Monday, November 1, 2004

previous entry | main | next entry | TrackBack (3)

A question for polling geeks

This Josh Marshall post raises a question that's been bugging me for the last 48 hours:

The final FOX news poll -- with calls on Saturday and Sunday only --has Kerry over Bush 48% to 46% among likely voters. Among registered voters it's Kerry 47%, Bush 45%. Among those who've already voted, it's Kerry 48%, Bush 43%. (emphasis added)

Here's a more in-depth story by Dana Blanton on Fox's results, which notes, "about one in five voters report they have already voted by early or absentee ballot, and these voters break for Sen. Kerry by 48 percent to 43 percent." I can't find that figure anywhere in Fox's .pdf report of the results, but there it is.

Here's my question -- this confirms other reports I've heard saying that the early vote favors Kerry [But see the update to this post below--ed.] So what does this mean for the election? There are three possibilities:

1) As in 2000, the polls for Election Day are missing turnout and early voting, and so the final vote tally will mirror the early voting and Kerry will win handily;

2) The early voters are disproprtionately likely to vote for Kerry so they have no bearing on the final outcome;

3) The polling of early voters relies on too small a sample and should be ignored.

Most cognoscenti seem to assume (2). My question is, why? The one argument that makes sense to me is that early voting is a sign of intensity of preferences, and the ABB vote is more intense than the ABK vote.

UPDATE: Stop the presses! CBS News also has early voting results -- but they have Bush beating Kerry!

Early voters split about evenly, one-third each between Democrats, Republicans and Independents. They are a bit older: one-quarter are 65 or over, and eight in ten are above age 45. President Bush holds a lead among them (51% to 43%). (emphasis added)

Let's take a moment to allow the heads of those obsessed with media bias to explode at the thought that FOX has a poll favorable to Kerry while CBS has one favorable to Bush.

However, the large contrast between the CBS and FOX results lead me to think that the answer to my original question is actually (3).

One final question -- the Fox result has 9% of voters voting for someone other than Bush or Kerry, and the CBS result has 6% of voters doing that. Who else are they voting for besides Nader?

ANOTHER UPDATE: Over at Tapped, Garance Franke-Ruta has early voting numbers for Florida (a third of the vote cast; 51 to 43 Kerry) and Iowa (a quarter of the vote cast; 52 to 41 Kerry). However, Franke-Ruta seems to buy hypothesis (2) -- early voters are more likely to go for Kerry. Link via Kevin Drum, who offers a hypothesis on why this might be true: "memories of Florida combined with news of Republican efforts to suppress voting have probably motivated Kerry voters to vote early in greater numbers than Bush voters due to their distrust of the voting process."

posted by Dan on 11.01.04 at 03:12 PM


The cognoscenti are right.

Polls published before election day don't separate on their samples among those who voter early and those who vote on election day.

If anything, the fact that more people have voted early bodes well for Bush. Why? Because at the time people cast their ballots early, the polls favoured Bush more than they presumably will on election day.

Therefore, Bush gets more votes than in that cluster of voters than what he ll get on election day.

That of course pressuposes that the pollster models are accurate, that is, they haven't underestimated turnout. If they have, their likely voters are probably tilting Republican, so the numbers coming out of early voters bode well for Kerry.

posted by: Nick Kaufman on 11.01.04 at 03:12 PM [permalink]

"If anything, the fact that more people have voted early bodes well for Bush. Why? Because at the time people cast their ballots early, the polls favoured Bush more than they presumably will on election day.

Therefore, Bush gets more votes than in that cluster of voters than what he ll get on election day."

Which might be interesting except for the fact that the polls show Kerry up among early voters. So, if Bush is getting more votes among early voters than during election day, he's in big trouble.

And, every early vote for Kerry is one fewer person that needs to be GOTV on election day.

posted by: Aaron on 11.01.04 at 03:12 PM [permalink]


-Kerry voters, being disproportionately government workers/college students/unemployed have a lot of time on their hands.
-Bush supporters are stupider and cant understand the intricasies of voting absentee.. as opposed to complicated things like the butterfly ballot.
-Democrats have had it pounded in their heads for the last 4+ years that Karl Rove is hiding in their voting booth waiting to disenfranchise them with extreme prejudice.
-Democrats tend to be nonconformists. Voting when the man tells you to is for squares.
-Republicans, being fascists at heart, enjoy standing in dreary lines at the whim of the government.
-The Dead, understanding the value of time, vote early.
-Those old ladies manning the polls give my dreadlocks dirty looks.
-Halliburton is monitoring the junior high school libraries.
-Need to be good and liquored up to punch the Kerry bubble, easier to do at home. Cant they just have an 'Anyone but Bush' bubble?

Real answer: Anger. Cant wait to get rid of Bush.

posted by: Mark Buehner on 11.01.04 at 03:12 PM [permalink]

You misunderstood me, I think (and know I support the Dems).

The point is this. If the polls don't sample out early voters, then the fact that most early voters have gone for Kerry simply means that the particular segment of the population tilts heavily towards Kerry. Unless the pollsters have underestimated turnout(which is likely and in which case, Kerry is favored), Bush will make up for that margin on election day.

The point then is that when that sample of voters voted, the overall Bush numbers were better than on election day. Therefore, there's a good chance that even though Bush gets trounced by Kerry among early voters, he's getting a higher percentage among that particular set of voters than he would, if those voters voted on election day with the rest of the country.

posted by: Nick Kaufman on 11.01.04 at 03:12 PM [permalink]

I have a good theory on this. I think that in general, Bush supporters see voting early and by absentee ballot as somewhat sacrilegious and are more likely to abstain from using these methods for moral reasons. This is reflected in the “vote fraud” v. “vote suppression” argument that has been going on. Democrats favor maximum turnout with vote fraud as a side effect. Republicans tend to think that the act of voting is worthy of respect and should not be tarnished by double voting, felons voting, etc. Republicans see absentee ballots and early voting as easy ways to commit fraud, and so they will not use them. Democrats see them as useful tools to get out the vote, and are therefore more likely to use them.

posted by: PaulNoonan on 11.01.04 at 03:12 PM [permalink]

"I think that in general, Bush supporters see voting early and by absentee ballot as somewhat sacrilegious and are more likely to abstain from using these methods for moral reasons."

My republican relatives all voted early. I don't buy this argument for a second. And to imply that one sanctioned method for voting is morally superior to another is repugnant.

posted by: dcrall on 11.01.04 at 03:12 PM [permalink]

Historically, early voters have always been disproportionately Republican. If these reports are true, it is an ominous omen for the BC'04 campaign. It could mean that rumors are true. Overwhelming new Kerry voters.

posted by: manyoso on 11.01.04 at 03:12 PM [permalink]

We voted early because we wanted to avoid the possibility of a mixup at the polls (we have moved), and because a conservative talk radio host was promoting that idea as a way to avoid someone else showing up and "stealing" your vote at the polls (in Wisconsin there is lots of talk of fraud).

As for "other than Bush or Kerry" (or Nader), write in votes are a way of registering a protest against the choices of both major parties. And I sense an unprecedented amount of the sentiment: "this is the choice the process has presented us with --- get serious! You other guys can choose between these two; I want neither and my antipathy to both is equal albeit for different reasons."

posted by: Terry Ott on 11.01.04 at 03:12 PM [permalink]

"Who else are they voting for besides Nader?"

Well, if some clown called me at home over the weekend and asked who i voted for, my answer would probably be Calvin Coolidge.

posted by: Mark Buehner on 11.01.04 at 03:12 PM [permalink]

If I'm not mistaken, the same FOX poll also showed Kerry up among men, but down among women, which makes me very suspicious about the quality of this particular sample. On FOX News Sunday, Bill Kristol pointed that out, and jokingly wondered if FOX shouldn't be praised for releasing the survey anyway.

posted by: Gabriel Pentelie on 11.01.04 at 03:12 PM [permalink]

The other possibility is that the early vote is more prone to vote fraud. (See Ohio)

posted by: Bithead on 11.01.04 at 03:12 PM [permalink]

To dcrall:

Why is it repugnant to suggest that one method for voting is morally superior to another? Is it better to vote on Florida's punch card butterfly ballot, or an optical scanning machine? There is an efficiency argument here too, but by and large, the reason people hate the punch card system is that it tends to undercount votes, which is not a good thing. Likewise, absentee ballots and early voting make committing vote fraud easy. I am from Milwaukee WI. It would have been simple for me to vote there by absentee ballot and at the same time vote in person with no fear of getting caught. Plus check this out:

And I didn't say all republicans abhor early voting, I just think that more of them do. I have no problem with your family members all voting early, I’m just trying to explain this phenomenon (which, according to Dan’s update might not exist anyway).

To manyosos:

That may be historically true, but the recent phenomenon of diligent “get out the vote” drives by MTV and the like has shifted early voting and absentee voting to the democratic side. Many groups advertise the convenience of absentee ballots, especially as a draw to younger and elderly voters.

posted by: Paul.Noonan on 11.01.04 at 03:12 PM [permalink]

Simple explanation for why Fox's pre-electionday poll of early voters shows Kerry ahead and CBS's shows Bush ahead... there's still plenty of time for those fates to be reversed. CBS is a partisan hack network watched mainly by Democrats that wants Kerry to win; Dems will see Kerry is down and turn out to vote tomorrow to put him ahead. Same story for Fox, just an R affiliation instead. Seems pretty obvious to me.

posted by: Tom on 11.01.04 at 03:12 PM [permalink]

In a lot of states, early voting is new, so your last guess, #3, is correct. The only thing where it could be useful is in assessing how much vote occurred before a late-breaking October Surprise influenced the rest.

Both sides are motivated, so I view it as meaningless. After two or three elections, they might be able to extrapolate something more useful from it perhaps. But for now, I'd watch 'likely voters' and wait for the verdict on other new variables like new registrants and mobile phone users.

As I've noted, the range of possibilities remains too close to call.

posted by: Kevin Hayden on 11.01.04 at 03:12 PM [permalink]

Two conservatives at work have early voted, and they wrote in John McCain. Don't ask me why, exactly. I like McCain, but wouldn't consider writing him in as my choice this year.

I also know people who will be voting for Badnarik. (I live in Texas, and we're sure that the state will go Bush. There's a bit of freedom to "take a stand" and cast a fun vote.)

posted by: ugly baby on 11.01.04 at 03:12 PM [permalink]

Who are people voting for other than Bush, Kerry, Nader?

I know at least 3 who are voting for Badnarik (Libertarian).

posted by: Sam on 11.01.04 at 03:12 PM [permalink]

If I am not mistaken, the CBS results are from averaging all the early voting states together. That would not include any deep blue states, like Vermont or Rhode Island, but some deep red states, like Texas. Thus, the average is skewed towards Bush.

I've seen early results in states like Florida and Iowa where Kerry is ahead by at least 10 points. That's why I am a little excited.

posted by: Brian on 11.01.04 at 03:12 PM [permalink]

I think voting other than at the polls on election day or by absentee ballot accompanied by a statement as to why one cannot be at the polls on election day is just sick and wrong. And I'm a Republican, so if I'm representative of Republicans Bush will poll better among election day voters than among the early voters.

But two early posters here are right, the answer is behind Door No. 3. Rules for early voting vary from state to state, and I doubt very much that pollsters have been able to control for that variable. Polls in individual states (i.e. polls of voters who are all playing under the same set of rules) may be more accurate. I don't think the national polls are helpful at all by this stage.

posted by: Zathras on 11.01.04 at 03:12 PM [permalink]

Allegedly, Univ of Miami exit-polling shows early voting in Miami-Dade going 2-1 to Kerry. This is a county that Gore won, but not by a large margin. Either Kerry voters are disproportionally turning out early in massive numbers or the election is turning his way.

posted by: Zaoem on 11.01.04 at 03:12 PM [permalink]

I want a recount of these exit polls.

posted by: Mark Buehner on 11.01.04 at 03:12 PM [permalink]

The Des Moines Register reports that 27% of Iowans have already voted, that Kerry has captured 52% of those voters to Bush's 41% -- and that, among the remaining 73% of Iowans, Bush and Kerry are tied.

So, clearly the Democrats are thumping the GOP in the effort to get Iowans to vote early. The question is whether both sides' efforts to raise voter turnout (or, in many states, Democratic efforts to encourage voters vs. GOP attempts to suppress them) will turn in the same performance on Election Day as they have before then.

I suspect the GOP will lift their game somewhat on the big day, but the Democrats will continue to outperform: Of the remaining Iowans, the ones who are evenly split, more Kerry voters than Bush voters will actually make it to the polls.

The GOP might have better luck in Nevada or other states where the early Democratic turnout isn't so dominating (Kerry is up by about 1-2% in Nevada right now, I think), but I suspect the GOP has already lost Iowa, Florida and Ohio -- and, with Ohio, the election.

posted by: Scott Forbes on 11.01.04 at 03:12 PM [permalink]

I honestly don't believe that 7% of Iowa voters voted for neither Bush nor Kerry. My guess is that a big chunk of that 7% just didn't want to answer the exit pollers, so they made up something.

posted by: Bernard Yomtov on 11.01.04 at 03:12 PM [permalink]

Is it just a question of which states offer early voting? A cursory glance at this list and it seems like it's more common in red states and swing states than the blue strongholds (especially the Northeast), but I didn't check carefully.

posted by: Katherine on 11.01.04 at 03:12 PM [permalink]

The extra numbers probably refused to say who they voted for. Not uncommon.

Red states outnumber blues by a lot in early voting. There's California and Oregon for the blue states, and swing states are Florida, Iowa, Wisconsin, New Mexico and Nevada. But the rest are red states. No Illinois, New York or any New England states (except Vermont). In 2000 Bush won early voting by 12 points. So if current early voting is Bush up 8 or Kerry up 6 than Bush is in big trouble. The better indicator is the swing states. Reports from Florida are huge lines and great enthusiasm in Democratic areas. Maybe the Republicans will match it tomorrow but it isn't there yet.

posted by: Elrod on 11.01.04 at 03:12 PM [permalink]

Post a Comment:


Email Address:



Remember your info?