Income and political attitudes are powerfully correlated in cross‐sectional data, yet research based on panel data finds at most a weak correlation. In this paper, we examine this puzzling pattern by exploring the long‐term evolution of attitudes over the life cycle. We evaluate the predictions of five different explanations on the relationship between attitudes and income experiences. These explanations focus on, respectively: socialization, anticipation, myopic self‐interest, learning and status maximization. We employ accelerated longitudinal design models using data on core political values that span up to sixteen years from the British Household Panel Survey. Our findings reconcile the mixed evidence in the literature: the correlation between income and political attitudes, strong in cross‐sectional studies but weak in short panel studies, emerges because attitudes crystallize – slowly but systematically – as income evolves over the life cycle. This pattern is most consistent with the learning explanation.