You are viewing an obsolete version of the DU website which is no longer supported by the Administrators. Visit The New DU.
Democratic Underground Latest Greatest Lobby Journals Search Options Help Login

Reply #33: This is getting SO silly.... [View All]

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
This topic is archived.
Home » Discuss » Topic Forums » Election Reform Donate to DU
Febble Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jul-15-06 02:50 AM
Response to Reply #22
33. This is getting SO silly....
Edited on Sat Jul-15-06 03:16 AM by Febble
We have a thread, again, about the exit polls, and whether they are evidence of fraud.

OTOH, and myself (Elizabeth Liddle) have spent a fair amount of time, as have others, looking into this question, and have concluded that they probably don't, and indeed I have come to the converse conclusion, that they are actually counter-evidence of million-vote scale fraud, although they are not nearly sensitive enough instruments to detect fraud on a smaller scale. Nonetheless, there is some evidence in the exit poll data that the exit poll discrepancies, at precinct level, were higher in urban areas where older technology was used, which may suggest that some kind of miscounting associated with older technology differentially benefitted Bush.

Now, I may be wrong. OTOH may be wrong. When it comes to these kind of data, all anyone can offer is probabilities, not certainties. And there may be ways of interrogating the data that I have overlooked. I have certainly spent the best part of the last couple of years figuring out new ways of interrogating the data that might shed light on the issue. So I quite accept that there are arguments and counter arguments that could be made.

But I do not see them being made on this thread. I see attacks on OTOH's integrity. On other threads I see attacks on mine. They are completely irrelevant (apart from being offensive). If an argument is worth making it is worth making. I am interested in arguments. I am interested in counter-arguments. But I am not interested in posts that seem to conclude from any post that questions the proposition that the exit polls evince fraud that the poster
  • thinks the election was fair and uncorrupt
  • thinks that DREs or other digital voting systems are reliable or secure
  • thinks that Bush deserved to win
  • is a Republican plant
  • wants to "undermine" the exit polls
  • is not interested in evidence

OTOH made a substantive series of responses to BeFree's OP. So, leaving aside, for a moment, your unsupported inferences about OTOH's integrity, let's look at your counter-points.

Your response to the point that the exit polls were almost as far off in 1992 is to argue that the poll in 2004 was larger and more sophisticated. I don't know what your evidence is for this. What may have been more sophisticated were the models used to make the projections from the various data sources (which, as usual, included the vote-returns) but I am not aware of any changes to the data collection protocols, nor am I aware that the sample sizes were any larger. Indeed, in 2004, the mean precinct sample size was only 80, for a target sample size of 100, so the total sample may even have been smaller. Non-response was a big problem in 2004. 53% is a low response rate.

Your response to OTOH's second and third points is the word "sophistry". I'm meeting this word increasingly. Unless you can say what it means, there can be no response.

Your next point is:

Do you have evidence of this.... are you saying that before their were any official numbers the exit pollsters already had doubts about their numbers. Because why? And if this was the case then why was the data released raw on all the networks without a warning to this effect. And since it was isn't it more than probable that this is yet another post-facto BS explanation from Mitofsky.

Well, Mitofsky is on record as saying he knew the raw numbers were off before polling closed, but Mystery Pollster points to direct evidence that this was not "de facto BS" Mitofsky from Wonkette, who leaked the information at 5.40pm on election day itself:

*** There appear be problems with exits in the following states that could be tipping numbers toward kerry: MN, NH, VT, PA, VA, CT, DE. described only as serious issues were looking at. so i would not put too much faith in those results.

As for your second point - none of the data was "released raw" - it was all weighted, precisely because that's what pollsters do when they have reason to believe their sample is biased, and is why no projections were made on the basis of the exit poll samples alone. The idea that the numbers CNN put up were "raw" is false, although it is true that those numbers were (mostly) not weighted by the vote-returns as at that point, not many vote-returns had entered the data-stream. But they were weighted by other stuff, because already the pollsters knew that their sample had some "'serious' issues" indicating a pro-Kerry bias.

Then you deduce "bad faith" from the non-release of the data, and claim that it was "scrubbed". Well, I don't call posting "data" (actually cross-tabulations from partially weighted data) for hours on the internet "non-release" - I saw them myself. Sure, they were replaced later by fully weighted cross-tabs, because the assumption would be that the fully weighted crosstabs would be more accurate than the partially weighted cross-tabs. That assumption has been legitimately brought into question, but there is no a priori reason to suppose that the interim cross-tabs were "scrubbed" in "bad faith", and the fact that in their January evaluation (pages 21 and 22) E-M actually published their "Call 3 estimates" - in other words, the "raw" estimates made at close-of-poll, weighted only by geographic stratum and not by factors designed to compensate for sampling bias - looks to me like an act of good faith, not bad. Not only did they publish those Call 3 estimates, they also published additional tables showing how representative their precinct samples were of each state (good), and pinpointing the source of the discrepancy to within-precinct level.

There are certainly some anomalies in the reweighted cross-tabs that are worth examining, although the extent of the reweighting required (because the discrepancy was fairly large) means that both weighted and unweighted cross-tabs need to be treated with some caution. Once you think you have a biased sample you have to make a guesstimate as to where to apply the weights. If the weights are applied to the wrong characteristics, you won't necessarily get a better answer. But bear in mind that they knew before the first results were in that the sample had problems.

However, even that being the case, the anomalies are less anomalous than they appear once examined closely, and mostly concern the past reported vote question. OTOH researched answers to this question from past exit polls, and found a clear tendency for voters to over-report having voted for the incumbent at the previous election, including voters who claimed, in 2000, to have voted for Clinton in 1996. Strange it may be, but it seems to happen. In which case, the responses to the past vote question in 2004 are no longer anomalous (unless you want to postulate massive pro-Gore fraud in 2000).

"They actually said nothing for weeks infact" isn't a fact. Here is a transcript from a pbs interview with Mitofsky on November 5th 2004:

Data release: privacy is not a "specious argument". It is absolutely essential. The data were collected under promise of confidentiality. A great number of personal details were collected and are in the public domain. Release of data that would allow precincts to be identified could quite easily result in respondents being identified, and would run counter to the ethical guidelines of the pollsters own professional organisation. People keep asserting that confidentiality is not a problem. It is a huge problem. It would be less of a problem if those questionnaires were not already archived, but they were, and they always are. So turn off the caps lock and consider what you are demanding: access to data that would compromise the confidentiality of detailed personal information given by volunteer participants under promise of non-disclosure.

And as for your last point: it is completely unsupported. OTOH has worked harder than almost anyone I know to find out what the exit polls can tell us about what happened in 2004, and the assertion that "he never questioned his assumption" that he knew what happened is ludicrous. My email inbox is stuffed with questions from OTOH regarding just about any assumption either of us has ever made. And your accusation that he "resolutely refuse{s} to examine or discuss any of the facts and instead choose{s} to snipe unceasingly from the sidelines" is similarly ludicrous. His own homepage is stuffed with papers that examine and discuss facts in copious detail, his posts are full of discussions and facts, though typically responded to by sniping posts that avoid engaging either.

Neither OTOH nor I have any fear of facts. I wish the same could be said of others who, like both of us, believe that Election Reform is an urgent and serious requirement.

edited for clarity
Printer Friendly | Permalink |  | Top

Home » Discuss » Topic Forums » Election Reform Donate to DU

Powered by DCForum+ Version 1.1 Copyright 1997-2002
Software has been extensively modified by the DU administrators

Important Notices: By participating on this discussion board, visitors agree to abide by the rules outlined on our Rules page. Messages posted on the Democratic Underground Discussion Forums are the opinions of the individuals who post them, and do not necessarily represent the opinions of Democratic Underground, LLC.

Home  |  Discussion Forums  |  Journals |  Store  |  Donate

About DU  |  Contact Us  |  Privacy Policy

Got a message for Democratic Underground? Click here to send us a message.

© 2001 - 2011 Democratic Underground, LLC