What Happens Around a Message Matters Too
It seems reasonable that hearing a different opinion might help us reconsider our own. That would make a neat explanation. Yet a research paper about political disagreement shows why even a sensible expectation needs more than one explanation.
In a 2018 field experiment, Bail and colleagues asked frequent American X/Twitter users to follow accounts sharing political messages from the other side. Participants were offered financial incentives, and researchers compared subsequent attitudes with those of a control group.
Even a sensible expectation needs more than one explanation.
The result is not the whole story
Republican participants assigned to follow the opposing-view bot expressed more conservative views afterwards. The estimated change among Democrats was smaller and not statistically significant. The researchers did not find the straightforward movement towards agreement that a simple exposure argument might suggest.
That finding matters, but it does not tell us what every participant was thinking. Did someone feel challenged, dismiss the source, become more certain or simply react to the unusual experience? Those are all possible interpretations.
There is a difference between comparing groups and knowing what happened inside each person. The experiment provides the first kind of evidence, which can be extremely useful. It does not provide a complete account of how every individual interpreted the messages.
Psychology gives us questions about attention, identity and judgement. Sociology asks where commitments and relationships fit. Neither set of questions, on its own, establishes an answer for this particular experiment.
The setting makes a difference
There is also the platform. Following a bot for a month is not the same experience as discussing an issue with a trusted person. X/Twitter's public messages, the accounts selected and the study's incentives all formed part of the conditions being examined.
Van Dijck and Poell (2013) describe how platform principles such as connectivity and popularity organise communication. That perspective reminds us to ask how an encounter is arranged before assuming that the message alone produced the response.
But institutional explanations have limits too. Knowing how a platform operates does not tell us exactly why a particular person shifts position. A persuasive account still needs evidence about the people, their circumstances and the outcome measured.
Keeping questions open
What I value about reading the study this way is the discipline of resisting the first explanation that sounds complete. We can take the result seriously while admitting that another setting might produce another pattern.
We would also need to consider how participants usually encountered disagreement. A month of experimentally arranged exposure may tell us something important about online communication, but it should not stand in for every conversation that happens outside that setting.
Check out my recent EMPOWERVERSE explainer, considering media change across several levels.
Before we say that a message changed someone's mind, we may want to know what was measured, which conditions made the encounter possible and which explanations the evidence cannot yet separate.