I’ve stumbled across a paper you might find interesting. It’s a year old but it just came to my attention just now because it’s apparently just found a publisher. The paper was written by Douglas J. Ahler of Florida State University and Gaurav Sood and is titled “The Parties in Our Heads: Misperceptions about Party Composition and Their Consequences”. You can read the paper in full here. Here’s its abstract:
We document a large and consequential bias in how Americans perceive the major political parties: people tend to considerably overestimate the extent to which party supporters belong to party-stereotypical groups. For instance, people think that 32% of Democrats are LGBT (vs. 6% in reality) and 38% of Republicans earn over $250,000 per year (vs. 2% in reality). Experimental data suggest that these misperceptions are genuine and party specific, not artifacts of expressive responding, innumeracy, or ignorance of base rates. These misperceptions are widely shared, though bias in out-party perceptions is larger. Using observational and experimental data, we document the consequences of this perceptual bias. Misperceptions about out-party composition are associated with partisan affect, beliefs about out-party extremity, and allegiance to one’s own party. When provided information about the out-party’s actual composition, partisans come to see its supporters as less extreme and feel less socially distant from them.
The authors make a number of interesting observations. For example, Republicans greatly overestimate how many Democrats are union members (36.5% vs. actual 10.5%) while Democrats even more greatly overestimate how many Republicans earn more than $250,000 (31.5% vs. actual 2.2%). They also find something interesting: the more closely you follow the news the more likely you are to make these mistakes.
Here’s what to my eye is their most important observation:
Believing that opposing partisans hold more extreme policy preferences, and feeling more socially distant from them, are both liable to cause citizens to become less receptive to out-party communications and less likely to consider voting for that party. This may happen because people come to see opposing partisans as working on behalf of the interests of a few groups (at the expense of other groups or even the national interest) (e.g., Bawn et al. 2012), or because they think that the opposing party supports more extreme policies that it does, or potentially even because they distrust elites representing disliked groups.
Let that sink in for a bit. Lack of understanding of the people in the other party promotes the interests of the more extreme members of both parties at the expense of sympathy and moderation. No wonder we’re so screwed up.
Overall a decent paper with conclusions that shouldn’t surprise anyone who has spent much time on internet forums discussing politics. One thing I’d add is that partisans, in my experience, frequently believe themselves to be experts on the out-group but are often completely wrong.
One shortcoming is that the paper only looks at the perceived composition of the opposing party for a particular group. For example, the percentage of Democrats that are LGBT. To me, that only tells half the story – you also need to consider the percentage of LGBT people who are Democrats. This paper doesn’t address that part of it.
I’ve linked to it many times here, but Brink Lindsey’s essay on partisanship from several years ago sums up most of this paper’s conclusions in a shorter and more readable package:
https://www.cato-unbound.org/2009/02/04/brink-lindsey/partisanship-still-half-empty
The key passage IMO:
If you are interested in politics, you probably end up on the internet or listening to talk radio or watch cable TV. You get to hear extreme versions of the opposition at those sites. Plus the aggregator sites make things even worse. They amplify everything so that everyone in the bubble carries the story convincing people the story is real.
Steve
“…..the percentage of Democrats that are LGBT. To me, that only tells half the story – you also need to consider the percentage of LGBT people who are Democrats. This paper doesn’t address that part of it.â€
I would suggest that the high percentage of LGBT who are Democrats derives from their bias based upon a minority of Republicans, the so-called religious right, and the presence of cause seeking zealots in the Democrat Party. And to take things further, I bet you’d find a large fraction of both party identifiers whose libertarian leanings cause them to shrug their shoulders at LGBT, except when the LGBT community starts telling them how their sexual leaning is going to impact them.
“There’s no epistemologically sound reason why one’s opinion about, say, the effects of gun control should predict one’s opinion about whether humans have contributed to climate change or how well Mexican immigrants are assimilating — these things have absolutely nothing to do with each otherâ€
I’d disagree with that. A general philosophy of individual liberty and not central authority, and respect for the responsibilities that come with citizenship lead one away from gun control, taxing and dictating energy consumption choices and responsibility free immigration practices, all much more closely associated with one of the two major parties.
Read this earlier when I wasn’t able to read the pdf and now I’m too exhausted, but I’m flummoxed by how a group could believe 32% of Democrats are LGBT and its not “expressive responding, innumeracy, or ignorance of base rates.” That number seems so extreme that it could be the result of an accumulation of all three. For instance, people do seem to greatly overestimate the number of minorities, perhaps because that which is different stands out more.
OK, I found this:
“Surprisingly, providing base rates appears to make participants less accurate”
(!!!) They note that the test interface may be part of the issue, but without quite understanding the interface, it does seem like that self-admitted issue just moves some potion of the problem to innumeracy and perhaps the whole exercise is, as I would intuit, a game of whack-a-mole.
I think that some of the contributing factors are that African Americans, Hispanics, and Asians are not distributed uniformly throughout the country and the places that punch above their weight in terms of opinion-making, e.g. Los Angeles, New York, Washington, have much higher percentages of minorities than most of the country does. Just because LA is half Hispanic leads Angelenos to believe that the whole country is, too. African Americans, Hispanics, and Asians are all wildly over-represented on primetime TV programs. If it’s on TV it must be true.
I spent most of last week with a client in Racine, Wisconsin. You’d get a very different impression of the demographic makeup of the U. S. from Racine than from Chicago but guess what? Racine is more like the U. S. as a whole than Chicago and in both places African Americans and Hispanics are more numerous than in the overall U. S. population.