« PRC Poll: Kerry Up 6 in Battleground, Tied Nation-wide | Main | Marist Poll: WH Race Dead Heat »

Harris Poll: Kerry Ahead by 7 in Swing States, Down 2 Nation-wide

George Bush leads John Kerry 48-46 percent of nation-wide LV's in a Harris Poll conducted 10/14-17 (MoE +/-3%). In 17 "swing states" the poll found Kerry leading 51-44 percent.

Comments

My summary of the WSJ-Zogby results ( http://online.wsj.com/public/us ):

- Considering only those states won solidly by either candidate plus those battleground states that are outside the margin of error gives Kerry 243 Electoral Votes to Bush's 189 EVs. (Remember 270 EVs to win.)
- This means Kerry needs another 27 EVs. Ignoring Florida (since Dems can't expect much there) this means Ohio (20 EVs) plus 7. A win in Missouri (11) or Iowa (7) or 2 smaller states would do it. My favorite would be a steal of all 9 of Colorado's EVs since the state was not originally considered a swing but is now within 1-2% w/ Bush below 50%.
- If all leaners in the poll are considered (including those within the margin of error) it gives Bush a win with 274 EVs to Kerry's 264. This means it's going to be a GOTV battle all the way - As if anyone didn't already know that.

Ray, I know this blog is dedicated to polls, but I wanted to update you on what in happening in Dallas.

After two days of early voting and including the absentee votes, the number of ballots cast in Dallas County has already exceeded 25% of the early vote and 10% of the total vote from 2000.

That would (very unscientifically) suggest a projected increase in voter turnout of about 18%. So it looks like the the voter turnout in Dallas County might be as high as 70%.

From my observation, both parties have energized the voters but I have never seen this level of Democratic activism.

I have no idea how this squares with the polling models but thought you might like the information.

When you say "Dems can't expect much" in Florida, are you just assuming that cheating and intimidation will swing the state for Bush? Because from what I've seen, very recent poll numbers are rapidly closing there, and it may be a more likely Kerry win than Ohio.

Some very interesting internals from the Harris poll.

Of their 820 successful interviews, 65 people met all these criteria:

- They say they are registered to vote.
- They say they are "absolutely certain" to vote.
- They chose not to vote in 2000, although they were old enough.

All 65 plan to vote for Kerry!


(This is discussed in a MyDD recommended diary entry.)

Based on what I read at Real Clear Politics it looks like Harris' issued the results cited here (Bush +2), but also results that are much more favorable for Bush (+8).

I'm a Kerry supporter, but I'd rather have both analyses given. If the +8 is bs, then it would be very appreciated to hear why (as you've done with the Gallup polls).

To David Montgomery:

To answer your question (Harris poll internals), see globecanvas post just above yours. When Harris limited LV to those "absolutely certain" to vote, the result was Bush +2. When they further limited it to "absolutely certain" AND "voted in 2000," the result was Bush +8. You decide which is bs. What I wonder is what the RV results would have been in this poll.

David M: Harris's two sets of results are (1) including and (2) excluding new voters who chose not to vote in 2000, as described in my comment above.

Including those 65 voters, Harris came up with 48/46 Bush-Kerry. Excluding those 65 voters (all Kerry voters, as noted above) from the results, Harris came up with 51-44 Bush-Kerry.

The 17 swing state average is irrelevant because of the winner take all rule in the electoral college. What matters are states that might actually swing from the Gore column (2000) to the Bush column in 2004, or Bush colum (2000) to the Kerry column (2004).

Reviewing the latest state polls, the only net changes appear to be:
Ohio to Kerry for 20 (still too close to call)
New Hampshire to Kerry for 4
Wisonsin to Bush for 10 (still could change)
Iowa to Bush for a 7 (this could decide the election)

So even if Kerry wins Ohio, he would still lose the election if Bush wins Wisconsin and Iowa. All conjecture, but my real point is watch Wisconsin and Iowa, not Ohio and Florida, and definitely not Pennsylvania. Why Clinton is wasting time in PA is beyond me, but maybe he or Kerry think it's in trouble. If so, Kerry has no chance of winning. Same with Florida and Bush of course.

Wow. If you've not read the Harris report, do. And this will go a long way towards understanding those Gallup results. globecanvass has given the punchline.

If the LV status involves being registered to vote, saying you're absolutely certain to vote, and (if old enough) you voted in 2000, then Bush leads 51-43.

If they drop that third issue (voting behavior in 2000), then Bush leads only 48-46.

In the battlegrounds, under the first criterion, Bush and Kerry are tied at 47%. Under the second criterion, Kerry is up 51-44.

Wow.

Recall that Gallup uses voting behavior in 2000 to determine their LV's. Thus they are ruling out those who didn't vote last time and say they are certain to this time, who, if the Harris poll has it right, are overwhelmingly going towards Kerry.

(a) Props to Harris for listing both definitions. That helps a lot in the interpretation of the data. Would that Gallup would do the same.

(b) I wonder if some of those saying they didn't vote last time actually did vote for Gore and are choosing not to mention that fact. If so, that would help to interpret the fairly consistent report that in these surveys, voters favored Bush over Gore.

it could be the october surprise;

if the press (or Kerry campaign) pursue the pat robertson statement on CNN, the elecotrate would see one evangelical leader is calling the other a lair!

either way, so long to the base!!!!

Posted by David Montgomery at October 21, 2004 02:39 AM
Post a comment
David,
I have not looked at the internals of the Harris poll, but I noted that the results posted are likely voter
samples. We all know the problems with these models. Put more weight into RV samples,
and independent polls versus medial polls. Also
watch the approval numbers, several polls are
showing that number around 44% for Bush. You can expect that to be Bush's number on November 2, 2004.

Not posted is the 8 point lead by Bush if Harris uses its traditional likely voter model. The only way it's a 2 point lead is if you use some new method that counts voters that "say" they are definitely going to vote, but didn't vote last time if they were eligible.

So, if the Harris people are right, the only way it's even close, is if armies of people that haven't bothered to vote in the past, suddenly develop the enthusiasm and minimal amount of self discipline that it takes to actually vote. And, even if every single one of these people, for the first time do show up and vote, he still loses. So why is this good news for Kerry?

As pleased as I am to hear reports of heavy early voting, I'm concerned that many of these people would have voted anyway, probably for Kerry. "I'm going to vote that W. out of office the first chance I get!" Are the high early vote totals indicative of an increase in total voters, or just a displacement forward of the normal voters?

If the latter, we could be in trouble.

while not Good news, I think it is not bad news for Kerry. for following reasons.

At 2 points election is tied. News reports were saying that Bush is moving away from Kerry, not quite accurate. that can lead to lower turnout if people believe its in the bag for Bush.

At 48% support its a bit dangerous for Bush. Its on the cusp not enough to win or lose, an incumbent president in the middle of a 'war', with the power of the presidency behind, should not be polling 48% 12 days before an electio

Bill M: The important result from the Harris poll is that swing states are tied 47/47 by the traditional definition of "likely voter", but Kerry leads 51-44 in those swing states when including voters who say they are "absolutely certain " to vote, despite having opted not to vote in 2000.

Kerry doesn't need "armies" of these newly-enthusiastic voters -- he just needs some of them. And according to Harris, there are a LOT of them (about 9% of their total sample fits this definition).

Bill-

It's good news because it explains the LV discrepancies. And there's a perfectly plausible explanation for why the 2% figure is a lot closer to the truth--There were a fair number of Democrats in 2000 who were convinced that Bush was actually a moderate and had Clinton fatigue, so didn't show at the polls. This time, they won't make the same mistake.

Given that the battleground numbers don't break out the states, they aren't as informative as I'd like them to be, but it is still reassuring to see Kerry at least tied, and, if the above interpretation is correct, well in the lead.

As for states...the four you list (Ohio, Florida, Wisconsin, Iowa) are clearly among the most competitive. But there are others that are also very close. On the Kerry side, there's New Hampshire. On the Bush side, there's New Mexico and that Maine district. I'm keeping a close eye on those. And state results are, on an individual basis, less predictable, so I'm not ruling out Kerry snaring Nevada, Colorado, Virginia (4% Bush lead at the latest SUSA poll and a Democratic governor), North Carolina (3% Bush lead at the latest SUSA poll and the Democratic VP nominee), or West Virginia (5% Bush lead in the latest Mason Dixon). Each is unlikely to go to Kerry, per the state polls at least, but odd things happen. From the Bush side, I'd not rule out Minnesota (2% Kerry lead in the latest Market Shares), Michigan (5% Kerry lead in the latest Research 2000), Pennsylvania (6% Kerry lead in the latest SUSA), or Oregon (6% Kerry lead in the latest Research 2000).

So...I see 7 states (one of which is just one district in Maine) to watch really closely, with 9 others gaining at least some attention.

Your basic conclusion...winning Ohio might not be enough for Kerry seems on target. He also needs to get one or the other of Iowa or Wisconsin, and if he loses Wisconsin, he needs either that Maine district or New Hampshire.

Or he needs Florida and either Iowa, Wisconsin, the Maine district, or New Hampshire.

Or one of the longer shots could come into play for either party.

The Catholic bishops are stepping into politics in a way they never have before. In some parts of the country, parishoners are being told that a vote for Kerry is a sin which they will have to go to confession for. I have not seen any polling on this point, but I assume that such pronouncements from the pulpit would have the effect of swinging Democratic Catholic votes over to Bush, in a way that no one has anticipated because the Church has never gone so far before. Could Kerry lose the election because of this effort by the U.S. bishops?

Zogby on KLSD radio in San Diego right now. Still says it's to close to call, virtual tie (46/45 Bush). Really nothing new.

Notes:

Undecided: 5 %

1/5 - President deserves to be reelected
2/5 - Somebody new
Kerry has just not closed the deal "yet"
Expects 2/1 break against incumbent and that he suspects Kerry will win.

Other polls - Zogby says plenty of new voters that they are capturing. This includes "cell phone only" voters. Kerry has cut into military families significantly.

For some of the "non-voters" from the last election, it must have been the first election in which they were eligible to vote. So here's a question, is there any data on whether people typically do vote in the first election in which they are eligible to vote?

I suspect that if you're, say, 40 years old and haven't voted in the past you are unlikely to vote this time around. But if you are 22 or 23 and didn't vote when you were, say, 18 or 19 it's not necessarily true that you are less likely to vote now.

But is there any data about this? It seems much too strong of a screen to screen out as unlikely everybody who was old enough to vote, but didn't vote.

I'm looking forward to seeing numbers from early voting turnout in other states.

Anyone have any idea what that is looking like?

Austin also is having very high turnout in Texas. Austin tends to vote very Democratic--sort of an anamoly in Texas, but it sounds like Dallas and Houston may be more divided than in the past as well.

Numbers from other states anyone??

>Not posted is the 8 point lead by Bush if Harris uses its traditional likely voter model. The only way it's a 2 point lead is if you use some new method that counts voters that "say" they are definitely going to vote, but didn't vote last time if they were eligible.

Given the large number of new voter registsrations, it doesn't seem appropriate to discount new voters. That's what Harris is saying, and it makes sense. There's no reason to stick with traditional voting models when it's obvious that the electorate this year is considerably more fired up about the election than is usually the case.

Bill M,
Here's my reasoning, FWIW:
1) In 2000, even though nearly all polls showed Bush a couple of points ahead right down to the wire, about half a million more people voted for Gore than for Bush; this seems to mean that all the polls are oversampling Republicans and their top-line results should be discounted.
2) Now if that 65/820 (~8%) figure is at all meaningful, even if only half of them actually drag themselves to the polls it'll amount to several million former non-voters turning out for Kerry, and _no_ corresponding movement from the bench towards Bush. (Where are Rove's 4 million missing evengelicals?) Now, if everyone who voted in 2000 votes the same way this year - a very favorable assumption for Bush, as anecdotal evidence indicates a large number of shifts from Bush to Kerry or to couch-potato status, as well as from Nader to Kerry, and almost none in the opposite directions - and if new adults and new citizens exactly balance out those deceased since 2000 or convicted of felonies in states where that costs them voting privileges - Bush is toast. If he gets _all_ the votes he got in 2000, and Kerry gets all the Gore votes _plus_ a million crossovers from Nader and 3 million newly energized former non-voters, Kerry wins by 4.5 million, and the poultry industry will get a great boost from providing all the egg which will be found on the face of the polling industry.

That Bush approval rating seems to me to be the most telling number out of the lot. If he's getting around 44% approval he's going to get about 44% of the vote. Bush is in real bad shape for a sitting president.

Tony,
Interesting to read your rundown of states and the myriad of potential outcomes. Thanks for pointing out the sleeper district in Maine. Wouldn't we all be surprised if they decided the election.

On ABC the Note, they mention a possibility that would drive everybody crazy:
Bush picks up Wisonsin and New Mexico, while
Kerry picks up Ohio and New Hampshire leaving a tie and throwing the election to the house.

Of course, there's also the "landslide factor" cited by Mickey Kaus, which seems like a reasonable possibility. The theory is that enough voters will want to avoid having lawyers and judges decide the outcome that they'll side with whoever is leading on election day.

P.S. As a resident of the great state of Virginia, I can confidently state that John Kerry has about as much chance of winning Virginia, as the Red Sox have of winning the......never mind.

Another update on early voting in Texas.

Carolyn noted the high turnout in Travis county, this is also being reported in Bexar county (san antonio) and in El Paso county, which Gore won in 2000 with 58% of the vote, early voting is 40% higher than in 2000.

If this is indicative of the country as a whole, it is very good new for Kerry.

2000 Nader Voters?

I haven't seen much in the way of polling on those who voted for Nader in 2000. Here are some totals for battleground states:

NADER 2000

Ohio----------117,000
Colorado------91,000
Iowa-----------29,000
Wisconsin----- 94,000

My guess is that Kerry will receive a 50% boost from these
figures, so from OH add 58,000 votes, Iowa add 14,500 etc.
Does this seem too optimistic? Too conservative?

I have yet to meet anyone who voted for Nader in 2000 who is not planning on voting for Kerry.

riverrat, if the Catholic bishops are as effective in swaying Democrats to vote for Bush as they are in convincing people not to use birth control then Kerry will win by a landslide. Most American Catholics are much more willing to disobey the church hierarchy (except for the mean-mouth conservative Catholics that aren't going to vote Kerry anyway) than you'd think. We Americans hate being told what to do - it's in our culture.

Pdb, it's a reassuring scenario but I think it boils down to Florida and Ohio.

I think if Bush is leading the polls in Ohio and Florida come election day, Kerry is toast, because if the election is close in those states the Republican Secretaries of State will game the count for Bush.

One wonders whether Bush's national numbers are being distorted by big leads in solid red states, in the way Kerry gets a boost from big leads in NY, Cal. How else to understand polls like this?

Riverrat:

Actually, I heard that just the other day the Vatican officially cleared Kerry of heresy charges brought on him by the conservative bishops.

I frankly think this is good news for Kerry. If the president's approval rating is this low, then he's clearly in trouble. No incumbent with an approval rating that low has ever been re-elected.

new AP poll shows Kerry up 49-46 among likely voters. Internals look bad for Bush again (47% approval rating).

http://www.startribune.com/stories/587/5044847.html

Coldeye,
Not to ne naive -- because I know some "gaming" is bound to happen -- but I think the reality is that if BUSH is not ahead in Ohio and Florida by at least 4-5 points on Election Day, BUSH is toast. He needs that big a lead to counteract the last-minute undecideds that break 2-1 or 3-1 for Kerry; and the intense voter turnout among new voters, young voters and 'cellphone-onlys' not captured in the polls.

The Catholic bishop effect likely is in the polling numbers already. The ones I know who are taking the most anti-Kerry stance are in Colorado and Missouri, and I wouldn't be surprised to see them have *some* effect there, though perhaps they'll shift others over to oppose the Republicans because of fear of the religious element taking over the party.

Bill M.

Here are three different semi-plausible ways of getting a tie, if there really were just the 7 states in play (Ohio, Florida, New Hampshire, Wisconsin, Iowa, New Mexico, and the Maine district). I'll list only the changes from 2000, with all other states going as then:

1) Kerry gets Ohio and New Hampshire, Bush Wisconsin and New Mexico.

2) Kerry gets Ohio, Bush gets Wisconsin and that Maine district

3) Kerry gets Florida, Bush gets Wisconsin, Iowa, and the Maine district.

Kerry gets Florida and New Hampshire, with Bush getting Wisconsin, Iowa, and New Mexico.

Note that the main effect of New Hampshire in these scenarios is to overcome any effect of the Maine District. Of course, if others of the states come into play, that could all change.

harvard survey of college students.

62 kerry
36 bush

Well wouldn't a tie be interesting. New House with a Republican majority picks Bush to be pres and new Senate with a Democrat majority pics Edwards to be VP.

2 thoughts:
1) Good news: an additional significance of the comparison between the 2 Harris LV models is the ABSENCE of Bush voters in the "certain to vote but didn't vote 4 years ago" group. This does not augur well for Karl Rove's grand strategy of mobilizing the supposed 4 million evangelicals who abstained last time.
2) I still don't get how large (7-10 point) leads for JK in "battleground state" polls can be reconciled with (i) small GB leads in national polls and (ii) rough parity in polls in most of the individual battleground state polls (esp. Fla. and Oh.).

Hey c'mon, peoples. That Maine district has only 1 EV, right? In NH, here, we have 4.

I've averaged Bush support since Sept. 1 from all of the major polls tracked (likely voter) by the Polling Report. I recognize that the polls all have slightly different data collection methods, likely voter models, etc., and the Polling Report doesn't track every poll, so this isn't perfect. But generally speaking, I think this provides a nice overall guide to the trends in the race.

Overall, since Sept. 1, Bush has received an average of 48.3% of the vote. This should be worrisome enough for the Bush campaign. The total, however, masks the a trend that is decidedly downward:

Sept. 1 to Sept 15 Polls: 49.8%
Sept. 16 to Sept 30 Polls: 48.7%
Oct 1 to Oct 15 Polls: 47.7%
Oct 16 to Oct 20 Polls: 47.0%

With Bush support at 47%, Kerry pulling in about 45% and almost all the rest undecided and historically likely to vote for Kerry, I think Kerry is in excellent position to win.

phatcat-

Thanks for that analysis. Interesting.

Marcel-

I want those 4! They provide some safety margin. It was just in those scenarios I cited that NH mostly takes care of the Maine district. In others, NH might make a different crucial difference.

Various-

Thanks for pointing out the importance of the data for the absence of those evangelicals.

Various/Bill M.

Cook has out their updated electoral scorecard. They have 9 tossup states. Five went to Gore in 2000 (Minnesota, Wisconsin, Iowa, New Mexico, Pennsylvania), 4 to Bush (Ohio, Florida, New Hampshire, Nevada). They have three lean Kerry's in Michigan, Maine, and Oregon. And they have three lean Bush's in Colorado, North Carolina, and West Virginia. OK. North Carolina does now join my list of watch states. They have 15 EV's, which would be huge.