Bell Curve The Law Talking Guy Raised by Republicans U.S. West
Well, he's kind of had it in for me ever since I accidentally ran over his dog. Actually, replace "accidentally" with "repeatedly," and replace "dog" with "son."

Friday, January 01, 2010

2010 Predictions

So, what's the point of following politics really closely if you aren't willing to go out on a limb and be a prognosticator? I'm sure there are mathematical models out there based on polls, voting history, and economic data that can predict 2010, but they are likely of limited value here. Such models are good explanatory (in reverse) but depend on data we don't have yet and have large margins of error, large enough to consume the tighter races and. Applying such metrics in advance reminds me of NFL referees who spot the football based on their own hunch, then measure it with a micrometer as if it were an exact science. I am going to make predictions based on how I expect the world to look in the summer of 2010.

Background: I presume that the campaigns of November 2010 will be "set" by summer 2010. What I mean by this is by August 2010, we will know what issues are going to be salient, which races are going to be contested, and the voters will have formed their impressions about (a) how Obama is doing (b) how the Democrats are doing on the Hill and (c) how the economy is going. So after Labor Day 2010, the arguments raised by the candidates will take place within this existing framework. I presume it will not be possible after that point to make large changes to public opinion about how the economy is going or what issues matter. (Only a whopping scandal, terrorist attack, or other massive unforeseeable event can really change the layout of campaigns from that point forward). Thus I base my predictions largely on how I expect the parties and candidates to position themselves in relation to this set battlefield.

Base predictions on issue salience-
Economy: the GDP will be enjoying its 5th straight quarter of growth by July 2010, and the unemployment rate will have begun to dip below 10%. Public opinion will amenable to the argument that the Obama administration has "fixed" the economy from the state it was left in by Bush. This will be the most important issue in the 2010 races.

Foreign policy: This is rarely an important issue. It will be mixed, in any event: Afghanistan will be an unpopular war, but the public will have no significant reason to fault Obama on foreign policy.

Deficits: Important if connected to a weak economy and fear of sliding back to recession; unimportant by itself.

Social issues: Independent voters hate these issues. They are unlikely to be a big deal because the GOP is getting tired of running on an anti-gay platform. Despite brutal and vocal Republican opposition, no cause in America has done as well in the past decade as gay rights.

Health Care Reform: Not an issue except for the party base on both sides.

Taxes: not an issue in 2010, because no promise to cut taxes will be taken seriously or perceived as responsible.

Likely themes: Democrats will look backwards to their economic accomplishment and warn of a return to Bush. Republicans will run hard against Barack Obama (NoBama) and deficits.

Likely mistakes: Democrats run the risk of downplaying their accomplishments, and weakening the base support/response, in the mistaken belief that they need to run to the right to win. Republians are a more right-wing, conservative bunch than even 2008 (smaller electorate) and they run the risk of having a bitter, extremist campaign that turns off the center. Democrats, in other words, run the risk of taking their base for granted, while Republicans run the risk of playing only to their base.

How to succeed: For Democrats, put Barack Obama out front and cheer for him. Bring back the love. For Republicans, urge caution and moderation.

Democrats will lose 10-15 seats in the house. These will be almost entirely marginal southern/border states where a smaller, more conservative midterm electorate cannot support them. The most written-about losses will be in Virginia, probably Dist. No. 5.

Pickups by D: NH, Missouri
Why: NH is trending blue and Carnahan can win MO given the economic issues.

Pickups by R: Arkansas, Colorado, Delaware
Why: Arkansas hates Obama. Colorado will not have the Obama excitement in 2008. Delaware will not vote for Biden's son for senator over its incumbent rep, Castle (R).

Seats that will be hard-fought but will not change hands: Nevada (Reid), Florida (Open-R), LA (Vitter), Penn (Specter), Ohio (Voinovich).
Why: Florida and Louisiana are trending Republican, while Nevada and Penn are going blue. These are very powerful demographic forces that aren't going to be turned back. As tough as things may look for Reid now, Nevada has never been more Democratic. Ohio will sink back to its Republican ways in an open seat in an off-year election.

Seats that are supposed to be hard-fought, but won't be: CT (Dodd), Illinois, Kentucky
Why: These races are beltway stories. Connecticut isn't going to elect a right-wing Republican, and there's no other kind available. Ditto for Illinois. Kentucky's Bunning may be unpopular, but KY isn't going to elect a Democrat now. Not given Obama's unpopularity in the Appalachia-Ozark area.

Predicted Senate: 59D - 41R.
Note that the projected pickups are all pretty tight races. We could legitimately see Dems from between 57-62.


Dr. Strangelove said...

LTG writes that mathematical models for election predictions are likely of limited value 11 months out. His point is well-taken, but there are two types of models worth distinguishing: correlative and causative. (And really, they need not be mathematical.)

There are a lot of correlative models floating around there, masquerading as science. For example, I heard an "expert" on the radio yesterday say that January was a harbinger month for the entire year, that the major stock indexes generally follow the trend from January for the rest of the year. He noted this had not been true in 1982, 2003, or 2008, but he expected it would be true for 2010. "But why would that be?" I wondered. The expert offered no explanation for the trend nor for his reason to believe it would hold this year, and the interviewer did not bother to ask for one. They both just chuckled knowingly about the oddities of the market.

This is exactly the kind of crap that makes people distrust science... Because people palm it off as science when it is nothing more than an observation of some coincidences. This great little NY Times article from 1992 says it all. The article was a lengthy debunking of the various methods of predicting presidential elections, such as looking at Crook County, Iowa's vote, the World Series winner, etc. And then it concludes with this gem:

"Mr. Lewis-Beck noted that there is one rule that has managed to survive completely intact, and that is that no left-handed President wins a second term. The rule worked for President James Garfield, President Gerald Ford and now President Bush. It does not bode particularly well for President-elect Clinton, a southpaw."

So you see, there was no chance Clinton could win a second term, because he's left-handed! Science proved it, right? Ugh. All I'm saying is, give me a causative model and then we can actually test the model in ways other than just looking at the final prediction: We can check the links in the chain. Fortunately, the best modelers (e.g. Nate Silver) have good explanations for the factors they include. They also realize their margins of error are large enough that, as LTG said, many races will get caught in the no-man's land between the points.

The Law Talking Guy said...

I really want to be careful because I don't want people to think I am pooh-poohing mathematical modeling per se. Formulae end up looking like the following "if the candidate is of the incumbent party and the economy is growing at X%GDP and so forth, then we can expect a likelihood of victory..." These have been shown to yield pretty good results, particularly when applied a few days before the election. But, as Dr.S suggests, this is correlative in nature.

What I am suggesting is that such electoral modeling requires good data inputs to work, and there won't be any for months to come.

So what we can do now is to make assumptions about what data will matter, then make educated guesses as to what the data might look like. Indeed, that's all anybody can really do until September. When you're picking all the inputs and deciding which ones matter and how much to weight them, you are basically picking the outputs too. It doesn't matter if you do this through formal modeling or informal analysis.

These "predictions" I'm making are really about laying down the basics that we know usually matter (1) which races have incumbent senators (2) which party is in power in Congress and in the presidency? (3) how does the candidate's party match up generally with the voting patterns/trends of their state (4) what is the economy like nationally (5) what is the economy like in that state (6) is the president's party popular? (7) are there any other particular issues or personalities of unusual importance?

The Law Talking Guy said...

Welll, things got grimmer for the Dems. Byron Dorgan (D-ND) will retire. While his seat was no doubt safe for him, it is unlikely that North Dakota would elect another Democrat to the Senate. There is a possibility that Earl Pomeroy will run, who is the at-large rep for ND and a Democrat, but he will likely continue in his place in the at-large seat in the House, largely uncontested for the booby prize of ND politics.

At this point, I would say that keeping 60 votes is going to be even harder, then. Expect 56-61 seats.

The Law Talking Guy said...

Now word a few hours later that Dodd will resign in Connecticut. I expect that Dems will retain the seat, if only because CT Dems are so angry about Lieberman that they will organize and vote. But it will take even more money now, and that means less for another seat somewhere else. On balance, to the advantage of the GOP.

Raised By Republicans said...

Dr. S. I think you are taking Dr. Lewis-Beck out of context. He was certainly being sarcastic and criticizing exactly the kind of silly "theories" you are complaining about.

Lewis-Beck is actually one of the world's leading experts on our ability (or inability) to predict election outcomes. I've heard the man speak several times and I can assure you that he was not seriously proposing that left handedness means anything.

Raised By Republicans said...

LTG, any one want to take a bet that Lieberman endorses the Republican candidate to replace Dodd?