Gallup's Dirty Secret

by Chris Bowers, Fri Sep 17, 2004 at 07:28:16 PM EST

Kerry leads Bush by ten points among independents who are registered to vote, 51-41. And yes, that was in the poll they released yesterday.

I obtained this information from Steve Soto, who obtained it directly from Gallup. Steve also has the breakdown of Gallup's poll sample by Party ID (read the entire article, it is excellent), which is horribly, horribly weighted in favor of Republicans.

By re-weighting Gallup's Party ID internals (Bush 95-4 among Rs, Kerry 85-11 among Ds, Kerry 51-41 among I/O's) according to 2000 Party ID turnout (40.3 D, 36.5 R, 23.2 I/O), the race is tied at 48.

Folks, we have an even election.




Tags: General 2008 (all tags)

Display:
Permalink :: 21 Comments :: Post a Comment


Kerry's Iraq Position over time: CONSISTENT! (none / 0)

Interesting, I'm going to look into these numbers with more specificity. On another note:

Here is the text of Kerry's speech from the Senate floor about two years ago (October 9, 2002) where he states his position on Iraq before the war began:

http://www.independentsforkerry.org/uploads/media/kerry-iraq.html

Note especially the sections highlighted in blue.  Here is Kerry's position today:

http://www.johnkerry.com/issues/national_security/iraq.html

Did Kerry "flip-flop" in his position on Iraq?  No.  In fact, he has been completely consistent.  His position is the same now as it was then.  He stated the conditions under which he would not support the President's military actions.  Those conditions have been met - the President failed to take Kerry's advice, and because of that, we are sinking $200 billion into Iraq and 1,000 of our troops have died.  Kerry opposes Bush's approach in Iraq, but he said he would oppose it before Bush began the war, and Kerry has the same plan now that he had before the war.  Kerry has been completely consistency.  What am I missing?!

If you want to know Kerry's positions, you have to read original documents.

by drplaud on Fri Sep 17, 2004 at 06:40:45 PM EST

Re: Gallup's Dirty Secret (none / 0)

Love your optimism, but it is probably fair to say that the party ID number is at least partially tied to candidate preference. Certainly not to the tune of 13 points, but a little. Clearly, Bush is leading by a couple points. And with the debates coming and Kerry really finding his anti-Bush voice, I'm in a very good emotional place this weekend.

Go get yourself into a canvass or phone bank and finish this off. Start tomorrow!

by AS on Fri Sep 17, 2004 at 07:12:40 PM EST

That's fine with me (none / 0)

I'm happy to go into the election assuming that Kerry's down a couple of points.  It will motivate Kerry voters to get out there.
by jonweasel on Fri Sep 17, 2004 at 08:11:33 PM EST
[ Parent ]

Re: Gallup's Dirty Secret (none / 0)

There is no way in hell that Bush is going to win this election while losing among Indy's by 10%

by Jerome Armstrong on Fri Sep 17, 2004 at 07:19:19 PM EST

I agree completely (none / 0)

If it's true that all of the reason Bush is up at all right now is that the Dem base is a tad soft, and Kerry is doing extremely well among independents, then Kerry is in excellent shape.  It's always easy to shore up your base; I would imagine that Kerry's recent hard-hitting attacks on Bush are going a long ways towards that sort of thing.
by jonweasel on Fri Sep 17, 2004 at 08:13:35 PM EST
[ Parent ]

Vote Map (none / 0)

So what's up with the Electoral Vote Map in the left column?  They don't say, that I can find, where they get their data or how current it is.  According to the map Kerry has lost about 100 electoral votes in the last week or so.  What gives?
by Alan S on Fri Sep 17, 2004 at 07:20:25 PM EST

Down one everywhere (none / 0)

He's down one in CO, MN, OH and PA. Some weird new NJ poll showed Bush ahead there. That account for the entire gap.
by Chris Bowers on Fri Sep 17, 2004 at 08:20:58 PM EST
[ Parent ]

Re: Down one everywhere (none / 0)

I think a couple of other recent NJ polls have shown Kerry's numbers there weakening.

I'm wondering if McGreevy might be dragging down Kerry's numbers there?

I would dearly love to know what exactly is going on with New Jersey. Unfortunately unless someone like Zogby does a poll there we may not know.

Do any of the polls that have been showing a close race in NJ have internals availible?

by ces on Sat Sep 18, 2004 at 12:22:00 PM EST
[ Parent ]

Re: Gallup's Dirty Secret (none / 0)

That's nice, but now CBS/NYT has Bush up by 9.  They can't ALL be wrong.

And citing Zogby isn't very reassuring - Pollkatz has him as the most pro-Kerry pollster.

by BigModerate on Fri Sep 17, 2004 at 08:32:55 PM EST

Re: Gallup's Dirty Secret (none / 0)

Actually, they can be. According to the poll internals, 33%R to 31%D to 29%I. More revealing however, is the Bush v. Gore numbers: 36% voted for Bush, 28% for Gore. Relative to the 2000 election, those numbers seem very off. Also, in this poll 32% of those polled did not vote in the previous election. According to 2000 Exit polls, 13% of voters had not voted in the previous election.

Although I am not sure I did it correctly, I attempted to adjust only the Bush v Gore Numbers, and came out with a 48.5% Bush, 46% Kerry result.

Levy's Law-The truth is always more interesting than your preconception of what it might be.
by bigring55t on Fri Sep 17, 2004 at 11:12:24 PM EST
[ Parent ]

Re: Gallup's Dirty Secret (none / 0)

Hell yea they can all be wrong, look at the two most recent elections in large Democratic countries, and even here in 2000. Conservatives were supposed to win in Canada and India-- they got stomped. Bush was supposed to win by the consensus of a little over 2%, and they were wrong by 3%.

The only problem with these type of polls is that they snowball their own ignorance. Look at Fineman:

Kerry's strategy all along has been to windsurf himself into the position of being the unavoidable choice: not beloved, he's never had any illusions about that, but the Necessary Man. If voters see the war in Iraq as an unforgivable mistake--if they see overwhelming evidence of that on TV--then Kerry still has a chance.

In the meantime, his fate depends on the polls. If the consensus of them doesn't start moving in his direction--very soon--his defeat will become a self-fulfilling prophecy (no one wants to vote for a loser) and the civil war already rumbling beneath the surface of the Democratic Party will burst into the open. He got good news from one poll (Pew) but very bad news from the one that counts the most, Gallup, which now shows him 13 points behind the president among likely voters. (Most of the closer polls are among registered voters.)

Well, the "one that matters most" shows Kerry leading by 10% among the voters that matter most.
by Jerome Armstrong on Sat Sep 18, 2004 at 01:58:11 AM EST
[ Parent ]

Re: Gallup's Dirty Secret (none / 0)

That "polls will determine the outcome" theme is one of the most ridiculous things I have ever heard.  If that was true, no one would have voted for Michael Dukakis, and his poll numbers would have continued dropping.  But his poll numbers increased from mid-October (39%) to just before the election (44%) to the actual election (46%).

There is absolutely no evidence that polls have any impact on the actual votes cast in an election.

by PonyFan on Sat Sep 18, 2004 at 05:10:57 AM EST
[ Parent ]

gallup/nyt (none / 0)

right now it`s  obvious that everyone is just poll crazy, a new poll every day. it`s pretty obviou is that bush is up 5-8 points right now but the interesting thing is kerry is up 14 points in the battleground states where it really counts.i hate to say it but the only poll that counts is on nov2nd, but if these pollsters tell us it`s over it may depress turnout on both sides.
by EXBUSHVOTER on Fri Sep 17, 2004 at 08:39:22 PM EST

Re: gallup/nyt (none / 0)

Where did you get your info about Kerry being up in the battle
states?
by gail on Fri Sep 17, 2004 at 09:55:39 PM EST
[ Parent ]

Bush is not up by 5-8 points... (none / 0)

...it's a tie (or quite possibly a slight Kerry lead).

Everybody poll that doesn't oversample Republicans shows a tie (within one point on either side).  AND, if you adjust the numbers in the fucked up polls to the proper D/R/I numbers, they even show a tie.

PLUS, every poll that included Sept. 15th (Rosh Hashanah) undersampled Jews (who wouldn't answer the phone and/or not be home that day), which, in 2000, only voted 19% Republican.

by Geotpf on Fri Sep 17, 2004 at 10:50:54 PM EST
[ Parent ]

Gallup Talking Points (none / 0)

The media don't care about anything but a good story.

Give them one!!!

by Ddeele on Fri Sep 17, 2004 at 10:24:31 PM EST

Re: Gallup Talking Points (none / 0)

Oops.

Use this link intead

by Ddeele on Fri Sep 17, 2004 at 10:26:37 PM EST
[ Parent ]

Can't beat one flawed poll with another (none / 0)

While the Gallup poll can't withstand serious scrutiny, neither can the practice of using of reweighting the party ID internals with 2000 Party ID turnout. First, as a pollster noted on this site recently, party ID is NOT a constant.  More importantly, though, the turnout by party from one election to the next is serially independent -- there's no necessary connection. (Just think of turnout in 1994 compared to turnout in 1998.) If the turnout this year turns out to be similar to 2000 in terms of party ID, then that will be pure coincidence. You can't build into a model the assumption that it will. For instance, go back and try predicting the 2000 election based on party ID internals that year but after having reweighted them according to the turnout by party ID in 1996. And then try the same thing for 1996 by using the turnout by party in 1992. Your results won't be much better than the polls we're excoriating.
by smartliberal on Fri Sep 17, 2004 at 11:03:13 PM EST

Nah... (none / 0)

...party ID is NOT a constant.  However, it's CLOSE to a constant-it doesn't move that far that fast.  Dems (in the real world) always have an advantage over Reps in party ID-although actually getting them to the polls is another matter.  HOWEVER, every indication I've seen is that Dems are fired up more than ever, with so many new voter registering that many states have huge backlogs processing the forms.

by Geotpf on Fri Sep 17, 2004 at 11:07:44 PM EST
[ Parent ]

Re: Gallup's Dirty Secret (none / 0)

In light of this information from Gallup about their likely voter percentages (D/R/I), which are far removed from recent election numbers, what's next is to ask Time and Newsweek what they use to determine their likely voter party-ID percentages.  Because they have been showing the same 10-13% leads for Bush.

Actually a similar examination of these forecast turnout ratios from all the polls would be quite enlightening.  If we can refigure the Gallup numbers by using a more realistic turnout, we could produce new numbers for any other poll.

by Elwood Blues on Sat Sep 18, 2004 at 02:04:15 AM EST

Is it the Excitement Filter? (none / 0)

As I understand it, Gallup asks if people are excited about the election, and uses this (not alone, but as a major filter) to identify "likely voters."

Now people who expect their candidate to win are more likely to say they're excited. And people who expect their candidate to lose are less likely to say they're excited. And people who see major media coverage of their candidate are more likely to say they are excited.

So the excitement filter builds in three biases: It increases volatility (media events affect the filter), it favors percieved frontrunners, and it favors incumbents (who get so much more media coverage). But, as we see, real elections show little volatility, no edge from perception, and a loss from incumbency.

by Left for the Left on Sat Sep 18, 2004 at 07:49:25 AM EST


You are not logged in.

In order to post a comment, you must be logged in. If you have a member account, please log in to comment.

If not, you can make an account right here. It's quick and free.

Login
Username:

Password: