Affective and social political polarization—a dislike of political opponents and a desire to avoid their company—are increasingly salient and pervasive features of politics in many Western democracies, particularly the United States. One contributor to these related phenomena may be increasing exposure to online political disagreements in which ordinary citizens criticize, and sometimes explicitly demean, opponents. This article presents two experimental studies that assessed whether U.S. partisans’ attitudes became more prejudiced in favor of the in-party after exposure to online partisan criticism. In the first study, we draw on an online convenience sample to establish that partisan criticism that derogates political opponents increases affective polarization. In the second, we replicate these findings with a quasi-representative sample and extend the pattern of findings to social polarization. We conclude that online partisan criticism likely has contributed to rising affective and social polarization in recent years between Democrats and Republicans in the United States, and perhaps between partisan and ideological group members in other developed democracies as well. We close by discussing the troubling implications of these findings in light of continuing attempts by autocratic regimes and other actors to influence democratic elections via false identities on social media.

Image recognition systems offer the promise to learn from images at scale without requiring expert knowledge. However, past research suggests that machine learning systems often produce biased output. In this article, we evaluate potential gender biases of commercial image recognition platforms using photographs of U.S. members of Congress and a large number of Twitter images posted by these politicians. Our crowdsourced validation shows that commercial image recognition systems can produce labels that are correct and biased at the same time as they selectively report a subset of many possible true labels. We find that images of women received three times more annotations related to physical appearance. Moreover, women in images are recognized at substantially lower rates in comparison with men. We discuss how encoded biases such as these affect the visibility of women, reinforce harmful gender stereotypes, and limit the validity of the insights that can be gathered from such data.