OK, Here's My Election Prediction. What's Yours?
The shouting has almost stopped. Weeks ago the voting began, but Tuesday the balloting storm reaches landfall. It is time for pontificators to prognosticate. So here you go. My call:
Romney 52%, 277 electoral votes; Obama 47%, 261 electoral votes. Romney wins Florida, North Carolina, Virginia, New Hampshire, Iowa, Colorado, and Wisconsin -- and loses Ohio, Nevada, and Pennsylvania.
Why?
Yes, Michael Barone comes in 38 electoral votes higher (see his Examiner rundown). Figure that, between us, we've bracketed the range. For my part, I am giving Romney all the states in which Rasmussen has him up even by a point, and all the ties (with the exception of Ohio) where Rasmussen shows a tie. I am giving the president all of the states where he is up even by a point.
Why do the ties go to Romney?
At the recent American Enterprise Institute briefing on the election, Karlyn Bowman, AEI's polling expert (actually, AEI has a number of polling experts, including Barone, who also sat on that day's panel), noted that undecideds generally break between 55% and 60% for the challenger. That would give Romney the advantage in each of the poll-tied states. He won't necessarily win all those states, of course, but by giving up Ohio I've built in a cushion.
I endorse Michael's reasoning about fundamentals driving elections, as well as about state-by-state dynamics, even if I don't endorse all his picks.
On top of that, I am very impressed with Gallup's profile "What the Electorate Looks Like in 2012." The sample is of all likely voters Gallup contacted in their nightly tracking between October 1-24 -- 9,424 respondents all together, significantly larger than any other published sampling this year.
You may have read about the findings. While likely voter demographics this year (gender, age, education, ethnicity, regional balance) are pretty much the same as in 2008, party balance has shifted markedly. Last time the Democrats had a ten point edge on the Republicans. This time, the Republicans lead by a point (36-35). With independents breaking to Romney and Republicans significantly more passionate than Democrats, how plausible is it that the race is really tied, much less that the president leads?
At the New York Times' 538 blog, Nate Silver dismisses the plausibility of a Romney victory, putting the odds at 16%. Silver has a naive faith that all polls are created equal. One thing everyone should have figured out by now: it ain't so, particularly for polls with comparatively small sample sizes, which is the case with most state polls. Averaging a bunch of dreck findings doesn't give you a smaller margin of error. It gives you a bigger pile of dreck.
- Comment (44)
- · Quote
- · UnfollowFollow (7)
Comments:
May '11
Re: OK, Here's My Election Prediction. What's Yours?
Clark - if you are correct, then the increased stomach acid, decreased sleep, butterflies; all of those things that have made this past week just as bad as any 24 hour period before a big speech will be worth it.
Re: OK, Here's My Election Prediction. What's Yours?
We have exactly the same prediction. Here's my map:
Aug '10
Re: OK, Here's My Election Prediction. What's Yours?
Romney 295 - Obama 243.
Romney gets OH and CO, IA, WI, NH. Why?
Apr '11
Re: OK, Here's My Election Prediction. What's Yours?
Obama 288 Romney 250
Romney takes Colorado and Iowa but loses Virginia.
Apr '11
Re: OK, Here's My Election Prediction. What's Yours?
If Gallup is right that the demographics of likely voters will remain largely the same as 2008, that's a big danger sign, no matter what they are finding on Party ID. We really need the number of blacks and hispanics to be less than 2008;I don't know if the white crossover vote is enough to get it done.
Edited on November 4, 2012 at 7:05amJun '10
Re: OK, Here's My Election Prediction. What's Yours?
Well, it's hard to tell. The fraud has already started in NC.
Tonight, a guy I know said his wife, a friend of hers and the friend's daughter went to early vote. The friend and her daughter were told they had already voted [which they hadn't]. They called the Elections Board and were told that the time to contest voting ended two weeks ago. They said "but voting only started a week ago". They were told the next time to contest voting was in mid-November [a little late]
They were able to get the Elections Board to call up their supposed vote, and were told they had voted for Obama - which they weren't going to do. They will follow up.
While at the voting place, 4 busloads of people arrived and, as each person got off the bus, a person handed them a piece of paper, and pointed out things on the paper. The people took the papers and went into the voting place. They didn't know if they were "same day" registrants, union people or what.
Oct '10
Re: OK, Here's My Election Prediction. What's Yours?
Although I will settle for any Romney electoral victory, I’m with Frozen Chosen with 295 for Romney. The enthusiasm demonstrated by recent relative crowd size convinces me Romney will prevail in Ohio. Even Stevie Wonder couldn’t draw a crowd for Obama there.
May '10
Re: OK, Here's My Election Prediction. What's Yours?
Romney 317, 55% of vote; 54 GOP Senate; +2 GOP in House. Bluest state that goes for Romney: MN.
Mar '11
Re: OK, Here's My Election Prediction. What's Yours?
While I'd really like Romney to win Ohio, I'd take great pleasure in destroying the "You have to win Ohio to win" conventional wisdom. In a way, I think a Romney victory sans Ohio would perhaps drive a stake into the heart of the notion that many Ohions have: that the old union factory jobs paradigm can make a comeback. It can't.
Mar '11
Re: OK, Here's My Election Prediction. What's Yours?
Frozen Chosen: Romney 295 - Obama 243.
Romney gets OH and CO, IA, WI, NH. Why? · 47 minutes ago
It would be fantastic if that IA recall succeeds. I'm so tired of judges acting like they're our gods.
Edited on November 4, 2012 at 6:01amNov '11
Re: OK, Here's My Election Prediction. What's Yours?
I don't know, I really don't know. I'm stumped. My instinct is telling me that Romney is going to come through. My mind is evaluating the fact that only conservatives are predicting a Romney victory, and that they're arguing with the polls; that's a bad sign. But I'm also seeing Romney and Obama spending their last hours in decidedly blueish states, and they have the best data available.
Both the red states and the blue states seem to be redder, with the swing states just barely holding on for Obama. So he probably has the edge. But his focus on Ohio, etc., seem to have left some of the bluer purple states vulnerable, and if undecideds break for Romney and turnout favors Republicans, it might not be pretty for Obama.
Jun '12
Re: OK, Here's My Election Prediction. What's Yours?
I went with 52:47 too, and I think I agreed with all of your state calls except I give Mitt Ohio: http://j.mp/Y9TsOd That puts it at 295:243. Though I expect Ohio to take a while for a solid count (maybe not until they sort through the provisional ballots), so I'm hoping that Wisconsin & Iowa will spare us the stress of waiting for Ohio's tally.
Edited on November 4, 2012 at 6:40amMar '11
Re: OK, Here's My Election Prediction. What's Yours?
My prediction - the Silent Majority take their revenge and Mr Barone is right.
Edited on November 4, 2012 at 6:45amJul '12
Re: OK, Here's My Election Prediction. What's Yours?
I don't think the sample sizes are what's wrong with Nate's predictions... it's the statistical bias in the polls themselves (projecting based on 2008 trends).
May '12
Re: OK, Here's My Election Prediction. What's Yours?
Here's mine: No Iowa or New Mexico, but Ohio and Wisconsin come through
Re: OK, Here's My Election Prediction. What's Yours?
Alas, I don't share this board's optimism. I think we lose Ohio, Wisconsin, Nevada, NH, and maybe Virginia too, though I think we have good odds of getting Colorado and Iowa. I think Romney is going to fall just a bit short.
Aug '12
Re: OK, Here's My Election Prediction. What's Yours?
Obama, 300
Sep '12
Re: OK, Here's My Election Prediction. What's Yours?
wmartin: Obama 288 Romney 250
Romney takes Colorado and Iowa but loses Virginia. · 5 hours ago
I don't see there is any way for Obama to win VA. NV is going blue. I'm hopeful for an Ohio victory and I'm with Barone that Pennsylvania has a good chance of flipping red. New Hampshire as well. Wisconsin, Michigan, Minnesota all stay blue. Romney 305. Obama 233.
http://www.270towin.com/2012_election_predictions.php?mapid=bgLQ
Oct '12
Re: OK, Here's My Election Prediction. What's Yours?
Obama 294, Romney 244
Popular vote prediction: 49.8% to 48.8%
Edited on November 4, 2012 at 11:17amSep '12
Re: OK, Here's My Election Prediction. What's Yours?
I'll have to quibble with one of Clark's points. Errors from small sample sizes are statistical errrors, and you do, in fact, reduce these by averaging several small samples together. Think of it this way: add up the number of Romney voters across all small polls, add up the number of Obama voters across those polls, and divide each by the total number. You've effectively got one big poll.
We do the same mathematical operation in astronomy, when we add together several short camera exposures of the sky to make (effectively) one long exposure. Short exposures have large percentage errors in a star's brightness (we call this the "noise"), while long exposures have low noise. But by adding several short exposures, we reduce the uncertainty in the brightness, just as if we had one long one.
Now, what you don't fix are the systematic errors, and if those polls have different systematic biases, then combining them this way leads to a result with hard-to-quantify systematics.