Advertisement

Mechanisms of change in an adapted marijuana e-CHECKUP TO GO intervention on decreased college student cannabis use

Published:January 25, 2021DOI:https://doi.org/10.1016/j.jsat.2021.108308

      Highlights

      • Randomized clinical trial of an adapted marijuana e-CHECKUP TO GO compared to a healthy stress management condition
      • Program effects on reductions in heavy use were transmitted by decreased marijuana use while studying.
      • Marijuana e-CHECKUP TO GO may be most effective at reducing student marijuana use while studying.

      Abstract

      The objective of this study was to test indirect effects of the Marijuana e-CHECKUP TO GO program on college students' frequent marijuana use through decreased use in specific social and academic activities. This study randomly assigned college students who reported frequent marijuana use (i.e., approximately five times per week) in fall 2016 to receive Marijuana e-CHECKUP TO GO or healthy stress management (HSM) strategies. The final baseline sample included 298 participants. Path analyses tested direct program effects on marijuana use at six-week posttest, as well as the indirect effect via use within four activities frequently participated in by college students: socializing, being physically active, studying, and being in class. Direct Marijuana e-CHECKUP TO GO effects on reductions in frequent use were transmitted by decreased marijuana use while studying and no use while socializing, being physically active, or in class. Marijuana e-CHECKUP TO GO may be most effective at reducing use of marijuana among college students while studying.

      Keywords

      1. Introduction

      The rate of past 30-day marijuana use for young adults (aged 19–28) has significantly increased over the past 5 years to current estimates of 24% (
      • 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, 1975–2018: Volume II, college students and adults ages 19–60.
      ). Marijuana use is associated with lower GPA, discontinuous enrollment, delay of graduation, less time studying, and drop-out (
      • Becker S.P.
      • Langberg J.M.
      • Luebbe A.M.
      • Dvorsky M.R.
      • Flannery A.J.
      Sluggish cognitive tempo is associated with academic functioning and internalizing symptoms in college students with and without attention deficit/hyperactivity disorder.
      ;
      • Bell R.
      • Wechsler H.
      • Johnston L.D.
      Correlates of college student marijuana use: Results of a US National Survey.
      ;
      • Suerken C.K.
      • Reboussin B.A.
      • Egan K.L.
      • Sutfin K.G.
      • Wagoner J.S.
      • Wolfson M.
      Marijuana use trajectories and academic outcomes among college students.
      ). Regular marijuana use is also related to decreased cognitive function and increased emotional problems (e.g., depression, anxiety), both of which may affect student achievement and retention (
      • Becker S.P.
      • Langberg J.M.
      • Luebbe A.M.
      • Dvorsky M.R.
      • Flannery A.J.
      Sluggish cognitive tempo is associated with academic functioning and internalizing symptoms in college students with and without attention deficit/hyperactivity disorder.
      ;
      • Bruffaerts R.
      • Mortier P.
      • Kiekens G.
      • Auerbach R.P.
      • Cuijpers P.
      • Demyttenaere K.
      • Kessler R.C.
      Mental health problems in college freshmen: Prevalence and academic functioning.
      ). For example, heavy use is associated with decreased higher-order cognitive processing (e.g., executive function), and decrements in learning and memory, self-conscious awareness, and IQ (
      • Batalla A.
      • Bhattacharyya S.
      • Yucel M.
      • Fusar-Poli P.
      • Crippa J.A.
      • Nogué S.
      • Martin-Santos R.
      Structural and functional imaging studies in chronic cannabis users: A systematic review of adolescent and adult findings.
      ;
      • Filbey F.
      • Yezhuvath U.
      Functional connectivity in inhibitory control networks and severity of cannabis use disorder.
      ;
      • Meier M.H.
      • Caspi A.
      • Ambler A.
      • Harrington H.
      • Houts R.
      • Keefe R.S.
      • Moffitt T.E.
      Persistent cannabis users show neuropsychological decline from childhood to midlife.
      ;
      • Zalesky A.
      • Solowij N.
      • Yucel M.
      • Lubman D.I.
      • Takagi M.
      • Harding I.H.
      • Seal M.
      Effect of long-term cannabis use on axonal fibre connectivity.
      ).
      The academic, cognitive, and psychosocial consequences associated with misuse of marijuana by college students serve as rationale for developing effective marijuana misuse interventions for college students. However, to date, few evidence-based interventions for marijuana use reduction among college students exist. Research has shown personalized normative feedback (PNF) interventions, such as Alcohol e-CHECKUP TO GO, typically implemented to reduce alcohol misuse among college students, correct misperceptions of alcohol use norms (
      • Larimer M.E.
      • Cronce J.M.
      Identification, prevention, and treatment revisited: Individual focused college drinking prevention strategies 1999–2006.
      ;
      • Neighbors C.
      • Larimer M.E.
      • Lewis M.A.
      Targeting misperceptions of descriptive drinking norms: Efficacy of a computer-delivered personalized normative feedback intervention.
      ). One study showed that Alcohol e-CHECKUP TO GO reduced normative perceptions of peer drinking, positive alcohol expectancies, and alcohol use. Based on the efficacy of Alcohol e-CHECKUP TO GO, researchers have developed similar interventions for marijuana misuse (
      • Blevins C.E.
      • Walker D.D.
      • Stephens R.S.
      • Banes K.E.
      • Roffman R.A.
      Changing social norms: The impact of normative feedback included in motivational enhancement therapy on cannabis outcomes among heavy-using adolescents.
      ). For example, BASICS (
      ; Brief Alcohol Screening & Intervention for College Students) and the recently developed CASICS (Cannabis Screening & Intervention for College Students) include 2 in-person sessions that last 60–90 min with a trained facilitator. Session content includes assessment of student alcohol or marijuana use patterns, history, and use-related consequences, personalized feedback, and strategies to reduce substance related risks. While BASICS is well established in the literature (
      • Fachini A.
      • Aliane P.P.
      • Martinez E.Z.
      • Furtado E.F.
      Efficacy of brief alcohol screening intervention for college students (BASICS): A meta-analysis of randomized controlled trials.
      ), CASICS is currently being delivered on campuses in the absence of published studies on its efficacy.
      Marijuana e-CHECKUP TO GO is a commercially available, online intervention providing personalized PNF designed to motivate college students to reduce marijuana use by correcting misperceived descriptive norms (i.e., misperceptions of the prevalence of use) and providing marijuana use education (
      • San Diego State University Research Foundation
      Marijuana eCHECKUP TO GO (eCTG) for universities & colleges.
      ). Although widely implemented, few studies have tested the efficacy of Marijuana e-CHECKUP TO GO. One exception—a study of 245 college student abstainers—found that participants reported more precise perceptions of descriptive (i.e., use prevalence) and injunctive (i.e., attitudes about use) norms at one-month posttest than did an assessment only control group (
      • Elliott J.C.
      • Carey K.B.
      Correcting exaggerated marijuana use norms among college abstainers: A preliminary test of a preventive intervention.
      ). A second study demonstrated Marijuana e-CHECKUP TO GO intervention effects on decreasing “extreme” descriptive norms of college students who report marijuana use, but not on student use or consequences of use for students who reported relatively heavy (i.e., 2+ times per week) use (
      • Elliot J.C.
      • Carey K.B.
      • Vanable P.A.
      A preliminary evaluation of a web-based intervention for college marijuana use.
      ). Our research group worked with the developers of Marijuana e-CHECKUP TO GO to adapt the program to include content focusing on increasing knowledge and use of protective behavioral strategies (PBS), strategies individuals can use to prevent or reduce substance use that have been found to mediate college student alcohol misuse outcomes in three randomized controlled trials (
      • Barnett N.P.
      • Murphy J.G.
      • Colby S.M.
      • Monti P.M.
      Efficacy of counselor vs. computer-delivered intervention with mandated college students.
      ;
      • Larimer M.E.
      • Lee C.M.
      • Kilmer J.R.
      • Fabiano P.M.
      • Stark C.B.
      • Geisner I.M.
      • Neighbors C.
      Personalized mailed feedback for college drinking prevention: A randomized clinical trial.
      ;
      • Murphy J.G.
      • Dennhardt A.A.
      • Skidmore J.R.
      A randomized controlled trial of a behavioral economic supplement to brief motivational interventions for college drinking.
      ). Some of the PBS added to the intervention include avoiding marijuana use before school or work, limiting use to weekends, avoid mixing with other drugs, and avoiding use to cope with emotions (
      • Pedersen E.R.
      • Huang W.
      • Dvorak R.D.
      • Prince M.A.
      • Hummer J.F.
      The Protective Behavioral Strategies for Marijuana Scale: Further examination using item response theory.
      ).
      Researchers conducted a pilot study (
      • Riggs N.R.
      • Conner B.
      • Parnes J.E.
      • Prince M.
      • Shillington A.
      • George M.W.
      Marijuana e-CHECKUP TO GO: Effects of a personalized normative feedback intervention for college student heavy marijuana use.
      ) of this adapted version of the Marijuana e-CHECKUP TO GO with students who reported frequent (i.e., 5 times per week average reported use frequency) cannabis use with the expectation that students who report frequent use would have a) the greatest misperceptions of social norms, b) the greatest need for PBS education, and c) the greatest need for intervention given their elevated risk for negative consequences of use. Thus, the study anticipated that those who report frequent use would be most responsive to the PNF intervention. This study demonstrated that among Marijuana e-CHECKUP TO GO participants, females reported greater use of PBS; more precise descriptive norms; and being high fewer hours per week, days per week, and weeks per month than participants in a comparison condition receiving strategies for healthy stress management (HSM), but that there were no direct program effects on marijuana use consequences (
      • Riggs N.R.
      • Conner B.
      • Parnes J.E.
      • Prince M.
      • Shillington A.
      • George M.W.
      Marijuana e-CHECKUP TO GO: Effects of a personalized normative feedback intervention for college student heavy marijuana use.
      ). Effect sizes ranged from small to medium and were entirely due to reductions in use among e-CHECKUP TO GO participants rather than between-group differences in use escalation.
      The study did not test indirect intervention effects on marijuana use through hypothesized mechanisms of change w in the first round of analyses. These analyses go beyond testing whether an intervention was efficacious to testing how the intervention contributed to decreased marijuana use. One important potential mechanism of change for this study was marijuana use by college students in specific social contexts (i.e., socializing, physical activity, studying, and in the classroom). A better understanding of the specific social contexts within which the Marijuana e-CHECKUP TO GO program reduces marijuana use will inform continuous program improvements, including adaptations that support reduced use in social contexts that may not be affected by the intervention in its present form. As such, this study is the first to test indirect program effects on marijuana use during specific activities to determine where college students reduced their use as a result of the Marijuana e-CHECKUP TO GO program and whether reductions in use during these activities are contexts within which direct program effects result in reduction in frequent marijuana use.
      The purpose of this study was to test the indirect effects of the Marijuana e-CHECKUP TO GO program on frequent marijuana use through reductions in use during four activities commonly in which college students commonly participate: being social/partying, being physically active, studying, and in class. Significant indirect effects would demonstrate specific pathways, or activities, through which Marijuana e-CHECKUP TO GO had its effects. This study hypothesized that there would be significant indirect effects of the intervention on cannabis use through reductions in marijuana use during time spent in each of these four specific activities.

      2. Method

      2.1 Participants & procedures

      Undergraduate college students were recruited in the Fall of 2016 via emails to on-campus residents and fraternity/sorority life, on-campus fliers, Facebook advertisements, and word-of-mouth. Students expressing interest in the study were e-mailed a screener to determine eligibility (see Fig. 1, Consort Flow Diagram). Eligibility criteria were that participants were 18 years of age or older, an undergraduate university student, self-reporting recreational marijuana use (i.e., non-medicinal) of at least twice per week. Of the 918 completed screeners, 527 (57%) met eligibility requirements. Participants were invited to participate on a rolling basis until the target number of 300 participants was achieved. One additional student was added to the study sample in between the time the target number of participants was achieved and when study staff were able to discontinue enrollment. Thus, the sample at baseline was 301 participants. Participants received $20 for completing the baseline survey and $10 for completing the 6-week posttest survey. The research described was conducted in accordance with the Institutional Review Board at the Colorado State University.
      Participants were randomly assigned to either the Marijuana e-CHECKUP TO GO (n = 146) or HSM (n = 155) comparison condition. Prior to intervention, all participants completed a 203-item survey asking about participants' personal substance use, perceived marijuana use norms, and PBS use. Baseline survey responses from 3 participants indicated that they did not meet study eligibility (n = 1 no reported use, n = 2 non-students), despite indicating as such on the screener. These participants were removed from further analyses. Therefore, the final baseline sample included 298 participants (PNF = 144, 48%; HSM = 154, 52%). The sample was 51% male and had a mean age of 19.97 years (SD = 2.0). No significant differences existed between the two study conditions on sex, racial/ethnic background, or age (
      • Riggs N.R.
      • Conner B.
      • Parnes J.E.
      • Prince M.
      • Shillington A.
      • George M.W.
      Marijuana e-CHECKUP TO GO: Effects of a personalized normative feedback intervention for college student heavy marijuana use.
      ).
      Following survey completion Marijuana e-CHECKUP TO GO participants received PNF regarding personal marijuana use, perceptions of marijuana use norms versus actual use prevalence at their university and nationally, and suggested PBS (
      • Riggs N.R.
      • Conner B.
      • Parnes J.E.
      • Prince M.
      • Shillington A.
      • George M.W.
      Marijuana e-CHECKUP TO GO: Effects of a personalized normative feedback intervention for college student heavy marijuana use.
      ). Comparison condition participants were provided with strategies for HSM (e.g., deep breathing, mindfulness, exercise). Participants were then sent up to three e-mails at four-day intervals inviting them to complete the same survey at 6-week posttest. Two hundred and twenty-seven (75%) participants (PNF = 109, 48%; HSM = 118, 52%) completed this survey. Retained participants reported significantly fewer hours high per week (t = −3.71, p < .001), hours high per use day (t = −3.60, p < .001), and days high per week (t = −2.46, p < .05) than those who did not complete six-week posttest surveys. Retained participants were also significantly less likely to be male than female (OR 0.52, 95% C.I. 0.30–0.89, p < .05). However, there were no statistically significant differences in the number of retained vs. non-retained participants across condition (OR 1.05, 95% CI 0.62–1.79, p > .05). There were also no statistically significant differences by condition in the attrition of students who identify as male or who report frequent use.

      2.2 Measures

      The independent variable of primary interest was intervention condition (PNF = 1, HSM = 0). The dependent variable was Periods High per Week. To calculate Periods High per Week, we asked participants to indicate if they are typically high during 6-h time blocks of each day of the week. The total number of endorsed time blocks during a typical week were summed to evaluate Periods High per Week. This method of assessing marijuana use has been validated in previous research (
      • Pearson M.R.
      • Liese B.S.
      • Dvorak R.D.
      • Marijuana Outcomes Study Team
      College student marijuana involvement: Perceptions, use, and consequences across 11 college campuses.
      ).
      The mediator variables were the proportion of time high while partying/socializing, exercising/playing sports, studying, and in class were measured by asking participants “During a typical school week, how many hours do you estimate you spend in total: partying/socializing, exercising/playing sports, studying, or in class?” with open text fields for each activity. They were then asked to report the number of hours they engaged in each activity during a typical week while under the influence of marijuana. Proportions were calculated by dividing the number of hours participating in each of these activities while under the influence of marijuana by the total amount of time participating in the four activities. The proportion of time high while participating in each activity was computed, rather than the total amount of time high while engaged in each activity, due to potential fluctuations in the number of total hours engaged in the activities over the study duration. At baseline, the proportion of time participants reported being high while social, physically active, studying, and in class was 0.65, 0.21, 0.17, and 0.11, respectively. Hours participating in activities high raw proportions are best modeled using beta regression models (cf.
      • Ferrari S.L.P.
      • Cribari-Neto F.
      Beta regression for modelling rates and proportions.
      ). Table 1 provides univariate descriptive statistics for each variable at baseline, by intervention condition. There were no significant differences between the two conditions.
      Table 1Variable univariate statistics at baseline by treatment condition.
      ConditionMarijuana e-CHECKUP TO GOHSM
      (n = 144) (n = 109)(n = 154) (n = 118)
      Baseline

      Mean (SD)
      Follow-up

      Mean (SD)
      Baseline

      Mean (SD)
      Follow-up

      Mean (SD)
      Number of time periods high per week9.24 (6.89)7.43 (6.67)8.90 (6.60)8.41 (6.11)
      High while social/partying0.62 (0.31)0.64 (0.32)0.66 (0.32)0.63 (0.31)
      High while exercising/playing sports0.18 (0.32)0.19 (0.35)0.23 (0.33)0.19 (0.32)
      High while studying0.19 (0.28)0.13 (0.24)0.16 (0.23)0.14 (0.22)
      High while in class0.10 (0.22)0.09 (0.21)0.11 (0.24)0.10 (0.21)
      Note: HSM = healthy stress management. For all High While… variables values represent the proportion of time high while engaging in each activity.

      2.3 Analyses

      In the present study, we used a “per-protocol” analysis (
      • Ranganathan P.
      • Pramesh C.S.
      • Aggarwal R.
      Common pitfalls in statistical analysis: Intention-to-treat versus per-protocol analysis.
      ) to handle attrition to assess mechanisms of change. Per-protocol analysis includes data from those who complete all aspects of the study. First, we tested the direct Marijuana e-CHECKUP TO GO program effects on use during the four activities potentially mediating direct intervention effects on marijuana use at the six-week follow-up using separate beta regressions (
      • Ferrari S.L.P.
      • Cribari-Neto F.
      Beta regression for modelling rates and proportions.
      ). Beta regressions are appropriate when the dependent variable ranges from 0 to 1 and is the best practice for analyzing proportions as outcomes. In beta regression models the extremes (i.e., 0 and 1) should be transformed using the following (y ∗ (n − 1) + 0.5)/n where y is each score on the dependent variable and n is the sample size (
      • Smithson M.
      • Verkuilen J.
      A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables.
      ). Beta regression analyses were conducted in the statistical software R (
      • R Core Team
      R: A language and environment for statistical computing, version 3.5.3. R Foundation for Statistical Computing, Vienna, Austria.
      ) with the betareg package (
      • Francisco C.-N.
      • Zeileis A.
      Beta regression in R.
      ). Results were used to build hypothesis testing models by trimming activities not affected by the intervention to create the final indirect effects model.
      Next, a residualized gains path analysis was conducted to test for indirect effects of treatment on marijuana use via activities that were identified in the direct effects tests using the MPlus 8 statistical package (
      • Muthén L.K.
      • Muthén B.O.
      Mplus user’s guide.
      ). The marijuana use variable, a highly skewed count variable, was analyzed using negative binomial regression for paths leading to marijuana use. The residualized gains approach controls for baseline values of each outcome in predictions of follow-up values (cf.
      • Pearl J.
      Lord’s paradox revisited – (Oh Lord! Kumbaya!).
      ). In this case baseline levels of proportion of time high while studying predicted proportion of time high while studying at follow-up, and baseline marijuana use predicted follow-up marijuana use. Sex was also controlled for on both proportion of time high while studying and marijuana use follow-up variables. Biological sex (male = 0, female = 1) was included as a covariate due to research demonstrating that males are approximately twice as likely to report heavy marijuana use (
      • 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. 1975–2016: Volume II, college students and adults ages 19–55. Ann Arbor: Institute for Social Research.
      ). Due to the negative binomial specification of the outcome variable (i.e., marijuana use at follow-up), typical model fit statistics are not available because count regression models are estimated using maximum likelihood with robust standard errors (MLR). MLR relies on raw data rather than means, variances, and covariances which eliminates the ability to calculate typical model fit indices. Further, the best practices approach for testing indirect effects is to use the product of coefficients method (
      • Hayes A.F.
      Beyond Baron and Kenny: Statistical mediation analysis in the new millennium.
      ). The product of coefficients method violates the normality assumption making p-values not trustworthy. Instead, the best approach for determining significance of indirect effects with count distributed outcomes is to evaluate Monte Carlo Confidence Intervals (MCCIs,
      • Preacher K.J.
      • Selig J.P.
      Advantages of Monte Carlo confidence intervals for indirect effects.
      , as implemented in:
      • Selig J.P.
      • Preacher K.J.
      Monte Carlo method for assessing mediation: An interactive tool for creating confidence intervals for indirect effects [Computer software].
      ). MCCIs that do not include zero are considered to be statistically significant. Further, regression coefficients from negative binomial regressions can be exponentiated to calculate Rate Ratios to ease interpretation (
      • Hilbe J.M.
      Negative binomial regression.
      ).

      3. Results

      3.1 Direct effects models

      The direct effects beta regression model results are presented in Table 2. Direct Marijuana e-CHECKUP TO GO program effects on proportion of time high while participating in each of the four activities demonstrated direct program effects on proportion of time high while studying (b = −0.31, SE = 0.14, p = .02, OR = 0.73), but not with proportion of time high while socializing/partying, being physically active, or in class. This can be interpreted as those in the Marijuana e-CHECKUP TO GO program reported 27% lower proportion of time high while studying compared to those in the healthy stress management condition at follow-up. Based on the results of the direct effects tests, only proportion of time high while studying was included in the path model examining indirect effects.
      Table 2Beta regression results examining the direct effects of treatment condition predicting proportion of time engaging in activities while high.
      Predictor variableEstimateStandard errorP-Value
      Proportion time high while studying at follow-up
      Treatment condition−0.310.140.02
      Proportion of time high while studying at baseline4.300.33<0.01
      Proportion time high while in class at follow-up
      Treatment condition−0.080.140.55
      Proportion of time high while in class at baseline4.340.40<0.01
      Proportion time high while exercising/playing sports at follow-up
      Treatment condition0.070.170.65
      Proportion of time high while exercising/playing sports at baseline2.920.33<0.01
      Proportion time high while partying/socializing at follow-up
      Treatment condition0.090.170.57
      Proportion of time high while partying/socializing at baseline1.630.28<0.01
      Note: Treatment condition coded 0 = Healthy Stress Management Condition; 1 = marijuana e-CHECKUP TO GO condition.

      3.2 Indirect effects model

      The indirect effects model results are presented in Table 3. As reported by
      • Riggs N.R.
      • Conner B.
      • Parnes J.E.
      • Prince M.
      • Shillington A.
      • George M.W.
      Marijuana e-CHECKUP TO GO: Effects of a personalized normative feedback intervention for college student heavy marijuana use.
      for individual outcomes, Marijuana e-CHECKUP TO GO participants significantly decreased weekly use as indicated by a significantly lower self-reported 6-week marijuana use when compared to participants in the HSM comparison condition. Among the covariates, sex did not predict either follow-up marijuana use or follow-up proportion time high while studying; however, baseline marijuana use and baseline proportion of hours high while studying were significantly associated with the respective 6-week follow-up variables. Moreover, a significant indirect effect (b = −0.03, SE = 0.02, MCCI = −0.064, −0.004) demonstrated that the Marijuana e-CHECKUP TO GO program effect on marijuana use was partially explained by decreased time high while studying.
      Table 3Indirect effect analysis results: treatment condition → proportion of time high while studying → marijuana use.
      Direct effects
      RREstimateSEP-Value
      Marijuana use at follow-up (negative binomial)
       Treatment condition0.82−0.190.070.01
       Periods of time high while studying at follow-up1.820.600.240.01
       Sex0.92−0.080.070.30
       Marijuana use at baseline1.080.080.010.00
      Periods of time high while studying at follow-up
       Treatment condition−0.050.020.01
       Periods of time high while studying at baseline0.800.060.00
       Sex0.000.020.94
      Indirect Effect
      EstimateS.E.MCCI
      Treatment condition → periods of time high while studying at follow-up → marijuana use at follow-up−0.030.02−0.064, −0.004
      Note: RR = rate ratio; SE = standard error; MCCI = Monte Carlo Confidence Intervals.

      4. Discussion

      Study results add to the limited research on web-based PNF approaches to marijuana use reduction among college students. Our previous research established a small direct effect of the Marijuana e-CHECKUP TO GO program on reducing cannabis use. What we did not know at the time was the mechanism of change for that direct effect. Results add to prior research by demonstrating that decreases in marijuana use are partly due to decreases in the proportion of time that college students are high while studying. Specifically, college students participating in Marijuana e-CHECKUP TO GO who reported frequent marijuana use reported 27% lower proportion of time high while studying compared to those in the healthy stress management condition at follow-up. Further, the indirect effects model showed that time spent high while studying mediated the relation between the treatment condition and marijuana use. In other words, the treatment condition decreased time spent studying while high, which decreased marijuana use. Future research should explore whether reductions in the amount of time studying high translates into increased achievement and retention.
      The Marijuana e-CHECKUP TO GO program did not decrease the proportion of time high while socializing, in class, or being physically active. One potential explanation for null findings in the socializing/partying context is that college students may choose to first decrease use within solitary contexts and/or those considered to be most normatively irresponsible (e.g., while studying) versus social contexts considered to be relatively normatively responsible (e.g., socializing/partying). College students who report frequent marijuana use may also initially reduce use within contexts where they feel most in control of their use (i.e., studying, which in many cases is a solitary activity) and are less likely to be socially sanctioned for not using (i.e., being social or partying with others who use heavily).
      Null findings for being high while in class are notable given academic similarities between studying and being in class. One potential explanation here is a floor effect in the proportion of time spent high while in class. Specifically, participants reported being high only 11% of the time while in class. This relatively low amount of time using in this context may have rendered it difficult to detect significant reductions in use. Future studies with larger samples and greater power to detect group differences will help to determine whether the Marijuana e-CHECKUP TO GO program can significantly reduce use while in class.
      Another notable finding was the lack of reduction in the time spent high while engaging in physical exercise. This may have resulted because the PNF intervention did not directly address physical exercise. Prior research has shown that the explicit recommendation of physical activity in brief cannabis intervention results in greater reductions in cannabis use (
      • Prince M.A.
      • Collins R.L.
      • Wilson S.D.
      • Vincent P.C.
      A preliminary test of a brief intervention to lessen young-adults’ cannabis use: Episode-level smartphone data highlights the role of protective behavioral strategies and exercise.
      ). Future adjustments to the intervention should address exercise or physical activity explicitly.
      This study shows the importance of measuring use during specific activities in which college students frequently participate for a better understanding of the collegiate contexts where use is decreasing, as well as for systematic intervention adaptation and improvement. For example, future efforts to decrease marijuana use in social situations, where use is most prevalent, could adapt the intervention to increasingly emphasize PBS, or alter messaging around PBS that can be used in social contexts including partying. Such program adaptations have the potential to enhance already promising intervention effects.

      4.1 Limitations

      Study results should be considered in light of study limitations. There was 24% attrition of participants at six-week post-test. Further, participants not completing post-test surveys reported using marijuana at significantly higher rates and were more likely to report being male. Therefore, whether PNF was an effective intervention approach for those using at the highest frequency is unclear. Data were self-reported, a strategy limited by several threats to the internal validity including recall, which may be of particular concern due the effects of frequent marijuana on certain aspects of memory (
      • Podsakoff P.M.
      • MacKenzie S.B.
      • Lee J.Y.
      • Podsakoff N.P.
      Common method biases in behavioral research: A critical review of the literature and recommended remedies.
      ). Related, the program developers developed the mediating variables, and they lack established psychometric properties. The relatively brief six-week interval between baseline and follow-up surveys was also a limitation, a longer follow-up window may have shown that the intervention was more effective. Finally, daily diary or ecological momentary assessment strategies would be beneficial to provide a more nuanced view of marijuana use and time spent high while engaging in a variety of activities.

      4.2 Conclusions

      This study builds on research that initially supported marijuana e-CHECKUP TO GO as a low-cost, easily diffused, web-based PNF approach to reduce marijuana use among college students. Importantly, the research reported herein demonstrates one pathway through which intervention effects were transmitted. Findings support that college students who report frequent use reduced their use while studying from 19% to 13% of the time, which mediated the relation between the treatment condition and self-reported marijuana use. Given that studying is likely to be more effective when sober compared to when high following marijuana use, the intervention may also be able to address concomitant academic difficulties associated with frequent marijuana use while in college. Study results describe a mechanism of change operating as a result of this PNF intervention. Frequency of marijuana use was the recruitment criteria for the current study, not a desire to reduce marijuana use. In other words, direct and indirect study effects occurred for individuals who were not specifically looking to alter their marijuana use. This may explain the effect size while also identifying a mechanism of change, among those who are not seeking to change their behavior.

      CRediT authorship contribution statement

      Each author contributed to the manuscript across the following domains. Mark A. Prince: conceptualization; methodology; data analysis; editing; draft preparation. Alexander J. Tyskiewicz: reviewing and editing. Bradley T. Conner: conceptualization; methodology; editing; draft preparation. Jamie E. Parnes: data curation; analyses; draft preparation. Audrey M. Shillington: reviewing and editing. Melissa W. George: reviewing and editing Nathaniel R. Riggs: conceptualization; methodology; data analysis; editing; draft preparation.

      References

        • Barnett N.P.
        • Murphy J.G.
        • Colby S.M.
        • Monti P.M.
        Efficacy of counselor vs. computer-delivered intervention with mandated college students.
        Addictive Behaviors. 2007; 32: 2529-2548
        • Batalla A.
        • Bhattacharyya S.
        • Yucel M.
        • Fusar-Poli P.
        • Crippa J.A.
        • Nogué S.
        • Martin-Santos R.
        Structural and functional imaging studies in chronic cannabis users: A systematic review of adolescent and adult findings.
        PLoS One. 2013; 8: e55821
        • Becker S.P.
        • Langberg J.M.
        • Luebbe A.M.
        • Dvorsky M.R.
        • Flannery A.J.
        Sluggish cognitive tempo is associated with academic functioning and internalizing symptoms in college students with and without attention deficit/hyperactivity disorder.
        Journal of Clinical Psychology. 2013; 70: 388-403
        • Bell R.
        • Wechsler H.
        • Johnston L.D.
        Correlates of college student marijuana use: Results of a US National Survey.
        Addiction. 1997; 92: 571-581
        • Blevins C.E.
        • Walker D.D.
        • Stephens R.S.
        • Banes K.E.
        • Roffman R.A.
        Changing social norms: The impact of normative feedback included in motivational enhancement therapy on cannabis outcomes among heavy-using adolescents.
        Addictive Behaviors. 2018; 76: 270-274
        • Bruffaerts R.
        • Mortier P.
        • Kiekens G.
        • Auerbach R.P.
        • Cuijpers P.
        • Demyttenaere K.
        • Kessler R.C.
        Mental health problems in college freshmen: Prevalence and academic functioning.
        Journal of Affective Disorders. 2018; 225: 97-103
      1. Dimeff L.A. Brief alcohol screening and intervention for college students (BASICS): A harm reduction approach. Guilford Press, 1999
        • Elliot J.C.
        • Carey K.B.
        • Vanable P.A.
        A preliminary evaluation of a web-based intervention for college marijuana use.
        Psychology of Addictive Behaviors. 2014; 28: 288-293
        • Elliott J.C.
        • Carey K.B.
        Correcting exaggerated marijuana use norms among college abstainers: A preliminary test of a preventive intervention.
        Journal of Studies on Alcohol and Drugs. 2012; 73: 976-980
        • Fachini A.
        • Aliane P.P.
        • Martinez E.Z.
        • Furtado E.F.
        Efficacy of brief alcohol screening intervention for college students (BASICS): A meta-analysis of randomized controlled trials.
        Substance Abuse Treatment, Prevention, and Policy. 2012; 7: 40
        • Ferrari S.L.P.
        • Cribari-Neto F.
        Beta regression for modelling rates and proportions.
        Journal of Applied Statistics. 2004; 31: 799-815
        • Filbey F.
        • Yezhuvath U.
        Functional connectivity in inhibitory control networks and severity of cannabis use disorder.
        American Journal of Drug and Alcohol Abuse. 2013; 39: 382-391
        • Francisco C.-N.
        • Zeileis A.
        Beta regression in R.
        Journal of Statistical Software. 2010; 34: 1-24
        • Hayes A.F.
        Beyond Baron and Kenny: Statistical mediation analysis in the new millennium.
        Communication Monographs. 2009; 76: 408-420
        • Hilbe J.M.
        Negative binomial regression.
        Cambridge University Press, 2011
        • Larimer M.E.
        • Cronce J.M.
        Identification, prevention, and treatment revisited: Individual focused college drinking prevention strategies 1999–2006.
        Addictive Behaviors. 2007; 32: 2439-2468
        • Larimer M.E.
        • Lee C.M.
        • Kilmer J.R.
        • Fabiano P.M.
        • Stark C.B.
        • Geisner I.M.
        • Neighbors C.
        Personalized mailed feedback for college drinking prevention: A randomized clinical trial.
        Journal of Consulting and Clinical Psychology. 2007; 75: 285-293
        • Meier M.H.
        • Caspi A.
        • Ambler A.
        • Harrington H.
        • Houts R.
        • Keefe R.S.
        • Moffitt T.E.
        Persistent cannabis users show neuropsychological decline from childhood to midlife.
        Proceedings of the National Academy of Sciences of the United States of America. 2012; 109: E2564-E2657
        • Murphy J.G.
        • Dennhardt A.A.
        • Skidmore J.R.
        A randomized controlled trial of a behavioral economic supplement to brief motivational interventions for college drinking.
        Journal of Consulting and Clinical Psychology. 2012; 80: 876-886
        • Muthén L.K.
        • Muthén B.O.
        Mplus user’s guide.
        7th ed. Los Angeles, CA1998–2017
        • Neighbors C.
        • Larimer M.E.
        • Lewis M.A.
        Targeting misperceptions of descriptive drinking norms: Efficacy of a computer-delivered personalized normative feedback intervention.
        Journal of Consulting and Clinical Psychology. 2004; 72: 434-447
        • Pearl J.
        Lord’s paradox revisited – (Oh Lord! Kumbaya!).
        2016 (Technical Report)
        • Pearson M.R.
        • Liese B.S.
        • Dvorak R.D.
        • Marijuana Outcomes Study Team
        College student marijuana involvement: Perceptions, use, and consequences across 11 college campuses.
        Addictive Behaviors. 2017; 66: 83-89
        • Pedersen E.R.
        • Huang W.
        • Dvorak R.D.
        • Prince M.A.
        • Hummer J.F.
        The Protective Behavioral Strategies for Marijuana Scale: Further examination using item response theory.
        Psychology of Addictive Behaviors. 2017; 31: 548-559
        • Podsakoff P.M.
        • MacKenzie S.B.
        • Lee J.Y.
        • Podsakoff N.P.
        Common method biases in behavioral research: A critical review of the literature and recommended remedies.
        Journal of Applied Psychology. 2003; 88: 879-903
        • Preacher K.J.
        • Selig J.P.
        Advantages of Monte Carlo confidence intervals for indirect effects.
        Communication Methods and Measures. 2012; 6: 77-98
        • Prince M.A.
        • Collins R.L.
        • Wilson S.D.
        • Vincent P.C.
        A preliminary test of a brief intervention to lessen young-adults’ cannabis use: Episode-level smartphone data highlights the role of protective behavioral strategies and exercise.
        Experimental and Clinical Psychopharmacology. 2020; 28: 150-156
        • R Core Team
        R: A language and environment for statistical computing, version 3.5.3. R Foundation for Statistical Computing, Vienna, Austria.
        • Ranganathan P.
        • Pramesh C.S.
        • Aggarwal R.
        Common pitfalls in statistical analysis: Intention-to-treat versus per-protocol analysis.
        Perspectives in Clinical Research. 2016; 7: 144
        • Riggs N.R.
        • Conner B.
        • Parnes J.E.
        • Prince M.
        • Shillington A.
        • George M.W.
        Marijuana e-CHECKUP TO GO: Effects of a personalized normative feedback intervention for college student heavy marijuana use.
        Drug and Alcohol Dependence. 2018; 190: 13-19
        • San Diego State University Research Foundation
        Marijuana eCHECKUP TO GO (eCTG) for universities & colleges.
        (Retrieved from)
        • 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. 1975–2016: Volume II, college students and adults ages 19–55. Ann Arbor: Institute for Social Research.
        The University of Michigan, 2017: 445
        • 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, 1975–2018: Volume II, college students and adults ages 19–60.
        The University of Michigan, Institute for Social Research2019
        • Selig J.P.
        • Preacher K.J.
        Monte Carlo method for assessing mediation: An interactive tool for creating confidence intervals for indirect effects [Computer software].
        (Retrieved from)
        http://quantpsy.org/
        Date: 2008, June
        • Smithson M.
        • Verkuilen J.
        A better lemon squeezer? Maximum-likelihood regression with beta-distributed dependent variables.
        Psychological Methods. 2006; 11: 54
        • Suerken C.K.
        • Reboussin B.A.
        • Egan K.L.
        • Sutfin K.G.
        • Wagoner J.S.
        • Wolfson M.
        Marijuana use trajectories and academic outcomes among college students.
        Drug and Alcohol Dependence. 2016; 162: 137-145
        • Zalesky A.
        • Solowij N.
        • Yucel M.
        • Lubman D.I.
        • Takagi M.
        • Harding I.H.
        • Seal M.
        Effect of long-term cannabis use on axonal fibre connectivity.
        Brain. 2012; 135: 2245-2255