Could punch-card voting skew elections in Ohio?
Tuesday, August 17, 2004

Article published Tuesday, August 17, 2004, at Columbus Dispatch.

Sampling of entire state refutes selective error-data

By John R. Lott Jr.

There is a corrosive perception that the voting system in parts of the United States systematically prevents people from voting and that this particularly discriminates against blacks. Litigation over punch-card voting machines tried unsuccessfully to derail the 2003 election in California to recall the governor, and now the American Civil Liberties Union has brought a similar lawsuit in Ohio.

The ACLU is not alone: As many as 18 percent of blacks nationally and 20 percent of 18- to 24-yearolds claim they donít believe their votes are counted accurately. Thus, this is a hard issue to ignore.

Since the 2000 Florida presidential election, the question has been how elections officials can prevent nonvoted, or so-called spoiled, ballots. These occur when voters either mark too many candidates (overvoting) in a race or do not vote for any candidate (undervoting).

Much of the debate centers on whether these nonvotes are intentional or the result of problems using punch-card machines.

Over the past three presidential elections in Ohio, punch cards have produced higher rates of nonvoted ballots than other voting machines do. Votomatic punch cards used in 69 of Ohioís 88 counties averaged a 2.4 percent nonvoted ballot rate. By comparison, electronic machines had a 1.1 percent rate, levers 1.5 percent and optical scans 2 percent.

The focus on the presidential race is understandable, given the experience in Florida, but it is also quite misleading. In races for Congress and state legislature, Votomatic machines actually do much better than electronic and lever machines and perform similarly to optical scans.

This result is natural, because voters simply donít know or care as much about other races as they do who wins the presidency. Interestingly, the drop-off in voting for other races is much less for punch cards than for other types of voting machines. For example, compared with the 1.3 percent difference between voting systems in presidential races, the nonvoted ballot rate for Ohio Senate races for Votomatic machines is almost 10 percent, while the rates for electronic and lever machines is 18 percent.

Even after accounting for factors that could affect nonvoted ballot rates, such as income and education and the number of candidates in a race, switching from Votomatic punch cards to electronic or lever machines would result in about 200 more nonvoted ballots in the average Ohio ward of 1,696 voters.

This pattern has held true for decades. Even an expert hired by the ACLU, professor Herb Asher at Ohio State University, also found that punch-card machines overall had much lower rates of nonvoted ballots than other machines during the 1978 election. Once this was clear, the ACLU did not call him to testify during the trial in July.

Why punch cards do so well down the ballot is simple. The more effort or time it takes to vote, the fewer races voters vote in. For example, recent research points to problems with the electronic machines regarding "the willingness of voters to navigate through multiple ballot screens before casting a vote (and) delays caused by the use of the review feature when coupled with extended ballots." Whatever their faults, punch cards are relatively quick and simple to use.

Most important for the ACLUís case, my research found that Votomatic machines were the only ones that consistently had lower nonvoted ballot rates for blacks than for whites. The Datavote punch-card machines that were used by only one county in 2000 and optical scans used by 11 counties were the worst for blacks, with electronic and lever machines varying, depending upon which race one examined.

Yet, even then, the race of voters only explains a small 0.4 percent to 3 percent of the variation in nonvoted ballot rates, itself an already small number.

With all the debate over voting machines, one would think that they must be too complicated for many people to figure out. But neither education nor income is related to nonvoted ballot rates. For education, the nonvoted ballot rate is high for those with less than a ninth-grade education, low for those with some high school, high for highschool graduates, low for college graduates and generally higher again for those with post-graduate degrees. Instead of deep conspiracy theories, some voters were probably more conflicted over whom to vote for and decided not to support anyone in some races.

My data, as well as Asherís, examine the entire state, not just a few counties, as all the experts the ACLU used did. Nor has the ACLU ever explained exactly why it thinks that blacks have a more difficult time using punch cards.

The ACLUís lawsuit seems designed to maximize confusion, not just in Ohio but across the nation. If it wins, then any close election at least could be challenged in the press.

But whatever short-term political gains, the unfounded claims of selective disenfranchisement risk poisoning the political debate for years to come.

John R. Lott Jr., a resident scholar at the American Enterprise Institute, served as a statistical expert for the U.S. Commission on Civil Rights and USA Today in their evaluations of the 2000 election in Florida.†

The other side of the debate is presented here:

Yes: Miscast ballots easily could decide close races
Tuesday, August 17, 2004

There is a real possibility that the 2000 voting debacle in Florida will be replayed in Ohio come Nov. 2.

Thatís because about 70 percent of Ohio voters will use the punch-card machines similar to those that caused such "hangingchad" turmoil four years ago. The votes of many of these people will not be counted. As Ohio Secretary of State J. Kenneth Blackwell has acknowledged, the 29 counties with the highest rates of uncounted votes in the 2000 presidential election all used punch cards; the seven counties with the lowest rates used other equipment.

Punch cards result in a large number of lost votes. For example, voters using punch cards in Montgomery County are more than four times more likely not to have their votes counted than a voter in Franklin County, which uses more-modern voting technology. Statewide, at least 55,000 presidential votes will be lost because of unreliable voting equipment.

People of all races will see their votes denied because of punch cards, but blacks will be especially hard hit. Largely because of the educational disparities in the state, punch cards generate a substantial "racial gap," resulting in an unusually high number of lost votes in heavily black precincts. In effect, the punch-card system functions as a sort of technological literacy test.

Worse still, in a close election, the punch cardís margin of error could exceed the margin of victory. Given that Ohio is a key swing state, those errors could decide who occupies the White House for the next four years.

How did we end up in this predicament? That question has many experts scratching their heads.

The public wasnít aware of the problems with punch cards until the 2000 election, but a handful of social scientists had studied them long before that. One expert, Roy Saltman of the National Bureau of Standards, recommended as early as 1988 that the punch-card system be abandoned.

Since 2000, study after study has found that punch cards are unreliable. Some states have replaced their systems with better technology. Other states, including Ohio, have failed to do so, leaving concerned citizens with no choice but to turn to the courts.

In 2002, the American Civil Liberties Union brought suit in Ohio, on behalf of some of those people. The ACLU seeks to have punch-card systems replaced with more reliable technology that will prevent mistaken "overvotes" and give voters the opportunity to check for errors. There are electronic and paper-based systems that have this feature, commonly known as second-chance voting, because it allows voters to verify their choices before casting their ballots.

Concerns over electronic voting are no excuse to keep using punch cards. If counties are worried about electronic-voting security, they may choose a paper-based system that allows second-chance voting.

Lack of funds isnít an excuse, either. Ohio already has received $132 million in federal funds to upgrade its voting system. And to his credit, Blackwell has acknowledged that punch cards need to be replaced with second-chance systems.

As he put it: "In a study of Ďoverí and Ďunderí voting in Ohio, it was clearly demonstrated that punch-card voting was unreliable to the extent votes cast by thousands of Ohioans were not being counted in the final election tabulation."

Blackwell has also written: "With Ohio slated by both national parties as a battleground state, the possibility of a close election with punch cards as the stateís primary voting device invites a Florida-like calamity."

Blackwellís admissions are consistent with the findings of experts throughout the country. Ironically, the lone exception is a report provided by the stateís paid expert, John R. Lott Jr., of the right-leaning American Enterprise Institute. Lott concedes that punch cards do poorly in presidential contests but he concludes that they do better than some other systems in so-called downballot races, such as state legislative contests.

The problem with this analysis is that voters in different parts of the state have different candidates on the ballot. Some of these races are competitive, while others are not. In some but not others, candidates are running unopposed. Some races involve popular candidates, while others do not. In short, thereís no way of telling whether voters intentionally have chosen to abstain in downballot races. Itís therefore fallacious to attribute the nonvoted ballots in those races to newer voting technology.

Ohio, which promised last year that punch cards would be gone by this election, would have been better off upgrading its voting technology instead of spending thousands of taxpayersí dollars for a flawed report.

Whatís unclear is whether Ohioans will be forced to continue using punch cards in 2005 and 2006.

The trial of the ACLUís case was continued last month and is to resume in late September. We only can hope that this case results in a court order that will ensure that 2004 is the last time that our votes are left hanging by a chad.

Daniel P. Tokaji is an assistant professor of law at Ohio State University and a member of the legal team in the American Civil Liberties Unionís challenge to punch-card voting.

tokaji . 1†

Home (description of book, downloadable data sets, and discussions of previous controversies)

Academic papers:

Social Science Research Network

Book Reviews:

For a list of book reviews on The Bias Against Guns, click here.

List of my Op-eds

Posts by topic

Research finding a drop in violent crime rates from Right-to-carry laws

Appalachian law school attack

Baghdad murder rate

Arming Pilots

Fraudulent website pretending to be run by me

The Merced Pitchfork Killings and Vin Suprynowicz's quote

Ayres and Donohue

Stanford Law Review

Mother Jones article


Craig Newmark

Eric Rasmusen

William Sjostrom

Dr. T's

Interview with National Review Online

Lyonette Louis-Jacques's page on Firearms Regulation Worldwide

The End of Myth: An Interview with Dr. John Lott

Cold Comfort, Economist John Lott discusses the benefits of guns--and the hazards of pointing them out.

An interview with John R. Lott, Jr. author of More Guns, Less Crime: Understanding Crime and Gun Control Laws

Some data not found at

Updated Media Analysis of Appalachian Law School Attack

Since the first news search was done additional news stories have been added to Nexis:

There are thus now 218 unique stories, and a total of 294 stories counting duplicates (the stories in yellow were duplicates): Excel file for general overview and specific stories. Explicit mentions of defensive gun use increase from 2 to 3 now.

Journal of Legal Studies paper on spoiled ballots during the 2000 Presidential Election

Data set from USA Today, STATA 7.0 data set

"Do" File for some of the basic regressions from the paper