The Limbaugh Effect

I just posted this in the comments section of someone else's blog, so I figured I ought to post it to my blog as well. For people who doubt the possibility of the Limbaugh Effect being significant in Indiana, here's the math that I did from CNN's election and exit poll data. (Note: I still hate CNN, but I do like their election tracker stuff.)

Total Indiana democratic primary votes (99% reporting): 1,272,471

Given the choice of Clinton or McCain, 16% of democratic primary voters chose McCain, of which 41% voted FOR HILLARY in the primary -- So that's 1,272,471 * .16 * .41 = 83,474 McCain/Republican supporters who voted for Hillary in the primary, even though they won't vote for her in the general.

Given the choice of Obama or McCain, 18% of democratic primary voters chose McCain, of which 12% voted FOR OBAMA in the primary -- So that's 1,272,471 * .18 *.12 = 27,485 McCain/Republican supporters who voted for Obama in the primary, even though they won't vote for him in the general.

Net gain for Hillary: 83,474 - 27,485 = 55,988 votes

Hillary's margin of victory: 14,413 votes

So without the votes of people who will vote McCain over their own democratic choice in the fall, Obama wins by about 40,000 votes. Obviously, the numbers above are estimates because they're based on the exit polls, but when the estimated net gain for Hillary is almost four times the margin of victory, that's a legitimate concern.

--

Incidently, this also allows us to put a lower bound on complete assholes in Indiana at 83,474 + 27,485 = 110,959 complete assholes, or 1.76% of the Indiana population. Go Hoosiers!

--

Update:

I've seen a few places reporting that Obama's camp is saying that 7% of the Indiana democratic primary turnout is attributable to Limbaugh's Army. That is also supported by the same exit poll numbers, though admittedly it rounds up to the nearest percent, and ignores the 2% of the turnout who did the opposite of Limbaugh's suggestion and voted for Obama even though they won't vote for him as the nominee. The 7% number from Obama's camp is calculated as follows:

Percentage who would vote for McCain over Clinton: 16%
Percentage of that 16% that voted for Clinton: 41%
Percentage of total electorate who voted for Clinton but would vote for McCain over her in the general: 0.16 * 0.41 = 0.0656 = 6.56% (rounds up to 7%).

--

Update:

Most of the media articles I saw on this topic today poo-pooed the idea that the Limbaugh thing had much of an effect. Here are the arguments I saw, and why I think they are less compelling than the argument I posted.

Argument #1: Hillary won self-declared Democrats 52-48 by exit poll data. Some sites just stop there, saying "See, she won democrats straight up, so forget Limbaugh's people, democrats got what they wanted." That's a very solid argument…as long as you think independents don't matter. Since I’m registered as Non-Partisan, that idea naturally offends the hell out of me. Obama won independents 54-46.

Argument #2: Hillary also won self-declared Republicans 54-46, and some sites argue that the data shows those Republicans picked her as genuinely the better candidate; better on actual issues like the war and economy, or on qualities such as leadership. There's a major problem with that analysis: The questions about the issues and leadership in the exit polls asked voters to pick from Obama and Clinton only. McCain is nowhere in those questions, so while some of those people certainly were being genuine in their support, that analysis completely fails to detect those who still think McCain is better than either one of them. Just because a group of people likes Candidate A over Candidate B does not mean they like Candidate A over Candidate C. Limbaugh voters are a blind spot in this analysis, so can it really say much about the effect of Rush Limbaugh? Nope.

Argument #3: Some articles I read have argued that it is legitimate for the Republicans to vote in the Democratic primary even if they intend to vote for McCain in the fall, because they are just trying to give themselves the best two options. I can see how that might occur to people, but it's still unethical. If you intend to vote for McCain, but then vote in the Democratic primary as a Second Choice, you're manipulating the nomination process. You're injecting your second choice vote into the count of Democrats' first choice votes, without any real intention of backing the Democratic candidate. If you intend to vote for John McCain in the election, congrats, he's on the ballot. The Republicans had a primary in Indiana too, as meaningless as it was, and if your first choice is McCain, then your place was there. This argument is morally crippled.

This whole thing has interested me more in a Data-versus-Media Perception way than in a political way. I have a friend who pointed out that it is not particularly tactful for Obama to make a big deal out of this, and I would agree with that – his performance last night all but seals the nomination for him, and there is no need for him to irritate people by saying “Oh yeah, by the way, I should have won Indiana too.” Even by my analysis, he only would have won by a percent or two, hardly anything to brag about.

More irritating to me, however, is that for all the dismissals in the media of a significant Limbaugh effect in Indiana, I have yet to find a good counterargument that provides its own compelling data and/or a reason to dismiss the data that I (and the Obama camp) used. If you see one, send it my way. I think the real reason it’s being dismissed is the same reason I was skeptical at first myself – it’s ugly, and we don’t want it to be true, because of what it says about some voters and about the vulnerabilities of our election system. Maybe the exit polls were way off, absurdly off, maybe they were incompetently run, maybe the sample size was too small (1881 respondents, 0.15% of total voters), I don’t know – but based on the data we have, I think Limbaugh voters just might have flipped the Indiana primary election.