Gallup continues to tighten. (-9/-14 vice 11/17)

[FBaggins]

PRINCETON, NJ -- Republicans remain in position to win control of the House of Representatives in next week's midterm elections, although Democrats are doing slightly better now than they were early in October. Gallup's latest two-week average on its generic ballot for Congress shows Republicans retaining a 48% to 44% margin among all registered voters, a 52% to 43% margin among likely voters in a high-turnout scenario, and a 55% to 41% margin in a low-turnout scenario. These likely voter advantages for the Republicans are slightly smaller than in previous weeks, reflecting in particular increased Democratic strength over the most recent days of interviewing.

An increase in Democratic positioning in the last weeks of the midterm campaign is not historically atypical, nor is it necessarily surprising in light of the campaigning and efforts Democrats are putting forth as they work to narrow the enthusiasm gap Republicans have enjoyed all year.

However, Democrats would need to generate a substantial narrowing of the margin among likely voters over the last week of this campaign in order to prevent Republicans from gaining enough seats to take over control of Congress. Gallup's statistical estimates based on historical U.S. House-vote data by party suggest that the Republicans need a 52% share of the two-party national House vote to be in a position to win control of the House.




http://www.gallup.com/poll/127439/Election-2010-Key-Indicators.aspx

http://www.gallup.com/poll/143996/Republicans-Remain-Control-Race-House.aspx

[ProSense]

1. Do you know if Gallup took this into account

MAJOR FLAW revealed in polling, excludes 1/2 to 1/3 of age groups Obama won 66-32 in 2008!

[FBaggins]

2. Yes. They responded quite effectively

To that as well as concerns over Hispanic assumptions.

[pnwmom]

3. What was the gist of their response? n/t

[FBaggins]

4. That the author didn't know how to read a poll

And essentially made up his own numbers.

http://2010central.gallup.com/2010/10/more-on-gallups-likely-voter-models.html

[ProSense]

5. That's "effectively" explained?

One of the specifics of Alan’s micro-analysis of estimated votes among smaller subgroups focused on “nonblack, nonwhites.” That’s actually a group not represented in the cross-tabular data we typically use and was not in the data provided Alan, but one Alan apparently attempted to identify by performing his own calculations. (Typically a scholar would contact us or inquire about aspects of the data they are unsure of, but I don’t believe we heard from Alan on this one.) In this particular case, we would have told Alan that nonwhites in our usual procedures is a broad, mixed group of respondents, including blacks, Hispanics, Asians, other races, and a significant number of respondents who chose not to identify their race. Alan attempted to make guesses or assumptions about the composition of this group, and made an assumption as a result that Hispanics in the likely voter sample must be too Republican in voting orientation.

In fact, like most pollsters, we typically are cautious and do not report data for subgroups when there is low sample size involved. Hispanics are one of these. Certainly our analyses of broad, aggregated datasets has shown that Hispanic registered voters as a national group skew Democratic, as we have pointed out many times. But Gallup has also shown that Hispanics’ support for President Obama tumbled into the low 50s earlier this year and has only recently recovered some. Further, among likely voters within all subgroups, those most likely to vote this year are disproportionately Republican in their orientation compared to the subgroup as a whole.

In any instance, shifts in the voting estimates of a relatively small segment of voters is not going to change the overall ballot estimate for Republicans and for Democrats by more than a point or two. Of course, shifts in the voting estimates of larger segments of voters -- like whites -- will affect the overall ballot estimates. Alan, in fact, criticizes the report of voting choices of whites in the Sept. 27-Oct. 3 dataset as too Republican. But that’s basically tautological, as I mentioned above, given the large percentage of the sample that is white. Yes, if one thinks that the overall ballot is implausibly Republican, then one is going to think that the ballot among whites is implausibly Republican as well, and vice versa.

Summary: We drop samples we consider too small and whites are the majority.

Gallup leans Republican.

[FBaggins]

8. Yes.

Summary: We drop samples we consider too small and whites are the majority.

That isn't even close to what they're saying. They don't report subgroups that are too small to have a reasonable statistical margin of error. That doesn't mean they aren't included in the totals.

You're making the same error as that thread a few days ago that claimed that a given poll didn't survey anyone in the 18-29 age group (which was flat wrong). It's just that even with a representative sample of that age group, the sub-sample margin-of-error would have been 10%... and no polling firm is going to report crosstabs with such a large MOE.

Gallup leans Republican.

Sigh. Do you even look these things up before you make such wild claims? Republicans have complained about Gallup for years and now, because you don't like the results, they lean Republican? I'm sure that six of the last eight democratic presidential candidates would be surprised to hear that (since only two of them outperformed Gallup's number). The last six mid-term races are pretty much split (though in the only two races where they were off by more than about a point, we underperformed).

[ProSense]

9. "Republicans have complained about Gallup for years and now" So?

Republicans complain about the media and voter fraud.

[FBaggins]

10. "So?" is answered quite well by the part of the post you're conveniently ignoring.

The statistics don't back up your claim that Gallup leans republican. (Which, BTW, is not the same thing as saying that I agree with their current LV assumptions).

But given the fact that a number of well respected long-time democratic analysts (Cook, Silver, etc) are getting the same treatment... I suppose Gallup shouldn't be surprised.

[ProSense]

11. You choose to take Gallup at its word

The flaws pointed out were convincing, and Gallup's response lame.

Charlie Cook has made enough mistakes as well ridiculous comments over the years.

Nate spent almost all of his time prior to going to the NYT criticizing the hell out of all the polls.

Up until Pew made the point about cell phones versus landlines, none of these polling organizations highlighted the impact of this omission.

So don't pretend that they're above criticism. It's not the first time that polling organizations have gotten caught on a flawed bandwagon. They shift their polling methods from time to time, and that leaves room for error.

[FBaggins]

12. vs. the word of someone who doesn't HAVE the data but makes it up anyway? Yes.

It's the ultimate straw man.

The claim about the hispanic sub-sample was entirely made up. There was nothing in the Gallup data to base the claim on (other than unsupported assumptions). It isn't possible for such a claim to be "convincing" unless you're predisposed to wanting ANY explanation for why the results aren't what you want.

Charlie Cook has made enough mistakes as well ridiculous comments over the years.

He's also been right more often then just about anyone else... for several cycles now.

Up until Pew made the point about cell phones versus landlines, none of these polling organizations highlighted the impact of this omission.

Riddle me this. How many of the major polling firms do you think actually omit cell phones from their sample?

Look back in 2006 and 2008 and you can see republican posts that sound just the same way. When the polls are against you, you grasp at any straw to explain why they're wrong. For some people it helps them sleep at night. Maybe it's cell phones (as if we didn't have those two years ago), maybe it's weekend polling (when everyone just knows that republicans are out having picnics and unable to answer)... or it's daytime calling that only picks up those unemployed minorities and not the republicans who have jobs... or it's simply the bias of the "left-wing media" that wants to bias the results of the election.

The story is the same. When your team is down by two touchdowns to your biggest rival, the true-believing fans are certain that the refs are in the tank for the opponents... or the other side is cheating and getting away with it... or the instant replay official didn't have the right angle.

So don't pretend that they're above criticism.

Of course they're not (and I've pointed out at least one of them in the LV modeling)... but that doesn't mean that every straw someone grasps at is actually valid. This is the third time that someone has claimed that one poll or another is leaving out younger voters... and each time they've been flat wrong. Now we also have a claim that they're data re" hispanic support is way off... except that no such data has been reported.

[pnwmom]

6. Thank you. I don't see where the writer addressed the issue

of young people being underrepresented in the sample.

[polichick]

7. Gallup isn't a source I trust. nt