Silly Rabbit - Data's Not for Analyzing!

David Weigel | May 17, 2007

Weird Janet Elder column on whether people lie about voting for black candidates: the evidence tells her that they don't and she writes the column as if they do. After some background on Virginia's 1989 election (the one where a black Democrat underperformed the exit polls):

In a more recent election, the Tennessee Senate race between Harold Ford, a black Democrat and Bob Corker, a white Republican, the questions about race and polling played out differently.

Pete Brodnitz who did polling for the Ford campaign, said Mr. Ford was initially hurt by what had become conventional wisdom that polls overstate the strength of black candidates. It caused people to discount Mr. Ford, he said.

“In the beginning everybody said he didn’t have a chance, he can’t raise money, he’s never run statewide, even if I see data that he is competitive,” he said.

The point is that the people who discounted Ford were wrong. On election day he was lagging badly in the polls, and at the end of the night he lost by only 3 points. Forty-one percent of whites said they'd vote for him and forty percent did vote for him.

Earlier post about this stuff here. It's not a huge issue apart from the way it pushes a false, pity-party narrative about Barack Obama.

Comments

|5.17.07 @ 9:07AM|#

Is this an effort to guilt White people into voting for Black candidates they're otherwise ambivalent about?

Because, as a Whyte dood, there's nothing I'd rather listen to than a grievous sense of entitlement set to a dulcet whine. No, no, really, please, tell me how you lost your campaign because my stubbornly hateful fellow palefaces can't get over their narcissism and bigotry to imagine that a Black man could possibly do a good job at anything not involving flying balls, feats of speed and strength, or song and dance.

Vote for Barack or you're a Klansman!

Douchebags in every direction, I swear.

|5.17.07 @ 9:32AM|#

Steven Dubner at the Freakonmonics blog posts comments by ABC News polling director, Gary Langer. Langer notes, "The reality is that pre-election polling relies on accurately modeling who is or isn't going to vote. It's plenty likely to be bad modeling - not lying respondents - that causes the estimate to be blown."

The Link
http://www.freakonomics.com/blog/2007/05/09/guest-blog-whos-to-blame-for-inaccurate-election-polls/

Tiffany Case|5.17.07 @ 9:32AM|#

"Because, as a Whyte dood"

Willard Whyte?

Your casino system is impressive. How on earth did you ever get those diamonds back?

|5.17.07 @ 10:05AM|#

"...the evidence tells her that they don't and she writes the column as if they do."

Welcome to the mainstream media. You must be new here.

Ellie|5.17.07 @ 10:06AM|#

Janet Elder took advantage of my white guilt! Which is only to be used for good, like over-tipping and supporting Barack Obama.

|5.17.07 @ 10:14AM|#

I'm constantly bemused by pundits who claim that when an election doesn't follow the pre-election opinion polls, that there must be something wrong with the election. If your statistical model fails to predict what happens in the real world, there is something wrong with it, or with the data you've plugged into it. The sample size could be too small, the screen may not a good match for those who actually turn out, one or more of the survey questions could be flawed, etc. The point that Brodnitz makes, that a certain demographic of late deciders tended to break for the conservative, Republican candidate seems plausible.

I would have no problem voting for a candidate of either sex or any ethnic background, provided that he or she substantially agreed with me. I don't want a representative who looks like me. I'd rather have one who thinks like me.

Kevin

Scooby|5.17.07 @ 11:17AM|#

But if the election doesn't match the polls, it is obvious evidence of ballot box tampering, intimidation of minority voters, and/or other election fraud.

Or, it could be faulty modeling. Or even a little of all of the above (just because the polls are faulty doesn't mean that there isn't electoral hanky-panky going on).

I think that, in the case of the Jesse Jackson polling data (hypothesis: white people are more likely to react positively to Jesse Jackson's candidacy with a black poll-taker than with a white poll-taker) can be explained by a sort of response bias different from the one Mr. Edelman alleges. Maybe black poll-takers took laughter and "yeah, right. I'd vote for Jesse" as a positive response, while white poll-takers saw through the sarcasm.

|5.17.07 @ 11:19AM|#

Using the Corker - Ford Race in Tennessee was a bad example anyway. I and all of my closest friends thought he was white up until his supporters bitched about some particularly mean anti-Ford campaign ad came out, which was about his promiscuity and being invited to the Playboy mansion. The ad didn't say anything about him being black, but his supporters said the attack ad was covertly implying that all black people are promiscious because they get invited to the Playboy mansion. Or something. Regardless, I did a double-take because the man is seriously whiter than I am, and I'm Irish.

bubba|5.17.07 @ 11:44AM|#

Wasn't Bobby Jindahl another example of a non-white badly underperforming his poll numbers?

|5.17.07 @ 11:54AM|#

David-Adam,

Do you not get it, lads? The Irish are the blacks of Europe. And Dubliners are the blacks of Ireland. And the Northside Dubliners are the blacks of Dublin. So say it once, say it loud: I'm black and I'm proud.

|5.17.07 @ 12:11PM|#

Filthy Irish!

|5.17.07 @ 12:11PM|#

:-).

|5.17.07 @ 2:07PM|#

Yeah! Sometimes polls are right! Ron Paul will win the GOP nomination! Woooo!

...sorry, I'm on a delusion-fueled endorphin high for the next dozen months. When the janitors enter to sweep up the confetti at the 2008 Republican Convention, I'll be collapsing into severe clinical depression...