Weekend Update: General Election Simulations
It's time for another summary of the Votemaster's state-by-state polling data. I've been doing this weekly for quite a while now, running a simple Monte Carlo simulation of the general election using the most recent polls from each state (averaged with any other polls taken within 7 days of the most recent).
I repeat that this is not a prediction of what will happen in November; at best it's a summary of current state-by-state polling (and some of the most recent polls are quite old). "Mystery Pollster" Mark Blumenthal does a good job outlining the many grains of salt with which one should take such simulations.
Here's this week's results:
Obama wins 82.0%, averages 283.3 EV
McCain wins 13.8%, averages 254.7 EV
Electoral tie 4.1%
Clinton wins 100%, averages 334.1 EV
McCain wins 0%, averages 203.9 EV
In terms of how often Democrats win simulated matchups, there's virtually no change from last week, although Clinton's average electoral vote total went up, and McCain's chances of beating Obama dropped, with the change mostly going to an increased chance of a 269-269 tie.
The big gain for Obama this week is taking the lead in Wisconsin, and he's also widened his lead in Iowa beyond the polling margin of error. Balancing that, most recent polls in Michigan show Obama dropping from 1 point back to 4, and a week-long Quinnipiac poll in Ohio showed Obama trailing McCain by 4, whereas the most recent Survey USA poll showed Obama ahead by 8. Also notable (but not yet affecting the simulation) is a poll from Mississippi showing Obama down by just 6.
Clinton took a 9 point lead in Kentucky (compared to a 12 point deficit before), showing a huge post-primary bounce, and she's also now leading in Nevada by 5 (the prior poll, from April, had McCain up by 11). She is also polling better in Mississippi, now trailing by 10 instead of 20, although it's premature to consider that state "in-play". Travis Childers's recent win is providing "reverse coattails", boosting the Democrats' standing in the state.
I still believe Obama would be at least as strong a general election candidate as Clinton. I would attribute his trailing now in current polling to the nature of the campaign right now: both Clinton and McCain are fighting Obama, whereas Obama is fighting only McCain right now. This suggests Clinton is basically not being attacked right now by anyone, and that would clearly change if she wins the Democratic nomination.
Both Democrats are currently in good shape, but the winning percentages especially can be quite volatile. Clinton's map is the more traditional Democratic map, and she's currently winning previous key states Ohio, Florida, and Pennsylvania. If she holds her lead in North Carolina, she could afford to lose two of the other big 3 and still win. By contrast, Obama is also ahead despite trailing by 10 in Florida, and leading by just 1 in Ohio. He can plausibly win despite losing both states.
I repeat that this is not a prediction of what will happen in November; at best it's a summary of current state-by-state polling (and some of the most recent polls are quite old). "Mystery Pollster" Mark Blumenthal does a good job outlining the many grains of salt with which one should take such simulations.
Here's this week's results:
Obama wins 82.0%, averages 283.3 EV
McCain wins 13.8%, averages 254.7 EV
Electoral tie 4.1%
Clinton wins 100%, averages 334.1 EV
McCain wins 0%, averages 203.9 EV
In terms of how often Democrats win simulated matchups, there's virtually no change from last week, although Clinton's average electoral vote total went up, and McCain's chances of beating Obama dropped, with the change mostly going to an increased chance of a 269-269 tie.
The big gain for Obama this week is taking the lead in Wisconsin, and he's also widened his lead in Iowa beyond the polling margin of error. Balancing that, most recent polls in Michigan show Obama dropping from 1 point back to 4, and a week-long Quinnipiac poll in Ohio showed Obama trailing McCain by 4, whereas the most recent Survey USA poll showed Obama ahead by 8. Also notable (but not yet affecting the simulation) is a poll from Mississippi showing Obama down by just 6.
Clinton took a 9 point lead in Kentucky (compared to a 12 point deficit before), showing a huge post-primary bounce, and she's also now leading in Nevada by 5 (the prior poll, from April, had McCain up by 11). She is also polling better in Mississippi, now trailing by 10 instead of 20, although it's premature to consider that state "in-play". Travis Childers's recent win is providing "reverse coattails", boosting the Democrats' standing in the state.
I still believe Obama would be at least as strong a general election candidate as Clinton. I would attribute his trailing now in current polling to the nature of the campaign right now: both Clinton and McCain are fighting Obama, whereas Obama is fighting only McCain right now. This suggests Clinton is basically not being attacked right now by anyone, and that would clearly change if she wins the Democratic nomination.
Both Democrats are currently in good shape, but the winning percentages especially can be quite volatile. Clinton's map is the more traditional Democratic map, and she's currently winning previous key states Ohio, Florida, and Pennsylvania. If she holds her lead in North Carolina, she could afford to lose two of the other big 3 and still win. By contrast, Obama is also ahead despite trailing by 10 in Florida, and leading by just 1 in Ohio. He can plausibly win despite losing both states.
Advertisement
Very cool! What a nifty way to aggregate all the polling data.
'commeneded!
May 31, 2008 2:57 PM | Reply | Permalink
Thanks once again, Fossberry. And thanks from this disgruntled erstwhile E-V supporter for noting that many polls are quite old.
As we've said before, you build your projections based on the polls you've got, not on the polls you wish you could get. Or something like that.
Perhaps new polling numbers fall under the category of "known unknowns".
May 31, 2008 3:24 PM | Reply | Permalink
You're welcome. But I "projections" is probably too strong a word, because since I'm using 4% as the margin of error for the polls, the model doesn't allow for anything other than errors in sampling by pollsters. So the possibility of changing opinions is not reflected in these percentages, otherwise the winning percentages of the Democrats would be much lower.
May 31, 2008 4:12 PM | Reply | Permalink
Thanks cute white dog. I'm very pleased to see Obama's numbers improving in these match ups, because that bodes well for Nov.
I find it heart breaking that Hillary is winning even more convincingly, and her numbers are stronger than his and improving as well, and yet the CW is that she has lost.
I really believe dems ignore her strength at their peril.
May 31, 2008 4:28 PM | Reply | Permalink
I find it heart breaking that you can't see the strength it shows to fight two candidates at once and still lead against the unopposed candidate.
May 31, 2008 5:48 PM | Reply | Permalink
When I was in college I remember reading a little book called "How McGovern Won the Presidency", which explored flaws of polling in the context of the 1968 election. I suspect this year will repeat the complications of 40 years ago.
There are two reasons I think this year is going to shatter the expectations of number-crunching. First is the makeup of the Republican voters.
Over the past generation, the Republican party has managed to assemble three different groups under one party banner: religious conservatives, fiscal conservatives, and social conservatives. Republican candidates have succeeded by convincing each of these groups that "I'm just like you".
Although the specific arguments for a candidate having a similar viewpoint may not have been entirely convincing, in most cases were accompanied by some evidence - or at least a sufficient absence of evidence to the contrary - that they actually shared the views of the targeted voters.
In this go-around, I see McCain's strategy as rather flawed. If we were looking at a contest between the 2008 Obama and the 2000 McCain, I'd agree with your conclusions, but McCain's currently stated positions are in many ways contrary to the positions he held 8 years ago. I think this will provide fuel for sowing doubt among those who have been following the party line reflexively.
I think McCain's positions are weaker than Obama's for holding the fiscal conservatives, roughly the same on core issues (except for Iraq) for the social conservatives, and wobbly for the religious conservatives. Remember, this is the guy who said Falwell was an agent of intolerance, and compared him to Louis Farrakhan.
The second complication is on the other side if the fence. I'm not sure that Obama's numbers will yield what the early polls say they will. Some voters will vote on the basis of race as a deciding factor, and I'll wager that a good chunk of those folks aren't sharing that bias with pollsters.
Given McCain's weaknesses with the R base, I don't think past history is as strong a measure this time 'round. There's also a lot of Bush backlash sentiment across the country - just look at the recent Congressional elections. Obama is not a slam dunk, but in many ways he's a thought provoker for those (most?) of the Republican voters who don't entirely share McCain's position.
On the other hand, Hillary is a lightning rod for many of these folks. If she prevails, those red states are far more likely to remain red. If Obama is the nominee, I think we may have a big re-shading of the map, even if just for one election cycle.
May 31, 2008 7:34 PM | Reply | Permalink
I should make clear the above numbers aren't my predictions, they're the results of simulations using the most recent data I can get. I agree with much of your comment about the problems McCain may have holding his coalition, and that Obama may well reshape the electoral map.
I think there's certainly much reason for optimism, but I should hope Obama supporters don't get complacent. McCain has strength in much of the country, and he could win. Against either Democrat, I should add.
May 31, 2008 8:03 PM | Reply | Permalink