Paul S. Herrnson, Michael J. Hanmer, and Benjamin B. Bederson (University of Maryland)
Richard G. Niemi (University of Rochester)
Frederick G. Conrad and Michael W. Traugott (University of Michigan)

October 2007

Many election jurisdictions across the United States offer voters a “straight-party” option, whereby an individual can cast a vote for all candidates in a given party at once, whether it be with one touch on a direct recording electronic (DRE) voting system screen, by filling in one oval, or connecting one arrow on a paper ballot. Straight-party options, motivated primarily by partisan interests, were designed to discourage voters from splitting their tickets or from failing to vote for offices below the top of the ticket. In this respect, the features have been a success. However, with regard to the voters’ ability to use the straight-party option to vote with confidence and without errors, the jury is still out.

As part of our research on new voting systems (to be published in the forthcoming Voting Technology: The Not-So-Simple Act of Casting a Ballot, Brookings Institution Press, 2007), we performed an analysis of the straight-party option, with a focus on voters’ ability to cast their votes as intended, satisfaction with the voting system, and ability to vote without asking for help.

The study investigated six voting systems:

• Diebold AccuVote-TS (a touch-screen DRE)
• ES&S Model 100 (an optically scanned paper ballot)
• Hart Intercivic eSlate (a dial and buttons DRE)
• Nedap LibertyVote (a full face membrane button DRE)
• Avante Vote-Trakker (a touch-screen DRE with voter verifiable paper trail)
• Zoomable Prototype (a touch-screen DRE with ability to zoom in or out on the ballot)

We recruited a diverse pool of 1,540 individuals in Maryland, Michigan, and New York and conducted a mock election asking them to vote for the same candidates and issues on each of the six voting systems and, after voting on a given system, to answer questions about that system. The ballot—a realistic combination of partisan and nonpartisan offices and ballot propositions—was the same for all voters, except that some used the standard office bloc format while others contained a straight-party option (or a ballot organized on the basis of party in the case of the Nedap, which could not then be programmed with a straight-party feature).

Though the straight-party option was not intended as a way to reduce voter errors, it might nevertheless have this effect. Assuming the single required mark or touch is done correctly, it eliminates the possibility of a single oversight, slips, or other mistake affecting what may be more than a dozen other votes. Yet a reduction in errors is not guaranteed. We found that many voters do not understand the concept of a straight-party vote, perhaps because they were never properly introduced to the idea. Most civics classes, after all, emphasize that citizens should “vote for the person, not the party” and are unlikely to discuss a nuance of voting that is directly contradictory to this lesson. Moreover, because only about fifteen states offer voters a straight-party option, just a fraction of the electorate has been exposed to a ballot with such a feature.

One of the most serious, but fortunately rare, errors was when voters who used the straight-party option accidentally selected the political party for which they did not intend to vote. Regardless of whether this was due to a slip of the finger on the three touch-screen systems, an inability to accurately negotiate the dial and buttons on the Hart InterCivic system, or plain confusion on what a straight-party vote means when filling in the oval on the ES&S paper ballot, this error has considerable consequences for the voter. It highlights one of the pitfalls of ballots that incorporate a straight-party option—a single incorrect action can result in one’s voting incorrectly for many offices.

Our research shows that the presence of a straight-party feature lowered voter accuracy both for partisan offices and also for nonpartisan offices and ballot propositions. This increase in errors was partly a result of individuals not voting for the nonpartisan offices and/or propositions because they apparently thought that the straight-party mechanism applied to the entire ballot, despite instructions to the contrary. Others who completed the ballot made different errors. For example, on the vote for probate judge in our simulated election (an office for which our instructions asked voters to change their votes after initially casting them), individuals voted with less accuracy on the ballot with the straight-party feature than on the ballot without it, even though the feature had nothing to do with this nonpartisan office. This dynamic serves as a reminder that there are likely to be spillover effects when voters are confronted with nonstandard tasks or complex ballots.

Voters who used a straight-party option were also confronted with another complication—referred to as casting a straight-party vote with exception. In our experiments, voters were asked to cast a straight-party Republican ticket but then change their vote for the U.S. House to a vote for the Democratic candidate. This “exception” procedure, available in a number of states, results in the vote for the individual office overriding the straight-party vote. Many voters appeared confused and failed to properly complete this task; as a result, voter accuracy plummeted to about 50 percent on one of the systems. Some voters, no doubt, simply forgot what they were supposed to do and their mistake was in not following instructions, rather than trying to follow them and failing. Nonetheless, the evidence shows that many voters are derailed by complexities involving the straight-party option and that the interaction between ballot style and voting interface can exacerbate the situation.

The complexity of the straight-party option led to only slightly lower levels of voter satisfaction, and, even then, not on all the systems. However, voters using the straight-party option were much more likely to request help, especially on the paper/ballot optical scan system. The types of help for which voters asked when using this system, moreover, provide evidence of voter confusion about both the concept and procedure for casting a straight-party vote. Perhaps the strongest proof of the difficulties voters encountered was provided by the paper ballots themselves, as many voters filled in the oval for straight-party Republican and then filled in the oval for every Republican candidate—an unnecessary and redundant effort. Voters are evidently more confused by a paper-based system that gives them the opportunity to vote straight party and then also to vote individually for a party’s candidates than they are by a DRE system that allows them to vote straight party and then automatically advances them down the ballot to nonpartisan offices. We found this across all sites in the field studies, including those in Michigan, where, in actual elections, voters have a straight-party option and most voters use a paper ballot/optical scan system.

Overall, our results suggest that debates over the inclusion of the straight-party option should be extended beyond the usual arguments about partisan advantage to include concern about the confusion that could result when voters use these types of ballots. Our findings also suggest that instruction in the mechanics of voting is an essential part of civic education and that it should include teaching not just about why voters should vote, but also about the types of ballots and interfaces they will encounter in elections.

Paul Herrnson is a Professor and Michael Hanmer an Assistant Professor in the Department of Government and Politics at the University of Maryland. Benjamin Bederson is an Associate Professor in the Department of Computer Science at the University of Maryland. Richard Niemi is the Don Alonzo Watson Professor of Political Science at the University of Rochester. Frederick Conrad is an Associate Research Scientist at the Institute for Social Research and Michael Traugott a Professor of Communication Studies and an Adjunct Professor of Political Science at the University of Michigan.

Viewpoint is an occasional feature analyzing various election reform issues.