Who Evaluates a Presidential Candidate by Using Non-Policy Campaign Messages?

• DOI: 10.1177/106591290505800105
• Author: Marisa A. Abrajano
• Date: 01 March 2005
• Published date: 01 March 2005
• Subject Matter: Articles

PRQ_March05_II 3/24/05 9:18 AM Page 55
Who Evaluates a Presidential Candidate
by Using Non-Policy Campaign Messages?
MARISA A. ABRAJANO, NEW YORK UNIVERSITY
This article tests the hypothesis that low-education voters are more likely to evaluate a candidate using per-
sonalistic or non-policy campaign messages than are more educated voters. The Latino electorate in the U.S.
presents an ideal case study, given that both Presidential candidates in the 2000 election directed personalis-
tic campaign messages toward them. Latinos with low-levels of education should be the most likely to evalu-
ate a candidate using personalistic campaign cues since processing and understanding these messages require
little in stored political information. Analysis of self-reported responses from the Latino Voter Survey of 2000
indicates that low-education Latinos are more likely than are high-education Latinos to use non-policy cues
when evaluating a candidate. This finding implies that vote choice is structured differently for Latinos with
varying levels of education. To test this implication and to confirm the finding from the self-reported responses,
I estimate a model of Latino vote choice for the 2000 Presidential election. Probit analysis shows that high-
education Latinos are indeed more likely to use factors that are informationally demanding, such as candidates’
issue positions and ideology than are low-education Latinos.
The 2000 U.S. election saw both Presidential candi- question: Which Latinos are more likely to evaluate a Pres-
dates, George W. Bush and Al Gore, actively courting
idential candidate who uses these non-policy campaign
the Latino vote. Both campaigns invested heavily in
cues? I contend that Latinos with lower levels of stored
Spanish-language media and made frequent appearances in
political knowledge are the ones who will evaluate a candi-
Spanish neighborhoods (Segal 2002; West 2000). Anecdo-
date from personalistic cues. To test this hypothesis, I use
tal evidence further suggests that both Presidential candi-
the Latino Voter Survey of 2000 and estimate a model that
dates inundated Latinos with personalistic campaign mes-
examines the determinants of using non-policy campaign
sages and activity.1 Examples of such efforts included
cues in a respondent’s evaluation of a candidate. Education,
George W. Bush’s touting of his half-Latino nephew, George
which serves as a proxy for political information, is
P. Bush, and Gore proudly announcing that his grandson
expected to be the primary explanatory variable. If it is
was born on Cinco de Mayo, which is the Mexican day of
indeed the case that non-policy campaign cues influence
Independence. These campaign activities exemplify person-
low-education Latinos to a greater extent than more edu-
alistic, or non-policy related, campaign efforts, since they do
cated Latinos, it would suggest that voters with low-educa-
not reveal any information about the candidates’ issue posi-
tion levels are more susceptible to certain types of campaign
tions or policy stances. However, not all voters are equally
messages than are voters with higher levels of education.
affected by these campaign cues. As Zaller (1992: 123)
If this finding proves true, it would also imply that vote
notes, voters with low levels of political awareness have a
choice is structured differently for voters with varying levels
higher probability of accepting a message, given that they
of education. Thus, a second hypothesis asserts that high-
receive the message, than do voters with greater levels of
education Latinos will be more likely to use issues and ide-
political awareness.2
ology in their vote choice than will low-education Latinos.
Based on Zaller’s research and candidates’ campaign
High-education Latinos possess the necessary amounts of
behavior toward Latinos in 2000, I pose the following
stored political information to vote according to the spatial
model of voting (Downs 1957), whereas low-education
Latinos may be less likely to vote in this way, given the high
1 For media accounts, refer to Anderson 2000, Calvo 2000 and Fountain
informational burden required from the spatial model
2000.
2
(Campbell, et al. 1964).
While Zaller notes that voters with mid-levels of political awareness pos-
sess the highest probability of attitude change, I am focusing on Zaller’s
point that low-education voters have the highest probability of accepting
LITERATURE REVIEW
a message, even if it might not result in an attitude change. Since I am
not interested in whether one’s attitude changes, I merely want to show
The importance of political information in one’s ability to
that voters with the lowest amount of political awareness are the most
make informed vote decisions has been an extensively
susceptible to campaign messages, to a greater extent than mid- and
researched topic, and it is widely acknowledged that much
high-awareness voters.
of the U.S. electorate is politically uninformed (Gerber and
NOTE: Thanks to Jonathan Nagler for his most helpful comments, feed-
Greene 1998; Husted, Kenny, and Morton 1995; Lupia and
back, and advice.
McCubbins 1998; Niemi and Junn 1998; Smith 1989; Sni-
Political Research Quarterly, Vol. 58, No. 1 (March 2005): pp. 55-67
derman, Brody, and Tetlock 1991) and that this distribution
55

---

PRQ_March05_II 3/24/05 9:18 AM Page 56
56
POLITICAL RESEARCH QUARTERLY
of political information systematically varies from individual
less information, e.g., Latinos, are either underrepresented or
to individual. Two factors help to alleviate this unfortunate
altogether overlooked. Because surveys are biased toward the
reality. First, Popkin (1994) asserts that information short-
groups who give substantive responses, and low political
cuts overcome this problem, and that such behaviors are
information respondents like Latinos often provide either
grounded on the basis of “low-education rationality.” His
“no opinion” or “do not know” responses, their opinions are
explanation rests on the assumption that “views the voter as
often underrepresented in the surveys. Moreover, for those
an investor and the vote as a reasoned investment in collec-
respondents who are more informed, their beliefs will be
tive goods, made with costly and imperfect information
consistent with the answers that they provide, but for those
under conditions of uncertainty.” Because information is dif-
who either provide “don’t know” responses or “no opinion”
ficult to attain and thus provides voters with little incentive
responses, beliefs may not be accurately reflected in their
to gather it, information shortcuts function as a mechanism
responses. Given that elected officials today give consider-
that reduces these costs. Such heuristics can take the form of
able weight to public opinion polls in the formation of policy
a reliance on campaign messages, experiences from daily
decisions (Althaus 1998), minority policy interests, espe-
lives or using information experts, such as political com-
cially Latino concerns, may not be properly represented.
mentators, as sources of information (Lupia 1994; Popkin
The literature on heuristics suggests that its content often
1994). However, not all voters rely on the same type of
focuses on policy, or if using an individual or political com-
heuristics. Popkin and Dimock (1999, 2000) demonstrate
mentator, the voter is receiving policy-related information
that the type of information shortcut voters’ use is highly
from that individual. However, what happens when such
dependent on their stored levels of political knowledge.
cues are absent of policy? Along with demographic features
Popkin and Dimock’s (2000) empirical research finds that
that make minority groups less likely to have high levels of
the less knowledgeable an individual is, the more likely that
political information, anecdotal evidence suggests that the
individual is to perceive foreign policy issues as threatening.
content of minority-specific campaign messages differs from
In addition, these low-information voters have greater diffi-
the mainstream campaign messages. Most notably, Presi-
culty placing foreign policy issues within the broader politi-
dential campaigns in the past decade, and especially in the
cal context. In another work by Popkin and Dimock (1999),
2000 election, devoted considerable time and energy in tar-
they find that voters who are less politically informed are
geting Latino voters, a minority group that has increased in
more likely to evaluate a candidate based on their personal
its share of the U.S. electorate. While efforts to woo these
characteristics as opposed to their policy positions.
voters are not new, anecdotal evidence suggests that a con-
Information shortcuts are not the only way that voters may
sistent theme in these efforts focuses on symbolic and non-
behave as if they were fully informed. Page and Shapiro
policy campaign cues.3 For instance, Latinos have been
(1995) contend that when voters’ decisions are aggregated,
exposed to campaign cues such as Ford’s inability to eat a
the random errors of each voter cancels out. Thus, it may be
tamale (Popkin 1994), Bush and Gore’s attempts to com-
the case that statistical aggregation overcomes individual
municate in Spanish during the 2000 election, George W.
shortcomings. This assertion is demonstrated by Page and
Bush’s touting of his half Latino nephew, and Gore proudly
Shapiro, who contend that on average individual behavior is
sharing to...