Ballot Order Favors First Names by 10%; Let’s Rotate!

South Dakota’s elections will be rigged by the luck of draw. NPR this morning highlights research on ballot order that supports results political scientists have reported for yearsappearing first on the ballot helps a candidate win more votes. NPR cites Sam Houston State University professor Darren Grant, who finds that in down-ticket races among less well-known candidates, “going from last to first on the ballot raises a candidate’s vote share by nearly ten percentage points.”

South Dakota’s election officials draw lots to determine ballot order. On general election ballots, the Secretary of State draws lots for each party, and candidates in all partisan races appear in the order drawn for their parties. This institutional choice thus randomly assigns up to a ten-percent advantage to certain candidates and parties.

No state action should place a thumb on the electoral scale. South Dakota should reform its ballot rules to eliminate this first-slot bias and more fairly and accurately measure the will of the voters. But how do we erase this first-slot bias? A ballot has to list the candidates in some order, so no matter what, on each individual ballot, candidate #1 will enjoy an advantage over candidate #2.

The practical solution is to distribute different ballots to different voters. Dr. Grant writes that Ohio, New York, and North Dakota “use a precinct-level rotation system, moving the first name to last in each successive precinct.” Different ballots for each precinct (or, in some South Dakota communities, each voting center, where anyone from any precinct can stop by to vote) give each candidate a shot at being first, thus canceling out the advantage. However, depending on the math, some candidates would still get an edge. In a district with an odd number of precincts (District 9, District 28, District 35…), a two-candidate race would give one candidate the first-slot boost in one more precinct than the other. In any district where the number of precincts is not a multiple of the number of candidates (for example, where there are eight precincts and three candidates, or sixteen precincts and five candidates, or, as in District 28, a prime number of precincts, which never works out!), some candidate or candidates will still enjoy some first-slot advantage. A potential 10% lucky edge in just one precinct is fairer than a 10% lucky edge in every precinct, but it’s still not truly fair.

The Secretary of State could rotate candidate order by ballot. Within one box of ballots, the first would list candidates in the drawn party order, say DEM–LIB–GOP–CON. The next sheet in the box would list all candidates LIB–GOP–CON–DEM, the next GOP–CON–DEM–LIB, and so on. Each voter who comes to any precinct table would get a different ballot from what the preceding voter got, thus canceling out the first-slot bias for all but the last couple-three voters.

Counting such rotated or randomized ballots poses practical challenges. Human canvassers couldn’t shorthand their scans and check the same candidate’s name every time they see Bubble #1 filled; eyeballers would have to slow down and read each ballot, which invites delay and error. Optical scanners would also need more information: we’d either have to sort all the ballots into separate stacks—one pile for the DEM-first ballots, one for the GOP-first…—and reprogram the optical scanners for each run, or, more simply, we’d need to include one extra column on the ballot indicating candidate/party order to the scanner.

We could save the Secretary of State any complicated printing by computerizing all ballots: put touch screens in every polling place, and have the computer rotate the names for each voter. However, we can’t mail iPads to absentee voters (remember, previous Secretary of State Jason Gant couldn’t even keep track of thirty iPads in his office). Some voters still need paper ballots. I also hear a fair amount of voting-machine angst from voters. They want less black-box technology, not more, taking and counting their votes.

Reducing first-slot bias complicates ballot printing and counting, but we can do it. For the sake of more fair elections, we should do it.

13 Responses to Ballot Order Favors First Names by 10%; Let’s Rotate!

  1. Troy Jones

    Wow, 10% is a big number. If true, because the Democrats were listed first, here is who wouldn’t be in the legislature from the last election:

    Scott Parsley, Dem from Madison
    Bernie Hunhoff, Dem from Yankton (no wonder he decided not to run)
    Paula Hawks, Dem from Hartford (Probably wouldn’t be your US House nominee either)

    Do you really want to contemplate more Republicans in the Legislature and how much worse Wismer, Weiland, and Robinson would have been beat if they hadn’t been first on the ballot?

  2. There is a Dr. Gant who studies this ballot pedia stuff? That is just insaner than most. No relation to Jabba the Gant, I can only hope. The whole thing is just silly. People need to take a little responsibility and color the dot next to the fellow they want to elect. If you cant do that you are too stupid to vote.

  3. Douglas Wiken

    Or subtract 10% of votes from first on the list.

    Actually, I kind of doubt that first on the ballot may only have an impact of initiated and referred votes where there are multiple choices.

  4. Douglas Wiken

    I meant to say I kind of doubt being first on the ballot has much impact except for initiated and referenda when there is a bunch of them to read and that may not change the ratio between pro and con.

  5. Troy, yes, I really do want to take that chance. Pointing to one election where the luck of the draw favors us ignores the next election where it favors our opponents. I don’t want anyone of any party winning office on blind luck. I don’t want any institutional bias changing the outcome of our elections. I want every election to reflect as accurately as possible the will of the people.

    I take the same stance on Amendment T: when I am Majority Leader after the Census, I will wish we Democrats could carve up the map to dilute Republican votes and solidify the Democratic grip on the Legislature. But I’ll still vote for independent redistricting on principle.

    Ditto Amendment V: an open non-partisan primary will box Dems out and put two Republicans on the general election ballot in some districts. I can tolerate that possibility, because the open nonpartisan primary gives more people an opportunity to vote.

    Principle is more important than occasional partisan advantage.

  6. Douglas, the new research cited does not address ballot questions, only candidates. Grant’s findings on the ballot-order effect would suggest that “Yes” should have an advantage over “No” on ballot measures, since we always list responses in that order. Grant doesn’t speak to ballot-order in terms of which measure is listed first, just order within each ballot line. He does find that ballot order has a greater effect in races that appear down-ticket, but that appears to be less a product of voter fatigue and more a product of voters having less information about the down-ticket races. Stepping out on a shaky limb, I would speculate that the order in which ballot measures are listed will have less effect on the Yes/No split for each one than will the amount of information voters have on each measure. In the case of candidates, down-ticket position is often a good proxy for voters’ information level: Commissioner of School and Public Lands is toward the bottom, and almost nobody knows who’s running for it or what the office does. With ballot measures, ballot position is determined by the order in which sponsors get their petitions in and thus is unlikely to have any correlation with how much voters know about each measure. IM 23 may be the last item on the ballot, but if the unions spend big bucks on their TV ads, it may have the highest top-of-mind awareness of all ten ballot measures.

  7. The statements highlighted are ripe for Election Fraud:

    “We could save the Secretary of State any complicated printing by computerizing all ballots: put touch screens in every polling place, and have the computer rotate the names for each voter.”

    “Some voters still need paper ballots. I also hear a fair amount of voting-machine angst from voters. They want less black-box technology, not more, taking and counting their votes.”

    More electronics in voting will never make the election outcomes change for the good. Never. I have been involved for 50 years and never saw any better voting outcomes by adding electronics to the physical process of voting. Never accept touchscreen anything in the voting booth except ADA. Only accept ADA if it can be verified.

    Voting rights are important. We need to teach civics and end the idea of lazy voting by encouraging civic responsibility. There is no winner takes all in civic responsibility. Governing should never be a super bowl contest. Politics and governing are not the same thing. We all must have a voice in the process. The more our politicians make it “my way or the highway” the easier it is for the voter to take the highway.

    So the above statements only encourage the operators and operatives of the elections to claim easier voting encourages more voters when in fact it will make it easier to steal elections away from the voters. No one teaches the consequences of lazy voting.

    Well, hello Mr. Trump, how are you and the Clintons today?

    Lazy voting will destroy our democracy and encouraging it will only make it fail faster.

  8. Troy Jones


    I guess my points weren’t clear.

    1) I’m with Doug. I don’t think ballot order makes any difference and certainly not 10%. The assertion it is 10% makes me discount the analysis in whole.

    2) If I were in your shoes (your candidate’s getting trounced despite this “advantage), I wouldn’t admit how far removed your candidates are from average South Dakotans.

  9. Bruce, voter education and participation are still my preferred solutions to any bias in the election system. Still, if we can identify a factor like ballot order that could be putting an unearned 10% in some candidates’ columns, and if we can fix that problem with simple changes in how we print ballots, we should make those changes.

    Troy, you can’t just assert away an analysis that challenges your assumptions. The data and analysis are there. They say what they say. You can say you don’t like it, but to justify rejecting policy changes to address that proven problem, you need to counter with data and analysis of at least equal rigor to support your assertion.

  10. Cory,

    If the SoS already draws lots for whose name appears first, the system is already addressing the first-name-favor bias, at least at the party level on a longitudinal basis.

    There is no state favortism, so everyone gets equal protection under the law.

    The rabbit hole you’re chasing is to eliminate the first-name-favor bias in individual elections. I don’t in principle think that’s a bad idea, but given the resources it would take to design, administer, and count the ballots as you suggest, and validate the results… well, I think those resources could be better allocated elsewhere (like extending voting booths to more isolated communities).

    I performed a survey of Dell Rapids for my Master’s Thesis, and we implemented four distinct rotations to eliminate first-response and positive/negative perception bias. It was a herculean effort to ensure the surveys were correct, and were coded properly for sorting. Now imagine doing that for every municipality, special government, county, and state election.

    Even if you create ballot software that randomizes order, that’s a significant financial investment; software subscriptions aren’t cheap. Then there’s the worry about tampering. I doubt anyone would hack our sanitation board election, but state elections are definitely worth hacking.

    I would posit South Dakota’s voting system is probably plagued less by the first-name-favor bias and more by the letter-behind-the-name bias.

  11. Wayne, the SOS lot-drawing does not address the bias. It imposes that bias on every ballot, giving a whole party slate of candidates a built-in and undeserved advantage. GIving every party a chance at that unfair advantage each election does not change the fact that the advantage is unfair and clouds expression of the voters’ will.

    The resources required to truly cancel out this bias are not overwhelming. If we accept electronic voting, the resources are trivial. (I can envision the software requiring maybe ten lines of extra code—is that really that expensive?) If we stick with scantron ballots, the resources are an extra dot of ink on each ballot and perhaps some fractionally increased processing time. If we have to hire grad students to implement to ballot rotations, then yeah, the county’s pizza budget will skyrocket. ;-)

    I will agree that letter-behind-the-name bias is the greater influencer. I’ll flip that advantage shortly. The Secretary of State should look for a way to counteract that other bias.

  12. Darin Larson

    Cory, it looks like the Dems lost the drawing of lots for this election. Republicans are first on the ballot. We have our work cut out for us once again.

  13. I always run assuming we’re behind and the odds are against us. Charge!