Today at 11 a.m. in the atrium of the Maine State Archives, Secretary of State Matt Dunlap will engage in a quirky ritual of Maine democracy. He will determine, through some kind of random drawing, the order of the citizen initiative referendum questions on November’s ballot.
As far as I can tell, Maine is the only state to employ this method for determining ballot order. Most other states with initiative processes place questions on the ballot in the order they are approved.
While this ordering may not have as much of an effect on the outcome of these initiatives as the campaigns for and against them, or even the way the questions are worded on the ballot (a determination made just last week), there is a small chance that where a question appears on a ballot will make a difference in whether or not it passes.
With five citizen initiatives this year, more than ever before in Maine history,* which of them comes first on the ballot and which comes last could be more consequential than usual.
What we know
It’s long been received political wisdom that questions higher on the ballot do better. In 2012, California Governor Jerry Brown even went so far as to change state law to make sure a tax measure he backed was placed higher than a rival proposition. The referendum he supported won and the later-listed measure failed.
Political scientists, however, don’t yet agree on whether such a general effect actually exists.
There have been several studies of concepts like “downballot drop off” and “choice fatigue” among voters. One study, for instance, looked at natural experiments in California (where there are many referendums and they sometimes appear at different places on the ballot in different jurisdictions based on the number of races in various precincts). The authors found that “Facing more decisions before a given contest significantly increases the tendency to abstain or rely on decision shortcuts, such as voting for the status quo.”
Another study, however, which examined some of the same data in California and similar results from Texas, and which used a different methodology, found “the absence of evidence that being listed at the top compared to the bottom of the ballot has any effect on approval.” The authors concluded that it was actually the overall number of propositions, and not their order, that affected approval rates.
Field experiments have been similarly mixed. One study actually examined those 2012 California tax referendums by commissioning a poll in another state (Florida). They asked about a set of hypothetical referendums, including two that were similar to the two California propositions, and rotated the question order. They discovered a small benefit when one of the questions was placed before the other, but no effect that rose to the level of statistical significance.
What we might assume
Altogether, the research seems to indicate that the effect of question order is probably small, and is based on a more complex set of circumstances than just a single question’s position.
The ordering of specific questions on related issues might have some impact. For example, If voters first encounter (and vote in favor of) the Stand Up for Students measure to increase taxes slightly on the wealthy to increase school funding, they may be less worried about revenue shortfalls when they encounter the marijuana legalization question and less likely to be swayed by the idea that it would bring in new tax dollars.
Strength of support, rather than just overall preference, may also play a role for some measures. Public opinion polls often show that only a minority of voters are opposed to gun safety measures like Maine’s background check referendum, but that they often hold those beliefs more fervently. If that question is at the bottom of the ballot, perhaps opponents will be more likely to seek it out, experience less drop-off than supporters and have an outsized influence on the final results.
On the other hand, that same kind of dropoff could be beneficial for the ranked choice voting referendum. If it were at the bottom of the ballot, fewer voters that aren’t as interested in voting (and potentially also less interested in passing new voting reforms and more likely to default to the status quo) may vote on the question.
The potential effects of ballot position may even be mitigated or enhanced by the campaigns for and against other measures. If supporters or opponents of the initiative in the fifth position on the ballot mount a massive turnout operation, perhaps drop-off would decrease for the first four as well.
Then, there’s the elephant and the donkey in the room. It’s not yet clear what voters Donald Trump and Hillary Clinton will each bring to the polls, what referendums they’ll support or oppose and how likely they’ll be to vote on downballot questions.
We will never really know all the effects that ballot order has on referendums like these, but at least after today’s ritual we’ll know what it will mean to vote Yes or No on Questions 1,2,3,4 and 5 in November.
Update: With the drawing concluded, the order will be 1. marijuana legalization 2. school funding 3. firearm background checks 4. minimum wage increase 5. ranked choice voting.
* (While there are a record five initiatives on the ballot this year and one bond question, that’s not actually the record for total measures. The 2009 election saw one people’s veto, four initiatives, one bond question and one constitutional change, for a total of seven ballot questions.)