via evenings out are presented in Table 4. Because of space limitations, neither wave-level results for age and other covariates, nor any person-level associations, are presented.

Table 4.

Multivariable wave-level associations between social roles and driving after marijuana use and 5+ drinking among U.S. young adults ages 19–30: Mediation via frequency of evenings out

graphic file with name jsad.2021.82.584tbl4.jpg

X (independent variable) a (XàMediator) Est. [95% CI] b (MediatoràOutcome) AOR [95% CI] c′ (XàOutcome) AOR [95% CI] (a*b) Indirect AOR [95% CI]
Drive after marijuana use
Live with parents (vs. other) 0.344 [0.250, 0.438] 1.38 [1.14, 1.68] 1.39 [0.77, 2.50] 1.12 [1.04, 1.21]
Married (vs. other) -0.763 [-0.851, -0.675] 1.38 [1.14, 1.68] 0.44 [0.22, 0.89] 0.78 [0.67, 0.91]
Parent (vs. other) -0.605 [-0.701, -0.509] 1.38 [1.14, 1.68] 0.83 [0.40, 1.71] 0.82 [0.72, 0.92]
Attendance (vs. not attending)
Part time 0.157 [0.030, 0.284] 1.38 [1.14, 1.68] 0.98 [0.46, 2.08] 1.05 [1.01, 1.12]
Full time 0.544 [0.468, 0.620] 0.99 [0.53, 1.87] 1.19 [1.07, 1.34]
Employment (vs. not employed)
Part time -0.048 [-0.152, 0.056] 1.38 [1.14, 1.68] 0.58 [0.31, 1.08] 0.98 [0.94, 1.02]
Full time -0.408 [-0.504, -0.312] 0.55 [0.28, 1.08] 0.88 [0.80, 0.95]
Drive after 5+ drinking
Live with parents (vs. other) 0.344 [0.250, 0.438] 1.34 [1.15, 1.56] 1.38 [0.92, 2.05] 1.11 [1.05, 1.18]
Married (vs. other) -0.763 [-0.851, -0.675] 1.34 [1.15, 1.56] 0.73 [0.45, 1.18] 0.80 [0.71, 0.90]
Parent (vs. other) -0.605 [-0.701, -0.509] 1.34 [1.15, 1.56] 0.78 [0.50, 1.22] 0.84 [0.76, 0.92]
Attendance (vs. not attending)
Part time 0.157 [0.030, 0.284] 1.34 [1.15, 1.56] 1.73 [0.98, 3.07] 1.05 [1.01, 1.10]
Full time 0.544 [0.468, 0.620] 1.43 [0.90, 2.28] 1.17 [1.08, 1.28]
Employment (vs. not employed)
Part time -0.048 [-0.152, 0.056] 1.34 [1.15, 1.56] 1.24 [0.77, 2.01] 0.99 [0.95, 1.02]
Full time -0.408 [-0.504, -0.312] 1.73 [1.02, 2.94] 0.89 [0.83, 0.94]
Open in a new tab

Notes: Ns(unwtd.) = 7,037 waves from 1,873 individuals. All estimates obtained from weighted models. OR = bivariate odds ratio; AOR = multivariable odds ratio; CI = confidence interval. Bold font indicates associations significant at p < .05 or stronger. All models simultaneously included all wave-and person-level indicators for all five social roles; wave- and person-level indicators for average miles driven, past-30-day marijuana use frequency, and past-30-day alcohol use frequency; age and age2 at the wave level, and gender and race/ethnicity at the person level. Mediation was tested with one social role per model. All measures person-centered as wave specific value minus personal mean.

Direct associations between wave-level change in social roles and change in evenings out were observed (parameter a). Respondents reported more evenings out at waves when they lived with parents (vs. not; nonstandardized multivariable regression coefficient estimate [Est.] = 0.344) or were attending school/college part- or full time (vs. not attending, Ests. = 0.157 and 0.544, respectively). Respondents reported fewer evenings out at waves when they were married (vs. not, Est. = -0.763), parents (vs. not, Est. = -0.605), or employed full time (vs. not employed, Est. = l-0.408).

Direct associations were observed for wave-level change in evenings out and likelihood of DMU and D5D at any specific wave (parameter b). At waves when respondents reported more evenings out, they reported higher likelihoods of DMU (AOR = 1.38) and D5D (AOR = 1.34).

With regard to direct associations between wave-level social role change and DMU or D5D likelihood at any specific wave after controlling for change in evenings out (parameter c′), DMU likelihood was significantly lower at waves when respondents were married (vs. not, AOR = 0.44). D5D likelihood was significantly higher at waves when respondents were employed full time (vs. not employed, AOR = 1.73).

Regarding parameter a*b, evidence was observed for both mediation associations (i.e., wave-level change in evenings out transmitted at least part of an observed social role total association with DMU or D5D likelihood) and indirect associations (no total association with wave-level social roles and DMU/D5D likelihood was observed, but social role change was indirectly associated DMU/D5D likelihood by associations with wave-level change in evenings out). Mediation associations were observed for wave-level change in marriage (DMU) and employment (D5D). Indirect associations were observed for wave-level change in living with parents (DMU and D5D), marriage (D5D), parenthood (DMU and D5D), college attendance (DMU and D5D), and employment (DMU).

Discussion

Among this national sample of young adults followed from ages 19 to 30, DMU prevalence peaked at age 21; D5D peaked at age 25. When we controlled for person-level differences, wave-level change in all social roles examined was associated with DMU and D5D likelihood through either mediation or indirect associations with wave-level change in evenings out. Wave-level change in specific social roles was associated directly with DMU or D5D likelihood. Prevention and intervention efforts to reduce DMU and D5D may be strengthened by focusing on subgroups undergoing specific social role transitions, and by further researching how socialization change may affect at-risk subgroups.

DMU and D5D likelihood varied as young adults changed how often they spent evenings out, which changed during social role transitions, even after controlling for person- and wave-level average miles driven and past-30-day marijuana and alcohol use frequency. The implication is not to stop evenings out (although that would clearly reduce driving after substance use). Additional research is needed exploring what specific characteristics of evenings out (where, when, and with whom) are most associated with DMU and/or D5D. Resulting knowledge may help strengthen interventions shown to reduce impaired driving, such as electronic screening and brief interventions (Community Preventive Services Task Force, 2013), multicomponent interventions with community mobilization (Shults et al., 2009), and availability of factors such as ride-hailing services (Burtch et al., 2021). Such research also may improve efficacy of event- and location-based interventions, such as those associated with birthday celebrations (Neighbors et al., 2012; Steinka-Fry et al., 2015) and college fraternities and sororities (Scott-Sheldon et al., 2016).

Awareness of which social role transitions result in higher socialization likelihood—and potentially higher likelihood of DMU/D5D—may help target prevention/intervention messaging. Evenings out increased within individuals when they transitioned (a) from either not attending or attending part time to full-time school/college attendance and (b) from full-time employment to unemployment. DMU and D5D prevention and intervention efforts may be relevant to incorporate into (a) incoming full-time student orientation programming at schools/colleges and (b) programming for the unemployed, such as transitional jobs programs. Health care providers also may make use of the current study’s findings: If individuals are or have recently undergone role transitions associated with higher risk (out of marriage, out of full-time employment to unemployment, returning to live with parents, or starting full-time attendance at school/college), risk of DMU/D5D may increase due to more evenings out; health care providers could consider screening and (if relevant) brief intervention, which is a recognized component of strategies to reduce impaired driving (Esteban-Muir, 2012).

Use of multilevel modeling identified complex differences in person- versus wave-level social role associations with DMU and D5D that are important for high-risk subgroup identification. Bivariate DMU likelihood was lower among respondents who spent more waves attending school/college full time compared with those not attending. However, controlling for these person-level differences, bivariate DMU likelihood at any particular wave increased when respondents transitioned from not attending to attending full time. These “oppositional” results speak to the importance of recognizing how between- and within-person associations may have meaningfully different intervention and prevention implications. Efforts to reduce DMU may be particularly needed among those who are less likely to attend school/college throughout young adulthood. Yet, the transition from not attending to attending full time appears to be a high-risk window for young adults, regardless of their overall education career.

Although partly mediated by evenings out, direct associations remained between wave-level change in some social roles and DMU or D5D. Transition to marriage was associated directly only with DMU. Many factors associated with marriage (change in social networks, routines, responsibilities, etc.) would be expected to relate similarly to DMU and D5D. One possible difference may be related to research that has observed strong associations between the transition to marriage and a decrease in crime (Sampson et al., 2006). Data collection for the current study occurred from 2002 to 2019; state legalization of adult recreational marijuana use first occurred in 2012, and federal policy continues to prohibit use. Via social control, transition to marriage in a policy environment prohibiting recreational marijuana use may lower the likelihood of engaging in behavior that may result in police involvement. If so, this association may weaken as both legal status and public acceptance of marijuana changes. Transition to full-time employment was associated directly only with D5D. Research has linked social motives, conformity pressures, and increased stress with heavy alcohol consumption in the workforce (Gunstone & Samra, 2019; Roche et al., 2015). The current study controlled for person- and wave-level past-30-day alcohol use frequency in terms of number of occasions but did not control for occasion-level consumption. If full-time employment is associated with unique drinking motives and pressures that, in turn, are associated with binge drinking, the likelihood of driving after such drinking could be expected to increase during such social role transition.

The lack of findings supporting mediation of social role associations with DMU or D5D likelihood via either perceived risk or religiosity should not be taken to mean these factors are not associated with DMU or D5D likelihood. Previous work found within-person change in religiosity mediated associations between social roles and use of both marijuana and alcohol (Staff et al., 2010); use likelihood is clearly associated with DMU and D5D risk. Future within-person research using items that specifically measure the perceived risk of driving after marijuana and alcohol use on a range of dimensions (physical safety, risk of arrest, social acceptance, etc.) would be informative.

Limitations

These findings should be considered within their limitations. Estimates are based on nationally representative samples of 12th-grade U.S. students; those who drop out of school before 12th-grade are excluded, and school dropout is associated with increased alcohol and marijuana use (Tice et al., 2017). All data were self-report. The measures used for at-risk driving are limited in that it is not possible to ascertain the extent to which drivers were actually impaired, either in respect to degree of consumption or time since use (i.e., how proximal driving was to a state of impairment). Monitoring the Future does not have specific measures of perceived risk of driving after alcohol or marijuana use. Last, as noted in Supplement Appendix A, attrition across young adulthood is a limitation, somewhat mitigated via weighting adjustments.

Conclusion

Within-person social role change was strongly associated with within-person change in evenings out, which was associated with change in DMU and D5D likelihood. Efforts to reduce the harms associated with these driving behaviors may be strengthened by identification of subgroups undergoing specific social role transitions and expanding understanding of how socialization change is associated with driving after substance use.

Footnotes

This research was supported by National Institute on Drug Abuse awards R01DA001411 and R01DA016575. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References

  1. Armentano P. Cannabis and driving: A scientific and rational review. Retrieved from https://norml.org/marijuana/library/cannabis-and-driving-a-scientific-and-rational-review/ [Google Scholar]
  2. Arnett J. J. Emerging adulthood. A theory of development from the late teens through the twenties. The American Psychologist. 2000;55:469–480. doi:10.1037/0003-066X.55.5.469. [PubMed] [Google Scholar]
  3. Asbridge M., Hayden J. A., Cartwright J. L. Acute cannabis consumption and motor vehicle collision risk: systematic review of observational studies and meta-analysis. BMJ. 2012;344:e536. doi: 10.1136/bmj.e536. doi:10.1136/bmj.e536. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Azofeifa A., Rexach-Guzmán B. D., Hagemeyer A. N., Rudd R. A., Sauber-Schatz E. K. Driving under the influence of marijuana and illicit drugs among persons aged ≥16 years — United States, 2018. Morbidity and Mortality Weekly Report. 2019;68:1153–1157. doi: 10.15585/mmwr.mm6850a1. doi:10.15585/mmwr.mm6850a1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bachman J. G., Johnston L. D., O’Malley P. M., Schulenberg J. E., Miech R. A. The Monitoring the Future project after four decades: Design and procedures. (Monitoring the Future Occasional Paper No. 82). Ann Arbor, MI: Institute for Social Research, University of Michigan; 2015. Retrieved from http://monitoringthefuture.org/pubs/occpapers/mtf-occ82.pdf. [Google Scholar]
  6. Bachman J. G., O’Malley P. M., Schulenberg J. E., Johnston L. D., Bryant A. L., Merline A. C. The decline of substance use in young adulthood: Changes in social activities, roles, and beliefs. Mahwah, NJ: Lawrence Erlbaum Associates; 2002. [Google Scholar]
  7. Bachman J. G., Wadsworth K. N., O’Malley P. M., Johnston L. D., Schulenberg J. Smoking, drinking and drug use in young adulthood: The impacts of new freedoms and new responsibilities. Mahwah, NJ: Lawrence Erlbaum Associates; 1997. [Google Scholar]
  8. Berning A., Compton R., Wochinger K. Results of the 2013-2014 National Roadside Survey of alcohol and drug use by driver. (Traffic Safety Facts Research Note. Report No. DOT HS 812 118). Washington, DC: National Highway Traffic Safety Administration; 2015. [Google Scholar]
  9. Bingham C. R., Elliott M. R., Shope J. T. Social and behavioral characteristics of young adult drink/drivers adjusted for level of alcohol use. Alcoholism: Clinical and Experimental Research. 2007;31:655–664. doi: 10.1111/j.1530-0277.2007.00350.x. doi:10.1111/j.1530-0277.2007.00350.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Borodovsky J. T., Marsch L. A., Scherer E. A., Grucza R. A., Hasin D. S., Budney A. J. Perceived safety of cannabis intoxication predicts frequency of driving while intoxicated. Preventive Medicine. 2020;131:105956. doi: 10.1016/j.ypmed.2019.105956. doi:10.1016/j.ypmed.2019.105956. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Burtch G., Greenwood B. N., McCullough J. S. Ride-hailing services and alcohol consumption: Longitudinal analysis. Journal of Medical Internet Research. 2021;23:e15402. doi: 10.2196/15402. doi:10.2196/15402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Centers for Disease Control and Prevention. What you need to know about marijuana and driving. 2017 Retrieved from https://www.cdc.gov/marijuana/pdf/marijuana-driving-508.pdf. [Google Scholar]
  13. Chang I., Lapham S. C., Barton K. J. Drinking environment and sociodemographic factors among DWI offenders. Journal of Studies on Alcohol. 1996;57:659–669. doi: 10.15288/jsa.1996.57.659. doi:10.15288/jsa.1996.57.659. [DOI] [PubMed] [Google Scholar]
  14. Chihuri S., Li G., Chen Q. Interaction of marijuana and alcohol on fatal motor vehicle crash risk: A case-control study. Injury Epidemiology. 2017;4:8. doi: 10.1186/s40621-017-0105-z. doi:10.1186/s40621-017-0105-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Community Preventive Services Task Force. Preventing excessive alcohol consumption: Electronic screening and brief interventions (e-SBI). The Guide to Community Preventive Services (The Community Guide) 2013 Retrieved from https://www.thecommunityguide.org/sites/default/files/assets/Alcohol-e-SBI.pdf. [Google Scholar]
  16. Compton R. Marijuana-impaired driving - A report to Congress (DOT HS 812 440). Washington, DC: National Highway Traffic Safety Administration; 2017, July. [Google Scholar]
  17. Delcher C., Johnson R., Maldonado-Molina M. M. Driving after drinking among young adults of different race/ethnicities in the United States: Unique risk factors in early adolescence? Journal of Adolescent Health. 2013;52:584–591. doi: 10.1016/j.jadohealth.2012.10.274. doi:10.1016/j.jadohealth.2012.10.274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Escobedo L. G., Chorba T. L., Waxweiler R. Patterns of alcohol use and the risk of drinking and driving among US high school students. American Journal of Public Health. 1995;85:976–978. doi: 10.2105/ajph.85.7.976. doi:10.2105/AJPH.85.7.976. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Esteban-Muir R. P. Community-based impaired-driving programs: Local ordinances and other strategies addressing impaired driving (Report No. DOT HS 811 678). Washington, DC: National Highway Traffic Safety Administration; 2012. Retrieved from https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/811678.pdf. [Google Scholar]
  20. European Monitoring Centre for Drugs and Drug Addiction and Canadian Centre on Substance Use and Addiction. Cannabis and driving: Questions and answers for policymaking. Luxembourg: Publications Office of the European Union; 2018. Retrieved from https://www.emcdda.europa.eu/publications/joint-publications/cannabis-and-driving_en. [Google Scholar]
  21. Fan A. Z., Grant B. F., Ruan W. J., Huang B., Chou S. P. Drinking and driving among adults in the United States: Results from the 2012-2013 national epidemiologic survey on alcohol and related conditions-III. Accident Analysis and Prevention. 2019;125:49–55. doi: 10.1016/j.aap.2019.01.016. doi:10.1016/j.aap.2019.01.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Goodman S. E., Leos-Toro C., Hammond D. Risk perceptions of cannabis- vs. alcohol-impaired driving among Canadian young people. Drugs: Education, Prevention & Policy. 2020;27:205–212. doi:10.1080/09687637.2019.1611738. [Google Scholar]
  23. Greene K. M., Hedstrom A. M., Murphy S. T. Driving/riding after alcohol and marijuana use among young adults: Is residing with family protective? Traffic Injury Prevention. 2019;20:679–684. doi: 10.1080/15389588.2019.1641597. doi:10.1080/15389588.2019.1641597. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Gunstone B., Samra S. Drinking behaviours and peer pressure in relation to alcohol: Drinkaware Monitor 2019. London, England: YouGov and Drinkaware; 2019. Retrieved from https://media.drinkaware.co.uk/media/novbxssv/drinkaware-monitor-2019-peer-pressure-final.pdf?v=0.0.9. [Google Scholar]
  25. Heeringa S. G., West B. T., Berglund P. A. Applied survey data analysis, second edition. Boca Raton, FL: CRC Press, Taylor and Francis Group; 2017. [Google Scholar]
  26. Impinen A., Mäkelä P., Karjalainen K., Haukka J., Lintonen T., Lillsunde P., Ostamo A. The association between social determinants and drunken driving: A 15-year register-based study of 81,125 suspects. Alcohol and Alcoholism. 2011;46:721–728. doi: 10.1093/alcalc/agr075. doi:10.1093/alcalc/agr075. [DOI] [PubMed] [Google Scholar]
  27. Kelley-Baker T., Berning A., Ramirez A., Lacey J. H., Carr K., Waehrer G., Compton R. 2013-2014 National Roadside Study of alcohol and drug use by drivers: Drug results (Report No. DOT HS 812 411). Washington, DC: National Highway Traffic Safety Administration; 2017, May. [Google Scholar]
  28. Li M.-C., Brady J. E., DiMaggio C. J., Lusardi A. R., Tzong K. Y., Li G. Marijuana use and motor vehicle crashes. Epidemiologic Reviews. 2012;34:65–72. doi: 10.1093/epirev/mxr017. doi:10.1093/epirev/mxr017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lipari R. N., Hughes A., Bose J. The CBHSQ Report: December 27, 2016. Rockville, MD: Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration; 2016. Driving under the influence of alcohol and illicit drugs. [PubMed] [Google Scholar]
  30. Moan I. S., Norström T., Storvoll E. E. Alcohol use and drunk driving: The modifying effect of impulsivity. Journal of Studies on Alcohol and Drugs. 2013;74:114–119. doi: 10.15288/jsad.2013.74.114. doi:10.15288/jsad.2013.74.114. [DOI] [PubMed] [Google Scholar]
  31. National Highway Traffic Safety Administration. Presence of drugs in drivers. Retrieved from https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/documents/13839-drugged_facts_flyer_101918_v8_002.pdf. [Google Scholar]
  32. National Highway Traffic Safety Administration. Alcohol-impaired driving (Traffic Safety Facts. Report No. DOT HS 812 864). Washington, DC: National Highway Traffic Safety Administration; 2019. [Google Scholar]
  33. National Institute on Drug Abuse. Marijuana research report. 2020 Retrieved from https://www.drugabuse.gov/download/1380/marijuana-research-report.pdf?v=d9e67cbd412ae5f340206c1a0d9c2bfd. [Google Scholar]
  34. Neighbors C., Lee C. M., Atkins D. C., Lewis M. A., Kaysen D., Mittmann A., Larimer M. E. A randomized controlled trial of event-specific prevention strategies for reducing problematic drinking associated with 21st birthday celebrations. Journal of Consulting and Clinical Psychology. 2012;80:850–862. doi: 10.1037/a0029480. doi:10.1037/a0029480. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Nezlek J. B. Multilevel modeling for psychologists. In: Cooper H., Camic P. M., Long D. L., Panter A. T., Rindskopf D., Sher K. J., editors. APA handbooks in psychology. APA handbook of research methods in psychology, Vol. 3. Data analysis and research publication. American Psychological Association; 2012. pp. 219–241. Retrieved from . [DOI] [Google Scholar]
  36. Ogourtsova T., Kalaba M., Gelinas I., Korner-Bitensky N., Ware M. A. Cannabis use and driving-related performance in young recreational users: A within-subject randomized clinical trial. CMAJ Open. 2018;6:e453. doi: 10.9778/cmajo.20180164. doi:10.9778/cmajo.20180164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Oh S., Vaughn M. G., Salas-Wright C. P., AbiNader M. A., Sanchez M. Driving under the influence of Alcohol: Findings from the NSDUH, 2002-2017. Addictive Behaviors. 2020;108:106439. doi: 10.1016/j.addbeh.2020.106439. doi:10.1016/j.addbeh.2020.106439. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. O’Malley P. M., Johnston L. D. Driving after drug or alcohol use by US high school seniors, 2001-2011. American Journal of Public Health. 2013;103:2027–2034. doi: 10.2105/AJPH.2013.301246. doi:10.2105/AJPH.2013.301246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Preacher K. J., Zyphur M. J., Zhang Z. A general multilevel SEM framework for assessing multilevel mediation. Psychological Methods. 2010;15:209–233. doi: 10.1037/a0020141. doi:10.1037/a0020141. [DOI] [PubMed] [Google Scholar]
  40. Ramaekers J. G., Berghaus G., van Laar M., Drummer O. H. Dose related risk of motor vehicle crashes after cannabis use. Drug and Alcohol Dependence. 2004;73:109–119. doi: 10.1016/j.drugalcdep.2003.10.008. doi:10.1016/j.drugalcdep.2003.10.008. [DOI] [PubMed] [Google Scholar]
  41. Raudenbush S. W., Bryk A. S. Hierarchical linear models: Applications and data analysis methods. 2nd ed. Thousand Oaks, CA: Sage; 2002. [Google Scholar]
  42. Roche A. M., Lee N. K., Battams S., Fischer J. A., Cameron J., McEntee A. Alcohol use among workers in male-dominated industries: A systematic review of risk factors. Safety Science. 2015;78:124–141. doi:10.1016/j.ssci.2015.04.007. [Google Scholar]
  43. Rogeberg O., Elvik R. The effects of cannabis intoxication on motor vehicle collision revisited and revised. Addiction. 2016;111:1348–1359. doi: 10.1111/add.13347. doi:10.1111/add.13347. [DOI] [PubMed] [Google Scholar]
  44. Ronen A., Gershon P., Drobiner H., Rabinovich A., Bar-Hamburger R., Mechoulam R., Shinar D. Effects of THC on driving performance, physiological state and subjective feelings relative to alcohol. Accident Analysis and Prevention. 2008;40:926–934. doi: 10.1016/j.aap.2007.10.011. doi:10.1016/j.aap.2007.10.011. [DOI] [PubMed] [Google Scholar]
  45. Sampson R. J., Laub J. H., Wimer C. Does marriage reduce crime? A counterfactual approach to within-individual causal effects. Criminology. 2006;44:465–508. doi:10.1111/j.1745-9125.2006.00055.x. [Google Scholar]
  46. Schulenberg J. E., Johnston L. D., O’Malley P. M., Bachman J. G., Miech R. A., Patrick M. E. Monitoring the Future national survey results on drug use, 1975-2019: Volume II, college students and adults ages 19-60. Ann Arbor, MI: Institute for Social Research, The University of Michigan; 2020. Retrieved from http://monitoringthefuture.org/pubs/monographs/mtf-vol2_2019.pdf. [Google Scholar]
  47. Schulenberg J. E., Maggs J. L. A developmental perspective on alcohol use and heavy drinking during adolescence and the transition to young adulthood. Journal of Studies on Alcohol, Supplement. 2002;14:54–70. doi: 10.15288/jsas.2002.s14.54. doi:10.15288/jsas.2002.s14.54. [DOI] [PubMed] [Google Scholar]
  48. Scott-Sheldon L. A. J., Carey K. B., Kaiser T. S., Knight J. M., Carey M. P. Alcohol interventions for Greek letter organizations: A systematic review and meta-analysis, 1987 to 2014. Health Psychology. 2016;35:670–684. doi: 10.1037/hea0000357. doi:10.1037/hea0000357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Selig J. P., Preacher K. J. Monte Carlo method for assessing mediation: An interactive tool for creating confidence intervals for indirect effects. 2008, June Retrieved from http://quantpsy.org/ [Google Scholar]
  50. Sewell R. A., Poling J., Sofuoglu M. The effect of cannabis compared with alcohol on driving. American Journal on Addictions. 2009;18:185–193. doi: 10.1080/10550490902786934. doi:10.1080/10550490902786934. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Shults R. A., Elder R. W., Nichols J. L., Sleet D. A., Compton R., Chattopadhyay S. K. Task Force on Community Preventive Services. Effectiveness of multicomponent programs with community mobilization for reducing alcohol-impaired driving. American Journal of Preventive Medicine. 2009;37:360–371. doi: 10.1016/j.amepre.2009.07.005. doi:10.1016/j.amepre.2009.07.005. [DOI] [PubMed] [Google Scholar]
  52. Singer J. D., Willett J. B. Applied longitudinal data analysis: Modeling change and event occurrence. New York, NY: Oxford University Press; 2003. [Google Scholar]
  53. Sloan F. A., Eldred L. M., Davis D. V. Addiction, drinking behavior, and driving under the influence. Substance Use & Misuse. 2014;49:661–676. doi: 10.3109/10826084.2013.858167. doi:10.3109/10826084.2013.858167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Snijders T., Bosker R. Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage; 1999. [Google Scholar]
  55. Sommet N., Morselli D. Keep calm and learn multilevel logistic modeling: A simplified three-step procedure using Stata, R, Mplus, and SPSS. Revue Internationale de Psychologie Sociale. 2017;30:203–218. doi:10.5334/irsp.90. [Google Scholar]
  56. Staff J., Schulenberg J. E., Maslowsky J., Bachman J. G., O’Malley P. M., Maggs J. L., Johnston L. D. Substance use changes and social role transitions: Proximal developmental effects on ongoing trajectories from late adolescence through early adulthood. Development and Psychopathology. 2010;22:917–932. doi: 10.1017/S0954579410000544. doi:10.1017/S0954579410000544. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Steinka-Fry K. T., Tanner-Smith E. E., Grant S. Effects of 21st birthday brief interventions on college student celebratory drinking: A systematic review and meta-analysis. Addictive Behaviors. 2015;50:13–21. doi: 10.1016/j.addbeh.2015.06.001. doi:10.1016/j.addbeh.2015.06.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Substance Abuse and Mental Health Services Administration. Center for Behavioral Health Statistics and Quality, National Survey on Drug Use and Health, 2017 and 2018. 2019 Retrieved from https://www.samhsa.gov/data/report/2018-nsduh-detailed-tables. [Google Scholar]
  59. Sukhawathanakul P., Thompson K., Brubacher J., Leadbeater B. Marijuana trajectories and associations with driving risk behaviors in Canadian youth. Traffic Injury Prevention. 2019;20:472–477. doi: 10.1080/15389588.2019.1622097. doi:10.1080/15389588.2019.1622097. [DOI] [PubMed] [Google Scholar]
  60. Tice P., Lipari R. N., Van Horn S. L. The CHBSQ Report. Rockville, MD: Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration; 2017. Substance use among 12th grade aged youths, by dropout status. Retrieved from https://www.samhsa.gov/data/sites/default/files/report_3196/ShortRe-port-3196.pdf. [PubMed] [Google Scholar]
  61. Traffic Injury Research Foundation. Rising marijuana use by drivers and growing prevalence among drivers in fatal crashes warrants concern. 2019 Retrieved from https://www.globenewswire.com/news-release/2019/11/28/1953678/0/en/Rising-marijuana-use-by-drivers-and-growing-prevalence-among-drivers-in-fatal-crashes-warrants-concern.html. [Google Scholar]
  62. Vaez M., Laflamme L. Impaired driving and motor vehicle crashes among Swedish youth: An investigation into drivers’ sociodemographic characteristics. Accident Analysis and Prevention. 2005;37:605–611. doi: 10.1016/j.aap.2005.03.001. doi:10.1016/j.aap.2005.03.001. [DOI] [PubMed] [Google Scholar]

Articles from Journal of Studies on Alcohol and Drugs are provided here courtesy of Rutgers University. Center of Alcohol Studies

RESOURCES