Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Sep 29.
Published in final edited form as: Psychol Addict Behav. 2015 Jul 13;29(3):576–589. doi: 10.1037/adb0000098

Testing an Expanded Theory of Planned Behavior Model to Explain Marijuana Use among Emerging Adults in a Pro-Marijuana Community

Tiffany A Ito 1, Erika A Henry 1, Kismet A Cordova 1, Angela D Bryan 1
PMCID: PMC4586319  NIHMSID: NIHMS691333  PMID: 26168227

Abstract

Opinions about marijuana use within the US are becoming increasingly favorable, making it important to understand how psychosocial influences impact individuals’ use within this context. Here we use the Theory of Planned Behavior (TPB) to examine the influence of initial attitudes, norms, and efficacy to resist use on initial intentions, and then the effect of initial intentions on actual marijuana use measured one year later using data drawn from a community with relatively high use. We expanded the traditional TPB model by investigating two types of normative influence (descriptive and injunctive) and two types of intentions (use intentions and proximity intentions), reasoning that exposure to high use in the population may produce high descriptive norms and proximity intentions overall, but not necessarily increase actual use. By contrast, we expected greater variability in injunctive norms and use intentions and that only use intentions would predict actual use. Consistent with hypotheses, intentions to use marijuana were predicted by injunctive norms (and attitudes) and in turn predicted marijuana use one year later. By contrast, descriptive norms were relatively high among all participants and did not predict intentions. Moreover, proximity intentions were not predictive of actual use. We also found that increasing intentions to use over a one year period predicted greater use. Given the greater efficacy of theory-based as compared to non-theory-based interventions, these findings provide critical information for the design of successful interventions to decrease marijuana-associated harms.

Keywords: Theory of Planned Behavior, Substance Use, Marijuana


Recreational marijuana use became legal in multiple states in the US for the first time in history in 2012 (Colorado Amendment 64, 2012; Washington Initiative 502, 2012) and the majority of Americans favor legalization of marijuana (Gallup, 2013). The perception of risk associated with marijuana use has also declined, particularly among adolescents and emerging adults (Johnston, O’Malley, Miech, Bachman, & Schulenberg, 2014). This changing social landscape highlights the importance of better understanding the psychosocial process that affects marijuana use. This is important from a theoretical perspective, but also as a foundation from which to design effective interventions to reduce marijuana use, of which they are currently few, if any, especially among non-dependent users (e.g., Elliot, Carey, & Vanable, 2014; Stephens et al., 2007). Given the greater effectiveness of theory-based interventions as compared to those that are not theory-based (Glanz & Bishop, 2010), the present study examines marijuana use within the framework of the Theory of Planned Behavior (TPB), a powerful model for explaining health behavior, to determine whether the components of the theory are actually associated with both the intention to engage in a behavior and behavioral engagement per se.

Marijuana Use and the TPB

The TPB is a widely-used health behavior theory that provides both a robust and parsimonious account of the factors that predict health behaviors (c.f., Sutton, 2004). According to the TPB, behavior is most proximally predicted by intentions or motivations to engage in that behavior. Intentions are, in turn, predicted by attitudes and norms related to that behavior, as well as perceived behavioral control or efficacy over the behavior (Ajzen & Fishbein, 1980; Armitage Conner, Loach, & Willetts, 1999; Conner & McMillan, 1999; Fishbein & Ajzen, 1975; McMillan & Conner, 2003). In the model, attitudes reflect general positive and negative feelings toward the behavior in question, norms refer to subjective perceptions about the beliefs that important others hold toward the behavior (e.g., participants’ perceptions that important others think they should engage in the behavior), and perceived behavioral control refers to subjective perceptions about the ease of performing the behavior. In the context of substance use, attitudes and norms are usually measured with respect to engaging in the behavior (e.g., how positively one feels toward using marijuana), whereas perceived behavioral control is usually operationalized as efficacy to resist using the substance.

The TPB has been very successful in predicting use of a number of substances (e.g., Armitage et al., 1999; Conner, Warren, Close & Sparks, 1999). Moreover, recent meta-analytic work convincingly demonstrates that experimental manipulation of attitudes, norms, and self-efficacy results in theoretically predicted changes in intentions and behavior (Sheeran et al., 2015). Of particular importance, however, this recent meta-analysis failed to identify a single study experimentally testing the TPB in the domain of marijuana use. With the increasing favorability toward marijuana use and its legalization, it seems critical to examine the TPB as the basis for possible interventions to prevent or decrease marijuana use.

Although experimental manipulations of TPB constructs have not been done in the context of marijuana, studies have examined the degree to which TPB variables account for variability in marijuana use in correlational designs. These studies confirm the association of TPB constructs with marijuana use, showing that more positive attitudes, more positive perceived norms, and lower efficacy to refuse marijuana use predict greater intentions to use marijuana, and that intentions in turn predict marijuana use (Ajzen, Timko, & White, 1982; Armitage et al., 1999; Connor & McMillan, 1999; Hames, Evangeli, Harrop, & di Forti, 2012; Lac, Alvaro, Crano, & Siegel, 2009; Malmberg et al., 2012; McMillan & Connor, 2003). Most of these past studies, however, are limited by their reliance on samples with low absolute levels of marijuana use or lack of information about overall levels of use within the sample (Ajzen et al., 1982; Armitage et al., 1999; Connor & McMillan, 1999; Lac et al., 2009; Malmberg et al., 2012; McMillan & Connor, 2003). In addition to characterizing relations within a restricted range of possible use, these studies also cannot reveal how the model operates within a broader cultural context of favorability toward marijuana, which we refer to here as a pro-marijuana community. Together, these features of past studies make it difficult to determine the extent to which the TPB accounts for marijuana-related intentions and behavior across the full range of use levels and within a larger cultural context of relative favorability toward use. Both are important to consider given expected increases in use that are likely to accompany legalization, and the favorability toward marijuana use that is driving legalization.

While we generally expect the relations specified by the TPB to hold in a pro-marijuana community, we also think further specification of the standard theoretical constructs may be required when rates of use within the population are high. The operation of norms – which we expect to be relatively favorable within a community supportive of use – is one source of possible divergence from past research. Norms usually predict intentions, and by definition, communities favorable toward marijuana use should be perceived as possessing favorable norms. This might suggest that intentions will be correspondingly high. However, not everyone will intend to use a substance, even within a highly supportive environment. Thus, the extent to which norms predict intentions and subsequent use in a pro-marijuana environment is unclear.

We think the answer critically depends on the type of norm being considered. Past work has distinguished between two types of normative influence: descriptive and injunctive (Cialdini, Reno, & Kallgren, 1990; Deutsch & Gerard, 1955; Connor & McMillan, 1999). Descriptive norms refer to characterizations of behavior that are perceived as typical whereas injunctive norms refer to rules or beliefs about what one ought to do. Of importance, the two types of norms can diverge, and show differential impact on behavior (Cialdini et al., 1990; Connor & McMillan, 1999). The distinction may be particularly important to consider as marijuana use becomes more prevalent; perceiving that others around you are favorable toward marijuana use may increase perceptions of descriptive norms but not necessarily injunctive norms. That is, one could perceive that the behavior is typical yet not feel it is what one ought to do. For instance, while students may acknowledge high rates of marijuana use exist among their peers, they themselves may feel using marijuana will interfere with their school work and thus not intend to use it. Similarly, employees may feel that many of their peers outside work use marijuana yet not intend to use themselves because their employer prohibits marijuana use. From this analysis, relatively high levels of descriptive norms could exist in a pro-marijuana community alongside greater variability in injunctive norms. In recognition of the importance of these distinctions, more recent treatments of the TPB highlight the benefit of including both kinds of norms (Ajzen, 2011; Fishbein & Ajzen, 2010). This has been done in past studies of marijuana use, with variable results in the degree to which descriptive and injunctive norms predict marijuana use (Ajzen et al., 1982; Armitage et al., 1999; Connor & McMillan, 1999; Elliot & Carey, 2012; McMillan & Connor, 2003). Not all studies have measured both types of norms, and the level of marijuana use in the local context was not always reported so the degree to which the norms have different effects within a population of high use remains unclear.

Just as a pro-marijuana community might be associated with perceptions of more favorable descriptive norms, it might also increase expectations of being around marijuana use. It is not known, however, what the effects of expected proximity to marijuana use are on one’s own use. While merely being in the presence of other marijuana users does not necessarily imply personal use, peer usage is known to facilitate use of marijuana and other substances (Andrews, Tildesley, Hops, & Li, 2002; Ennett & Bauman, 1991, 1994; Simons-Morton & Farhat, 2010; Maxwell, 2002; Urberg, Degirmencioglu, & Pilgram, 1997). Moreover, the desire to make new friends and satisfy belonging needs (Baumeister & Leary, 1995) can be particularly high among emerging adults – the age range within which marijuana use is currently the highest (SAMHSA, 2011) – potentially intensifying the effects of peer use in this age range. This analysis suggests the benefits of separately examining intentions to use marijuana as well as intentions to be around those who are using.

In addition to differentiating between meaningfully separable aspects of norms and intentions, we also examine the temporal stability of intentions. Stability of intentions is considered a prerequisite for creating the causal link between intentions and behavior (Ajzen, 1991; Ajzen & Fishbein, 1980), is thought to represent the “strength” of intention-related cognitions, and is associated with stronger intention-behavior relationships (Conner, Sheeran, Norman, & Armitage, 2000; Sheeran & Abraham, 2003). Stability of intentions is of particular interest here both because our data were collected during a period of cultural change within the larger historical context and because of our interest in marijuana use among emerging adults just starting college, where the new environment can expose students to new situations, norms, and peers.

Overview and Hypotheses

The present research examines factors predicting marijuana use by expanding on the traditional TPB to more explicitly test the model across the full range of possible marijuana use (from never users to nearly daily users) and within the current cultural context of relatively favorable attitudes toward marijuana. We also do so across a relatively long time period, using attitudes, norms, and refusal self-efficacy (RSE) to prospectively predict marijuana use one year later. Given that adolescents and emerging adults are the most vulnerable to the negative consequences of heavy marijuana use (Bava & Tapert, 2010), including impairment of some aspects of memory and attention (Crane, Schuster, Fusar-Poli, & Gonzalez, 2013), and that marijuana use is particularly high among 18-to-25-year-olds (SAMHSA, 2011), we particularly focus on the ability of the TPB to predict actual marijuana use within this age range. In order to provide a test of the operation of the TPB within a context of relative favorability toward marijuana use, data were collected within a population of relatively high overall levels of use.

We hypothesize that attitudes, injunctive norms, refusal self-efficacy, and use intentions will be related to current marijuana use, with more favorable attitudes, injunctive norms, and use intentions among those who currently use more marijuana, but lower RSE as a function of use. However, because we expect both descriptive norms and proximity intentions to derive from perceptions of peer behavior, descriptive norms and proximity intentions may be relatively high among all participants in this pro-marijuana community, regardless of their own level of use. These hypotheses were tested cross-sectionally within initial TPB and marijuana use data assessed at a baseline measurement.

To test theoretical accounts of what predicts marijuana use, we assess how baseline TPB variables predict baseline intentions. In general accord with the TPB, attitudes, norms, and RSE should predict intentions. However, these relations may vary somewhat depending on the specific type of norms and intentions being examined. We specifically expect use intentions to be a function of attitudes, injunctive norms, and RSE. However, because we suspect that all participants will expect relatively high proximity intentions, we do not expect individual variability in attitudes, injunctive norms, and RSE to predict proximity intentions. To the degree that both descriptive norms and proximity intentions are influenced by the pro-marijuana cultural content, descriptive norms may be predictive of proximity intentions.

Finally, we examine the degree to which the two different types of intentions prospectively predict marijuana use one year later. Past research with the TPB suggests that intentions to use marijuana will predict actual marijuana use behavior. To the degree that proximity intentions suggest greater exposure to peer use, proximity intentions may also predict marijuana use. By contrast, if proximity intentions are elevated for all participants, expectations of peer use may be not predict own actual marijuana use. To the degree that stability over time indicates a stronger intention-relation relation, we expect the temporal stability of intentions to independently predict behavior.

All hypotheses were separately assessed for males and females to determine the degree to which gender affects psychosocial influences on marijuana use. We had two primary motivations for doing so. The first is that gender differences in motivations to engage in substance use, use trajectories, consequences of use, barriers to treatment, and patterns of relapse have been observed for other substances (Brady & Randall, 1999; Schepis et al., 2011; Walitzer & Dearing, 2006) but relatively few studies have examined gender differences in aspects of marijuana use. Moreover, those that have report mixed results (Kandel & Chen, 2000; Ridenour, Lanza, Donny, & Clark, 2006; Schepis et al., 2011; Wagner & Anthony, 2007), clearly indicating the need for more research. Our second motivation derived from findings suggesting that females may be more sensitive to some of the psychosocial variables measured here. In particular, females are sometimes found to be more sensitive to peer influences than males (e.g., Brown, 1982; Mizuno, Seals, Kennedy, & Myllyluoma, 2000; Rudolph & Conley, 2005). This could result in a bigger impact of normative influence on marijuana use for females than males. At the same time, meta-analysis indicates that the absolute size of the gender difference in social influence is quite small (Eagly & Carli, 1981). Given the mixed results in past research, we had no basis for strongly predicting gender differences in the present research. However, given potential implications for prevention and treatment, it is important to determine whether gender differences exist.

Method

Participants

Participants were 370 University of Colorado students taking part in a three-year longitudinal study (see Table 1 for sample characteristics). As evidence of marijuana favorability in this context, students from the university from which we recruited reported using marijuana within the past 30 days at more than double the rate (38.2%) of other US college students (15.8%) (American College Health Association, 2011). In order to ensure a sampling of a wide range of use, participants were recruited as never users (i.e., never tried marijuana), infrequent marijuana users (i.e., used marijuana four times or less per month for less than three years) and frequent marijuana users (i.e., used marijuana an average of 5 days a week or more for at least the past year), with eligibility determined via phone interview prior to enrollment. Although we wished to sample a full range of recreational marijuana users, we did not specifically focus on recruiting dependent users. Examination of responses to the Marijuana Dependence Scale (Stephens, Roffman, & Curtin, 2000), a scale based on Diagnostic and Statistical Manual of Mental Disorders (4th ed.) criteria for dependence (e.g., “When I smoked marijuana, I often smoked more or for longer periods of time than I intended”; “I need to smoke more marijuana to achieve the same ‘high’”) that was completed during the first experimental session showed that even though marijuana use was high among the frequent marijuana users in this sample, frequent users on average endorsed only 4.30 of 10 symptoms of dependence. Finally, because the full protocol also included electroencephalography (EEG) measures, individuals who reported a history of head trauma, neurological disorder, or the use of prescription medication (with the exception of oral contraceptives or medical cannabis) were excluded from the study. Only the measures of interest to our current hypotheses will be described in detail, but where appropriate (e.g., when they preceded the measures of interest), other measures collected will be noted.

Table 1.

Sample Characteristics by Marijuana Use Group Collapsed Across Subject Gender

Marijuana Use Group
Never Infrequent Frequent F or χ2 Value
Demographics
 Year 1 N 126 145 99
 Year 2 N 114 133 91
 Gender (% female) 50.0% 51.7% 51.5% 0.09
 Age (Year 1) 18.30 (0.46) 18.39 (0.52) 18.34 (0.50) 0.99
 Ethnicity (% White) 70.6% 71.7% 79.6% 2.63

Year 1 TPB
 Attitudes 1.46 (0.80)a 3.14 (1.30)b 4.91 (0.92)c 299.48**
 Descriptive Norms 4.90 (1.24)ab+ 4.67 (1.09)a+ 4.97 (0.90)b 2.58+
 Injunctive Norms 2.29 (1.07)a 3.74 (1.11)b 5.06 (1.17)c 173.11**
 RSE 5.90 (0.21)a 5.40 (0.56)b 3.89 (1.13)c 245.82**
 Use Intentions 1.15 (0.66)a 3.79 (1.71)b 6.43 (0.69)c 549.49**
 Proximity Intentions 4.71 (1.81)a 6.31 (0.89)b 6.89 (0.30)c 103.72**

Temporal Stability Year 1 to Year 2
 Use Intentions 0.33 (1.20)a+ 0.00 (1.69)a+ −0.42 (1.14)b 7.33**
 Proximity Intentions 0.24 (1.70)a 0.07 (1.34)a −0.06 (0.84)a 1.28

Marijuana Use
 Days of Year 1 Use 0.00 (0.00)a 1.68 (1.90)b 25.95 (3.74)c 4430.84**
 Days of Year 2 Use 0.37 (1.13)a 3.35 (4.98)b 23.76 (7.99)c 586.49**

Note. RSE = Refusal Self-Efficacy. Numbers in parentheses are standard deviations. Marijuana use reflects the number of days the substance was used in the 30 days prior to the first session in both Year 1 and Year 2 laboratory appointments and so have ranges of 0–30. Other possible ranges are (1) Attitudes, Norms, Use Intentions, and Proximity Intentions 1 – 7, (2) RSE 1 – 6, (3) Temporal Stability of Use Intentions −4.5 – 4.5, and (4) Temporal Stability of Proximity Intentions −5 – 5. Higher values indicate greater attitude positivity, norm positivity, ease of abstaining, and increasing use and proximity intentions from Year 1 to Year 2. F and χ2 values reflect the test of the omnibus Marijuana Use Group main effect. Means within the same row with different letter subscripts differ at p ≤ .05.

*

p ≤ .05

**

p ≤ .001. Main effects and pairwise comparisons with + differ marginally at p ≤ .10.

Of the 369 participants who provided racial information, 2 identified as Black, 13 as Asian, 12 as Hispanic, 1 as Pacific Islander, 3 as East Indian, 1 as Middle Eastern, 66 as multi-racial, and 271 as White. Eight participants were sophomores but the remaining 98% were freshman. (Of the eight sophomores in the sample, two each were never and infrequent users, and four were frequent users.) Thirty two participants included in analyses did not return to complete Year 2. Those who failed to return in Year 2 did not differ from those who did in the demographic characteristics of gender, age, race, or marijuana use group, nor on the TPB variables of attitudes, injunctive norms, RSE, use intentions, or proximity intentions (all p’s > .21). The only significant difference was on descriptive norms, with those who were retained in the study reporting lower descriptive norms in Year 1 (M = 4.78) than those who failed to return for Year 2 (M = 5.34, t(41)=3.44, p = .001). Of these variables, the lack of a difference on marijuana use is especially critical because it indicates that there was no differential attrition on the construct of primary interest. We are uncertain about the meaning of the difference in descriptive norms, especially given that it was not accompanied by differences on other TPB constructs. An additional nine participants were initially enrolled but dropped from the study because they either were determined to have provided incorrect marijuana use information at enrollment (n=4) or failed to return for their second session in Year 1 (at which the TPB was measured).

Self-Report Measures

Marijuana use

Marijuana use was assessed using the Time-Line Follow Back (TLFB; Sobell & Sobell, 1992), a calendar assisted structured interview in which participants were asked to indicate over the past 30 days the quantity of marijuana used. Marijuana use was quantified as total number of days participants reported using in the 30 days prior to their laboratory session.

Theory of Planned Behavior (TPB)

TPB variables were assessed with measures adapted from Bryan, Rocheleau, Robbins, and Hutchison (2005) and following Ajzen (1991). General attitudes toward smoking marijuana use were assessed with five items (e.g., “For me, smoking marijuana regularly in the next 12 months would be…”). All items were measured on a 7-point scale anchored with Bad/Good, Unpleasant/Pleasant, Valuable/Worthless, Harmful/Beneficial, and Healthy/Unhealthy. A single mean attitude score was created with higher scores reflecting more positive attitudes about marijuana (α = .93).

Descriptive norms for marijuana use were assessed with two items that reflect the perceived prevalence of marijuana use (“Most people my age have tried marijuana,” “Most people my age smoke marijuana regularly”). A single mean descriptive norm score was created with higher scores reflecting greater prevalence of marijuana use (r = .58, p < .01). Injunctive norms for marijuana use were assessed with three items that address perceptions of what others believe one ought to do (“My friends think I should not smoke marijuana,” “My parents think I should not smoke marijuana,” “People who are important to me think I should not smoke marijuana”). These items were reverse scored, and a single mean injunctive norm score was created with higher scores reflecting greater acceptance of use among close others (α = .80). Both norm constructs were rated on a 7-point scale ranging from “Strongly Disagree” to “Strongly Agree.”

Nineteen items assessed refusal self-efficacy for marijuana (e.g., “I feel confident that I could refuse to smoke marijuana if a friend offered it to me”) on a 6-point scale ranging from “Very sure I could NOT resist” to “Very sure I could resist.” A single mean refusal self-efficacy (RSE) score was created with higher scores reflecting greater ease abstaining from marijuana use (α = .97).

The marijuana use intentions scale was comprised of 3 items assessing readiness to use (likelihood of smoking marijuana, purchasing marijuana paraphernalia, buying marijuana). A single mean use intention score was created with higher scores indicating greater intentions to use marijuana (α = .91). The marijuana proximity intentions scale was comprised of 4 items that address willingness to be in situations where marijuana is present (e.g., hanging out with friends who use marijuana, going to parties where people will be smoking marijuana). A single mean proximity intention score was created with higher scores indicating greater intentions to be in situations where others are using marijuana (α = .90). All intention items were assessed on a 7-point scale from “Not at all likely” to “Very Likely.”

The directional temporal stability of intentions (TSI) was calculated by a procedure adapted from Broaddus, Schmiege, and Bryan (2011). The TSI was based on the average of two indices assessing change in intentions between Year 1 and Year 2. For the first index, we computed the difference between Year 1 and Year 2 intention values for each individual intention item, calculated as Year 2 – Year 1 for each item. We then computed a mean change in intentions across all items. For the second index, we obtained a count of the number of intention items that changed between Year 1 and Year 2, taking into account whether that change was positive or negative such that any decrease in intentions resulted in subtracting one from the total, and an increase in intentions resulted in adding one to the total. Because both are face valid of measures of stability, we simply took the average of the two, calculating a separate TSI score for use intentions (TSI-use) and proximity intentions (TSI-proximity). A score of zero indicates no change from Year 1 to Year 2, while positive values reflect increases in intentions from Year 1 to Year 2 and negative values indicate decreasing intentions from Year 1 to 2.

Procedure

Potential participants were recruited via email invitations to their university account and advertisements on campus. Those who were interested in the study were initially interviewed on the phone by study personnel to determine eligibility (see eligibility criteria in Participants). Participants who met criteria for inclusion were invited to participate in two sessions a year for three total years. Data in the present analyses come from the first four sessions. Because our interest is in how psychosocial factors predict actual use, our primary analyses use attitudes, norms, RSE, and intentions measured in Year 1 to predict use in Year 2. We focus specifically on data from the first two years of assessment because this period captures the first two years of college for the majority of our students, a time when marijuana use risk is increased (e.g., Fromme, Corbin, & Kruse, 2008) and students are potentially exposed to new social influences. It is also a period where young adults begin taking more personal responsibility for life choices. The first two laboratory sessions at which baseline data were obtained occurred on average within 5.77 days of each other (SD = 5.12). The first Year 2 session occurred approximately 12 months after the first Year 1 session (M = 362.77 days, SD = 19.30 days). The second Year 2 session occurred on average 5.79 days later (SD = 6.53). Participants were instructed to abstain from alcohol for 24 hours, recreational drugs (including marijuana) for 6 hours, and caffeine and cigarettes for 1 hour prior to each laboratory session.

In the first session in Year 1, participants were consented and then breathalyzed to ensure a breath alcohol concentration (BAC) of zero. Adherence to other abstinence requirements was verified verbally. Participants then completed the TLFB followed by a number of questionnaires not relevant to the present analyses. Upon returning for their second laboratory session participants were breathalyzed to confirm a BAC of zero, given a urine toxicology screen to help assess compliance with our other abstinence requirements1, then had electrodes affixed for EEG recording. Participants then completed a go/no-go and gambling task (Yeung & Sanfey, 2004) not relevant to the present hypotheses, followed by the TPB survey.

Participants returned for two follow-up sessions approximately 1 year after their baseline sessions. Procedures were similar to those in Year 1, with current marijuana use assessed in the first session and TPB assessed in the second session. In Year 2, only intention items from the TPB measure are of interest to the current hypotheses as reassessing these constructs allowed us to compute TSI values. TPB measures were preceded by six tasks not relevant to the current hypotheses. These included two switching tasks (i.e., two category switch, four category switch; Kiesel et al., 2010) followed by an additional four tasks taken from the Kit of Factor-Reference Cognitive Tests (the Hidden Figures Test, Identical Figures Test, Making Groups Test, and Different Uses Test; Ekstrom, French, Harman, & Dermen, 1976) prior to completing the TPB measures. Participants received $25, $65, $25, and $75 for sessions 1–4, respectively. The full protocol was approved by the University of Colorado IRB.

Analytic Strategy

We first performed preliminary descriptive analyses to assess the relation between marijuana use and TPB variables, with separate analyses representing marijuana use either categorically or continuously. Categorical analyses were done with one-way ANOVAs using a 3-level Marijuana Use Group variable based on the participant’s use at time of study enrollment (i.e., never, infrequent, frequent). Continuous analyses consisted of bivariate correlations between number of days marijuana was used in past 30 days from the TLFB and TPB variables. These latter analyses were performed with marijuana use obtained in both Years 1 and 2. To assess the impact of gender, ANOVAs were also conducted adding a Participant Gender variable and bivariate correlations were computed separately for men and women. Marijuana Use Group × Participant Gender ANOVAs were also conducted with TLFB marijuana use as the dependent measure to determine whether use in Years 1 and 2 differed for men and women. Finally, Cohort was entered into the ANOVAs to assess whether date of data collection with respect to passage of laws affecting marijuana legalization affected responses.

Our primary theoretical questions concerning the impact of psychosocial factors on marijuana use were assessed with path models estimated with Mplus Version 7.0 (Muthén & Muthén, 2012) using full information maximum likelihood (FIML) estimation of missing data. Because the model χ2 tests are sensitive to sample size, we follow established guidelines and general benchmarks by primarily assessing model fit based on comparative fit index (CFI > .90), root mean square error of approximation (RMSEA < .08), and standardized root mean square residual (SRMR < .08). We first evaluate a basic model assessing how baseline (Year 1) attitudes, descriptive norms, injunctive norms, and RSE predict baseline use intentions and proximity intentions, and how the two types of intentions in turn predict Year 2 use. These models also control for Year 1 use. We next examine the effect of stability in proximity and use intentions over the year interval with expanded models that include the temporal stability of use and proximity intentions as predictors of marijuana use. For both types of models, separate models were estimated for males and females to allow assessment of whether the processes that predict marijuana are the same in both genders.

Results

Preliminary Analyses

Table 1 shows basic descriptive data from our sample collapsed across subject gender. Separate Marijuana Use Group (never, infrequent, frequent) one-way ANOVAs on TPB variables revealed significant Marijuana Use Group effects on attitudes, injunctive norms, RSE, and both intention types such that attitude positivity, acceptance of use among close others (i.e., injunctive norms), and use and proximity intentions increased as a function of use. As expected, Marijuana Use Group had the opposite effect on RSE, with frequent users reporting the lowest refusal self-efficacy (i.e., the most difficulty resisting marijuana use) and never users reporting the highest refusal self-efficacy.

TSI-use also showed an inverse relation with Marijuana Use Group, with frequent users showing decreasing intentions from Year 1 to Year 2 relative to never and infrequent users. The direction of these effects makes sense given that frequent users were near the scale maximum for use intentions in Year 1 (M = 6.43), making change in intentions among these participants most likely in the direction of decreasing intentions to use. Of course, frequent participants could have reported no intention to change their use, so the results are noteworthy for indicating that on average our frequent users expressed a slight desire to decrease their use.

Of interest, neither descriptive norms nor TSI-proximity intentions varied as a function of Marijuana Use Group. The omnibus Marijuana Use Group main effect was marginally significant for descriptive norms, and differences among Marijuana Use Groups were computed for exploratory purposes (see Table 1). Given the power afforded by the sample size and the overall high means among all types of users, we are cautious about interpreting these as indicative of meaningful differences in descriptive norms as a function of Marijuana Use Group. For TSI-proximity, the pattern of means was directionally similar to the TSI of use intentions (decreasing intentions among heavier users) but the difference was not statically significant.

To examine potential gender effects and their interaction with marijuana use, we also conducted separate 3 (Marijuana Use Group) × 2 (Participant Gender) ANOVAs on attitudes, norms, RSE, intentions, and TSI. There were Subject Gender main effects on two TPB variables: attitudes (F (1, 363) = 5.00, p < .05) and use intentions (F (1, 364) = 5.06, p < .05). Males reported more positive attitudes (M = 3.29) and use intentions (M = 3.94) than females (M = 3.05 and 3.66, respectively).

The only TPB variables for which there were Marijuana Use Group × Subject Gender interactions were attitudes (F (2, 363) = 4.73, p < .01) and injunctive norms (F (2, 364) = 3.43, p < .05). It is important to note that males and females showed the same general effects of Marijuana Use Group for both variables (see Supplemental Table 1). That is, both attitudes and injunctive norms were the least positive among never users, of intermediate positivity for infrequent users, and the most positive among frequent users for both males and females. The simple effect of Marijuana Use Group is significant for both males and females for both attitudes and injunctive norms (Table S1). For attitudes, the interaction is attributable to significantly more positive attitudes among male infrequent users than female infrequent users (t(363)=3.68, p < .001). For injunctive norms, tests of the simple gender effects within each level of use showed more positive injunctive norms among female frequent users than male frequent users (t(364)=−2.38, p < .02). The pattern of Marijuana Use Group effects did not differ for males and females for descriptive norms, RSE, use intentions, proximity intentions, TSI-use, and TSI-proximity but means for these variables are shown separately as a function of Marijuana Use and Gender for descriptive purposes in Table S1.

We also performed 3 (Marijuana Use Group: never, infrequent, frequent) × 2 (Subject Gender: male, female) ANOVAs on days of marijuana use, quantified from the TLFB. There were significant Marijuana Use Group main effects on days of marijuana use in both Years 1 and 2. In both cases, the expected linear increase as a function of Marijuana Use Group was observed (Table S1). There were also significant Subject Gender main effects for both variables, with males reporting more days of use in both years, (F (1, 364) = 10.83, p = .001 and (F (1, 332) = 16.65, p < .001, respectively). The Marijuana Use Group × Subject Gender interactions were also significant for both variables, (F (2, 364) = 3.54, p < .05 and (F (2, 332) = 6.26, p < .005, respectively). Tests of the simple gender effects within each level of use showed more days of marijuana use in both Year 1 and Year 2 among males than females for both infrequent and frequent users (see Table 2) (Year 1: Infrequent: t (364) = 1.95, p = .052; Frequent: t (364) = 3.56, p < .001; Year 2: Infrequent: t (332) = 3.43, p = .001; Frequent: t (332) = 4.02, p < .001). The interaction seems to reflect gender differences in the size of the differences between use groups.

Table 2.

Bivariate Correlations Among Variables

Variable 1 2 3 4 5 6 7 8 9 10
Males
1. Year 1 Marijuana Use
2. Year 2 Marijuana Use   0.92**
3. Attitudes   0.66**   0.67**
4. Descriptive Norms   0.01   0.07 −0.04
5. Injunctive Norms   0.54**   0.57**   0.68**   0.05
6. RSE −0.70** −0.66** −0.58** −0.05 −0.43**
7. Use Intentions   0.76**   0.79**   0.80** −0.06   0.67** −0.66**
8. Proximity Intentions   0.40**   0.43**   0.50**   0.06   0.58** −0.34**   0.53**
9. TSI-Use −0.20* −0.05 −0.21**   0.06 −0.09   0.16* −0.28** −0.03
10. TSI-Proximity −0.07 −0.03 −0.02 −0.11 −0.07   0.06 −0.02 −0.26**   0.18*

Females
1. Year 1 Marijuana Use
2. Year 2 Marijuana Use   0.89**
3. Attitudes   0.77**   0.71**
4. Descriptive Norms   0.14+   0.14+   0.00
5. Injunctive Norms   0.67**   0.59**   0.72**   0.12
6. RSE −0.82** −0.74** −0.69** −0.11 −0.58**
7. Use Intentions   0.77**   0.70**   0.85**   0.04   0.70** −0.76**
8. Proximity Intentions   0.44**   0.39**   0.49**   0.12+   0.59** −0.45**   0.55**
9. TSI-Use −0.14+   0.03 −0.23** −0.05 −0.22**   0.19* −0.38** −0.19*
10. TSI-Proximity −0.07   0.02 −0.07 −0.06 −0.17*   0.11 −0.07 −0.33**   0.26**

Note. RSE = Refusal Self-Efficacy. TSI-Use = Temporal Stability of Use Intentions; TSI-Proximity = Temporal Stability of Proximity Intentions.

+

p ≤ .10

*

p ≤ .05

**

p ≤.01.

Data were collected from August 2009 to May 2013, a period spanning passing of new state regulations formalizing a system for medical marijuana dispensaries (May, 2010) and passage of a state amendment legalizing possession and sale of marijuana for recreational use for those 21 and over (November, 2012) (data collection ended before the legalized sale of marijuana began in January, 2014). Additional analyses were conducted to assess potential effects of cohort on our outcomes of interest by including it as a factor in the ANOVAs, defined as initial study enrollment in 8/09–5/10, 8/10–5/11, or 8/11–5/12. There were no effects of cohort, either as a main effect or interaction with the categorical Marijuana Use Group variable, on any TPB variables or on TLFB use in either year. This indicates stability of psychosocial variables and marijuana use within our sample across a three year span.

In addition to assessing the relationship between our categorical Marijuana Use Group variable and TPB constructs, we also examined the relationship between continuous measures of marijuana use and TPB constructs. Table 2 presents the correlations between TPB variables and Year 1 and Year 2 marijuana use from the TLFB. Correlations are presented separately for males and females. Results were largely similar for males and females and reflect the same pattern of relations between marijuana use and TPB variables reported in Tables 1 and S1. Specifically, Year 1 marijuana use was related to all TPB constructs for males and females except descriptive norms and TSI-proximity (although the correlation of descriptive norms with use was marginally significant for females). Similar relations were seen between Year 2 marijuana use and TPB constructs, with Year 2 marijuana use significantly correlating with all TPB constructs for males and females except descriptive norms, TSI-use, and TSI-proximity (with the correlation between of descriptive norms with Year 2 use again marginally significant for females but not males).

Use and Proximity Intentions Predicting Year 2 Marijuana Use

We next evaluated our primary theoretical questions about the degree to which psychosocial variables predict marijuana use with the path models in Figure 1. These models are similar to traditional TPB models, but with measures of two separate aspects of norms (injunctive and descriptive) and intentions (use and proximity). The question of interest was how baseline (Year 1) attitudes, descriptive norms, injunctive norms, and RSE predict baseline use intentions and proximity intentions, and how Year 1 intentions in turn predict Year 2 use, controlling for Year 1 use. The model exhibited good fit to the data for males: χ2 (6, N = 181) = 33.16, p < .001; CFI = .97, RMSEA = .16 (90% confidence interval of RMSEA (.11, .21)), and SRMR = .03. Model variables accounted for 73% of the variance in use intentions, 36% of the variance in proximity intentions, and 86% of the variance in Year 2 marijuana use. Model fit was also good for females: χ2 (6, N = 189) = 2.68, p = .85; CFI = 1.00, RMSEA < .001 (90% confidence interval of RMSEA (.00, .05)), and SRMR = 01. Model variables accounted for 79% of the variance in use intentions, 37% of the variance in proximity intentions, and 79% of the variance in Year 2 marijuana use. As can be seen, the basic framework of the TPB was confirmed, with more positive attitudes, more positive injunctive norms, and lower RSE associated with greater use intentions among both males and females. Of interest, descriptive norms were not related to use intentions for either gender. Proximity intentions were similarly predicted by more positive attitudes and injunctive norms among males, but only by injunctive norms for females. There was no relation of RSE or descriptive norms to proximity intentions for either males or females. The latter result contradicts our expectation that high perceived descriptive norms would be associated with high intentions to be around marijuana use. Controlling for baseline marijuana use (which was a large predictor of use in Year 2), use intentions significantly predicted marijuana use 12 months later but proximity intentions did not. However, this was only the case for males. Among the females in our sample, baseline marijuana use was a significant predictor of Year 2 use, but use intentions were not.

Figure 1.

Figure 1

Model of Year 2 marijuana use. Numbers on arrows are standardized loadings. Numbers on boxes are the amount of variance accounted for by the model. Numbers before slash represent values for males, while numbers following the slash reflect values for females. *p ≤ .05, **p ≤ .001, ***p ≤ .001. +p = .08.

Temporal Stability of Intentions Predicting Year 2 Marijuana Use

Figure 1 shows that intentions to use marijuana predict actual marijuana use a year later among males, but that intentions to be around others who are using marijuana are unrelated to Year 2 use among both males and females. We next wanted to assess how stability of intentions over a year interval predicts use so we expanded on the model in Figure 1 to include two additional constructs: the temporal stability of use intentions and the temporal stability of proximity intentions (see Figure 2). As before, this model was estimated separately for males and females. The model exhibited good fit to the data for both genders. For males, χ2 (8, N = 181) = 34.30, p < .001; CFI = .98, RMSEA = .14 (90% confidence interval of RMSEA (.09, .18)), and SRMR = .02. Model variables accounted for 89% of the variance in marijuana use 12 months later. For females, χ2 (8, n = 189) = 6.98, p =.54; CFI = 1.00, RMSEA < .001 (90% confidence interval of RMSEA (.00, .08)), and SRMR = .01. Model variables accounted for 83% of the variance in marijuana use 12 months later. Note that the paths from Year 1 Marijuana Use to attitudes, injunctive norms, descriptive norms, and RSE assess identical relations to those assessed in Figure 1, as do the paths from attitudes, injunctive norms, descriptive norms, and RSE to use intentions and proximity intentions. Those path coefficients are accordingly the same in both sets of models. Of primary interest in this model are the paths from the TSI variables to Year 2 use. As can be seen, TSI-use predicted Year 2 use, with increasingly positive use intentions associated with greater Year 2 marijuana use for both males and females. TSI-proximity, in contrast, was unrelated to use for both males and females. Of interest, when the two TSI variables are included, use intentions in Year 1 predicted Year 2 marijuana use for both males and females.

Figure 2.

Figure 2

Model of Year 2 marijuana use including temporal stability of intentions. Numbers on arrows are standardized loadings. Numbers on boxes indicate the amount of variance accounted for by the model. Solid lines reflect paths for which we had hypotheses; dotted lines indicate paths without hypothesized relationships. Numbers before slash represent values for Males, while numbers following the slash reflect values for Females. *p ≤ .05, **p ≤ .01, ***≤ .001. +p = .08.

We did not have hypotheses about the relation of attitudes, injunctive norms, descriptive norms, and RSE to TSI-Use and TSI-Proximity (to indicate this, these paths are shown by dotted lines in Figure 2). There was only one path among this set that reached conventional levels of significance: for females, more positive injunctive norms were associated with increasingly positive TSI-Proximity. This path was not significant for males. Attitudes were a marginally significant predictor of TSI-Use for males, with more positive attitudes in Year 1 associated with decreasing intentions to use marijuana in Year 2.

Discussion

The increasing acceptance of marijuana use in the U.S. highlights the benefit of understanding factors that predict its use, both from a basic theoretical perspective, and also to inform the design of effective interventions. Our basic descriptive analyses first provide a unique snapshot within the current context of relative favorability toward marijuana of factors known to predict substance use. Level of marijuana use was linearly related to many but not all of our measured variables. At levels of higher use, both male and female participants reported (1) more positive attitudes toward marijuana, (2) more positive injunctive norms, (3) lower confidence in their ability to refuse using marijuana, (4) higher intentions to use marijuana, and (5) higher intentions to be in the proximity of people using marijuana. We think it notable that descriptive norms did not show this association with use and instead all participants perceived relatively favorable descriptive norms. That injunctive norms did differ as a function of use supports our prediction that the changing status of marijuana use may create a differentiation in types of normative perceptions, with relatively high perceptions of whether marijuana use is typical but greater variability in whether individuals feel it is something they ought to do. Although proximity intentions did vary as a function of current use, with higher intentions to be around marijuana users among heavier users, proximity intentions were also relatively high among the other use groups. Thus, even individuals who had never used marijuana expected to be around marijuana use. Another noteworthy feature of the descriptive results is the degree of similarity for males and females. The Marijuana Use Group main effect on attitudes and injunctive norms was moderated by gender, but in both cases, males and females showed similarly more positive attitudes and injunctive norms as a function of marijuana use.

Given the predictive relation between intentions and behavior observed in other domains (e.g., Webb & Sheeran, 2006), we examined the relation of attitudes, norms, and RSE to intentions to use and be in the proximity of marijuana, again seeing a difference between injunctive and descriptive norms. Consistent with the TPB framework, attitudes, injunctive norms, and RSE predicted use intentions among both males and females, respectively explaining 73% and 78% of the variance in our path models. By contrast, descriptive norms did not predict intentions to use marijuana, suggesting that perceiving high peer use does not necessarily increase one’s own intentions to use. In fact, there was a marginally significant relationship for males showing the opposite effect such that perceptions of higher peer use were associated with lower use intentions. The small size of this effect (standardized path coefficient = −.07) suggests more work should be done before conclusions are drawn about this relationship other than to say that high perceptions of peer marijuana use do not necessarily translate into high use intentions. Contrary to use intentions, intentions to be in the proximity of marijuana users were associated only with more positive injunctive norms among both males and females, and with more positive attitudes for males.

Considering the important question of what predicts marijuana use a year later, there were significant bivariate correlations between use intentions expressed in Year 1 and Year 2 marijuana use for both males and females. However, when amount of use in Year 1 was included as a predictor within the basic TPB model in our path models (Figure 1), the relation between use intentions and Year 2 use was significant only for males. By contrast, proximity intentions did not predict Year 2 use for either males or females in our path models. The latter finding supports our prediction that the changing status of marijuana use may create differences in the way different types of intentions affect behavior, with actual marijuana use less affected by intentions to be around others who are using, particularly in an environment when expectations of being around marijuana use are relatively high.

In addition to the basic TPB model predicting future marijuana use, we also examined the effect that the directional temporal stability of intentions may have in predicting future marijuana use. Consistent with the assumption that stability of intentions is a prerequisite for creating the causal link between intentions and behavior (Ajzen, 1991; Ajzen & Fishbein, 1980), considering TSI showed that both use intentions in Year 1 and change in use intentions from Year 1 to Year 2 significantly predicted Year 2 marijuana use among males and females. By contrast, neither proximity intentions nor the stability of those intentions predicted Year 2 use.

Gender Effects

Analyses were stratified by gender to assess the degree to which psychosocial variables explaining marijuana use differ or are the same for males and females. There were relatively few gender differences, suggesting similarity in the degree to which attitudes, norms, and RSE affect intentions to use and be around marijuana for males and females. However, one striking divergence was the absence of a relation between Year 1 use intentions and Year 2 use among females in the path model when the TSI variables were omitted. The effect of past marijuana use was particularly strong for females in the model omitting the TSI variables (standardized path coefficient of .85 compared to .77 for males). Inclusion of this strong predictor may have overwhelmed the impact of use intentions on actual use for females. However, this path from use intentions to use was significant for both males and females when the TSI variables were included (and the bivariate correlation between use intentions and Year 2 use was sizeable for both genders). We are presently unable to determine the exact reason for the different relation between use intentions and Year 2 use for males and females in some models. Thus, while we highlight the many similarities in how the TPB model fit for males and females, we also note there may be subtle differences between males and females in how the variables relate to each other that suppress the effects of use intentions on Year 2 use in some situations. The potential sources of these gender differences require additional research.

Different Types of Norms and Intentions

We outlined our expectation that participants would be sensitive to the relatively high rates of marijuana use in their community, making it relevant to assess different kinds of norms and intentions. Results confirmed our predictions by showing different effects for different types of norms and intentions, both at the mean level and in terms of their predictive relation with use. This highlights the benefits of separately assessing descriptive and injunctive norms, and use and proximity intentions. Given their different predictive relations in this context, failing to separately assess them may obscure theoretically and practically (e.g., for interventions) important subtleties in the way norms and intentions affect marijuana use.

As we consider the potential impact of increasing marijuana use and legalization, our data suggest that merely expecting to be around high levels of use will not necessarily lead to increases in one’s own use. Likewise, perceiving high levels of use around one does not necessarily increase intentions to use. However, while we found no direct link between high descriptive norms or proximity intentions and own use, perceiving high use may not be completely benign. Among those who are never or infrequent users, perceiving high descriptive norms and/or having high proximity intentions could contribute to feeling poor social fit. Because belonging and acceptance have broad effects on psychological and physical well-being (e.g., Baumeister & Leary, 1995), this raises the possibility that high descriptive norms and proximity intentions could have detrimental effects in other ways. We also note that high descriptive norms and/or proximity intentions could have different effects in different populations. For instance, although high proximity intentions had no discernable detrimental effects in this sample, they might be much more problematic among treatment-seeking or dependent users. Finally, it should be considered that the descriptive and injunctive norms differ in level of specificity; descriptive norms asked about peers in general and injunctive norms asked about specific close individuals. It is possible the greater predictive relation of injunctive norms to marijuana use is affected by the greater level of specificity in the referent for the norms.

The lack of relation between descriptive norms and intentions, and between proximity intentions and Year 2 marijuana use may initially seem at odds with past research showing a link between peer behavior and future substance use (Andrews et al., 2002; Ennett & Bauman, 1991, 1994; Simons-Morton & Farhat, 2010; Urberg et al., 1997). One explanation may be that peer effects are often reported in terms of age of initiation (e.g., Maxwell, 2002), whereas our study assessed level of use post-initiation.

When considering solely studies of marijuana use with participants in the same range as the present sample, effects of norms on use have been variable. Consistent with the present results, injunctive norms sometimes predict intentions (Ajzen et al., 1982; McMillan & Connor, 2003), but other studies find that descriptive norms also predict intentions (Armitage et al., 1999; Connor & McMillan, 1999), and not all studies measure both types of norms. One factor that may affect the predictive potency of descriptive norms is the level of marijuana use within the local environment. We have argued that when marijuana use is pervasive – as in our sample – descriptive norms can be relatively high among all individuals, reducing their impact on behavior. It is unclear how prevalent marijuana use was within any of the past samples, although given the publication dates, we assume the prevalence to be if anything lower. Measurement issues may also affect the relation between descriptive norms and use. Our descriptive norm variable was intended to assess perceived typicality of use among peers, leading us to ask participants about use among most people their age. The descriptive norm variables in Connor and McMillan (1999) and McMillan and Connor (2003) were broader, including items asking about use among one’s best friend, friends in general, family, and health experts. Something about the consideration of these additional referents, and perhaps the information participants drew on to generate expectations for these additional referents (e.g., participants probably relied on first-hand experience to judge whether friends use marijuana, but are not likely to have first-hand experience with the marijuana use behavior of many health experts) may be increasing the prediction of actual marijuana use. Finally, we note that the time frame of prediction in the present study was much greater than in past studies, which assessed the relation between norms and intentions/use from 1 week to 6 months. Explicit evaluation of the degree to which the features we identify here moderate the impact of peers on use would be useful in future research.

A final difference between the different types of intentions we observed was in the degree to which our two intention measures were explained by attitudes, injunctive norms, descriptive norms, and RSE. Whereas these variables accounted for 73% and 78% of variance in use intentions for males and females, they only accounted for 36% and 37% of variance in proximity intentions for males and females. This suggests that additional factors may account for proximity intentions.

Temporal Stability of Intentions

Our path models including the TSI variables are broadly consistent with extant theorizing that consistency in intentions over time is an independent predictor of behavior, at least with respect to TSI-Use. In addition to showing this basic effect, the TSI variables are also of value in characterizing the overall level of consistency in marijuana intentions in this age range. The descriptive data in Table 1 indicate very little intention to change behavior over a one year period at the mean level, although there is variability across the sample. While there was a reliable effect of Marijuana Use Group on TSI-Use, with decreases in intentions among frequent users as compared to never and infrequent users, the differences were small in absolute magnitude. What is more notable is that across all levels of marijuana use, and for both TSI-Use and TSI-Proximity, students in this population were reporting little change in intentions. This is potentially encouraging for those who entered the study as non-users, indicating that beginning college does not radically increase their intentions to use marijuana and be around those who are using, even when they perceive descriptive norms to be relatively favorable toward marijuana use. However, by the same token, the results show that those who were already using on a nearly daily basis show little intention to decrease their use over a year period. The present data focus on one important life transition and period of time. It would be informative to examine stability of intentions in other periods of potential change, such as the period surrounding college graduation.

A final issue for consideration is what factors account for stability in intentions. Our path models indicate that in this sample, neither attitudes, injunctive norms, descriptive norms, nor RSE predicted TSI-Use or TSI-Proximity (although attitudes were a marginally significant negative predictor of TSI-use for males only). This might primarily be a function of the relatively low amount of instability in our sample. However, it could also reflect that factors other than attitudes, injunctive norms, descriptive norms, and RSE are important determinants of stability. Given the predictive potential attributed to stability of intentions, further research on factors that predict stability would be beneficial.

Limitations and Future Directions

Although the results of the present study are novel and provide insight into marijuana use in a pro-marijuana college community, there are limitations. For one, while we think sampling from a population with relatively high use makes our results relevant to understanding predictors of marijuana use within the current context of rapidly changing attitudes and laws toward its use, they cannot definitively speak to the impact of legalization as all data were collected before recreational use become legal within the state. However, short of waiting until after legalization occurs, examining predictors of use within pro-marijuana communities provides the best insight into possible effects of legalization. Another concern may be the legal changes that occurred over our data collection time period. While the state amendment legalizing recreational use was passed during data collection, there were no cohort effects on any of our measured variables. This indicates stability in psychosocial variables and marijuana use within our sample across a three year span and suggests that our results are unlikely to be due to changes in marijuana use across our sampling period (e.g., much more positive attitudes coupled with elevated marijuana use in data collected most recently). This conclusion is further supported by larger scale surveys within this population which show similar levels of marijuana use among students at this campus before and after the vote to legalize marijuana use. Specifically, self-reported use within the past 30 days was similar in spring 2011 (38.2%, n = 509) and spring 2013 (32.5%, n = 1064) (American College Health Association, 2011, 2013) (the amendment legalizing marijuana use was passed in November, 2012).

While we view the pro-marijuana environment as a strength of the context in which the data were collected, there could be concern about it creating a lack of variability within our variables. As the means and standard deviations reported in Table 1 attest, there was sufficient variability in all variables. Even for descriptive norms and proximity intentions, which tended to be high in our sample, there was variability in responses. Still, the sociocultural context does need to be considered in terms of generalization. We intended these data to speak to relations among psychosocial variables within pro-marijuana communities; they may not generalize to populations with much lower levels of use and/or much less acceptance of marijuana use.

Another potential limitation of this study is its reliance on self-report measures. One factor mitigating against this concern is that our measure of marijuana use, the TLFB, shows high agreement with biochemical measures of substance use (e.g. 87.3%–90.9% agreement for cannabis across studies; Hjorthoj, Hjorthoj, & Nordentof, 2012). Moreover, while it is possible that participants do not have full conscious access to constructs such as their attitudes toward an object or behavior (Nisbett & Wilson, 1977), the significant predictive relations we found between TPB variables and marijuana use one year later indicate that even if measured with some error, self-reported attitudes, intentions, and RSE account for significant variance in marijuana use.

Conclusions

The results provide important insight into the determinants of marijuana use within a pro-marijuana community. Within the context of relatively high favorability toward marijuana and levels of marijuana use, the current results are the first to show that use and proximity intentions are differentially predictive of marijuana use. Specifically, the driving force in predicting marijuana use among college adolescents in a pro-marijuana community was use intentions and their increase over time. The results also highlight the different effects of descriptive and injunctive norms. Perceptions of use among peers were relatively high among all participants and did not predict use intentions. By contrast, injunctive norms did vary as a function of use group and predicted use intentions. Thus, the results strongly favor the importance of the perceived expectations of important others over what peers are perceived to be doing.

Supplementary Material

Tables

Acknowledgments

This research was supported by a grant from the National Institute on Drug Abuse (R01 DA024002, Tiffany Ito PI).

Footnotes

1

Because the urine tests lacked the sensitivity to detect whether substances had been used within our abstinence periods (e.g., the tests were sensitive to marijuana use within the past 3–10 days whereas our abstinence period for marijuana was 6 hours), they were used primarily to supplement our verbal abstinence instructions. Participants were told they would have to complete a urine test to verify their abstinence and were not told of the test’s sensitivity, so we expected knowledge of the impending test to provide extra motivation to comply. For any participant showing evidence of recent substance use, we told them of their positive test and interviewed them in more detail about their period of abstinence. The majority of positive tests were for marijuana among known users. Given that our frequent users were reporting more than 20 days of use per month on average, we would expect most to be positive on a test sensitive to marijuana use up to 10 days previously. Among the 12 positive tests for other substances (4 for amphetamines, 1 for methamphetamine, 4 for cocaine, and 3 for opiates) all provided credible explanations that their use occurred outside the abstinence windows. The determination that these participants had met the abstinence criteria was admittedly subjective but even if they misrepresented their use, we had little reason to expect that recent substance use would bias the pattern of relations among variables being assessed here.

References

  1. Ajzen I. The Theory of Planned Behavior. Organizational Behavior and Human Decision Processes. 1991;50(2):179–211. [Google Scholar]
  2. Ajzen I. The Theory of Planned Behaviour: Reactions and reflections. Psychology & Health. 2011;26:1113–1127. doi: 10.1080/08870446.2011.613995. [DOI] [PubMed] [Google Scholar]
  3. Ajzen I, Fishbein M. Understanding attitudes and predicting social behaviour. Englewood Cliffs, NJ: Prentice-Hall; 1980. [Google Scholar]
  4. Ajzen I, Timko C, White JB. Self-monitoring and the attitude-behavior relation. Journal of Personality and Social Psychology. 1982;42(3):426–435. [Google Scholar]
  5. American College Health Association. National College Health Assessment (ACHA-NCHA) Web Summary. 2011 Available at http://www.acha-ncha.org/docs/ACHA-NCHA-II_UNDERGRAD_ReferenceGroup_DataReport_Spring2011.pdf.
  6. American College Health Association. National College Health Assessment (ACHA-NCHA) Web Summary. 2013 Available at http://www.acha-ncha.org/docs/ACHA-NCHA-II_UNDERGRAD_ReferenceGroup_DataReport_Spring2013.pdf.
  7. Andrews JA, Tildesley E, Hops H, Li F. The influence of peers on young adult substance use. Health Psychology. 2002;21(4):349–357. doi: 10.1037//0278-6133.21.4.349. [DOI] [PubMed] [Google Scholar]
  8. Armitage CJ, Conner M, Loach J, Willetts D. Different perceptions of control: Applying an extended Theory of Planned Behavior to legal and illegal drug use. Basic and Applied Social Psychology. 1999;21(4):301–316. [Google Scholar]
  9. Baumeister RF, Leary MR. The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin. 1995;117(3):497–529. [PubMed] [Google Scholar]
  10. Bava S, Tapert SF. Adolescent brain development and the risk for alcohol and other drug problems. Neuropsychology Review. 2010;20(4):398–413. doi: 10.1007/s11065-010-9146-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Brady KT, Randall CL. Gender differences in substance use disorders. Psychiatric Clinics of North America. 1999;22(2):241–52. doi: 10.1016/s0193-953x(05)70074-5. [DOI] [PubMed] [Google Scholar]
  12. Broaddus MR, Schmiege SJ, Bryan AD. An expanded model of the temporal stability of condom use intentions: Gender-specific predictors among high-risk adolescents. Annals of Behavioral Medicine. 2011;42(1):99–110. doi: 10.1007/s12160-011-9266-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Brown BB. The extent and effects of peer pressure among high school students: A retrospective analysis. Journal of Youth and Adolescence. 1982;11(2):121–133. doi: 10.1007/BF01834708. [DOI] [PubMed] [Google Scholar]
  14. Bryan A, Rocheleau CA, Robbins RN, Hutchison KE. Condom use among high-risk adolescents: Testing the influence of alcohol use on the relationship of cognitive correlates of behavior. Health Psychology. 2005;24(2):133–142. doi: 10.1037/0278-6133.24.2.133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Cialdini RB, Reno RR, Kallgren CA. A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology. 1990;58(6):1015–1026. [Google Scholar]
  16. Colorado Amendment 64. Amendment 64: Use and Regulation of Marijuana. 2012 http://www.colorado.gov/cs/Satellite?blobcol=urldata&blobheader=application%2Fpdf&blobkey=id&blobtable=MungoBlobs&blobwhere=1251834064719&ssbinary=true.
  17. Conner M, McMillan B. Interaction effects in the Theory of Planned Behaviour: Studying marijuana use. British Journal of Social Psychology. 1999;38(2):195–222. doi: 10.1348/014466699164121. [DOI] [PubMed] [Google Scholar]
  18. Conner M, Sheeran P, Norman P, Armitage CJ. Temporal stability as a moderator of relationships in the Theory of Planned Behaviour. British Journal of Social Psychology. 2000;39(4):469–493. doi: 10.1348/014466600164598. [DOI] [PubMed] [Google Scholar]
  19. Conner M, Warren R, Close S, Sparks P. Alcohol consumption and the Theory of Planned Behavior: An examination of the cognitive mediation of past behavior. Journal of Applied Social Psychology. 1999;29(8):1676–1704. [Google Scholar]
  20. Crane NA, Schuster RM, Fusar-Poli P, Gonzalez R. Effects of cannabis on neurocognitive functioning: Recent advances, neurodevelopmental influences, and sex differences. Neuropsychology Review. 2013;23(2):117–137. doi: 10.1007/s11065-012-9222-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Deutsch M, Gerard HB. A study of normative and informational social influences upon individual judgment. Journal of Abnormal and Social Psychology. 1955;51(3):629–636. doi: 10.1037/h0046408. [DOI] [PubMed] [Google Scholar]
  22. Eagly AH, Carli LL. Sex of researchers and sex-typed communications as determinants of sex differences in influenceability: A meta-analysis of social influence studies. Psychological Bulletin. 1981;90(1):1–20. doi: 10.1037/0033-2909.90.1.1. [DOI] [Google Scholar]
  23. Elliott JC, Carey KB. 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. doi: 10.15288/jsad.2012.73.976. [DOI] [PubMed] [Google Scholar]
  24. Elliott JC, Carey KB, Vanable PA. A preliminary evaluation of a web-based intervention for college marijuana use. Psychology of Addictive Behaviors. 2014;28:288–293. doi: 10.1037/a0034995. [DOI] [PubMed] [Google Scholar]
  25. Ekstrom RB, French JW, Harman HH, Dermen D. Manual for kit of factor-referenced cognitive tests. Princeton, NJ: Educational Testing Service; 1976. [Google Scholar]
  26. Ennett ST, Bauman KE. Mediators in the relationship between parental and peer characteristics and beer drinking by early adolescents. Journal of Applied Social Psychology. 1991;21(20):1699–1711. [Google Scholar]
  27. Ennett ST, Bauman KE. The contribution of influence and selection to adolescent peer group homogeneity: The case of adolescent cigarette smoking. Journal of Personality and Social Psychology. 1994;67(4):653–663. doi: 10.1037//0022-3514.67.4.653. [DOI] [PubMed] [Google Scholar]
  28. Fishbein M, Ajzen I. Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley; 1975. [Google Scholar]
  29. Fishbein M, Ajzen I. Predicting and changing behavior: The reasoned action approach. New York, NY: Psychology Press; 2010. [Google Scholar]
  30. Fromme K, Corbin WR, Kruse MI. Behavioral risks during the transition from high school to college. Developmental Psychology. 2008;44(5):1497–1504. doi: 10.1037/a0012614. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Gallup. For first time, Americans favor legalizing marijuana. 2013 http://www.gallup.com/poll/165539/first-time-americans-favor-legalizing-marijuana.aspx.
  32. Glanz K, Bishop DB. The role of behavioral science theory in the development and implementation of public health interventions. Annual Review of Public Health. 2010;31:399–418. doi: 10.1146/annurev.publhealth.012809.103604. [DOI] [PubMed] [Google Scholar]
  33. Hames A Evangeli M, Harrop C di Forti M. Understanding cannabis use in first-episode psychosis: An application of the Theory of Planned Behaviour. Early Intervention in Psychiatry. 2012;6(1):38–44. doi: 10.1111/j.1751-7893.2011.00316.x. [DOI] [PubMed] [Google Scholar]
  34. Hjorthoj CR, Hjorthoj AR, Nordentoft M. Validity of Timeline Follow-Back for self-reported use of cannabis and other illicit substances–systematic review and meta-analysis. Addictive Behavior. 2012;37(3):225–233. doi: 10.1016/j.addbeh.2011.11.025. [DOI] [PubMed] [Google Scholar]
  35. Johnston LD, O’Malley PM, Miech RA, Bachman JG, Schulenberg JE. Monitoring the Future national survey results on drug use: 1975–2013: Overview, key findings on adolescent drug use. Ann Arbor: Institute for Social Research, The University of Michigan; 2014. [Google Scholar]
  36. Kandel DB, Chen K. Types of marijuana users by longitudinal course. Journal of Studies on Alcohol. 2000;61(3):367–378. doi: 10.15288/jsa.2000.61.367. [DOI] [PubMed] [Google Scholar]
  37. Kiesel A, Steinhauser M, Wendt M, Falkenstein M, Jost K, Phillipp AM, Koch I. Control and interference in task switching—A review. Psychological Bulletin. 2010;136(5):849–874. doi: 10.1037/a0019842. [DOI] [PubMed] [Google Scholar]
  38. Lac A, Alvaro EM, Crano WD, Siegel JT. Pathways from parental knowledge and warmth to adolescent marijuana use: An extension to the theory of planned behavior. Prevention Science. 2009;10(1):22–32. doi: 10.1007/s11121-008-0111-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Malmberg M, Overbeek G, Vermulst AA, Monshouwer K, Vollebergh WA, Engels RC. The Theory of Planned Behavior: Precursors of marijuana use in early adolescence? Drug and Alcohol Dependence. 2012;123(1–3):22–28. doi: 10.1016/j.drugalcdep.2011.10.011. [DOI] [PubMed] [Google Scholar]
  40. Marsh P, Allen JP, Ho M, Porter M, McFarland FC. The changing nature of adolescent friendships: Longitudinal links with early adolescent ego development. Journal of Early Adolescence. 2006;26(4):414–431. doi: 10.1177/0272431606291942. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Maxwell KA. Friends: The role of peer influence across adolescent risk behaviors. Journal of Youth and Adolescence. 2002;31(4):267–277. [Google Scholar]
  42. McMillan B, Conner M. Applying an extended version of the Theory of Planned Behavior to illicit drug use among students. Journal of Applied Social Psychology. 2003;33(8):1662–1683. [Google Scholar]
  43. Mizuno Y, Seals B, Kennedy M, Myllyluoma J. Predictors of teens’ attitudes toward condoms: Gender differences in the effects of norms. Journal of Applied Social Psychology. 2000;30(7):1381–1395. [Google Scholar]
  44. Muthén LK, Muthén BO. The comprehensive modelling program for applied researchers: User’s guide. Los Angeles, CA: Muthén & Muthén; 2012. Mplus. [Google Scholar]
  45. Nisbett R, Wilson T. Telling more than we can know: Verbal reports on mental processes. Psychological Review. 1977;84(3):231–259. [Google Scholar]
  46. Ridenour TA, Lanza ST, Donny EC, Clark DB. Different lengths of times for progressions in adolescent substance involvement. Addictive Behaviors. 2006;31(6):962–983. doi: 10.1016/j.addbeh.2006.03.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Rudolph KD, Conley CS. The socioemotional costs and benefits of social-evaluative concerns: Do girls care too much? Journal of Personality. 2005;73:115–138. doi: 10.1111/j.1467-6494.2004.00306.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Schepis TS, Desai RA, Cavallo DA, Smith AE, McFetridge A, Liss TB, Potenza MN, Krishnan-Sarin S. Gender differences in adolescent marijuana use and associated psychosocial characteristics. Journal of Addiction Medicine. 2011;5(1):65–73. doi: 10.1097/ADM.0b013e3181d8dc62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Sheeran P, Abraham C. Mediator of moderators: Temporal stability of intention and the intention-behavior relation. Personality and Social Psychology Bulletin. 2003;29(2):205–215. doi: 10.1177/0146167202239046. [DOI] [PubMed] [Google Scholar]
  50. Sheeran P, Maki A, Montanaro E, Bryan AD, Klein WMP, Rothman AJ. The impact of changing attitudes, norms, and self-efficacy on health-related intentions and behavior: A meta-analysis Manuscript under review. 2015 doi: 10.1037/hea0000387. [DOI] [PubMed] [Google Scholar]
  51. Simons-Morton BG, Farhat T. Recent findings on peer group influences on adolescent smoking. Journal of Primary Prevention. 2010;3(4):191–208. doi: 10.1007/s10935-010-0220-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sobell LC, Sobell MB. Timeline Follow-back: A technique for assessing self-reported ethanol consumption. In: Allen J, Litten RZ, editors. Measuring Alcohol Consumption: Psychosocial and Biological Methods. Totowa, NJ: Humana Press; 1992. pp. 41–72. [Google Scholar]
  53. Stephens RS, Roffman RA, Curtin L. Comparison of extended versus brief treatments for marijuana use. Journal of Consulting and Clinical Psychology. 2000;68:898–908. [PubMed] [Google Scholar]
  54. Stephens RS, Roffman RA, Fearer SA, Williams C, Burke RS. The Marijuana Checkup: Promoting change in ambivalent marijuana users. Addiction. 2007;102:947–957. doi: 10.1111/j.1360-0443.2007.01821.x. [DOI] [PubMed] [Google Scholar]
  55. Substance Abuse and Mental Health Services Administration. Results from the 2010 National Survey on Drug Use and Health: Summary of National Findings. Rockville, MD: 2011. (NSDUH Series H-41, HHS Publication No. (SMA) (11-4658)). [Google Scholar]
  56. Sutton S. The Sage Handbook of Health Psychology. London: Sage; 2004. Determinants of health-related behaviours: Theoretical and methodological issues; pp. 94–126. [Google Scholar]
  57. Urberg KA, Degirmencioglu SM, Pilgrim C. Close friend and group influence on adolescent cigarette smoking and alcohol use. Developmental Psychology. 1997;33(5):834–844. doi: 10.1037//0012-1649.33.5.834. [DOI] [PubMed] [Google Scholar]
  58. Wagner FA, Anthony JC. Male-female differences in the risk of progression from first use to dependence upon cannabis, cocaine, and alcohol. Drug and Alcohol Dependence. 2007;86:191–198. doi: 10.1016/j.drugalcdep.2006.06.003. [DOI] [PubMed] [Google Scholar]
  59. Walitzer KS, Dearing RL. Gender differences in alcohol and substance use relapse. Clinical Psychology Review. 2006;26(2):128–48. doi: 10.1016/j.cpr.2005.11.003. [DOI] [PubMed] [Google Scholar]
  60. Washington Initiative 502. Summary of Initiative 502. 2012 http://www.leg.wa.gov/House/Committees/OPRGeneral/Documents/2012/I-502%20summary.pdf.
  61. Webb TL, Sheeran P. Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin. 2006;132(2):249–268. doi: 10.1037/0033-2909.132.2.249. [DOI] [PubMed] [Google Scholar]
  62. Yeung N, Sanfey AG. Independent coding of reward magnitude and valence in the human brain. Journal of Neuroscience. 2004;24(28):6258–6264. doi: 10.1523/JNEUROSCI.4537-03.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Tables

RESOURCES