E-Voting News and Analysis, from the Experts

Thursday November 04, 2004

Analysis of e-voting vs. optical scan in Florida

Filed under: — Joseph Lorenzo Hall @ 4:17 pm UTC

Charlie Strauss of Verified Voting New Mexico has has analyzed data from Florida (provided by Kathy Dopp from FL officials). Some interesting trends seem to appear… I’m not sure I fully understand this yet. Maybe someone with more time on their hands can examine this critically and post comments.:

TREND 1: optical scan shows much less predictable voting trends than e-voting

  1. Both parties show greater party defection on optical scan than e-voting
  2. Of the two democrats show slightly more of this discrepancy
  3. The large discrepancies from the trend line hurt Democrats just slightly more often than Republicans but this may not be statistically significant.

TREND 2: The smaller the precinct the greater the party defection

  1. There is very strong effect of the smaller the precinct the more voters crossing over to Republican. This vanishes in large precincts.
  2. [In] every case the trend favored the Republican party.


The URI to TrackBack this entry is: http://evoting-experts.com/wp-login.php/wp-trackback.php/wp-admin/wp-admin/wp-admin/wp-trackback.php/wp-trackback.php/wp-trackback.php/wp-trackback.php/59

  1. I don’t see how we can conclude anything from this data. There are so many confounding variables here: turnout rates (which could potentially vary by party, or by precint), undervote rates (which could also vary), and the like. It is hard to separate these from the accuracy of the recording method.

    I’m not sure I understand the observation of “more variability on DREs than on optical scan". If I understand this correctly, it seems to be based on three “Democratic optical scan” and two “Republican optical scan” points. Perhaps the next thing to do would be to look carefully at those precints in detail.

    The question raises the hypothesis that perhaps DRE machines tend to make voters more loyal. If this hypothesis were true, one would expect it to hold throughout the nation; that ought to be something that could be put to the test.

    I think we have to be cautious how we interpret statistics. In general, there are two modes of statistical analysis: exploratory (where one searches for a hypothesis), and evaluation (where one tests a specific hypothesis). It is important not to confuse the two. At this point, this is clearly very much at the exploratory stage.

    Comment by David Wagner — Thursday November 04, 2004 @ 5:33 pm UTC

  2. Average US turnout was 60%, I believe. That leaves lots of room for for this possibility: non-registered republicans show up, registered Democrats to not. This means the GOP is better at getting the vote out. If voting was 100%, or closer to it, it might be reasonable to talk about cross-overs and loyalty. Otherwise, that is entirely supposition.

    Comment by Michael Lines — Thursday November 04, 2004 @ 6:13 pm UTC

  3. Here’s another analysis that shows defection rates of 6% in optical scan equipment and 1.2% in eVoting machines:


    Comment by Bill — Friday November 05, 2004 @ 1:25 pm UTC

  4. I forgot to point to Kathy’s data: http://ustogether.org/Florida_Election.htm

    Comment by joe — Friday November 05, 2004 @ 1:47 pm UTC

  5. Looks like the collision of four noncontroversial effects:

    1) Small counties can’t afford e-vote machines yet

    2) Democratic vs. Republican tendency is well-correlated to population density.

    3) Bush won

    4) In partisan counties, voters across the line are likelier to go
    with the guy their neighbors favor.

    Comment by Jon — Friday November 05, 2004 @ 2:01 pm UTC

  6. Needed: Careful E-Voting Correlation Study
    Tuesday’s election created lots of data about voting patterns in places that used different voting technologies. Various people have done exploratory data analysis, to see how jurisdictions that used e-voting might differ from those that did not. See, …

    Trackback by Freedom to Tinker — Friday November 05, 2004 @ 4:21 pm UTC

  7. Re: #2 –
    Mike, if Republicans weren’t registered, then they couldn’t vote.

    Comment by Joe Murphy — Monday November 08, 2004 @ 10:41 am UTC

  8. I have looked very hard at Kathy Dopp’s data. It’s called a “correlation without causation” in political science. Yes, there is a correlation. No, it is not caused by the op-scan machines themselves. It is caused by crossover voting by Dixiecrats who register as Democrats and vote for Republican presidents. This voting pattern has a long history in the South.

    Counties with poorer residents with less education have not yet acquired the newer and more expensive E-touch machines. Poorer, less well-educated Democrats also crossover and vote for Republican presidents much more often than their wealthier, better educated party brethren who live in metropolitan and suburban areas. Those latter areas have acquired the new machines. The 2004 voting results mirror the ones in 2000 county-by-county. No big, sweeping changes.

    Kathy Dopp’s analysis is worthless. What do you expect from a former high school math teacher who’s trying to play political scientist?

    Comment by Mark B. — Monday November 15, 2004 @ 12:03 am UTC

  9. @ Mark B: While there may be merit to your argument that there were “No big, sweeping changes” in the 2004 voting results as compared to 2000, I feel that the ad Hominem attack on Kathy Dopp in the last paragraph was very much uncalled for. The source of an argument cannot invalidate the argument itself. You come to the conclusion that the analysis is worthless. Maybe you should leave it at that.

    Comment by Tim M — Thursday November 18, 2004 @ 7:29 am UTC

RSS feed for comments on this post.

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <a href="" title="" rel=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>



Powered by WordPress