Discussion (Psychology)

Name Recognition and Candidate Support

Author(s): Cindy D. Kam and Elizabeth J. Zechmeister

Source: American Journal of Political Science , October 2013, Vol. 57, No. 4 (October 2013), pp. 971-986

Published by: Midwest Political Science Association

Stable URL: http://www.jstor.com/stable/23496668

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms

Midwest Political Science Association and Wiley are collaborating with JSTOR to digitize, preserve and extend access to American Journal of Political Science

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

Name Recognition and Candidate Support

Cindy D. Kam Vanderbilt University Elizabeth J. Zechmeister Vanderbilt University

The mass media devote a great deal of attention to high-profile elections, but in American political life such elections are the

exception, not the rule. The majority of electoral contests feature candidates who are relative unknowns. In such situations,

does name recognition breed contempt, indifference, or affection? Existing work presents modest theory and mixed evidence.

Using three laboratory experiments, we provide conclusive evidence that name recognition can affect candidate support,

and we offer strong evidence that a key mechanism underlying this relationship is inferences about candidate viability. We

further show that the name-recognition effect dissipates in the face of a more germane cue, incumbency. We conclude with

afield study that demonstrates the robustness of the name-recognition effect to a real-world political context, that of yard

signs and a county election.

"... there is only one thing in the world worse than being talked about and that is not being talked about."

— Oscar Wilde ( The Picture of Dorian Gray)

X olitical candidates dedicate significant resources to attempts to imprint their names on the public.

.A. They post yard signs, send mailers, print bumper stickers, distribute flyers, and make media appearances, among other things. Many of these messages contain only the candidate's name, and they likely draw very little con

scious consideration by citizens. Such messages aim to increase the candidate's name recognition, but there is no academic consensus on whether mere name recognition actually affects candidate support.

On the one hand, some researchers see the wisdom in

promoting a candidate's name, thus subscribing to Stokes and Miller's classic statement that "recognition carries a positive valence; to be perceived at all is to be perceived favorably" (1962, 541). On the other hand, some second Abramowitz's seminal rebuttal that "while mere name

recognition [does] not breed contempt, neither [does] it breed affection" (1975, 674). Bartels takes a middle ground, arguing that "in electoral politics, mere public familiarity, although far from sufficient to ensure a can didate's success, does appear necessary" (1988, 57).

In this article, we ask: Does name recognition matter,

why, and under what conditions? We argue that name recognition, as a heuristic for decision making, can play an important role in electoral politics. We find that it matters in a direct way, by increasing voters' support for

a candidate, and in an indirect way because recognition signals viability, which increases voters' support for a can

didate. In short, name recognition has both direct and indirect effects on candidate support. We also demon strate that there are limits to name recognition: name recognition matters primarily in the absence of more ger

mane political information. In sum, mere recognition of

a candidate's name primarily matters in low-information elections.

Why should we care about low-information elec tions? For one thing, these elections are the rule, not the

Cindy D. Kam is Associate Professor of Political Science, Vanderbilt University, PMB 0505, Nashville, TN 37203-5721 ([email protected]). Elizabeth J. Zechmeister is Associate Professor of Political Science, Vanderbilt University, PMB 0505, Nashville, TN 37203-5721 ([email protected]).

Laboratory data were collected through the Research on Individuals, Politics, and Society Laboratory at Vanderbilt University and The Omnibus Project at the University of California, Davis. We thank Jennifer Anderson, Margarita Corral, Desmond Dennis, Carl Palmer, Matt Simpson, Aimee Sobhani, Matthew Taylor, and Stephen Utych for research assistance. Many thanks to John Geer, Don Green, and Ben Highton for useful advice. We are also grateful to seminar participants at the Center for the Study of Democratic Institutions at Vanderbilt University, the Center for the Study of Democratic Politics at Princeton University, the American Politics Workshop at MIT, the Political Psychology Workshop at the University of Chicago, and the Interdisciplinary Workshop on Politics and Policy at the Center for Political Studies at the University of Michigan for constructive comments. Data and replication files are available at the AJPS Data Archive on Dataverse (http://dvn.iq.harvard.edu/dvn/dv/ajps).

American Journal of Political Science, Vol. 57, No. 4, October 2013, Pp. 971-986

©2013, Midwest Political Science Association DOI: 10.1111/ajps. 12034

971

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

972 CINDY D. KAMAND ELIZABETH J. ZECHMEISTER

exception, in American politics. Citizens are frequently asked to weigh in on races where the most effective piece of

information—partisanship—is unavailable. These races include prominent positions, such as mayor, but also less

visible positions, such as court clerk, public defender, school board members, city council members, and lo cal authority positions, such as port commissioner and fire commissioner. Additionally, political primaries re quire citizens to adjudicate between candidates who are indistinguishable on party lines.1 Further, and as we will detail later, only a very small percentage of the ballots we

sampled included information about incumbency status.

These low-information races often appear on the same ballot as more visible state and national contests that

yield high rates of participation.2

In an overwhelming number of electoral decisions, voters confront a set of names and have little to no ad

ditional information about them.3 And, in such contexts

where voters are almost blindly casting around for their best guess as to their preferred candidate, cues matter (Schaffner and Streb 2002). One possible cue, we say, is name recognition.

Theoretical Perspectives on Name Recognition

While name recognition is often discussed by both political activists and academics, there is no em pirical consensus—and no comprehensive theoretical perspective—on whether mere name recognition matters

in politics, why, and under what conditions. With respect to whether it matters, on the one hand, some scholars

'We conducted a content analysis of sample ballots used in four midsized U.S. cities (El Paso, TX; Nashville, TN; Portland, OR; and Omaha, NE) from 2007 to 2011. For each electoral decision on the ballot, we coded ( 1 ) whether party affiliations appeared next to the candidates' names; (2) whether a candidate's incumbency was stated; and (3) the type of race (primary or a general elec tion). Of the 417 electoral decisions voters were asked to make, 76.5% provided no information about the partisan affiliation of the candidates, either because the race was nonpartisan (44.8% of decisions) or because the race was within a primary (31.7% of decisions).

2 While citizens are less likely to vote in a down-ballot election with no partisan information, a fair number still do (Schaffner and Streb 2002).

3Scholars studying these low- and no-information races have ex amined the extent to which voters are influenced by the order in which the candidate names appear (Miller and Krosnick 1998), gender, and race and ethnicity (e.g., Matson and Fine 2006). Both these studies suggest the relevance of name recognition but do not directly test for its effect.

suggest that any causal relationship between name recog nition and candidate success is spurious: an empirical correlation may exist but only because name recogni tion is a proxy for other attributes of the candidate, such

as spending, personal qualities, public image, and per formance, which are the real drivers of electoral success

(e.g., Abramowitz 1975; Jacobson 2009). On the other hand, other studies suggest an important relationship be

tween name recognition and candidate support, in var ious types of elections and across various types of can didates. Thus, name recognition has been linked to the electoral success of incumbent congressional candidates and their challengers (e.g., Mann and Wolfinger 1980; Panagopoulos and Green 2008); congressional primary candidates (e.g., Burden 2002); presidential primary can didates (e.g., Bartels 1988; Lenart 1997); widows (e.g., Solowiej and Brunell 2003); and "celebrity" candidates who can sometimes capitalize on a highly visible public image (Squire 1995). In contrast to the old adage that "fa miliarity breeds contempt," this line of research suggests that familiarity breeds support.

Belief that name recognition matters has spurred research, for example, on how incumbents and chal lengers might increase their name recognition and, sub sequently, their vote shares. Within this research, schol

ars have argued that campaign spending is particularly effective in increasing name recognition and support for candidates, in particular challengers (e.g., Jacobson 2006; Panagopoulos and Green 2008); constituency ser vice is useful to securing an incumbency advantage (Serra and Cover 1992); outreach in person, by mail, and via the media is effective in increasing name recognition (Goldenberg and Traugott 1980; see also Prior 2006); and, an overlap or congruence between media markets and electoral districts in particular for House elections can increase recognition, especially to the benefit of chal lengers (Campbell, Alford, and Henry 1984; Levy and Squire 2000).

Much less research addresses the question of why name recognition matters. For the most part, this has been

a secondary concern of political scientists. Some research

does offer suggestive, but conflicting or unresolved, expla

nations. Thus, for example, Jacobson (1978) uses survey and campaign data to show that spending affects name recognition, in particular for challengers; and, further, his

study shows that name recognition affects evaluations of the candidates, and these in turn influence vote choice. However, Jacobson (2009) also offers evidence that the

nature of these evaluations is not predetermined; more familiar candidates do not always receive more favor able evaluations. At the same time, other research finds

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE SUPPORT 973

little relationship between increased name recogni tion and other assessments germane to vote choice. Abramowitz's (1975) study finds no evidence of a re lationship between recognition and performance assess ments. Coleman and Manna (2000) show that, while cam

paign spending increases recognition of candidates and their stances as well as warmth toward candidates, it does

not affect assessments of the candidates' trustworthiness,

efficacy, or involvement.

Drawing on research by scholars in the fields of psy chology and consumer marketing, we argue that there are two complementary pathways through which name recognition might lead to candidate support. The direct

causal pathway suggests that individuals will often simply

favor objects to which they have been exposed. According to Zajonc's (1968) mere exposure effect, familiarity (or "perceptual fluency") with a stimulus, induced by mere exposure to it, leads to warmer feelings toward it. While

most of Zajonc's early work used exposure to stimuli vis ible to the naked eye, Kunst-Wilson and Zajonc found that exposure delivered via subliminal presentation also increased liking for a variety of novel objects, conclud ing that "individuals can apparently develop preferences for objects in the absence of conscious recognition and with access to information so scanty that they cannot as certain whether anything at all was shown" (1980, 558). Zajonc (2001) suggests such an effect may occur because increases in familiarity, in the absence of negative infor mation, signal something about the benign, safe nature of

the stimulus.4 In his study of party primaries, Bartels ar

ticulates a version of this mere exposure hypothesis when

he speculates that voters "might respond quite unthink

ingly to changes in simple political stimuli, such as the frequency with which candidates' names appear on televi sion and in the newspapers" ( 1988,111 ). In a similar vein, research on exposure to and recognition of brand names in consumer marketing reaches similar conclusions: in dividuals are more willing to consider the purchase of a product brand to which they had been previously exposed (e.g., Coates, Butler, and Berry 2004, 2006; Holden and Vanhuele 1999). Altogether, this research suggests that exposure to a name might directly affect vote choice.

"Others have posited a similar logic to explain classic mere exposure effects (see, for example, Burgess and Sales 1971), leading scholars to examine the consequence of exposure to stimuli accompanied by a positive or a negative tag. In light of this research, our design tested for whether such an affective tag mattered, and we find no support for this expectation within this research design (see the online supplementary information).

We also argue that name recognition can affect can didate support through an indirect causal pathway. We propose that an indirect causal pathway arises because recognition prompts voters to make inferences about the candidate, and these inferences influence overall support. As one example, Jacoby and colleagues' (1989) influential piece on the "false fame" effect demonstrates that prior exposure to unfamous names can lead subjects to infer mistakenly that familiar, recognized names are those of

famous people. In addition, extensive research from the decision sci

ences on the "recognition heuristic" argues that recog nition of an attitude object can lead individuals to make

ecologically valid, inductive inferences about that attitude

object. The ecological validity of the recognition heuris tic allows individuals to make fast and accurate infer

ences because the information environment is structured

such that "recognition is correlated with the criterion be

ing predicted" (Goldstein and Gigerenzer 2002, 78). For example, recognizing one city (over another) provides an ecologically valid heuristic for inferring whether the recognized city has a larger population than the unrec ognized one (Goldstein and Gigerenzer 2002); recogniz ing one political party over another provides an ecologi cally valid heuristic for inferring whether the recognized

political party will outperform the unrecognized party (Gaissmaier and Marewski 2011). In electoral politics, the information environment supports the ecological va

lidity of using recognition to predict electoral outcomes: candidates and parties who receive more coverage and have more campaign advertisements (and thus are more

recognizable) also happen to do better in elections (Gaiss maier and Marewski 2011; Marewski et al. 2010).

Building from and contributing to this work on the recognition heuristic, we propose that the inferences de rived from recognition may then affect choice. Indeed, scholars advancing the recognition-heuristic literature have argued that vote choice may be influenced by voters'

tendencies to "rely on episodic knowledge and recogni tion to infer which political parties have the best chances of being elected" (Marewski et al. 2009, 2237; see also Gaissmaier and Marewski 2011; Marewski et al. 2010). In assessing the extent to which name recognition in duces inferences about political candidates, we consider three types of inferences that citizens might make in the

absence of any additional heuristic beyond mere recogni tion: traits, experience, and viability. The evidence, we will show, indicates that individuals do not make inferences

about traits and experience from recognition; however,

they do appear to make inferences about viability, a find

ing that is consistent with the work on the recognition

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

974 CINDY D. KAM AND ELIZABETH J. ZECHMEISTER

heuristic as applied to politics (Gaissmaier and Marewski 2011).5

While the recognition literature mostly focuses on explicit recognition, we suggest that implicit recognition may induce both the direct and indirect effects on choice

that we have outlined here. Recognition-heuristic scholars

do suggest the short-cut could work on an implicit level. For example, Goldstein and Gigerenzer discuss a priming study conducted by Standing (1973), in which a very short, explicit priming of numerous novel photographs

two days prior to a recognition test was sufficient to induce

above-chance recognition rates, even though the authors note these participants "must have felt they were making a

fair number of guesses" (2002,77). Items that are "merely

recognized" (to use Goldstein and Gigerenzer's term) may be items that individuals at a gut level believe they have

heard of before, even if they cannot trace the source of that

exposure or recall to any additional learned information

about the object. The recognition-heuristic scholarship has focused nearly exclusively on nonfictitious targets that

subjects explicitly self-report as recognized or not (but see

Hilbig 2010 for a discussion). In brief, we complement and extend this line of research in two ways. First, we argue and test the notion that exposure and recognition can operate on an implicit level, where subjects can be unaware of being exposed to a stimulus and unaware of recognizing the stimulus. Second, we argue that implicit recognition can lead not just to inductive inferences about

the candidate's viability but also to overall support for the candidate.

Assessing Name Recognition in the Lab: Studies 1-3

We begin with evidence from three laboratory studies that apply a subliminal priming technique. Distilling the effect of name recognition on candidate support is tricky using standard observational data, since a relationship between name recognition and political support could arise from confounding factors, such as candidate spend

ing or candidate quality (Abramowitz 1975; Jacobson 2009). A series of laboratory studies provides leverage on internal validity, by enabling us to estimate precisely

the cause-and-effect relationship between the key inde pendent variable (name recognition) and the dependent variable (candidate support).

5Krosnick et al. (1992) also find that exposure to stimuli influences a narrow, rather than broad, range of assessments.

With the laboratory design, we utilize a subliminal stimulus. The use of subliminal priming allows us to avoid demand effects that might be present in a more explicit test of name recognition (Orne 1962). Because the design directs subjects to stare at the screen while the prime "hits" subjects between the eyes, it allows us a more precise administration of the treatment than, for example, a study that presents the prime consciously but unobtrusively. The design allows us to identify whether

name recognition works on a subconscious level, without

conscious awareness of citizens. Finally, the design denies individuals the ability to identify the study as the source of their feelings of recognition.6 In terms of external va lidity, our use of a subliminal priming paradigm to induce

"mere exposure" to a hypothetical candidate's name ap proximates the incidental and accidental ways in which political stimuli often reach citizens, an assertion that we support in Study 4, below.

The first laboratory study examines the effect of name

recognition alone. Subjects are randomly assigned to a group that receives a name embedded in a subliminal prime, or a control group that does not receive the name. The results demonstrate that, absent additional infor

mation, name recognition influences candidate support.

Moreover, mediation analyses support our conjecture that assessments of viability may underlie increased candi date support for those more familiar with the candidate's

name. In short, we argue and offer strong evidence that name recognition registers an effect, not because it in duces considerations about candidate traits and experi ence but, instead, because it evokes effects that are akin

to bandwagoning. The second and third studies add an additional, politically relevant piece of information, in cumbency, which allows us to demonstrate the limits of name recognition against this stronger cue.

Subliminal Priming and Name Recognition: Study 1

In Study 1, 497 young adults were recruited from polit ical science and humanities courses at a medium-sized,

private, southern university in autumn of2009. They par

ticipated in exchange for extra credit. Subjects completed

a prestimulus questionnaire consisting of demographics, personality, and mood questions. They participated in an

individuals may discount recognition if the source of recognition is the laboratory or another source that would be deemed irrelevant to the decision at hand (see Marewski et al. 2009, 2010). We tested to see if they could detect the prime in the experiment, and they could not. Results appear in the online supporting information.

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE SUPPORT 975

unrelated study7 and then were randomly assigned to one

of two conditions for a subliminal priming period.8

During each of the 32 trials of the subliminal priming period, subjects were exposed to a fixation point (a dot) in the center of the screen, a forward mask of nonsense

letter strings, a candidate name (if they were in the exper

imental group) or no name (if they were in the control

group), a string of letters (nonsense string, positive word,

or negative word), and then a backward mask of non sense letter strings).9 The sequence of each trial appears in Table 1.

Immediately following the subliminal priming pe riod, subjects answered a distracter question10 and then received the following text:

We are interested in how candidates' names shape

first impressions. Since we are interested in first

impressions, please answer the following ques tions as quickly but as accurately as you can. Imagine two candidates are running for politi cal office: Mike Williams and Ben Griffin. If you

were eligible to vote in this election, for which candidate would you vote?

The selection of the candidates' names was deliberate.

We wanted to create a situation in which one candidate

would be more appealing, so that we could more easily de

7Subjects were evenly divided across men and women. They leaned Democratic, with 28% identifying as Republican, 45% identifying as Democrats, and the rest as Independent/Other. On the standard 7-point ideology measure, the average respondent was close to the midpoint of the scale. With respect to race, 7% of subjects identified as Asian, 5% as Hispanic, 10% as black, and 5% as other. The average age was 19.7. In the unrelated study, subjects were randomly assigned to watch audiovisual clips and then answered a series of questions. When we controlled for treatment in the unrelated study, the results were identical. Interaction terms capturing assignment into these studies were insignificant.

Additional details of the design are in the online supporting information.

'These instructions preceded the task: "The next task is part of a study of how people perceive different types of letter combinations. During the next few minutes we'll show you a series of letters and potentially some words. Later we may ask you questions about letter combinations. The images will appear only briefly. Between letter combinations, you should focus on the dot in the center of the screen. Press the space bar when you are ready to begin." The use of fixation points, forward masks, and backward masks is standard in subliminal priming procedures. The duration of the primes was determined after a wide canvass of existing works on subliminal priming (van den Bussche, van den Noortgate, and Reynvoet 2009). To determine whether subjects were able to detect the subliminally presented stimuli, we administered both recognition and recall tasks at the end of the study; these responses indicate that the level of conscious awareness of the primes was quite low, as noted in the online supporting information.

10"Which of the following letter combinations do you like most?"

Table 1 Subliminal Priming Design

Subliminal Priming Name Condition

Dot on screen (1000 ms) KQHYTPDQFPBYL

(500 ms) GRIFFIN (40 ms)

NONSENSE, POSITIVE, OR NEGATIVE

STRING (25 ms) PYLDQFBYTQKPH

(100 ms)

Control Group

Dot on screen (1000 ms)

KQHYTPDQFPBYL (500 ms)

NONSENSE, POSITIVE, OR NEGATIVE

STRING (40 ms)

PYLDQFBYTQKPH

(100 ms)

tect movement away from that baseline preference. Con

sequently, we gave one candidate a more familiar name and one candidate a less familiar name, and we placed the more familiar name first, given research on primacy

effects (e.g., Krosnick and Alwin 1987).11 For the famil iar name, we chose the surname "Williams" because it

is the third most popular surname in the country from the 2000 Census. We chose the first name "Mike" because

"Michael" is a very popular boy's name. For the less fa miliar candidate (Ben Griffin), we selected the last name "Griffin," which is ranked 114th on the list—familiar

enough not to be considered strange, but far less pop ular than "Williams." The first name ("Ben," from "Ben

jamin") was intentionally selected to be reasonably famil iar but far less popular.12 By keeping sex constant and using names that do not signal race/ethnicity, we elim inated other means by which voters might distinguish between the candidates.

With this intentional selection of names, we a pri ori expected Mike Williams to garner more support than Ben Griffin. And, consistent with this expectation, sub

jects in the control condition did favor Williams: 67% of

uWe could have executed a fully factorial design, where we sub liminally primed each name. Given the limited number of cases we were working with and our initial interest in exploring classical conditioning, we opted to subliminally prime only the less popular name. If we had included a condition where we primed the more popular name, we would still have expected to see an increase in candidate support. However, it is possible that the size of the recog nition effect could be smaller when a name is already sensed as popular.

I2Surname popularity is from (http://www.census.gov/genealogy/ www/data/2000surnames. From 1959 to 2008, "Michael" has ranked as the most or the second most popular boy's name, accord ing to the Social Security Administration. "Benjamin" has generally ranked in the 20s and 30s for the last few decades (http://www. ssa.gov/OACT/babynames/).

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

976 CINDY D. KAM AND ELIZABETH J. ZECHMEISTER

Figure 1 The Effect of Subliminal Priming on Candidate Support

0.6

c 0.4

| 5 0

00

1 0.3 c o

■■E o

0.1 |

0 4

Control Treatment

" -2

-4

.£2 -6

s c

SE

-10

Control "Ireatmer

Note: Bar graphs with 95% confidence intervals. Differences across conditions significant at ρ < 0.01.

them said they would vote for him and only 33% said they would vote for Griffin. Did subliminally priming the

name "GRIFFIN" affect candidate support? It did, as the first panel in Figure 1 demonstrates. Because we have ran domly assigned subjects to be in the treatment or control conditions, the difference in means provides us with an unbiased estimate of the treatment effect, without having

to control for the litany of variables typically enlisted in observational data analysis.13 Figure 1 compares the pro portion of subjects who said they would vote for Griffin,

by condition. As noted above, only 33% of the control group subjects would have voted for Griffin. Among the treated subjects, who were subliminally primed with the name "GRIFFIN," 46% said that they would have voted for Griffin. The difference in proportions is statistically

significant at ρ < 0.01. Subjects were also asked to evaluate each of the candi

dates using the feeling thermometer. We construct a feel

13Random assignment ensures that any preexisting differences across treatment and control conditions are in expectation zero. That said, any given experiment may suffer from incidental imbal ance across cells. We looked at differences in individual characteris

tics (sex, partisanship, ideology, race, age, and political awareness) across the two conditions; a Hotelling vector of means test of the equality of means could not be rejected at ρ < 0.05, suggesting that the means across our conditions are indistinguishable from each other.

ing thermometer differential which theoretically ranges from —100 (advantage Williams) to +100 (advantage Griffin), with 0 indicating neutrality. Did subliminally priming the name "GRIFFIN" influence subjects' sup port for Griffin? As we can see from the second panel of results in Figure 1, it did. In the control group, Griffin has

almost a 6-point disadvantage compared with Williams. In the treatment group, Griffin has only a 1-point disad vantage compared with Williams. These differences are statistically distinguishable from each other at ρ < 0.01.

Together, these results demonstrate that mere exposure to

a candidate's name significantly affects candidate support,

as measured by both vote choice and affect.14

Inferences from Name Recognition

There maybe more to the story than mere exposure simply

leading to support. Consistent with research on the recog nition heuristic, we suspect that some sort of inferential process may occur: that is, individuals may make infer ences about a candidate's traits, experience, or viability.

14For each dependent variable, controlling for sex, partisanship, ideology, race, age, and political awareness yielded identical results. For vote choice and feelings, the nonparametric Wilcoxon Mann Whitney test yielded similar results: a p-value of 0.01 and 0.03, respectively.

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE SUPPORT 977

Figure 2 The Effect of Subliminal Priming on Inferences

0.45

0.4

=0.35

0.3

0.25

Note: Bar graphs with 95% confidence intervals. Difference across conditions significant for Viability only, at ρ < 0.01.

We included a battery of trait questions aimed at ascer taining whether or not the subliminal priming led subjects

to make positive trait inferences about the primed can didate. The four-item trait battery focused on Griffin.15 We combined responses to these four trait items into an additive scale, ranging from 0 (most negative on all items)

to 1 (most positive response on all items).16 Across the treatment and control groups, as shown in Figure 2, there

were no statistically distinguishable differences in average

trait assessments (ρ ~ 0.24). Subliminally priming sub jects with the name "GRIFFIN" did not make them more likely to draw positive inferences about Griffin's personal traits.

Perhaps recognition led subjects to make inferences about candidates' qualifications for office (e.g., their past political office). To see if this was so, we asked subjects

15Subjects were asked, "In your opinion, how well does the phrase 'He is honest' describe him?" Four response options were given: "Extremely well"; "Quite well"; "Not too well"; and "Not at all." Questions about leadership, empathy, and intelligence followed.

16The scale is highly reliable: Cronbach's α = 0.84. Factor analysis yielded one eigenvalue = 2.22, with the next highest eigenvalue at 0.02. The individual-item factor loadings landed in a narrow range, from 0.67 to 0.80. The scale mean is 0.48 with standard deviation of 0.20.

how many times Griffin had held previous office.17 The plurality of subjects guessed that Griffin had held office once in the past. But, there were no statistically significant

differences between the treatment and control groups (ρ ~ 0.77), again suggesting that if subjects were mak ing inferences from recognition, these were not inferences

about experience. A final possibility is that subjects made inferences

about viability from recognition. To determine whether this was so, we asked subjects, "If you had to make a prediction, which candidate do you think will win the election?" Partitioning by condition, we see significant differences, as shown in the last panel of Figure 2. In the control condition, only 34% of subjects thought that Griffin would win. In the treatment condition, 46% be

lieved Griffin would win, a significant difference at ρ < 0.01. These results suggest that inferential processing is at work—an inferential process based not on traits or experience, but on viability.

These results relating recognition to viability are con sistent with scholarship on the recognition heuristic. The recognition-heuristic literature argues that recognition can serve as an ecologically valid heuristic, so long as

17The question read: "How many times would you guess that BEN GRIFFIN has held previous office?" Responses ("Never," "1," "2," or "3 or more") are scaled 0 (Never) to 1 (3 or more).

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

978 CINDY D. KAM AND ELIZABETH J. ZECHMEISTER

there is an ecological correlation between the mecha nism for recognition (e.g., media coverage) and the cri terion measure (e.g., population size, corporation size, tennis-tournament success, disease prevalence, or elec toral success). For example, Gaissmaier and Marewski (2011) show that subjects infer that a recognized politi cal party in Germany is likely to receive more votes than

an unrecognized party; they argue that this recognition based heuristic is ecologically valid because the structure of the information environment creates a positive rela tionship between the medium that promotes recognition

(newspaper coverage of parties) and electoral success. Marewski et al. (2010) note that recognition of political parties and candidates predicts who will win the election, and they demonstrate the ecological rationality for that connection. Thus, the relationship we uncover between recognition and viability judgments is an ecologically ra tional heuristic and conforms well to this established lit

erature. The lack of relationship between recognition and traits also makes sense, as there is not as likely to be an

ecological correlation between, say, media coverage and, say, positive traits or experience.

While our results are consistent with those predicted

by the recognition heuristic, they also extend this line of work by suggesting that inferences about viability can influence vote choice. Why would inferences about via bility predict a willingness to support a candidate? Cer tainly, viability is important in the strategic voting cal culus in multicandidate elections and primaries, but we think there are additional explanations that can apply, re gardless of election type. First, inferences about viability might induce behavior akin to bandwagoning because the majority opinion serves as a rational heuristic for infer ring the candidate's general quality: public opinion polls indicating electoral strength could serve as a proxy for the wisdom of the many. This informational heuristic is separable from the strategic consideration and can ap ply easily in two- or multicandidate elections. Second, inferences about viability might matter because individ uals simply prefer to back a winner (for discussions, see

Abramowitz 1987; Bartels 1988; Kenney and Rice 1994). Again, this psychological desire to back a winner is separa ble from the considerations that enter in multicandidate,

strategic voting.

Viability as a Causal Mechanism?

We next investigate the extent to which inferences about

viability might account for the change in candidate pref erence: that is, do viability inferences mediate the rela tionship between treatment and candidate preference? To

answer this question, we use the causal steps approach (Baron and Kenny 1986) with a Sobel test (Sobel 1982)

Table 2 Causal Steps Approach

Model 1 Model 2 Model 3

DV: Vote DV: Griffin's DV: Vote

for Griffin Viability for Griffin

Subliminal Priming 0.12" 0.13" 0.06

Name Condition (0.05) (0.05) (0.04)

Viability 0.51"*

(0.04) Intercept 0.33 0.34 0.16

(0.04) (0.04) (0.04) N 496 496 496

Note: Table entry is the OLS coefficient with standard error below. **p < 0.05; ***p < 0.01, two-tailed.

and the average causal mediation effect (ACME) approach (Imai, Keele, and Yamamoto 2010).

The causal steps approach (Baron and Kenny 1986) uses four steps to identify whether a covariate mediates

the effect of some other covariate (typically the experi mental treatment). For ease of presentation, we use linear probability models in the causal steps analysis. In Step 1, we identify the relationship between the treatment and the dependent variable. As shown in Table 2, the treat ment has a significant effect on vote choice. In Step 2, we

identify the relationship between the proposed mediator (perceptions of Griffin's viability) and treatment: as the results in Table 2 show, the treatment has a significant effect on perceptions of Griffin's viability. In Step 3, we

show that the proposed mediator (perceptions of Grif fin's viability) significantiy predicts vote choice, even in the presence of the treatment variable. Finally, in Step 4, we compare the estimated coefficients on treatment, across Models 1 and 3. The effect of the treatment is cut

in half, and the effect of the treatment is not even signif

icant once viability perceptions are included in Model 3. A Sobel (1982) test yielded a z-value of 2.95, ρ < 0.001, suggesting that viability inferences do significantly me diate the effect of the subliminal priming condition on

vote choice. Thus, inferences about Griffin's viability do

partially mediate the relationship between the subliminal priming treatment and vote choice.

One drawback to the above analyses is that they use

linear probability models on binary mediator and out come variables. To test for the robustness of the find

ing, we analyzed the continuous outcome variable of Griffin's feeling thermometer advantage. In that case, we

find similar evidence suggesting that viability mediates

the treatment's effect on candidate support.18

18Using the causal steps approach, the effect of the treatment declines from 4.98 (s.e. = 2.05, ρ < 0.02) in Model 1 to 3.10

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE SUPPORT 979

As a final test, we used the average causal media tion effect approach developed by Imai and colleagues (Imai, Keele, and Yamamoto 2010) to allow for speci fying the mediation and outcome models using the fa miliar probit specification, with bootstrapping. This ap proach suggests that 47% of the total effect of the treat

ment occurs via mediation, which is largely consistent with the causal steps comparison (Step 4).19 Overall, for both outcome variables and across the three pro cedures, our results suggest that the subliminal prim ing treatment increases Griffin's perceived viability as a candidate, which then translates into higher support, as measured by vote choice and feeling-thermometer advantage.

Theoretically, as we discussed earlier, there are good reasons to believe that inferences about viability influence

support, as individuals tend to prefer candidates who are

more likely to win and who are of high overall quality (a consideration they might infer to the extent they believe

the candidate has wide support). That said, we pause here to caution that, because we did not experimentally

manipulate viability, our empirical findings are suggestive

of a mediation process but not determinative of one.20 Nonetheless, the analyses we present here are consistent with the notion that, if anything, mere recognition affects

more than just support, but not much more: that is, it affects inferences about viability and not inferences about

experience or traits.

The Limits of Subliminal Priming and Name Recognition: Studies 2 and 3

Study 1 shows that subliminally priming a candidate's name has a significant and sizable effect on candidate preference, an effect that appears to register via inferences

about viability. These findings occurred under minimal information, when subjects were given no information about the candidates aside from their names—an infor

(s.e. = 1.92, ρ ~ 0.11) in Model 3. Inferences about Griffin's via bility significantly predict Griffin's FT advantage (b = 14.91, s.e. = 1.73, ρ < 0.001). Sobel z-value = 2.43, ρ < 0.001.

"When we use the ACME approach with the continuous FT ad vantage variable, viability accounts for 37% of the treatment effect.

20An alternative interpretation of the results is that subjects ra tionalized a vote for Griffin by increasing their evaluation of his viability. We believe the evidence allows us to discount this pos sibility. If, indeed, subjects were seeking to rationalize their vote choice, then we would expect evaluations of Griffin's traits and ex perience to also be enhanced in the treatment condition, but they were not. Moreover, the viability question was asked after the trait battery and the experience question, thus dismissing any possibility of simple decay effects.

mation context that mimics many electoral decisions for American voters.

Are there limits to the usefulness of the name

recognition heuristic in electoral politics? Scholarship on heuristic decision making holds that the extent to which certain cues are used might depend on the menu of avail able cues as well as their applicability to the decision task

at hand (Higgins 1996; see also Marewski and Schooler 2011). As Marewski et al. (2010) document, the name recognition heuristic is ecologically valid in politics in that a correlation exists between media mentions, fly ers, and posters of political parties and candidates and election results; thus, name recognition is an ecologically valid heuristic on which voters might rely in making infer

ences about viability (see also Goldstein and Gigerenzer 2002). However, when other factors (partisanship, sex, race or ethnicity, appearance, or incumbency) are avail able, we propose that name recognition could become less relevant to preferential choice for three reasons: (1)

those factors may heavily influence vote choice directly, thus drowning out any name-recognition effect; (2) those factors may reduce reliance upon recognition in directly

determining vote choice; and (3) those factors may min imize the weight placed on viability as it indirectly influ ences vote choice.21 This line of reasoning does depart from what might be hypothesized if we were to general ize from the recognition-heuristic literature to the degree

that work suggests that the recognition heuristic is a non

compensatory, ecologically rational strategy, one that is used even in the presence of other, potentially pertinent

cues such as party affiliation and size (e.g., Marewski et al.

20 1 0).22 We note, though, that a key difference between

this work and our study is that we are principally focused

on candidate support, which is a preferential choice and thus different from the inductive inferences that are typi

cally the focus of research on the recognition heuristic. In

21Some scholarship on the recognition heuristic suggests that it is frequently a dominant strategy for individuals making inferences, even in cases where contradictory information that one would reasonably conclude is more valid is made available (see Pachur, Broder, and Marewski 2008; Marewski et al. 2010). We test recog nition against one of the most ubiquitous cues in the vote-choice literature: incumbency status; however, we could have found dif ferent effects if we had introduced a different piece of information, such as candidate gender, ethnicity, or age.

22As we indicated at the start, low-information elections are the

norm, and many of these likely appear to voters as little more than a toss-up choice between a set of relative strangers; in such a context, information about incumbency status exerts a significant effect, inducing both a strong tendency toward selecting the incumbent and, as we show here, overriding the effects of name recognition. It is possible that the effects of incumbency itself would be diminished by the presence of more information about the candidates and, as well, that name recognition might matter more in cases in which that additional information is unfavorable toward the incumbent.

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

98ο CINDY D. KAMAND ELIZABETH J. ZECHMEISTER

very recent work that has investigated consumer choice (rather than probabilistic inferences), Oeusoonthornwat

tana and Shanks (2010) do find that recognition affects product preference, but in a compensatory fashion.

In Studies 2 and 3, we focus on incumbency status, a cue that may be arguably more germane to vote choice

than name recognition, since incumbency at a minimum signals actual political experience.23 In low-information races that lack party cues, where sex, race, and ethnic ity are constant, incumbency might in fact be the only additional cue available to voters. For example, ballots in the states of California and Arkansas allow candidates

to list their occupations or current public office, which

in essence explicitly allows candidates to signal incum bency (Niemi and Herrnson 2003). Incumbent status (or current job position) also appears for statewide judicial candidates in Minnesota, Michigan, and Oregon. In our content analysis of sample ballots, we found that explicit incumbency cues were present in 8.9% of electoral de cisions (judgeships in Oregon). In this next study, we directly confront the extent to which incumbent or chal

lenger status might or might not mitigate the effects of name recognition.

Study 2 repeated the procedures in Study 1 for a ran domly selected portion of the subjects (the other subjects were routed to a different study).24 After answering ques

tions about the Griffin-Williams race, subjects were then told:

During the next few minutes we'll show you an other series of letters and potentially some words.

Later we may ask you more questions about let ter combinations. Between letter combinations,

you should again focus on the dot in the center of the screen. Press the space bar when you are ready to begin.

Subjects were again randomly assigned to one of two experimental conditions, in which they were exposed to a subliminally primed name ("JENKINS") or were not. All

other aspects of the subliminal priming procedure were identical to Study 1. The questions were also identical,

23 Our use of incumbency also allows us to weigh in, to at least some extent, on a debate in the literature regarding the heuristic value of information about incumbency status (in a study of low information judicial elections, for example, Klein and Baum 2001 find no significant effect of incumbency on electoral choice). That is, in contrast to scholars who identify a positive role for incum bency as a cue, other scholars have suggested that "voters are not strongly attracted by incumbency per se" (Jacobson 2009, 138; see also Krasno 1994).

24 A programming glitch ruined the first two days of data for this study, leaving us with 121 subjects in the treatment group and 33 in the control group.

with one exception. Immediately following the subliminal priming phase, subjects were asked a distracter question25

and then were asked to make a judgment between two new

hypothetical candidates.26

In the first question about the race, the vote-choice

question provided two response options: "Incumbent IOHN DAVIS" or "Challenger MILT JENKINS." Also in contrast with Study 1, across treatment and control groups, there were no statistically distinguishable dif ferences attributable to the experimental treatment. For

the vote-choice comparison,27 29% of treated subjects preferred Jenkins compared to 27% of control subjects, with a difference of proportions test suggesting no dis tinguishable differences (ρ ~ 0.85). Treated subjects gave the incumbent a 10.8-point advantage (s.e. = 2.60) on the feeling thermometer, while control subjects gave the incumbent a I2.8-point advantage (s.e. = 5.36), with a ί-test of means yielding no distinguishable differences (ρ ~ 0.74). There were no differences across the treated and control groups on trait evaluations ( ί-test of means

insignificant, ρ ~ 0.54), judgments of experience (ns, ρ ~ 0.84), nor viability inferences (ns, ρ ~ 0.83).

It is worth noting that the incumbent candidate was

judged extremely likely to win (84.9% of control sub jects and 83.3% of treated subjects guessed the incumbent would win). And, there is a strong relationship between vote choice and viability estimates (r = 0.46). But, these results suggest that an explicit, politically relevant heuris

tic (the incumbent/challenger tag) effectively eliminates

the effect of the recognition heuristic in predicting vote choice. These results, however, derive from a study that was administered immediately following Study 1 with a dramatically smaller sample size. In order to determine whether subject fatigue or power could have been re sponsible for these null findings, we replicated Study 2 at

a different location, with a different subject pool, and ad ministered it as a stand-alone subliminal priming study.

In Study 3, subjects were recruited from political sci ence courses at a large, public, western university. The 339

subjects were randomly assigned to either the treatment

25 "Which of the following letter combinations do you like most?"

26 As before, the selection of candidate names was deliberate, such that the first and last names for the incumbent ("John Davis") were far more commonplace than those for the challenger ("Milt Jenk ins"). In the field study presented later, we find that preferences for "Ben Griffin" and "Milt Jenkins" were perfectly equivalent in the control condition for that study; thus, while we prime differ ent names across the laboratory studies, we find no discernible difference in baseline preferences for those names.

27Hotelling's vector-of-means test suggests no differences in co variates across treatment and control groups, so we conduct simple differences of proportions and means tests.

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE SUPPORT ç8l

group (Ν = 259) or the control group (N = 80).28 The study was administered using the same computer software

with the same interface and identical questions. As with Study 2, the incumbent was far more popular than the challenger; indeed, the proportion of subjects supporting the incumbent versus the challenger was nearly identical in Study 3:72% of subjects preferred the incumbent com

pared with 28% of subjects who favored the challenger. Across the treatment and control groups, there were no statistically distinguishable differences in vote choice (ρ ~ 0.62), FT advantage (ρ ~ 0.61), trait evaluations (ρ ~ 0.92), experience (ρ ~ 0.42), or viability (ρ ~ 0.46). Again, the incumbent candidate was deemed to be highly viable (83.5% of control subjects and 86.8% of treated subjects estimated he would win), and there is a strong relationship between viability and vote choice (r = 0.38).

But, the name-recognition effect falls short in the pres

ence of the incumbent/challenger tag. In sum, replicating

Study 2 in a new venue, with a larger sample size and a fresh set of subjects, reiterates our findings above: the

presentation of the more applicable, politically relevant incumbent/challenger tag dramatically eliminates the effect of the name-recognition heuristic on candidate support.29

Studies 2 and 3 suggest that one explicit cue (incum

bent/challenger status) can trump another (name recog nition). That is, we find evidence of a substitution effect,

whereby individuals rely less on a particular cue when it

is presented alongside another, more attractive (poten tially more germane) cue. Such a substitution effect may be most likely to obtain in this particular case, where one

of the cues is based on feelings or some "nonanalytic" criterion (Jacoby et al. 1989).

By providing our subjects with the analytic cue of in cumbent/challenger status, we gave them an alternative, more applicable criterion by which they could assess the hypothetical candidates. These results differ from those

28 Subjects were evenly divided across men and women. They leaned strongly Democratic, with 17% identifying as Republican, 60% as Democrats, and the rest as Independent/Other. On the standard 7-point ideology measure, the average respondent was somewhat liberal. In terms of race, 26% of subjects identified as Asian, 12% as Hispanic, 2% as black, and 12% as other. Average age was 20.3.

29 Our original pilot study provides supporting evidence here. There, we presented short candidate biographies describing the candidates' families and occupations. The biographies also in dicated that one candidate was running for reelection (Mike Williams), and the other candidate was the challenger (Ben Grif fin). The treated group was subliminally primed with the name "GRIFFIN," and there was no treatment effect. Studies 2 and 3, which we describe in the text, represent a more precise test of the incumbency cue; the pilot study suggests that the null name recognition effect holds when additional information appears with incumbency/challenger status.

that would be hypothesized if we were to generalize from the recognition-heuristic literature: as we noted, that line

of work suggests that the recognition heuristic is a non compensatory strategy, one that is used even in the pres ence of other, potentially pertinent cues (Marewski et al. 2010). But, recall that the recognition-heuristic literature is based primarily upon probabilistic inferences, not mat ters of taste. Our results are consistent with those of Oeu

soonthornwattana and Shanks, who, like us, investigate the effect of recognition not on probabilistic inferences but on preferences (in their case, preferences for con sumer products such as peanut butter and other consumer

products). As they note, "there is nothing special about

recognition other than the fact that it is a highly accessi ble cue, one that can be contradicted or compensated for by other information" (2010,318). Our results also differ

from those reported by Panagopoulos and Green (2008),

who find a borderline significant result that nonparti san radio ads mentioning both candidates (and cueing incumbency) benefited challengers at the incumbents' expense. One way to explain the difference is the deci sion context. While the radio ads aired by Panagopoulos and Green (2008) did cue incumbency (by mentioning the chance to reelect candidate X), there is no evidence to suggest that the immediate decision context (i.e., the

written ballot) in the 49 mayoral contests explicitly cued incumbency, as was the case in our laboratory study. The information about incumbency status may not have been processed or retained by subjects. It is possible that the

name of the challenger became more recognizable as a consequence of the radio ads, and thus the challenger received a boost because the incumbency cue was not ex

plicitly activated in the voting booth. As such, the limiting case that we constructed in our lab for Studies 2 and 3

displays that name recognition can be shut down in the presence of an explicit alternative analytic criterion, but in

practice, it may continue to affect decision making in the voting booth, given the paucity of information available to citizens when they cast their votes.

Assessing Name Recognition in the Field: Study 4

While we believe the laboratory experiments pro vide convincing evidence of a causal relationship be tween name recognition and candidate support, we acknowledge that these results may be criticized on ex ternal validity grounds. External validity refers to the ex

tent to which causal findings can be generalized to other

people, other settings, and other operationalizations

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 20hu, 01 Jan 1976 12:34:56 UTC

All use subject to https://about.jstor.org/terms

982 CINDY D. KAMAND ELIZABETH J. ZECHMEISTER

(Campbell and Stanley 1963; for discussions, see Druck man and Kam 2011; Kam, Wilking, and Zechmeister 2007). We designed our field study primarily to address the generalizability of our particular operationalizations, but it has the added benefit of speaking to the generaliz ability of our results to other people and other settings.

In terms of operationalizations, our laboratory stud ies utilize subliminal priming, but in political life most, if not all, stimuli are not subliminal, but supraliminal— that is, visible to the naked eye.30 Even so, visible stimuli are not necessarily consciously processed. Citizens haphaz ardly come across myriad political advertisements, bill boards, yard signs, bumper stickers, and other visual dis plays in their daily lives, but they are likely not consciously

aware of them, or more importantly, do not consciously call them to mind in the voting booth.

We extend our laboratory studies to examine a real world example in which name recognition might matter: political yard signs. An ordinary citizen may drive by political yard signs and not notice them, but they might still affect support for the candidate. Many party activists

and political candidates believe yard signs are effective, but the subject has received scant scholarly attention.31 We

expect political yard signs to influence candidate support in much the same way as we find in our laboratory study.

To test this assertion, we implemented a small-scale field

study. The field study was conducted in a midsized city in May 2011, two months prior to an August county wide election. The study was timed so that candidates had begun to campaign for these local offices, but the election was not yet salient. At the time of the study, a small scattering of yard signs supporting candidates for council, mayor, and vice-mayor were beginning to appear in the area. For the yard signs, we used the name Ben Griffin, one of the names used in the laboratory studies.32 The treatment consisted of four professionally

printed, 18" χ 24" yard signs (a standard size), placed, with homeowner permission, in pairs roughly 100 feet apart from each other on what we will call / Street. The yard signs were simple in design, as shown in Figure 3. A

30A well-known exception is the RATS ad (Weinberger and Westen 2008).

3lKaid (1977) and Sommer(1979) provide some evidence that yard signs relate to electoral outcomes; Panagopoulos (2009) shows that yard signs can affect rates of political participation. Huckfeldt and Sprague (1992), moreover, demonstrate that yard signs can affect individuals' assessments of the electoral preferences present in their neighborhood or, in other words, the viability of candidates within a group of voters. None of these works have directly connected yard signs to a psychological theory of name recognition.

32By using a fictitious candidate, we avoided interfering with the outcome of the political race.

Figure 3 Field Study Yard Sign

BEN GRIFFIN

wide blue bar appeared at the top and the bottom of the

sign; in the middle, the name of our fictitious candidate was displayed in capital letters with the first name in red and the last name in blue.33

We placed the signs in front of a particular home on J Street in order to take advantage of natural assignment to driving routes, an assignment that is orthogonal to our study's purposes. The study population consisted of par ents with at least one child enrolled in a local elementary school. According to our data, roughly 85% of children at this school commute to campus by car. The school has a strict policy regarding the routes by which car drivers can

approach the school, a process that conveniently channels about half of all drivers down J Street, with the rest ar

riving via one of the other authorized routes. This strict policy, channeling one portion of drivers along J Street and the other portion of drivers along other routes, al lows us to exploit natural and exogenous assignment to driving routes.

Three days after the signs had been placed on the lawn, the elementary school Parent-Teacher Organization e-mailed the school's body of parents a link to a short Internet survey. Parents (one per household) were asked to complete the survey in order to earn $5 per survey for

the school. The sample consists of 82 car drivers. Of these,

46 (or 56%) indicated that they approached the school via the J Street route; the remainder constitutes the control

34 group.

33We used both red and blue to avoid signaling a particular political party.

34The school is centrally located within a fairly affluent commu nity. There are approximately 500 students enrolled. None of the suburban streets sees much traffic, other than at drop-off and

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE SUPPORT 983

Table 3 Analysis of Field Study

p-value Vote for Vote for for target Griffin Jenkins placebo (Target) (Placebo) difference

Treatment Group 0.239 0.130 0.09

(0.06) (0.05) Control Group 0.139 0.139 1.00

(0.06) (0.06) p-value of 0.13 0.91

treatment effect

Note: Table entry is the cell mean with standard error below.

The survey was designed to test the effect of name recognition on candidate support. At the time of the study,

five incumbents for the five council at-large seats had declared their intentions to run. The first question on the survey asked respondents to indicate their top three choices for these council seats. The seven response options were the five actual incumbent candidates plus two of the

fictitious candidates featured in our laboratory studies: Ben Griffin and Milt Jenkins.

Did recognition spurred on by political yard signs increase support for Ben Griffin in the treatment group? To determine if this is so, we examine the extent to which

survey respondents selected Ben Griffin as one of their top three choices for council. As shown in Table 3, in the control condition, only 13.9% of respondents placed Ben

Griffin among their top three choices, but in the treat ment condition, 23.9% of respondents placed Ben Griffin among their top three choices. This 10 percentage point difference is sizable given the modesty of the treatment. In

light of the small sample size, it is statistically significant

at generous levels (ρ ~ 0.13, one-tailed). Could these differences be explained by a simple pref

erence for fictitious names, or an anti-incumbent streak,

among the treatment group? This is, of course, a potential

vulnerability of our design, since it relies upon naturally occurring assignment to conditions (rather than random assignment). To protect our inferences against this pos sibility, we included a placebo name, Milt Jenkins, as a choice. Analysis of the placebo name shows that there is no difference between the treatment and control groups

in preference for Milt Jenkins: 13.9% of control subjects and 13.0% of treatment subjects placed Milt Jenkins in their top three (ρ ~ 0.91).

pick-up. There is no clear socioeconomic difference in the house types/neighborhoods that feed into either route to the school.

Finally, we can examine the rates of selection of the two fictitious names, within each condition. Among the treated subjects, 23.9% of subjects placed Ben Griffin in the top three set, but only 13.0% placed Milt Jenkins in the top three set, a statistically significant difference at ρ < 0.09, one-tailed. Among the control subjects, 13.9% placed Ben Griffin in the top three set, and the iden tical percentage, 13.9%, placed Milt Jenkins in the top three set. The results from this field study lend generaliz

ability to the claim established in our laboratory studies: name recognition increases candidate support in low information elections.

Conclusions

In this article, we have focused on the effects of name

recognition on electoral decision making. With our first laboratory experiment, we showed that subliminal pre sentations of a hypothetical candidate's name have signif icant effects on vote choice, affect, and inferences about

viability. These effects are large in magnitude and re flect an interconnected set of attitudes about a particular candidate. The results reflect Zajonc's (1968) concept of perceptual fluency, whereby mere exposure induces lik

ing. The results also reflect the literature on the "recog nition heuristic" (e.g., Gaissmaier and Marewski 2011; Marewski et al. 2009,2010), whereby recognition leads to inferences about viability. Our results reinforce and con nect these two literatures by identifying a direct pathway

between name recognition and candidate support and by

identifying an indirect pathway from name recognition to vote choice, through viability assessments. Moreover, our results demonstrate that recognition can operate on a subconscious level and does not require voters to explic itly report on whether or not they recognize a particular candidate.

We further considered the limits of name recogni tion, in particular assessing whether a more applicable cue would substitute for, and therefore eliminate the effects

of, name recognition on vote choice. The two supplemen

tary lab studies showed that the effects of mere recogni tion dissipate when a more politically relevant criterion (incumbency) explicitly appears in the decision context.

Our use of subliminal primes in the three laboratory

studies allows us to achieve high levels of internal validity

while still mimicking real-world psychological processes,

specifically those in which political information comes by

incidentally, and is thus subconsciously processed, leav ing individuals, at the time of decision making, unaware

of how they may be influenced by the information. Such

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

984 CINDY D. KAMAND ELIZABETH ]. ZECHMEISTER

processes are ubiquitous in political life. Our final demon stration, a small-scale field study, establishes the general

izability of our claims. Taking advantage of natural varia tion in driving routes, we showed that a modest three-day

intervention can increase candidate support in an actual electoral context. The field study extends the laboratory

experiments by identifying generalizability not only in operationalization (a yard sign as opposed to subliminal priming) but also in who is studied (parents as opposed to college students) and within what context (an actual election as opposed to a hypothetical one). In contrast to

existing scholarship that suggests that name recognition does not directly influence candidate support, we find clear evidence of a causal link. The convergence in our lab and field studies demonstrates that in low-information

races, name recognition can increase candidate support. In our lab experiment, we kept the level of exposure

constant among those who were treated with the sub liminal priming named trials. A natural extension would

examine whether the relationship between exposure and candidate support would vary depending upon the degree

of exposure. Relating this to the real world of politics, we

might consider whether it matters if an individual drives by a particular candidate's yard sign once, 20 times, or 100 times. Our present data cannot speak to this partic ular question, but varying the frequency of exposure is certainly something that our research design could easily accommodate.

Additionally, in each experiment, we primed only one name, in order to obtain the cleanest test of the re

lationship between name recognition and candidate sup port. And, we primed the less innately "familiar" name to

mimic the incumbent/challenger dynamics around which the name-recognition literature has revolved. With a fully

factorial design, we would have manipulated the more familiar name, and we could have investigated whether repeated exposure affects more and less familiar names disproportionately. It is possible that name-recognition effects are stronger in races where support for one candi date is less one-sided, where the signal of whom to support

is less clear. While we do not have the data to adjudicate on this possibility, it does present another way to consider

the limiting conditions for name recognition's effect on

candidate support. In our studies, we elected to prime only one name at a time. But citizens may be exposed to more than one candidate in an election. Future work

could certainly examine this question by priming more than one name and by examining the extent to which the degree of exposure varies as well.

We chose to focus on the attenuating effect of incum

bency on the relationship between name recognition and

candidate support, given the prominent role that incum

bency plays in the name-recognition literature. There are,

however, several additional moderating factors that could be investigated in the future. For example, one might be interested in the extent to which other cues, such

as partisanship, race, ethnicity, or sex could attenuate, or conversely accentuate, the relationship between name

recognition and candidate support. One could also give more consideration to whether any press is good press; that is, is a negative affective reaction to a name likely to

benefit or harm a candidate? Zajonc's (2001) perceptual fluency mechanism suggests that mere exposure primar

ily works if a stimulus is initially judged to be affectively

neutral or positive in nature. In our study, we purpose fully selected names that were neutral, and our results are consistent with that intention.35 If, instead, a candidate's

name triggers some negative reaction, then it is possible that repeated exposure to the name may not necessar ily influence candidate support: indeed, it could reduce support.36 This is, however, an open question, and our argument here departs from the predictions based on the recognition heuristic. Goldstein and Gigerenzer (2002) note that repeated exposure, even when tied to a very negative association, can still lead to name recognition: they point to the visually shocking series of Benetton ads

as evidence that name recognition can skyrocket even in the presence of affectively repellant images. However, their discussion of the Benetton ads focuses solely on name recognition per se and not on evaluative judgments. Because we are interested not only in name recognition per se but also in the evaluative judgments that arise from

name recognition, we suggest that repeated exposure in the presence of affectively negative stimuli could limit the

effect of name recognition on candidate support. In considering the relevance of our findings, we have

focused principally on the plethora of low-information elections at the subnational level and in primary elections in the United States. In other countries and party systems,

35Prior to our study, we searched the names on Google and did not encounter any evidence that either name was associated with a well-known past scandal or other negative event.

36We can imagine two processes that could lead to this result. First, the candidate's name could be subconsciously associated with neg ative considerations; in the lab, this association could be created by "tagging" the name with a negative word or image, but on the basis of our research we would suspect that this negative tag would have to be highly charged in order to have an effect (see footnote 4 and the online supporting information). Second, the candidate's name could introduce additional, analytical information; for example, a candidate's name could prompt a racially resentful citizen to infer that the candidate is a racial minority and therefore to devalue that option. This possibility seems more plausible when we consider real-world scenarios and, as well, parallels our argument about incumbency status diminishing the effect of name recognition; if a comparatively neutral piece of information can drown out the effects of name recognition, then negative analytical information should do the same.

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

NAME RECOGNITION AND CANDIDATE S UPPORT 985

however, individuals routinely face elections with more than two dominant choices and where partisanship serves as a weaker cue. In those contexts, name recognition could

exert particularly important effects even in national elec tions, in particular with respect to candidates promoted by smaller, less visible parties, as shown in Gaissmaier and Marewski's (2011) study of national- and state-level elections in Germany.

In the United States, the mass media devote a great deal of attention to high-profile national and state elec tions. In American political life, however, these high profile elections are the exception rather than the rule.

The vast majority of decisions confronting voters feature names of candidates who may be virtually unknown to them. What we have shown is that in the absence of an

explicit alternative analytic criterion—such as that re garding incumbent/challenger status—citizens may rely

on name recognition as a heuristic. They will use this name-recognition heuristic to make inferences about can

didates' viability, and these inferences about viability will shape their voting decisions and degree of affect for the candidates. From these four studies of name recogni tion and vote choice, we conclude that Oscar Wilde was

right—at least when it comes to political candidates in low-information contexts.

References

Abramowitz, Alan I. 1975. "Name Familiarity, Reputation, and the Incumbency Effect in a Congressional Election." Western Political Quarterly 28(4): 668-84.

Abramowitz, Alan I. 1987. "Candidate Choice Before the Con vention: The Democrats in 1984." Political Behavior 9(1): 49-61.

Baron, Reuben M., and David A. Kenny. 1986. "The Moderator Mediator Variable Distinction in Social Psychological Re search: Conceptual, Strategic, and Statistical Considera tions." Journal of Personality and Social Psychology 51(6): 1173-82.

Bartels, Larry M. 1988. Presidential Primaries and the Dynamics of Public Choice. Princeton, NJ: Princeton University Press.

Burden, Barry. 2002. "When Bad Press Is Good News." Harvard International Journal of Press/Politics 7(3): 76-89.

Burgess, Thomas D. G., II, and Stephen M. Sales. 1971. "Atti tudinal Effects of Mere Exposure." Journal of Experimental Social Psychology 7(4): 461-72.

Campbell, Donald T., and Julian C. Stanley. 1963. Experimen tal and Quasi-Experimental Designs for Research. Boston: Houghton Mifflin.

Campbell, James E., John R. Alford, and Keith Henry. 1984. "Television Markets and Congressional Elections." Legisla tive Studies Quarterly 9(4): 665-78.

Coates, Sarah L., Laurie T. Butler, and Dianne C. Berry. 2004. "Implicit Memory: A Prime Example for Brand Considera tion and Choice." Applied Cognitive Psychology 18:1195-211.

Coates, Sarah L., Laurie T. Butler, and Dianne C. Berry. 2006. "Implicit Memory and Consumer Choice: The Mediating Role of Brand Familiarity." Applied Cognitive Psychology 20: 1101-16.

Coleman, John J., and Paul F. Manna. 2000. "Congressional Campaign Spending and the Quality of Democracy." Journal of Politics 62(3): 757-89.

Druckman, James N., and Cindy D. Kam. 2011. "Students as Experimental Participants: A Defense of the 'Narrow Data Base'." In Handbook of Experimental Political Science, ed. James N. Druckman, Donald P. Green, James H. Kuklinski, and Arthur Lupia. New York: Cambridge University Press, 41-57.

Gaissmaier, Wolfgang, and Julian N. Marewski. 2011. "Fore casting Elections with Mere Recognition from Small, Lousy Samples: A Comparison of Collective Recognition, Wisdom of Crowds, and Representative Polls." Judgment and Decision Making6(1): 73-88.

Goldenberg, Edie N., and Michael W. Traugott. 1980. "Con gressional Campaign Effects on Candidate Recognition and Evaluation." Political Behavior 2(1): 61-90.

Goldstein, Daniel G., and Gerd Gigerenzer. 2002. "Models of Ecological Rationality: The Recognition Heuristic." Psycho logical Review 109(1): 75-90.

Higgins, E. Tory. 1996. "Knowledge Activation: Accessibility, Applicability, and Salience." In Social Psychology: Handbook of Basic Principles, ed. E. Tory Higgins and Arie W. Kruglan ski. New York: Guilford Press, 133-68.

Hilbig, Benjamin E. 2010. "Precise Models Deserve Precise Mea sures: A Methodological Dissection." Judgment and Decision Making5(4): 272-84.

Holden, Stephen J. S., and Marc Vanhuele. 1999. "Know the Name, Forget the Exposure: Brand Familiarity versus Mem ory of Exposure Context." Psychology & Marketing 16(6): 479-96.

Huckfeldt, Robert, and John Sprague. 1992. "Political Parties and Electoral Mobilization: Political Structure, Social Struc ture, and the Party Canvass." American Political Science Re view 86(1): 70-86.

Imai, Kosuke, Luke Keele, and Teppei Yamamoto. 2010. "Iden tification, Inference, and Sensitivity Analysis for Causal Me diation Effects." Statistical Science 25(1): 51-71.

Jacobson, Gary C. 1978. "The Effects of Campaign Spending in Congressional Elections." American Political Science Review 72(2): 469-91.

Jacobson, Gary C. 2006. "Campaign Spending Effects in U.S. Senate Elections: Evidence from the National Annenberg Election Survey." Electoral Studies 25(2): 195-226.

Jacobson, Gary C. 2009. The Politics of Congressional Elections. 7th ed. New York: Pearson Longman.

Jacoby, Larry J., Colleen Kelley, Judith Brown, and Jen nifer Jasechko. 1989. "Becoming Famous Overnight: Lim its on the Ability to Avoid Unconscious Influences of the Past." Journal of Personality and Social Psychology 56(3): 326-38.

Kaid, Lynda Lee. 1977. "The Neglected Candidate: Interpersonal Communication in Political Campaigns." Western Journal of Communication 41(4): 245-52.

Kam, Cindy D., Jennifer R. Wilking, and Elizabeth J. Zechmeis ter. 2007. "Beyond the 'Narrow Database': An Alternative

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms

CINDYD. KAMAND ELIZABETH J. ZECHMEISTER

Convenience Sample for Experimental Research." Political Behavior 29(4): 415-40.

Kenney, Patrick J., and Tom W. Rice. 1994. "The Psychology of Political Momentum." Political Research Quarterly 47(4): 923-38.

Klein, David, and Lawrence Baum. 2001. "Ballot Information and Voting Decisions in Judicial Elections." Political Research Quarterly 54(4): 709-28.

Krasno, Jonathan S. 1994. Challengers, Competition, and Reelec tion. New Haven, CT: Yale University Press.

Krosnick, Jon Α., and Duane Ε Alwin. 1987. "An Evaluation of a Cognitive Theory of Response-Order Effects in Survey Measurement." Public Opinion Quarterly 51(2): 201-19.

Krosnick, Jon Α., Andrew L. Betz, Lee J. Jussim, and Ann R. Lynn. 1992. "Subliminal Conditioning of Attitudes." Personality and Social Psychology Bulletin 18(2): 152-62.

Kunst-Wilson, William Raft, and Robert B. Zajonc. 1980. "Affec tive Discrimination of Stimuli That Cannot Be Recognized." Science 207(1): 557-58.

Lenart, Silvo. 1997. "Naming Names in a Midwestern Town: The Salience of Democratic Presidential Hopefuls in Early 1992." Political Behavior 19(4): 365-82.

Levy, Dena, and Peverill Squire. 2000. "Television Markets and the Competitiveness of U.S. House Elections." Legislative Studies Quarterly 25(2): 313-25.

Mann, Thomas E., and Raymond E. Wolfinger. 1980. "Can didates and Parties in Congressional Elections." American Political Science Review 74(3): 617-32.

Marewski, Julian N., Wolfgang Gaissmaier, Lael J. Schooler, Daniel G. Goldstein, and Gerd Gigerenzer. 2009. "Do Voters Use Episodic Knowledge to Rely on Recognition?" Presented at the Proceedings of the 31st Annual Conference of the Cognitive Science Society.

Marewski, Julian N., Wolfgang Gaissmaier, Lael J. Schooler, Daniel G. Goldstein, and Gerd Gigerenzer. 2010. "From Recognition to Decisions: Extending and Testing Recognition-Based Models for Multialternative Inference." Psychonomic Bulletin & Review 17(3): 287—309.

Marewski, Julian N., and Lael J. Schooler. 2011. "Cognitive Niches: An Ecological Model of Strategy Selection." Psycho logical Review 118(3): 393-497.

Matson, Marsha, and Terri Susan Fine. 2006. "Gender, Ethnic ity, and Ballot Information: Ballot Cues in Low-Information Elections." State Politics and Policy Quarterly 6(1): 49-72.

Miller, Joanne M., and Jon A. Krosnick. 1998. "The Impact of Candidate Name Order on Election Outcomes." Public

Opinion Quarterly 62(3): 291-330.

Niemi, Richard G., and Paul S. Herrnson. 2003. "Beyond the Butterfly: The Complexity of U.S. Ballots." Perspectives on Politics 1(2): 317-26.

Oeusoonthornwattana, Onvara, and David R. Shanks. 2010. "I Like What I Know: Is Recognition a Non-Compensatory Determiner of Consumer Choice?" Judgment and Decision Making5(4): 310-25.

Orne, Martin T. 1962. "On the Social Psychology of the Psy chology Experiment." American Psychologist 17: 776-83.

Pachur, Thorsten, Arndt Broder, and Julian N. Marewski. 2008. "The Recognition Heuristic in Memory-Based Inference: Is

Recognition a Non-Compensatory Cue?" Journal of Behav ioral Decision Making 21: 183-210.

Panagopoulos, Costas. 2009. "Street Fight: The Impact of a Street Sign Campaign on Voter Turnout." Electoral Studies 28(2): 309-13.

Panagopoulos, Costas, and Donald P. Green. 2008. "Field Ex periments Testing the Impact of Radio Advertisements on Electoral Competition." American Journal of Political Science 52(1): 156-68.

Prior, Markus. 2006. "The Incumbent in the Living Room: The Rise of Television and the Incumbency Advantage in U.S. House Elections." Journal of Politics 68(3): 657-73.

Schaffner, Brian F., and Matthew J. Streb. 2002. "The Parti san Heuristic in Low-Information Elections." Public Opinion Quarterly 66: 559-81.

Serra, George, and Albert D. Cover. 1992. "The Electoral Con sequences of Perquisite Use: The Casework Case." Legislative Studies Quarterly 17(2): 233-46.

Sobel, Michael E. 1982. "Asymptotic Confidence Intervals for Direct Effects in Structural Equation Models." In Sociological Methodology, ed. S. Leinhardt. Washington, DC: American Sociological Association, 290-312.

Solowiej, Lisa, and Thomas L. Brunell. 2003. "The Entrance of Women to the U.S. Congress: The Widow Effect." Political Research Quarterly 48(4): 891-917.

Sommer, Barbara. 1979. "Front Yard Signs as Predictors of Elec tion Outcome." Political Methodology 6: 237-40.

Squire, Peverill. 1995. "Candidates, Money, and Voters: Assess ing the State of Congressional Elections Research." Political Research Quarterly 48(4): 891-917.

Standing, Lionel. 1973. "Learning 10,000 Pictures." Quarterly Journal of Experimental Psychology 25, 207-22.

Stokes, Donald E., and Warren E. Miller. 1962. "Party Govern ment and the Saliency of Congress." Public Opinion Quar terly 26(4): 531-46.

Van den Bussche, Eva, Wim Van den Noortgate, and Bert Reynvoet. 2009. "Mechanisms of Masked Priming: A Meta Analysis." Psychological Bulletin 135(3): 452-77.

Weinberger, Joel, and Drew Westen. 2008. "RATS, We Should Have Used Clinton: Subliminal Priming in Political Cam paigns." Political Psychology 29(5): 631-51.

Wilde, Oscar. [1891 ) 2010. The Picture of Dorian Gray. London: Bibliolis Books.

Zajonc, Robert B. 1968. "Attitudinal Effects of Mere Exposure." Journal of Personality and Social Psychology 9(2): 1-27.

Zajonc, Robert B. 2001. "Mere Exposure: A Gateway to the Sub liminal." Current Directions in Psychological Science 10(6): 224-28.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's website:

• Experimental Design • Covariates

• Subliminal Stimuli Recall and Recognition

This content downloaded from �������������184.54.114.89 on Sat, 27 Jun 2020 18:50:32 UTC�������������

All use subject to https://about.jstor.org/terms