Marijuana policy opinions in Washington state since legalization: Would voters vote the same way?

Abstract

Background

In 2012, voters in Washington state approved Initiative 502 (I-502) which legalized recreational marijuana use at the state level. This study examines the relationship between demographics, marijuana and alcohol use, and voting outcomes, as well as how these variables relate to (i) whether voters would still vote the same way (a reflection of satisfaction with the new policy) and (ii) the likelihood of using marijuana purchased from legal retail stores.

Methods

The sample consists of 2,007 adult Washington state residents recruited through Random Digit Dial between January and October 2014. Bivariate tests and multivariable regressions were used for analyses.

Results

Less than five percent of those who voted for marijuana legalization would change their votes, whereas 14% of those who voted against legalization would change their votes. In multivariable models controlling for demographics, substance use, and marijuana-related opinions, those who voted for legalization had half the odds of changing their votes than those who voted against it. Among past-year non-marijuana users, almost 10% were somewhat/very likely to use marijuana if they could buy it from a legal store. Past marijuana use, the belief that adults should be allowed to grow marijuana for personal use, and the belief that marijuana is not very risky for health were all related to increased likelihood of using marijuana purchased from legal stores.

Conclusion

Since November 2012, support for marijuana legalization in Washington state has increased; accounting for the proportion of voters who would change their votes suggests that I-502 would pass today with even more votes in favor.

1. INTRODUCTION

In November 2012, voters in Washington state approved Initiative 502 (I-502) which legalized marijuana at the state level as of December 1, 2012. Initiative 502 authorized the Washington State Liquor Control Board (WSLCB) to regulate and tax marijuana for persons 21 years of age and older and to add a new threshold for driving under the influence of marijuana. Along with reduced social sanctions and risks of criminal penalties, this increased availability could affect opinions about marijuana and related policies.

Few studies regarding public opinions about marijuana legalization have been published in the three years since marijuana has actually become legal in a number of US states. Most recently, a study of marijuana-related voting behaviors among college students in Wisconsin and Washington states showed that 46.3% of Washington participants reported voting in favor of legalizing recreational marijuana, while 50.6% of Wisconsin participants indicated they would vote for a similar initiative in their state (Moreno, Whitehill, Quach, Midamba, & Manskopf, in press). Perhaps not surprisingly, more positive attitudes towards marijuana use and stronger intentions to use marijuana in the next six months were both related to increased odds of voting in favor of/would vote in favor of legalization (Moreno et al., in press).

In terms of the general public, support for legalization in the U.S. appears to be increasing, as more and more states continue to legalize both recreational and medicinal marijuana (Benac, 2013). In 1969, only 12% percent of American voters in the USA supported legalization, which more than tripled to 44%–52% of voter support by 2012 (Carroll, 2005; Newport, 2012). A 2015 Gallup poll showed that 58% of Americans believe marijuana should be legal in the US, which is the maximum support found in Gallup’s 46 years of polling of opinions regarding marijuana legalization (Jones, 2015).

Internationally, recreational marijuana use is currently legal or non-penalized in Uruguay and the Netherlands, while nations including Australia and New Zealand are considering legalizing medicinal marijuana (Shipton & Shipton, 2014). General population surveys in Israel (where medicinal marijuana is legal) and Norway (where medicinal marijuana is illegal) show that 78% of adult Israelis and 51% of adult Norwegians are in favor of medicinal marijuana, and that beliefs regarding the medicinal benefits of marijuana influence support for legalization of medicinal marijuana more than concerns regarding addiction or spillover effects (Sznitman & Bretteville-Jensen, 2015). Similar general population surveys in Canada have shown that the majority opposes the prohibition of recreational marijuana use, despite the awareness of potential negative health effects (Hathaway, Erickson, & Lucas, 2007). However, no study to date has examined opinions regarding the legalization of recreational marijuana use among the general population since I-502 passed in November 2012.

Notably, another historic policy change—the privatization of liquor sales (I-1183) —occurred in Washington state in 2011, approximately one year prior to the passing of I-502. Voters appear to be expressing regret regarding I-1183 (Subbaraman & Kerr, in press)(Parry, 2016), in that almost 20% of those who voted in favor of I-1183 would change their votes to “No” given what has happened since privatization went into effect (e.g., increased availability; (Subbaraman & Kerr, 2016). Even when accounting for those who would change their votes from “No” to “Yes” on I-1183, Subbaraman & Kerr (in press) found that I-1183 would most likely not pass if the election were to occur today. This finding explicitly demonstrates that people who were in favor of a particular policy at one point in time can and do change their minds once the policy has been enacted and consequences have been observed.

1.1 Rationale for current study

This is the first study to identify demographic and substance use variables that correlate with how people voted on marijuana legalization, as well as whether voters would still vote the same way based on what has happened since legalization went into effect. A number of states are expected to have marijuana-related legislation on forthcoming ballots, and may follow Washington in legalization. Thus, understanding correlates of voting outcomes and marijuana-related opinions could inform how these elections might result. In addition, examining whether voters would still vote the same way gives us an idea of whether voters consider legalization in Washington a success. Specifically, this study examines the relationship between demographics, marijuana and alcohol use, and voting outcomes, as well as how these variables relate to (i) whether voters would still vote the same way (a reflection of whether people are satisfied with the new policy and its associated changes) and (ii) the likelihood of using marijuana if it could be purchased from legal stores. By capturing experiences with marijuana legalization via objective voting questions, we are able to better understand subjective reactions to marijuana legalization in Washington state.

2. MATERIALS AND METHODS

2.1 Sample

The sample consists of 2,007 adult (aged 18 and over) Washington state residents recruited in two separate waves (January–April 2014 and August–October 2014). Importantly, Washington marijuana stores officially opened in July 2014, which falls in the middle of the two waves and thus allows us to test for opinion changes post-store openings. Participants were reached via Random Digit Dial, which includes > 40% from cell phone exchanges. The American Association for Public Opinion Research (AAPOR2) cooperation rates were 50.9% (landline) and 60.9% (cell phone) in Wave 1, and 45.8% (landline) and 62.4% (cell phone) in Wave 2; the American Association for Public Opinion Research has extremely detailed standardized formulas for cooperation rates in US survey research (The American Association for Public Opinion Research, 2000). At survey completion, participants were issued $20 gift cards. Surveys lasted 30 minutes on average. This study was approved by the Public Health Institute’s Institutional Review Board. Informed consent was obtained from all participants.

2.2 Measures

Voting outcomes were determined using: “Did you vote in favor of proposition 502 legalizing marijuana production and sales to adults?” We examined correlates of voting vs. not voting, as well as voting “Yes” vs. voting “No” (among voters only). To understand whether opinions have changed since the initiation of the policy, we also examined correlates of voting the same way (vs. change their votes, among voters only). Changing vote outcomes were determined using: “Would you have voted the same way based on what has happened since legalization?” Response options for this question were “yes,” “no,” “I don’t know,” and “Refused.” The marijuana-related opinion outcome of interest was “How likely would you be to use marijuana if you could buy it legally in a store?” which was asked of non-current marijuana users only (i.e., no use in past 12 month) in order to understand demographics of potential new consumers. Response categories to the purchasing question were “very unlikely,” “unlikely,” “somewhat likely,” and “very likely”; we collapsed these categories into a binary response of “unlikely”(unlikely + very unlikely) and “likely” (somewhat likely + likely) for use in logistic regressions.

We used the following demographics as independent variables: gender, age, race (White, African American, Native American, Asian/Hawaiian/Pacific Islander, Multi-racial), ethnicity (Hispanic, non-Hispanic), education (high school diploma or less, some college, college graduate, graduate school), and employment (full-time employed, part-time employed, retired, unemployed). We also controlled for marijuana user status (lifetime abstainer, past user, current user), which was determined based on the questions “Have you ever used marijuana at any time in your life?” and “How often have you used marijuana, hash, or pot during the last twelve months?” Those who had used at least once in the last 12 months were classified as current users; drinking status (lifetime abstainer, past drinker, current drinker) which was assessed analogously; and weekly number of drinks, which was calculated from monthly frequency of drinking and average number of standard drinks per drinking occasion. Marijuana-related opinions included as covariates were “Do you think adults should be able to grow their own marijuana for personal use?” (possible responses: yes, no) and “How risky do you think weekly marijuana use is for a person’s health?” (possible responses: very, somewhat, a little, not at all, or good for health).

2.3 Statistical Analyses

First, we used bivariate chi-squared and t-tests to detect possible differences in demographics, marijuana user status and policy opinions between waves and across voter categories (voted Yes, voted No, non-voters). Then we assessed whether demographic and substance use variables were associated with whether respondents voted and how respondents voted using logistic regressions. We next examined correlates of changing one’s vote on I-502 with logistic regression (among voters only) and stratified by how respondents actually voted, using logistic regression to identify correlates of changing one’s vote separately for those who voted for and those who voted against I-502. Finally, we used logistic regression to identify correlates of respondents’ likelihood of using marijuana if it could be purchased in legal retail stores. Sampling weights were used in all analyses to adjust for the probability of selection introduced during the sampling design. All analyses were performed in Stata V.13, StataCorp, College Station, TX, USA.

3. RESULTS

Table 1 describes demographics, drinking variables, and I-502 policy opinions for the Wave 1 (N=1,202) and Wave 2 (N=805) samples overall, as well as for each voting group (voted Yes, voted No, did not vote) for Waves 1 and 2 combined. Table 1 also includes bivariate tests of differences in demographics and substance use between Waves 1 and 2 (column 1), and across voting groups (columns 2–4). Because the Waves 1 and 2 samples did not significantly differ on any variable, we combined the two to increase power in multivariable models described below (while retaining a Wave control variable for examining effects of time). Conversely, voting groups differed significantly (P<0.05) on all demographic, substance use, and policy opinion variables in bivariate tests (except for Hispanic ethnicity). Within the two combined waves, 44.6% voted in favor of legalization, 38.5% voted against legalization, and 14.4% did not vote.

Table 1.

Demographics, marijuana use and policy opinions for Washington state residents

	SAMPLE OVERALL		VOTED YES on I-502	VOTED NO on I-502	DID NOT VOTE on I-502
	Wave 1 (N=1,202)	Wave 2 (N = 805)	Waves 1 and 2 (n =895)	Waves 1 and 2 (n =772)	Waves 1 and 2 (n =289)
Male (%)	49.5	49.6	49.6	48.3	52.4
Age (%)
18–29 years	22.6	22.8	21.4	14.1	40.4***
30–39 years	17.7	16.4	18.5	11.8	23.2
40–49 years	16.5	17.3	18.2	18.0	12.1
50–64 years	25.6	25.5	26.7	29.4	17.3
65+ years	17.6	17.9	15.2	26.8	7.0
Education (%)
High school graduate or less	34.8	35.3	27.5	37.0	48.1***
Some college	34.9	33.8	36.8	33.2	31.0
College graduate	17.9	16.0	19.5	16.1	12.4
Graduate school or more	12.5	14.9	16.1	13.7	8.5
Employment (%)
Full or part-time	60.4	59.7	64.7	54.5	58.2***
Retired	18.1	19.0	17.0	26.0	8.3
Unemployed	21.6	21.3	18.3	19.4	33.5
Race (%)
White	81.1	81.3	82.4	85.0	71.4*
Black	4.2	4.4	4.7	2.8	6.4
American Indian/Alaska Native	3.4	1.9	2.4	3.2	2.6
Asian/HI/Pacific Islander	7.8	7.4	6.2	5.8	12.5
Ethnicity (%)
Hispanic	9.3	9.5	9.8	8.6	8.4
Non-Hispanic	90.4	89.9	89.8	90.8	91.1
Drinking status (%)
Lifetime abstainer	7.0	7.8	2.4	10.8	13.0***
Past drinker (no drinking in past 12 months)	24.7	22.8	17.5	32.8	22.6
Used in past 12 months	68.3	69.4	80.2	56.4	64.4
Average weekly # drinks (mean (SE))	4.3 (0.3)	6.5 (0.9)†	7.0 (0.7)	3.7 (1.0)	5.0 (0.8)*
Marijuana use status (%)
Lifetime abstainer	42.6	36.8	23.8	61.3	35.5***
Past user (no use in past 12 months)	34.2	36.2	40.4	28.9	33.4
Used in past 12 months	23.2	27.0	35.8	9.8	31.1
Do you think marijuana should be legal for adults? (%)
Yes	64.6	66.9	95.4	26.8	69.2***
No	29.9	27.5	3.4	64.6	23.5
I don’t know	5.3	4.9	1.2	7.6	6.5
Do you think adults should be able to grow their own marijuana for personal use? (%)
Yes	59.3	61.1	83.6	27.7	67.6***
No	36.6	35.2	13.5	67.7	28.9
I don’t know	3.6	3.2	2.6	4.2	2.6
How risky do you think weekly marijuana use is for a person’s health? (%)
Very	17.9	18.1	3.5	39.3	14 2***
Somewhat	25.5	22.6	20.0	30.2	22.0
A little	21.0	25.0	29.5	14.4	24.3
Not at all	20.1	19.5	29.2	7.1	20.2
Good for health	9.6	8.5	13.2	2.6	12.1
I don’t know	5.2	5.5	3.6	5.8	6.7
Would you still vote the same way?(%)
Yes	88.6	88.0	92.2	83.4***	N/A
No	8.8	8.9	4.8	14.0
I don’t know	2.4	2.8	2.8	2.3
How likely would you be to use marijuana if you could buy it legally in a store?c (%)
Very unlikely	70.0	74.3†	57.3	85.1	75.2***
Unlikely	17.5	16.2	23.1	12.3	12.5
Somewhat likely	7.3	4.8	11.7	0.8	7.4
Very likely	3.9	4.2	6.9	1.8	4.1
I don’t know	1.1	0.1†	0.5	0.0	0.8

^aDifferences tested between Waves 1 and 2 for sample overall (first column) and across voting groups for Waves 1 and 2 combined (columns 2–4)

^bOnly asked of those who reported using marijuana in the past 12 months (Wave 1 n = 213; Wave 2 n = 168)

^cOnly asked of those who reported never using marijuana in the past 12 months (Wave 1 n = 985; Wave 2 n = 628)

Boldface indicates statistical significance:

^***P<0.001;

^**P<0.01;

^*P<0.05;

^†P<0.10

3.1 Correlates of I-502 voting outcomes

Table 2 displays adjusted odds ratios from logistic regressions of voting outcomes. In terms of voting vs. not voting on I-502, older age, higher education, and current alcohol use were all related to higher odds of voting. The odds of voting appeared to increase as age increased: those aged 40–49 had 3.00 times the odds and those aged 65+ had 5.61 times the odds of voting at all compared to those aged 18–29 years (all Ps < 0.001). Those with more than a high school education had approximately double the odds of voting compared to those with only a high school education or less. Compared to lifetime alcohol abstainers, current drinkers had almost double the odds of voting. Finally, Asian/Hawaiian Islanders/Pacific Islanders had about half the odds of voting than Whites, while Hispanics had almost four times the odds of voting than non-Hispanic Whites.

Table 2.

Adjusted odds ratios with 95% confidence intervals from weighted regressions of demographics and substance use variables regressed on Initiative 502 voting-related outcomes

Predictors	Outcomes
	Voted on I-502 (vs. Did not vote)a (Total sample, N = 1,800)	Voted Yes on I-502 (vs. Voted No)b (Voters only, n = 1,547)
Male (vs. Female)	0.85 (0.60, 1.20)	0.84 (0.64, 1.10)
Age (vs. 18–29 years old)
30–39	1.40 (0.85, 2.30)	1.20 (0.69, 2.07)
40–49	3.00 (1.751, 5.10)***	0.72 (0.42, 1.25)
50–64	3.46 (2.11, 5.68)***	0.55 (0.34, 0.90)*
65+	5.61 (2.80, 11.26)***	0.50 (0.28, 0.89)*
Education (vs. High school graduate or less)
Some college	1.70 (1.14, 2.55)*	1.54 (1.06, 2.25)*
College graduate	2.23 (1.39, 3.56)**	2.14 (1.45, 3.16)***
Graduate school or more	2.05 (1.21, 3.48)**	2.03 (1.33, 3.09)***
Employment (vs. Full or part-time)
Retired	0.96 (0.52, 1.77)	1.35 (0.89, 2.05)
Unemployed	0.64 (0.43, 0.96)*	0.79 (0.52, 1.20)
Race (vs. White)
Black	0.69 (0.31, 1.54)	1.41 (0.62, 3.20)
American Indian/Alaska Native	1.18 (0.34, 4.17)	0.65 (0.25, 1.72)
Asian/HI/Pacific Islander	0.49 (0.27, 0.89)*	1.01 (0.52, 1.97)
Multi-racial	0.54 (0.24, 1.23)	0.91 (0.37, 2.27)
Ethnicity (vs. Non-Hispanic)
Hispanic	3.80 (1.26, 11.47)*	1.11 (0.55, 2.24)
Marijuana use status (vs. Lifetime abstainer)
Past user (no use in past 12 months)	0.81 (0.53, 1.23)	3.70 (2.70, 5.06)***
Used marijuana in past 12 months	0.88 (0.55, 1.40)	8.18 (5.29, 12.65)***
Drinking status (vs. Lifetime abstainer)
Past drinker (no drinking in past 12 months)	1.79 (0.93, 3.45)†	1.70 (0.86, 3.35)
Used alcohol in past 12 months	1.94 (1.06, 3.55)*	3.16 (1.66, 6.02)***
Average weekly # drinks, past 12 months	1.00 (0.99, 1.01)	1.00 (0.99, 1.01)
Wave (Wave 2 vs. Wave 1)	0.92 (0.66, 1.29)	0.88 (0.67, 1.15)

Boldface indicates statistical significance:

^***P<0.001;

^**P<0.01;

^*P<0.05;

^†P<0.10

^aF = 5.82, Prob > F = 0.000

^bF = 9.90, Prob > F = 0.000

Among voters, correlates of how individuals voted on I-502 were similar to correlates of whether people voted. Individuals aged 50–64 and 65+ had about half the odds of voting Yes (vs. No) compared to those aged 18–29 years old. Higher education (vs. high school or less) was related to approximately double the odds of voting Yes on I-502, while current drinking (vs. lifetime abstention) was related to approximately triple the odds. Compared to lifetime marijuana abstainers, past users had 3.70 times the odds of voting Yes while current users had 8.18 times the odds (Ps<0.001).

3.2 Success of I-502

The outcome of whether voters would vote the same way vs. change their votes was used as a proxy to determine the success of I-502. In a multivariable model including all voters and controlling for how people voted (not shown), the odds of wanting to change one’s vote among Yes voters were 0.46 the odds of wanting to change one’s vote among No voters (P<0.036, 95% CI: 0.23, 0.95). We then stratified by how people voted to better understand any differences in correlates of wanting to change one’s vote between the Yes and No voters. Table 3 (column 1) presents results from the No voters; older age and higher education were significantly related to lower odds of wanting to change one’s vote (Ps<0.05). Marijuana-related opinions were also significantly associated with wanting to change one’s vote in that those who were unsure about growing for personal use had higher odds of wanting to change their votes than those who disagreed with growing for personal use. On the other hand, those who were unsure about marijuana’s riskiness had 0.04 the odds of wanting to change their votes compared to those who reported that marijuana is very risky. Notably, age, education, and marijuana-related opinions were similarly associated with wanting to change one’s vote among voters overall and among those who voted yes (results not shown).

Table 3.

Adjusted odds ratios with 95% confidence intervals from weighted regressions of Initiative 502-related outcomes

Predictors	Outcomes
	Would change vote (vs. Not)b (Voted No only, n = 690)	Likely to buy MJ in legal store (vs. Not)a, c (Non-current users only, n = 1,473)
Male (vs. Female)	1.47 (0.79, 2.74)	0.91 (0.57, 1.44)
Age (vs. 18–29 years old)
30–39	0.53 (0.17, 1.71)	0.78 (0.35, 1.73)
40–49	0.25 (0.09, 0.65)**	1.08 (0.50, 2.32)
50–64	0.36 (0.15, 0.89)*	0.85 (0.44, 1.63)
65+	0.37 (0.11, 1.20)†	0.53 (0.22, 1.27)
Education (vs. High school graduate or less)
Some college	0.32 (0.15, 0.69)**	0.97 (0.54, 1.74)
College graduate	0.31 (0.13, 0.75)**	0.91 (0.48, 1.72)
Graduate school or more	0.43 (0.16, 1.16)†	0.54 (0.26, 1.14)
Employment (vs. Full or part-time)
Retired	0.54 (0.17, 1.68)	1.56 (0.74, 3.30)
Unemployed	0.69 (0.30, 1.60)	1.05 (0.58, 1.90)
Race (vs. White)
Black	1.51 (0.31, 7.39)	1.76 (0.65, 4.77)
American Indian/Alaska Native	1.26 (0.18, 9.10)	1.34 (0.39, 4.81)
Asian/HI/Pacific Islander	2.71 (0.79, 9.28)	0.27 (0.05, 1.41)
Multi-racial	0.64 (0.09, 4.79)	0.85 (0.21, 3.48)
Ethnicity (vs. Non-Hispanic) Hispanic	2.45 (0.59, 10.12)	1.37 (0.44, 4.21)
Marijuana use status (vs. Lifetime abstainer)
Past user (no use in past 12 months)	0.95 (0.49, 1.85)	2.46 (1.41, 4.28)**
Used marijuana in past 12 months	0.51 (0.10, 2.53)	N/Aa
Drinking status (vs. Lifetime abstainer)
Past drinker (no drinking in past 12 months)	0.62 (0.20, 1.96)	1.42 (0.36, 5.58)
Used alcohol in past 12 months	0.75 (0.25, 2.27)	2.09 (0.57, 7.70)
Average weekly # drinks, past 12 months	1.00 (0.99, 1.01)	0.99 (0.98, 1.01)
Do you think adults should be able to grow MJ for their own personal use? (vs. No)
Yes	1.45 (0.63, 3.32)	1.88 (1.04, 3.41)*
I don’t know	4.61 (1.49, 14.28)*	1.25 (0.34, 4.59)
How risky do you think weekly MJ use is for a person’s health? (vs. Very)
Somewhat	0.63 (0.31, 1.28)	2.21 (0.83, 5.89)
A little	0.74 (o.30, 1.84)	4.05 (1.59, 10.32)**
Not at all	1.44 (0.38, 5.50)	4.96 (1.82, 13.52)**
Good for health	0.51 (0.02, 11.44)	12.55 (4.11, 38.31)***
I don’t know	0.04 (0.004, 0.37)**	3.09 (0.93, 10.28)†
Wave (Wave 2 vs. Wave 1)	1.16 (0.66, 2.05)	0.69 (0.43, 1.10)

Boldface indicates statistical significance:

^***P<0.001;

^**P<0.01;

^*P<0.05;

^†P<0.10

^aPurchasing question only asked of respondents who reported no cannabis use in the past 12 months

^bF = 1.87, Prob > F = 0.005

^cF = 3.22, Prob > F = 0.000

3.3 Correlates of using marijuana purchased in legal stores (among non-current users)

The final outcome of interest here was respondents’ likelihood of using marijuana if it were purchased legally in a store. This question was only asked of non-current users only (i.e., those who had not used in the past 12 months, which includes lifetime abstainers as well as those who had used in their lifetimes but not in the past 12 months). As shown in Table 3 (column 2), past marijuana users had 2.46 the times of responding that they were somewhat/very likely to use marijuana if they could buy from legal stores than lifetime marijuana abstainers (P<0.01). Those who think adults should be able to grow marijuana for personal use had 1.88 the odds of being somewhat/very likely to use marijuana purchased from legal stores (P<0.05). Similarly, the odds of being likely to use marijuana purchased in legal stores appeared to increase as one’s belief in the riskiness of marijuana decreased (Ps<0.01).

4. DISCUSSION

Most notably, those who voted for marijuana legalization had half the odds of wanting to change their votes compared to those who voted against it. Thus, support for marijuana legalization appears to have increased; if we reclassify the 4.8% Yes voters who would change their votes to Nos and the 14% of No voters who would change to Yeses, we can see that the proportion of voters in favor of I-502 has increased since the 2012 election.

Interestingly, 26.8% of those who voted against marijuana legalization reported that marijuana should be legal; this could be because marijuana advocates were not satisfied with how I-502 was written (Leon & Weitzer, 2014), and/or could imply that support for legalization has increased since November 2012. In addition, multivariable models show that those who voted Yes were more likely to keep their votes the same than those who voted No, which suggests that any immediate repercussions that No voters might have expected have not occurred. Among No voters, younger age and less education were related to changing one’s vote; perhaps these groups may have more malleable opinions than older or more educated voters.

We also found that older age, higher education, and current alcohol use were significantly related to voting (at all) on I-502, while younger age, higher education, lifetime marijuana use, and current drinking were significantly related to voting Yes. Voting groups differed on all demographics, substance use and policy opinion variables, except Hispanic ethnicity; the non-voter group appears to be driving these differences. Hispanic ethnicity was the only factor that did not differ across the three voting groups, perhaps because Hispanics have the lowest voter turnout of racial/ethnic groups in the US (McDonald, 2016). Those who did not vote on I-502 were significantly younger than those who did vote, partly because a small proportion (~2%) of survey respondents was not of the US voting age of 18 years in 2012. Greater quantity/frequency of marijuana use were related to significantly higher odds of support for marijuana legalization in the 2007–2011 Monitoring the Future samples, though 17% of non-users also reported supporting legalization (Palamar, 2014).

The correlates of voting on I-502 are the same as correlates of voting on I-1183 (Subbaraman & Kerr, 2016), except that spirits drinkers were more likely to vote (vs. not vote) on I-1183. Spirits drinker status was not related to voting on I-502. Furthermore, 53.0% of eligible voters voted on I-1183 (Wyman, 2011) while 81.3% of eligible voters voted on I-502 (Reed, 2012). In the US, initiative measures can be proposed by individual citizens or corporations, and differ from legislation written by the state or local government. Elections for initiative measures can occur every year, while presidential elections occur every four years. Initiative 502 was on the ballot during a presidential election year, which might explain differences in voter turnout between the two measures; the US Census Bureau’s Current Population Survey does show that presidential election years consistently have higher voter turnout than other years (McDonald, 2014). Or perhaps voters are more interested in marijuana legalization than privatization of liquor sales.

In terms of the correlates of voting Yes, the same set of variables—older age, having at least a Bachelor’s degree, and marijuana users—were all significantly related to supporting marijuana legalization in a large (N=1,081) random household survey of adults in Toronto, Canada (Hathaway et al., 2007). We also found that older age, higher education, and ambivalent marijuana-related opinions were similarly associated with wanting to keep one’s vote the same regardless of whether the subpopulation was all voters, those who voted in favor of I-502, or those who voted against the initiative; this finding may indicate that these factors are predictive of stronger voting convictions rather than perceptions regarding I-502 in particular.

Past marijuana were more likely to use marijuana if they could buy it from legal stores than lifetime abstainers, which implies that marijuana legalization could be a gateway for return to marijuana use among those who have not used in the past year. This suggests that marijuana legalization might increase the number of current (but not lifetime) users, possibly due to decreased stigma and probability of legal trouble. These results could be used for estimating marijuana market sizes, as well as for predicting potential new/returning consumers. In terms of clinical implications, physicians should consider discussing intentions to restart marijuana use with past users, especially if marijuana could potentially interact with any medical treatments.

We also found that younger people were less likely to vote at all while still more likely to vote in favor of legalization. The current results are extremely similar to those found in the recent study of marijuana-related voting behaviors among college students. In the college student study, half of respondents voted for/would vote for legalization and 6–16% reported either not voting in the Washington election or indicated that they would abstain from voting in a similar election if one were to occur in Wisconsin (Moreno et al., in press); in the current study, 44.6% of respondents in this study reported voting for legalization and 14.4% did not vote. In addition, stronger intentions to use marijuana were associated with favoring legalization in the college student study; although we did not ask about intent to use marijuana in the future, we did find that current and past marijuana use increased the relative odds of voting in favor of legalization by 8.18 and 3.70 respectively. Interestingly, neither current nor past use was associated with favoring legalization in the college student study (Moreno et al., in press); these discrepancies could be due to differences in covariates across multivariable models, as the prior study included marijuana attitudes, intentions, behaviors and perceived norms.

Perceptions of the riskiness of marijuana are likely related to more liberal marijuana policies. Analyses using the National Survey on Drug Use and Health data and testing for temporal changes in marijuana attitudes and use outcomes in Colorado between 2003–2011 showed that the proportion of respondents reporting “great risk” from using marijuana 1–2 times/week dropped significantly among all age groups, and specifically from 45% to 31% among those aged 26 and older (Schuermeyer et al., 2014); this drop was related to several 2009 policy changes that lessened penalties and restrictions on marijuana use. Thus, states that have legalized or will legalize marijuana in the future need to be prepared to appropriately warn the public of potential harms related to marijuana use.

Public perceptions of the health risks and/or benefits from marijuana use are likely influenced by the media as well, especially because the majority of Americans get their public health information from the media (Brodie, Hamel, Altman, Blendon, & Benson, 2003). A study of news stories from high circulation/viewership news outlets for the years 2010–2014 (with a focus on the four states with legalized recreational marijuana: Alaska, Colorado, Oregon, and Washington) categorized a sample of news stories as containing 1) only pro-legalization arguments, 2) only anti-legalization arguments, 3) both pro- and anti- arguments, or 4) no arguments (McGinty et al., in press). Within Washington state, stories from the newspaper print versions of the Spokesman Review (considered more conservative, based on its 2012 Presidential candidate endorsement) and the Seattle Times (considered more liberal) were both most likely to contain no arguments regarding legalization and both least likely to contain only pro-legalization arguments. Stories from the Seattle local television outlets were uniformly spread across the four categories, while the distribution of stories within national news outlets (e.g., New York Times, PBS News) also appeared unbiased (McGinty et al., in press). Thus, it is unlikely that the discussions regarding marijuana legalization in the general media have affected our results.

4.1 Study strengths and limitations

Data come from a large, general population sample that was surveyed before (Wave 1) and after (Wave 2) marijuana retail stores opened in Washington state. Because voters and non-voters tend to have similar policy opinions (Leighley & Nagler, 2007), and because survey weights were used in order to standardize results according to Washington’s demographic breakdown, the current results likely reflect public opinions regarding Washington’s legalization of marijuana. However, less than two years had passed between the policy change and data collection, implying that there may be longer-term consequences that sway public opinion in the future. Furthermore, adverse public health effects of the policy change may take more time to emerge, e.g., the negative effects of smoking marijuana could result in higher incidence of respiratory disorders and/or potentially higher rates of marijuana use among adolescents could lead to more neurocognitive problems among young adults in the future. It is also possible that prevalence for support has increased since legalization simply because marijuana is currently legal and there may be people who tend to favor the status quo; however, this type of bias would not substantively change the current findings. Furthermore, the survey is cross-sectional and thus susceptible to reporting biases, especially because self-report regarding marijuana use and related policy opinions may be influenced by social acceptability issues.

4.2 Conclusion

Since November 2012, support for marijuana legalization in Washington state has increased; more than a quarter of those who had voted against legalization now report supporting legalization, and accounting for the proportion of voters who would change their votes suggests that I-502 would pass today with even more votes in favor. Thus, marijuana legalization may not have come with the immediate drawbacks expected by those who were once against it. However, the effects of policy changes on public health and social welfare often take time to develop. Thus, states leaning towards marijuana legalization might consider waiting to see whether adverse consequences arise in Washington over a longer period of time.

Biographies

Meenakshi Sabina Subbaraman, PhD, is a Biostatistician at the Alcohol Research Group, Public Health Institute in Emeryville, CA. She is currently working on various projects related to cannabis and alcohol policies, and recently received funding from the US National Institutes of Health to examine how cannabis use affects alcohol treatment outcomes. She completed her MS in statistics at Stanford University and her PhD in epidemiology at UC Berkeley.

William C. Kerr, Ph.D., is a senior scientist at the Alcohol Research Group, Public Health Institute, and is Director of ARG’s National Alcohol Research Center focused on the epidemiology of alcohol problems and alcohol-related disparities. Dr. Kerr received his Ph.D. in economics from the University of California-Davis. Since joining the Alcohol Research Group in 2001 Dr. Kerr has pursued research in the areas of alcohol policy, measurement methodology, trends in US alcohol consumption with a focus on age-period-cohort decompositions and relationships between alcohol use patterns and health outcomes. He has published over 80 articles as well as reports and book chapters. He is currently leading a projects evaluating the privatization of the spirits market in Washington and analyzing inter-relationships between alcohol consumption patterns and health problems in a life-course perspective.