Ballot order effects

All states have laws that mandate the order in which candidates appear on the ballot. If the order in which candidates appear on the ballot influences election outcomes, then these laws will advantage some candidates over others, whether this advantage is due to incumbency, party, or alphabetical order. Concern over the advantage presumably conveyed by “ballot-order effects” has led some states to require names to be randomly listed on the ballot.

This explainer was last updated on April 20, 2022.

Background

A broad range of state and local laws govern the order in which candidates are presented to voters on the ballot. Depending on where in the United States they vote, the first candidates that voters see on their ballots may be incumbents, representatives of the two major political parties, representatives of the party that won the largest vote share for governor in the previous election, the first to have filed their paperwork to run, candidates with names at the beginning of the alphabet, or a random selection from among the candidates running. (Table 7 of a working paper by Daniel Ho and Kosuke Imai contains an instructive summary of ballot ordering conventions by state as of 2004.)

These approaches to ordering candidates on the ballot may influence election outcomes and have been the subject of some controversy. For researchers and policymakers alike, the relevant question is whether appearing earlier on the ballot improves a candidate’s chances of winning an election. To avoid conferring electoral advantages upon the first-listed candidates, some states use randomized ballot ordering.

Academic research into ballot-order effects has mostly confirmed common-sense beliefs on the subject: that appearing first increased a candidate’s chances of performing well in an election, and that incumbents often appeared first. However, this research has also suggested that ballot-order effects are probably not equally strong across types of elections and types of candidates. Furthermore, the cognitive mechanism that leads to ballot-order effects, if they do exist, are unclear.

California

In 1975, California moved from listing incumbents first on the ballot to listing candidates based on a randomized order. Under this approach, candidates are listed in alphabetical order, but instead of using the traditional alphabet, state officials list candidates according to an alphabet generated by drawing 26 letters in random order (Cal. Elec. Code § 13112(a) (2003)). California’s supreme court outlined the rationale for this change in Gould v. Grubb, holding that “a significant advantage accrues to a candidate by virtue of a top ballot position.” The court cited work by James W. Scott, who studied 10 districts in California without incumbents running and concluded that approximately 5 percent of each candidate’s vote share could be attributed to their position on the ballot. The court required statewide elections to adopt a process that gave every candidate an equal chance of appearing first on the ballot.

Today, many states list candidates on the ballot randomly, although there does not appear to be a comprehensive listing of which states randomize ballot names and how the randomization is implemented.

Academic Research on Ballot Order Effects

For decades, researchers have studied whether the order in which voters see candidates on the ballot has any impact on election results. Some scholars have found that being listed first on the ballot translates into an electoral advantage, while others have found little or no evidence of a relationship between where a candidate appears on the ballot and the final election results.

A series of studies have presented evidence for ballot order effects, studying a variety of U.S. elections. The following is a sample of this research.

Joanne E. Miller and Jon Krosnick found that House and Senate candidates in the 1992 and 2000 elections in Ohio benefited from significant ballot order effects.
Krosnick, Miller, and Michael Tichy found in 2003 that appearing first on the ballot in the 2000 presidential election increased George W. Bush’s vote share by almost 10 percentage points relative to appearing last.
Jonathan Koppell and Jennifer Steen studied the 1998 Democratic primary election in New York City, where the order in which candidates were listed on the ballot rotated by district. They found that, in almost 90% of precincts, candidates received a larger share of the vote when listed first on the ballot relative to being listed in any other position. In almost 10% of cases where candidates listed first got a bump in vote share, that bump was larger than the winner’s margin of victory
Examining elections for city council and school board seats across California, Marc Meredith and Yuval Salant found that candidates listed first win between four and five percent more than they might be expected to if they were listed later on the ballot. This effect appeared consistently across offices and was not sensitive to whether elections were timed to coincide with statewide elections.

Some scholars, however, caution against interpreting the results of these studies as evidence that being first on the ballot always results in an unfair electoral advantage. Some have pointed out that the vast majority of studies concerning ballot order effects concentrate on the effect of being first, and ignore the implications of most other ballot positions. This leaves open the possibility that, while the difference between being listed first and being listed last may be large, the difference between being listed first and being listed second may be relatively small.

For instance, in their study of the 1998 California General election, R. Michael Alvarez, Betsy Sinclair, and Richard L. Hasen found that some candidates lost votes by appearing first on the ballot and that candidates listed at the bottom of the ballot won additional votes as frequently as candidates who were listed first.

In their work on California statewide elections, Ho and Imai reported that, in general elections, ballot order only seems to impact candidates from minor parties. Candidates from major parties appear unaffected. Their results for primary elections, however, still suggest substantial ballot order effects exist, especially for minor party candidates and candidates running in nonpartisan races.

Another important vein of academic research on ballot order effects focuses on why voters might choose candidates listed first over candidates listed later in the ballot order. Research that reaches back more than half a century has shown, for instance, that students taking multiple-choice exams tend to guess from among the first few of their listed options.

In the context of politics, scholars have suggested that voters with little to no information on the candidates in a particular race search their memories for anything that might compel them to vote for a particular candidate. Voters start at the top of the list and repeat this for candidates in order, but tend to get tired of repeating this process for large numbers of candidates. As a result, the candidates at the top of the list tend to be the candidates for whom voters have found the most reasons to vote. Another explanation holds that, if the cost of making a mistake is low, voters simply settle for the first option in a list because they cannot think of a clear reason not to do so. These mechanisms are extremely difficult to test, and some scholars have suggested that the explanation that voters simply choose the first candidate they cannot think of a reason not to choose is not enough to explain the ballot order effects we see in practice.

Search

Websites