All we need is the candidate’s face: The irrelevance of information about political coalition affiliation and campaign promises

Abstract Recent research has indicated that judgments of competence based on very short exposure to political candidates’ faces reliably predict electoral success. An unexplored question is whether presenting written information of the kind to which voters are typically exposed during an election alongside candidates’ faces affects competence judgments. We conducted three studies using photographs of 16 pairs of competing politicians in 16 medium-sized towns of northeast Italy as stimuli. Study 1 confirmed the external validity of earlier research in which participants were exposed to candidates’ faces without providing any other information. Study 2a showed that competence judgments were not subject to in-group favoritism: candidates’ faces were presented alongside information about the political coalition to which they belonged (center left; center right) to participants who declared a left or right political orientation. Finally, Study 2c compared the competence inferences made in Study 1 (face-only condition) with those of Study 2a (face plus political coalition label) and with new inferences (Study 2b) based on candidates’ faces plus information about campaign promises (greater equality; lower taxes). The results showed that automatic competence inferences are not substantially modified when relevant written information is presented alongside candidates’ faces.


ABOUT THE AUTHORS
Lucia Mannetti is a full professor of Social Psychology at Sapienza, University of Rome. Her research interests focus on epistemic motivations, group and intergroup processes, and persuasive communication.
Ambra Brizi is a post-doctoral fellow at Sapienza, University of Rome. She received her PhD in social psychology from Sapienza, University of Rome in 2011 and has research interests in intergroup relationship and motivation.
Jocelyn Belanger is an assistant professor of psychology at the New York University in Abu Dhabi. He received his PhD from the University of Maryland in 2013 and his research interests in social cognition, goal pursuit, and interpersonal relationships.
Ilaria Bufalari is an assistant professor of social psychology at Sapienza, University of Rome. She received her PhD in cognitive neuroscience from Sapienza, University of Rome in 2008. Her research interests focus on body representations and face perception.

PUBLIC INTEREST STATEMENT
Recent research has challenged the assumption of citizens, as rational actors, giving their votes on the basis of informed and systematic evaluation of candidates. Some recent studies point out that judgment of competence based on short exposure to political candidates' faces predicts success in political elections. Results of our study: (a) confirm validity of previous results in the Italian context; (b) show that there is no evaluative bias due to the fit between participants political preferences and candidate political affiliation, and (c) show that verbal information concerning political affiliation or campaign promises does not substantially modify fast inferences based on candidates' faces. Our results suggest that, at least in the contemporary Italian context, political parties are not any more able to influence citizens' choices and that actual voters may neglect relevant information such as political coalition affiliations and candidates' main political goals, when casting their vote.

Introduction
In order to explain voting behavior, modern political scientists have mostly assumed that citizens are rational actors, making choices based on relevant information, while being unaffected by irrelevant cues (Kuklinski & Quirk, 2000;Quattrone & Tversky, 1988). The assumption of rationality is, more or less explicitly, at the heart of citizens' trust in the democratic system, leaving everyone hoping that their fellow citizens are good at choosing the best leaders and the right policies.
However, this assumption has been challenged by research in cognitive psychology and behavioral economics (Ariely, 2009;Tversky & Kahneman, 1973, 1974. There is mounting evidence that in order to cope with the complexity of life human beings routinely rely on heuristics and automatic information processing and make relatively parsimonious use of more reflective ways of thinking (System 1s and 2; Kahneman, 2003;Kahneman & Frederick, 2002;Lieberman, McBratney, & Krovitz, 2002;Petty & Cacioppo, 1986;Sloman, 1986). Consistent with these notions, research in the political domain has revealed that voters are usually not well informed about policies (Converse, 1964;Zaller, 1992), do not understand economic issues (Bartels, 2005;Caplan, 2007), have idiosyncratic view of ideological differences (Conover & Feldman, 1981) and inconsistent preferences (Quattrone & Tversky, 1988), and largely predicate their evaluation of political candidates and political issues on emotions (Lodge & Taber, 2005).
Supporting the idea that people are often cognitive misers, recent research has indicated that judgments of competence based on very brief exposure to political candidates' faces reliably predict electoral success (Hall, Goren, Chaiken, & Todorov, 2009;Lenz & Lawson, 2007;Sussman, Petkova, & Todorov, 2013;Todorov, Mandisodza, Goren, & Hall, 2005). In fact, explicit instructions to make a judgment based on careful deliberations elicit competence judgments that are less predictive of electoral outcomes than those made rapidly and intuitively between 250 and 2,000 ms (Ballew & Todorov, 2007, Study 2).
If competence judgments based on candidates' faces were highly accurate, then the fact that they influence voting behavior would not be troublesome. Unfortunately, research on accuracy of social attributions based on images of faces (Todorov, Olivola, Dotsch, & Mende-Siedlecki, 2015) suggests that such inferences are rather inaccurate and that people are not aware of their degree of accuracy. In other words, individuals "may be better off ignoring subtle facial cues" and focusing on "other sources of information when they are presented with facial cues" (Todorov et al., 2015, p. 533). In the political domain, the accuracy of inferences based on faces has been examined by Olivola and Todorov (2010, Study 2), who asked participants to guess the political affiliation (Democrat or Republican) of US candidates both when participants had to make the inference only on the base of candidates' faces and when they were informed about the composition of the pool of candidates' photos (base rate information). Although participants performed significantly better than chance when they did not know the base rate, they were not able to integrate information from base rates and from photos properly; they over weighted appearances and underused information about base rate, thus confirming a well-known effect (base rate fallacy or neglect) in social attribution research (e.g. Borgida & Brekke, 1981;Borgida & Nisbett, 1977;Pennycook & Thompson, 2016). In another study, Olivola, Sussman, Tsetsos, Kang, and Todorov (2012, Study 1) examined the predictive impact of inferences about political affiliation based on facial features. They found that "candidates running in right-leaning states benefited from looking more stereotypically Republican than their rivals. In contrast, there was no relationship between political facial stereotypes and electoral success in liberal states" (pp. 607-608). The same pattern of results was replicated in the second experimental study: whilst right-leaning participants favored Republican-looking candidates, leftleaning participants did not show any political facial bias.
One critical question that has been left underexplored in previous studies is whether presenting more specific information alongside candidates' faces influences competence judgments and their utility as a predictor of electoral outcomes. This appears particularly important given that in real-life elections, voters are exposed to additional relevant information alongside candidates' faces. Indeed, the abundance of information surrounding candidates often competes with, and thus may reduce the effect of automatic inferences based on candidates' faces (Hall et al., 2009). To our knowledge, there is only one study that has tackled this, by showing participants video clips of a political debate with and without sound (Benjamin & Shapiro, 2009), and asking them which candidate they thought would win the election. Results of this study, beyond providing further supporting evidence of the predictive power of inferences of personal attributes based on "thin-slices", show that participants exposed to video clips with sound (in comparison to silent video clips) were better at guessing political positions of the candidates but less able to guess election outcomes.
One of the first pieces of information citizens received during electoral campaigns is the candidates' political affiliations. The relevance of this information is controversial as some political scientists believe that voters have become less loyal to political parties, a phenomenon which has been termed the "decline of the parties" (Wilson & DiIulio, 1995), whilst others (e.g. Bartels, 2000) have pointed out that parties still have a significant impact on voting behavior, and that party identity can be considered a "brand" that influences evaluations of candidates (Hoegg & Lewis, 2011). Another piece of information that is frequently associated with a candidate's face is his or her main campaign promises, that is, the main goals the candidate intends to pursue. These promises are frequently cited during electoral campaigns and are printed on campaign signs alongside the candidate's face. Information about political affiliation and campaign promises may appear weakly related to a candidate's competence in comparison with information about incumbency, previous experience in office, endorsements by third parties, the candidate's origins, or even socio-demographic information such as education and occupation; however, our aim was not to assess whether participants (and actual voters) would use relevant information to arrive at more accurate competence judgments. Our aim was simply to determine whether or not stimuli similar to actual political signs (i.e. consisting of candidates' faces plus political party or campaign promises) would induce inferences that differ from those based solely on images of candidates' faces. In other words, our aim was to assess what extent competence inferences based on faces might be influenced by verbal information about political categories that-for at least some citizens-functions as a group label (e.g. we (in-group) are leftwing, they (out-group) are rightwing). Olivola et al. (2012) demonstrated a political facial bias amongst right-leaning electors, which can be seen as another example, in the political domain, of classical in-group favoritism (Tajfel & Turner, 1979;Turner, 1985; more recently Eriksson & Funcke, 2015); however our aim was to assess whether brief verbal information-about political affiliation or campaign promises-positioned below an image of a candidate's face would influence left-and right-leaning participants' competence judgments. If participants' inferences were affected by in-group favoritism, competence judgments would be less predictive of actual election results. These results would suggest that in actual elections, when voters are frequently informed about-at the very least-candidates' political affiliations and main campaign promises, the impact of candidates' faces is reduced. If, however, information about political affiliation did not influence participants' competence judgments, this would support the suggestion that political parties are in decline (Wilson & DiIulio, 1995), showing that even individuals who declare a broad political orientation do not trust in-group members more than out-group ones. It is worth noting that this hypothesis is supported by a recent survey showing that only 5% of Italians trust political parties, although 70% trusts the police, the most trusted institution (Demos survey for the La Repubblica newspaper, 2013). In this paper, we present three studies. In the first study, we tested our first hypothesis, based on Todorov et al.'s results (e.g. 2005) that in the case of people from Central Italy competence judgments based solely on seeing political faces can predict the election outcomes in northeast Italy. The aim of this preliminary study was to provide further evidence for the external validity of previous results.
The aim of the Study 2a was to investigate whether competences judgments are influenced by the condition of fit or misfit between candidates' political affiliation and the participant's political preference. Based on the Italian political situation at the time of data collection, we used two political affiliation labels, "central left coalition" and "central right coalition", both including at least two parties.
Then, the aim of the Study 2b was to test the effect of another piece of information very often added to the candidate's photo on campaign signs, namely campaign promises.
Finally, the aim of Study 2c, was to compare the competence inferences made in Study 1 (faceonly condition) with those of Study 2a (face plus political label) and Study 2b (faces plus campaign promises).
If participants' inferences were influenced by information about candidates' political affiliations and campaign promises, then competence judgments based solely on candidates' faces should differ from those based on candidates' faces plus information about political affiliation and from those based on candidates' faces plus information about campaign promises with respect to ability to predict election outcome. If, on the contrary, competence judgments based on short exposure to candidates' faces are made very quickly, they should be unaffected by further information requiring more cognitive elaboration such as political affiliation or campaign promises. In this case, competence judgments based on different information would all be equally predictive of actual electoral outcomes.

Participants
Seventy undergraduate students from Sapienza, University of Rome (M age = 24.20, SD age = 3.34) participated in the study on a voluntary basis.
The study was approved by the Ethics Committee of Psychology Research of Sapienza University (n°22-CED01).

Stimuli and procedures
To further test the external validity of previous results (e.g. USA: Sussman et al., 2013;Hall et al., 2009;Italy: Castelli, Carraro, Ghitti, & Pastore, 2009), we selected 16 pairs of politicians who competed in the elections of May 2011 in 16 cities in northeast Italy. All candidates were standing in medium-sized towns (between 15,000 and 21,000 inhabitants) in northeast Italy and were men with the same ethnic background who belonged to Civic Lists. We excluded very small towns as we assumed that in such cases, the voters were likely to have first-hand knowledge of local candidates. We also excluded major cities because the candidates' faces might have appeared on National TV programs or newspapers and could thus be familiar to our participants in Rome.
All 16 pairs of images of political candidates were presented on a computer screen, and participants indicated which of each pair they perceived as more competent. The positions of the candidates were counterbalanced across participants (see Figure 1).
All pictures were standardized: head shots, size 104 × 147 pixels, on a gray background. The pictures were 6 cm distant. The experiment was run using SuperLab software which randomized the location of presentation of the winning and losing candidates (right or left-hand side of the screen).
For each pair of politicians, participants were instructed to select the most competent candidate by pressing the "A" to select the candidate presented on the left or "L" key to select the candidate on the right. Before they left, the laboratory participants were asked if they had recognized any of the politicians during the experiment. None of the participants recognized any of the candidates and they were unaware that some had already been elected.

Results
Data were analyzed following the procedure described by Na, Kim, Oh, Choi, and O'Toole (2015). We counted the number of times participants chose the election winner as the most competent candidate of the pair and calculated a correct choice score (range: 0-16) for each participants and the percentage of successful candidates rated as more competent.
In order to assess whether judgments by people from central Italy based solely on candidates' faces predict election outcomes in northeast Italy, we correlated perceived competence judgments with the actual election outcome (dichotomous variable: 1 for winners; 0 for losers).
Perceived competence judgments were positively and significantly correlated with the actual results [r = .56, p = .001].

Discussion
In our first study, we replicated the Todorov's results in Italy such as in Castelli et al. (2009). The judgments about the competence of politicians from northeast Italy made by students in central Italy were positively correlated with actual election results.

Studies 2a, 2b and 2c
The general aim of this second set of studies was to assess whether judgments of competence were influenced by verbal information presented alongside candidate faces and whether such an effect would alter the predictive power of these judgments with respect to election outcome (percentage of votes). In Study 2a, the additional information was information about candidates' political affiliation, whereas in Study 2b, it was about campaign promises. Finally, in Study 2c, we compared the predictive power of all the different types of stimuli.

Study 2a
In this study, we not only assessed the impact of added information about candidates' political affiliation, but also whether judgments about competence were influenced by the fit between the candidates' political affiliations and the participant's political preference. For this purpose, we asked participants to indicate their political orientation.
As discussed in the introduction, an extended tradition of research on intergroup perceptions (Tajfel & Turner, 1979;Turner, 1985; more recently Eriksson & Funcke, 2015) strongly implies that people are prone to in-group favoritism when making inferences based on stable positive features of in-group and out-group members; in other words they tend to perceive in-group members in more positive terms than out-group members. In the context of this study this would manifest as left-leaning participants rating left-wing candidates as more competent than right-wing candidates and one would expect this bias to reduce the predictive power of their judgments. One could also hypothesize, based on the notion of the decline of parties, that participants do not really identify with the political coalition they favor and do not perceive candidates as belonging to their in-group or out-group. If this were the case, one would not expect information about political affiliation to bias competence inferences.

Participants
The participants were 76 undergraduate students from Sapienza, University of Rome (M age = 25.15 years, SD age = 2.14).

Stimuli and procedures
Participants viewed 16 pairs of candidates, one candidate in each pair was tagged as belonging to the left-wing coalition and the other tagged as belonging to the right-wing coalition. All candidates' faces were presented in two conditions: once tagged with "central left coalition" and once tagged with "central right coalition"; half the participants viewed each candidate with one label and half with the other label (see Figure 2).
All 16 pairs of images of political candidates were presented on a computer screen and participants were asked to indicate which member of each pair they perceived as more competent.
In all conditions, the positions (left and right of the screen) of candidates' photographs were counterbalanced across participants.
All pictures were standardized as in Study 1. Participants were also asked to indicate their own political orientation ("centre right"; "centre left"; "other" [which had to be specified]). Before they left, the laboratory participants were asked if they had recognized any of the politicians during the experiment. None of the participants recognized any of the candidates and the participants were unaware that some had already been elected.

Results
The data were analyzed at the participant level. We used a 2 (candidate's political affiliation: centre right vs. central-left) × 2 (participant's political preference: centre right vs. centre left) ANOVA with percentage of cases in which the winner was selected as most competent as the dependent variable to assess whether participants' competence judgments were influenced by the fit between their political preference and the political affiliation of the candidates. Forty-two participants self-defined as left-wing, 26 as right-wing and 8 did not specify their political preferences and were excluded More interesting, however, is that the interaction also was not significant (F = .27, p = .60, 2 p = .00), which suggests that perceived competence judgments were not biased by the fit between the candidate's political affiliation and the participant's political preference. On average, participants selected the winner as the most competent in 9.04 out of 16 choices (56.52%), which replicates previous findings (e.g. Todorov et al., 2005) on the relationship between the perceived competence judgments and electoral success. This result supports the second hypothesis that is the equivalence of political signs with information concerning political coalition added to candidate's face to political signs with only candidate's face as far as the predictive power of competence inferences are concerned.

Study 2b
The aim of the this study was to test the effect of another piece of information very often added to the candidate's photo on campaign signs, namely campaign promises. The campaign promises used in the study were selected on the basis of a pilot investigation in which participants, who were drawn from the same population as those for the experimental study, were requested to indicate the extent to which they associated various campaign promises with the political left or right. We chose "greater equality" as the left-wing promise and "lower taxes" (see Figure 3) as the right-wing promise, as these were judged, respectively, as the most left-and right-wing (the other choices were "power to the people"; "faith and family"; "greater social justice"; "lower crime").

Participants
The participants were 77 undergraduate students from Sapienza, University of Rome (M age = 23.17, SD age = 3.73), who were exposed to the stimuli: face plus campaign promise "greater equality" (n = 38) or "lower taxes" (n = 39).

Stimuli and procedures
Participant viewed 16 pairs of candidates, in which one candidate was tagged with the left-wing campaign promise ("greater equality") and the with the right-wing campaign promise ("lower taxes"). In all other respects, the procedure was as in Study 2a.

Results
As before, the data were analyzed at participant level. A one-way ANOVA with percentage of cases in which winner was chosen as most competent as the dependent variable showed that campaign promise did not affect competence inferences (F(1,75) = .121, p < .729, 2 p = .002). The mean number of cases in which the winner was selected as most competent was 9.18 when participants were exposed to the left-wing promise and 9.0 when they were exposed to the right-wing promise.

Study 2c
The final study consisted of a comparative analysis of the "accuracy" of competence inferences (i.e. the percentage of cases in which the winner was judged the more competent candidate) across the five conditions.
To summarize, 70 participants were exposed to the face-only (condition 1), 38 to the face plus "centre left" political label (condition 2), 38 to the face plus "centre right" political label (condition 3), 38 to the face plus "greater equality" promise (condition 4) and 39 to the face plus "lower taxes" promise (condition 5).
All pictures were standardized as in Study 1 and before they left the laboratory participants were asked if they had recognized any of the politicians during the experiment. None of the participants recognized any of the candidates and they were unaware that some had already been elected.

Results
On the basis of the results of Studies 2a and 2b, we hypothesized that all the additional information conditions would produce similarly successful predictions of electoral outcome to the face-only condition. This means to adopt the null hypothesis as the research hypothesis, or to test the equivalence of the different treatments (stimuli conditions). We are aware that this analytical procedure has been a subject of debate (e.g. Rouder, Speckman, Sun, Morey, & Iverson, 2009), but we think that in this instance, the reliability of the dependent measure (used in several previous studies) and the use of the different levels of analysis (participant and candidate) improves its reliability.

Analysis at participant level
We compared the predictive power of competence judgments based on candidates' faces across the five different conditions using a one-way ANOVA with the condition as the independent variable and the number of times in which each participant had chosen the winner as the most competent as the dependent variable. If the additional information presented alongside the faces influenced our participants' judgments of competence one would expect to see an effect of condition. On the contrary, if further information concerning political affiliation and campaign promises was not taken into account by participants, there would be no significant difference between conditions. The results confirmed our second hypothesis: there was no effect of condition (F(4, 220) = 1.22; p = .30, 2 p = .02), indicating that in none of the four conditions in which additional relevant information was presented alongside the candidates' faces did judgments of competence differ from the face-only condition (condition 1). On average, participants selected the winner as the most competent in 9.45 out of 16 choices (59.08%), once again confirming previous results relating to the predictive power of citizens' perceived competence judgments with respect to observed electoral outcomes. We also ran an ANOVA in which the condition variable was dichotomized (face-only vs. face plus additional information conditions). In this analysis, the effect of condition just failed to reach significance (F(1, 223) = 3.614, p < .06, 2 p = .016), confirming again their predictive equivalence.

Analysis at candidate level
In this analysis, we followed the procedure described by Willis and Todorov (2006) and used the candidate as the unit of analysis; in other words, we counted the number of participants that had chosen each of the 32 candidates as the most competent in a pair, and then divided this number by the number of participants in that condition, thus obtaining five competence ratings for each candidate. As suggested by Rosenthal (2005), we used Cronbach's α coefficient to estimate the reliability of the five sets of competence ratings: face-only condition (condition1): Cronbach's α = .94; face plus "centre left" label (condition 2): α = .80; face plus "centre right" label (condition 3): α = .82; face plus "greater equality" promise (condition 4): α = .92; face plus "lower taxes" promise (condition 5): α = .89. We had a data matrix consisting of five competence ratings for each of the candidates (derived from participants' selection of the most competent of the pair in the five different conditions) and the percentage of votes that each candidate had actually obtained in the election. We computed an average "face-plus" competence rating based on the four conditions with face plus relevant information and correlated this and the "face-only" competence rating with the percentage of votes obtained by each candidate in the election; both correlations were positive and significant. A comparison of the two coefficients (Fisher's z = −.51, p < . 61) confirmed the participant level analysis results, showing that face-only competence ratings (r(31) = .63, p < .01) were as predictive of election outcomes as "face-plus" competence ratings (r(31) = .54, p < .01).

General discussion
The results of these studies provide further evidence of the predictive power of competence inferences based on brief presentations of electoral candidates' faces (Study 1) but they also provide some preliminary evidence for: (a) the absence of an in-group bias related to the fit between the candidate's political affiliation and the participant's political orientation (Study 2a) and (b) the absence of any interference from verbal information with competence inferences based on faces (Studies 2a, 2b, and 2c).
With regard to the predictive power of competence inferences based on brief exposures to political candidates' faces, the results of Study 1 are consistent with previous studies conducted in the USA (Hall et al., 2009;Lenz & Lawson, 2007;Sussman et al., 2013;Todorov et al., 2005) and in Italy (Castelli et al., 2009). They showed that the competence judgments expressed by young people in Rome based on presentations of candidates' faces were strongly related to actual election outcomes in medium-sized towns in northeast Italy.
Study 2a provided preliminary evidence for the absence of an in-group bias induced by providing information about the condition of fit or misfit between candidates' political affiliation and the participant's political orientation. Although these results are consistent with current Italian political debate and actual election outcomes characterized by strong growth, in support for a new movement that identifies itself as "non-partisan" (the "Five stars movement"), they are based on only one experimental study in which participants (potential voters) assessed a specific combination of information, namely the candidate's face and political affiliation.
The absence of an in-group bias (i.e. candidates who share the rater's ideological leaning are not rated more competent) seems to conflict with previous research, for example Jahoda (1954) found that "in judging other people, subjects tend to associate favorable personal characteristics with political attitudes of which they themselves approve" (p. 330) and Bull and Hawkes (1982) reported similar findings.
The present results can be explained in terms of the "halo effect" (Thorndike, 1920); similar findings were reported by Herrmann and Shikano (2015) who noted that participants were more likely to attribute their own political views to candidates they perceived as attractive and competent. Similarly, Riggle, Ottati, Wyer, Kuklinski, and Schwarz (1992) showed that in the absence of other information voters' judgments of politicians were affected by physical attractiveness; however these effects disappeared in presence of other specific indicators of candidates' political orientation.
One limitation of our experiments is that our stimuli consisted of candidates' faces with the political affiliation labels beneath the faces, which may have increased the impact of faces by making them the first information perceived by participants. Further research should be done to determine whether variables relating to the provision of information about political affiliation such as position, different size of it, or even the introduction of traditional party logos influences the extent to which such information affects participants' (and by implication voters') inferences. A further limitation of Study 2a is the young age for our participants: the absence of in-group bias in this Italian sample may be typical of the young population, but not of older people. Finally, it is also possible that forcing participants to choose between two candidates with opposite political affiliations alerted them to the research goal and prompted them to correct their bias. All these possibilities deserve further investigation.
Finally, the studies presented here provide the first empirical evidence that the predictive power of rapid inferences about competence (based only on faces) is not substantially modified by presentation of relevant written information along with candidates' faces. Neither information about political affiliation nor campaign promises appears to alter the predictive power of competence judgments. Although this study focused on competence judgments rather than guesses about election outcomes, our results appear very similar to those of Benjamin and Shapiro (2009): in both cases, in fact, added verbal information that should be relevant to the election outcome did not alter the predictive power of participants' judgments.
Although the verbal information added to the candidates' faces was very similar to that presented in actual campaign signs, the graphic format was different and not professional, so our finding that such information is largely irrelevant to potential voters' judgments should be replicated in studies based on stimuli consisting of a different combination of visual and verbal information.
The fact that the written information provided in our studies did not substantially modify the predictive power of competence judgments suggests that the people who actually voted in the elections concerned may also have neglected relevant information such as political affiliation and candidates' main political goals when casting their vote. This hypothesis should be tested empirically by interviewing actual voters.
While the rational voter model assumes that voters carry out an extended search for information relevant to choosing political representatives and process it in depth, our results suggest that contemporary voters might not have the minimal epistemic motivation required to process even very short labels stating political affiliation or campaign promises.
Further research should be conducted with samples more varied in terms of age and political orientation, using different stimuli and should include monitoring of visual observation (eye tracking) and physiological indices of emotional activation. Such research would improve our understanding of the processes underlying the predictive power of competence ratings derived from facial impressions. Does the power of such ratings derive from an attentional preference for faces relative to verbal stimuli? Do faces elicit greater affective reactions? These issues could be fruitfully probed in future research.

Funding
The research was founded by Sapienza University, 2014 [grant number C26A149HHJ].