Voting Machines, Race, and Equal Protection

Abstract

61 IMAGINE taking a test for which there are five different graders, and your score depends on which grader corrects your exam. You may feel the test capricious and unfair, especially if you received a poor grade. That is the situation with voting in the United States today. The United States employs five different types of voting technology: handcounted paper ballots, punch cards, optically scanned paper ballots, lever machines, and electronic voting machines (also known as direct recording electronic machines, or DREs for short). The incidence of unmarked, uncounted, and spoiled ballots—called residual votes—depends on which equipment a locality uses. The presidential election contest in Florida in 2000 exposed an important weakness with some of these technologies, especially with punch card ballots. Subsequent study has shown that over the last four presidential elections punch cards and electronic voting machines have produced higher rates of unmarked and unrecorded ballots than hand-counted paper ballots, optically scanned paper ballots, and lever machines.1 It is unclear whether the equal protection standard in Bush v. Gore requires uniform technologies.2 The first legal challenges to the nonuniform system of voting equipment are underway. One of these complaints, Common Cause v. Jones,3 invokes Bush v. Gore to test the equal protection claim generally and makes a specific claim for equal protection of racial groups. The general equal protection claim holds that different technologies count votes differently. Specifically, punch cards may have higher rates of uncounted votes and spoiled ballots. Voters residing in counties using punch card technologies are less likely to have their votes counted. A number of studies have addressed this matter, for example the Caltech/MIT Report, and this claim appears correct. In this paper, I consider the second argument—equal protection of racial groups. This argument rests on two premises. First, punch cards are an inferior technology for casting and counting votes. Second, counties with higher densities of minority voters disproportionately use inferior technologies, especially punch cards. A more subtle version of the racial argument holds that some technologies might be particularly problematic in areas with high densities of minorities. Some technologies may have disproportionately high rates of uncounted, unmarked, and spoiled ballots in counties with high numbers of nonwhite voters, owing to the particular difficulty implementing the equipment in those counties. These race-based claims may be relevant not only for the equal protection argument, but also for challenges under the Voting Rights Act. Is there statistical evidence to support the race-based equality argument? This paper examines the statistical evidence for the argument that race and voting equipment together affect the casting and counting of votes in ways disadvantageous to minority voters. I focus on