Who Wins? A 2012 Projection Based on the Partisan Skew Hypothesis

In my recent (and surprisingly popular) post about the weirdly good polls the Democrats have been seeing recently, I suggested that the polls might be suffering from an unidentified systemic source of bias.  I promised a followup discussing the implications of systemic polling bias on the presidential race.  Since we are now only three days away from Election Day, I’ve decided to go whole hog and write up my final projections for the race at the same time.

But first, polling bias and its effects.  You might be asking yourself, “Why is James putting himself out on a limb endorsing this fringe electoral theory a week before the elections?  Why not just sit back and wait for the fallout?”  Frankly, it’s because I want the credit if I’m right.  I think the theory has merit, and I am looking forward to seeing the glory that will accrue to me on Wednesday morning.  (If I’m proved wrong, then you will forget.  This is the magic of punditry.  Win-win for me.)  In fact, the falsifiability of this thesis is one of its most exciting aspects.

So suppose I am right.  The polls are biased; the Democratic turnout advantage is smaller  than projected.  Does Romney win?  Well, that depends on how large the bias is, and I can’t even begin to hazard a reasonably useful guess about that.  Normally, I would check the polls, but the polls are precisely the problem here.  Some have pointed at Gallup’s recent voter ID surveys as evidence that Republicans will achieve turnout parity in 2012 — but Gallup’s polls are already showing much better results for Romney than the polling consensus, which suggests that their model is simply better attuned to Republican voters, fully explaining their voter ID results.  We have no way of determining whether their model is superior to the others on the market, and, indeed, we should assume that it is wrong, because it is a dissenter.  We really are just guessing.

But before we try to hazard a guess, let’s look at the big picture of polling bias.  A few days ago, I grabbed two polls from Ohio virtually at random.  I checked their crosstabs to obtain their results by party, then rescaled the polls to show how the poll would have likely turned out if there had been different numbers of Republicans and Democrats in the sample.  Here is the snazzy chart:

You’ll notice the “inferred” results from the SurveyUSA poll.  The partisan ID categories SUSA provided were “Republican”, “Democrat”, and “Independent”, but the totals added up to only 96%, and (just using those 96%) the final poll results I calculated were off from SUSA’s published results by about a point.  I concluded that SUSA had excluded self-identified third-party voters from their partisan ID results, and I did some math to figure out roughly what their 4% had contributed to the final poll results.  (Which makes me, I don’t know, Yoda?)  Hopefully, I done right.

I deliberately used polls that showed both bad and good results for Mr. Romney.  At the time the chart was made, the FiveThirtyEight average showed Ohio at Obama+2.3 (it is now Obama+2.9).  You’ll notice that the PPP poll is a bit lower than that (Obama+1), while the SUSA poll is a bit higher than that (Obama+3.5).

However, both polls show the same strong partisan ID in favor of the Democrats, and D+8 and D+7, respectively.  Ohio’s official partisan turnout in 2008, the Democratic wave election, was D+8.  (In reality, it was likely closer to D+5.)  In 2010, the Republican wave year, it was R+1.  My hypothesis is that these partisan ID’s are several points more Democratic than reality will reflect on Tuesday.  For instance, if all we do is rescale these polls so that they show the 2008 D+5 result, rather than their current vast partisan margins, both polls tighten up significantly: SUSA’s 3- or 4-point race becomes a 1- or 2-point race.  PPP’s 1-point lead for Obama suddenly flips to a 1-point lead for Romney.

And that’s assuming the Democrats do about as well in crushing Republican turnout as they did in 2008!  Further adjustments to account for theoretical “missing Republicans” only shift the polls more in Romney’s favor.  These changes are very steady, which allows us to calculate a rule: every extra percentage point voter ID for Republicans translates to a 0.8% change in the race.

I should emphasize now that I looked at only two polls, which means this post should be considered strictly unscientific.  My reviews of other recent polls strongly suggests that my findings would be generalizable, but I can’t swear to it.  I’ve proved nothing, and that’s even if you assume my fringe theory is correct to begin with.  All I have done is set the stage for my personal 2012 election projections.  (However, there is this really nifty blog, which I came across while writing this post.  They are reaching similar conclusions, but doing a ton more data processing than I am.  Check it out!)

So, now that we have quantified the effect of statistical polling bias, we have to ask ourselves what will be the magnitude of the bias on election day?  How many voters will the Democrats and Republicans actually turn out?  Normally, we would ask the polls.  Instead, I just have to guess.

There’s really no hiding that.  It’s a guess.  I can educate myself as much as I can.  I can look at the best available numbers on this year’s enthusiasm gap. I’ve got the Washington Post’s cool story on Obama’s defectors.  I can grab Pew’s 2008 voter ID findings.  And then I can notice that their voter ID numbers are clearly wrong (they suggest a popular vote margin of +7.5 for President Obama; the reality was +6), and rebalance their D+7 sample to reflect, with some confidence, the actual partisan ID gap in 2008 (I end up with D+5).  I can even look to my own scant research into swing state partisan changes during 2008-2012.

But, in the end, it’s still a guess.  To me, it appears that the President will probably not meet his 2008 turnout numbers, even with early voting plumping him up.  That sets his ceiling at D+5.  At the same time, this has none of the hallmarks of the 2010 Republican wave, where the margin was about D+0 (or D+1).  Everything about the race so far suggests that the parties are about evenly matched.  So let’s split the difference and predict that this race will end up at a partisan ID of about D+3.

If I am correct (and remember I am just guessing, which makes this more witchcraft than science), then the polls will be off by a significant margin.  They have been showing an average partisan bias of about D+8, according to my recent findings.  My D+3 guess would make their partisan shares about five points off.  As we discussed above, every point of change in partisan share is worth about 0.8 points in the final margin.

We can now calculate the expected bias (the difference between the final polls and the actual returns) in this race: 0.8% * 5 = +4% Obama.  In other words, I believe the polls are inflating President Obama’s margin by about 4 points.

Right now, based on current polls, FiveThirtyEight projects the final popular vote at 50.6% Obama, 48.4% Romney (+2.2 Obama).  Based on my guess about the true partisan ID in this race, I project an actual result of 48.6% Obama, 50.4% Romney (+1.8 Romney).  If I am correct, then this is a very close race, but Mr. Romney will win it about 3 times out of 4.

My projected battleground state margins:

Colorado: Romney+2.5
Florida: Romney+4.4
Iowa: Romney+0.9
Nevada: Romney+0.1
New Hampshire: Romney+0.7
North Carolina: Romney+6.5
Ohio: Romney+1.1
Pennsylvania: Obama+1.4
Virginia: Romney+2.8
Wisconsin: Obama+1.1

This is not quite as strong for Romney as it looks… but it is not bad.  It leaves him quite a lot of paths to victory, and very few for the president:

Finally, even though it’s stupid to project a literal electoral map (they are too fuzzy!), I can’t resist making the attempt.  The beauty of it is, I win either way: I predicted back in March that Romney could not win the electoral college.  So now I’m predicting that he can, and here’s how he does it:

However, please bear in mind that, even if I am right about everything, this is still a very close race.  Even if you accept all my theories and guesses as accurate, President Obama still has a very good chance of picking up an Iowa or an Ohio and ruining the electoral math for Mr. Romney.

Of course, if I’m not right about everything, it will be a very short night.

And don’t underrate that possibility.  There is a very sizable chance that I am dead wrong.  For all we know, the polls could be biased against the Democrats and they’re about to sail to a blowout win!  Because of the great uncertainty about my predictions, the tenuousness of Romney’s lead even in my projection, and the historical strength of polling, I am still going into election night with a realistic belief that President Obama is 60-75% likely to be re-elected.  Mitt Romney remains the underdog.

On the bright side, we should know relatively early whether I was right or the polls were.

P.S. It would be really interesting to find out what effect my predicted polling bias could have on the Senate races.  But there aren’t enough stolen hours in the world for me to do that write-up before the election.

P.P.S. I predicted Gore in ’00 and Kerry in ’04, but Obama in ’08 and the Republican wave in ’10.  One way or another, his election will be the tiebreaker!  (Well, I also predicted Dole in ’96, but, to be fair, I was 7.)

 

Posted in Analysis, Horse Race, Politics | 1,796 Comments

Romney/Obama as Baseball: 3 November: THE LAST INNING

It is now the top of the 9th and final inning.  Despite a few scattered base hits, the score has not changed since Romney’s big inning in the 6th.  The 8th inning saw five strikeouts between the two teams, but you wouldn’t know it from all the foul balls and near-misses the batters knocked around before finally eating dirt.

The score remains Obama 7, Romney 6 as both ballclubs prepare for their final at-bats.  This is baseball, folks: anything could happen!

Still, the Obamaniacs will win this game 84% of the time, which is the same probability FiveThirtyEight.com gives the President’s re-election campaign.

(NOTE: I don’t entirely buy into Mr. Silver’s model at the moment, as I suspect the polls are somewhat biased this cycle.  See my explanation here.)

Posted in Horse Race, Politics | 2,149 Comments

The Great Democratic Turnout Miracle

During this cycle, likely voter polls have consistently shown more Democrats showing up to vote than Republicans.

Actually, that is a bit of an understatement.  It is perfectly normal for Democrats to outnumber Republicans, even at the polls, even in strong Republican years (like 1984 and 2010).  Historically speaking, an equal number of Democrats and Republicans showed up to vote in the presidential election of 2004.  Independents sided with the Democrats that year, but the Republicans made up for it with especially strong turnout in battleground states.  In 2010, the Republicans and Democrats were again equal, but it became a wave election because independents sided with Republicans.  2008, by contrast, was the perfect storm for Republicans.  Democrats had the largest turnout margin in history (+7% versus Republicans) and overwhelmingly won independent voters to deliver a huge electoral territory to Sen. Barack Obama.  So predicting that the Democrats will have a turnout advantage in the election is like predicting that Pennsylvania will go blue: once the flirting’s over, the answer is always yes.

What’s weird during this cycle is that likely voter polls have consistently shown Democratic turnout crushing Republicans by a margin that augur not only the end of the Romney candidacy but possibly the Republican party itself.

Continue reading

Posted in Horse Race, Mere Opinion, Politics | 2,546 Comments

Romney/Obama as Baseball: 26 October

It looks like the Obamaniacs have finally got their pitching problems worked out!  All they had to do was put their main man on the mound.  He’s not doing anything special, nor anything we haven’t seen a million times before, but it’s doing the job and holding the line.

As we enter the home stretch (did you enjoy Lena Dunham‘s rendition of “Take Me Out to the Ball Game”?), the pace of the game has slowed dramatically — which is good for Team Obama, which continues to hold a narrow lead of Obama 7, Romney 6 here in the Top of the 8th.  Romney’s baserunner has not been able to advance to second (and scoring position) thanks to a ground out from this week’s debate.

Team Obama will win this game 74% of the time. According to FiveThirtyEight.com, President Obama has a similar probability (73%) of winning re-election.

Posted in Horse Race, Politics | 2,119 Comments

Romney/Obama as Baseball: 18 October Update

In the bottom of the 6th, Fightin’ Joe Biden came in to pinch hit for an exhausted David Axelrod. He made a lot of noise, fouled a lot of balls, and got the Romulans jeering that he was all talk, but finally managed to knock a soft base hit down over first baseman Paul Ryan’s head to drive in the runner on second. Team Obama retook the lead, Obama 7 – Romney 6, then two strikeouts ended the inning.

In the top of the 7th, down a run with just a little time left to make it up, Team Romney has certainly resumed the position of the underdog… but it looks like Obama relief pitcher Steph Cutter is tiring out a lot faster than she should. Romney’s boys get a base hit and a sacrifice fly to put a runner on second with one out here in the top of the 7th.

Team Obama will win this game 65% of the time. According to FiveThirtyEight.com, President Obama has the same probability of winning re-election.

Posted in Horse Race, Politics | 1,580 Comments

Romney/Obama as Baseball

Photo: Steve Mitchell-US PRESSWIRE, 15 July 2012

A few weeks ago, I was watching Mitt Romney’s polling numbers dive, and, on both sides of the aisle, the conversation — and the conventional wisdom — was starting to ask not, “Who will win the election?” but, “Why will Mitt Romney lose it?”  Nate Silver, of FiveThirtyEight.com, was kind enough to place a number on the probability of President Obama’s re-election, as he does every day, but the number was more or less meaningless.  Just how likely is 25%, in real-world terms?  Should I join the recriminations crowd, or wait patiently to see whether things changed?

Mr. Silver suggested an answer in an off-hand reference to the apolitical sport where he got his start as a practical statistician: baseball.  He linked me to the Win Probability Inquirer at the Hardball Times, and off I went!  After all, I’ve only seen four presidential elections, and they’re very complicated games, but I have seen a lot of baseball over the years (thanks, Dad!).  If you’re down 2 runs in the top of the first inning, I have a gut sense of what that means — and I know how different it is from being down 2 runs in the 6th. So all I needed to do was look at the daily election forecast at FiveThirtyEight, find a baseball scenario that matched that probability, and, bam, suddenly I knew where the race really stood.

Continue reading

Posted in Horse Race, Politics | 2,110 Comments

Impromptus on the Vice-Presidential Debate, 12 October 2012

Alright, took me a while to pin this down, but I’ve figured out how I feel about the debate.

First, I think Vice Pres. Biden lost tonight. It is true that, at this hour, it seems that a slim but certain majority believe he won the debate, and that would seem to settle that. But I suspect that, although Biden’s chokehold on civil discourse gave him a technical victory, his obnoxious display will actually make the electorate less likely to vote for him. We shall see how this hypothesis plays out in this week’s polls. (We will also need to keep an eye on partisan ID results, to find out whether Mr. Biden successfully re-energized his base.)

Second, I do not think that Rep. Ryan won tonight. Veep Biden’s loss was entirely at his own hands. Ryan seemed content to get out of the way, conceding the last word to Biden on many occasions and abiding by the will of the moderator like the polite young man he is. He never said the words, “Mr. Vice President, that is a bald-faced lie,” and he could have done it at least three times. Tactically speaking, that was probably wise — but it is a gamble, since he is counting on the electorate to reject Mr. Biden’s totally inappropriate display.

Continue reading

Posted in Horse Race, Mere Opinion, Politics | 2,090 Comments

The Appropriateness of Constitutional Amendments

In Minnesota’s ongoing debate over the constitutional questions before voters this fall, opponents have occasionally raised a question that has nothing to do with the wisdom of the policies themselves.  “Is it appropriate to enshrine this policy in the Minnesota state constitution?” they will ask, or, if they are in a somewhat more confrontational mood, “Aren’t you just using the referendum to bypass the legislative process” — by which they mean the governor’s veto pen — “and enact your party’s initiatives by any means necessary?”

Usually, this question is asked in the traditional civic spirit that pervades American politics, by which I mean bald-faced hypocrisy.  In 2012, with a Democratic governor and Republicans in control of the entire legislature, it is Democrats angrily demanding that Republicans cease this coup d’etat against the typical legislative process.  In 2008, when the power structure was reversed, so was the question: Republicans accused Democrats of abusing the amendment process to pass their agenda into law.  (The 2008 “Legacy Amendment” proposal, which sought a new sales tax to fund Art and Nature, passed handily.)  Every single self-styled constitutional conservator I have met in 2012 voted for the Legacy Amendment in 2008.  Now they raise the very argument they ignored in 2008 to attack the amendments they don’t like.  The reverse is true for those who questioned the Legacy Amendment’s legitimacy four years ago.  It is difficult to take so many unserious people seriously.

But the question itself is a good one, and a sincere and inquisitive reader recently asked it of me, which always demands a thorough reply.

Continue reading

Posted in Law, Mere Opinion, Politics | 2,726 Comments

Go Ahead. Abolish the Electoral College.

…but don’t come crying to me when you realize what you’ve done.  If you want to increase the bitterness, brutality, and extremism of American politics, by all means, eliminate the Electoral College.  Otherwise, think again.

People get all surprised when I say this out loud.  Two of the least popular things in America are the bitter tone of our political discourse and the Electoral College.  The people, by and large, would love to be rid of both.  Moreover, I’m too lazy to go find polls to back this up, but I seem to recall reading that decrying the bitterness of politics correlates with favoring a national popular vote (and being a less informed and engaged voter, and voting for Democrats, incidentally; zing).  Now that there is a plausible anti-Electoral College campaign underway that is, at this writing, a little more than halfway to its goal of a national popular vote, it’s time to discuss the relationship between the College and the health of our politics.

Continue reading

Posted in Mere Opinion, Politics | 2,210 Comments