Position Effects With Party Column Ballots

• Author: R. Darcy
• Published date: 01 December 1986
• Date: 01 December 1986
• DOI: 10.1177/106591298603900406
• Subject Matter: Articles

POSITION EFFECTS WITH
PARTY COLUMN
BALLOTS
R. DARCY
Oklahoma State University
ART
of the folklore of politics is that being first on a ballot is worth
at least some votes in the general election. A number of states, out
of a sense of fairness, as a result of political compromise, or as a con-
sequence of court orders, have attempted to nullify supposed ballot posi-
tion effect by rotating ballot positions in general elections. This has been
done in several ways. Alaska completely shuffles ballots before distribut-
ing them to voters. In this way the candidate on the top for one voter will
be on the bottom for the very next voter at that voting station. Ohio, Idaho,
and several counties in Colorado rotate by precincts. California rotates by
state assembly districts; Kansas and Illinois rotate by counties; Kentucky
rotates by congressional districts. New Mexico conducts a lottery each year
to determine state-wide ballot position. In this way, either in each election
or across several elections, each party or each candidate is in the first posi-
tion roughly an equal number of times.
To assess the consequences of ballot rotation, interviews by phone and
in person were conducted with election officials in thirteen states and with
certain local election officials where rotation is practiced: Franklin County,
Ohio; Chicago; and Denver County, Colorado. In addition, interviews were
conducted with senior officers of two major companies providing election
equipment and ballots to a number of states. These interviews were struc-
tured and averaged over one hour each. Published testimony by other elec-
tion officials in various federal court cases was also reviewed.
It is clear that ballot rotation is not without its costs. First, rotation in-
creases the costs of preparing ballots. Generally, rotation increases the num-
ber of ballot styles and that alone increases printing costs. Additional costs
are incurred by ballot randomization and by special numbering required
by the rotation. An official for a printing company estimated Alaska’s ro-
tation increases its printing costs by 40 percent.
Rotation adds greatly to the administrative burdens of the election. For
one county in Ohio (Franklin) a complex ninety-five page document was
Received: 10/09/85
Revision Received: 1/8/86
Accepted for publication: 1/14/86
NOTE: This research was supported in part by the Office of the Attorney General of the
State of Oklahoma. The author is grateful to Jim Frankes and Thomas Spencer of that
office for their help and encouragement. Susan Cook of the Payne County League of
Women Voters and Claire Baker of the Oklahoma State University Residence Halls
Association are due special thanks for helping field the experiments reported here. Phil
Outhier prepared the data for analysis and otherwise assisted with this project. Charles
Hadley and Robert England helped to improve this writing. Election officials from thir-
teen states were very helpful and generously shared their insights.

---

649
required to specify the precinct rotation for the 1984 presidential election.
For counties using punchcards, specific computer counting programs must
be written and tested for each ballot type. Additional training is also re-
quired for the temporary help administering the election. Bain and He-
cock (1957) cite several instances of failure to follow rotation procedures
properly in Michigan and voice the suspicion that such failures are
widespread.
Rotation increases the chances of error in vote counting. State election
officials indicated a number of instances of failure to count ballots properly
due to a confusion at the county level over the rotation scheme (see Growe
1985; see also Simak v. Kusper 1985). While this is less a problem with vot-
ing machines, it appears to be a major problem with paper and punchcard
ballots.
Finally, rotation can add to the burden on voters who must hunt through
complex ballots for candidates. Judge Marshall, in Simak v. Kusper noted
that what the voter wants must also be considered in these cases. He
concludes:
what they want is a ballot which enables them to find their candidates with
reasonable speed. And those who are waiting in line to vote want them to
find their candidates with reasonable speed. (Simak v. Kusper 1985: 16).
His feeling was that rotation schemes can make long and already complex
ballots more difficult for voters.
Problems of expense, burdens on election officials and voters must be
borne as a necessity if, indeed, there is an advantage to being first on bal-
lots. If there is no position effect, then states are incurring unnecessary ex-
penses, elections are being made unnecessarily more difficult to administer
and voters unnecessarily are being given more confusing ballots. Yet there
is no evidence that there is ballot position advantage in general elections.
PREVIOUS RESEARCH
Research from the 1950s and 1960s as well as more dated anecdotal
evidence indicates that there are position effects in non-partisan and parti-
san primary elections -
elections in which there is no party guide for voters
(see Bain and Hecock 1957; Byrne and Pueschel 1974; Dana 1912; Mueller
1969, 1970; Scott 1972; White 1950). Other research seems to indicate that
where voters are asked to elect several candidates to one office, as is the
case for multi-member legislative districts (Darcy, Welch and Clark 1985),
there is also a positional advantage. This advantage is not for position on
the ballot per se; it is for position within the set of candidates of a faction
or a party (see Brooks, 1921; Robson and Walsh 1974; Upton and Brook
1974).
Researchers have touched on American-style partisan elections involv-
ing only one winner. This work finds no significant position advantage in
general elections when party is indicated (Coombs, Peters and Strom 1974;
Mueller 1969; Scott 1972). Despite the lack of evidence some authors still
are reluctant to conclude ballot position effects are absent in general elec-

---

650
tion situations. For example,...