Wednesday, January 19, 2005

HALF-TRUTH: Mitofsky Solves The Mystery of the Exit Poll Discrepancy

U.S. votergate media blackout
Finally, on the eve of the Presidential inaugauration (timely, wouldn't you say), Mitofsky's conclusions for the much anticipated initial investigation into the exit poll discrepancy for the 2004 election have been released to the public (see his full report here).

As reported by CNN, here are his main assertions as to why Mitofsky believes there was a significant discrepancy between the exit poll numbers and the actual results:
Distance restrictions from polling places imposed upon the interviewers by election officials at the state and local level.

Weather conditions, which lowered completion rates at certain polling locations.

Multiple precincts voting at the same location as the precinct in the exit poll sample.

Interviewer characteristics, such as age, which were more often related to the errors last year than in past elections.
The numbers in the report are, indeed, interesting. For example, look at the numbers for the "interviewer distance from the polling exit":

Interviewer Distance
NOTE: WPE = Within Precinct Error. A negative number indicates that the exit poll differed by the main results in Kerry's favour. Similarily, a positive WPE would indicate that the exit poll overestimated the Bush numbers.

There definitely seems to be a co-relation between Republican voters not completing exit polls due to how far the exit poll interviewers are away from the polling building. What does this mean? What are we to assume is going on here? Do Republicans overwhelmingly park their cars closer to the polling building? Are Democrats more likely to take a bus after voting (the bus stop being in the general location of the exit poll station)? No explanation is given by Mitofsky as to why this phenomenon occurs. Further examination is definitely needed here!

With respect to interviewer age, there also seems to be a slight co-relation with the WPE. That is, the older the interviewer, the less the exit poll discrepancy:

Interviewer Age

So, here Mitofsky insinuates that Republicans are less likely participate with younger exit poll workers (The "contempt for youth" effect). If Republicans tend to discriminate based on "youth", I would love to see the stats co-relating the WPE against whether the exit poll worker was "black" or "white". Alas, Mitofsky did not provide that analyses -- since only 7% of the workers were African-American.

All in all, do these numbers indicate that there was a "shy Republican" effect with the variables looked at in the Mitofsky analyses? It seems that way. However, the most obvious detail of every variable looked at in this study has been completely ignored by Mitofsky and the mainstream media.

THE WPE IS ALWAYS SIGNIFICANTLY NEGATIVE IN THE MOST IDEAL CIRCUMSTANCE.

That is, even if Mitofsky implemented all of the changes he recommends, there still would be a significant discrepancy.

Don't believe me? Look at all of the systematic bias variables looked at above and then look at the most "ideal" situation:

1. Distance Restrictions -- PROBLEM: THE FURTHER AWAY FROM THE BUILDING, THE LESS LIKELY REPUBLICANS WERE TO RESPOND

There was still a -5.3 and -6.4 discrepancy with the mean WPE even when the exit poll workers were "inside" or "right outside" the entrance.

DISCREPANCY STILL EXISTS!

2. Weather Conditions -- PROBLEM: BAD WEATHER MADE REPUBLICANS LESS LIKELY TO RESPOND

There was still a -6.2 discrepancy with the mean WPE when the weather was NOT a factor.

DISCREPANCY STILL EXISTS!

3. Multiple Precincts -- PROBLEM: i) the interviewer was not able to interview voters only from the sample precinct; ii) either the reported vote or the exit poll is not only from the sample precinct alone.

There was still a -6.3 discrepancy with the mean WPE when there was ONLY 1 precinct.

DISCREPANCY STILL EXISTS!

4. Interviewer Characteristics -- PROBLEM: THE YOUNGER THE AND MORE EDUCATED THE INTERVIEWER, THE LESS LIKELY REPUBLICANS WOULD RESPOND

There was still a -7.0 discrepancy with the mean WPE when the age of the interviewer was between 55-64. A similar discrepancy still exists when education and gender is taken into consideration.

DISCREPANCY STILL EXISTS!

Has Mitofsky proven that a co-relation exists between these above variables and the Republicans willingness to participate in the exit poll? It seems that way.

HOWEVER, the big question is this:

Has Mitofsky proven that these variables account for the significant discrepancy between the exit poll numbers and the actual results of the election?

That answer is a BIG RESOUNDING "NO".

According to Mitofsky's results, there would STILL be a significant discrepancy even if all of his recommendations were followed. Why no mention of this extremely important detail? The silence, as we say, is deafening.

A crude analogy would be to say that a bowl of sh*t would taste bad because one put too much pepper on it. However, I think we would all agree that if we take away the pepper, it would still taste bad. (I hope no one argues with me on that particular detail).

So remember: When the editorials from the mainstream give you the pablum that it has been proven by Mitofsky that the exit poll discrepancy was a result of Republicans not wanting to participate (due to the variables listed above) -- please write them and set them straight!!! THERE IS STILL AN "UNACCOUNTED FOR" DISCREPANCY.

I would love for Dr. Freeman to do another follow-up to his yet unrefuted research, showing that these variables listed by Mitofsky do not adequately account for the overall significant discrepancy between the exit polls and the official election results. I'm sure we will hear from him soon!

Mitofsky (and the mainstream media) still assume that there was no significant COUNTING ERROR (malicious or benign). The evidence of widespread irregularites across the U.S. and, most notably, in Ohio, paints Mitofsky (and the press) as being a bit naive (or careless) in this regard. As he states in his report:
It is difficult to pinpoint precisely the reasons that, in general, Kerry voters were more likely to participate in the exit polls than Bush voters. There were certainly motivational factors that are impossible to quantify, but which led to Kerry voters being less likely than Bush voters to refuse to take the survey.
How about this reason: More people voted for Kerry than what was officially recognized.
Image Hosted by ImageShack.us
Name:
Location: Vancouver, British Columbia, Canada