Political ideology and party affiliation as determinants of attitudes toward legalization of Marijuana in the United States: Implications for policy conversations

This study examined data gathered in the General Social Survey in the years 2018, 2022, and 2024 (Davern et al., 2024, 2025; Smith et al., 2019).² The GSS, a full-probability survey, has monitored social trends in the US since 1972. Data and codebooks for the current study were downloaded from the website of the National Opinion Research Center at the University of Chicago (see gss.norc.org). In 2018, the GSS surveyed 2348 individuals, asking 1447 about the legalization of marijuana. With listwise deletion in logistic regression analyses, the current study included responses from 1322 (91.4%) of the 1447. The 2022 GSS included 4149 respondents and consisted of both the GSS cross-section and an oversampling of minorities from the NORC AmeriSpeak^® panel. Of 4149 respondents, 1123 were asked about marijuana, and the current study includes 949 (84.5%) of the 1123. Lastly, in 2024, the GSS surveyed 3309 individuals, asking 867 respondents about marijuana legalization; the current study included responses from 741 (85.5%) of the 867. In all, the study included responses from 3012 (87.6%) of 3437 who expressed an opinion on legalization of marijuana. As indicated, year of GSS study served as an explanatory measure in a logistic regression analysis containing all respondents.

Table 2a contains model statistics for the overall analysis. Each model in the three-step logistic regression procedure showed improvement over the model before it. Cox & Snell pseudo-R² values increased at each step as well. Table 2b contains the results of the three-step regression analysis that first included (a) GSS year and the five demographic measures, then (b) GSS year, five demographic measures, and political ideology, and finally (c) GSS year, five demographic measures, political ideology, and party affiliation as determinants of attitudes toward legalization of marijuana.

In Table 2b, Model 1 shows that respondents in 2022 and 2024 were significantly less likely to oppose legalization, as were males, younger respondents, and individuals other than those who had not graduated from high school. Relative to respondents on the Pacific coast, those who lived internationally at age 16 were more likely to oppose the legalization of marijuana, as were individuals in the South Atlantic, East South Central, West South Central, and Mountain regions.⁵ Individuals in these regions tend to be more conservative than individuals living the Northeast and on the Pacific coast,⁶ and in fact Model 2 indicates that when political ideology was entered as an explanatory measure, three of the four regions identified in Model 1 lost statistical significance. As shown in Model 2, liberals and moderates appeared significantly less likely to oppose legalization, while demographic variables apart from region showed similar effects to the estimate in Model 1. Those patterns held in Model 3, although fewer estimates for ideology showed significance, and an overall effect for race appeared. Examining the model, it appears whites were more likely to support legalization and African Americans were more likely to oppose it, but neither race category exceeded chance. Conversely, all categories in the party identification measure were statistically significant, with Democrats and Independents appearing more supportive of legalization. Interaction terms did not show statistical significance, and thus neither ideology nor party identification showed changes across the three years of GSS surveys.

But the presence of party affiliation did affect ideology estimates when each period was examined independently. For example, in 2018, four categories of ideology – extremely liberal, liberal, slightly liberal, and moderate – showed initial significance; however, when party identification was added, only one category – extremely liberal – remained significant. In 2022, the same ideology estimates showed initial significance, and the inclusion of party affiliation resulted in the slightly liberal and moderate categories losing significance. Finally, in 2024, three categories of ideology showed initial significance, and when party identification was entered, none remained significant. Additionally, for the first time, ideology lost overall significance as an explanatory measure. Thus, while the effects of ideology and affiliation on attitudes toward marijuana legalization did not show changes in themselves across GSS periods, the inclusion of party identification did affect ideology to some extent in each period. While the significance levels of other explanatory measures remained largely the same, categories of ideology changed. The results offer limited support for group identity as a determinant of attitudes toward marijuana legalization.

This research examined political ideology and party affiliation as determinants of attitudes toward the legalization of marijuana. A stepwise logistic regression analysis showed that survey participants, in general, became more supportive of legalization after 2018. Males expressed greater support, as did younger respondents, those with a high school diploma or higher, and those who lived in the Pacific region at age 16. Differences in the categories of political ideology appeared significant prior to the inclusion of party affiliation; while some differences remained after its inclusion, affiliation appeared a stronger determinant. Significant differences across nearly all affiliation categories appeared, as did an overall effect. Democrats and Republicans who were “strongly” committed to their parties showed the most (Democrats) and the least (Republicans) support for marijuana legalization, lending support to social identity theory and the notion of group partisanship in American politics.

But as Back et al. (2022) observed in an article about cannabis legalization in Sweden, the “party line” does not explain attitudes in all instances. Participants in their study received partisan cues that either endorsed or objected to legalization. When an ingroup party opposed the policy, in line with the participant's view, and the outgroup party supported the policy, partisans remained opposed. But when an ingroup party endorsed the policy, which went against a participant's beliefs, and an outgroup opposed the policy, partisans retained their prior attitudes; that is, their views in opposition did not weaken in the second scenario. Recent policy actions in the United States provide opportunities to examine similar phenomena.

In December 2025, US President Donald Trump signed an executive order downgrading marijuana (cannabis) from a Schedule I to a Schedule III Controlled Substance at the federal level (Broadwater and Southall, 2025). Schedule I is the most restrictive class and includes drugs such as heroin and LSD, both of which have strong potential for dependency and no accepted medical use. Schedule III substances have some potential for dependency but also have legitimate medical indications (e.g., anabolic steroids, ketamine). In general, reclassification of marijuana lessened the gap between federal and state laws, as medical marijuana is available in 48 states, Washington DC, and three US territories, and it is legal for recreational use in 24 states (Davis and Yelenik, 2025).

While individuals across the political spectrum support the medical use of marijuana, support for recreational use is less consistent (see McGinty et al., 2017). Experimental research might examine attitudes toward the change from Schedule I to Schedule III both between and within political parties. To the extent that group identity influences attitudes, one might observe differences based on party identification. But within a given party, how might individuals who supported the classification of marijuana under Schedule I communicate their attitudes given that a US president in their own party changed its status? In times of intense partisanship, would they stand by their prior attitude or express an opinion consistent with the party ingroup? Additionally, to what extent might attitudes shift when the idea for a change in classification is attributed to former US President Joseph Biden instead of Donald Trump? Biden had previously introduced a decriminalization initiative that included a possible change in classification (Broadwater and Southall, 2025).

Regarding limitations of the current study, this research examined three GSS datasets with notably different sample sizes. While logistic regression analyses were robust to these differences, sample sizes would ideally be closer in size. Additionally, in staying consistent with past GSS questions about marijuana, the GSS did not distinguish between medical and recreational use but asked a more general question. Future research should examine whether group dynamics observed in the current study vary based on medical and recreational use of marijuana. While the study used social identity theory and self-categorization to create a conceptual framework, it should be noted that the study did not test the theories directly by way of experimental design; rather, it used the theories to generate expectations for statistical patterns. Lastly, because statistical analyses relied on variables included in the three GSS datasets, use of single-item measures was necessary. Future research should include measures that lend themselves to the formation of indices.

Overall, this study found limited support for social identity theory, which suggests that individuals seek group memberships in order to build self-esteem and develop a sense of belonging. Statistical analyses showed stronger, more consistent effects for political affiliation than for political ideology as an attitudinal determinant. These findings culminated in the loss of ideology as a statistically significant explanatory measure in the 2024 General Social Survey.