Exploring the Predictors and Outcomes of Perceived Algorithm Responsiveness in Social Media Feeds: Combining Longitudinal and Computational Methods

Perceptions of how algorithms curate content are integral to how people use social media and its effects. Building on the algorithm responsiveness process, we explore whether perceived algorithm responsiveness (PAR) was associated with the message cues of algorithmically curated content. We coupled longitudinal data from a 7-day, twice-daily diary survey with computational digital traces from Twitter (now X) to operationalize algorithmic message characteristics and to understand the antecedents and outcomes of PAR ( N  = 10,337 tweets). Results suggested that message cues from algorithmic content in social feeds (i.e., closeness, popularity, valence, and format) were associated with perceived message personalization, but message personalization was the only message characteristic associated with PAR. PAR predicted feeling more connected to others on Twitter, along with message-level factors: closeness to the poster and message personalization. This work contributes to the algorithm responsiveness process, demonstrating that characteristics of algorithmically curated messages predict how individuals experience algorithms.