OP62 Does smoking, drinking, physical activity, and self-reported health mediate the relation between education and voting during midlife? Evidence from the 1958 national child development study. (3rd September 2019)

• Record Type: Journal Article
• Authors: Gagné, T; Schoon, I; Sacker, A
• Is Part Of: Journal of epidemiology and community health. Volume 73(2019)Supplement 1
• Page Start: A30
• Page End: A31
• Publication Date: 2019-09-03
• Subjects: Voting -- Education -- Health inequalities; Public health -- Periodicals; Epidemiology -- Periodicals; 614.4
• DOI: 10.1136/jech-2019-SSMabstracts.64
• Languages: English
• ISSNs: 0143-005X

Abstract

Abstract : Background: Education is a strong predictor of voting in most Western countries. New studies, predominantly from the United States, question the role of health as a mechanism reinforcing social inequalities in voting over the life-course. In the United Kingdom, we previously found that heavy smoking, physical inactivity, poor self-reported health, and, to a lesser extent, drinking over the recommended limit were each associated with a lower probability of voting in the 1958 National Child Development Study (NCDS). Building on these findings, this study examines the proportion of the association between educational attainment and voting that is mediated through these health indicators. Methods: We used the data of 6, 166 NCDS participants who responded to each sweep between the ages of 23 and 50. We examined educational attainment at the age of 23, smoking, drinking, physical activity, self-reported health at the ages of 23, 32, and 42, and voting behaviour at the age of 42, 46, and 50. Adjusting for non-response and attrition using inverse-probability weights and missingness using multiple imputation, we compared prevalence ratios to examine: 1) the direct effect of having a higher level of education on voting when accounting for these health indicators, and 2) the proportion of its total effect that is eliminated by this adjustment. Results: Contrasting extreme groups, we found that participants with a degree (NVQ5/6) at the age of 23 were 25%, 28%, and 32% more likely to have voted in the last general election compared to those with no qualifications at the ages of 42, 46, and 50. Adjusting for health indicators at the ages of 23, 32, and 42, participants with a degree remained 19%, 24%, and 27% more likely to have voted at the ages of 42, 46, and 50, respectively. This translates into an average proportion eliminated of 22%, 18%, and 16% across these age points. Testing mediators separately, we found that smoking, physical activity, and self-reported health were each likely to contribute to the 'education-voting' association. Discussion: In keeping with health promotion principles, health represents beyond the absence of disease a resource for individuals, their social network, and their communities. Our findings suggest that health and its behavioral determinants are likely to explain a portion of social inequalities in voting over the life-course. Research and intervention should address the specific health-related mechanisms through which current electoral processes may unequally influence voter turnout across social groups.