Sources of information, salience and novelty, and belief reversion

In this year’s MAPOR panel, Ken Blake’s presentation looked at belief that Barack Obama was born outside the United States, and Doug Hindman talked a bit about right-wing denial of the September unemployment numbers from the Bureau of Labor Statistics. Both issues are good example cases for a phenomenon that we might call belief reversion.

Let’s take birtherism first. Ken noted that belief that Obama was born elsewhere declined considerably between polls conducted right before and right after the April 2011 release of Obama’s long-form birth certificate. This should not come as a surprise, of course — we expect people to absorb new information from the media and incorporate it into their structures of existing knowledge and belief. Indeed, that existing context plays a role in the nature of the outcome we observe. Before and after the release, Gallup found the biggest changes among Republicans, likely explained by the fact that they had the highest levels of birtherism to start with. Specifically, Republicans went from 43% birther to 23% birther in the span of a couple weeks. Gallup appears to have considered the question settled at that point, but YouGov followed up their similar 2011 results by returning to the question in January and July of 2012. Before the release of the certificate, 30% of Republicans believed Obama was born in the U.S., compared with 47% after the release. By July 2012 that number was back down to 31%, and Republicans were actually more likely to specifically disbelieve it (rather than say they don’t know) than they were before the release.

Of course, birtherism spent several years growing into a sibboleth on the right before the long-form certificate was released. That it might be central to Republican or conservative identity during this time period is not terribly surprising. How what about reactions to the September unemployment numbers? While unemployment had been an issue throughout the Obama presidency, the specific notion that the September report was doctored, and that unemployment had therefore not declined, only arose after Jack Welch’s tweet to that effect. In other words, this particular belief was not something that had spent years percolating. Yet, Brian Schaffner finds this same reversion phenomenon occurring in beliefs about this issue, and doing so in just the span of a few days. Before the report, belief that unemployment had either increased or decreased in the past year followed a predictable ideological pattern (predictable based on the incumbent party in the White House, that is): Liberals were most likely to say it had decreased, while conservatives were most likely to say it had increased, with moderates in the middle on both questions. But the immediate aftermath of the report’s release, each group went up in belief that it had decreased (which is the correct answer as of both August and September) and down in the belief that it had increased. That is, new information pointing out the decrease in unemployment became readily available and highly salient to expressing beliefs about the change in unemployment. Over the next three days, liberals and moderates took that information in and became more and more correct; conservatives, however, quickly reverted to their previous beliefs. By the last day the poll was in the field, they were back to the same level they’d been before the report was released.

Finally, one more example that may shed some light on how approaches to different pieces of information plays a role here. Republican officials continue putting far more time and effort into the Benghazi consulate attack issue than do their Democratic counterparts, and Pew finds that reflected in the attention being paid to the issue by rank and file Republicans. Twice as many Republicans as Democrats say they are following the issue “very closely,” which is likely a product of both individual content choice and channel selection — that is, that they’re more likely to watch Fox, and Fox is more likely to cover Benghazi. But the final result is a dramatic difference in what kinds of information Republicans a) have available, and b) find salient when expressing beliefs about this issue.

So, to summarize:

  • Exposure to new, corrective information can apparently overwhelm whatever it is that leads to an incorrect partisan belief, but that information may only remain salient or accepted for a finite period of time. The response to the BLS numbers suggests it could only be a few days.
  • Jack Welch’s involvement in the BLS case also suggests that either novelty of information (Welch’s tweet obviously came after the release of the report) or elite challenge could make news information either less salient or less acceptable to partisans.
  • The traditional knowledge gap idea that more educated people are better available to get information from the media relies on an assumption of relatively balanced media diets. But if we don’t make that assumption, and instead allow for selective exposure and different agendas across media outlets, we can see a situation in which partisans will have different sources of information upon which to base their beliefs, even before taking interpersonal sources into account.
  • A likely outcome of the sudden salience of correct information, followed by a relative decline in salience of that information, is for beliefs to become more correct, and then revert to their previous incorrect state as the information balance reverts. This is really just a particular case of a more general process of shifting and churn within one’s set of available information — we could also see situations in which incorrect beliefs appear for the first time, or get stronger, then revert or change as saliences shift.

How to test this? Scheduled releases of information, such as BLS reports, provide an excellent opportunity for close examination of the before and after phenomenon. Beyond that, looking at very specific news and information sources, as well as attention paid to them and to different issues, would help provide a sense of individuals’ available information in a way that the typical “How frequently to do watch cable news programming?” style of question can’t do.

Share this post:

    Tags: , , , , , , ,