Evaluating the Effect of Retail Marijuana Legalization on Parent Marijuana Use Frequency and Norms in U.S. States With Retail Marijuana Legalization

Abstract

Objectives:

To examine post-retail marijuana legalization (RML) change in marijuana use frequency and pro-marijuana norms among parents.

Methods:

The Intergenerational Study, a longitudinal panel of parents (N = 668) and children, followed participants from 2002 to 2018, when parents were 27 and 43 years old, respectively. Three quarters of participants (74%) lived in an RML state and 142 (21%) had used marijuana in the 8 years prior to RML. Piecewise growth modelling compared pre- and post-RML slopes of use frequency and pro-marijuana norms.

Results:

Frequency of use and pro-marijuana norms increased following legalization in both RML and non-RML states, though norms rose significantly faster in RML states. Growth in use was primarily driven by new users of marijuana. There were no differences in frequency of marijuana use or pro-marijuana norms by race/ethnicity, gender, or education.

Conclusions:

An increase in marijuana use frequency associated with RML among parents poses risk to both parents' well-being and the health of their children. A faster pace of increase in pro-marijuana norms in RML states may signal continued increases in use in the future.

1. Introduction

Since the first retail marijuana legalization (RML) legislation passed in Washington and Colorado in 2012, public health concerns have risen that RML would lead to higher rates of youth use. Although initial evaluations of RML's effect on youth use have shown mixed results (Anderson, Hansen, Rees, & Sabia, 2019; Bailey et al., in press; Dilley et al., 2019; Harpin, Brooks-Russell, Ma, James, & Levinson, 2018; Mason et al., 2016), studies have found that children's pro-marijuana norms and other risk factors have increased (Fleming, Guttmannova, Cambron, Rhew, & Oesterle, 2016). Increase in adult use, and particularly in parent use, is a key risk factor for increasing availability of marijuana to youth (Bailey et al., 2016; Duncan, Duncan, Hops, & Stoolmiller, 1995; Hawkins, Catalano, & Miller, 1992). Several recent studies have reported a rise in marijuana use prevalence among adults in the time period following legalization (Cerdá et al., 2019; Everson, Dilley, Maher, & Mack, 2019; Substance Abuse and Mental Health Services Administration), including among parents (Goodwin et al., 2018). What is less clear is whether increase in prevalence indicates new users of marijuana, and whether there may be accompanying increases in frequency of use among existing users. It is also unknown the extent to which the effects of RML are limited to those states that have legalized marijuana compared to states that have not. Further, more information is needed about differences by ethnicity, gender, and education in post-RML marijuana use. Finally, comparing increases in marijuana use to changes in use of other legal substances (e.g., alcohol) is also important in order to distinguish between a general trend in greater substance use and an increase in marijuana use in particular. The current investigation uses data from a longitudinal study of parents to a) examine frequency of marijuana use and pro-marijuana norms before and after legalization in RML and non-RML states, compared to frequency of alcohol use; b) understand to what degree increase in use is attributable to new or existing users; and c) compare changes in frequency of use and norms among demographic groups.

Data from the National Survey on Drug Use and Health (NSDUH) shows a steady increase in the prevalence of past-year marijuana use among U.S. adults age 26 and older, from 12% in 2010 (Azofeifa et al., 2016) to 16% in 2018 (Substance Abuse and Mental Health Services Administration). This is also true among parents, whose past-month prevalence of marijuana use rose from 4.9% in 2002 to 6.8% in 2015 (Goodwin et al., 2018). Scant evidence exists about whether marijuana use increased equally in RML and non-RML states and whether increases are due to new users. We found only one study using the NSDUH dataset addressed these questions. Led by Cerdá and colleagues (2019), this study reported that between 2012 and 2016 there was an increase in marijuana use prevalence in both RML and non-RML states, but that the increase in past-month (but not past-year) prevalence of marijuana use in RML states was steeper than in non-RML states. The authors state that there were potentially new users of marijuana, even in states without legalization, but that new use was concentrated in RML states. However, an analysis based on a longitudinal sample of Washington State parents concluded that the increase was largely due to a greater frequency of use among existing users (Kosterman et al., 2016). Further investigation into this issue is necessary.

Another unanswered question concerns changes in frequency of marijuana use following legalization. The Cerda study found that prevalence of use on 20 or more days in the past month increased among adults following RML (Cerdá et al., 2019). However, cross-sectional studies with a single point of contact with participants may be particularly susceptible to social desirability bias, with participants less likely to report their (illegal) use in pre-RML years because they had limited trust in the researchers conducting the survey. Longitudinal studies build trust by establishing longstanding relationships with participants and earning their trust.

In addition to tracking changes in marijuana use, several studies have addressed change in pro-marijuana norms, through all focused on adolescent or college student populations. Reporting pro-marijuana norms is an important precursor to use initiation, and of heavier use (Wu, Swartz, Brady, Hoyle, & NIDA AAPI Workgroup, 2015), and pro-marijuana norms have been shown to increase following passage of medical marijuana laws (Blavos, Glassman, Sheu, Thompson, & DeNardo, 2019; Cerdá, Wall, Keyes, Galea, & Hasin, 2012). Among adolescents and young adults, pro-marijuana norms were shown to increase following legalization, and increase at a higher rate in RML states (Koval, Kerr, & Bae, 2019; Wallace et al., 2019). However, it is not clear what impact RML had on adults' norms, nor whether RML had greater impacts on norms in RML states than non-RML states.

Finally, trends in use post-legalization by ethnicity have not been examined. An important driver of RML was the documented disparities between blacks and whites in how drug offenses were handled by the justice system (Alexander, 2010). Further, Americans of color generally experienced worse substance use outcomes prior to RML (Chartier & Caetano, 2010; Leventhal, 2015; Sahker, Toussaint, Ramirez, Ali, & Arndt, 2015). Post-legalization, it is important to understand whether RML reduced disparities or perpetuated them, and whether race/ethnicity moderates any effects of RML on marijuana use. We are aware of no study that has compared the changes in adult marijuana use by race post RML. Similarly, as there had been documented differences in rates of marijuana use among men vs. women (Hemsing & Greaves, 2020) and those with and without a college degree (Center for Behavioral Health Statistics and Quality, 2016), understanding of moderation by gender and by education is essential to crafting policy that minimizes health disparities in adult use.

The current study will shed light on how marijuana legalization has shaped parents' marijuana use and norms in a longitudinal community panel of parents followed from age 27 to age 42. Typically, marijuana and other drug use begins a steady decline in frequency and prevalence in the mid to late 20s. Marijuana legalization presents a policy intervention that is hypothesized to change the course of marijuana use and norms among this panel. We expect that both frequency of use and pro-marijuana norms will increase following legalization (Hypothesis 1 and Hypothesis 2), compared to the rate of alcohol use (a counterfactual comparison; Hypothesis 3). Next, analyses will examine whether frequency of use and pro-marijuana norms differed by RML; both are expected to increase faster in RML states (Hypothesis 4). We will also examine whether increases in frequency of use were due to new users of marijuana versus an increase in frequency of use among those who had been using before RML (Research Question 5). No hypotheses are provided for the research questions due to lack of prior findings. Finally, we compare frequency of use before and after RML for black vs. white participants, as well as by other ethnicity, gender, and education (Research Question 6). We make no predictions for the research questions due to a lack of available data. Results from this study will add to the still small literature on changes in marijuana use and norms among adults after RML. This study will be the first to examine how RML has affected frequency of marijuana use among parents. Since parent substance use, including marijuana, is strongly related to child substance use (Bailey et al., 2016; Henry & Augustyn, 2017; Hill, Sternberg, Suk, Meier, & Chassin, 2018; Kosty et al., 2015) and other child health outcomes (Epstein, Bailey, Furlong, Steeger, & Hill, 2019), changes in parent behavior may signal increased risk for children.

2. Methods

2.1. Participants

The current study uses data from the Seattle Social Development Project - The Intergenerational Project (SSDP-TIP) (Bailey, Hill, Epstein, Steeger, & Hawkins, 2018), a longitudinal study of parents and children (N = 426 families). SSDP-TIP participants include a biological parent, their oldest biological child, and a second caregiver (usually the other biological parent) where appropriate. Target parents were drawn from another longitudinal study, the Seattle Social Development Project (SSDP), which has followed them since 1985 when they were 10 years old. TIP families were surveyed 10 times between 2002 and 2018. Data in Waves 1 - 7 were collected prior to legalization (2002 - 2011) and Waves 8 - 10 were collected after the opening of marijuana dispensaries in 2014 (2015 - 2018). The current analysis uses data from all biological parents in TIP (N = 668; SSDP parents N = 426, second caregiver N = 242). The SSDP parents in the study were, on average, age 27 at Wave 1 and age 43 at Wave 10; no age data was available on the other parent. The sample was gender diverse (57% female); 52% of parents identified as white, 21% identified as black, 20% identified as Asian American, and 6% identified as Native American. The majority (69%) held a Bachelor's degree or higher. During post-legalization waves of data collection (Waves 8 - 10), most parents lived in an RML state for at least a portion of the time (74%); 26% had lived in one of 24 states without RML. Procedures were approved by the University of Washington Institutional Review Board.

2.2. Measures

Parents reported their marijuana use frequency in the past year. Frequency of use was coded as 0 = Never, 1 = Monthly or less, 2 = Weekly or less, 3 = Multiple times a week, and 4 = Daily use.

Pro-marijuana norms were assessed with a single item which asked parents to rate whether they thought it was "ok for adults to smoke marijuana." Response options included "YES!" "yes," "no," and "NO!" The two yes responses were combined as were the two no responses for a dichotomous measure of marijuana use approval (0 disapprove, 1 approve).

Parents also reported on past-month alcohol use frequency coded as 0 = None, 1 = 1 - 3 times, 2 = 4 - 9 times, and 3 = 10+ times.

RML state status was computed from parents' self-report of the state where they lived in Waves 8 - 10. By 2018, parents lived in 5 RML states (WA, CO, OR, AK, CA) and 24 non-RML states. For those who had ever lived in a state that had already legalized retail marijuana between 2015 and 2018, RML was coded 1, 0 otherwise.

Gender (male =1), race/ethnicity (Black, Asian, Native, White = referent), and education (BA degree vs. less than a BA) were self-reported.

2.3. Analysis

All variables were coded as categorical and analyzed in Mplus (Muthén & Muthén, 1998-2018) using the Maximum Likelihood estimator with robust standard errors (MLR). Piecewise latent growth models were used to estimate and compare pre-legalization and post-legalization slopes in marijuana and alcohol use and marijuana norms. Figure S1 (Supplement) shows a conceptual model of the analyses. Pre-legalization slope was estimated in Waves 1 - 7 (2002 - 2011), post-legalization slope was estimated in Waves 7 - 10 (2011 - 2018), with a single intercept held at the last time point before legalization (Wave 7, 2011). Both linear and quadratic slopes were tested, but no quadratic slope factors were significant; pre- and post-legalization slopes were specified as linear. Each parent or caregiver was included as a separate participant in the dataset; shared variance within parent pairs was modeled by clustering by family ID. Models stratified by demographic variables were computed by using the KNOWNCLASS function of Mplus. The MODEL CONTRAINT function (Wald test) was used to compare means of slopes across classes. Missing data was handled using Full Information Maximum Likelihood (FIML).

3. Results

3.1. Descriptive findings

About a quarter of the sample (28%) reported any pre-RML use from ages 27 (2002) to 36 (2010). From year to year, past-year prevalence declined significantly from 25% of the sample reporting using use at age 27 to 15% at age 36 (see Figure 1). Following RML, 33% reported any use between 2011 and 2018; annual prevalence increased from 15% at age 36 to 26% by age 42, which is a divergence from the normative stabilization in substance use with age. Pre-RML, parents' pro-marijuana norms did not change between ages 27 and 36, ranging from 28% and 26% agreeing that marijuana use was "ok for adults." Pro-marijuana norms rose sharply, however, following RML—almost doubling from age 36, with 53% of parents at age 42 saying that marijuana use was ok.

Figure 1: Marijuana use before and after legalization.

Three quarters of parents (74%) had lived in a state that had legalized marijuana between 2015 and 2018. Parents who lived in a legalized state were more likely to be female, but there were no differences by race/ethnicity, education, or pre-RML marijuana use. Because of the difference in gender by RML context, gender was included in analyses for Hypothesis 4 and Research Question 5.

3.2. Changes in marijuana use and norms compared to alcohol use

Table 1 summarizes piecewise regression results and shows pre- and post-RML slopes and their differences. The slope of frequency of use was negative and significant pre-RML; post-RML, the slope was positive and significantly different from zero (Table 1: full sample). There was a significant difference in the change in frequency of marijuana use before and after legalization, confirming Hypothesis 1. Figure 1 shows the pattern of parent marijuana use from age 27 (2002) to 42 (2018); for ease of interpretation, Figure 1 shows only the pattern of ever use of marijuana, although the full frequency scale was used in the regression.

Table 1.

Changes in marijuana use and norms and alcohol use before and after marijuana legalization.

		Marijuana use			Marijuana norms
	Pre-slope mean	Post-slope mean	Slope difference	Pre-slope mean	Post-slope mean	Slope difference
Full samplea	−.12	.12	p < .001	−.01ns	.17	p < .001
By state policy
Non-RML	−.09	.11	p < .05	−.01ns	.10	p = <.05
RML	−.06	.11	p < .001	−.02ns	.21	p < .01¥
New vs. existing usersb
New	-	.22	p < .01	-	-	-
Continuing	-	.07	p = ns¥	-	-

Note.

^aN = 668. ns = not significant. Slopes are all significant at p < .05, except when indicated by ^ns.

^bN = 511.

^¥Post-legalization slope in RML states > non-RML states (p < .01).

There was little change in pro-marijuana norms before RML, and the slope was not significantly different from zero. However, post-RML slope in pro-marijuana norms was positive and significantly different from zero. In accordance with Hypothesis 2, the two slopes were significantly different from each other.

Consistent with Hypothesis 3, there were no changes in alcohol use following RML. All alcohol use frequency slopes pre- and post-RML were flat, with slopes not significantly different from zero (Table S1 and Figure S2 in Supplement).

3.3. Changes in marijuana use and norms in RML and non-RML states

To test Hypothesis 4, we examined whether the increase in marijuana use and pro-marijuana norms was unique to parents who lived in RML compared to non-RML states (see Table 1). Intercept and slopes were regressed onto gender to account for uneven gender distribution in RML and non-RML states. Figure 1 shows that frequency rates and changes in use were very similar, regardless of state RML status. In both RML and non-RML states, the slope of marijuana use frequency decreased slightly before RML and increased thereafter; the increase in slope was not significantly different in RML compared to non-RML states.

Across RML contexts, pro-marijuana norms remained flat before legalization and increased thereafter (Figure 2). In both RML and non-RML states, the pre- and post-legalization slopes were significantly different from each other. However, the rate of increase in post-legalization slope was significantly greater in RML than non-RML states.

Figure 2: Marijuana attitudes before and after legalization.

3.4. New and continuing users after legalization

Research Question 5 asked whether the increases in marijuana use frequency following RML were due to additional new users of marijuana compared to continuing use among those prior users, and whether these changes varied by RML context. Analyses for this aim examined the post-legalization slope from age 36 to 42 (N = 511) and compared marijuana use among those who had reported no marijuana use in the pre-legalization period (2002 - 2011; N = 369) to those who reported use in the 10 years prior to RML (N = 142; see Figure 3). Table 1 shows a large and significant increase in use frequency among previous nonusers (new) and a small but significant increase among previous users (continuing). The two slopes were significantly different from each other, indicating that the majority of growth was explained by new users after legalization, but that some increase in use frequency among existing users also occurred.

Figure 3: Marijuana use following legalization.

3.5. Demographic differences in marijuana use and norms

To address Research Question 6, we stratified analyses of marijuana use and norms by race/ethnicity, gender, and education in three separate models (see Table S1 and Figure S3 in Supplement). There were significant increases in use among black and white parents following RML, but not among Asian American or Native American parents; however, the rate of increase did not differ between groups, including when comparing black and white parents. Use increased at equal rates among men and women, and among those with and without a BA degree. With respect to pro-marijuana norms, there was a significant increase for all demographic groups except Native Americans, and the rate of increase was similar across groups.

4. Discussion

The current study examined changes in marijuana use and norms among a community sample of parents following legalization of retail marijuana. An increase in use among parents is notable from a public health perspective and is an important metric to monitor as parental use may place their own children at increased risk of using marijuana and other substances (Bailey et al., 2016; Bailey, Hill, Oesterle, & Hawkins, 2006; Hill et al., 2018). Analyses indicate that, among parents, the frequency of use and level of pro-marijuana norms increased following legalization, whereas alcohol use frequency remained flat. Moreover, the increase in use was largely, though not entirely attributable to an increase in new users.

Increase in frequency of marijuana use among parents was observed in states both with and without RML legislation. Although it is possible that some residents of non-RML states crossed state borders to purchase marijuana (e.g., from Idaho to Washington), the majority of non-RML state parents in this study did not live in a state that bordered one with RML (e.g., Maryland) during the years examined. Instead it is possible that RML has changed the overall normative structure around marijuana use nationwide. This is supported by our findings that pro-marijuana norms also rose sharply among parents following state-specific legalization in both RML and non-RML states. Norms rose faster in RML states, which could indicate that greater increases in rates of use in RML states will follow, which further underscores the need for continuing observation.

The current study also found that the majority of growth in parents' marijuana use is due to users who had not reported use in the previous eight years (2002 - 2010), though continuing users were also using at slightly higher rates. New users are a concern for public health because of increased exposure to the drug, both for parents themselves, who may risk escalation of use and subsequent addiction, and for their children. However, an increase in frequency among existing users presents a different challenge. If existing users are using marijuana more frequently, use higher potency marijuana (potency of marijuana has increased dramatically in the past few decades; Wilson, Freeman, & Mackie, 2019), or use marijuana and alcohol simultaneously, there are concerns about possible effects on traffic accidents, overdose, or other consequences of being under the influence. Both of these findings underscore the need to continue to monitoring prevalence, frequency, and amount consumed by parents (and other adults) overall, and curbing unsafe behaviors such as combining marijuana and alcohol and driving while impaired.

Finally, the current study is one of the first to examine shifts in marijuana use frequency among different demographic groups. Of most interest was the question of whether use might increase at a greater rate among black parents. Despite black and white Americans using marijuana at similar rates (Substance Abuse and Mental Health Services Administration, 2019), blacks were disproportionately affected by laws designed to punish marijuana possession, experiencing much higher rates of arrest and incarceration compared to whites who committed similar offenses (American Civil Liberties Union, 2013). Results from the current study showed that the rate of increase in marijuana use was not significantly different among black compared to white parents. In order to maintain commitment to social justice, it is important to monitor both rates of marijuana use among black Americans and the rates of arrests due to possession both in RML states and nationwide.

There are also several limitations of this study that need to be considered. The sample primarily resided in Washington State and consisted of parents of biological children; as such it may not be generalizable to all adults in the United States. Because no data were collected between 2011 and 2015, it is not possible to estimate whether rates of use increased immediately after RML legislation or were already on the rise. Future analyses should also examine whether changes in use frequency for new or continuing users differ by RML vs. non-RML states, and whether there are differences in how frequency in marijuana use is changing for black Americans in states with different RML statuses.

This study has a number of strengths that make the results a unique contribution to the literature on marijuana legalization. First, this study used prospective longitudinal data, which is rare in the field. Second, because parents were, on average, age 36 at the time of legalization, their frequency of use were not expected to change developmentally, which allowed us to better attribute changes to RML and the changing climate around marijuana in the United States. Third, the study included a) a sufficient sample size; b) a number of data points to determine changes in trajectory of marijuana use; and c) assessment of marijuana use frequency, as opposed to use/non-use, as is common in existing literature. Fourth, the inclusion of alcohol as a counterfactual outcome showed that alcohol use rates did not change during the corresponding time period of marijuana use. This finding increases confidence in the results. Finally, marijuana use change among parents, specifically, is an important behavior to monitor because of the associated increased risk of marijuana use for their children.

Legalization of marijuana was associated with a change to greater use among parents during a time in their development (late 30s) when substance use generally decreases, and with a greater acceptance of marijuana use. Increased use among adults may give rise to other problem behaviors, such as driving under the influence, and has potential for increased rates of marijuana dependence. Although occasional adult use may not be of public health concern, a broadening in marijuana use and acceptability among parents could increase availability to their children. Marijuana use among adolescents has been linked to a number of adverse outcomes, including other substance use, school dropout, and mental health problems (Hall, 2015; Volkow, Baler, Compton, & Weiss, 2014), suggesting that close monitoring of the trend in adult use is needed. At the same time, parenting interventions that help parents communicate with their children about the use of marijuana, set clear rules and guidelines, and administer consistent and appropriate consequences and reinforcement need to be tailored to be acceptable to parents who use marijuana.

Highlights.

• Following marijuana legalization in 2012 in Washington and Colorado, rates of marijuana use have increased among parents in both states that have and have not passed retail marijuana legalization (RML); by comparison rates of alcohol use have remained flat

• Increased marijuana use is largely due to new users, both in RML and non-RML states

• There were few differences in patterns of marijuana use by gender, race/ethnicity, or education

• Pro-marijuana use norms have increased during that time period as well

• Pro-marijuana use norms have increased at a higher rate in states that have passed RML, compared to non-RML states