Abstract
Aging is typically associated with characteristic changes to cognitive function, personality, and political orientation. While online data collection for psychological research has greatly increased in frequency, there has been little systematic examination of whether online samples are appropriate for studying aging. Here, we examine whether typical age-related differences in cognitive function, personality, and political orientation are replicated in online samples. We measured cognitive performance using tests of vocabulary, processing speed, memory, and attention developed by the TestMyBrain Project; personality using the Big Five Inventory; and political orientation following the ANES survey format. At least 200 participants each were sampled from three crowdsourcing websites (Amazon MTurk, CloudResearch MTurk Toolkit, and Prolific) and three panel recruitment websites (Lucid, CloudResearch PrimePanels, and Qualtrics Panels), with age ranging from 18 to 92 years across platforms. On all six platforms, consistent with established findings, age was positively correlated with vocabulary performance and negatively correlated with processing speed. Additionally, in all six samples, consistent with prior studies, age was associated with higher Agreeableness, lower Neuroticism, and greater political conservatism. There were some differences between crowdsourced and panel samples, however. Performance on cognitive measures was broadly better for crowdsourced samples. The correlations between age and Openness and Extraversion differed between crowdsourced and panel samples, with trends in panel samples likely more comparable to the ground truth. Finally, MTurk produced some discrepant effects of age on cognition relative to other crowdsourced platforms. Beyond these differences, though, our results are broadly encouraging for the prospect of studying aging via online experiments.