« Scripps Survey Research Center: Kerry Leads by 5% | Main | Take Two Articles and Call Me in the Morning »

Kerry Up by 1 in WP/ABC Tracking Poll

Today's WP/ABC tracking poll has Kerry up by 1 among RVs. (I explain my preference for the WP RV data in this post.). But the LV data are not much different: a dead heat.

Both results differ substantially from today's Zogby tracking poll result, which had Bush up by 4 and occasioned considerable rending of garments in Democratic circles.

What accounts for the differences between the two polls? Certainly not their party ID distributions. Zogby, as is well-known, weights to the 2000 exit poll distribution, which was +4D. WP, on the other hand, does not (though, as I explain in the post cited above, they may use some modified party-weighting scheme, at least on their LV data). Based on data WP releases on vote by partisanship, their LV sample appears to be currently running at about +2-3D on party ID (no comparable data are released on RVs, so no comparable estimate can be made). That means the differences between the two polls, controlling for party ID distribution, are actually slightly larger than they seem on the surface--if you weighted the WP LV sample to the Zogby/exit poll distribution, Kerry would actually be ahead of Bush by a point, rather than tied.

So: what does account for the differences between the two polls? Try this one on for size: Zogby has Kerry winning Democrats by just 64 points (79-15), while Bush wins Republicans by 83 points (89-6)--almost a 20 point gap in partisan support margins. In contrast, in the WP LV poll, Kerry wins Democrats by 73 points (85-12), while Bush wins Republicans by 80 points (90-10)--a much more modest 7 point gap in partisan support margins (commensurate, incidentally, with the gap we saw in 2000 between partisan support margins).

In short, the primary driver of the differences between the WP and Zogby tracking polls is the unusually low margin for Kerry among Democrats in the Zogby poll. I personally don't find that margin particularly plausible, since it is contradicted by the WP poll, by numerous other recent polls and by the 2000 exit poll results.

Lesson: don't let any particular tracking poll result get you down and, to the extent possible, look at a poll's internals to see what might be driving the result.

A note on Nader: A front-page story in The New York Times today (and see also Ryan Lizza's piece in the new TNR) on the threat Nader poses to Kerry no doubt caused some additional garment-rending in Democratic circles. And there's no doubt he does pose some kind of threat--pretty much by definition even a very small Nader vote could tip a state if that state were close enough.

But the smaller Nader's vote, the less likely such an outcome will occur. And Nader's vote is likely to be quite small indeed. Currently, he is running at 1-2 percent in the national polls and most relevant state polls as well. Looking at analagous polls from the analagous time period in the 2000 campaign, he was running at 3-5 percent of the vote.

Take Zogby's tracking poll as an example. In 2000, Nader was polling 5 percent at this time in October. This year in the same poll, he's getting 1 percent support, very close to the support being received by Badnarik, the Libertarian candidate.

So: worry some about Nader. But not too much. I'd worry much, much more about turnout and mobilization.

Comments

Okay, I did freak out at Zogby today, but I'm chilled now...

What do you think the odds are that those polled purposefully lie about what their party preference is? If it is known ahead of time that the pollster weights the survey based on party preference, I can tip the survey in favor of Bush if I state that I am a Democrat but am voting for Bush. The converse is also true. Is there any scuttlebutt out there that operatives are encouraging their followers to misstate their party affiliation when polled?

Frankly, I'm not worried about the Zogby numbers. There's no reason why Bush's numbers ought to have gone up, plus he's still under 50%. If he can't break 50 in at least a few polls, he's in trouble, since late deciders always go for the challenging party.

As for Nader, I can't see him getting much more than 1.5%, maybe not even that much. He was polling 4% in 2000 and only got 2.7%, and it's reasonable to expect that some of his supporters will defect when they see that there's a real possibility of Bush remaining president. Almost half of Nader's 2000 supporters defected at the last minute, so why not again?

If Nader gets 1-2% of the vote, how many of those people would vote for Kerry even if Nader didn't exist? If Nader voters are otherwise non-voters (or Socialist Worker Party voters), it doesn't matter.

This issue is up my alley, because I am a progressive and not a DLC Democrat at all. Many people I know supported Nader in 2000 -- but somehow managed in NY to support Hillary against Lazio. (Kind of odd to be FOR a Clinton Democrat in NY against an honorable Republican, but not nationally against a self-evident turkey. I supported both Democrats, because, when the chips are down, the national Republican Party has NOTHING to offer positively either in Congress or the White House, while the Democrats are at least the lesser of two evils -- on many issues no different but other significantly so.)

On Nader -- He's polling around 1%, and Andrew Kohut says his numbers are going down and down steadily. In swing states it should even be lower. But in one region -- the upper midwest (Iowa, Minn, and Wisc) he appears to have a substantial base, more likely to tip a state there all other things being equal than anywhere else. A concerted effort by the Michael Moore progressives for Kerry in that region, figuring out where the pockets of Nader voters are and with 527's buying up media time in those less than superpricey markets with progressive voices like Chomsky should be able to make a real dent right in that area. A strategic approach to the problem would be helpful.
Nader's alliance with sleazy corporate Republicans in the election (openly) should be flaunted in front of these voters, then have trusted progressives who support Kerry make the case for Kerry Edwards to the Left RATHER THAN wasting the time and concern of the candidates themselves.
---------------
For some voters, really getting down to brass tacks about unemployment and the strategic deficit as a lever against future Social Security, as well as the draft should win over both mainstream AND progressive voters. These are areas that gain broader support in BOTH directions at the mass level rather than sacrificing the swing voters to win progressives.
-------------
I'm much more concerned about the fallout from the 'nuisance' article -- which is what it was. I even read a poster at this site whose views I highly respect praising the picture of Kerry in the article.
Its distortions were crafted to be subtle, but they were distortions and not merely differences of interpretation. And they were systematic and deliberate, which is probably why Kerry was (by the reporter's own account) so suspicious of him. He was clearly fishing for materiel to use against Kerry and has done the job. Kerry MUST respond more effectively, both to the distorted article and then the second layer of distortion from the Dick Morrises of the world.
It is a case of highbrow demagogy -- more dangerous as you can't always recognize it, but having spent 15 years of my life in and around Harvard and Berkeley (degrees from each) I do know how to smell it a mile off. Slicker than the SwiftBoat smear but no less slimy.

I've long thought that the Dems had benefitted from not having a someone run on their fringe, unlike Republicans, who have been losing votes to the Libertarians and other right-wing fringe parties for years. The Dems will just have to get used to dealing with a crankish fringe siphoning off the occasional percentage point on their left.

At this point, anybody who is actually going to cast a vote for Nader is someone who, absent Nader, probably wouldn't vote for Kerry anyway - they'd vote for some other third party (what's Peace & Freedom doing this year?), write in "Donald Duck", or just not vote at all (and sneer about how "all those parties are owned by the same corporate fascists, maaaaan.") Those votes aren't really "lost" to the Democrats. The Dems should just plow forward and concentrate on registering and getting the votes of people in critical battleground areas.

That doesn't mean the Dems should make it easy for Nader - they should work to stymie his candidacy at every level, if only to hone their election law skills. Hell, they could try a bit of old Clinton-style triangulation and run against Nader and his crazier ideas, to burnish their moderate appeal.

Based on what I've seen in all the polling thus far, I believe we are looking at an electoral landslide for Kerry on Nov. 2.

First of all, incumbents do not win re-election with approval ratings in the 40s, with 56% of voters consistently saying the country is on the "wrong track" and when they are presiding over a wobbly, if not failing, domestic and foreign policy (see Jimmy Carter).

Second, numerous pollsters including Ruy have said that within the last 3 weeks of a campaign, the only number that counts is not the spread but the INCUMBENT's percentage. In all of the polls taken over the last week, Bush is at between 44% and 48%, meaning that is almost certain to be the MAXIMUM he will get of the popular vote. Not a good sign for the president.


Third, and perhaps most important, I am convinced that Democrats and Kerry voters are being DRAMATICALLY undersampled in every one of these polls. The most clear evidence of this are the state polls in New Jersey that show Bush within 5 points or less. Now, Gore won New Jersey by 15 POINTS in 2000, and there are actually more registered Dems there now. And nothing in the dynamic of New Jersey has changed to explain a 10-point swing from 2000 (I don't buy the 'terrorism has touched them' argument because that is not showing up in the NYC polls). The only plausible explanation is that a substantial number of Democrats and Kerry voters are not being reached by the pollsters in New Jersey -- and that trend obviously would hold true in every other state. Therefore I would wager that any state in which Bush is not at least 5 points ahead on Election Day he will lose.

Anyone have any thoughts on these observations?

i agree about not being concerned about tracking poll changes. although i think of zogby as being one of the best pollsters, i don't think things change that fast unless something big happens. just a few days ago kerry was up by 3 or 4 in zogby's poll.

Difference of 4 with MOE +/-3? I don't think you
really needed to actually explain the difference, did
you?

Although I tend to buy your argument about weighting
by party, I don't see how you can just reweight the
way you do in all your posts here. If party ID is correlated
with other demographic categories, then won't
merely reweighting make the other demographics wrong,
given that the samples have *already* been reweighted
to fix those demographics?

For example, if men are more likely to be GOP, then
reweighting a survey with too many GOP would also
make the survey too female. That doesn't seem like
an improvement to me. Obviously, including
party ID as part of the reweighting matrix (where the
relationship between party ID and age/sex/income
is taken into account) is reasonable, but that's not
what you've done here (unless you have all this
data, which I didn't think was so easily obtainable).

According to these two news stories just out, Kerry appears to be surging in Florida. After several weeks of being behind by 3-5 points, two polls out Friday show he is tied in one and 4 points AHEAD in the other. Here are the links:

http://www.bocaratonnews.com/index.php?src=news&category=Local%20News&prid=9844

http://www.washingtonpost.com/wp-dyn/articles/A36823-2004Oct15.html

Mark,

Just to play devil's advocate -- someone I know suggested that NJ might be a lot tighter than expected this year due to a backlash against the Dems from the McGreevey scandal.

Mark -

I have the same take on all this as you do.

I just read the Big FIVE-OH a few days ago, about the maximum for the incumbent between the last poll and the election day numbers, and I think it is completely valid that that last poll % is his election day %, too.

I have been saying for several weeks (months, really) that if Bush is not leading by MORE than five in a state he is toast. My reason has to do with the massive turnout that is coming - and will blow the pollsters out of the water. The 200 election was boring, boring, boring, and the turnout of 50.8% reflects that. I can't see the turnout being less that 56% (I am expecting 60%), and I also cannot see how Bush can get more than a 20% share of the increase in vote totals. I have so far only heard of a 1 million increase in GOP voters.

Anybody else that has better figures on that?

SO many people stayed home in 2000 because they were bored or because they bought into the Rove lie that Bush was the same as Gore anyway, so what difference does it make?

The rest of the increase will go to Kerry, who did not bore us - even though the press seems to think he does. Hatred for Bush is rampant, and I don't think the media and pollsters have a clue.

I agree that this is all going to come down to turnout. I have another question, though.

Again this morning, Zogby shows Bush ahead 48/44. The explanation is that more Republicans are voting for Bush than Dems are voting Kerry (by a pretty large margin, I think; 14% of Dems are polling for Bush, 5% Repubs for Kerry). I'm just finding this sort of odd, and wonder if you have any idea of what gives with this.

Addy

to whoever played devils advocate on NJ,

This is the Bush feint into blue territory. Remember the California feint in 2000? That's what these stories are.

As for McGreevy, he's such a local politician that not even any New Jersyites link him with the national party. I mean he was just an obscure state legislator before 2001. Besides which, governors can;t really do much for their national party. Republican gov. Ernie Fletcher is very unpopular in KY, but Bush will still win there. Ditto on NJ.