With all the concern and speculation that's been swirling around the 2004 presidential election, it's useful to take a look at some data. The Ohio Secretary of State's office has posted the results of the November 2004 election. In conjunction with this list of voting equipment used from the Akron Beacon-Journal, this information allows an initial assessment of how the different types of voting equipment used in Ohio performed in this election.
Three different types of voting technology were used on November 2, 2004: punch card ballots, optical scan ballots, and electronic voting machines. As explained below, the initial results indicate that, as in prior elections in Ohio and elsewhere, punch cards performed much more poorly than other types of equipment. It also suggests that, if and when Ohio converts to better voting equipment, tens of thousands of votes will be saved.
Ohio's Voting Equipment
Ohio is one of the last bastions of the punch card ballot, with approximately 72% of Ohioans using this voting method. That's a higher percentage of punch-card voters than in any other swing state. With the punch card system, voters mark their ballots by using a stylus to punch a pre-scored "chad." Among the 68 counties using the punch card in the 2004 general election were large urban counties such as Cuyahoga (Cleveland area), Montgomery (Dayton), Summit (Akron) and Hamilton (Cincinnati), as well as many smaller rural counties.
The next most commonly used type of voting system in the general election was the Direct Record Electronic or "DRE" system, used by approximately 16% of Ohio voters. There are actually several types of DRE equipment used in different counties within the state, including older push-button machines used by voters in Franklin County and newer "touchscreen" voting machines used by voters in Auglaize and Mahoning Counties. The Secretary of State's office lists the type of DREs used, along with instructions on how each type of equipment is operated.
The third and final type of voting equipment used in Ohio 's November 2004 election is the optical scan or "Marksense" type system, used by approximately 12% of voters. With this system, voters make their choices by marking a piece of paper, which is then counted by machine.
There are two basic varieties of optical scan voting system: precinct-count and central-count. With precinct-count optical scan systems, voters feed their ballots into a counter at the polling place, and are notified if they've cast an inadvertent "overvote" - that is, if they've marked more than one candidate for a particular office. With central-count optical scan systems, ballots are counted at a central location, so no such notification is possible. Only three counties used precinct-count optical scan in the last election (Allen, Hardin, and Lucas), according to this AP report.
How Did Ohio 's Voting Equipment Fare in '04?
The metric that's most often used to measure voting system performance is "residual votes." That term is used to describe ballots for which no valid vote was cast. It includes both overvotes and undervotes (ballots on which no choice is read). Overall, there were a total of 94,488 residual votes in Ohio 's November 2004 presidential election. Of those, the substantial majority (76,398) were cast using punch card equipment.
Election Law @ Moritz has prepared a table which shows the total number of votes for each candidate (pdf file) in the 2004 presidential election, as well as the number and percentage of "residual votes." Put simply, the residual vote rate is the percentage of ballots for which no valid vote was recorded.
Overall, the residual vote rate in Ohio 's November 2004 presidential election was 1.65%. That's down slightly from the November 2000 election, in which the residual vote rate was 1.89%.
The table also indicates the type of system that each county uses for in-precinct voting. In 2004, the counties that used each of the three types of equipment for in-precinct voting had the following residual vote rates:
|Type of Voting Equipment||Residual Vote Rate|
A few words of caution are appropriate in considering the meaning of these numbers. First, not every residual vote represents an error on the part of the voter or the equipment. In some cases, voters may intentionally undervote - that is, they may choose to abstain from the presidential race. Surveys of past elections indicate that approximately 0.3 to 0.7% of voters choose not to vote for any candidate. If those numbers were the same in Ohio this year, it means that the substantial majority of those who cast residual votes didn't mean to do so.
It's also important to bear in mind that equipment isn't the only factor that can affect the residual vote rate. The demographic characteristics of certain counties can affect the rate of residual voting. For example, there's evidence that voters of lower income and education levels tend to have higher residual vote rates, particularly with punch cards. A study by Michael Tomz and Robert P. Van Houweling shows that there's a racial gap in residual voting - i.e., that there's a higher rate of residual voting in predominantly black precincts - with punch card and central-count optical scan equipment. Tomz and Van Houweling also show that the use of electronic voting machines virtually eliminates that gap.
A final point to keep in mind in reviewing the Ohio data is that, in some counties, absentee and provisional voters use a different type of equipment than other voters. For example, in Franklin County, most voters who appeared at their polling place used electronic voting machines. But absentee voters and those who cast provisional ballots used the punch cards in Franklin County .
Overall, Franklin County had a 1.45% residual vote rate in the 2004 election. At this time, we don't know for sure how many residual votes were cast using punch cards and how many were cast using electronic voting machines. It's very likely, however, that this residual vote rate is being driven up by provisional and absentee voters using punch cards. That was true in the 2000 election, when Franklin County had an overall residual vote rate of 0.89%. This overall number obscured the fact that Franklin County voters using electronic voting machines registered a very low 0.6% residual vote rate, while those using punch cards had a whopping residual vote rate of 3.6% - about six times as high.
For these reasons, countywide and statewide residual vote rates provide only a rough measure of voting system performance. Still, it's interesting to note that the high rate of residual voting with punch cards in Ohio is consistent with what other studies have found, including one by Henry Brady and his colleagues at the U.C. Berkeley Survey Research Center. That study also showed that residual vote rates decline when counties move from punch cards to electronic or precinct-count optical scan voting systems.
What's in Store for Ohio Voters
Secretary of State Blackwell has recently announced that precinct-count optical scan systems will be used as the state's primary voting device starting in 2006. Of course, the state had previously said that it would get rid of its punch card machines by 2004, but failed to do so. But if Ohio follows through on this plan, it will be good news for Ohio 's punch card counties, which can expect to see a higher percentage of their votes counted in the next election cycle.
Moving to precinct-count optical scan isn't necessarily the best thing for all voters. As Ruth Colker has noted, it's imperative that Ohio provide equal access for people with disabilities. For people with visual impairments in particular, electronic voting provides unparalleled advantages, since it has an audio feature that allows them to vote independently. The Help America Vote Act requires at least one accessible DRE unit at each polling place for voters with disabilities by 2006.
Moreover, it's not at all clear that precinct-count optical scans will lower residual vote rates, for those counties that are moving from lever or electronic voting machines. Lucas County 's experience provides an interesting example. Lucas County used a lever voting machine in 2000 and had an exceptionally low residual vote rate of 0.57% overall. In the 2004 presidential election, however, the county converted to a precinct-count optical scan system and saw its residual vote rate climb to 0.70%. That's still pretty low - but not as low as with the old-fashioned lever voting machine. Nevertheless, Ohio counties are required to replace punch card ballots, as a condition of the state's decision to accept funds under Title I of the Help America vote Act.
Still, precinct-count optical scan appears to do pretty well from the standpoint of accuracy. The three counties that used precinct-count optical scan as their primary method of voting in 2004 had a 0.84% residual vote rate. And there's no question that the 72% of Ohioans who used punch cards in Election 2004 will benefit from the replacement of this equipment, whether it's with precinct-count optical scan or electronic voting technology. The early data from this election provide further evidence that the "hanging chad" punch card should be retired, before it results in yet another disputed election.