At MAPOR on Friday, we had another terrific session dealing with the belief gap and extremes in partisanship — a direct follow-up to the belief gap session we had in 2011. We were fortunate enough to get the entire set of participants back together, and had a big audience and a good discussion at the end.
Closing out the panel, Doug Hindman presented a nice overview of how knowledge gaps and belief gaps differ, which is really a narrow set of fairly subtle characteristics. He began his presentation by noting that he’s come more and more to see belief gaps as a particular kind of knowledge gap. Specifically, these are instances where the “knowledge” deals with politically disputed information, and where partisan elites (Doug mentioned “elected officials,” but I think we need to be able to include people like, for example, Rush Limbaugh here) make claims both supporting or denying that knowledge. Consistent with Zaller’s work on elite cuing, we should (and do) see partisans — particularly stronger and more educated partisans — taking up the incorrect beliefs put forth by their partisan elites.
This is also consistent with my findings, presented just one slot earlier, that partisans can respond to hostility toward their partisan identity by putting forth exaggerated incorrect beliefs. In particular, my study found that strong Republicans were especially susceptible to priming effects on their partisan beliefs about Barack Obama (e.g., birtherism) when primed to think negatively about Obama, and to policy-related beliefs when primed by a negative assessment of Mitt Romney.
All of this is to say that one way we might model belief gaps is by looking at how group norm adherence moderates knowledge intake and retention, traditionally seen as operating through education. Because political parties are highly sorted by ideology (which has not always been the case), partisan affiliations are highly relevant reference points for group norms regarding political issues. But crucially, they may also be primed by a stimulus during measurement (i.e., something in the survey design) or before measurement through some informational stimulus. Perhaps more importantly, these norms need not come from political group affiliation, though that’s all we’ve studied in the belief gap context so far.
Here’s an update to the issue typology I speculated about last year:
|Consensus||Obama born in U.S.||Global temperatures will rise||Climate change|
|Disputed||Roe v. Wade lowered crime rate||Economy will improve next year||Gun ownership and safety|
|Unknowable||2000 election stolen||Jesus will return someday||God guides all events|
|Consensus||Lincoln killed by Booth||World will end in December||Vaccines can cause autism|
|Disputed||JFK killed by Oswald||Green Bay will win the Super Bowl||Attachment parenting is harmful|
|Unknowable||Atlantis sank||Ragnarok will happen someday||Alien life exists|
For the politicized issues, it’s clear to see where group norms might come from and what they might be. Apart from partisanship and ideology, religion and religiosity play key roles in many of these examples. But what could account for possible belief gaps in the non-politicized (or not yet politicized) issues? It is just as likely that non-political group norms drive these beliefs. First, religion also likely plays a role in in issues we think of as non-politicized. As I noted in an earlier post on fringe belief entrepreneurship, belief that vaccines cause autism is strongly related to use of Christian media. Such a belief could also be related to affiliation with anti-corporate groups, or expressing an anti-corporate identity; alternative media sources could also provide such information. Fringe groups and media (e.g., Fortean Times) also would likely predict belief in the existence of alien life, and certainly the role of group affiliation and identity is self-evident in beliefs about sports.
My takeaway from this is that we should be examining the belief gap as the outcome of a moderating mechanism operated by a balance of group identity salience across one’s entire identity. For example, someone who considers herself both a Democrat and a Catholic will be subject to elite cues related to both parts of her identity, but how she balances those parts at any given time will help to explain how she expresses beliefs related to issues for which each identity is relevant. That balance is malleable, as demonstrated by my priming findings, and as it varies, we should expect expressed beliefs and belief gaps to vary. Identity salience is just one factor to emerge through these presentations; in a subsequent post, I’ll look at a related phenomenon that I’ll call information salience and its relationship to belief reversion.