On Chuck Woolery and Americans’ Stereotypes of Muslims

by John Sides on April 21, 2013 · 22 comments

in Public opinion

My Wonkblog post today begins with a tweet from Chuck Woolery and segues into a forthcoming article by Kimberly Gross and me about American stereotypes of Muslims and their consequences for policy attitudes.  Here is a graph showing how white non-Muslim respondents stereotyped Muslims, Muslim-Americans, and several ethnic groups on four different trait dimensions:


In the post, I write:

…on average these respondents rated both Muslims and Muslim-Americans as more violent than peaceful and as more untrustworthy than trustworthy. Put in percentage terms, 45 percent of respondents placed Muslim-Americans on the “violent” side of the scale, and 51 percent placed Muslims on this side of the scale. Given that Dzhokhar Tsarnaev was an American citizen, it is notable that respondents do not appear to distinguish between Muslims and Muslim-Americans.  Both groups are stereotyped in much the same way.

At the same time, Muslims and Muslim-Americans were perceived as more hardworking than lazy and as more intelligent than unintelligent.  Gross and I argue that this pattern fits the prevailing images of Muslims that Americans are exposed to in the news and entertainment media.  Muslims are portrayed, intentionally or not, as devious and violent more often than they are portrayed as lazy or dumb.

More here.


RobC April 21, 2013 at 4:56 pm

Your Wonkblog essay states, “These surveys suggest that many Americans do not distinguish between the vast majority of peaceful Muslims and the very small number of Muslims who commit violent acts, as Tamerlan and Dzhokhar Tsarnaev are alleged to have done.”

Is that what the surveys suggest? Were respondents asked about the vast majority of various groups or about the groups in general? In other words, were they essentially asked to stereotype the groups?

Let’s consider a hypothetical in which several campus rapes had occurred and all had been committed by members of a particular fraternity–Sigma Epsilon Xi. A survey is conducted of members of the university community, asking them their opinion of how peaceful/violent various groups are, including ΣEX, the faculty, sorority sisters and the university chaplains. Remember, only a minority of the members of ΣEX are rapists. What’s the “correct,” non-stereotyped answer? That all those groups have the same 4.0 score for peaceful/violent? If respondents on average score ΣEX members as more violent than university chaplains, have they failed to distinguish between a violent minority and the majority of law-abiding ΣEX members? Or are they answering the question in the only logical way–apart from the way suggested by the movie “War Games”: the only way to win is not to play.

Maybe that’s the answer. The only logical way to respond to a survey of attitudes about groups is not to respond.

John Sides April 21, 2013 at 5:33 pm

RobC: The battery of stereotype items began with this preamble, borrowed from the American National Election Studies:

“Now we have some questions about different groups in our society. We’re going to show you a seven-point scale on which the characteristics of people in a group can be rated. In the first statement a score of 1 means that you think almost all of the people in that group are ‘hard-working.’ A score of 7 means that you think almost all of the people in the group are ‘lazy.’ A score of 4 means that you think the group is not toward one end or the other, and of course you may choose any number in between that comes closest to where you think people in the group stand.”

People were given the name of the various groups — in this order: whites first, then blacks, Hispanic-Americans, and Asian-Americans in random order, and last Muslims or Muslim-Americans (randomized)– and asked to place each group on those scales. Each group-scale combination (20 in all) appeared on a different screen in the on-line instrument.

Given the wording of the question, a person who places Muslims at 7 on the peaceful-violent scale is saying that “almost all” Muslims are violent. A person who said 5 or 6 is communicating, in essence, that most Muslims are violent or that Muslims are, on average, more violent than peaceful.

To give an answer that does not conform (at least somewhat) to the stereotype “Muslims are violent,” respondents had to indicate a 1, 2, or 3 on the scale — something that more than half did not do.

Sebastian April 21, 2013 at 10:40 pm

do we know enough about what people are actually answering in a question like that?
I’m not a public opinion guy, but this seems like a very demanding type of question, especially considering that NES is pretty long&tiring anyway. I’m perfectly willing to believe that many, even a majority of Americans hold negative stereotypes about Muslims and doesn’t distinguish care much whether they’re American or not. (cf. any popular and non-moderated comment thread on the internets…)

But look at the peaceful/violent value for Whites – according to the question, the value of a bit of 3 means that whites believe, on average that almost half of all whites are violent. That just doesn’t seem plausible to me. I think it suggests that a large share of the respondents reads the question much more along the lines of what RobC suggests.
I’m willing to be convinced that I’m wrong about this, but I’d want to see some research on what the instrument actually measures.

John Sides April 21, 2013 at 10:49 pm

Our questions were included in the CCES, not the ANES, so the instrument wasn’t nearly as long. You’ll also see in the paper that use a different ANES measure — a “feeling thermometer” for Muslims — in a parallel analysis and get similar results: less favorable views of Muslims compared to other groups, which are then associated with attitudes toward the War on Terror.

Sebastian April 21, 2013 at 11:20 pm

Thanks. I’ll read the paper later, but I suspect I won’t object much with your general claim of stereotypes about Muslims along the lines of what’s portrayed in media/popular culture. And I think the finding about lazy is quite cool.
I was mainly taking issue with: ” A person who said 5 or 6 is communicating, in essence, that most Muslims are violent or that Muslims are, on average, more violent than peaceful. ” in your response to Rob. I realize these are comments on a blogpost, so I don’t expect you to measure every word, it just sounds overly optimistic about the degree to which people answer the question as it is asked. In other words – I believe that you’re measuring people’s real stereotyped views of Muslims as more violent than other groups. I’m just arguing that people who answer with >=4 don’t necessarily believe that more Muslims than not are violent (because I don’t believe the >30% (guesstimating) who answered >=4 for Whites believe that).

John Sides April 22, 2013 at 9:31 am

About 12% of white respondents placed “Whites” on the violent side of the spectrum (that is, at 5, 6, or 7), compared to the 51% who placed Muslims there. Ultimately, in the paper, we are using the stereotype measures in the relative sense that you describe — i.e., how much they see Muslims are more violent than their own in-group. It’s a fair question how people “use” those stereotype batteries — there is clumping at the midpoint (4) as in many survey questions that use ordinal scales — but I don’t think it’s true, as RobC posited, that it is impossible to respond in ways that do not indicate negative stereotypes.

John Jay April 22, 2013 at 12:02 am

Getting at the same thing as some of the other commenters: do you have reliability and validity data for the battery of stereotype items?

John Sides April 22, 2013 at 9:32 am

John Jay: It depends on what you consider reliability and validity data. Why don’t you look at the paper and then tell me if there is something else you’d like to see?

John Jay April 22, 2013 at 10:05 am

Will do. Thanks.

John Jay April 22, 2013 at 7:38 pm

I read the methods. When I say reliability and validity, I mean the terms as psychological researchers do. Does the battery produce consistent results (reliability)? Does it measure what it purports to measure (validity)? Here’s a summary of the concepts as psychologists think of them (forgive me if you’re familiar with them already):


I don’t see data in the paper pertaining to the reliability and validity of the stereotype items.

John Sides April 22, 2013 at 7:59 pm

John Jay: I’m familiar with the concepts. The question is what evidence you would like to see. There are multiple types of reliability and validity, as the link you include makes clear. What specific tests, correlations, models, etc. are you thinking about?

John Sides April 22, 2013 at 8:19 pm

John Jay: I should say that I’m not being evasive or willfully obtuse here. I honestly don’t know what you would like to see. For example, the paper shows that stereotypes of Muslims correlate strongly with a broader measure of “ethnocentrism.” Is that evidence of validity, in your mind? Moreover, we posit that the stereotype items can be divided into two dimensions, and then use measurement models to show that a two-dimensional model fits the data better than a one-dimensional model. Is that evidence of validity?

John Jay April 22, 2013 at 8:42 pm

Thanks for the reply. I didn’t think you were being purposefully evasive or obtuse. I think that’s some evidence of validity, yes.

MiLO April 21, 2013 at 6:01 pm

An implication from the four graphs is that white opinion would place Black Muslims (with or without a National of Islam affiliation) in THE most unfavorable category. Just an observation.

I wonder: Has this approach ever been used with stereotypical and non-stereotypical images of the groups? It’d be interesting to see how results may shift in response to images rather than names of groups.

Also, how good is it from a civic standpoint to treat “Muslims” and “Muslim-Americans” as being without race or ethnicity? I ask given that Muslims are, as Malcolm X observed on his pilgrimage to Mecca, “all colors, from blue-eyed blonds to black-skinned Africans.” However, there is, I assume, a certain Muslims and Muslim-Americans in the minds of white respondents. You could even use images of Muslims from different racial and ethnic categories, again, to get at differences.

I look forward to reading the paper.

JG April 21, 2013 at 7:50 pm

While Muslims and Muslim-Americans are not a racial or ethnic category they are a ‘behavioral out-group.” Despite the conceptual difference I suspect that if a question were asked about Arabs and Arab-Americans the responses would not vary significantly.

MiLO April 21, 2013 at 6:09 pm

“We also found that black and Latino respondents tended to view Muslims in the same way as whites.” Would you mind linking to graphs with those findings? We almost never get a sense of what non-whites think on such issues or towards such groups.

John Sides April 21, 2013 at 10:46 pm

We don’t have a graph on this, but the paper has a footnote. Apologies.

Andrew Gelman April 21, 2013 at 7:46 pm

All I can say is, if Chuck Woolery’s the best the bigots can come up with, they’re in big trouble. Talk about a Grade-Z celebrity.

Kevin April 22, 2013 at 10:53 am

Wait just a dang minute!!!

Am I to understand that Chuck “I’d Like to Buy a Vowel [in 1978]” Woolery is quoting political science research?

Is this was we meant by “public outreach”? Or should I say “p_bl_c __tr__ch”?

Kevin April 22, 2013 at 10:55 am

D’oh! Sorry. Giving exams today…. reading comprehension of the students is rubbing off on me…

Anyway, kudos for quoting Chuck Woolery in a poli sci blog post!

Steve Sailer April 22, 2013 at 8:09 pm

These seem like pretty accurate stereotypes.

Sally Edelstein September 23, 2013 at 10:52 am

In a time of heightened sensitivity over stereotypes, years of ethnic and racial labeling have largely been erased from advertising.
During the post-war years when it seemed the only risk of offending others was if we suffered the unforgivable shame of halitosis, a series of ads ran that would not only raise a politically correct eyebrow today, its offensive nature could very well spark angry, violent protestations. Take a look http://wp.me/p2qifI-1GA

Comments on this entry are closed.

Previous post:

Next post: