Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: Hand Recounts of votes recorded on DREs



Vern Williams, wrote, in part:

> It would be interesting if you would compare the accuracy of the
> DRE voting systems with Mark Sense and see which is more accurate.
> Maybe the ACM would be interested in those results.

The table below is from the Caltech/MIT report.  It measures residuals,
not accuracy directly.  (Accuracy of an output can't be measured
without something to compare the output against, which is expensive
for some of these technologies and impossible for some.)   We know
residuals never reach zero, but reduced residuals are taken to
indicate increased accuracy.


                        Table 3
      Residual Vote as a Percent of Total Ballots Cast
                 By Machine Type and Year
        US Counties, 1988-2000 Presidential Elections

        Residual Votes as a Percent of All Ballots

Machine Type            1988    1992    1996    2000
-----------------       ----    ----    ----    ----
Paper Ballot            2.2     1.4     2.1     1.3
Lever Machine           2.0     1.5     1.7     1.7
Punch Card
  VotoMatic             2.9     2.2     2.6     3.0
  DataVote              3.7     2.4     2.1     1.0
Optically scanned       2.5     2.4     1.5     1.2
Electronic (DRE)        3.5     2.5     2.9     1.6
Mixed                   2.1     1.4     1.5     2.7

Overall                 2.5     2.0     2.1     2.0


The first paragraph of the Conclusions section of the Caltech/MIT
report reads:

  "Paper ballots, lever machines, and optically scanned ballots
   produce lower residual vote rates on the order of one to two
   percent of all ballots cast over punch card and electronic
   methods over the last four presidential elections."


Regards,
Dave.