Friday, January 22, 2010

In the Same Boat

A new Gallup poll concludes that the strongest predictor of anti-Muslim sentiment is ... anti-Jewish sentiment. Presumably, this indicates that both anti-Jewish and anti-Muslim sentiments still are primarily associated with the American right to far-right. The overlap doesn't really surprise me, though I'd be curious to see whether it is as extensive in Europe.

One thing I do wonder about, though, is reporting bias. Presumably, liberals who harbor negative feelings about Jews or Muslims are less likely to admit them, even to themselves, as such prejudice is dissonant with the prevailing self-image of what it means to be a good leftist. I doubt that Tom Hickey, for example, would be keen on confessing anti-Jewish prejudice, even to a pollster.

But taking the poll at face value (and to be clear, I feel reasonable in asserting that the poll captures a goodly portion of the social positions of Jews and Muslims in America), it does reinforce something Jews have known ever since we set foot on these shores: the only way to insure our security, is to insure the security of all vulnerable minorities here. Anytime America looks poised to create another division between "real" and "fake" Americans, Jews are at risk of falling into the latter camp. We're all in the same boat here.

9 comments:

N. Friedman said...

Your statement that "A new Gallup poll concludes that the strongest predictor of anti-Muslim sentiment is ... anti-Jewish sentiment" is a misinterpretation. Read the poll itself. What the poll actually correlates on the point is stated in the conclusion of the poll:

Self-reported prejudice toward Jews is most strongly associated with “a great deal” of prejudice toward Muslims. Th ough some see Muslims and Jews as representing competing loyalties in the Middle East conflict, prejudice toward Jews is associated not with solidarity, but disdain for Muslims.

In fact, your conclusion that this is a right wing thing is not shown either. If we assume - and this may or may not be the case but it seems a reasonable supposition - that church attendance is a sign of conservatism, then you are clearly wrong. The conclusion of the poll states: "Frequent religious service attendance is associated with reports of “no prejudice.” While some assume that commitment to one’s faith makes one less accepting of followers of another, our findings suggest the opposite. Those who report attending a religious service more than once a week are more than twice as likely to report no prejudice and more than half as likely to report extreme prejudice versus those who attend less often."

The other two findings are interesting as well. The one that explains why your interpretation is wrong is that Americans are more than twice as likely to report having negative feelings about Muslims as compared to holding negative feelings about Buddhists, Christians, or Jews. Which is to say, there are a lot more people who claim to be prejudiced about Muslims than claim to be prejudiced towards Jews.

PG said...

How is "predictor" (which in the statistical sense does not mean cause but rather correlation) a misinterpretation?

In "Variables Associated With Prejudice Toward Muslims (Respondents self-reporting 'a great deal' of prejudice toward Muslims)," "Prejudice toward Jews (a great deal)" is by far the highest. "The variable most strongly linked to self-reported prejudice toward Muslims is self-reported prejudice toward
Jews. Respondents who say they feel 'a great deal' of prejudice — or extreme prejudice — toward Jews are about 32 times as likely to report feeling 'a great deal' of prejudice toward Muslims. While Jewish-Muslim relations sometimes suffer because of the turbulence of the Israeli-Palestinian conflict, among other reasons,
these findings point to an area of potential cooperation for the two communities in addressing a common
concern of prejudice toward each group. Previous Gallup research indicates that, compared with other
religious groups in the U.S.,
Muslim Americans and Jewish
Americans are most similar
in terms of political ideology,
education, and political party
identification."

And if we're going to try to figure out from the poll whether political conservatives are more likely to report prejudice against Muslims, you might want to look at "Race (Nonwhite)," which is a stronger predictor of No Prejudice At All than greater-than-once-a-week religious attendance. (The poll said religious attendance; "church attendance" is your own reading.)

Nonwhites are much less likely to be politically rightwing than whites are, if one judges by voting patterns (and photos from tea parties, CPAC, Republican conventions, etc.). Nonwhites are 2.3 times more likely than whites to feel no prejudice toward Muslims.

N. Friedman said...

PG,

Because it is not a good predictor. That is why.

And Church and religious attendance, in the context of most of the US, is pretty much the same thing.

N. Friedman said...

My apology, PG. I left out the reason.

It is not a good "predictor" because not enough people claim to have prejudices against Jews to make prejudice towards Jews a predictor of prejudice towards Muslims.

Cycle Cyril said...

In the report of the actual poll only one percent of those polled had a "a great deal" of prejudice toward Jews. This is a mere 10 people out of the 1,002 polled. The margin of error for the poll is +/- 3%.

Statistically this is meaningless. You can get one percent of almost any poll to agree to a bizarre belief. Hell, more people probably believe they were abducted by aliens. That this small group also has "a great deal" of prejudice against Muslims is thus unremarkable.

I tend to think that the sponsors of this poll are attempting to tell a compliant audience, such as Washington Post reporters, that they should not be critical of Islam, and reveal its nature, because it is the same as anti-Semitism.

PG said...

It is not a good "predictor" because not enough people claim to have prejudices against Jews to make prejudice towards Jews a predictor of prejudice towards Muslims.

What would you consider a sufficient n of "people who claim prejudice against Jews" here? Please provide your basis in statistical methodology for requiring that n. My understanding was that you were criticizing David's interpretation of the poll results, not the poll's own methodology.

PG said...

"This is a mere 10 people out of the 1,002 polled."

I hope no one reading these comments is assuming Cycle Cyril is being accurate about the poll. His assertion that only 10 people reported having a great deal of prejudice against Jews is belied by p. 25 of the PDF of the report.

Cycle Cyril said...

I shall concede that instead of one percent it is 1.5 percent. But if this makes any more significant for you then you need to take an intro stat class. (Of note and to be obsessive it is slightly less than 1.5 which is why, I presume, they rounded it down to 1 and not to 2%)

If 100 percent of those with "a great deal" of prejudice towards Jews also had "a great deal" of prejudice towards Muslims it only accounts for 15 of the 87 respondents.

The correlation they are trying to foster has no significance. What might be interesting (at least to me though it very well would have no true significance) is the reverse.

PG said...

CC,

I didn't criticize your statement of the percentage; I criticized your assertion that the percentage represented 10 people. This indicates that you didn't actually read the report very carefully, which in turn makes your other assertions about what the poll does and doesn't say rather suspect.

If the total basis for your claim, that the percentage of people with a great deal of prejudice toward Jews is statistically insignificant and thus incapable of being a predictor, is that any percentage lower than the maximal margin of error is not statistically significant, then your stats class must not have been too fantastic. Maximal margin of error just tells us how well the sample maps onto the population it's intended to represent; it doesn't tell us whether a given statistic within the poll is significant.