Reply #134: Return of Service....

[anaxarchos]

You have asserted that "Bush won because of a dramatic red-shift of Gore voters from 2000, at approximately twice the rate of blue-shift in the other direction".

Your premise was "Bush gets 14.6% of Gore voters, while Kerry gets 7.2% of Bush2k voters".

Your only supporting evidence was the 2004 NES panel study.

I believe your “supporting evidence” in no way supports your premise:

1) There is inherently no reason to believe that a poll of 1100 “eligible voters” conducted after the election is more reliable than a poll of 13000 actual voters conducted on the day of the election. If I understand the claim, this is the reason that exit polls are “an advance”, i.e. that they are able to reliably eliminate response error inherent in the questions “Did you vote?” and “Who did you vote for?”. I am not talking about the retrospective question (yet). As ONE supporting piece of evidence, I cite:

Robert H. Prisuta, A Post-Election Bandwagon Effect? - Comparing National Exit Poll Data with a General Population Survey, SRMS/ASA 1993

available here:

http://www.amstat.org/sections/srms/Proceedings/papers/...

This is as exact a match to your argument as you are ever likely to find in research and Prisuta’s findings imply the exact opposite of your claim. I tried to bring this up with Febble (I guessed where you were going) but she chose to blow me off:

Available here (post #126):

http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x379573#389726

2) There are both extrinsic and intrinsic reasons to regard the NES as inherently unreliable on this issue. This response error is the oldest, most persistent, and largest in the NES, from its inception.

I cite as my authority on this...................... You.

(and, luckily, one hundred assorted other, fucking boring, NES Technical Reports, Research Summaries, Voter Verification Reports and other odds and ends of research that no one in prison should be forced to read).

You have been trying to tell everyone who will listen (and many who won’t) about “false memory”, etc.. By golly, you’re right. And how exactly did it go away when you decided to use it to reset the vote and the exit polls?

We can go down this very deep rat hole if you want to but I think it will be the end of the game and of my ability to stay awake. Why would we believe that you are not simply re-sampling the response error that is already stipulated?

I don’t believe anyone at the NES would support using the 2004 Panel Study in the way you have just done. But, hey, what do I know? Prove me wrong….

3) I believe you are misusing the research (an outrageous comment, I know, from an interloper like me). At the least, you are being very free with what part of your claims is actually “supported by a preponderance of research”, what part seems to you to be supported by any research at all, and what part is only according to you. In fact, I don’t remember you ever citing ANY basic research. You cite numbers, you say what they mean, and then you move on. Febble is actually worse. I ask her for a reference and she cites a source that says “people sometimes have false memories” … No shit! This is quite a catch-all. I can hardly quibble if you give me diddle. No wonder TIA says “False Memory, False Memory, False Memory”.

For starters, I think that “false memory” is largely a misnomer in this case or perhaps an artifact of previous research. I know that some researchers think that it is. If so, they claim a highly selective, “respondent-specific”, and difficult to quantify (unlike “Did you vote”) “false memory”, which applies under certain conditions. And as far as the NES goes, I know that efforts to fix this response error as “false memory” have largely failed. I cite as evidence:

Belli, Robert, Santa Traugott, and Steven J. Rosenstone. 1994. Reducing Over-Reporting of Voter Turnout: An Experiment using a ‘Source Monitoring’ Framework, NES Technical Report Series, No. nes010153.

Available here:

ftp://ftp.nes.isr.umich.edu/ftp/nes/bibliography/documents/nes010153.pdf

On the other hand (couldn’t resist) there is, what appears to me, to be the preponderance of thinking. That view attributes the response error to “the social pressure to vote”, particularly among those who were predisposed to vote but didn’t. This is indirectly implied even in the UK poll citations that Febble sent me. In those, the terms “misremembrance” and “lie” are used interchangeably, a dead giveaway as to whether this response error is a sin of “omission” or a sin of “commission” . More germane is the following citation which claims to have reduced the response error in the NES by nearly half by essentially letting people off the hook on the social pressure:

Duff, Brian, Michael J. Hanmer, Won-ho Park, and Ismail K. White. 2004. How Good is This Excuse?: Correction of Overreporting of Voter Turnout Bias in the 2002 American National Election Study, NES Technical Report Series, No. nes010872.

Available here:

ftp://ftp.nes.isr.umich.edu/ftp/nes/bibliography/documents/nes010872.pdf

If I understand the implications of this last, then how exactly does “false memory” even apply in exit poll “retrospective” questions? Exactly the same? Not at all? Cite your research.

I think the era of “false memory” has to end. I think you have to cite specific research, its implication, etc. The current path is guaranteed to insure disbelief (coming directly after rBr, etc.) as well as to confuse us as to when you are speaking for yourself and when you speak for “the dead” (i.e. “authority”).

Believe it or not, I say all of this point #3 without rancor. This is the web and not a journal…. But, you get my point.

One last issue: You just told TfC : “My reading of the General Social Survey results is that false recall favoring the incumbent pretty reliably increases over time.”

That disagrees directly with Presser who claims the exact opposite:

Stanley Presser, "Can Context Changes Reduce Vote Over-reporting?," Public Opinion Quarterly, Winter, 1990.

You might mention when you are being “controversial”.

4) I believe there is a logical fallacy in your presentation. If your premise were true, THEN one explanation of the exit poll retrospective question could be “false memory”. But this “explanation” actually drives the initial premise. You are pulling from the ass end of the donkey (sorry, couldn’t resist). The initial premise could have been cherry picked to produce this result or it could have come from your astrologer. Ultimately, this is a circular argument, with each end ultimately relying on the other. The donkey has its head up its ass.

To prevent this, you have to present independent, empirical evidence

I am not protesting but don’t blame me if the earth swallows you up.

5) I think many of the responses that you received, immediately went to the heart of the matter. BillBored went straight to the exit polls… “look at the party split”, etc. Chi asked the obvious historical question… “when has a red-shift like that ever happened when everything else went blue”. MelissaG said: “Wait a second. I thought you said you had evidence”. TfC was, in some ways, the best of all: “But, but… that’s not “plausible”…”. All of this was premature because you hadn’t made a case yet. But it IS the reason I hang around this board.


I claim first point by right of “reason” and “virtue” (unless you protest)……

Luv-15…

What else ya got?