Tuesday, May 27, 2008 at 7:09 pm by Darryl
Poll Analysis: McCain Loses the Lead to Obama
| Obama | McCain |
| 50.1% probability of winning | 49.3% probability of winning |
| Mean of 268 electoral votes | Mean of 270 electoral votes |
Just yesterday, Sen. Barack Obama was trailing Sen. John McCain. A Monte Carlo analysis of state head-to-head polls suggested that, if the election were held now, Obama would only have a 38.5% chance of winning a general election. With the addition of one new poll today, Obama snatches back the lead from McCain.
After 10,000 simulated general elections, Obama wins 5,008 times, but he also takes the 62 ties, and McCain wins 4,930 times. In other words, if a general election were held today, Obama would have a 50.7% probability of winning (that is 50.1% plus 0.6% for the ties) and McCain would have a 49.3% probability of winning.
Here is the distribution of electoral votes [FAQ] from the simulations:
- 10000 simulations: Obama wins 50.1%, McCain wins 49.3%.
- Average ( SE) EC votes for Obama: 268.1 ( 15.5)
- Average (SE) EC votes for McCain: 269.9 ( 15.5)
- Median (95% CI) EC votes for Obama: 270 (242, 298)
- Median (95% CI) EC votes for McCain: 268 (240, 296)
| State | EC Votes | # polls | Total Votes | % Obama | % McCain | Obama %wins | McCain %wins |
|---|---|---|---|---|---|---|---|
| Alabama | 9 | 1* | 812 | 39.5 | 60.5 | 0.0 | 100.0 |
| Alaska | 3 | 2 | 1001 | 45.7 | 54.3 | 0.1 | 99.9 |
| Arizona | 10 | 1 | 561 | 43.9 | 56.1 | 0.3 | 99.7 |
| Arkansas | 6 | 1 | 480 | 40.6 | 59.4 | 0.0 | 100.0 |
| California | 55 | 4 | 2591 | 56.8 | 43.2 | 100.0 | 0.0 |
| Colorado | 9 | 1 | 450 | 53.3 | 46.7 | 92.5 | 7.5 |
| Connecticut | 7 | 1* | 1476 | 59.8 | 40.2 | 100.0 | 0.0 |
| Delaware | 3 | 1* | 553 | 55.0 | 45.0 | 98.6 | 1.4 |
| D.C. | 3 | 0 | (100) | (0) | |||
| Florida | 27 | 2 | 1671 | 46.8 | 53.2 | 1.0 | 99.0 |
| Georgia | 15 | 3 | 1733 | 42.9 | 57.1 | 0.0 | 100.0 |
| Hawaii | 4 | 1* | 546 | 66.3 | 33.7 | 100.0 | 0.0 |
| Idaho | 4 | 1* | 553 | 42.9 | 57.1 | 0.1 | 99.9 |
| Illinois | 21 | 1* | 546 | 65.9 | 34.1 | 100.0 | 0.0 |
| Indiana | 11 | 1 | 1215 | 50.5 | 49.5 | 66.8 | 33.2 |
| Iowa | 7 | 2 | 940 | 53.4 | 46.6 | 98.4 | 1.5 |
| Kansas | 6 | 1 | 445 | 38.2 | 61.8 | 0.0 | 100.0 |
| Kentucky | 8 | 2 | 991 | 36.1 | 63.9 | 0.0 | 100.0 |
| Louisiana | 9 | 1* | 465 | 44.1 | 55.9 | 0.9 | 99.1 |
| Maine | 4 | 1 | 445 | 57.3 | 42.7 | 99.7 | 0.3 |
| Maryland | 10 | 1* | 577 | 57.0 | 43.0 | 99.9 | 0.1 |
| Massachusetts | 12 | 1* | 450 | 56.7 | 43.3 | 99.6 | 0.4 |
| Michigan | 17 | 2 | 877 | 48.3 | 51.7 | 15.9 | 84.1 |
| Minnesota | 10 | 2 | 1449 | 56.0 | 44.0 | 100.0 | 0.0 |
| Mississippi | 6 | 1 | 558 | 41.9 | 58.1 | 0.0 | 100.0 |
| Missouri | 11 | 2 | 1856 | 48.0 | 52.0 | 4.6 | 95.4 |
| Montana | 3 | 1 | 569 | 43.9 | 56.1 | 0.3 | 99.7 |
| Nebraska | 5 | 2 | 961 | 37.8 | 62.2 | 0.0 | 100.0 |
| Nevada | 5 | 1 | 430 | 46.5 | 53.5 | 6.8 | 93.2 |
| New Hampshire | 4 | 3 | 1329 | 49.4 | 50.6 | 28.7 | 71.3 |
| New Jersey | 15 | 1* | 707 | 63.6 | 36.4 | 100.0 | 0.0 |
| New Mexico | 5 | 2 | 983 | 52.3 | 47.7 | 91.8 | 8.2 |
| New York | 31 | 2 | 976 | 57.9 | 42.1 | 100.0 | 0.0 |
| North Carolina | 15 | 5 | 2899 | 46.3 | 53.7 | 0.0 | 100.0 |
| North Dakota | 3 | 1* | 218 | 46.3 | 53.7 | 13.0 | 87.0 |
| Ohio | 20 | 3 | 2012 | 50.0 | 50.0 | 50.6 | 49.4 |
| Oklahoma | 7 | 1* | 569 | 40.1 | 59.9 | 0.0 | 100.0 |
| Oregon | 7 | 1 | 455 | 57.1 | 42.9 | 99.7 | 0.3 |
| Pennsylvania | 21 | 4 | 3079 | 53.5 | 46.5 | 100.0 | 0.0 |
| Rhode Island | 4 | 1* | 572 | 58.2 | 41.8 | 100.0 | 0.0 |
| South Carolina | 8 | 1* | 554 | 48.4 | 51.6 | 20.6 | 79.4 |
| South Dakota | 3 | 1* | 221 | 39.8 | 60.2 | 0.2 | 99.9 |
| Tennessee | 11 | 1* | 450 | 42.2 | 57.8 | 0.1 | 99.9 |
| Texas | 34 | 2 | 1001 | 44.9 | 55.1 | 0.0 | 100.0 |
| Utah | 5 | 1 | 537 | 30.4 | 69.6 | 0.0 | 100.0 |
| Vermont | 3 | 1* | 576 | 68.4 | 31.6 | 100.0 | 0.0 |
| Virginia | 13 | 3 | 1725 | 48.6 | 51.4 | 9.6 | 90.4 |
| Washington | 11 | 2 | 1088 | 56.2 | 43.8 | 100.0 | 0.0 |
| West Virginia | 5 | 1* | 540 | 39.8 | 60.2 | 0.0 | 100.0 |
| Wisconsin | 10 | 1 | 450 | 47.8 | 52.2 | 16.0 | 84.0 |
| Wyoming | 3 | 1* | 508 | 39.4 | 60.6 | 0.0 | 100.0 |
* denotes that an older poll was used
Details of the methods are given in the FAQ.
The most recent analysis in this and other match-ups can be found from this page.
Wednesday, May 28th, 2008 at 6:25 am
like i was saying what happened is the expiration of older “less favorable for obama” polls in IN and OH.
The next event of this nature will occur in NH which (if there are no new polls unfavorable to Obama) will go blue in your model in a few days also.
Your expiration method really needs more polls to prevent things from changing simply on expiring polls versus new polls. I see the method improving dramatically with fresh polls closer to the GE. That’s perhaps a DUH! comment.
however the statement i.e. “With the addition of one new poll today, Obama snatches back the lead from McCain.”
…is really not what caused a change in the %. It was the expiration of an Ohio poll wouldnt you say?
I guess the inaccuracy of the predictive element (which is what I think bothers a few most vocal commentors …or perhaps its the model outcome LOL) is highlighted this far away from an actual contest.
More polls with more frequency = better model.
But what I guess is an issue for me is that the reliance on older polls that get more than a month old goes to a singular poll. THat singular poll granted gets an * when it gets to more than a month old.
It would seem to me that the average of expired polls would be a bit better as a “second poll” as it averages errors of more than one expired poll (albeit 2 to 3 “even older” polls than the single one “recent poll with an *” relied upon). Once a poll is that old it just seems its elevating its inherent error by using a single poll when there are in fact maybe 3 to 4 polls that are maybe at least useful if not as “recent” or unrecent if you get my point.
Wednesday, May 28th, 2008 at 8:45 am
Tone,
You are correct that the results changed because of the expiration of some old polls, as is frequently the case. If I had had more time yesterday, I would have pointed out what states changed and what effect they had. I also might have pointed out that the single most likely outcome was a win for McCain (the mode with about 11% mass). Alas…there was Drinking Liberally to attend to…
“It would seem to me that the average of expired polls would be a bit better as a “second poll†as it averages errors of more than one expired poll (albeit 2 to 3 “even older†polls than the single one “recent poll with an *†relied upon).”
I considered doing something like this, but decided against using more older polls. Doing so would result in a larger sample size. Additionally, I would need a new rule-set to decide when polls should be combined and when they shouldn’t.
“Once a poll is that old it just seems its elevating its inherent error by using a single poll when there are in fact maybe 3 to 4 polls that are maybe at least useful if not as “recent†or unrecent if you get my point.”
It depends on what error you are talking about. Using the single most recent poll might increase the bias if the poll is no longer representative because of secular change in candidate preference. On the other hand including only one poll keeps the sampling error pretty high.
As you point out, the old poll phenomenon will be less of an issue as the frequency of polling picks up. And, as I have pointed out before, states without current polls are probably not as “interesting” (i.e. results are not close to 50-50, minimal candidate presence, no scandals, etc.) so that nobody is willing to pay to have ‘em polled.
Wednesday, May 28th, 2008 at 12:39 pm
thanks daryl
Wednesday, May 28th, 2008 at 8:32 pm
Do you think maybe Barr and Nader should be included in future analyses?
Wednesday, May 28th, 2008 at 11:12 pm
Brian Boru,
They will be implicitly included if pollsters include them as a category. I’ll not report on the results though. Rationale: Barr and Nader will not likely get any votes in the Electoral College.
They will be implicitly included because, essentially, poll “votes” that go for Barr and Nader will reduce the sample size “voting” for McCain and the Democratic nominee. The mean and sampling error will be affected.