<p>Supplementary appendix.</p>

Despite social media characteristics that include homophilous social networks, algorithms that can reinforce biases, and social bots that promote falsehoods, there is no evidence of a strong association between social media use and misperceptions. This project offers the strongest empirical test of this relation to date, providing additional evidence of the minimal effects of social media use on misperceptions.

Data come from two independent large-<i>N</i> panel studies, each paired with online tracking data, one focused on the 2020 election and the other on the COVID-19 pandemic. Within-between regression models show that social media use was not associated with a statistically significant decline in Americans’ ability to discriminate between true and false claims on any of the four platforms examined—Facebook, Twitter, YouTube, and Instagram.</p></div>