by: Grebner
Tue Jul 22, 2008 at 01:19:03 AM EDT
There’s plenty of “conventional wisdom”, but very little solid knowledge, about the effect of ballot order on voter choice. Finally somebody with statistical competence has taken the trouble to gather a large dataset and analyze it correctly.
The short answer is that being listed first is worth about two percentage points, except in partisan general elections, where the effect seems to disappear.
Every state’s election system is different, and many contain weird, even unique, twists. The voters are more or less the same, but the election machinery creates different patterns in different states. The point of that caveat is that the phenomenon demonstrated in California will apply to Michigan, but applying it will require extrapolation and guesswork.
(Taken from “Estimating Causal Effects of ballot Order from a Randomized Natural Experiment: The California Alphabet Lottery, 1978-2002“, by Daniel E Ho and Kosuke Imai, Public Opinion Quarterly, Summer 2008.
For the past 30 years, California has put the candidates for a given office into random order, and then has rotated that list in each State Assembly district. The result has been a huge collection of data (ten statewide offices, times 24 statewide elections, times 80 Assembly districts, generally with several candidates in each). But the data is very hard to analyze, because (for example) the rotation by Assembly district isn’t “random” – the districts are numbered from north to south. Without bothering with the details here, I’m convinced the authors figured out a correct way to control for all the cross-cutting factors, so they could focus on 1) who benefits, 2) in which kinds of elections, 3) under what circumstances, 4) by how much.
In partisan general elections, they found that candidates of third parties get about 0.3% more votes in any precinct they’re listed first than when their name falls in any other position. For many of the candidates, that seemingly tiny amount is enough to increase their vote by half. But there was no detectible difference for the Democratic or Republican Party candidates in general elections.
They had less data regarding non-partisan elections, but being listed first on the ballot appeared to be worth about 2.0 percentage points.
In major party primary elections, being listed first was worth about 2 percentage points.
Finally, it should be noted that although these results seem plausible as applied to Michigan, there is some danger than they may be affected by specific quirks of Californial election law. For one thing, the state government mails a non-partisan voter guide, containing all the candidates and offices, a few weeks before each election. For another – as weird as it seems – California ballots list the “occupation” of each candidate, which includes information that a particular person is the incumbent. The effect didn’t seem to be affected by the presence in the race of an incumbent, but as I said above, it’s the sort of special circumstance that makes it impossible to assume the effect would be exactly the same in Michigan.
If anything, providing “occupation” information on the printed ballot should reduce the effect of being first, since by providing cues, it should reduce guesswork. So the effect in Michigan may be slightly larger than in California.
This study considered only the effect of being listed first, ignoring possible benefits or harm from being second, or last. If there are consistent effects of being listed in other positions, they seem to be substantially smaller.
I haven’t discussed Michigan’s rules for printing names on ballots. They’re complicated, and create their own set of problems and questions. But it’s a pretty good bet that it’s better to be first than last.
Leave a Reply