What Might Be and Ought to Be
Predictions for the Presidential and Senate Races
With a few days left to the election, it's time to offer my prediction.
This is a somewhat unscientific projection. I don't use reams of data and fancy computer models like Nate Silver, and that's a good thing, because I think that pretends to a level of scientific precision that just isn't possible. Instead, I rely on a combination of the polls, a couple of decades of experience watching politics, and common sense.
I'm projecting a victory for Mitt Romney, 53% to 46%, with 315 Electoral College votes. In Tuesday's second most important election result, I'm predicting that Republicans will win 50 votes in the Senate, maybe 52.
That's the bottom line. Now let's talk about how I got to this result.
Let's start with the ultimate fundamentals of the race. Obama is running for re-election having presided over the slowest recovery on record, a long, agonizing slog of misery and hopelessness. The last two years have seen the slowest growth ever, outside of an actual recession. Unemployment has remained high for an unconscionably long period of time. It is not morning again in America and it doesn't even remotely feel like it. So Obama cannot take credit for the recovery from the 2007 to 2009 recession, since his main contribution was to stifle it.
Instead, his primary achievement in office was to shove through, over massive resistance, a massive expansion of the welfare state which remains unpopular. He has attempted to alter the balance between individualism and collectivism in this country and to expand the role of the state, but the American people have resisted this attempt at every step.
Finally, any advantage the president gained from his one big foreign-policy triumph, ordering the strike that killed Osama bin Laden, has been offset by the debacle in Benghazi, which was first and foremost a failure of top-level White House decision-making.
All of these are reasons why Obama shouldn't win re-election. If the American people agree, we would expect to see relatively low enthusiasm among Democratic voters, high enthusiasm among Republican voters, and a big lead for Romney among independent voters.
And that is exactly what we see.
This result, however, is obscured by the reported poll results. I've been discussing the polls pretty extensively in my RealClearPolitics newsletter, and the issue hasn't fundamentally changed since September. Two months ago, when the polls showed a huge lead for Obama, and Democrats started declaring the election over already, a lot of us pointed out that the polls assumed a massive Democratic turnout advantage. Basically, they were assuming that Republicans would be as demoralized as they were by McCain's campaign in 2008, and therefore they would stay home—while Democrats would be as delirious about re-electing Obama as they were about electing him in the first place, and they would flock to the polls in record numbers.
It's helpful to have a little background about polling. When you conduct a public opinion poll, you call up, say, a thousand people chosen at random. But to say that the sample is random is not to say that it is representative. You may happen, either by coincidence or through some undetected bias in your system, to contact a group with more whites or more blacks or more women or more upper-middle-class professionals or more union members, or whatever. This problem has only gotten worse, because most pollsters still call only traditional land line telephones, while many voters use only cellphones.
As you extrapolate from these thousand people to represent 120 million voters—each respondent standing in for 120,000 others—there is a huge potential for error if your sample of voters is tilted in one direction or another. So what all pollsters do is to rebalance their results to reflect what they know about the demographic makeup of the electorate. If they have too few poor people, for example, and too many in the middle class, they give extra weight to the poor people and less weight to the middle class to compensate.
Yet most pollsters don't re-weight their results to reflect party identification—i.e., whether they have more Democrats or more Republicans. Their justification is that while other demographic characteristics are fixed—i.e., whether you're black or white, or male or female—party identification is fluid. And after all, isn't a shift in loyalty from one party to another one of the things we're trying to measure? But this ignores the fact that shifts in voter identification from one party to another can be measured separately, both directly, through polls specifically targeting that issue, and indirectly, through other measures of enthusiasm for the two parties. And none of that data matches the voter turnout assumed by most mainstream polls.
The pollsters have been told time and again, for months, that polls which assume even greater turnout for Democrats than in 2008 are simply unrealistic. They have stubbornly refused to change anything about what they do. This is why I think that the biggest losers on Tuesday will be the mainstream pollsters, whose projections will be wildly off the mark.
A more realistic view of the polls should reflect the kind of results we see in a very recent Gallup survey which indicates that Republicans will turn out in equal or possibly larger numbers than Democrats, a huge swing from 2008. To factor that in, I think you have to "unskew" the polls by three to six percentage points, which yields a three-to-six point lead for Romney nationally and gives him the Electoral College votes of an awful lot of the swing states.
If you start with the RealClearPolitics poll average and add three points to Romney and subtract three from Obama, you get 50% for Romney to 44% for Obama. Assume that one percent of voters will go for third party candidates. Then assume that the remaining 5%, the as-yet-undecided voters, break slightly for Romney, so he gets an extra 3% and Obama gets an extra 2%. That's how you get to 53% for Romney and 46% for Obama. It's not quite a landslide—54% and above is getting into landslide territory—but it is a decisive result.
Note that the "unskewing" I'm doing here is relatively conservative. It doesn't assume that the polls are off by 6%, only 3%, which is pretty much in line with the results from previous years. And it assumes that undecided voters will break slightly for Romney, not overwhelmingly. I think if any of these assumptions are off, they are more likely to be too conservative and that Romney's lead will be bigger.
Correcting for the skew caused by unrealistic turnout assumptions allows us to focus back on what is really important from the poll results.
There are basically three ways President Obama can win re-election. He can get Democrats to turn out in much larger numbers than Republicans, he can get Republicans to cross over and vote for him, and he can win a majority of independent voters who aren't aligned with either party. In 2008, he did all three of these things. But this year, he doesn't have a turnout advantage. Polls also show that about 90% of both Democrats and Republicans are supporting their party's candidate, so crossover votes aren't likely to be a big factor either way. That leaves the election to be decided by independent voters, who favor Romney by a 10 to 12 percentage point margin.
That is the main reason other pundits on the right are also predicting a Romney victory. Michael Barone, an experienced observer with a literally encyclopedic knowledge of American politics—he is the principal author of The Almanac of American Politics—cites Romney's lead with independents as the basis for predicting that Romney will be the next president. I agree with his rundown of the various swing states, which yields a prediction of 315 electoral votes for Obama, though I also agree that he is going out on a "wobbly" limb in predicting that Romney will win Pennsylvania, so Romney could win with only 295 electoral votes.
Now, it's possible that I'm wrong and that Obama will squeak through with a victory by demagoguing the auto bailout in Ohio. From the beginning, I have held no illusions about Mitt Romney's weakness as a candidate; he is only the nominee because he turned out to be the best of a bad lot. But in his eagerness to run to the center and portray himself as a moderate in the last month of the election, Romney has created the impression—particularly in the final debate—that he mostly agrees with President Obama on the issues. This has been exaggerated, but there is enough truth to it to make it an effective argument. So Romney could come across as the "me-too" candidate who cedes the issues to his opponent, giving voters no reason to choose him over the incumbent.
I certainly think we can see this in the fact that Romney did not gain any real "Mitt-mentum" to follow up on the initial surge from his outstanding debate performance on October 3. A few weeks ago, he looked like he was in a position to secure a solid lead in the race, even measured by skewed polls. But he has since given back a point or two and runs within a few tenths of a point of Obama in the RealClearPolitics average. But I have not seen any sign of a collapse in support for Romney.
An interesting indication of the state of the race is what has been happening in the contest for control of the Senate, an election result that will be almost as important as the presidential race, since it will determine what the president is able to accomplish. Here, virtually all of the competitive races have narrowed to a dead heat in the past few days. See the RealClearPolitics Senate map. There are 14 races that are considered competitive, and 12 of them are now marked as pure toss-ups, where neither candidate has a definitive advantage in the polls.
Since I am calling ties in favor of the Republican—given the skew in the polls—I think Republicans have a very good change of winning seven of these races: in Indiana, Arizona, Montana, Nevada, North Dakota, Virginia, and Wisconsin. That would give them 50 votes, allowing Vice-President Paul Ryan to break ties in their favor, giving Republicans effective control of the Senate. Moreover, if Romney proves to have any "coattails," I think Republicans have a good chance to win in Massachusetts and Pennsylvania. If Romney does very well, he could even pull along Ohio, Connecticut, and Florida. Even Todd Akin is only a few points behind Claire McCaskill, after blowing up his campaign on the launch pad with his "legitimate rape" comments. If Republicans can win the Senate without him, I will be happy. But the fact that he even has a shot indicates the depth of popular resentment against the Democrats. So 56 seats is probably the upper bound for Republicans in the Senate, but I would consider 52 seats to be a very good showing.
Which is to say that the likely Senate results are consistent with my prediction for the presidential race, as you would expect them to be.
At any rate, the bottom line for this election is: will the actual results coincide with what they should be? Will the American people find President Obama's record as unsatisfying—not just on the dry, statistical level but on the sense-of-life "gut" level—as they ought to? Ultimately, my confidence in a win for Romney reflects my confidence that when the American people are presented with this kind of sense-of-life choice between individualism and government control, between striving for success and resigning themselves to failure and stagnation, they will make the right choice. They always have.
[fbshare type="button"] [twitter style="horizontal" float="left"]