ARG anti-Dean update -- skewing polls

[lib 4 all]

Take a look at this respondent polling data from the last few days. ARG is steadily reducing the numbers of Democrats polled and increasing the number of "undeclared" respondents.

Democrats ---- : 453 -- 445 -- 437 -- 436 -- 436 -- 432 -- 425
Undeclared --- : 169 -- 174 -- 176 -- 177 -- 177 -- 180 -- 185
Undeclared (%) : 27% - 28% - 29% -- 29% - 29% - 29% - 30%


This explains Clark's seeming climb over the last several days in the ARG tracking poll.
They are polling more "independents" and phasing out Democrats participating in their polling.

Something to be aware of. I thought there was something odd about it and have been skeptical.

[Renew Deal]

1. But I thought the Independents are going to Dean?

That's what Dean supporters on DU will tell you. Also, the poll said a couple days ago that Dean was leading independents. I don't think you are right.

[Bombtrack]

3. the independants who always vote republican in GE's will be for Dean

plenty of them, which is pretty aweful. Our primary system is just asking for sabotage.

[KaraokeKarlton]

6. edit

misread the post.

[Renew Deal]

9. Edit: It's going to be -12 in Manchester this week.

I hope you have your warm socks.

[KaraokeKarlton]

25. I live in Vermont

and it was colder than that a couple of days ago here.

Oh, and did you hear about Clark's flopped event in NH last night? I guess everyone was home watching the debate. Less than half of the number they expected showed up. There were a lot of empty chairs in the audience. Kerry is about to launch a strong offensive against Clark in NH. We'll see how much teflon Clark has. :D

[SahaleArm]

26. The Kerry proxy fight? Here Jean - do the dirty work for me...

Somewhere in the backroom of the Clark campaign Chris Lehane is laughing.

[Walt Starr]

7. Most ind3ependents who are backing Dean are of a couple of categories

1) New voters. These people have never ever voted before in their life and as such, will never ever be polled as a "likely voter". They are newly registered.

2) Disaffected voters. These voters have not voted for several elections. All of them sat out 2000. Many of them have not voted in over a decade. Many of them had to re-register. Because of this, not a single one of them will be polled as a "likely voter".

Frankly, I'm glad. It makes everybody (* included) "misunderestimate" (sic) Dean.

[lib 4 all]

14. perhaps a better title would have been "pro-Clark bias" then

it may not be anti-Dean per se, since he hasn't moved much either direction, but it does seem to explain the correlation between Kerry's drop vs Clark's gain.

[SahaleArm]

2. Until you realize that 1/3 to 1/2 of the voters will be independents...

because there is no Republican primary. Independent voters make up a plurality of NH voters; that's how McCain won in 2000.

[Kahuna]

5. Bingo!!!! You are correct...

...as usual! :D

[bearfartinthewoods]

4. how do they control the status of the people who answer

the phones? aren't they just calling random numbers?

[Marian]

8. or what about those who just mess

with the pollsters?


Polls/Schmolls!!!!




(edited for clarity - no snide remarks, please :evilgrin:)

[lib 4 all]

10. they may be over-sampling and then filtering out

the responses they don't want in order to create the appearance of a Clark surge. I think this explains the steady increase of "undeclared" vs the steady decline of registered Democrats.

[zeemike]

11. It’s obvious to me

Poll more independents and stop when you get the results you want.
It reminds me of a book I read long ago, “1001 ways to lie with statistics”.
Welcome to DU Lib!

[lib 4 all]

12. thank you

I've been a long time lurker.

;)

[SahaleArm]

13. Will you be complaining when 40% of the real votes are independent? n/t

[TeacherCreature]

17. You missed the point

The point is not to shut independents from voting (half my Dean meet-up is republican and third party voters). The point is that the way they are conducting the polls is making it look as if Clark is gaining ground in the democratic primaries (to create momentum)and he is not.

[SahaleArm]

20. Why? Should polls not accurately reflect NH voting tendencies?

[lib 4 all]

15. trend continues in just released numbers:

Democrats ---- : 453 -- 445 -- 437 -- 436 -- 436 -- 432 -- 425 -- 425
Undeclared --- : 169 -- 174 -- 176 -- 177 -- 177 -- 180 -- 185 -- 191
Undeclared (%) : 27% - 28% - 29% -- 29% - 29% - 29% - 30% -- 31%

[SahaleArm]

16. And Dean went up - There goes that theory... n/t

[TeacherCreature]

18. no

This means that Dean is also getting more independent voters. The polls that tell the story about the primaries are the ones taken from a sampling of democrats.

[SahaleArm]

19. Why is that? There are more independents in NH than Democrats

They are sampling voters - why sample more Democrats than will vote in the primary?

[hedda_foil]

21. The point is that the percentage sampled should remain the same.

This is a tracking poll, which means tracking the same population or statistical population sampling from day to day. What matters is not whether the ratio of Dems to Indies is 2/1 3/1 or 4/1, but that the ratio is the SAME each day. In this tracking poll, the change in the population sample is statistically HUGE, which shows irregularity in the polling technique. It's not about Dean, Clark or any other candidate but about the trustworthiness of the pollsters' technique.

[lib 4 all]

22. Right. The percentage of "undeclared" has increased by 5% over a week

and that means that the number of Democrats has decreased by 5% over that same time, creating a 10% spread.

[SahaleArm]

24. We haven't gone up or down 5% started at 29% - The spread is 4%...

And assumes Dean lost all that support to Clark which isn't true because Kerry lost more support.

[lib 4 all]

27. Dean and Clark may trade "undeclared" support

but the support Kerry has seemingly lost to Clark matches the ARG shift from Democrats to "undeclared" supporters exactly.

Kerry's support is going to be among Democrats, not "undeclared" voters. Every percent of increase in "undeclared" voters in the tracking poll creates a decrease of an equal percentage of Democrats.
If there was 27% "undeclared" respondents, then there were 73% Democrats. Now with 31% "undeclared" respondents that means there are 69% Democrats.

The effect is going to be double damage against candidates like Kerry who get their support from core Democrats. This explains why Kerry's support and Clark's support as a matter of percentage both trend exactly 2 times the shift in affiliation of respondents.

[SahaleArm]

23. What are we tracking - probable voting results or fantasy?

The question is likely voters and as long as the numbers are within MOE they are accurate (+/- 4%). Here's the past couple weeks of undecided's:

29% 28% 27% 28% 29% 28% 27% 27% 28% 29% 29% 29% 29% 30% 31%

[lib 4 all]

28. let's not forget that ARG has already shown an anti-Dean bias

illustrated here: http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=93666&mesg_id=93666

Whether or not this skewing of respondents is anti-Dean per se is difficult to say, but I believe the sum total of their actions is ultimately to push Clark. Hurting Dean is just one factor in this.

[unfrigginreal]

29. More grist for the mill

I think that you've nailed the reason that this pollster increased the percentage of undeclared. They are anti-Dean and they thought that increasing the undeclared in the polls would result in higher Clark numbers. The president of the firm just made some comments about the undeclared voters the other day but he didn't mention the reason that they've suddenly weighted their poll more heavily with these voters.

Tracking polls released yesterday by the American Research Group in Manchester showed (Ret.) Gen. Wesley Clark’s support among undeclared voters to be higher than among registered Democrats — and gaining slightly.

In a telephone survey of likely Democratic primary voters conducted Jan. 7-9, Clark was polling at 24 percent among undeclared voters, up two points from the previous sampling.

Former Vt. Gov. Howard Dean is still the strong front-runner at 36 percent among undeclared voters. Among registered Democrats, Dean gets 35 percent, and Clark gets 19 percent in the latest poll.

“Clark is becoming more attractive among the undeclared voters,” said Dick Bennett, president of ARG.

http://www.theunionleader.com/articles_showa.html?article=31420