request our brochure

It’s not just what is said but who says it that matters

September 19, 2023

In recent study published in Nature Human Behavior, researchers conducted a large-scale longitudinal field experiment to measure how identity cues shape content consumption and feedback online. MIT Sloan School of Management professor Sinan Aral and his colleagues from MIT and the Hebrew University of Jerusalem found that identity cues and identifying information applied to accounts and handles in social media caused changes in how viewers perceived the content they experienced, and whether they upvoted or replied to the content. 

The results showed the dramatic effects identity cues have on our perceptions of and engagement with online content. Identity effects accounted for up to 61 percent of the variation in voting, meaning that over half of the variance in users’ decisions to up-vote or down-vote content was explained by the presence or absence of identity cues. The presence of identity cues also caused viewers to evaluate content faster, implying greater reliance on initial “knee-jerk” reactions and System I thinking rather than longer, deliberative, System II thinking.

“These results imply people’s opinions and engagement behaviors online are not just about—or even largely about—what someone is saying or posting online, but rather they are even more about who they are and what identity markers are associated with them,” Aral said.

The researchers also demonstrated design solutions to curb reputation inequality and promote quality communication online. They showed, via simulation, that social platforms can improve content quality by recording some voting on anonymous content and including such votes on anonymized content as a ranking signal. 

We must start thinking deeply about platform designs that favor communication health,” said Sean Taylor, an independent researcher and co-author of the study. “One design we investigated was to include voting on anonymized posts in the ranking systems that display content to a platform’s users. Our simulations showed that including votes on anonymized content, which are more about what is being said than about who is saying it, can improve the quality of content online.”

Source: MIT Sloan