A discussion of Gallup's use of "likely voters". Pseudo-science?

[Karmadillo]

http://www.usatoday.com/printedition/news/20040830/a_likely30.art.htm

<edit>

The leading theory is that Kerry does worse among likely voters because the polling questions screen out some young and anti-Bush voters who didn't cast ballots in the last election.

Mark Mellman, pollster for the Kerry-Edwards campaign, says it's wrong to focus more on likely voters than on registered (among whom Kerry fares better).

“I have tremendous respect for Gallup. But this far out from an election, determining who are the likely voters is pseudo-science at best,” Mellman says.

“From day-to-day there are substantial variations in who's going to vote. You don't really know for sure until right before the election.”

USA TODAY has emphasized the likely-voter results, but other news organizations agree with Mellman that it's too soon to rely on likely voters. The Los Angeles Times on Thursday released its latest presidential poll and based its findings on registered voters.

more...

[blondeatlast]

1. I've been wondering how newly-reg'd voters fit into the mix.

Some of the newly-reg'd are older, too, and have never voted in their lives. They are so po'd at the chump in charge that they are finally ready to vote.

Do they get counted? I sincerely doubt it.

[AmerDem]

5. Simple...they don't

This is Gallup. Everyone knows Gallup's idea of "likely voter" polls. They are a scam all the way thru.

This whole debate stinks to high hell and i'm sure Kerry's people are going to go right along with it.

I have writing about this here for a few weeks now, very little interest from others.

[LuminousX]

2. Dewey Defeats Truman

A headline that became famous because of bad polling techniques. I think a similar effect if going on here.

Back then telephone polling was kind of new because telephones were kind of new. Only the well off had telephones, and thus only the well off got polled. The well off wanted Dewey.

Now we have a situation where everyone has a phone, but the young and political active are more likely to have a cell phone or not be at home to take landline calls. Thus, the polling is overlooking this segment of society and it is common knowledge the younger voters are anti-Bush.

[AmerDem]

3. I don't see why Kerry's campaign refuses

to be part of this hack set up. hell, shrub has yet to agree to excepting any debates at all. It boggles the mind that they would let this become an issue instead of saying the whole likely voter set up is a farce. If they agree to these condition. especially with Gallup of all groups behind it, they deserve all the backfire that will result.

[bobthedrummer]

4. I'd apply critical questions about face validity, meself.

O8)

[On the Road]

6. The Concept of Likely Voters is a Good Idea

because not all people vote in the same proportion. Usually it's a better indicator than registered voters, especially if they show both numbers. That way you can get an idea of the results with a large turnout.

In many elections in the past, new voters broke along the same lines as the rest of the population, so it would not make a big difference. Not this time.

The "likely voter" models should include a segment for new voters. No reason they couldn't do this. I'd like to see how it affects things. It should be somewhere between 5 and 10% of the total -- that can easily swing the results in a close election like this.

[barbaraann]

7. My mom worked for Gallup doing door-to-door surveys.

She says that she would always be sent to homes in one part of the income spectrum and that THAT is how they can manipulate results. I don't know if they do this with telephone polls, but that might be how they get the results the Bushes want.

[moondust]

10. I've often wondered about that.

If you simply call people in a predetermined area code with certain "desirable" prefixes, it would seem you could engineer whatever results you want. And thus guide public perception using polls. In the interest of saving some credibility you might try to be accurate on the day before the election.

I don't trust them.

[Gloria]

12. I fielded studies for Gallup when they were doing the door to door

omnibus surveys...lots of interviewers were going door to door across all types of neighborhoods and income levels....(I actually was an "Assiatant Study Director" so I worked with field and coding as part of my job of ultimately writing the reports).

This was when the founder was still roaming the halls and Gallup was independent. We only had about 6 phones for telephone polling back around 1980!!! The organization was very much 'academic' in orientation and Andy Kohut now of Pew was one of the alternating Presidents.

Gallup was sold out years ago to a firm out of Nebraska or somesuch...their Princeton office houses about 3 people and perpetuates the myth of Gallup.

[quaoar]

8. They also ask questions

Designed to screen likely voters. For instance, they might ask, "Do you know where your polling place is?" Someone who doesn't know might be checked off as unlikely to vote.

[xcmt]

9. Living proof?

I'll admit I'm one of the "unlikely" voters. I was of voting age in 2000, and I did not register or vote. It was a mistake, I know, and I've changed that. In 2000 I just wasn't interested in politics, all the pomp and circumstance and empty policy. Of course, the four years since have opened my eyes a little.

It bothers me now that, if Gallup called and wanted my input, they'd disregard me as an unlikely voter, even though I'm now registered and plan 100% to cross the street and vote Kerry this election.

It also eliminates everybody currently aged 18-21 from their polls, unless they've implemented some kind of control to include a sliver of such individuals in their polls, which I haven't heard about. These people are probably overwhelmingly Kerry.

[democraticinsurgent]

11. 3-7% have dumped landlines since 2000

this to me is the key stat. these people are likely significantly different than the rest of the electorate. they're younger, more hi-tech, less traditional, and to a degree more upscale.

big question is whether they are "likely voters". If they are likely voters in the same proportion and the rest of the electorate, and they are as different as i think they are, polling as we know it is going to fail us this year. this is scary when considering the diebold factor.

how will we know what really happened? exit polling? maybe, but what are the big networks doing on this in the wake of the VNS dissolution after 2000?

zogby is on to something with his internet polling, though it's too early to bank on it.

i think all the polls are wrong to a significant degree. i just don't know in what direction...

questions, questions...

;-)

[Gloria]

13. This topic came up yesterday...it must have been Staph's show on ABC

and I know Russert has discussed it. They're basing everything on 2000 data. I've heard Russert say that if there's a really big turnout, then the applecart is upset. George Will on ABC said that if the turnout is bigger than expected then they'll break for Kerry and Bush loses. I surmise that turnout will come from voters missed in polling.

Awhile back I heard a discussion on polling as part of a larger forum on CSPAN. It was Democratic election analysts. Skolnick (sp?) spoke about this whole issue, about how "likely voters" becomes a valid universe in the fall, but not months out (this was back a couple of months). He since has joined the Kerry campaign...within the last couple of weeks. What he mostly stressed in his talk was how it isn't until the fall that people start really paying attention and at that point, closer to the election,
"likely voters" really are more likely to actually become actual voters, thus a better indicator than "registered voters."

Skolnick specificallly mentioned Gallup, saying that they've used that universe since the dawn of time and simply stick to it for continuity at this point. He lumps this polling in with all the other "likely voter" scenarios--useless until the fall.

He also stressed the approval nos. as key indicators.......