scoring the pollsters

[windansea]

Tuesday, January 27 2004
THE NH RESULTS: So, Zogby pulled it out after all and found safe harbor in a double-digit Kerry win. Tomorrow New Hampshire will be forgotten and nobody will ever know or even bother to ask whether this was really ever the three point, dead-heat race Zogby said it was the day before yesterday.

Anyway, you may be wondering which of the pollsters performed the best in tracking the tricky turns of New Hampshire this past week. I sure was.

So I assembled a brief (and admittedly crude) scoring system based on two criteria. First, I've tallied the overall point differential by which a pollster missed each candidate's final numbers. For example, if a pollster projected Kerry to finish at 38% and Dean at 29%, and we now know the final returns were 39% Kerry and 26% Dean, the pollster missed Kerry's number by 1 point and Dean's by 3 for a total differential of 4. The pollster with the LOWEST total point differential came closest to predicting final outcome of the race.

Pollster Total Point Differential Correct Placements

CNN/USAT/Gallup (1/24-25) 7 4.5
Marist University (1/23-25) 7 3
Zogby (1/24-26) 8 3
UNH/Fox News (1/23-25) 9 3
ARG (1/24-26) 11 3
SurveyUSA (1/25) 13 3
Boston Herald/RKM (1/22-24) 14 3
Boston Globe/WBZ (1/24-25) 15 3
Suffolk University (1/26) 19 3
LA Times (1/20-23) 23 5
Franklin Pierce (1/20-22) 26 5

http://www.realclearpolitics.com/commentary.html#1_27_04_2342

Zogby probably would have been dead last if he didn't pull his last minute face saving switcheroo...go to the link above to see how Zogby "explained" the last minute change from 3% to double digit lead for Kerry...sounds fishy to me.

[windansea]

1. kick

:dem:

[West Coast Democrat]

2. Zogby putting Clark in a tie with Lieberman was uncalled for

Zogby clearly manipulated the numbers in an attempt to manipulate the voters.

[windansea]

3. yes..but his biggest mistake was saying

Kerry and Dean were only separated by 3% the day before...I am wondering what his agenda is

[RobertSeattle]

4. Thanks - we need more of this

We need more of these polls versus reality analyses.

[windansea]

7. no problem

even the number crunchers like to spin

[windansea]

9. kick n/t

[displacedtexan]

5. Zogby allowed the media to cite 1 out of 3 nights of a poll...

and he neglected to make them report that that one night total (200 people) would have a +/- 10 margin of error. I don't know why he did that, but the demographic professionals are outraged. Zogby has blackened their eye big time.

[windansea]

6. right...seems as if he was trying to influence voters

:thumbsdown:

[windansea]

8. kick

fast board today..hope people read this

[spooky3]

10. I am so glad you did this. I'd love to see it be a permanent part of DU.

Updated, of course, when the other primaries come in. Any chance you would be willing to do this?

[windansea]

11. Clear Politics did it not me

but if I see more poll analysis i will post it

[spooky3]

12. Thanks. I guess I didn't understand that from your post.

I will ask the administrators if we can link this or create some space on DU for poll analysis.

[windansea]

13. there's no single source that I know of

picked this one up from another site

[Jack_Dawson]

14. Are there no "democratic" pollsters or focus groups?

Why are we always subjected to Zogby and Frank Lunz and their ilk?

[windansea]

15. accurate polls are best

un biased info is what we want