« Hagee and the Jews | Main | The Truth Hurts »

Early Polling

23 May 2008 10:58 am

Brendan Nyhan says that the early state by state polling actually does have some predictive value, but that the important caveat to this is that the "toss-up" states are genuinely toss-ups. Tom Holbrook did an analysis comparing 2004 presidential election outcomes to polling in spring 2004 and found that "across all four months [March-June] the poll result called the wrong winner in 17 of the 36 cases in which Kerry's share of the two-party vote in trial-heat polls was between 47% and 53% (this excludes two cases in which the poll result was tied)."

To me, at the end of the day this essentially reenforces the idea that early polling isn't very valuable. We don't need a poll to tell us that John McCain's going to win Utah and Barack Obama's going to win Florida. We do need a poll to tell us Obama's narrowly ahead in Colorado and McCain's not a slight edge in Nevada but those polls are too close to be predictive. We also know that forecasts based on the fundamentals suggest that Obama will likely win the popular vote.

Share This

Comments (41)

Florida? Surely a slip.

We don't need a poll to tell us that John McCain's going to win Utah and Barack Obama's going to win Florida

Really? You are sure that Obama will win Florida?

Maybe "Florida" was just a misspelling -- maybe Matt was trying to spell "New York" or "Massachusetts".

We don't need a poll to tell us that John McCain's going to win Utah and Barack Obama's going to win Florida.

I like your optimism.

We don't need a poll to tell us that John McCain's going to win Utah and Barack Obama's going to win Florida.

Umm, Barack's behind in the FL polling I've seen.

We don't need a poll to tell us that ...Barack Obama's going to win Florida.

Uh, what?

I appreciate your optimism Matthew. But, Obama isn't going to win Florida, even if an early poll did say that (which I don't think it did). I polled my grandmother this morning, and Obama just doesn't have a shot (even with the enormous margin for error).

theeliotinside.blogspot.com

PROOFREAD, MOTHERFUCKER.

We don't need a poll to tell us that John McCain's going to win Utah and Barack Obama's going to win Florida.

No, we merely need a miracle to tell us either Democrat's going to win Florida, but I'm just being nitpicky.

First it was faze/phase and now this Florida business. Oh, Matt, all the blogging is getting to you. Maybe it's time for a vacation? I hear Little Compton, Rhode Island is a lovely spot this time of year.

Freudian slip, perhaps??

"Barack Obama's going to win Florida"


Haha, what the other 18 commenters said.

Matt, it is understandable that you have a policy of not changing typos - what else would you have time for? But this needs to be fixed. People will quote it as proof that you are an idiot, when you're not.

Ugh! Now I'm ashamed to be one of the many typo whiners. I hate those guys.

If another Dem wins the popular vote but not the electoral college (to make twice in 8 years), then I think this electoral college system is becoming disastrously rigged.

We no more need polls to tell us Utah going Republican then we need a dozen comments saying Florida was a slip up.

This post is better then the one yesterday, where he basically said to ignore these state by state polls. To predict the election results you need to start assigning states, and it’s interesting to know that Obama’s doing better in Colorado then in Ohio. This state by state polls are as good as anything to tell us how to do that.

Obviously its possible Obama could lose Colorado, and win Ohio. But it would be correct to place the odds of an Obama victory higher in Colorado then Ohion based on the state polling.

Matt's ignoring the fact that "the poll result was wrong in only 3 of the 44 cases in which Kerry's poll margin was outside this [6 point] range." But there are a lot of "swing states," where the poll results are outside this range. For example, current polls tell use that Hillary would win Ohio, but it's a tossup for Obama. Similarly, Obama would win Oregon, but it's a tossup for Hillary.

Also, not that Holbrook's analysis is based on only a single poll. Averaging multiple polls, as done by fivethirtyeight.com, for example, ought to do better still.

"....fundamentals suggest that Obama will likely win the popular vote."

Fat lotta good that does a man......

2000 U.S. Presidential Election Results

George Walker Bush
Richard Bruce Cheney
Party: REPUBLICAN
Home State: Bush - TX ; Cheney - WY
Electoral Votes: 271
Popular Votes: 50,456,002 (47.9%)

Albert Arnold Gore, Jr.
Joseph Isador Lieberman
Party: DEMOCRATIC
Home State: Gore - TN ; Lieberman - CT
Electoral Votes: 266
Popular Votes: 50,999,897 (48.4%)

It's not RIGGED. It just sucks. There's a difference.

I assume you meant California, not Florida.

The main fundamental is that over 50% of the country strongly disapproves of Bush. McCain just won't be able to overcome that; I'm amazed that people think he has a chance.

It is entirely possible that BO racks up huge popular vote leads in NY, Cal, and Ill and wins more overall votes as a result, with McCain taking the electoral college and the presidency. You all seem so full of Obamahope that you are discounting the reality that this is going to be a close electoral race.

Obama's mojo doesn't work on me. I'm not sure why any white male with a penis would vote for the guy.

It does not seem that republicans are so deeply convinced of a McCain win. They will not explode if he loses, because he ought to given the circumstances in which he has to run. It probably is the D's turn to run the executive branch for a while.

So, if a liberal democrat can't win this time, do you all agree that you should give up and not play this game anymore?

What would merit serious consideration of eliminating the electoral college is if Obama (or any other candidate, ever) got an actual majority of the votes cast and lost in the electoral college. If it's supposed to be "majority wins", no one actually won the 2000 election.

"I'm not sure why any white male with a penis would vote for the guy."

Who we supposed to vote for? McCain? Barr? Nader? And why? Keeping in mind that many white-males-with-penises believe that (a) the current foreign policy (which McCain supports) is destructive to American interests and (b) a protest vote (i.e., Barr or Nader) is pointless.

If your point is that white-males-with-penises should have voted for Hillary over Obama, nevermind.

But what about all of us white-males-without-penises?

Who should we vote for?

And what about the white-males-with-penises that have functioning brains?

But what about all of us white-males-without-penises?

Who should we vote for?

And what about the white-males-with-penises that have functioning brains?

You all seem so full of Obamahope that you are discounting the reality that this is going to be a close electoral race.

I don't think it's at all clear yet what the "reality" is. Typically, the popular race has to be very close for there to be the possibility for the electoral college winner not to be the popular vote winner.

Obama's mojo doesn't work on me. I'm not sure why any white male with a penis would vote for the guy.

I agree: who needs a president who has a bigger dick than you? This is why I refused to vote for Lyndon Baines Johnson in 1964. Also, I hadn't been born yet.

GOBama!

Obama's mojo doesn't work on me. I'm not sure why any white male with a penis would vote for the guy.

I agree: who needs a president who has a bigger dick than you? This is why I refused to vote for Lyndon Baines Johnson in 1964. Also, I hadn't been born yet.

What would merit serious consideration of eliminating the electoral college is if Obama (or any other candidate, ever) got an actual majority of the votes cast and lost in the electoral college.

Ha. Ha. Ha. Haven't you ever heard anyone argue that if we abolish the EC, the President will "only have to win the people in the cities"? You know what kind of people they're talking about, who live in cities? As the saying goes, it's not a bug, it's a feature. From their point of view, that would be the system working.

Holy shit, we ran guys with middle names Arnold and Isador against guys with middle names Walker and Bruce? No wonder they lost. Sort of.

Fortunately, I'm pretty confident that Hussein will kick Sidney's pampered ass.

I´m still prediciting an Obama landslide and thus think the study of 2004 is irrelevant. How about 1988 when Bush I won handily - wasn´t he behind Dukakis by 18 points or so in the spring of 1988? How about 1980, 1984, 1972 and 5 or 10 other election years? Comparisons of one year´s (2004) elections to it´s polls are statistically meaningless anyway

According to his basketball buddies, Obama's white genes are reflected in both his penis and jumping ability.

Polling this year, early or late, needs to be read differently.

Registrars all over the map are reporting increased voter registration and predicting significant increases (10-15%) in turnout in November. Many of these new registrants are young or minority, less likely to have landline phones or longtime fixed addresses, so less likely to be included in poll samples or rated as "likely" voters.

In the past, a poll with a point spread of +/- 4% was considered a statistical tie. This year, I would say +/- 8% to account for the volatility of the voter pool.

"According to his basketball buddies, Obama's white genes are reflected in both his penis and jumping ability."


Obama's basketball buddies can trusted to evaluate Obama's jumping ability, but not the size of his penis.

We'll need to ask Michelle about the latter.

"Obama's basketball buddies can trusted to evaluate Obama's jumping ability, but not the size of his penis."

You've obviously never been blindsided by a stiff back-pick.

We may not need a poll in Utah to tell us that Utah is McCain territory, but what about Virginia, Indiana, North Dakota, Nebraka's first district and Iowa? All but the last of those, you might initially think are uncompetitive for Obama -- and yet, they're close -- and the last, if 2004 were any use, we might imagine would be a swing state or McCain state -- it appears to be neither.

The point you're missing is that Hillary wins Obama's toss-up states by commanding margins. So we know Hillary would win; we don't know the same of Obama.

"But what about all of us white-males-without-penises?

Who should we vote for?

And what about the white-males-with-penises that have functioning brains?

Posted by peep | May 23, 2008 12:39 PM"

Maybe he only meant white males with multiple penises. I guess they have to all vote McCain, otherwise those extra penises just mean they like penises a little too much.

Obama will win Florida.

NOTE: Also important to remember that there's a big difference between the circumstances of June 2004, and June 2008. Kerry had the nomination wrapped up by March, and was well into general election campaign mode. Obama has been taking hits from the left (clinton) and the right (mccain and repubs) for months now. I'd say a study of accuracy of early polling in 2004 and this year's general election polls is like comparing apples and oranges (well, maybe not that different, but definitely like comparing grapefruit and clementines).

Perhaps the polls didn't reflect the results because Diebold's chairman really did deliver on his promise.


Comments closed June 06, 2008.

Copyright © 2008 by The Atlantic Monthly Group. All rights reserved.