- Roughly 13% of Republican and Democrat voters are misled by such ballot designs, leading them to cast votes against their intentions.
Authors of a new study on ballot design conclude that variable ballot orders and party labeling patterns mislead a substantial portion of North Carolina voters.
Researchers at Duke University’s Fuqua School of Business, colleagues Alessandro Arlotto, Alexandre Belloni and Saša Pekeč— along with Fei Fang of Yale University — examined public North Carolina election data from 2004 through 2020 to determine if ballot design features, such as flipping partisan orders or varying party labels, could affect election outcomes. According to their paper —Ballot Design and Electoral Outcomes: The Role of Candidate Order and Party Affiliation — roughly 13% of Republican and Democrat voters are misled by such ballot designs, leading them to cast votes against their intentions.
“In our analysis, we focus on ballot cues and examine whether partisan voters—those who vote by party even in nonpartisan contests—use the party order from a contest with party designations (i.e., the sequence in which parties in that contest appear on the ballot) to inform their voting in nonpartisan contests,” reads the working paper’s introduction.”If partisan voters cast their ballots assuming that the party orders of partisan and nonpartisan contests are the same, then their votes would not represent their intent when the two party orders actually differ.”
The effect can be seen in stark relief when comparing results of the 2016 presidential race between Donald Trump and Hillary Clinton, and those of the 2016 race for NC Supreme Court between Bob Edmunds and Michael Morgan, before all statewide judicial races carried partisan labels in North Carolina.
The races were on the same ballot, but with a flipped partisan ordering. The presidential candidates were listed on the ballot with their party labels: “Donald J. Trump (R)” was listed first, followed by “Hillary Clinton (D).” The relevant state laws prescribed that candidates from the party of the sitting governor (Republican Gov. Pat McCrory at the time) would be listed first in all partisan statewide races.
However, the Supreme Court candidates were listed without party labels, with Michael R. Morgan—a Democrat— listed first, and Robert H. Edmunds—a Republican— listed second.
Whereas the Republican, Trump (listed first), won North Carolina by 3.5%, the Republican Edmunds (listed second) lost to Morgan by nearly nine points.
The result had major implications for control of North Carolina’s highest court. Edmunds was running as an incumbent to keep his associate seat and maintain Republicans’ 4-3 majority, and Morgan’s victory flipped the court to majority-Democrat, setting the scene for partisan court conflicts between the branches of state government.
Was there something unique about Edmunds or Morgan that contributed toward such a departure in partisan outcomes? Not likely; the researchers found that adding party labels erased the “flip effect” entirely. So, control of the Tar Heel State’s high court may have been determined by ballot order mistakes cast by otherwise loyal partisan voters.
“Taken together, our results from judicial races with and without party designations suggest that including party designations for all contests on the ballot helps correctly capture the intent of the electorate,” state the authors in the paper.
Overall, the authors estimate about 12% of Democrats and nearly 14% of Republican partisan voters cast their votes incorrectly due only to the flip.
Notably, state lawmakers recognized the complications arising from judicial races’ lack of partisan labels, passing legislation in 2017 to include partisan affiliation for judicial races.
Ballot (re)design
Voters might be surprised how variable ballot designs have been over the years.
The researchers note that in the general elections of 2004, 2008, and 2012 presidential candidates along with their party affiliations were arranged alphabetically by party. Therefore, ballots featured the Democratic presidential candidate first, the Republican candidate second, followed by third-party candidates (in alphabetical order) and write-ins. In contrast, in those same elections statewide judicial candidates were arranged alphabetically by last name, without party designation included on the ballot, leading to the potential for the “flipped” scenario.
While lawmakers did add partisan labels for statewide Court of Appeals races, they did not for Supreme Court; that race was listed in alphabetical order, starting with H (as randomly selected), with no labels.
Following that election, lawmakers added partisan labels for all statewide judicial races, including the Supreme Court. Changes also required all candidates in any election ballot to be listed in either alphabetical or reverse-alphabetical order, according to a drawing conducted by the State Board of Elections.
The study authors estimated that, between 2004-16, a party order flip caused about 13% of the votes in the North Carolina judicial races to be mistakenly cast for the opposing party’s judicial candidate.
The addition of partisan labeling has coincided with Republican judicial candidates winning majorities on the NC Court of Appeals and NC Supreme Court in the elections since. The 2024 general election will again feature presidential and partisan judicial races, notably the lone race for Supreme Court between Republican Jefferson Griffin and Democrat Allison Riggs.
While those races now have partisan labels, many down-ballot races, such as contests for school boards or county commissions, will not. Though the net effect of mistaken partisan voting is often a statistical wash — Democrats and Republicans cast the mistaken votes about 13% of the time — the 2016 example shows how the flipping effect could prove politically consequential.
In 2024, such flipped down-ballot races could lead to net effects that make the difference between election victory or defeat in a close race.
“We show that electoral contests without party designations, appearing on ballots alongside races with party designations, can mislead a significant number of voters,” write the authors in the study’s conclusion. “Consequently, the outcomes of such races without party labels might not reflect the true preferences of the electorate.”
