Sept. 1, 2005
Residual votes declined in Florida, Michigan with updated voting equipment
WASHINGTON D.C.—Aided by new technology, voting behavior improved considerably in Florida and Michigan based on residual votes declining in 2004 compared with 2000, a new study co-authored by University of Michigan researchers indicated.
The study, funded by the National Science Foundation, is being presented today (Sept. 1) at the American Political Science Association annual conference. The four-day conference ends Sunday in Washington.
"The new voting technology is doing the job it is supposed to by reducing the number of votes that are lost or don't get counted. And it does not seem to produce any unintended consequences in terms of changes in other forms of voting behavior," said Michael Traugott, a senior research scientist at the Institute for Social Research and a professor in the Department of Communication Studies. The paper is one of several voting projects spearheaded by Traugott and other researchers.
"The Impact of Voting Systems on Residual Votes, Incomplete Ballots and Other Measures of Voting Behavior" was co-authored by Fred Conrad, U-M; Michael Hanmer, Georgetown University; Won-ho Park, University of Florida; Paul Herrnson and Benjamin Bederson, University of Maryland; and Richard Niemi, University of Rochester. Hanmer and Park are former U-M political science graduate students who worked on this three-year project.
The number and type of problems—from faulty voting equipment to untrained poll workers—experienced in the 2000 presidential election resulted in the need for new technology to remedy many of them. Nearly three years ago, President Bush signed the Ney/Hoyer Help America Vote Act (HAVA) into law, requiring states to replace outdated voting systems—mainly lever and punch-card voting machines—by 2006.
The U-M research examined whether those improvements put in place by the 2004 presidential election were working, based on two case studies: Florida and Michigan. These states were chosen because they represented different examples of the changes occurring in election administration. Florida had severe election administration problems in 2000, resulting in extensive reforms and technologies adopted in many counties. By the 2004 election, punch cards, lever machines and paper ballots had been eliminated for optical scan and direct-recording devices.
Michigan has the most decentralized system of election administration of any state in the country. The state has developed a plan to move to some form of optical character recognition or optical scanning system in compliance with HAVA. During the study, however, some Michigan jurisdictions still used paper ballots, punch cards and lever machines.
Presidential and Senate election results in 2000 and 2004 were compared in both states. In selected Florida counties, the residual rate—or uncounted votes based on over votes for one or more offices, ballots mismarked or other forms of "spoiled" ballots—declined in 2004. For the presidential race, the residual votes dropped more than 90 percent, from 5.24 percent in 2000 to 0.47 percent in 2004. The residual decline in Florida's Senate results was modest, from 6.43 percent in 2000 to 3.16 percent in 2004.
"This is not surprising, since voters rolling off down the ballot always result in residual vote rates being higher for the Senate," Traugott said. A "roll off" is the measure of how many voters who cast ballots for the highest office on the ballot do not vote for offices or issues that appear later in the ballot.
In all of the Michigan cities and townships, the residual rate dipped slightly from 1.4 percent in the presidential race in 2000 to 0.94 percent in 2004. Michigan did not have a Senate race in 2004, but the 2000 residual rate for that office was 3.46 percent.
In analyzing the voting equipment used in both states, residual votes were often high when punch cards were used. When punch cards, as well as paper ballots, were changed to newer technology—such as optical scanning systems—the residual declined significantly, the research showed.
The effect of equipment type on "roll-off" patterns across different offices and issues also was studied, dividing the total number of valid votes by the total of votes cast in the given election. In Florida, optical scanning systems showed the highest level of turnout in 2000 than any other machine throughout all races, except for the office of president. In addition, as voters move down the ballot to lower profile offices and judicial races, the difference became larger.
In 2004, two years after many Florida counties changed their voting system to optical scan or direct-recording electronic devices, the roll off was generally lower. The 2004 general election in Florida didn't have state cabinet elections, but eight constitutional amendment proposals appeared on the ballot.
The researchers also noted patterns when analyzing the machine types in Michigan. For example, in 2000 and 2004, roll off was substantially higher for lever machines than others.
Finally, the researchers said changing to new technologies did not lessen the need for research on the effects of voter education and training of poll workers, as well as programs designed to increase the general awareness of voting devices.
For additional information on Traugott, visit http://www.ns.umich.edu/htdocs/public/experts/ExpDisplay.php?ExpID=766
Contact: Jared Wadley
Phone: (734) 936-7819