Chair Hee, Vice Chair Shimabukuro, and Committee Members:
The League of Women Voters of Hawaii offers comments on SB 32, which would require the Chief Elections Officer to establish procedures to select a letter of the alphabet by lot and arrange candidate names on the ballot in alphabetical order by last name, beginning with the randomly selected letter. Our reading of the bill is that the proposed letter selection followed by alphabetic ordering would be the same throughout the State, without differences among legislative districts.
The League applauds the fact that this bill recognizes ballot design is important in producing a fair election, and hopes the Office of Elections will consider the good suggestions about the format of the ballot that are contained in this measure. But the main purpose of the bill appears to be addressing the well-known problem that coming first on the ballot increases a candidate’s total vote count.1 This phenomenon is well documentedin political science and psychology research. Estimates of the typical average advantage are varied, from two to three percentage points, depending on whether major or minor party.candidates are involved, whether it is a nonpartisan race or whether it is a primary or general election.2
This is not trivial theoretical research. In 2000 in California, George W. Bush got nine
percent more votes in the assembly districts where he was listed first than in the assembly
districts where he was listed last. This occurred in California even though California was also
rotating the name order across assembly districts. The nine percent advantage was present even when researchers took into account the fact that certain districts tend to vote Democratic and others tend to vote Republican. The authors concluded that although name order cannot swing the votes of decided voters, in a close race similar to this 2000 presidential race the winner can,and has been decided by the order of candidates’ names on the ballot.3
HB32 proposes a simple rotation scheme to mitigate this positional problem. Choosing a letter of the alphabet would introduce a small amount of randomization, but not much. 4Using the procedure described in the bill wouldn’t eliminate the problem – it would only lead to greater success for the candidate who was lucky enough to get selected first in the draw. I am also saying “simple” rotation because the bill does not propose to rotate the order of names across the State, as many states such as Ohio and California already do.
In an ideal world, you would want each possible ordering of candidates to be represented equally on our ballots. This is not accomplished with the alphabet drawing approach in this bill. For example, if there were 4 candidates for the Democratic primary election in Senate District 23, there would be 24 possible orders of candidates.5 Even with extremely careful printing and distribution by the Office of Elections, it would be impractical and confuse voters to print and distribute 24 different ballot orderings in this district. The Office of Elections currently makes sample ballots available for voters in advance of elections, and this “best practice” would be infeasible. With 24 different orderings how could voters feel assured they had received the correct ballot?
Coming up with a fairer approach to ballot ordering is definitely possible in Hawaii elections, when there are a limited number of candidates (e.g. the presidential candidates in a general election) and a relatively large number of election districts. Ohio, for example, tries hard to optimize fairness and accountability in this situation, by varying the name order across the state so that each possible order permutation appears an equal number of times, and observers can inspect ballots on Election Day to be sure the rotation was done properly. Idaho, North Dakota, Wyoming and a few other states use versions of this system. 6 But the larger the field of candidates the more complex the systems of these states become.
The League concludes that the intent of the proposed bill is good, but the methodology and implementation are very imperfect. Under these circumstances, there is no compelling reason to spend the money required to implement the proposed complex ballot order system rather than using the current approach of ordering names alphabetically. Thank you for the opportunity to submit testimony.
1 Krosnick, Jon A. and Joanne M. Miller, The Public Opinion Quarterly, Vol. 62, No. 3 (Autumn, 1998), pp. 291-3306 Krosnick, Jon A. “In the Voting Booth, Bias Starts at the Top,” New York Times, November 4,2006.
2 Ho, Daniel E. and Kosuke, Imai, “Estimating the Causal Effects of Ballot Order from a Randomized Natural Experiment: the California Alphabet Lottery, 1978-2002, The Public Opinion Quarterly, (2008) 72 (2): 216-240.
3 Jon A. Krosnik, Joanne M. Miller, and Michael P. Tichy, “An unrecognized need for ballot reform,”' in Ann N. Crigler, Marion R. Just, and Edward J. McCaffery (eds.), Rethinking the Vote:The Politics and Prospects of American Election Reform (New York: Oxford University Press, 2004), pp. 52, 53, 63
4 Another problem is that the letters of the alphabet are not equally like to be used at the beginning of last names. I know only one person in Hawaii whose last names begins with X,” but I know many people whose last names start with “L.”
5 (4 factorial, i.e. 4x3x2x1)).
6 Krosnick, Jon A. “In the Voting Booth, Bias Starts at the Top,” New York Times, November 4, 2006.