Do Feed Rankings Promoting or Demoting Political Distrust and Animosity Impact Affective Experience and Polarization?

In this study, we investigate the impact of feed rankings on feelings toward out-partisans and affective experience on social media. Using a browser extension, we run a field experiment on the platform X (Twitter). We focus on measuring the effect of increasing and decreasing social media users' exposure to content that supports attitudes potentially problematic for healthy democratic functioning. The intervention aimed at decreasing the exposure is achieved by reranking participants' feeds in real-time, penalizing the potentially problematic (“problematic”) posts to move them to a lower position. Conversely, increased exposure is achieved by reranking the posts to promote a problematic post from the inventory to a higher position. We are interested in measuring the treatment's effect on affective polarization and the emotions reported by the participants after consuming content on the social media platform. We measure these outcomes at two different temporal intervals: (1) “short-term,” assessed by in-feed surveys right after reading the feed content and (2) “long-term,” assessed by pre- and post-surveys after multiple days of intervention. In this document, we use "groups" to refer to the two experimental setups: the increased exposure group and the reduced exposure group. We use "condition" to refer to treatment and control in each group.