Can seeing opposing political points of view in your social media feeds push you to soften your own views? That possibility is a much-discussed option of late. There is plenty of research that shows individuals tend to cluster online into like-minded groups. That phenomenon extends to politics, which creates the “echo chambers” that seem to be eroding social discourse of late.
A group of researchers from Duke, Brigham Young University and NYU just published an experiment on this topic in the Proceedings of the National Academy of Sciences, and (no surprise) it is one of the PNAS website’s most read articles just now. Here’s a summary of the structure of their experiment:
- The researchers recruited 1,652 regular (3 times a week or more) Twitter users to take part in the study. Through both online questionnaires and analysis of whom these individuals followed on Twitter, the subjects were categorized as either “Democrats” (901 subjects) or “Republicans” (751).
- Each Twitter user was then offered $11 to follow a bot (an automated account) in their feed. What the subjects did not know was that they were randomly assigned to either a Democrat-leaning bot or a Republican one. In either case, the bot drew its messages from existing Twitter accounts of real people and institutions that politically align “left” or “right”. A Democrat, for example, might see tweets from a Republican member of Congress and a conservative think tank.
- Subjects completed two surveys about their political beliefs, one before they were exposed to the bots and again after a month and a half of seeing the bot tweets. They also completed weekly surveys and validated they were actually on Twitter and active by answering questions about what they saw in their feed.
In a nutshell, the study took a bunch of politically aligned people and randomly inserted messages into their Twitter feeds that either implicitly confirmed their beliefs or challenged them. The surveys the subjects took at the start and end of the study measured if their views had softened, hardened, or remained the same. Common wisdom at the moment would predict that those subjects who saw posts from the “other side” of the political spectrum might move towards the center.
In fact, seeing counterarguments to their political beliefs had either no effect on their thinking or actually pushed them further into their own camp:
- Subjects who identified as Democrats and saw Republican tweets remained just as committed to their points of view. The researchers noted that they actually became a little more “Blue” (the color of the party, to our non-US readers) but the amount was statistically insignificant.
- Republican subjects who saw Democratic tweets in their feed actually moved further to the right on the political spectrum by the end of the study. Unlike the Democrats, however, the shift here was statistically meaningful.
The upshot: while there is plenty of research that shows human-to-human interaction between politically-identified people serves to soften polarization, the same is apparently not true about social media. In the case of Democrats, they are unmoved. Republicans exhibit what the study’s researchers call a “backfire effect”, where viewing Democratic messages actually deepens their attachment to pre-existing beliefs.
The researchers for this study rightly point out that more work has to be done to understand why social media seems so ineffective at bridging the US’ political divide. For example, they do not know what creates the backfire effect among Republicans. And there is no color about what sorts of messages did sway those outliers from either party that actually moved towards the center.
As far as investment implications, this experiment is just one more example of how little we know about the effects of social media on a critical issue: its role in politics. The results neatly disprove the notion that there is an easy online fix to deepening political polarization. That’s important because politicians care deeply about this topic and they also write the laws and regulations that will one day govern this industry more closely.
See the whole study here: http://www.pnas.org/content/early/2018/08/27/1804840115