Romney's support among African-American voters: Zero percent

According to the latest NBC/WSJ poll:

This brings me perverse joy.

Well, I assume the poster known as OMG a black conservative will vote Romney, probably.

And Clarence Thomas. So that’s two.

Anybody think Condoleeza Rice votes for Romney?

I assume they rounded up.

I wouldn’t count on Alan Keyes or Herman Cain voting for Romney.

Nah, I believe she’ll vote for Obama. She did last time as well, didn’t she? I thought I read that somewhere.

You’re kidding, right? Alan Keyes is the Uncle Ruckus of American politics, not to mention a Birther. He wouldn’t vote for Obama, or any black man, if you paid him. And Cain seems to feel he has something to prove to white folks, in addition to being embarrassingly clueless, so yeah, I see him enthusiastically voting for Romney.

Well, yes. Humorous hyperbole. (Besides, any mention of Herman Cain is ipso facto funny. Which can’t be said of Keyes, more’s the pity.)

Hence the oh-so-urgent need to get the voter [del]suppression[/del] ID laws in place before the November election.

He could have as much as 3.1 percent right?

Or as little as -3.1 percent.

But that would probably be better news for Romney. At least he could claim the voter ID laws are needed, people are literally stealing his votes right?

Erhm so what about the remainder of six percent?

Probably no preference or undecided.

It’s not a very good poll and can’t really be used for anything besides making fun of Romney.

A 3.1 margin of error on a poll of a thousand people shows not even the pollsters were taking it seriously.

I don’t think that’s how margin of error works. I mean, they talked to those thousand people and got them to answer questions. There’s no margin of error on what they think; the pollsters know it exactly. The margin of error is how accurately those thousand represent the larger group they’re a part of.

Or am I misreading you? But a ~3% margin of error seems to be as accurate as most of the political polling I’ve been seeing, so there’s no reason to say they’re not taking it seriously.

Serious statistics question: since the “real” proportion can’t be “negative 3.1 per cent,” do stats sort of “bunch up” when you get towards an end-point? In other words, does the +/- 3.1 per cent you might expect in the middle turn in to + 6 per cent/ - 0 per cent at an end-point? Or does it go the other way, to, say, + 2 per cent/ - 0 per cent?

How do you mean? That’s the usual margin of error for N=1000


Of course, the margin of error for any subset of the poll would be larger. Without looking into the actual numbers, I’m going with about 14% of the respondents are black, as, IIRC, that’s their portion of the total US population. That would mean a margin of error of about 8% for that group.

Here’s a fact that might enlighten an alien trying to understand American politics, given the huge tendency of Blacks to vote Democratic:

Ignoring D.C. there are six states with populations more than 25% Black: Mississippi, Louisiana, South Carolina, Georgia, Maryland, Alabama. Except for Maryland, these are all strong Republican strongholds. In Alabama and Louisiana, Obama got less than 40% of the 2008 vote despite the large Black populations. These are lower percentages than Gore got in 2000, despite the higher Black turnout of 2008.

In case the dots are hard to connect, here’s a variation on the thread title:

Obama’s support among bigots: Zero percent

Condi endorsed Romney. I think it’s assumed that she voted for Obama because she refused to say who she was voting for in 2008.