NON-RESPONSE BIAS? HERE'S WHY IT HURT KERRY MORE THAN BUSH..

[TruthIsAll]

Another Dagger in the Heart of rBr.

From Statistics for Business and Economics: 
Heinz Kohler 
Second Edition

I quote from Chapter 2: Surveys and Experiments.

“Faulty questionnaire design, can, finally, be responsible for
NONRESPONSE BIAS, a systematic tendency for selected
elementary units with particular characteristics not to
contribute data in a survey while other such units, with other
characteristics, do. In the presence of this problem, even a
census based on a perfect frame, or perfectly selected random
sample, will fail. 

They will yield faulty conclusions because the data actually
collected will in fact constitute a convenience sample (for
example, of the MOST STRONGLY OPINIONATED PEOPLE among all the
people that were supposed to be in the survey).

 Questionnaire features that contribute to non-response bias
include a physically unattractive design; hard-to-read-print;
questions that are boring, unclear, or long and involved; an
excessive number of questions; bad sequencing of questions (so
that respondents are forced to jump back and forth from topic
to topic), and, in the case of multiple-choice questions, the
specification of answers that are not mutually exclusive or
are excessively restricted to particular points of view, while
omitting other possible views. 

Experience shows that HIGH-INCOME people and LOW-INCOME people
(unlike MIDDLE-INCOME PEOPLE) tend NOT to respond to surveys;
it is easy to see how the exclusion of either group is apt to
bias survey results.”

Now look at the National Exit Poll (13047) income demographic.
Low-income people (under $50,000) comprise 46% and high-income
(over $100,000) 18% of the voting mix. 

Kerry voters comprise the vast majority  (57%) of voters under
50,000. Bush voters are 54% of the high income group.

For Kerry the weighted total is .57 *. 46  = 26.2%
For Bush, the weighting is .54 * .18 =  9.7%.
Advantage Bush. Fewer non-respondents.

Strongly-opinionated people are people like the Republicans
whose riot terminated the Dade County Recount in 2000. They
would NOT be reluctant to fill out the survey - cause they
ain't shy. 
Advantage Bush. Fewer non-respondents.

On the other hand, as Peace Patriot has so eloquently pointed
out, Republicans who voted for Kerry could very well have been
reluctant responders. THAT'S A HYPOTHESIS WHICH MAKES SENSE.
THERE IS A RATIONALE TO IT. YOU CAN UNDERSTAND IT.

ON THE CONTRARY, rBr is TOTALLY DEVOID OF RATIONALE. NAYSAYERS
PROMOTE THE HYPOTHESIS WITHOUT CITING A SINGLE FACT TO BACK IT
UP. THEY JUST THROW IT OUT THERE BECAUSE THEY HAVE NO OTHER
EXPLANATION FOR THE 6% DIFFERENTIAL RESPONSE - EXCEPT FOR
FRAUD.

BASED ON THE INCOME DEMOGRAPHIC ALONE, COULD IT BE THAT KERRY
WON THE ELECTION BY THE SAME 53-47% DIFFERENTIAL? 
WHY WON'T THEY HYPOTHESIZE THAT?

And if there was a differential response, we have just
provided a RATIONALE for why it most likely would have hurt
Kerry more than Bush. In which case Kerry may have won the
election by more than the 51%-47.5% spread shown here:

INCOME	Mix	Bush	Kerry	Nader
0-$15K 	9.0%	33%	66%	1%
$15-30 	15.0%	39%	59%	2%
$30-50 	22.0%	47%	52%	1%

$50-75 	23.0%	53%	45%	2%
75-100 	13.0%	50%	49%	1%

100-150 11.0%	53%	45%	2%
150-200 4.0%	53%	47%	0
200+ 	3.0%	58%	41%	1%

	100%	47.54%	51.01%	1.45%
	122.26	58.12	62.36	1.77

[Starfury]

1. Makes sense to me!

I never felt the election was as close as even the initial exit polls were claiming. Maybe there was something to that feeling other than my wishful thinking. Thanks for posting!

[Febble]

2. Yup

You are getting there.

Keep reading. Non-response bias is a fascinating subject. As you will realise, of course, it is not simple - if only we could predict non-response on the basis of income alone!

Wouldn't life be easy for us guys with the spreadsheets!

But it's a start. Now, maybe try a chapter on political views....

And perhaps think about the question: what kind of voter (and remember there are reluctant Bush AND Kerry responders out there) might be shy? Not those loud Miami Dade types, no, definitely not. Not those guys with the brash bumper stickers, definitely not them. Kerry voters in evangelical churches - they might be shy, certainly. And what about Bush voters who normally don't vote at all? The ones who only turnout in very high turnout years? Like 1992, when the exit polls were also discrepant with the count? They might be shy. Now, which voter group is likely to be the shyest? It's going to be neck and neck....


Could be a few more Bushies. Could be a few more Kerryites. In which case, maybe there was fraud....

Decisions, decisions....

Math is SO much easier, isn't it? But psychology is so much fun.

[Sancho]

3. You've hit the nail on the head....!

That's why we need to profile the respondents with the raw data. EM KNOWs what the respondents look like, and they ain't telling....the fact that they don't release the data demonstrates what must me in the data!

[Febble]

4. No they don't have data on the non-respondents

well, literally they have data on what they LOOK like - but since when could you tell how a person voted by what they look like? You can only guess.

The whole point of non-response bias is that you don't have data on your non-responders. If you did, there wouldn't be a problem.

[Carolab]

6. What I want to know is....

were they men? Because on election night, Rove's argument was that the exit polls were wrong because men were underrepresented proportionately. How does that jibe with the rBr?

[kster]

5. KICK.NT