What we know about political misperceptions

In the wake of the publication of a Pew poll showing an increase in the false belief that Barack Obama is a Muslim, misperceptions have reached a new level of prominence in the national discourse, including a mention of the Muslim myth on Newsweek's cover. In addition, MIT political scientist Adam Berinsky released some new public opinion data on the topic Monday, so it seems like a good time to review what we do -- and don't -- know about misperceptions.

Belief in the Muslim myth has increased

As I noted at the time, the Pew poll found that the proportion of the public identifying Obama as Muslim increased from 11% in March 2009 to 18% in August 2010 and the proportion who didn't know his religion increased from 34% to 43%. This shift was corroborated by a subsequent Newsweek poll using somewhat different wording (PDF), which found that the proportion of the public saying Obama is Muslim had increased from 13% in June 2008 to 24% in late August 2010. (Time similarly found that 24% of Americans think Obama is Muslim, but no previous survey is available for comparison.)

Americans hold several false beliefs about Obama

In addition to the Muslim myth, polls have shown that a substantial fraction of the public believes Obama was not born in this country. Most recently, CNN found that 27% of Americans think Obama was "probably" or "definitely" born in another country (Berinsky similarly found in July that 27% said Obama was not born in this country.) The Newsweek poll mentioned above also found that 31% of Americans said the allegation that Obama "sympathizes with the goals of Islamic fundamentalists who want to impose Islamic law around the world" is "definitely" or "probably" true.

False beliefs about Obama are concentrated among Republicans

Pew found that the proportion of Republicans saying Obama is Muslim increased from 17% in March 2009 to 31% in August 2010 and the proportion who don't know increased from 28% to 39%. Similarly, CNN's poll showed that 41% of Republicans think Obama was "probably" or "definitely" not born here, a figure that corresponds closely to Berinsky's 46-47% (based on his bar chart). Newsweek also found that 52% of Republicans thought that the claim that Obama wanted to impose Islamic law was "definitely" or "probably" true. These figures are consistent with other polls showing differences by party in politically salient misperceptions (e.g., Iraq having WMD before the U.S. invasion). It's important to note that misperceptions are not confined to Republicans. Democrats, for instance, were far more likely than independents or Republicans to endorse the claim that the Bush administration was complicit in the 9/11 terrorist attacks in a 2006 Scripps poll, a 2009 PPP poll, and Berinsky's 2010 poll.

These large partisan differences in misperceptions appear to be the result of people's bias toward factual claims that reinforce their partisan or ideological views (selective acceptance). This pattern of motivated reasoning -- plus possible biases in the information to which people are exposed (selective exposure) -- appears to result in large partisan differentials in misperceptions along partisan or ideological lines (see my research here and here for more).

Misperceptions are not simply a function of ignorance

As Berinsky and many others have found, people who know more about politics (as measured by the questions political scientists typically use to measure political knowledge) tend to be less likely to hold false beliefs. However, that doesn't mean that the problem is simple ignorance. A better approach is to distinguish between ignorance (when you know you don't know the truth) and misinformation (when you falsely believe you know the truth). Politically salient misperceptions typically fall into the latter category, which is why they are so pernicious. For instance, I found that Republicans who believed they were knowledgeable about the Clinton and Obama health care plans were more likely to endorse false claims about them (Berinsky misstates my finding on this point).

In addition, elites often appear to play an important role in spreading false claims ranging from "death panels" to the Muslim myth and Obama supporting Islamic law. For this process to operate, partisans must be exposed to the message from elites, understand it, and integrate it into their belief system, which is not consistent with a simple story of ignorance. GW political scientist John Sides has provided evidence that is consistent with this account, showing that the persistence of the Obama Muslim myth increased more during Obama's presidency among Republicans with higher levels of education:

As TNR's Jon Chait notes, Princeton political scientist Larry Bartels similarly found that more knowledgeable partisans were more likely to develop false beliefs of economic performance that was consistent with their political views:

Voters’ perceptions may be seriously skewed by partisan biases. For example, in a 1988 survey a majority of respondents who described themselves as strong Democrats said that inflation had “gotten worse” over the eight years of the Reagan administration; in fact, it had fallen from 13.5 percent in 1980 to 4.1 percent in 1988. Conversely, a majority of Republicans in a 1996 survey said that the federal budget deficit had increased under Bill Clinton; in fact, the deficit had shrunk from $255 billion to $22 billion. Surprisingly, misperceptions of this sort are often most prevalent among people who should know better—those who are generally well informed about politics, at least as evidenced by their answers to factual questions about political figures, issues, and textbook civics.

The beliefs that people express aren't fixed

While the prevalence of these misperceptions has been repeatedly validated in national polls, it's important to note that the exact responses people provide will vary depending on question wording, context, etc. as in any other survey. For instance, in research with Reifler and Duke undergraduates, we found (PDF) that the presence of non-white interviewers appeared to influence how participants responded to corrective information about Obama's religion. Likewise, a study (PDF) recently published in the Journal of Experimental Social Psychology found that McCain supporters were more likely to accept the claim that Obama is Muslim when their racial identity was made salient.

Does this mean these beliefs aren't "real"? It's hard to know what that claim means. All survey responses are to a certain extent an artifact of the context in which they are solicited -- there is no way to measure what someone "really thinks." However, it's possible that people are expressing an ideological or partisan view as much as they are making a factual claim about the world. The strongest claim along these lines comes from Reason's Julian Sanchez, who suggests that misperceptions like the claim that Obama was not born in the U.S. are best conceptualized as "symbolic beliefs" rather than statements of what people believe to be literally true -- an argument that was subsequently endorsed by New York Times columnist Ross Douthat and ABC News polling consultant Gary Langer. Determining to what extent these beliefs are "symbolic" rather than literal is an important question for future research.

Update 9/16 1:26 PM: See John Sides for more on recent research into partisan bias in factual beliefs about politics.

[Cross-posted to Pollster.com and Huffington Post]

Comments

[Rob]

I think Brendan is onto something when he cites the possibility that symbolic beliefs are distorting the responses, i.e., that people may be answering in a certain way not because they truly believe something that isn't factual but because they want to convey their support or opposition to something or someone. Rigorous research would be great, but insight into the phenomenon doesn't need to wait for rigorous investigation. An academic or pollster could test a poll on a small sample and then interview the respondents to try to determine what motivated them to answer as they did. At a minimum this might help the design of the poll, as well as suggesting avenues for more rigorous research.

Having praised Brendan in the previous paragraph, let me chastise him for characterizing an opinion as to whether Obama sympathizes with the goals of Islamic fundamentalists as a misperception. Obama's sympathies are a matter of subjective judgment, not verifiable fact. To treat one person's opinion on the matter as a correct perception and another's as a misperception is simply to impose Brendan's own perception of the issue as the standard of truth. I may agree with him as to this particular perception, but I don't agree with perverting science in this way. Harry Cohn, the longtime head of Columbia Studios, said he had a foolproof way of knowing whether a picture was good. "If my fanny squirms, it's bad. If my fanny doesn't squirm, it's good." The screenwriter Herman Mankiewicz famously replied, "Imagine--the whole world wired to Harry Cohn's ass." In this case, the ass may be Brendan's but the arbitrariness is the same.

[David in Cal]

Nice post, Brendan.

Another important aspect of why people believe myths is the many elites these believers don't follow. A few elites may spread some of these myths, but many more elites deny them.

CSICOP fights false belefs about ghosts, mind-reading, spoon-bending, dowsing, etc. These beliefs are branded as false by many elites, yet it's the rare myth-believer who changes his/her mind.

Why do so many myth-believers ignore the majority of elites who are telling them the truth? It looks to me as if they don't trust our elites very much.

P.S. I'm particularly struck by the large number of 9/11 Truthers. I don't know of any elite person who's spreading the myth that Bush was somehow involved in the 9/11 attack, but the myth is simply not going away, at least according to survey results.

[David]

Would it be possible, in any way, to do follow-up interviews to find out whether people who say (on polls) that they believe Obama was born in Kenya etc actually do believe those things? To be honest, on the rare occasions when I get polled by phone, I answer the questions in a strategic way, to get my political opinions and choices across. If I were asked whether my economic situation had improved under Obama, I would probably say yes, even though in fact it has remained constant. That is because the question I really want to answer is, Do you support Obama? I would guess that most Republicans really want to answer a question such as, Do you utterly despise Obama and want him out of office? But instead they are asked, Do you believe he was born in Kenya? And, Do you think he is a Muslim? So they answer Yes to those, because those were the questions they were asked, and the only way they were given to express (however crudely) their disapproval.

[Rob]

I agree with David, and I'd go even further. I think sometimes people become annoyed as they perceive a bias in the questions being asked, and that colors their responses and may result in what is essentially contrariness. Also, many people simply don't understand the question they're being asked, especially in oral questionnaires. (Lots of effort spent in framing a question very precisely doesn't necessarily mean the nuances of the question are appreciated by the respondents.) Finally, the length of many questionnaires may produce annoyance in respondents and cause them to give unconsidered or contrary responses to the later questions.

Post-questionnaire interviews of respondents in a small-scale test could bring some of these possibilities to the surface. So could asking respondents to verbalize their thinking as they proceed through the questionnaire.

I hope there will be further consideration of the inadequacy of polls and questionnaires to obtain reliable information about respondents' knowledge and beliefs. The cynic in me says there won't be much, because what quantitative social scientists and pollsters want is quantifiable data they can slice and dice, not doubts about whether the data is meaningful.

[bnyhan]

The problem is that people's self-reports about what they "really" believe have all the same issues as regular survey questions. How do we know that those answers aren't symbolic or strategic? Also, asking people a question like that suggests that they should admit they don't really believe the misperception, which is a cue that could skew the answers. (Note also: Many of these polls ask approval questions and other questions about Obama. These questions are typically not the only way to express disapproval. See the questionnaires for more.)

[MartyB]

Brendan -

I find curious the sustained focus on the "Obama is Muslim" opinion, when in the recent polls, the people who "wern't sure" of Obama's religion was not only larger percentage but increaed by a significant amount more than the people who believed he is Muslim.

It appears to me this statistic might be more likely to be a key to understanding the opinion shift than "elites are spreading it", for which the evidence you have provided is rather sketchy.

My thought is that for whatever reason, people are in general more uncertain about Obama's religious beliefs than they were before, and some people - rather than offer a "I don't know" - choose an answer besides what they what they might be more familiar with (i.e. Muslim rather than Christian).

This is just speculation, of course, but so is most of the the "elites effect" you keep citing.

[Asics shoes]

If the money is in the hands of you, you don't want to this fetish grace, Because the sage once such instruction: hardworking than gold.Do you think so?

[Dopeyman]

"when you know you don't know the truth": Isn't that wisdom?