The Conceptualization and Measurement of Crossover Voting

• Author: Gary D. Wekkin
• Published date: 01 March 1988
• Date: 01 March 1988
• DOI: 10.1177/106591298804100107
• Subject Matter: Articles

THE CONCEPTUALIZATION AND MEASUREMENT OF
CROSSOVER VOTING
GARY D.WEKKIN
University of Central Arkansas
CIENTIFIC studies focusing exclusively on crossover voting did not
~~ appear until a Democratic party reform rule banning &dquo;open&dquo;
S presidential primaries was adopted during the early 1970s. The pur-
pose of the rule was to force open primary states such as Wisconsin and
Michigan to use party registration, pre-enrollment, or some other means
of preventing Republicans and other non-Democrats from participating
in the party’s presidential delegate-selection process. Those behind this
reform hoped to deprive non-mainstream candidates such as George Wal-
lace of electoral support from Republican and independent identifiers in
the afore-mentioned states’ contests in 1976 and thereafter. Although the
Democratic National Committee fought for ten years to enforce this rule
and went before the U.S. Supreme Court in order to win the right to do
so (Wekkin 1984), it now has relented in order to promote the party’s
competitive standing in certain states. Democrats in two states with open
primary traditions (Wisconsin and Montana) were granted special exemp-
tions to this rule for 1988; Democrats in a third state (Arkansas) already
have enacted an open presidential primary statute in the hope of exploit-
ing this weakening of national party resolve; and other states accustomed
to holding open or &dquo;crossover&dquo; (nominally &dquo;closed&dquo;) primaries may fol-
low suit before 1988.
Is crossover voting in fact a significant factor in the outcome of open
presidential primaries? The new and relatively small literature on this sub-
ject is divided concerning the extent and electoral effect of crossover vot-
ing in such contests. Ranney (1972) and Adamany (1976) have presented
evidence that crossover voting in three successive Wisconsin presiden-
tial primaries was extensive and diluted the participation of Democratic
identifiers by altering the margins of victory in those three contests. On
the other hand, Hedlund (1978), Hedlund, Watts, and Hedge (1982), and
Hedlund and Watts (1986) have argued that crossover voting in Wiscon-
sin presidential primaries since 1968 has not been unduly extensive, and
that the dilutive effect of participation by voters loyal to other parties
has been minimal. At this writing, the views of the Hedlund group ap-
pear to prevail in the literature on voting behavior.
Obviously, measurement of most forms of behavior depends a great
deal upon how the behaviors are conceptualized and operationalized. In
Received: December 5, 1986
Revision Received: April 2, 1987
Accepted for Publication: April 9, 1987
NOTE: The author wishes to thank the Research Council of the University of Central
Arkansas for its support.

---

106
this study I will examine some conceptual differences in the existing liter-
ature on crossover voting, offer for consideration an alternate, more dis-
criminatory conceptualization of crossover voting, and present some new
survey data which suggest that previously used conceptualizations have
resulted in either undermeasurement or overmeasurement of crossover
voting. Until there is agreeement on what crossover voting is and is not,
we
cannot know
how
much of it takes place, how much its impact may
be upon presidential primary outcomes, and thus cannot begin to form
and test hypotheses about the influence of factors such as primary for-
mat, type of partisan-enrollment procedure, and unchallenged incumbents
as variables in our primary-saturated presidential nomination process.
ON CONCEPTUALIZATION
A
fundamental problem in the current literature is the manner in which
various scholars have conceptualized and operationalized the term &dquo;cross-
over voting.&dquo; Ranney defines it as &dquo;voting in one party’s primary by ad-
herents of an opposing party or of no party at all&dquo; (1972: 34). By this
definition, all Republicans and all independents (including Democratic-
leaners) who cast ballots in a Democratic primary would be counted as
crossover voters. Adamany defines &dquo;crossover voters as those who do
not identify with the party in whose primary they cast ballots&dquo; (1976:
538). Again all independents, regardless of their leanings (or lack of them),
are counted along with opposing party-identifiers as crossover voters.
However, in operationalizing the partisan identification of the respon-
dents to his 1964, 1968, and 1972 Wisconsin samples, Adamany discards
the seven-point scale of partisan identification and classifies Republican-
leaning independents and Democratic-leaning independents as party iden-
tifiers, &dquo;because it is probable that under the mildest form of party regis-
tration, in which the partisan declares his affiliation at the polls, they
would be willing to volunteer a party identification in order to vote. Par-
ticipation in the opposition party primary would, therefore, be crossover
voting for...