Abstract Citizens are asked to make many judgments in politics, often in the face of scarce information and limited motivation. In making political judgments, citizens may relyupona variety of cues, including the partisanship, ethnicity, race, or sex of candidates. Some cues, however, are more democratically troublesome than others. Democratic norms of equality suggest that attitudes towards racial or ethnic groups should not influence citizens��� evaluations of candidates. Often, however, attitudes towards these groups do matter. This article identifies a lim- iting condition on the effect of group attitudes: the presence of a party cue. I demonstrate that attitudes towards Hispanics influence willingness to support a Hispanic candidate, but only in the absence of a party cue. The article also con- tributes to existing work by analyzing both explicit and implicit measures of attitudes towards groups. Explicit measures include stereotypes and feeling thermometers implicit measures are derived from a subliminal priming task. Subjects with positive attitudes towards Hispanics (whether these attitudes were measured implicitly or explicitly) were more likely to support the Hispanic can- didate, in the absence of party cues. Subjects with negative attitudes towards Hispanics were less likely to support the Hispanic candidate, in the absence of partycues.Thepresenceofpartycues,however,eliminatestheimpactofattitudes towards Hispanics on political choice. Keywords Implicit measures �� Subliminal priming �� Stereotypes �� Racism �� Party cues �� Heuristics This paper was originally presented at the 2004 Annual Meetings of the American Political Science Association, Chicago, IL. C. D. Kam (&) Department of Political Science, University of California, Davis, One Shields Avenue, Davis, CA 95616, USA e-mail: cdkam@ucdavis.edu 123 Polit Behav (2007) 29:343���367 DOI 10.1007/s11109-007-9030-0 ORIGINAL PAPER Implicit Attitudes, Explicit Choices: When Subliminal Priming Predicts Candidate Preference Cindy D. Kam Published online: 28 March 2007 �� Springer Science+Business Media, LLC 2007

Citizens are asked to make many judgments in politics, often in the face of scarce information and limited motivation. In making political judgments, citizens may rely upon a variety of cues, including the partisanship, ethnicity, race, or sex of candidates. Some cues, however, are more democratically troublesome than others. Despite democratic norms of equality and toler- ance, citizens��� attitudes towards racial or ethnic groups may influence their willingness to support a candidate of a particular ethnic or racial descent. Yet researchers face a steep hurdle in acquiring valid measures of citizens��� attitudes towards racial and ethnic groups and in identifying the impact of those attitudes on political choice. The pressures of social desirability and the conscious (or subconscious) suppression of attitudes towards racial and ethnic groups make it difficult for researchers to identify the political con- sequences of citizens��� attitudes towards racial and ethnic groups. In this article, I propose one solution to this measurement problem���an implicit measure of attitudes towards groups based on subliminal priming. Further, I demonstrate the conditions under which subliminal priming predicts candi- date preference. Using experimental data, I examine the impact of an implicit measure of attitudes towards an ethnic group on citizens��� willingness to support a minority candidate. I analyze the impact of ethnic attitudes in the presence and absence of more legitimate political information about candidates: their party affilia- tion. I find that attitudes towards Hispanics predict support for a Hispanic candidate, but only in the absence of party cues. Subjects with positive atti- tudes towards Hispanics (whether these attitudes were measured implicitly or explicitly) were more likely to support the Hispanic candidate, in the absence of party cues conversely, subjects with negative attitudes towards Hispanics (again, whether these attitudes were measured implicitly or explicitly) were less likely to support the Hispanic candidate in the absence of party cues. The presence of the party cue essentially eliminates the impact of ethnic group attitudes on political choice. Measuring Attitudes in the Blink of an Eye A relatively new body of work in social psychology has investigated the im- pact of implicit measures of attitudes on opinions and behaviors (for a review, see Fazio & Olson, 2003 Kihlstrom, 2004). The implicit measure used in this article, subliminal priming, is based upon associative network models of memory. These associative network models suggest that individuals store an array of considerations in memory that are cognitively and affectively con- nected, to varying degrees. An individual may possess an array of consider- ations about racial, ethnic, and social groups, or an array of considerations about political figures and political candidates (see, e.g., Lodge & Taber, 2000, 2005). The activation of one consideration in the associative network (e.g., a stereotype of a social group or a social group name) should facilitate the activation of other, related considerations (e.g., another stereotype of that 344 Polit Behav (2007) 29:343���367 123

social group or an affective reaction to that social group) (for additional de- tails, see Carlston & Smith, 1996 Lodge & Taber, 2000). In subliminal priming, associations in memory are measured using response latencies. First, a subliminal prime appears it is a stimulus (e.g., a word or an image) that is displayed to the subject outside of the subject���s conscious awareness. This prime is quickly followed by an attitude-object. The subject is asked to categorize that attitude-object, and the time it takes to categorize the attitude-object constitutes the relevant response latency. When the subliminal prime and the attitude-object are connected with each other (by affective or cognitive association), the response latency should be short. When the prime and attitude-object are not connected (or are more weakly connected), then the response latency should be long. The measure is implicit in two ways. First, the primes appear for such a short duration that they occur outside the conscious awareness of subjects. Subjects are not capable of guessing or recognizing the words that appear as subliminal primes. Second, the responses that are used to measure subjects��� evaluations of groups are response latencies rather than conscious self-reports to questions. Existing work, for example, has used group names, stereotypes, and even photographs as subliminal primes (Devine, 1989 Draine & Greenwald, 1998 Fazio, Jackson, Dunton, & Williams, 1995). The key feature of implicit measures of attitudes is that subjects are often unaware that their attitudes are being measured and are thus unable to exert conscious control over their responses. In this way, implicit measures of attitudes have several comparative advantages over explicit measures. The first is that when explicit measures are used, individuals may not reveal their true attitudes or preferences because of social desirability biases (Berinsky, 2004), thus attenuating the magnitude of the relationship that researchers identify between attitudes and political outcomes. A second comparative advantage of implicit measures is that individuals may not even be aware of their true preferences or attitudes (Banaji & Greenwald, 1994 Nisbett & Wilson, 1977). Although implicit measures may tap the underlying latent sentiment, they may or may not be strongly correlated with explicit measures, depending on the circumstances that face the perceiver for example, individuals who seek to manage the impression they make on the interviewer would display less correlation between their explicit and implicit measures that individuals not so motivated (Nosek, 2005 Nosek & Banaji, 2003). Implicit measures of attitudes may not necessarily dominate an individual���s subsequent attitudes or behav- iors, since perceivers can consciously over-ride them. However, they can be influential: implicit attitudes predict individual behaviors and choices as well as interpersonal interactions.1 Implicit measures of attitudes have begun to make their way into political science, largely due to the pioneering work of Stony Brook political psy- 1 For example, Fazio et al. (1995) suggest that whites��� implicit attitudes towards blacks are cor- related with less friendly interactions between white subjects and a black experimenter. Wilson, Lindsey, and Schooler (2000) find that implicit measures of hostility towards blacks are correlated with less physical contact between a white subject and a black confederate. Polit Behav (2007) 29:343���367 345 123

chologists, Lodge and Taber, and colleagues. In their work, Lodge and Taber have used the subliminal priming procedure to test their Hot Cognition hypothesis: that most socio-political groups, figures, symbols, and ideas con- tain an affective component, and this affective component is more accessible than their cognitive components. For example, Lodge and Taber (2000) ex- posed subjects to a story about a Congressperson, and then used the Con- gressperson���s name as a subliminal prime. They use reaction times to affect- laden targets (e.g., traits) versus cognitive targets (e.g., issues) as the depen- dent variable and find that responses to affect-laden targets are far faster, thus supporting the Hot Cognition hypothesis. In a separate study, Lodge and Taber (2005) showed that political figures, political issues, and political groups trigger affective responses that can be measured through the subliminal priming procedure. Finally, Lodge and Taber and colleagues have also used responses to the subliminal priming task as a dependent variable to identify subjects��� affective reactions to ingroups and outgroups. They find that sub- liminal priming responses reliably identify positive affect for ingroups over outgroups (Lodge, Taber, Burdein, & Yener, 2003). The present article builds upon this existing research. In their analyses, Lodge and Taber and colleagues have used responses to the subliminal priming procedure as a means of tapping individuals��� affective reactions to socio-political stimuli. That is, responses to the subliminal priming procedure have been used as a dependent variable. The present article uses the implicit measure of individuals��� affective reactions as an independent variable, to determine the political consequences of this innovative measure on an explicit political choice. Implicit measures of attitudes should be particularly useful in predicting political choice. Implicit measures are likely to have their greatest predictive power when citizens make casual, low effort inferences. Under these condi- tions, citizens are not necessarily devoting the requisite cognitive effort into monitoring the underlying attitudes that contribute to their judgments. Rea- soning from their dual-process model of attitude and behavior formation, Fazio and Olson (2003) suggest, ������When motivation and/or opportunity are low, behavior is expected to be largely a function of the automatically acti- vated attitude, and hence, the implicit measure should prove predictive������ (305). Implicit measures should be useful here due to the low-motivation, low- information environment. In such an environment, citizens are likely to en- gage in default, or automatic processing. As such, their implicit attitudes may creep into their candidate preferences, absent motivations to take in other, more politically relevant information. Candidates��� Characteristics and Citizens��� Preferences In evaluating political candidates, citizens are prone to rely on several cues, among them the perceived race or ethnicity of the candidate and the party affiliation of the candidate. The perceived race or ethnicity of a candidate 346 Polit Behav (2007) 29:343���367 123

might shape preference for the candidate by activating attitudes towards relevant groups. These activated attitudes can subsequently influence voter decision-making. Psychologist Jerome Bruner (1957) notes that, ������Perception involves an act of categorization������ (123) and that ������all perceptual experience is necessarily the end product of a categorization process������ (124). This ac- count suggests that the perceived demographics of candidates will likely evoke categories, and voter decision-making may be based on the attitudes associated with the demographic group(s) in which the candidate has been categorized. Voters may swerve away from a candidate because of the voters��� negative attitudes towards that candidate���s group(s) or may rally around a candidate because of the voters��� positive attitudes towards a candidate���s group(s).2 There are (at least) two ways of empirically identifying the impact of attitudes towards groups on citizens��� preferences for candidates: (1) explicit measures of attitudes towards groups and (2) implicit measures of attitudes towards groups. Nearly all of the existing literature on this topic relies on explicit measures: respondents��� consciously reported opinions, such as feeling thermometer scores or stereotype assessments. This approach has been predominantly analyzed in the context of white evaluations of black candi- dates. Terkildsen (1993), for example, proposes that stereotypes of blacks as a group are automatically activated by questions that ask the voter to evaluate a black candidate in her study, voters with negative attitudes towards blacks rated a black candidate less favorably than a white candidate. Citrin, Sears, and Green (1990) show that whites��� antiblack sentiment predicts reluctance to support black candidates.3 This article presents a methodological innovation 2 Citizens may also support candidates as a result of perceptions of shared demographic char- acteristics with the candidate. As Sigelman and Sigelman (1984) suggest, ������The literature on interpersonal attraction... provides abundant evidence that people consistently use similarity to themselves as a major basis for finding others attractive or unattractive������ (264). One mechanism underlying this proposition could be that voters who share similar demographic characteristics with a candidate believe that what the candidate does in office will benefit them conversely, voters who do not share those characteristics may see such a candidate as posing a threat to them. The present sample���s composition makes it difficult to test this proposition with much certainty however, the data analysis suggests that Hispanic ethnicity predicts willingness to support a Hispanic candidate, when group attitudes are not included in the model. The ethnicity effect fades away, however, once attitudes towards Hispanics as a group are incorporated into the model. Hence, a group-attitudes account is more consistent with the data than a shared demographics account. Voters might also use perceptions of candidates��� demographic characteristics as a way of predicting what policy actions the candidate will take in office as Pitkin (1967) notes, ������We tend to assume that people���s characteristics are a guide to the actions they will take������ (89). Similarly, McDermott (1997) notes, ������If voters stereotype candidates by group affiliation, then demographic characteristics should provide information based on commonly held social stereotypes������ (271). The candidate���s race/ethnicity might thus activate group-based stereotypes which then shape the vo- ter���s inferences about the candidate���s ideological positions, issue preferences, traits, or compe- tence in handling issues. 3 There has been little research on voters��� reactions to candidates of other ethnic backgrounds. One exception is Sigelman, Sigelman, Walkosz, and Nitz (1995), which includes one Hispanic candidate in their experimental manipulations. Polit Behav (2007) 29:343���367 347 123