Thursday, September 28, 2006

New Poll Lows

Now, apparently, people can just make up poll numbers from thin air and pass them off.

A New York Times story claims an Emily's list representative that says that Giffords has more than a twenty point lead on Graf.

I searched their site and Giffords site for any mention of this "poll" but couldn't find anything. Tedski seems to have got a release of the poll. But I would assume that he is on the mailing list.

If you cannot even place the actual results of the poll in the public domain, I am sure that the internals aren't going to be forthcoming. I can only assume that everyone involved is aware that this is a cartoon version of a poll. However, if you knew the poll was a lie, why would you release it?

Maybe so you can blame the actual election results on Diebold at a later date :)


sirocco said...

FWIW, I got a small summary of the poll as well. The provided details are:

Polled 400 likely voters, polling done Sep. 19-Sep. 21. MOE 4.9%.

Giffords leads Graf 54% to 29%. (Quick 3%, Nolan 2%, Undecided 13%). Yeah ... that's 101%, likely due to rounding ...

Graf is known by 80% of voters. 23% say they have a favorable opinion of him, 42% have an unfavorable opinion.

Giffords is known by 78% of voters. 45% have a favorable opinion of her, 17% unfavorable.

No information whatsoever on internals.

Even I, a strong Giffords supporter, filed this one under "wishful thinking".

sirocco said...


Speaking of the recent polls, did you see my post in the below thread ("GOP Coming") where I managed to find some numbers and do the math on others for the Zimmerman poll?

Based on those figures, I don't think women were over-sampled, nor were rural areas under-sampled, but I would be curious for your take.

John said...

I could not find much detail on the poll either except this on

I hope Gabby's camp keeps it up so her supporters get so confident that they don’t bother voting.

Brendan Greeley said...

Hi, I'm a producer for Open Source, a public radio show broadcast nationally out of Boston. We're trying to get a handle on the AZ 8th for a show this coming Monday. This blog was recommended by Rene Gutel at KJZZ in Phoenix; any chance I could talk to an author on the phone? You can find me at brendan radioopensource org


Framer said...


Sorry I haven't gotten back sooner. Not a lot going on locally to follow, so I have spent some with the family.

My contention was that women were oversampled, which they were, unless someone can demonstrate that women actually vote more than men in numbers greater than 4% in this district. Maybe not a huge difference, but not a negligable one either.

My other contention is that the sample size of the outlying counties was not large enough to determine that Gabby has a lead in the outlying counties. I most definately stand by that. if we use your 100 responants number, this is simply not a large enough sample size to determine anything with accuracy over 3 counties. You may be able to argue that the sample was sufficient for AZ 8 as a whole, but you cannot with a straight face make the micro analysis that the outlying counties are trending towards Giffords. Your data is simply not complete enough to do this.

And again, I am very leary of the fact that a party breakdown wasn't given. They did that previously and it is hugely important. It very much appears that they are hiding something.

The word that I get is that the Weekly poll has been done since Tuesday and that the candidates will get a sneak peek tomorrow. Not sure why the delay in getting it out, and I am definately surprised there haven't been leaks.

x4mr said...

Sirocco and Framer,

If it isn't already clear, respect both you guys, but I like your word "microanalysis" and to me it just occurs as early.

My take, FWIW, is that of the group of people that are definitely no kidding voting, we have a solid camp for Giffords, probably 30-40%, that will take extraordinary event to change. Graf also has a solid set of folks, probably a little smaller, say 25-35%, also needing something incredible to change.

Then there's the rest, and Darth Rove scheme if any, upcoming debate feast including two televised, and final two week ad blitzes containing footage, good or bad, of candidate performances, will do this thing.

Eager to see TW poll, and frankly expect a tightening.

I remain very curious, and frankly have no clue, what kind of turnout we are going to get this time around. Who decides to actually take the time come day of truth might surprise us.

Not sure about balance between mad as hell and gonna do something and mad as hell so F it all and stay home.

sirocco said...


No worries -- hope youhad a great time. Do you plan to pursue the radio show invite?

Anyhow, addressing your points:

1. Women make up 51% of the eligible voters in CD 8, so at worst they are over-sampled by 3% ... but ...

I can't find any data for CD 8 specifically. However, in looking just at AZ, in the 2004 primary elections, woman made of 59% of the voters. In the general, they made up 54%(CNN is the source for these figures).

If you are polling "likely" voters (which they did) AND women are more likely to vote than men (which history shows they are), I think I could actually make a stronger case women were slightly UNDER-sampled ... but either way, not to any significant degree.

2. Regardless of how large the "rural" counties are, their population remains small compared to that of Pima. 100 or so does look to be enough to draw statistically relevent values conclusions ... and my face is straight as I say that. :)

I do partially agree with you, however, in that with the smaller sample set any random difference (say too many women were included just within this set of 100, for example) will have a magnified effect on the outcome.

3. I entirely agree with you regarding the absence of a breakout by party affiliation. It sure would be nice to have.

In theory you could do some three-variable algebra to figure it out based on the party table, but that's more work than anyone should have to put in.

It's interesting to me the Weekly would sit on the poll results that long ... if it's true, any idea was to why they would? I'm coming up blank.

sirocco said...


You make a very good point ... regardless of what the figures are now, all they can really do is give both sides some idea of where they stand prior to the debates.

There is certainly more than enough time for things to change, and change drastically, either from something the candidates say/do or outside events they have no control over.

Framer said...

I would suspect that the Weekly Poll is going to be between 8 to 12% for Gabby. I'm kind of disappointed that the survey the Weekly is doing isn't a bit fresher. If the results were tabulated by Tuesday, then the calls had to have been made just after or concurrently with the Star poll.

I do like Kenski, however, and if she is involved with the poll, it should be worth a little more than what we have seen so far. She seems to have a bit more natural curiosity about relevant matters that are included in her polls. I would also expect a breakdown from her.

I'll see if I can use any of my sources to get a sneak peek. Now where did I put that number for Gabby's gardener. . .