Housewives never retire!? Gender biases in popular sample definitions for studies on the elderly

While research emphasized the risk of gendered sample selection bias among the elderly decades ago, the empirical literature on old-age inequalities remains largely unaware of it. This research note addresses this issue by investigating gendered sample selectivity for individuals aged 65 or older employing two common sample criteria: self-reported retirement status and pension receipt in countries covered by the Survey of Health, Ageing, and Retirement in Europe (SHARE). Findings show that more than half of older women are excluded when these criteria are applied. Gender selection bias varies widely across countries and is less pronounced in post-socialist or social-democratic welfare states. Visualizing work trajectories by sample status reveals that women with long unpaid care work periods and men with high self-employment, unemployment, and extended education levels are particularly likely to be excluded. Studies employing such sample criteria risk underestimating gender inequalities in pensions, health, and life satisfaction. The implications are severe for Southern, conservative, and liberal welfare states, and for cross-country comparisons, where sample bias often goes undetected due to its variability across contexts. While this article cannot offer a universal recommendation for sample definitions, it aims to promote less biased sample conceptualizations in studies of the elderly population.