October 30, 2003

More Punch Card Analysis by Henry Brady

Henry Brady has set up this page with his collection of analyses of error rates with punch card voting in California. Most important is his new document, "Detailed Analysis of Punch Card Performance in the Twenty Largest California Counties in 1996, 2000, and 2003." One of his conclusions:

    Summary of Results from 1996 to 2003 In 1996 ten of the twenty largest punch card counties (Alameda, Fresno, Los Angeles, Marin, Sacramento, Santa Clara, San Bernardino, San Diego, San Francisco, and Solano) used punch cards. By 2003 only five of the twenty used punch cards. The five counties that changed from punch cards to other systems are represented by solid triangles on Figure 3. This figure compares the residual vote performance of counties in 2003 (on the vertical axis) with their performance in 1996 (along the horizontal axis). All the counties that changed away from punch cards improved substantially. Their residual vote rates in 2003 were substantially below all the remaining punch card counties and similar to the residual votes for the non punch-card counties. The remaining punch card counties had an average residual vote rate of 7.77% in 2003, the counties that had employed non-punchcard systems since 1996 had an average residual vote rate of 2.14% in 2003, and those counties that had changed from punch cards between 1996 and 2003 had an average residual vote rate of 2.17% in 2003.

Brady also responds to an argument (made by, among others, Judge Kozinski at the oral argument in the en banc punch card lawsuit in the Ninth Circuit, Mickey Kaus, and L.A. Registrar-Recorder Conny McCormack) that errors in punch card votes can always be fixed by an after-the-fact recount of the votes:

    Recounts Prove there are no Problems Some observers have claimed that the 1% recounts done after the election prove that there are no problems with punch card systems because these recounts obtain almost exactly the same result as the initial count. This claim misses the point. My argument is that the biggest problem with punch cards is that they fail to record the vote. There are also sometimes problems with counting the vote, but these problems are much smaller than recording the vote. If, as I argue, votes
    are never recorded, then counting and recounting them will not reveal a problem or solve it. (Counting and recounting your pocket change after your wallet has been stolen does not solve the problem.) The difficulty is that some intentional votes are not recorded by punch cards. When better voting systems are put in place, then these votes are recorded.

Posted by Rick Hasen at October 30, 2003 09:44 AM