2004 Pollster Shootout! - Mason Dixon Vs SUSA vs POS vs The 'Bot

The Final results have been tabulated, and boy was it a barn burner for top honours in 2004!

The contestants were evaluated in 3 categories:

#1 - Correctly predicting the winners:

Rasmussen - 23.5 out of 24 (Called Iowa a tie)
Survey USA - 29/30 (Missed Iowa)
Mason-Dixon - 23/24 (They missed Minnesota)
Public Opinion Strategies 17/21

Fate was very cruel to POS in this Category

In Iowa they had Kerry + 1.3, Bush won by 0.7
In New Mexico they had Kerry up 0.6%, Bush won by 0.79%
In Wisconsin the had Bush up 1.2%, Kerry won by 0.38%

Three races they correctly has as a coin toss, three times it came up heads not tails... ouch! - In my mind, they didn't get these races "wrong" they were just unlikey.

They did however actually get Minnesota "wrong" in the sense of fundementally getting the race wrong - they had Bush up 2.1%, Kerry won by 3.48%

Ordinal rankings

The ' Bot => 1
SUSA => 2
Mason Dixon => 3
POS => 4

The two "robo-pollsters" take spots 1 and 2... !

Average Margin Of Error

POS - 2.07%
Mason-Dixon - 2.22%
'The Bot - 2.44%
SUSA - 2.96%

Ordinal rankings

POS => 1
Mason-Dixon => 2
Rasmussen => 3
Survey USA => 4

The Humans strike back with a 1-2 finish....

Three firms with an average magin of error of under 3% - All 4 firms can be very, very proud. That is excellent polling.

Bias

The bias, or rather lack there of, by Public Opinion Strategies and Mason Dixon is rather stunning...

On average, Mason Dixon had a pro-Kerry bias of 0.02%
On average, POS overstated Bush's support buy 0.04%
Rasmussen underpolled Bush by about 1.23% (statistically trivial) mostly in the South, and finally SUSA did have a bit of a pro-Kerry bias of about 3.5%, again mostly in the south...

Ordinal rankings

Mason-Dixon => 1
POS => 2
Rasmussen => 3
Survey USA => 4

Final Scores

Mason Dixon - 3 + 2 + 1 => Score of 6
POS - 4 + 1 + 2 => score of 7
Rasmussen => 1 + 3 + 3 => Score of 7
SUSA => 2 + 4 + 4 => Score of 10

For top honours we basically have a three way tie between Mason-Dixon, POS, and 'The 'Bot.

Survey USA gets a very, very honourable 4th place mention.

2004 Polling Summary

Most firms performed pretty much as expected in 2004...

Mason Dixon

They came in as "the champ" and while POS and The 'Bot basically battled the champ to a draw in 2004, you have to knock out the champ to take away the title.... Another very very solid year for Brad Coker and the boys...

This is FOUR election cycles in a row wher Mason-Dixon was the best pollster, even if you consider they "only" tied for #1 in 2004.

POS....

These guys are deep deep deep GOP.... so they get no respect.

But they don't need respect, they have huge piles of money as the defacto official GOP pollsters... These guys poll for about half the Senate and House, and looking at the numbers, you can see why...

The 'Bot...

A brutal 2000, a great 2004....

Did he actual "fix" things, or did he just get lucky....?

We will see, but let's give Scotty Boy his due in 2004... he had a great year.

SUSA.....

About what was expected - a slight Democratic bias, but overall a very respectable showing....

Research 2000

Again, about as expected... a solid firm with a mild Dem lean to it...

Zogby - Telephone and Internet

Yet another utter total massively pro dem biased statistical cluster F$%K that embarased both his firm and pollsters in general....

....in other words... steady as she goes, about as expected......

Gallup

My god, what happened...?

This is the question Gallup is asking themselves in a very public, humble way.

Gallup had a brutal 2004. But Gallup will, I predict, be back. This firm has too much pride, and history, and reputation, and money not to fix things.

Expect a very public "confession" from Gallup by about spring of '06, and a list of changes they plan to bring in.

	Welcome, Guest. Please login or register. Did you miss your activation email? June 18, 2013, 07:35:45 pm

News: Cast your ballot in the 2012 Mock Election!