Age and vote choice: Is there a conservative shift among older voters? Benny Geys, Tom-Reiel Heggedal, Rune J. Sørensen. Electoral Studies, Volume 78, August 2022, 102485. https://doi.org/10.1016/j.electstud.2022.102485
Abstract: Ageing is often believed to induce a movement towards the right of the political spectrum. Yet, empirical evidence remains inconclusive due to a dearth of longitudinal datasets covering multiple cohorts. Using eleven rotating panels of the Norwegian Election Studies (1977–2017) and exploiting first-derivative properties of the vote choice function, our empirical approach identifies non-linear life-cycle effects while controlling for cohort and period effects. Our main findings indicate that shifting towards the left is more likely among the young (under 40 years) whereas shifting towards the right occurs at an older age (over 55 years). Evaluating potential mechanisms, we find that individuals’ income, retirement, family status and political interest explain only a small part of the observed ageing effect. Life-cycle shifts in (some) policy preferences may play a bigger role. Finally, aging effects are similar across women and men, and only marginally stronger among groups with lower education and income levels.
Keywords: Population ageingVote choiceAge-period-cohortRotating panelNorway
4. Conclusion
Up to the 1960s, party choice was generally viewed as a result of stable party identifications and social class (“Michigan school”). Party preferences and political attitudes were assumed to be shaped during an individual's ‘formative’ or ‘impressionable’ years in adolescence and early adulthood, and to remain highly stable thereafter. Subsequent literature has criticized this view based on the recognition that citizens may well change their partisan preferences and ideological positioning over the life cycle. Consistent with the latter view, our analysis shows that people shift to more left-leaning parties up to midlife and then become more likely to swing back towards rightist political parties. These results are obtained using four decades of high-quality Norwegian Election Studies, which feature individual-level rotating panels that allow us to track party shifts over a fixed four-year election cycle.
While our analysis rests on a methodological innovation that allows identifying (non-linear) age effects while controlling for cohort and period effects, it naturally remains a challenge to track the exact mechanisms underlying this result. Nonetheless, our findings by and large reject the notion that social and political ageing can account for much of individuals' tendency to shift towards the right of the political spectrum as they age. This also implies that shifts in party choices over the life cycle are unlikely to be due to differences in individuals’ reliance on public sector transfers and services that accompany such social and political ageing. An alternative mechanism is that party choices can shift over the life-cycle due to changing perspectives on national identify and cultural values. Our evidence here provides partial support. While we find no evidence that fiscal policy preferences affect our estimated age affect, immigration, environmental and rural policy preferences do appear to account for a significant part of the observed ageing effect in vote choices.
Unfortunately, our data do not allow us to explore the role that changes over the life-cycle in individuals’ psychological traits may have on the ageing effects observed in our analysis. Previous research, however, has found some evidence suggestive of age-related psychological and personality changes. For instance, ageing has been linked to increasing levels of conservatism, authoritarianism, prejudice, and self-discipline, as well as lower levels of openness to change and cognitive flexibility (for reviews, see Tilley and Evans, 2014; Peterson et al., 2020). Evaluating whether – and, if so, to what extent – such psychological changes can account for the observed ageing effect would require direct measures of psychological traits at multiple points across the life-cycle. While these are not included in our dataset, we consider the further empirical verification of such psychological mechanisms an important avenue for future research.
Finally, we show that life-cycle effects in individuals’ vote choice are somewhat stronger among individuals with lower income levels. Hence, while the average aging effect observed within our overall sample is unaffected when controlling for income, different income groups within our sample display distinct (subgroup-average) age effects. This suggests that the increasing tendency among well-earning professionals to support left-wing parties in recent decades (e.g., Häusermann et al., 2012; Attewell, 2021; Gethin et al., 2022) need not exclusively derive from generational replacement or cohort effects – as recently argued by, among others, Gethin et al. (2022).