Partisan pressure and algorithmic promises: Exploring political discourse, self-censorship, and perceptions in online spaces

Abstract

This dissertation explores how partisan identity and social media platform design shape political expression and content perception online. Through three experiments, I investigate specific mechanisms that affect political discourse in digital spaces. In my first study, I investigate an understudied yet significant aspect of political polarization; namely, how partisan identity affects self-censorship \textit{within} political groups. Using a behavioral measure that compares participants' private opinions to their public statements in simulated chats, I find that Democrats were significantly more likely to withhold their opinions when speaking with fellow Democrats who disagreed with them than with Republicans who disagreed with them, while Republicans showed no significant difference in disclosure patterns regardless of their conversation partners' affiliation. In my second study, I explore how participants' perceived distances from their own party and the opposing party moderate their responses to being politically outnumbered in social media newsfeeds. I find that those who see themselves as ideological outliers within their party are more resilient to being surrounded by out-partisans, and actually find balanced environments more comfortable than in- and out-partisan echo chambers, while perceived distance from the opposing party leads to harsher evaluations of social media users. Finally, in my third study, I explore how giving users control over the type of content they consume affects their experience on social media. I find that offering users the choice to filter out toxic political content increases their satisfaction with our hypothetical platform. Paradoxically, however, those who opted to use this filtering option rated all posts they saw as significantly more hostile than control participants, even though everyone, regardless of experimental condition or choice, saw the same content. Combined, these findings reveal how both intra-partisan social pressures and platform design features can create invisible constraints on political expression online. This work challenges conventional assumptions about user autonomy in digital spaces and provides empirical evidence of the complex socio-technical factors that now mediate our digital public spheres.