One of the most heavily discussed questions in the context of belief polarization in recent times, is the contribution of social media technologies. The infamous “filter bubble” hypothesis suggests that that social media algorithms are, by design, driving ideological as well as affective polarization in users. Even though empirical evidence on this mechanism is mixed, at best, this assumption has become a fixture of any debate on polarization. Drawing from a secondary analyses of large-scale panel data and a systematic literature review of empirical research on social media’s polarization effects (both work in progress), I will argue that the question whether social media contribute to polarization has to be answered much more nuancedly. Findings of the two studies indicate that we are far from having consistent evidence on polarizing effects of algorithmic content selection. Rather, social media related polarization effects seem to result from users’ own content selection choices and their active participation in social media communications (i.e., posting, sharing, liking, etc.).