Thursday, November 29, 2012

Romney Campaign Still Math-Challenged

A bunch of liberals (here's one; here's another) had good fun yesterday with an op-ed by Romney strategist Stuart Stevens yesterday. The main point of the ribbing was that Romney's campaign apparently still doesn't quite understand that all the votes count, not just those from the groups they really liked.

But unless I missed it, no one noticed that Stevens still doesn't seem to have any idea of just how badly Romney lost. He says "Nor are we idiots because we came a little more than 320,000 votes short of winning the Electoral College in 2012."

320,000 votes? I don't think so. By my count, that's off by over 200,000 votes.

The final totals still aren't in, but looking around...

Romney won 206 electoral votes.

He lost Florida narrowly, by only 74,000 votes. Had he won Florida, he reaches 235.

Next closest was Ohio. He lost Ohio by 164,000 votes. FL and OH get him to 253.

Virginia was next. He lost Virginia by 149,000 votes. FL, OH, and VA put him at 266.

Oops! That's already 387,000 votes, and he's not there yet.

To get over the top, he would have had to win at least one more state. Pennsylvania and Colorado have been going back and forth as the votes are counted for the state that did it for Barack Obama...right now Obama's lead in Colorado is slightly less, and it's smaller, so we'll give him that. But Mitt Romney lost Colorado by 138,000 votes.

That means Romney needed not the 320K he said, but 525,000 votes, and again they're still counting in some of these. Regardless: Stevens was off by an impressive 64%. Which will presumably only increase as the last few votes are counted. That's a pretty big miss!

(Fine, you want to get technical? I'm sure that it's possible to find some electoral votes for fewer votes, but a whole lot less practical. For example, those last votes could have been found not in CO or PA, but in New Hampshire, which Obama won by a very slightly larger margin but of course with a lot fewer votes, only 40,000. But that's about it. Obama won Nevada by only some 68K votes, but that's only 6 EVs. Iowa, also 6 EVs, had about a 90K vote margin, so that doesn't really do anything for him. Obama's lead was under 100K in New Mexico and Delaware, but I don't seen any combination which lowers the overall total, and it's not as if that small number in Delaware was actually easier than the bigger number in Virginia).

Should we care? Oh, probably not, although I wouldn't put a whole lot of stock in any of the other empirical claims he makes in the op-ed (I haven't checked any of them). I don't really know where he gets the numbers; my guess is that it might be election night results, but who knows? I mean, I can't really blame Team Romney for not obsessively clicking on David Wasserman's wonderful spreadsheet every few hours, but then again if they want to write about the numbers, they might want to get them right.

Or, to make the obvious point, maybe that's exactly the kind of careful attention to reality and detailed quantitative study that was typical of how the whole campaign was run. But that's just silly talk. Right?


  1. Hey, I noticed that, then plum (plumb?) forgot it -- thanks for the reminder. My thought was to bring in Silver - to the effect that Romney would have had to win the national popular vote by perhaps 2% to overcome Obama's swing state advantage. That is, to move the 100-200k in key swing states would have required moving mountains.

  2. Ah, but let us also not commit the sin of "losing campaigns were poorly run and winning campaigns did everything right."

    Granted, I haven't seen much yet that would inspire confidence in Stevens, and tripling down on the 47% theme after the blowback to Romney's conference call is indicative of somebody who's pretty tone-deaf, which is a TERRIBLE quality in a politcal hack. But, I'm not willig to go out on that limb of blaming Stevens for the fundamentals.

    1. Sure. Odds are that even if it's totally true that Romney's campaign got the polling wrong that it didn't make much of a difference, if any.

      I've been meaning to write about this, however: the lead is up to 3.5% nationally, and >5% in the tipping point state(s). It's awful close to the point where we do have something real to explain beyond the fundamentals.

      Although today's GDP report may temper that a bit.

    2. Point taken.

      But, once we unwrap that goose (no, I have no idea what that saying is supposed to mean, and I just made it up, but I kinda like it), we have problems of candidate and of campaign. I would accept that campaign effects should be visible as baseline shifts in the polls during the campaign, or as a GOTV thing on election day differences. But where does that leave candidate effects? What would constitute evidence that an effect was due to the candidate's identity and not a campaign effect? I'm having trouble figuring that one out.

    3. I guess what I'm saying is that I'm a little uncomfortable with the following equation:

      Final vote = predicted vote based on fundamentals + candidate effect + campaign effect.

      I think there's error in there, so I'm not comfortable with the formulation of:
      Candidate effect = Final vote - predicted - campaign effect.

  3. The whole discussion is silly -- the quintessential math that Republicans do to make themselves feel better.

    Obama won the popular vote by 4.5 million. So what Stevens is saying is that if things had broken EXACTLY right for us, we could MIGHT have won the race with the most enormous split between the popular and electoral college results that has ever been seen in any election in American history.

    Rove's election night argument about Ohio was much more plausible.

  4. "And we would have gotten away with it too, if it wasn't for you still-enfranchised lower-middle class and poor voters!"

  5. Is it possible that what he meant was that *a switch of* 320,000 votes from Obama to Romney would have elected Romney?

    Either way, of course, the problem remains: the absolute number of votes you have to change may not seem that large but it requires winning at least one state which Romney lost by over 5 percent of the vote. (Of course a Romney-optimist's way of putting it would be "if we could just have persuaded 2.6 percent of voters in Colorado to change their minds, and a smaller percent in Virginia, Ohio, and Florida.")

    Anyway, using the number of voters whose minds you have to change as a basis is problematic for another reason: there just were not thaat many changeable voters. The campaign was more about getting your own voters to the polls than about persuading potential voters for the opposiiton to change their minds.

    1. My guess is that Stevens is looking at data from Politico and/or CNN that isn't being updated. They show a margin of about a million votes closer than the current count.

      In fact CNN's site now shows:

      Florida 73,000
      Ohio 103,000
      Virginia 116,000
      Colorado 113,000

      That's 405,000 in all.. but yeah, they'd have actually had to flip about 3.2 million Obama voters to gain 5% in the popular vote, and win those states by overcoming Obama's electoral college advantage.

      And yes, the myth of 'independent' voters. "Skewed" polls were a direct result of Republicans not wanting to identify with the crazies/tea party, and reclassifying as independents. So Romney won a GOP-leaning group of independents, but Obama won moderates, which may be a better measure of 'swing' voters.

    2. >but Obama won moderates, which may be a better measure of 'swing' voters.

      Not really. Democrats always win self-described "moderate" voters by a greater margin than that of the general populace. They won this bloc by 56% in 2012, 60% in 2008, 54% in 2004, 52% in 2000, and 57% in 1996.

      This is function, I suspect, of the negative associations of the word "liberal." Many voters with standard Democratic views prefer the word "moderate" to "liberal." Republicans, in contrast, are in general perfectly comfortable with the word "conservative." So "moderate" doesn't really give us any better a sense of who are the swing voters than "independent" does.

  6. It's worth pointing out that the Romney people thought whites would be around the same portion of the electorate they were in 2004. It was 87% in 1992, 83% in 1996, 81% in 2000, 77% in 2004, 74% in 2008, and 72% this year - a fairly even trend around -3% per year. Yet the Romney people bought into the false media narrative that 2008 was an outlier because Obama juiced minority turnout somehow and we would see a return to some imaginary trend.

  7. Stevens is playing an old rhetorical trick where you simply mention a number and then try to imply something about its relative smallness or largeness without providing enough context to justify that conclusion. One example of this gambit is the anti-Obama talking point from 2008 that Obama voted "present" over 100 times as a state senator. Of course, this failed to mention that his total votes numbered over 4,000, and so he voted "present" a relatively minuscule percentage of the time. But you wouldn't believe how many right-wingers I've encountered who think voting "present" was all Obama ever did--and it's quite possible all these people got that mistaken impression from a claim that wasn't technically a lie.

    Similarly, when Stevens describes Romney as "little more than 320,000 votes short of winning," he is implying that 320,000 votes isn't a big threshold to overcome, but he provides no support for this assumption other than in the way he phrases the sentence.

  8. Agree with ASP, Peter, and Kylopod

    The thing is: the 2004 comp *was* relatively plausible. No, it's not as if it was extremely likely that Kerry could have done something to move Ohio 2%, but it's one state, not multiple states, and 2%, not over 5%. It's at least vaguely plausible that a different allocation of resources might have made a difference.

    Four states, with at least one over 5%? It's not a landslide, but it's hardly a close race.


Who links to my website?