RESEARCH NOTE
Voters and the Personal Vote:
- ACounterfactual Simulation
DAMARYS CANACHE AND JEFFERY J. MONDAK, FLORIDA STATE UNIvERSITY
ERNESTO CABRERA, Ipsos-ASI LATIN AMERICA
Electoral formulas and campaign practices in some nations encourage
voters to consider the personal qualifications of candidates for legislative
office, whereas individuating information about candidates is unavailable
to voters in other nations. Where electoral rules and elite behavior foster
the personal vote, the personal vote flourishes. However, it is unclear
why voters respond to personal information about candidates. We test
two competing hypotheses: that the personal vote reflects a general
human tendency to attend to information regarding individuals' personal
qualifications; and that the personal vote is an acquired behavior that
emerges after voters are socialized to attend to personal information
about candidates. We use counterfactual simulations to test these
hypotheses. Specifically, we conduct laboratory experiments in which
subjects drawn from Mexico and Venezuela-nations in which electoral
rules ensure that the personal vote does not flourish-are provided the
opportunity to consider candidates' personal qualifications. Results are
consistent with the hypothesis that the personal vote stems from a general
tendency in human decision making.
Electoral systems vary in the extent to which they foster candidate-centered
campaigns and elections (Carey and Shugart 1995). In the United States, individuating
information about candidates is plentiful, and voters respond to that
information. Cain, Ferejohn and Fiorina (1987, 9) define the personal vote as "that
portion of a candidate's electoral support which originates in his or her personal
qualities, qualifications, activities, and record." Operationalizing the personal
vote with data on constituency service, Cain and his colleagues demonstrate the
importance of the personal vote in U.S. House elections. Related research on
American legislative elections has established that a legislator's involvement in
scandal adversely affects future electoral performance (e.g., Abramowitz 1991;
Political Research Quarterly, Vol. 53, No. 3 (September 2000): pp. 663-676
663
at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
PoliticalResearchQuarterl
Political Research Quarterly
Funk 1996; Peters and Welch 1980), and, more broadly, that a relationship exists
between incumbents' levels of competence and integrity, and electoral success
(e.g., Kulisheck and Mondak 1996; Luttbeg 1992; McCurley and Mondak 1995;
Mondak 1995a).
Our interest is in exploring the prerequisites of the personal vote. Three
requirements must be met for a personal vote to exist. First, a nation's electoral
formulas must allow for the personal vote. This requirement clearly is met in U.S.
elections. With partisan primaries, single-member districts, and a relatively weak
party system, electoral rules in the United States pose no obstacle to a personal
vote. Second, individuating information about candidates must be disseminated
to voters. Where electoral rules do not preclude a personal vote, legislative candidates
and scandal-hungry media have an incentive to ensure that personal
information reaches voters. Hence, in nations such as the United States, this
second requirement typically will be met. Third, regardless of electoral rules and
the campaign practices of candidates and journalists, a personal vote will not
exist unless voters add personal information about candidates to the decision calculus.
We focus on this third issue here. We are interested in whether voting on
the basis of a candidate's personal characteristics reflects a general human tendency
to prioritize individuating information, or, in contrast, whether such
voting constitutes an acquired behavior that emerges only if voters have been
socialized to view personal information as important.
Several researchers have posited that it is both natural and rational for voters
to attend to personal information about candidates (e.g., McCurley and Mondak
1995; Popkin 1991; Sullivan et al. 1990). Two factors underlie this perspective.
First, people regularly assess the competence and integrity of other people, and
we tend to be adept at doing so. Applied to the political arena, decision making
may be more efficient when voters evaluate candidates partly on the basis of their
personal attributes rather than solely on the basis of esoteric or ambiguous policy
stances. Second, personal traits are relevant considerations. Voters should prefer
that representatives be competent rather than incompetent, honest rather than
dishonest, and attentive to constituents rather than inattentive.
Although several analysts have argued that voters will respond to candidatespecific
information where possible, we should not assume this view to be correct
absent corroborating evidence. An alternate to this perspective is that the
personal vote reflects learned behavior that emerges only because it is encouraged
by certain media practices, and cultivated by political candidates. This view
suggests that rational voters prefer to choose between competing candidates
solely on the basis of impersonal criteria such as ideology, partisanship, and
policy positions. Where a personal vote exists, it does so because candidates and
media have emphasized personal factors over political ones. Voters are encouraged
to examine personal information, and they may be given little opportunity
to do otherwise.
664at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Voters and the Personal Vote
Determining which of these views best accounts for the personal vote is
important on two levels. First, on a micro level, exploration of the origins of the
personal vote may improve our understanding of the psychology of electoral
decision making. In at least some elections, voters do consider personal information
about the candidates. Does this behavior reflect citizens functioning as
competent and expedient decisionmakers, or does it signal an electorate
deprived ofsubstantive information and induced against its collective will to vote
on the basis of tangential personal considerations?
Second, on a macro level, insight regarding the foundations of the personal
vote may help us to foresee the effects of electoral reform. In those countries that
have moved to make elections more candidate centered, the short-term implications
of such changes will vary depending on whether voters must be socialized
to form their decisions partly on the basis of candidates' personal traits. We know
that electoral rules congenial to the personal vote and the dissemination of personal
information about candidates are necessary conditions for the personal vote
to exist. Voters are precluded from acting on personal information if such information
never reaches voters. But it is not clear that adoption of favorable rules
and the subsequent dissemination of personal information by candidates and
media are sufficient for a personal vote to emerge. Indeed, these factors would not
be sufficient if voters must learn to attend to candidate-specific information. In
this scenario, new electoral rules would pave the way for voters to see information
about the personal qualities of candidates, but voters might not immediately
recognize such information as being pertinent to electoral choice.
What would constitute a critical test of the competing perspectives we have
outlined? In any scenario in which the question is whether a behavior is natural
or acquired, the defining test is straightforward: select subjects who have not previously
engaged in the behavior in question, and see whether, when given the
opportunity, they engage in that behavior. Suppose, for example, that you purchase
a golden retriever. If you tell the dog to roll over, she will respond with a
blank stare. Rolling over is a learned behavior. But if you toss the dog into a pool,
she will swim. For dogs, swimming is an innate behavior.
We require an inherently artificial, or counterfactual, research design. We
must pluck voters from contexts in which electoral rules dictate that the personal
vote is not prevalent, and then determine whether they act on individuating
information about candidates when provided the opportunity. Kulisheck and
Mondak (1996) reported a laboratory experiment in which college students were
exposed to information regarding the competence and integrity of hypothetical
U.S. House candidates. Given the candidate-centered nature of U.S. elections, it
is not surprising that the effects were positive. We replicate this research design,
but with subjects drawn from two nations-Mexico and Venezuela-in which
electoral rules severely limit exposure to information about candidates' personal
characteristics. This means that our experiment simulates electoral reform in the
665at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Political Research Quarterly
sense that voters in our laboratory experienced the consequences of a new set of
rules-clear personal information about political candidates was now available.
Our empirical question thus is quite simple: will Mexican and Venezuelan subjects
behave like the previous U.S. sample, or will they ignore the unfamiliar candidate-specific
information?
STUDY DESIGN
Our experiments were conducted in Venezuela and Mexico in 1995. In this
section, we will provide a brief outline of the Mexican and Venezuelan electoral
contexts, followed by discussion of the procedures and manipulations used in
the experiments.
Mexico and Venezuela
Mexico and Venezuela are appropriate choices for present purposes provided
that extant electoral structures and campaign practices limit the personal vote in
those nations. If this prerequisite is met, then experimental study will permit us
to determine how Mexican and Venezuelan subjects respond to an unfamiliar
form of information. We believe that Mexico and Venezuela provide viable contexts
for the research question we have outlined.'
In Mexico, our experiment centers on the Assembly of Representatives of the
Federal District (local), and the Chamber of Deputies (national). Party selection
of candidates in Mexico greatly limits the role of a candidate's personal reputation.
Indeed, Carey and Shugart (1995: 426) offer the Mexican system as an
example of an electoral formula in which personal reputation is unimportant. In
Mexico, an added factor is that the nation imposes a one-term limit on service in
legislative posts (i.e., no legislator can be elected to consecutive terms in the same
office). Term limits were implemented in Mexico to maximize the power of party
leaders, and especially leaders of the nation's dominant party, the Partido Revolucionario
Institucional (PRI). Term limits strike two blows against the personal vote.
First, self-interested elected officials have no immediate incentive to cultivate a
personal reputation because reelection is a legal impossibility. Indeed, the central
assumption of the personal vote thesis is that legislators are motivated by the
desire to secure reelection (Cain, Ferejohn, and Fiorina 1987). This motivation
is absent in Mexico. Second, term limits in Mexico foster unflinching party loyalty,
meaning that there is a disincentive for the legislator to cultivate personal
Although Mexico and Venezuela are appropriate contexts for present purposes, so too are many
other nations. Familiarity and convenience also affected our decision to run our experiments in
Mexico and Venezuela. Two of us are natives of Latin America; at the time this paper's experiments
were conducted, one ofus was engaged in other research in Venezuela, and the other was teaching
in Mexico.
666at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Voters and the Personal Vote
support: the career-minded politician will not risk actions that run contrary to
the wishes of party leaders.
The bottom line is that Mexico's electoral system creates strong barriers to the
personal vote. Former candidates are sometimes recycled, and some politicians
do rotate among offices. Thus voters may, from time to time, be able to recall a
candidate's past performance in office. But aside from such distant recollections,
the personal vote is not a factor in Mexico. Mexico offers a research context in
which nonpersonal criteria function as central determinants of electoral outcomes.
The hypothetical elections contested in our laboratory likely marked the
first instance in which our subjects enjoyed a meaningful opportunity to vote on
the basis of candidates' personal qualities.
Venezuela's electoral formulas allow slightly greater room for a personal vote
than do Mexico's, but only because of changes implemented recently. We focus
on the Chamber of Deputies (national) and the District Council (local). Until
1993, Venezuela used closed, blocked lists for the Chamber of Deputies. A candidate's
personal reputation is of no consequence under such a formula, because
voters select from between party lists, and voters have no option to add or delete
names, or to change the order in which the names appear on the ballot (Carey
and Shugart 1995; Carey 1996). Crisp (1998: 24) notes that in Venezuela this
structure resulted in "nearly lock-step party discipline." A similar system was
used in elections for the District Council until 1989.
In 1993, Venezuela implemented a mixed formula for the Chamber of
Deputies; each voter can cast a ballot for a candidate in a single-member district,
and also for a party slate (Ellner 1993-94; Kulisheck and Canache 1998). Voters
were permitted to split their ballots in elections for the District Council in 1989,
and a mixed system (two-thirds of Council elected in single-member districts,
one-third proportional representation) was implemented in 1992 (Ellner 1993-
94). Because our experiment was conducted after Venezuela's initial mixed-system
election, subjects had little real-world experience in evaluating candidates' personal
qualities. The most vital information-how well legislators have performed
in office-is not available for the first election held under a new formula.
At the time our data were gathered, the personal qualities of political candidates
had not yet played a meaningful role in actual legislative elections in
Mexico and Venezuela. By drawing experimental subjects from these contexts,
we can determine if, given a favorable decision-making climate, voters will
attend to individuating information about legislative candidates.
Procedure
Procedures followed those used by Kulisheck and Mondak (1996). The 130
participants in Venezuela were students at two universities in Caracas, Universidad
Catolica Andres Bello and Universidad Sim6n Bolivar. We have 155 participants
from Mexico, and they were students at Universidad de las Americas,
667
at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Political Research Quarterly
Universidad Aut6noma de Mejico-Xochimilco, and Centro de Investigacion y
Docencia Econ6mica de Mejico. As noted above, each student was provided data
regarding two hypothetical legislative elections, one national and one local. All
renditions of the experiment followed a 3 x 3 between-subjects design, with
manipulations used to vary descriptions of the competence and integrity of candidates.
Students were assigned randomly to treatment conditions.
Experiments were run during regular class sessions. Students were read an
opening statement that explained the study as an inquiry regarding how media
might improve election coverage. Students were told that they would read information
about hypothetical candidates for two offices, and assess the utility of that
information. After the opening statement, participants completed a demographic
survey (age, sex, whether registered to vote, ideology, etc.). Next, they read descriptions
of the first pair of candidates. The descriptions resembled "Meet the Candidates"
stories published by many U.S. newspapers. Information about the two candidates
appeared in parallel columns, enabling direct comparison. The stories first
presented background information (name, education, occupation, family, and
political affiliation), then quotes from the hypothetical candidates explaining their
positions on crime, health care, and inflation, and finally information regarding the
candidates' levels of competence and integrity. The competence and integrity evaluations
were attributed to a fictitious nonpartisan political organization, Por un Voto
Consciente. Each candidate was rated separately on competence and integrity (high,
adequate, low). Students were asked to read the information regarding the first pair
of candidates, answer a series of questions (which candidate they would vote for in
an actual election; which candidate they preferred on the issues of crime, health
care, and inflation; whether the news story provided adequate information), and
then repeat the process for the second pair of candidates. Upon completion of the
study, participants were explained the project's specific purpose.
Manipulations centered on the competence and integrity of the first candidate
("Candidate A") in each election. All other information was held constant,
including evaluations of the competence and integrity of the second candidate in
each election (competence and integrity both held at adequate). The manipulations
were intended to be subtle. Many information-processing experiments
employ dramatic manipulations. For instance, studies on the effects of source
credibility may depict a message's source as either a high school student or a
Nobel Prize winner, or other comparable extremes. In contrast with such an
approach, we sought to present clear information about competence and
integrity, but to do so in a manner that would not overwhelm the experiments'
participants. In other words, we wanted to give the participants every opportunity
to choose between the two candidates on the basis of criteria other than
competence and integrity. Toward this end, we placed the competence and
integrity information last within the hypothetical news stories, after extensive
background and policy information. Further, the competence and integrity
668at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Voters and the Personal Vote
evaluations used a simple three-category typology, and little definitional detail
was offered regarding the organization Por un Voto Consciente.
Although the personal vote includes factors in addition to competence and
integrity, we limited our focus to these traits. One reason we did so stems from
the limits of experimental designs. In the laboratory, we can capture features of
real-world decision-making contexts while simultaneously creating counterfactual
contexts. Still, manipulations must be plausible. Information about competence
and integrity can be varied quite easily even for hypothetical candidates. In
contrast, it would be much more difficult to manipulate other components of the
personal vote, such as a voter's familiarity with a candidate's record of constituency
service. The second reason we focus on competence and integrity is to
bring comparability with past research conducted in the United States. Our study
replicates a laboratory experiment focused on voting in U.S. House elections
(Kulisheck and Mondak 1996). Hence, in addition to contrasts between Mexico
and Venezuela, we will be able to broaden the scope of comparison by discussing
similarities and differences in findings from three electoral contexts.
We use the laboratory experiment to create and study a counterfactual decisionmaking
environment. This inherent artificiality-coupled with our use of
nonrepresentative samples-clearly demands that our findings be viewed as only
suggestive, not conclusive. Fortunately, the pattern of results reported below is
straightforward and consistent, offering a revealing new perspective on the
nature of the personal vote.
RESULTS
Because the experiments systematically varied information regarding competence
and integrity, estimation of possible direct effects requires only that we
include the manipulations as independent variables in models of the vote choice.
Competence and integrity are coded 1 (low) to 3 (high). The dependent variable
in our analyses, the vote, is dichotomous (1 = respondent voted for candidate A;
0 = respondent voted for candidate B); hence, we explore the impact of these personal
qualities through estimation of a series of logistic regression models. We
will report four models, because parallel experiments were conducted conceming
local and national elections in both Mexico and Venezuela.
Each of the logit models includes two control variables designed to capture
nonpersonal elements of electoral choice. The first is Issue Proximity (6 =
respondent strongly favors the positions of Candidate A on the issues of crime,
health care, and inflation, to -6 = respondent strongly favors the positions of candidate
B on these issues).2 The second control variable is ideology. Subjects may
2 The initial scales for each issue ranged from 1 (e.g., subject strongly prefers candidate position
on crime) to 5 (subject strongly prefers candidate B). We rescaled these items to range from 2 to -
2, and then summed scores across the three issues.
669at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Political Research Quarterly
have been able to elicit ideological cues from the information about the candidates,
a possibility we test through inclusion of self-reported ideology in the logit
models (1 = left, to 10 = right).
We expect positive coefficients for our central independent variables, Competence
and Integrity Such results would indicate that the likelihood that a participant
voted for candidate A in a given hypothetical election increased as descriptions
of the two qualitative traits were varied from low, to adequate, to high. The
measures of issue proximity also should produce positive coefficients, meaning
that participants are expected to support those candidates whose positions they
favor on the issues of crime, health care, and inflation.
Logit estimates are depicted in Table 1. The data reveal that information about
the candidates' levels of competence and integrity did influence participants'
decisions regarding which candidates to support. The quality variables produce
statistically significant effects in six of eight tests (five at p < .05, one at p < .10),
the only exceptions being competence in the Mexican national election and
integrity in the Venezuelan local contest. Voters previously unaccustomed to
adding personal information about candidates to the decision calculus did so
upon exposure to such information.
The emergence of significant effects for the candidate quality variables allows
us to consider several more specific questions. First, are there any systematic differences
between results for Mexico and Venezuela, or results for the national and
local elections? Although we can speak to these questions only tentatively, the
answer in each case appears to be no. The coefficients for competence and
integrity vary in size across our four hypothetical elections, but this variance
seems unsystematic. For instance, the largest effects in Mexico are for the local
election, but in Venezuela competence and integrity bring the greatest joint influence
in the national contest. Conversely, the only insignificant coefficients
emerge in the national election in Mexico and the local election in Venezuela.
Candidate quality seemingly does not exert precisely the same level of influence
among all electorates, but current results reveal no conclusive indication that the
observed variance is anything but idiosyncratic.
A second question is how our results compare with those found in the United
States. Recall that our study is modeled on research concerning U.S. House elections
(Kulisheck and Mondak 1996). We have seen in Table 1 that six of the eight
coefficients for competence and integrity achieve statistical significance; the mean
value of the coefficients is .521. Results in the United States were very similar. Five
ofsix coefficients for the candidate quality variables were statistically significant in
the U.S. experiments, and the mean value of the coefficients was .403 (Kulisheck
and Mondak 1996: Table 2). Hence, laboratory experiments conducted using
three quite different samples have yielded consistent evidence regarding the
impact of candidate quality on the vote choice in legislative elections.
A third question concerns the substantive magnitude ofthe effects ofcompetence
670
at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Voters and the Personal Vote
TABLE 1
COMPETENCE, INTEGRITY, AND THE VOTE CHOICE: LOGIT ESTIMATES
Mexico Venezuela
Deputy Council Deputy Council
(national office) (local office) (national office) (local office)
Constant -2.387* -2.777** -1.152 -2.870*
(2.513) (2.206) (1.071) (2.126)
Issue Proximity 0.800*** 1.509*** 0.741*** 0.976***
(5.422) (5.627) (5.282) (5.373)
Ideology 0.088 -0.233# -0.310* 0.269
(0.845) (1.670) (2.034) (1.590)
Competence -0.001 0.942** 0.545* 0.657*
(0.003) (2.589) (1.915) (2.002)
Integrity 0.329# 1.006** 0.622* 0.070
(1.320) (2.799) (2.196) (0.215)
Number of Cases 155 153 130 128
Model Chi-Square 56.992 124.296 55.696 74.710
Note: Cell entries are coefficient estimates from logistic regression models in which the
dependent variable is the vote choice (1 = respondent voted for candidate A; 0 = respondent
voted for candidate B); t-values are in parentheses (one-tailed tests are used for issue
proximity, competence, and integrity).
#p<.10
*p < .05
**p < .01
***p < .001
and integrity. To address this issue, we calculate the estimated likelihood of a vote
for candidate A in each experiment across the three levels of competence and
integrity, while holding the control variables at their means. The results are
depicted in Table 2. The largest effect occurs for integrity in the local election in
Mexico, where the estimated likelihood of a vote for candidate A increases from
.249 to .712 across the three values of integrity, a swing of over 45 points. The
quality variables produce average swings of 24 points.
Subjects in our experiments were drawn from electoral contexts in which the
personal vote is rare, and yet these subjects responded strongly to information
regarding candidates' levels of competence and integrity. Indeed, the substantive
magnitude of current results equals what Kulisheck and Mondak reported in tests
conducted in the United States. Current findings clearly are more consistent with
the hypothesis that attention to personal characteristics is a general component of
human decision making than with the hypothesis that the personal vote is an
acquired behavior that flourishes over time given cultivation by political candidates
671at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Political Research Quarterly
TABLE 2.
ESTIMATED LIKELIHOOD OF VOTE FOR CANDIDATE A AS A FUNCTION OF
COMPETENCE AND INTEGRITY
Mexico Venezuela
Deputy Council Deputy Council
(national office) (local office) (national office) (local office)
Competence
High .300 .699 .747 .490
Adequate .300 .475 .634 .332
Low .300 .261 .498 .205
Integrity
High .373 .712 .762 .348
Adequate .300 .475 .632 .332
Low .235 .249 .479 .317
Source: Table 1.
Note: Cell entries indicate the estimated likelihood of voting for candidate A, with issue proximity
and ideology held at their mean values. In calculating estimates for Competence, the values of
Integrity is held at 2 (adequate); likewise, in calculating estimates for Integrity, the value of Competence
is held at 2 (adequate).
and news media. No socialization was necessary for participants in the experiments
to respond to information about a candidate's competence and integrity.
The skeptic might counter that we offered experimental subjects no choice but
to respond to our manipulations. We have several comments on this point. First,
the manipulations were subtle. Participants were exposed to a great deal of information
regarding the candidates, including the candidates' stances on three salient
policy issues. The information about competence and integrity was brief, and it
was presented without explanatory detail. Second, because the manipulations
were subtle, it was possible for participants to ignore the information about competence
and integrity. Issue proximity yielded substantively greater effects than
did competence and integrity in each election, and ideology also produced statistically
significant effects in two ofthe hypothetical contests. Further, the candidate
quality variables produced two null results in eight tests, meaning that some participants
could and did ignore our manipulations. Third, iflearning contributes to
the personal vote, then current results should have been weaker than those
reported in the United States. That is, even if the laboratory design ensures the
detection of positive effects-a claim we reject-it does not follow that those
effects should be as strong in Mexico and Venezuela as in the United States.
Our final response requires an added empirical test. If subjects reacted to the
manipulations because the study's design gave them no other choice, then it
672at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Voters and the Personal Vote
follows that the manipulations should have been most influential for those subjects
for whom issues and ideology were unrevealing. For example, a person who
strongly preferred one candidate to another on the issues had the opportunity to
ignore our manipulations, and to decide how to vote solely on the issues. If subjects
who could have ignored information about competence and integrity did
not do so, we would reject the possibility that our core results are an artifact of
this study's experimental design. Tests on this point require that we fold our
measures of issue proximity and ideology, and then interact these new variables
with competence and integrity The folded issue variable ranges from 0 (subject
prefers the two candidates equally on the issues) to 6 (respondent strongly
prefers one candidate to the other on all issues). The folded ideology variable
ranges from 1 (subject is a self-identified centrist) to 5 (subject is strongly liberal
or strongly conservative). Separate models were estimated for issue distinctiveness
and ideology. The models were of the form:
Vote = a + blIssue Proximity + b2ldeology + b3Competence + b4lntegrity +
b5lssue Distinction + b6(Issue Distinction X Competence) +
b7(Issue Distinction X Integrity).
Statistically significant negative coefficients on the interaction terms would indicate
that the impact of competence and integrity weakened for subjects who were
most able to distinguish between candidates on the basis of issue positions or
ideology. These tests produced 16 interaction terms across our four elections.
None of the 16 interactions yielded a significant negative effect, and in no
instance was model performance enhanced through inclusion of the new variables.
In short, the effects of the competence and integrity manipulations were
constant for all subjects. It is not the case that our manipulations produced
effects only for participants who had no other means to decide which candidate
to support.
IMPLICATIONS
Voters seemingly enjoy little choice on the matter of a personal vote. In the
United States, the candidate-centered nature of campaigns means that voters
often have no option but to consider information regarding the personal qualifications
of candidates. Elsewhere, in contrast, electoral formulas in some nations
are such that voters never receive, and thus cannot consider, this same form of
information. In this paper, we have sought to determine how voters would
respond if they were free to attend to, or to ignore, individuating information
about candidates. To pursue this question, we have had to construct an inherently
counterfactual research context. By drawing subjects from Mexico and
Venezuela, we could examine whether information about candidates' personal
qualifications resonates with voters who have not previously been socialized to
view such information as important.
673at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Political Research Quarterly
Results of our experiments are consistent with the hypothesis that the personal
vote reflects a general tendency in human decision making. We are all
much more practiced at evaluating people as people rather than as policymakers.
Therefore, provided that the requisite information is available, we prefer to judge
political candidates partly on the basis of their personal qualifications. In our laboratory
studies, Mexican and Venezuelan subjects responded to our manipulations
of information regarding the competence and integrity of political candidates.
Rather than ignoring this information, our subjects took it into account.
Moreover, the impact of competence and integrity in Mexico and Venezuela
equaled effects previously found in the United States, suggesting that political
socialization determines neither the existence nor perhaps even the magnitude of
the personal vote. Coupled with evidence that the particular personal traits
voters emphasize are common cross-nationally (e.g., Bean and Mughan 1989;
Pancer, Brown and Barr 1999), current results support the conclusion that candidates'
levels of competence and integrity are central components of the voter's
decision calculus across a wide array of electoral contexts.
Our results suggest a clear implication regarding the importance of electoral
rules: ifvoters inherently prefer to consider personal information when evaluating candidates,
then itfollows that the presence of a personal vote in a given electoral context
hinges on the existence of congenial electoral rules. In Mexico, for example, the virtual
absence of a personal vote reflects not the choice of voters to ignore individuating
information, but rather the impact of party-centered electoral formulas.
Voters view the competence and integrity of candidates to be meaningful
considerations, but these considerations are neutralized in electoral systems in
which the qualifications-and sometimes even the names and faces-of individual
candidates never reach the voters.
Insight regarding the nature of the personal vote helps us to understand not
only how voters vote, but also how well. Research conducted in the United States
suggests that term limits would disrupt the efforts of voters to screen incumbents
on the basis of competence and integrity, and thus would diminish the aggregate
quality of representation (Mondak 1995b). This same logic would appear to hold
with even greater power in those nations in which electoral formulas minimize
or fully preclude the personal vote. Where voters are unable to screen candidates
on the basis of personal qualifications, the system includes no mechanism to
foster either the retention of superior incumbents or the eviction of inferior ones.
In Mexico, for example, party selection of candidates coupled with the nation's
one-term limit means that quality is irrelevant. It makes all the difference if a
person is a PRI candidate. It makes no difference whatsoever if a person is a good
or bad PRI candidate.
Because voters apparently will attend to candidate-specific information when
given the opportunity to do so, our results suggest that electoral reforms
designed to weaken party control over the campaign process will succeed in
674at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Article
altering the nature of electoral choice. Venezuela is a case in point. Recent electoral
reforms in Venezuela have diminished party control over candidate selection,
ceding greater power to the electorate (e.g., Ellner 1993-94; Kulisheck and
Canache 1998; Shugart 1992). By weakening party control, electoral reforms in
Venezuela potentially facilitate the emergence of a stronger personal vote. Our
results suggest that Venezuelans are receptive to personal information about legislative
candidates. Because the nation's changing electoral structure will promote
the availability of such information, we foresee an increasing role for the personal
vote in future elections.
Given the rarity of the laboratory experiment in comparative political studies,
a few final words about the method we have used are in order. We believe that
laboratory research can help to broaden understanding of electoral choice. We
also feel that experiments (and other simulation techniques such as mathematical
models) should be viewed as complements to observational methods. In that
we have simulated electoral decision making in a counterfactual setting, our purpose
clearly has not been to observe how Mexicans and Venezuelans vote.
Instead, we have used the experiment to gain leverage on a fundamental question
of process. Laboratory research is highly useful for such purposes, and thus
we feel that the experiment can contribute to a multifaceted cross-national
research program.
REFERENCES
Abramowitz, Alan I. 1991. "Incumbency, Campaign Spending, and the Decline
of Competition in U.S. House Elections." Journal ofPolitics 53: 34-56.
Bean, Clive, and Anthony Mugham. 1989. "Leadership Effects in Parliamentary
Elections in Australia and Britain." American Political Science Review 83: 1165-
79.
Cain, Bruce, John Ferejohn, and Morris Fiorina. 1987. The Personal Vote: Constituency
Service and Electoral Independence. Cambridge, MA: Harvard University
Press.
Carey, John M. 1996. Term Limits and Legislative Representation. New York: Cambridge
University Press.
Carey, John M., and Matthew Soberg Shugart. 1995. "Incentive to Cultivate a Personal
Vote: A Rank Ordering ofElectoral Formulas." Electoral Studies 14: 417-39.
Crisp, Brian F 1998. "Institutional Design and Compromised Legitimacy" In
Damarys Canache and Michael R. Kulisheck, eds., Reinventing Legitimacy:
Democracy and Political Change in Venezuela. Westport, CT: Greenwood Press.
Dominguez, Jorge I., and James A. McCann. 1996. Democratizing Mexico: Public
Opinion and Electoral Choices. Baltimore, MD: Johns Hopkins University Press.
Ellner, Steve. 1993-94. "The Deepening of Democracy in a Crisis Setting: Political
Reform and the Electoral Process in Venezuela." Journal of Interamerican
Studies and World Affairs 35 (4): 1-42.
675at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from
Political Research Quarterly
Funk, Carolyn L. 1996. "The Impact of Scandal on Candidate Evaluations: An
Experimental Test of the Role of Candidate Traits." Political Behavior 18: 1-24.
Kulisheck, Michael R., and Damarys Canache. 1998. "Democratic Legitimacy
and Political Change in Venezuela." In Damarys Canache and Michael R. Kulisheck,
eds., Reinventing Legitimacy: Democracy and Political Change in
Venezuela. Westport, CT: Greenwood Press.
Kulisheck, Michael R., and JefferyJ. Mondak. 1996. "Candidate Quality and the
Congressional Vote: A Causal Connection?" Electoral Studies 15: 237-53.
Luttbeg, Norman R. 1992. "The Validity and Electoral Impact of Media Estimations
of 'Best' and 'Worst' State Legislators." Legislative Studies Quarterly 17:
395-408.
McCurley, Carl, and Jeffery J. Mondak. 1995. "Inspected by #1184063113: The
Influence of Incumbents' Competence and Integrity in U.S. House Elections."
AmericanJournal ofPolitical Science 39: 864-85.
Mondak, Jeffery J. 1995a. "Competence, Integrity, and the Electoral Success of
Congressional Incumbents." Journal of Politics 57: 1043-69.
Mondak, JefferyJ. 1995b. "Elections as Filters: Term Limits and the Composition
of the U.S. House." Political Research Quarterly 48: 701-27.
Pancer, S. Mark, Steven D. Brown, and Cathy Widdis Barr. 1999. "Forming
Impressions of Political Leaders: A Cross-National Comparison." Political Psychology
20: 345-68.
Peters, John G., and Susan Welch. 1980. "The Effects of Charges of Corruption
on Voting Behavior in Congressional Elections." American Political Science
Review 74: 697-708.
Popkin, Samuel L. 1991. The Reasoning Voter: Communication and Persuasion in
Presidential Campaigns. Chicago: University of Chicago Press.
Shugart, Matthew Soberg. 1992. "Leaders, Rank and File, and Constituents: Electoral
Reform in Columbia and Venezuela." Electoral Studies 11: 21-45.
Sullivan, John L., John H. Aldrich, Eugene Borgida, and Wendy Rahn. 1990.
"Candidate Appraisal and Human Nature: Man and Superman in the 1984
Election." Political Psychology 11: 459-84.
Received: July 12, 1999
Accepted for Publication: February 29, 2000
dcanache@gamet.acns.fsu.edu
676at Masarykova Univerzita on October 5, 2016prq.sagepub.comDownloaded from