« The tactical fallacy | Main | Twitter roundup »

September 06, 2010


Lots of food for thought in Brendan's tweets. Why do so many people say Obama is Muslim? Brendan appears to place much of the blame on "the role played by the media and political elites in misleading the public about Obama's religious beliefs." If that is true, however, then we are obliged also to ask which media and political elites misled the public about several myths mentioned by Douthat:

There’s the 32 percent of Democrats who blame “the Jews” for the financial crisis. There’s the 25 percent of African-Americans who believe the AIDS virus was created in a government lab. There’s support for state secession, which may have been higher among liberals in the Bush era than among Republicans in the age of Obama. And there’s the theory that the Bush White House knew about 9/11 in advance, which a third of Democrats endorsed as recently as 2007.
Also noteworthy is the study by Kosloff et al. that found this, according to the LifeScience.com report:
On average, McCain supporters said there is a 56 percent likelihood Obama is a Muslim. But among those who were primed to think about race, the likelihood jumped to 77 percent. Since the participants were primarily white, Kosloff said this shows that simply thinking about a social category that differentiated participants from Obama was enough to get them to believe the smear.
Did 21% of the respondents who were primed to consider race thereby become more receptive to their previous indoctrination by political and media elites, or is something else going on here?

Douthat writes of "understand[ing] those paranoias to be symbolic beliefs, rather than real convictions"--and that would seem to explain things like the Muslim myth and the 9/11 myth and the AIDS myth better than attributing the receptivity to those myths to nefarious political and media elites. In addition, the Kosloff et al. study is good evidence that questionnaire design may promote the expression of these contrarian symbolic beliefs.

Some surveys can be tested against reality. E.g., a voter preference poll taken shortly before an election can be compared to actual election results.

OTOH the 2007 poll showing that 33% of Dems believe that Bush knew about 9/11 in advance cannot be so tested. The best we can do is to look at people's general behavior and see if it's consistent with the poll results. Douthit quotes Sanchez: “you did not really see a lot of behavior consistent with millions upon millions of people being seriously convinced that their president was a treasonous mass murderer.” Exactly.

I am convinced that this poll result is not a real indicator of people beliefs. I reject the idea that 1/3 of Dems are paranoid nut-cases. IMHO this poll result doesn't match any true reality. I feel the same way about the polls regarding Obama's religion. There's nothing in the real world to test that poll against. I don't believe the poll results match any reality.

Once we accept the idea that some poll results don't match any reality, analysis of those results becomes less interesting. If a poll has unknown accuracy, who cares what it purports to show?

Based solely on the titles and cover pictures, both Liberal Fascism and American Taliban belong in the genre of 'my political opponents are monsters'. However, the content of Liberal Fascism did not argue that liberals were fascists. It showed that historically early Italian fascism was more supported by liberals than conservatives. If it made an argument, it was that it's unfair for liberals to call conservatives "fascists."

Based on the linked review, American Taliban lacks that sort of serious content. For that reason, I don't think it will sell nearly as well as Liberal Fascism did. (That's a prediction that can be tested against reality.)

Rob, the fact that I said that elites played a role in spreading the Obama Muslim myth doesn't mean I believe all myths are spread by elites -- there's obviously a lot of variance in the manner by which misperceptions spread.

Given that "there's obviously a lot of variance in the manner by which misperceptions spread," what's the basis for attributing the misperception about Obama's religion to media and political elites? With due respect, it seems at this point to be pure punditry. And that's fine. Brendan is as entitled to play pundit as anyone. But let's not confuse it with his day job.

I just got my tickets to see Markos Moulitsas on October 3rd!

Following up on my prior comment, Brendan's link to an MSU study says,
On average, participants who supported McCain said there is a 56 percent likelihood Obama is a Muslim. But when they were asked to fill out a demographic card asking for their own race, the likelihood jumped to 77 percent. Kosloff said this shows that simply thinking about a social category that differentiated participants from Obama was enough to get them to believe the smear.

Kosloff equated a person's response to a survey with the person's actual belief. IMHO that's clearly wrong. The percentage of people who actually believe Obama is a Muslim cannot be both 56% and 77%.

How could we find out how many people really believe Obama is a Muslim? There's no way to measure that figure. We cannot answer the question. We cannot even define what it would mean to really believe that Obama is a Muslim.

By comparison, suppose two alternative survey wordings gave substantially different percentages of people who said they would vote for Obama. These two results could be compared against the actual vote to see which one was more accurate. In this case, there's no problem with defining what it means to really intend to vote for Obama.

A story, possibly apocryphal, concerns a mathematics graduate student whose dissertation consisted of the derivation of the characterists of a certain class of mathematical structures. Supposedly, during his dissertation defence, one of the Professors was able to prove that no such mathematical structures could exist.

ISTM that there's a parallel with analyzing surveys that cannot be tied to real-world events. It's all well and good to discover that how one asks a question can lead to a difference in response. It's fine to discover that changes in the President's popularity have a certain effect on the responses. However, unless the survey response can to connected to something outside the survey, the main value of such discoveries would seem to be that they can lead to publications in professional journals.

The comments to this entry are closed.