An Examination of Political Orientation as a Determinant of Vaccination Outcomes and Health Mitigation Behaviours During the COVID-19 Pandemic

Abstract

Background: Vaccine hesitancy has been a longstanding challenge for public health. However, the COVID-19 pandemic brought a strong push to explore the issue with renewed vigor. Prior studies have identified social cognitive and demographic factors as determinants of vaccine hesitancy and mitigation behaviors more broadly. However, political factors seemed to be especially important during the COVID-19 pandemic. The literature examining political orientation as a predictor of COVID-19 vaccination and other mitigation behaviors focussed largely on the United States and the United Kingdom, with few studies exploring the issue within the Canadian context. This project utilized data from the Canadian COVID-19 Experiences Survey (CCES)—part of the Canadian COVID-19 Experiences Project (CCEP)—to assess the relationship between political orientation and COVID-19 mitigation outcomes in a national sample.

Specific Aims: The primary aim of this project was to test political orientation as a predictor of vaccination status and COVID-19 mitigation behaviour consistency, and to examine if this relationship differs by immunocompromised status (IC). The secondary aim was to examine the association between political orientation and reliance on a variety of information sources (e.g., social media, health professionals, family members). The third and final aim was to examine the association between political orientation and change in both vaccination status and mitigation behaviour frequency between Waves 1 and 2 of the survey. It was hypothesized that relatively more right leaning political orientation would be associated with lower likelihood of being vaccinated, lower consistency in mitigation behaviors, and more reliance on non-traditional information sources for information about COVID-19.

Methods: Data from Wave 1 (September 28th, 2021, to October 21st, 2021) and Wave 2 (March 3rd to March 21st, 2022) of CCES was used in a secondary analysis. Wave 1 had a total sample size of 1958, with 983 (50.2%) fully vaccinated, 848 (43.3%) unvaccinated, and 127 (6.5%) with 1 dose and no intent to finish. Wave 2 had a total sample size of 1848, with 1010 (54.7 %) fully vaccinated, 825 (44.6 %) vaccine hesitant, and 13 (7.0%) non-hesitant single-dose. Logistic regression was used to assess the association between political orientation and vaccination status. Multivariate general linear modelling was used to examine the association between political orientation, mitigation behaviour consistency and information source reliance. Likewise, multivariate general linear modelling was also used in prospective analyses to examine the association between political orientation and change in vaccination status and mitigation behaviour consistency. Moderation by immunocompromised status—for vaccine uptake and mitigation behaviour consistency—and covariates (demographic factors, etc.) was also examined.

Results: Relatively right-leaning political orientation was identified as a significant predictor of reduced odds of being vaccinated in Wave 1 (fully adjusted model: OR=.35; 95% CI = .30, .41) and Wave 2 (fully adjusted model: OR=.33; 95% CI = .29, .39). Moderation by immunocompromised status was observed to be significant in Wave 1 (IC OR=.58; 95% CI = .33, 1.02; non-IC OR=.31; 95% CI = .26, .36), but not in Wave 2. Relatively right-leaning political orientation was also significantly associated with decreased mitigation behaviour consistency in Wave 1 (fully adjusted model: masking F=3.988, p <.001, ηp2= .229; distancing F=3.494, p <.001, ηp2= .206; hand hygiene F=1.767, p <.001, ηp2= .116) and Wave 2. Moderation by immunocompromised status was significant for all three behaviours in Wave 1 for raw (masking F=5.463 (1, 1655), p =.0195; social distancing F=4.9329 (1, 1652), p =.0265; hand hygiene F=9.4275 (1, 1653), p =.0022) and partially adjusted models (only hand hygiene was significant in fully adjusted models). Stratifying by immunocompromised status revealed that effects were stronger for immunocompromised respondents than those who were not immunocompromised. In Wave 2, only masking was significantly moderated in unadjusted models, however the effect was stronger for non-immunocompromised respondents (Effect=.1065, p<.001) than for immunocompromised respondents (Effect=.0612, p=.0032). Reliance on all sources of information, except friends, was significantly predicted by relatively right-leaning political orientation in Wave 1. In Wave 2 only religion, print media, television, and other sources were significantly associated with relatively right-leaning political orientation. Lastly, prospective analyses indicated that relatively right-leaning political orientation was associated with significant decreases in masking, distancing, and hand hygiene between Waves 1 and 2 in raw and adjusted models (fully adjusted model: masking F=2.470, p <.001, ηp2= .221; distancing F=1.451, p =.005, ηp2= .143; hand hygiene F=1.280, p =.045, ηp2= .128). However, change in vaccination was not significantly predicted by political orientation, across any model.

Conclusion: Relatively right-leaning political orientation was observed to be a significant predictor of numerous COVID-19 related health and behavioural outcomes. Relatively greater right-wing orientation was associated with significantly reduced odds of being vaccinated, and reduced consistency of masking, social distancing, and hand hygiene. Immunocompromised status moderated these associations, in both expected and unexpected ways. The findings of this study were largely in line with existing literature, but provide important insight into the Canadian context, and may serve as a tool to guide future decision making for public health stakeholders.