CAIR exit poll: Trump got 13% of Muslim vote, doubling Romney

[Matty]

You must be logged in to read this quote.

https://www.cair.com/press-center/press-releases/13909-for-the-record-cair-releases-results-of-presidential-election-exit-poll.html

[Indy Texas]

Well obviously they'd be less enthusiastic for Hillary than they were to vote for the first Muslim president of the United States in 2008 and 2012.

[BaldEagle1991]

Self hating Muslims?

[PepeTalk]

Wow, I guess some Muslims agree with Trump. Like Jews voting for Democrats

[Beet]

Nice spin on getting 13% of the vote.

[#TheShadowyAbyss]

Still proud to be in the 74% who voted for Hillary Clinton. The 13% are probably the ultra-conservative Muslims who looked past Trump denigrating Muslims and hoped he'd repeal abortion rights and gay marriage or some other social conservative wedge issue.

[Indy Texas]

Still proud to be in the 74% who voted for Hillary Clinton. The 13% are probably the ultra-conservative Muslims who looked past Trump denigrating Muslims and hoped he'd repeal abortion rights and gay marriage or some other social conservative wedge issue.

Do conservative Muslims really care that much about abortion?

[Alcon]

Nice spin on getting 13% of the vote.

I don't think it's exactly insignificant that he doubled Romney's percentage showing.  It would be spin if he'd only listed "doubling Romney" in the thread title, but he indicated 13% too, so I don't see the problem.

[Beet]

Nice spin on getting 13% of the vote.

I don't think it's exactly insignificant that he doubled Romney's percentage showing.  It would be spin if he'd only listed "doubling Romney" in the thread title, but he indicated 13% too, so I don't see the problem.

But most people are reacting to the doubling part, not the 13%. What's the sampling error on a survey like this, anyway? A 6-7% shift is barely outside of noise. If you had asked me before seeing this study, I would have said probably 20%.

I just think it's bizarre how people expect any group to be so monolithic. Like we had complaints after the election that Hillary only got 90% of the black vote instead of 96% or something like that. Then suggesting that blacks cost Hillary the election. To me it's bizarre. People are human beings and when you have two individuals running in a competitive election, they will break different ways. To treat people as nothing more than their religion or skin color and expect them to be 95% in the tank for one candidate strikes me as strange. But other people seem to accept it.

[Alcon]

But most people are reacting to the doubling part, not the 13%. What's the sampling error on a survey like this, anyway? A 6-7% shift is barely outside of noise. If you had asked me before seeing this study, I would have said probably 20%.

Assuming it's a random, representative sample, a n=2000 survey has a margin of error of +/-2.2%, so that's well beyond the margin of error.  I have a low opinion of exit polls when it comes to the "random, representative" part, although this appears to be a phone survey so I'd have to know the methodology better to comment.

I just think it's bizarre how people expect any group to be so monolithic. Like we had complaints after the election that Hillary only got 90% of the black vote instead of 96% or something like that. Then suggesting that blacks cost Hillary the election. To me it's bizarre. People are human beings and when you have two individuals running in a competitive election, they will break different ways. To treat people as nothing more than their religion or skin color and expect them to be 95% in the tank for one candidate strikes me as strange. But other people seem to accept it.

I mostly agree, but two things.

1. The assumption that groups like blacks, Muslims, and especially Hispanics are homogeneous cuts across all political spectra.  How many white liberals would be totally shocked to learn that Trump received a big chunk of the Hispanic vote, and seems to have lost them by less than Romney?  A lot, I'd wager.

2. There are a lot of groups that contributed to Clinton's loss in a mathematical sense.  Non-college educated whites are a big chunk of it, but going from 98% to 90% among blacks, plus a lower black turnout, definitely didn't help Clinton, and quite probably lost her Michigan.  That's not nothing.

[Cruzcrew]

It seems odd that Romney did as bad with Muslims as he did with black voters.

[Mike88]

This election really confirms the old saying: "The more you hit me, the more I like you."

[Bigby]

Mitt's inauthentic image must have hurt with with minorities even moreso than it did with whites.

[Beet]

But most people are reacting to the doubling part, not the 13%. What's the sampling error on a survey like this, anyway? A 6-7% shift is barely outside of noise. If you had asked me before seeing this study, I would have said probably 20%.

Assuming it's a random, representative sample, a n=2000 survey has a margin of error of +/-2.2%, so that's well beyond the margin of error.  I have a low opinion of exit polls when it comes to the "random, representative" part, although this appears to be a phone survey so I'd have to know the methodology better to comment.

That means if they measured 6.5% support in 2012 with the same sample size, it could be anywhere from 4.3% to 8.7% support.

And 13% support in 2016 could be anywhere from 10.8% to 15.2% support.

In other words, it could be a shift from 8.7% to 10.8% support and be within the margin of error in both surveys. With larger numbers, a "doubling" of support would mean a lot more -- say, going from 20% to 40% would mean 1 in 5 votes flipped. With very small numbers, "doubling" means a lot less. At the extreme, going from 1% to 2% support is "doubling". 13% is a very small number, so it means less than the phrase "doubling" normally means.

I don't want to harp on it too much because now it looks like I'm getting defensive, but the above is part of why it seems like spin, also.

I mostly agree, but two things.

1. The assumption that groups like blacks, Muslims, and especially Hispanics are homogeneous cuts across all political spectra.  How many white liberals would be totally shocked to learn that Trump received a big chunk of the Hispanic vote, and seems to have lost them by less than Romney?  A lot, I'd wager.

Sure, but they're the same white liberals who can't imagine how Trump won white women. The social liberal assumption that people MUST vote their identity is absurd. It's the entire worldview that's the problem.

You must be logged in to read this quote.

Of course, when you lose by that little any factor probably cost her Michigan. But that doesn't excuse the absurd implication in these analyses that (a) it's these groups' fault for not supporting hard enough, or that (b) repeatedly getting 95% of the black vote with high turnout is a normal situation that should be expected, especially when the first black president is not on the ballot. ( Also considering the fact that Detroit and Flint are hardly thriving cities that have done wonderfully under the status quo Clinton was partially running on, but that's another story )

[Alcon]

That means if they measured 6.5% support in 2012 with the same sample size, it could be anywhere from 4.3% to 8.7% support.

And 13% support in 2016 could be anywhere from 10.8% to 15.2% support.

In other words, it could be a shift from 8.7% to 10.8% support and be within the margin of error in both surveys. With larger numbers, a "doubling" of support would mean a lot more -- say, going from 20% to 40% would mean 1 in 5 votes flipped. With very small numbers, "doubling" means a lot less. At the extreme, going from 1% to 2% support is "doubling". 13% is a very small number, so it means less than the phrase "doubling" normally means.

I don't want to harp on it too much because now it looks like I'm getting defensive, but the above is part of why it seems like spin, also.

Nothing you're saying is wrong, and I agree that the margin of error is high enough that saying it "doubled" is a bit misleading -- because there's a good chunk of probability that the difference less than doubled (although also a good chunk that it more than doubled).  But this is no more "spin" than saying someone "narrowly leads" in a survey is, which we do all the time when it's within MoE.  This is pretty standard language when it comes to presenting surveys.  I agree we should change the language we use to discuss surveys be more numerate, but the title isn't really irresponsible or misleading.

Of course, when you lose by that little any factor probably cost her Michigan. But that doesn't excuse the absurd implication in these analyses that (a) it's these groups' fault for not supporting hard enough, or that (b) repeatedly getting 95% of the black vote with high turnout is a normal situation that should be expected, especially when the first black president is not on the ballot. ( Also considering the fact that Detroit and Flint are hardly thriving cities that have done wonderfully under the status quo Clinton was partially running on, but that's another story )

I definitely don't endorse all interpretations or applications of that piece of data.  I'm saying that Clinton's losses among black voters (in % and turnout) aren't a pointless piece of trivia that's worth ignoring.  They have electoral implications at the margins, because most elections are won and lost at the margins.

For what it's worth, I completely agree with point B.  I think a lot of people over-read this shift when I think it's really a return to normal post-Obama than anything.  Setting Obama's showing as Clinton's baseline with black voters is silly.

[Associate Justice PiT]

     Exit polls also indicate that Trump outperformed both McCain and Romney with blacks and outperformed Romney with hispanics. His numbers with minorities were still weak, but stronger than anyone really expected.

[Beet]

Nothing you're saying is wrong, and I agree that the margin of error is high enough that saying it "doubled" is a bit misleading -- because there's a good chunk of probability that the difference less than doubled (although also a good chunk that it more than doubled).  But this is no more "spin" than saying someone "narrowly leads" in a survey is, which we do all the time when it's within MoE.  This is pretty standard language when it comes to presenting surveys.  I agree we should change the language we use to discuss surveys be more numerate, but the title isn't really irresponsible or misleading.

Your rephrase of my point doesn't really capture the full import. Going from say 20% to 40% with a MOE of 2% could also be 22% to 38% -- less than doubled. Yet this "less than doubled" would still be a gargantuan jump, and 'doubled' would be accurate. At the other extreme, going from 1% to 2% with a MOE of 2% could be 1% to 1%... far from doubled. Or the real number could be 1% to 4%-- quadrupled! Yet this quadrupling would represent only a shift of 3% of the population... quantitatively miniscule. In other words, as the absolute number reaches extreme limits, geometric expressions become both less accurate and less meaningful. In this case, even a doubling represents a shift of 6 in every 100 voters... In absolute terms quite small.

As far as whether it is misleading or spin, how are we to gauge this objectively? Put it in the context it's presented and look at its functional use. Before I posted, the conversation was dominated by this small minority of Muslims. Arguably they were being given outsized importance. I say this as someone who actually spoke to a Trump supporting Muslim in real life. His reason for supporting Trump? He said Clinton lied about her emails. And he thought a woman's menstrual cycle would make them too emotional. What'd days know-- Muslims are people with the ability to think in more than one dimension! Who would da thought? What's more remarkable is that 74% of this supposed sharia loving fifth column voted for a pro abortion rights woman to be CIC.

[Adam Griffin]

The CAIR link above says Trump "nearly doubled" support. Where are they getting their numbers for 2012? Because their own 2012 press release said that Romney got 4 percent (and McCain 2 percent) in their exit poll. According to that, he more than tripled it. Unless they revised their own exit poll after the fact...

Also, is it possible that maybe a community heavily maligned in this election and threatened with deportation/registries actually lied about their support (even to a sympathetic org), in the fear that the info might be obtained and used against them?

Also:

You must be logged in to read this quote.

Even when considering socioeconomic factors, that seems really high.

[#TheShadowyAbyss]

At my mosque, not a single member voted for Trump, at least to my knowledge, everyone was afraid that either their families would be kicked out (if they are immigrants) or that we'd all be heavily monitored and subjected to intensive scrutiny and xenophobic attacks.

[Alcon]

Your rephrase of my point doesn't really capture the full import. Going from say 20% to 40% with a MOE of 2% could also be 22% to 38% -- less than doubled. Yet this "less than doubled" would still be a gargantuan jump, and 'doubled' would be accurate. At the other extreme, going from 1% to 2% with a MOE of 2% could be 1% to 1%... far from doubled. Or the real number could be 1% to 4%-- quadrupled! Yet this quadrupling would represent only a shift of 3% of the population... quantitatively miniscule. In other words, as the absolute number reaches extreme limits, geometric expressions become both less accurate and less meaningful. In this case, even a doubling represents a shift of 6 in every 100 voters... In absolute terms quite small.

No, that's totally a good point too.  I didn't mean to gloss over it.  To make a related point, it's much more striking if a candidate goes from 10% with a group down to 6%, than if they go up to 14%, because the former means they lost a huge proportion of their supporters, while the latter means they only gained a small proportion of opponents.

As far as whether it is misleading or spin, how are we to gauge this objectively? Put it in the context it's presented and look at its functional use. Before I posted, the conversation was dominated by this small minority of Muslims. Arguably they were being given outsized importance. I say this as someone who actually spoke to a Trump supporting Muslim in real life. His reason for supporting Trump? He said Clinton lied about her emails. And he thought a woman's menstrual cycle would make them too emotional. What'd days know-- Muslims are people with the ability to think in more than one dimension! Who would da thought? What's more remarkable is that 74% of this supposed sharia loving fifth column voted for a pro abortion rights woman to be CIC.

It's a subjective call, but I think including the 13% figure in the title does the trick.  I realize that a lot of people will see "doubling Romney" and shut off their brains, not considering that it's relatively insignificant portion of both the Muslims electorate and previously-Democratic Muslims who switched.  But the title presents enough information for a reasonable person to make a contextually sound interpretation of the findings.  I do think it would be spin if they only included the "doubling" part in the title.

I suppose, to me, the difference is between knowingly including only information that will lead people astray, and including a reasonably full summary that will only lead people astray if they're being innumerate.

Anyway, this is probably getting a little subjective/in-the-weeds, sorry

[jaichind]

Just like Trump did better than Romney among Hispanics.  I think a bunch of Hispanics are for the Wall.  They just want to make sure they are on the right side of the wall first

[GLPman]

We live in such an interesting time.

[Sir Mohamed]

They saw him cheering, I guess.