Many of you have seen our new election simulator. This tool takes current polling in each state, turns the results into probabilities, and then runs simulated elections based on those probabilities.
Until the middle of last week, the simulator was behaving quite nicely (especially for a site like ours that will get more traffic if the race is competitive) — it was about 50-50 for Obama and McCain. However, toward the end of the week, Quinnipiac released polls that gave Obama a substantial lead in three pivotal swing states (PA, FL, OH) and other swing state polling moved toward Obama as well, including significant leads in IA and WI.
As a result of the above, if the election were held today (and if the polls are accurate…always a big if)…Obama would win. This can also be seen in the national polls that have come out over the weekend, with Obama ahead from 5 to as much as 15 points. (While the popular vote ultimately doesn’t matter, if a candidate wins by that much on Election Day, he isn’t going to lose the Electoral College vote).
The dilemma, of course, is that the election is not today and the current polling likely reflects a temporary post-nomination bump. We discuss that in another post.
So, it is important to note that the simulator doesn’t say Obama will win the election, it merely reflects what would likely happen based on current polls — if the election was today. Since Obama is currently polling well ahead in important swing states and there aren’t that many states currently polling inside 5%, this leads to a tighter range of outcomes and the near 100% victory for Obama in the simulations.
13 responses so far ↓
1 Joe // Jun 24, 2008 at 3:18 pm
About the simulator. Can you make it so it spits out state results based on what time the state’s polls closed in 2004 instead of randomly? it would make the simulator a little more realistic that way.
270toWin: We actually thought about doing it that way, but it was complicated enough just to get the statistics right! If we have some free time between now and Election Day, we’ll see what we can do.
2 Max // Jun 25, 2008 at 1:08 am
They last time I checked the gallop poll, he had only a 2% lead over McCain. Not to say that your information is incorrect, but I would assume that if what you said were true the lead would be higher. I also wouldn’t put those states in the win column for him yet. You may be right, and if so I’ll buy the first round, but I think the supporters of Clinton who have/will cross the line will make it a McCain victory come November. By the way, I love your map.
270toWin: Since we’re running an electoral college map site, nothing would be better for us than a close race. That said, the current polling is definitely favoring Obama.. he’s up double digits in two polls out this week. That kind of an advantage is most likely a temporary post-nominating bump, however. I would expect the race to get closer, and you may very well be right about the Clinton supporters. President Clinton’s statement of support for Obama yesterday was tepid, at best.
3 The Dude // Jun 25, 2008 at 4:13 pm
Max: I’m not even sure this IS his nomination bump. You could make the argument that Clinton supporters canceled that bump out. Rather I see this as the “Im running against a sell out, bump”. With unity bumps to come as feelings settle down between Clintonites and Obamaians.
As time goes by the correct choice is going to become more and more obvious. Be it McCain continuing to sell out to Bush’s Minions, selling out his Maverick past, future senior moments or just plain bad campaign management & strategy, the choice will be clear.
I think its funny that McCain comes out for off shore drilling the same day the Eron Loophole came to light. It just goes to show you how out of touch he really is.
I’ve said it over and over, this is gonna be a blow out. I look forward post ignorant/tyrant leadership, which is what McCain and Bush offer.
4 Joe // Jun 25, 2008 at 4:55 pm
Hey it’s me again.
I’ve noticed something after the seemingly hundreds of simulations I’ve run. I’ve seen Obama win only once in Virginia. One time. I’ve seen him win in many more simulations in unlikely states such as Montana, Kansas, Georgia, Arkansas, Nebraska, South Dakota, Mississippi, Louisiana, West Virginia, and even a couple of times in Indiana. Now he might be doing a little well in Indiana polling wise, but not in the other states I just listed. However, recent polls have had Obama winning Virginia by a small margin, or at least losing by a small margin. Surely I would have seen him win Virginia in more simulations by now.
Maybe the probabilities are a bit off?
5 JD // Jun 25, 2008 at 10:47 pm
I have run the simulation over 100 times and have yet to see a McCain victory, there was 1 tie. I’ve also had more results where Obama gets at least 350 electoral votes than I’ve had with McCain getting 260. Those results make it hard to take this simulation too serious. Even with Obama clearly having the advantage in pools now, odds are that McCain would edge out a few victories in that many simulations.
6 dd // Jun 26, 2008 at 2:42 pm
update polls for North Dakota and Mississippi they are not going to obama. a lot of obama electoral votes are barely dem like Ohio Pennslyvannia Oregon Virgina and Conn. they can switch quickly to repulican like Missouri did. mccain has a better chance still when you consider that Obama almost always polls better than he does
7 Peter // Jun 26, 2008 at 8:38 pm
Hey, with the simulator, I get some perposterous looking results because (I presume) each state is independent. I don’t think independence is a viable assumption here. If Obama doesn’t win Virginia, it’s near impossible for him to win N. Carolina. If you were going to do this, the best thing to do would be calculating potential margins in the closest swing states first, then apply these to the landslide states. Pretty much, the closer the swing results to the mean polling average, the less variance is allowed in ‘leaner’ states. So if Obama won Ohio by 1/2 of 1 pt, the North Carolina bell curve would become really narrow on the Obama side of the mean polling average, a 10 point victory might cause the st.dev. on the obama side to be increased.
8 Rob // Jun 26, 2008 at 11:39 pm
Obama is polling ahead in IN and ND. Time to move them to to the ‘in play’ group on the election map. Here in Kansas, he but the spread in half.
9 Jimbo // Jun 29, 2008 at 8:11 pm
I ran the simulator numerous, numerous times today and it did come up with one McCain win. The simple fact is that Obama can win in several instances with or without Michigan, with or without Ohio, w.o.w. Florida, w.o.w. etc. etc. If Obama can win just two of the threesome Nevada, Montana, New Mexico, that makes him very hard to beat. If he can win OH and PN, that makes him hard to beat. I’m admittedly a strong Obama supporter, but I think the simulator is realistic. Winning combinations for McCain have now decreased dramatically.
10 another jason // Jul 1, 2008 at 2:18 pm
You need to use Rasmussen polls only for your simulator. They have been, are, and will be more accurate than either Quinnipiac or SurveyUSA. Both of these polls are heavily biased in favor of Dem in certain states.
The reason I think Rasmussen is more accurate is it might have something to do with a guy’s name and reputation being on the line.
In states which aren’t in play (here, I’m thinking Oregon), I’ve seen polls from the two above which show a suspiciously close race.
It is as if partisans are in charge of them and by massaging sampling are baiting McCain to spend money somewhere he doesn’t have a chance in.
11 RF // Jul 4, 2008 at 1:44 am
JD, I’ve run it perhaps 75-100 times, too, and I’ve seen McCain win - I think - three times. I’ve also seen the dreaded 269-269 come up twice.
It’s going to be closer than the simulator is currently showing. McCain will play the same card Bush did in 2004, which is the “terror threat level” card , and unfortunately people will swallow it. However, I just read today about a poll that says fears of a terrorist attack are at their lowest since 9/11, so who knows if the fear-mongering will be successful enough to tip it for McCain? I predict Obama will win with 304 to McCain’s 234.
12 Ruairí // Jul 6, 2008 at 9:52 am
I agree with another jason (July 1st 2:18 pm) about Quinnipiac but SurveyUSA are reckoned to be among the most accurate pollsters of all. Check out this pollster rating by Nate Silver on his great site http://www.fivethirtyeight.com/search/label/pollster%20ratings
13 Andrew // Jul 7, 2008 at 7:48 pm
Absolutely agree with Jimbo here, if you check out pollster.com and compare swing states it confirms the Simulators results. Out of about 9 swing states McCain has to win 6-7, but the polls show Obama leading (beyond margin-of-error) in 5 and in the margin of error (in other words, tied) in 2 more.
Just remember what 270towin says: we’re 4 months away from the election with a lot of “ifs” involved.
Leave a Comment