November 10, 2004

More on voting…

Follow-up to Post–election statistical anomaly from blogistic regression

This picture says a lot, I think…
(in case it is not legible, the LH columns show Exit polls (Blue = Kerry) and the RH columns show the "actual" vote count. The states are
Row 1. Illinois, Maine, Wisconsin (all paper ballots)
Row 2. N. Carolina, New Hampshire, New Mexico (all electronic voting)
Row 3. Florida, Ohio, Pennsylvania (all electronic voting)

- 2 comments by 2 or more people Not publicly viewable

  1. I think this is rather misleading. Could you provide the source?

    If you examine the states in question on this website, you will see that all the states above had a variety of voting methods in use, depending on the county. The Verified Voting Foundation, which certainly has criticisms of some of the machines used, shows that in North Carolina (which I use just as an example) optical scan (central and precinct-based counting), touchscreen and other forms of e-voting, punch-cards, and (3 counties) paper ballot were in use. All of the graphs above must be based on some very limited number of counties, especially the paper ballots in Maine and Illinois. I do not have much confidence in this extremely partial "evidence." Is the exit poll data really available down to county and precinct level (as the votes are) to check out?

    As much to the point, what evidence is there that exil-polling was not subject to systematic distortions? "This post"@ on my blog refers to a report that thinks it was.

    Since you suggest that the picture says a lot, perhaps you would care to articulate some of it?

    10 Nov 2004, 21:25

  2. Jonathan Cave

    The source is a colleague at warwick – where he got it from I don't know. In any case, there are much more complete data available now. The more detailed (country-by-county) evidence shows that the worst (and most consistent) error rates come from the counties using the Diebold optical scanners, with touchscreens doing rather better.

    But my point is not whether this election was stolen from its rightful winner. However you slice it, an awful lot (or should that be the other way around?) of people wanted W. back. Rather, I am concerned that the method itself seems difficult (and in many cases impossible) independently to audit, and that as a result the credibility and thus the societal benefit of the US form of democracy is weakened.

    As for what I think the picture does say, it includes the following:

    The relationship between exit polls and tallies has traditionally been both close and relatively unbiased – in other words, error rates were small and as likely to be hgih as low (though I freely confess that I have not looked at the skeness and kurtosis to verify these statements). Now that relationship has not merely weakened, but shifted discontinuously. It is reasonable to ask why, and to ask what this means for the accuracy of other methods of measuring public opinion. If it holds for the exit polls, why not for the pre-vote polls? Why not for the vote itself (given that it remains only a random (though not unbiased) sample of public preferences?

    I know that e.g. the "Dixiecrat" phenomenon has been adduced to explain the widespread failure of what had until now been one of the most accurate estimates of actual tallies. So a longstanding relationship seems to have undegone a structural shift. At the same time, we have new technologies, a higher degree of polarisation, a higher turnout than before and a shift in the kind of issues ruling the campaign. Until we see a decent panel study, we won't know which is responsible.

    10 Nov 2004, 23:25

Add a comment

You are not allowed to comment on this entry as it has restricted commenting permissions.

November 2004

Mo Tu We Th Fr Sa Su
|  Today  |
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30               

Search this blog


Most recent comments

  • The source is a colleague at warwick – where he got it from I don't know. In any case, there are muc… by Jonathan Cave on this entry
  • I think this is rather misleading. Could you provide the source? If you examine the states in questi… by on this entry
  • Thank you Owen. This makes very interesting reading. I knew there had to be a way to blame Bill Gate… by Jonathan Cave on this entry
  • This is probably not what you were talking about, but it is very, very interesting nonetheless: Link… by on this entry
  • Not yet, but I'll post when I know more. by Jonathan Cave on this entry

Blog archive

Not signed in
Sign in

Powered by BlogBuilder