You are on page 1of 11

POLITICAL IDEOLOGY AND CONSUMPTION

The Face of Political Beliefs: Why Gender Matters


for Electability
AHREUM MAENG AND PANKAJ AGGARWAL

AB STR ACT The facial width-to-height ratio (f WHR) of a person is associated with dominance and leadership. Our
research examines the extent to which a perceiver’s political orientation biases their judgment of a political candidate’s
electability based on the candidate’s facial characteristics, and stereotypes associated with the gender of the candidate.
Four studies suggest that although f WHR is positively correlated with dominance evaluations of male faces, the same
attribution is less likely to be made for female faces. Furthermore, political conservatives show stronger bias than lib-
erals against female faces and are also less likely to elect female candidates associated with lower dominance. Finally,
although liberals show greater intentions to vote for females than for males, high f WHR has little effect on voters’
perceptions of females’ electability regardless of their political orientation.

F
acial images have an enormous impact on electoral signals two fundamental dimensions of human personality:
decisions. With simple photographs, voters can pre- valence and dominance (Carré, McCormick, and Mondloch
dict election winners and losers (Antonakis and Dalgas 2009). Faces with high fWHR are seen as dominant and ag-
2009) and tell political party affiliations (Rule and Ambady gressive but are also less likable (Stirrat and Perrett 2010).
2010). Recent studies further show that male and female Moreover, individuals with high f WHR are also perceived to
politicians’ facial cues—the degree of facial masculinity and be more suitable for leadership positions and indeed tend to
femininity— influence voters’ decisions differently for Re- achieve higher social rank. For instance, compared to the av-
publicans and Democrats (Carpinella and Johnson 2013a, erage American, US Presidents have higher fWHR (Lewis,
2013b; Carpinella et al. 2016). For example, visual informa- Lefevre, and Bates 2012). Similarly, CEOs (Alrajih and Ward
tion affects candidate evaluations more if it is incongruent 2014) and even Catholic Popes have higher f WHR (Hahn
with gender stereotypes (i.e., masculine female faces, femi- et al. 2017) than the average person. Thus, high fWHR is as-
nine male faces) than if it is congruent (Bauer and Carpinella sociated with higher rank in social hierarchy across organi-
2018). However, less is known about whether political ori- zations, countries, and value systems (Hahn et al. 2017). Fur-
entation of the observers also affects electability judgments thermore, CEOs with high fWHR deliver high ROI for their
of male and female candidates based on other facial cues firms (Wong, Ormiston, and Haselhuhn 2011), and sports-
that signal dominance. The purpose of the current research men with high f WHR perform better (Tsujimura and Ba-
is to fill this gap in the literature. nissy 2013). Thus, high fWHR not only is associated with
People routinely draw information about a person’s race, perceptions of higher rank and status (Maeng and Aggarwal
gender, age, and personality from their facial features (Oos- 2018) but also is linked to greater achievement in social sys-
terhof and Todorov 2008). An ecologically meaningful facial tems. Unfortunately, these studies are mostly correlational,
metric—the face width-to-height ratio (fWHR: bizygomatic typically studying existing leaders (majority of whom are
width divided by upperface height)—is a reliable cue that males), and treating female leaders as noise. Furthermore,

Ahreum Maeng (corresponding author: amaeng@ku.edu) is associate professor of marketing, School of Business, University of Kansas, 1654 Naismith Drive,
Lawrence, KS 66045, USA. Pankaj Aggarwal (pankaj.aggarwal@utoronto.ca) is professor of marketing, University of Toronto, 1265 Military Trail, Scarbor-
ough, Ontario M1C1A4, Canada. Financial support given to the second author from the Social Sciences and Humanities Research Council of Canada is grate-
fully acknowledged.

Issue Editors: Rashmi Adaval and Robert S. Wyer Jr.

Published online May 2, 2022.


Journal of the Association for Consumer Research, volume 7, number 3, July 2022. © 2022 Association for Consumer Research. All rights reserved. Published
by The University of Chicago Press for the Association for Consumer Research. https://doi.org/10.1086/719579
Volume 7 Number 3 2022 361

no research to our knowledge has examined judgments and Karau 2002). These findings suggest that the links be-
about the relationship of fWHR to the electability of polit- tween fWHR, dominance, and leadership might be applied
ical candidates, male or female. differently to females than to their male counterparts.
The findings are mixed with respect to the effect of
fWHR on the dominance associated with male versus female Influence of Stereotypes on Face Perception
faces. Some researchers have found that fWHR of both men Decades of research has documented the various down-
and women is positively correlated with perceptions of dom- stream consequences of social categorization. Historically,
inance (e.g., Geniole et al. 2015). Others have concluded it has been assumed that perceptual cues activate a categor-
that female fWHR is not linked to dominance (e.g., Mileva ical representation of gender stereotypes and that the goals
et al. 2014). As noted earlier, although individuals who have associated with this representation influence evaluative
attained high social rank, such as Presidents and CEOs typ- judgments (Ito and Urland 2003; Fiske, Cuddy, and Glick
ically have high f WHR (Lewis et al. 2012; Alrajih and Ward 2007). However, more recent evidence suggests that social
2014), those studied have been mostly males. Thus, the re- perceptions are not only sensitive to bottom-up facial cues
lationship between female fWHR and the attainment of so- but are also dynamically shaped by context and higher-order
cial rank is still unclear. social cognition in a top-down manner. Thus, a perceiver’s
In this research, we examine the effect of f WHR on judg- pre-existing knowledge and stereotypes can influence and
ments of leadership appropriateness based on perceived systematically bias social perceptions based on facial fea-
dominance of male and female faces. We find that although tures (Freeman and Johnson 2016).
high f WHR is associated with greater dominance and lead- Of the many top-down factors that influence face percep-
ership, stereotypes associated with the female social cate- tion, the impact of stereotypes has received the strongest
gory can inhibit these perceptions. We identify the mode- support. Stereotypes are semantic associations activated
rating role of voters’ political orientation on this effect. by social categories that automatically trigger implicit ex-
We surmise that political conservatives pursue hierarchy- pectations during face perception and consequently can bias
enhancing goals and hold stronger gender stereotypes, perceptions of visual facial information. This bias can occur
thereby exhibiting a stronger muting effect of leadership as- at an early stage of visual representation in the brain. Thus,
sociations for female faces. On the other hand, individuals working-class attire causes a racially ambiguous face to be
with liberal political orientation pursue hierarchy-attenuating categorized as Black, whereas upper-class attire leads the
goals and show less of this tendency. face to be categorized as White (Johnson, Freeman, and
Pauker 2012). Because men are stereotyped as aggressive
F A C E R A T I O, D O M I N A N C E , and women are stereotyped as docile, male faces are per-
AND ST ER EOTYPES ceived as angry and female faces are perceived as happy
Humans evolved during the Pleistocene era as hunters and (Hess, Adams, and Kleck 2004).
gatherers. This evolutionary history also seems to influence
our current attribution of leadership to individuals who Gender Stereotypes and Effect of fWHR:
possess masculine and dominant characteristics (Van Vugt An Evolutionary Perspective
2006). Leadership in ancestral humans was often a physical From an evolutionary perspective, gender stereotypes re-
activity such as hunting and warfare (Ahuja and Vugt 2010), flect the different functions ascribed to men and women
and so selection was based on cues of health, strength, and (Geary 1998). It has long been thought that differences in
an imposing physique. Consequently, there may have been parental reactions to male and female children can influence
an evolutionary bias to defer “leadership” to individuals with the different traits associated with the two genders. Wom-
such characteristics. en’s role as a parent often requires nurturance, leading them
Notably, this leader prototype does not seem to apply to to be seen as having “communal traits” (warmth and caring).
females. When behaving in line with the traits of prototypic In contrast, men are more involved in food gathering, and
leaders, women often receive negative reactions (Koenig providing security for their offspring and are seen to pos-
et al. 2011). Furthermore, female leaders are often seen as sess more “agentic” traits (e.g., authority and control). Be-
less likely to be successful due to the perceived incongruence cause these gender characteristics are evolutionarily selected,
between prototypic leader traits related to aggression and the male-agentic, female-communal links are rigid, and are
stereotypically female traits related to nurturance (Eagly by and large universally accepted gender characteristics.
362 Face of Political Beliefs Maeng and Aggarwal

We assume that these gender stereotype-based expecta- that signals dominance but is inconsistent with the female
tions can bias face perceptions. In particular, the processing stereotype of communality, conservatives are more likely to
of male faces is guided by the stereotype of males as dom- suppress such a cue than liberals. This, ironically, can re-
inant. Perceptions of the two categories (male and high sult in the perception of high-ratio female faces as low on
fWHR) consequently become inextricably linked and facili- dominance, and this effect might be especially true of
tate one another. Thus, high-ratio male faces are perceived conservatives.
as dominant, and this perception, in turn, leads them to be
evaluated as more suitable for leadership positions. On the H2a: Both conservatives and liberals associate high
other hand, the processing of female faces is guided by the fWHR male faces with high dominance.
stereotype of women as communal. Because the two cate-
gories (female and high fWHR) do not share characteristics, H2b: Relative to liberals, conservatives associate high
they are likely to inhibit processing of each other. Thus, fWHR female faces with low dominance.
fWHR will have less or no influence on the perceived dom-
inance and the consequent leadership evaluations for fe- Although both conservatives and liberals see high-ratio
male faces. Formally stated, male faces as high in dominance, this perception may have
different effects on leadership judgments. Conservatives,
H1: Facial width-to-height ratio (f WHR) will be more who prefer to see the world in a hierarchical structure are
strongly associated with dominance for male faces likely to perceive dominant-looking (high-ratio) male faces
than for female faces. as more suitable for political office. In keeping with their
desire to maintain the status-quo, however, they are likely
Political Ideology as Motivated Social Cognition to ascribe lower electability to females regardless of their
People often adopt a specific political ideology in an effort to fWHR. In contrast, liberals will resist conferring judgments
satisfy various social–cognitive motives. In particular, con- of greater leadership upon the more dominant-looking male
servative ideologies satisfy motivations to preserve the hier- faces because doing so would undermine their hierarchy-
archical status quo and reject social change, whereas liberal attenuating ideological goals. Since ascribing leadership sta-
ideologies satisfy a need for social change and egalitarian tus to females is consistent with their motivational goal of
goals (Jost et al. 2003). Prior research also shows that con- changing the status quo, liberals will attribute electability
servatives tend to be attracted to political leaders who signal to both high- and low-fWHR female faces. Hence, we predict
a diverse range of dominance cues (Hayes 2005). When pro- the following:
vided with personality descriptions of political candidates,
for example, conservatives show preference for political can- H3a: Conservatives are more likely than liberals to
didates who are described as bullies (Laustsen and Petersen vote for males with high-f WHR faces.
2017), have more masculine facial features (Laustsen and
Petersen 2016), and have deeper voices (Laustsen, Petersen, H3b: Liberals are more likely than conservatives to
and Klofstad 2015). vote for females than for males regardless of their
Despite very different preferences for hierarchical rela- f WHR.
tionship, both conservative and liberals will link high
fWHR male faces to dominance because the male-agentic Current Research
stereotype is a widely accepted evolution-based concept. We tested these hypotheses in four studies. Study 1A uti-
In general, although both conservative and liberal individ- lized an archival database to test the basic relationship be-
uals will suppress dominance cues from the female face ra- tween fWHR and perceived dominance. Study 1B replicated
tio because of the female-communal link, the two groups the findings of study 1A in a laboratory setting and exam-
are likely to show different perceptual patterns. Conserva- ined the underlying mechanism for the effect. Study 2 de-
tives hold stronger female stereotypes than liberals do termined the downstream effect of fWHR on the electabil-
(Tausch and Hewstone 2010), and political conservatism ity of political candidates and examined how the perceivers’
is positively associated with hostile sexism—hostility to- political orientation affected these judgments. Study 3 eval-
ward women who challenge male power (Austin and Jack- uated the ecological validity of the effect using real human
son 2019). Thus, when exposed to a high-ratio female face face images. Across these studies, we found support for our
Volume 7 Number 3 2022 363

premise that the effect of fWHR on perceived dominance If the face-ratio differences are seen as irrelevant in judging
and voting intention is different for male versus female the dominance judgments of females, the time to make
faces, and that this effect is moderated by evaluators’ polit- these judgments should be the same or less than the time
ical orientation. to judge male faces. However, if face ratio is in fact used
to judge the dominance of females but is then “corrected”
ST UDY 1 A: T HE E FFECT OF f W HR due to the stereotypes associated with women, the time
ON PERCEIVED DOMINANCE to judge the dominance of female high-ratio faces should
Method be greater than the time to judge the dominance of male
Study 1A utilized an archival database developed at the Uni- high-ratio faces due to the additional processing required.
versity of Chicago for use in scientific research (Ma, Correll, Study 1B examined this possibility.
and Wittenbrink 2015). For this study, we included 183
standardized photographs of White male (n 5 93) and fe- ST U DY 1 B: T HE R OL E O F G E N D ER
male faces (n 5 90) from the Chicago face database. We ST E REOT Y P E I N RE SP ON S E T IM E
used only White faces to minimize race- and ethnicity-driven The purpose of study 1B was twofold. First, we wanted to
effects. The data set includes extensive information about replicate the differential effect of fWHR on dominance
each face, including actual measurement of each face’s judgments of male versus female faces in a laboratory set-
width-to-height ratio as well as independent judges’ domi- ting to establish causality. Second, we wanted to examine the
nance ratings of each face. About 72 raters rated each face. process whereby gender stereotypes affected the differential
Our primary interest was to examine if there was a relation- effect of fWHR for male and female faces. We expected that
ship between fWHR and dominance, and if the relationship the communal stereotype associated with female faces would
between these two variables was moderated by the gender of be assimilated to the dominance cue for high-ratio female
the face. faces, resulting in greater time taken to make dominance
judgments of high f WHR female faces than of high f WHR
Results male faces.
We first regressed f WHR of both male and female faces on
their dominance ratings. Overall, the face ratio positively Method
predicted dominance ratings (B 5 :86, p < :05). When Five hundred forty-one respondents from an online panel
dummy-coded gender variable was entered into the model, participated for monetary compensation (284 male, 432 White;
the face ratio still positively predicted dominance ratings mean age 42.3). They were randomly assigned to one of
( B 5 :9, p < :01), but there was a significant interaction be- four conditions in a 2 (ratio: high vs. low)  2 (gender:
tween gender and fWHR (B 5 1:33, p < :05). Simple slope male vs. female) between-subjects design with political ori-
analysis revealed male fWHR to be the only significant entation as a continuous factor. Eight male and eight fe-
predictor for dominance rating (B 5 1:56, p < :001). The male faces were pooled from the Chicago face database
slope of female fWHR was not significant (B 5 :23, p 5 (Ma et al. 2015) as experimental stimuli. Four male and
:59; see fig. S1; figs. S1–S4 are available online). four female faces that were 1 SD below the mean f WHR
(male 5 1:85, female 5 1:84) were selected as low-ratio
Discussion faces, whereas four male and four female faces that were
The difference in influence of f WHR for male versus female 1 SD above the mean f WHR of each group were selected
faces could occur in two ways. First, the face ratio that is as high-ratio faces. All participants rated the perceived
used to infer males’ dominance might be considered irrele- dominance and aggressiveness of the stimuli along scales
vant when inferring females’ dominance. Alternatively, al- from 1 (not at all) to 7 (extremely), and the time to make
though the face ratio is used to infer dominance in both these responses was assessed. Participants’ ratings of the
males and females, its effects might be offset by the influ- four faces were averaged. They then completed a five-item
ence of the gender stereotypes (whereby women are associ- measure of their own political orientation (Evans, Heath,
ated with communality), which then muted the effect of and Lalljee 1996) along scales from 1 (strongly agree) to
fWHR on dominance judgments for females. One way to 9 (strongly disagree) where higher scores indicated greater
disentangle these two potential processes might be to assess conservatism. They then provided their personal demo-
the time required by the assessors to make these judgments. graphic information.
364 Face of Political Beliefs Maeng and Aggarwal

Results ciated with women is not relevant for low fWHR faces, the
Dominance Judgments. A 2 (face ratio)  2 (face gender) time to respond to low-ratio male and low-ratio female faces
covariance analysis of dominance ratings was conducted us- did not differ.
ing participants’ gender and its interaction with face gender
as covariates. The face gender by ratio interaction was sig- ST U DY 2 : T H E E F F EC T O F f W HR
nificant (F(1; 535) 5 37:1, p < :001, h2 5 :06). Specifically, ON DOMINANCE RATINGS
high-ratio male faces were perceived as significantly more A N D V OT I N G I N T E N T I O N S
dominant (M 5 4:1, SD 5 :86) than low-ratio faces (M 5 Study 2 examined how f WHR-based dominance ratings
3:36, SD 5 :94, p < :001, h2 5 :08). However, judgments would translate into political leadership evaluations. Rather
of high-ratio female faces (M 5 3:87, SD 5 0:91) did not than mechanically applying their perceptions of dominance
differ from those of low-ratio female faces (M 5 4:04, to voting intentions, people are likely to consider whether a
SD 5 :76; p 5 :09). candidate can help or hinder their own motivational goals.
Next, to investigate the role of political ideology, partici- We therefore proposed that political ideology would moder-
pants’ political orientation score was entered into a regression ate the link between dominance perceptions and voting
model along with their own gender and its interaction terms intentions.
with face ratio, face gender, and political orientation score
as covariates. Results revealed a significant interaction of face Method
ratio and gender (B 5 :69, SE 5 :32, p 5 :02, h2 5 :01). In Four gender-neutral human faces were generated using
particular, for male faces, both conservatives (1 SD above Facegen modeler 3.5 in keeping with prior research (Maeng
the mean) and liberals (1 SD below the mean) rated high-ratio and Aggarwal 2018). Each face was then manipulated on
male faces as more dominant than low-ratio faces (B 5 :68, width-to-height ratio by changing the upper lip to midbrow
SE 5 :15, p < :001, h2 5 :04 and B 5 :81 SE 5 :15, metric to generate faces with two different ratios, yielding
p < :001, h2 5 :05, respectively). In contrast, face-ratio neg- four high-ratio faces and four low-ratio faces (e.g., high ra-
atively influenced conservatives’ dominance ratings of female tio: 2.14 vs. low ratio: 1.58; see fig. 1).
faces (B 5 2:35, SE 5 :15, p < :05, h2 5 :01). However, the Three hundred ninety-seven individuals from an online
effect of face-ratio on liberals’ dominance ratings was not sig- panel (202 male; 291 White; mean age 34.3) participated
nificant (B 5 :02, SE 5 :15, p 5 :90). for payment. (A sensitivity analysis using G-Power suggests
small to medium effect size (f 5 0:23).) They were ran-
Response Time. A similar covariance analysis of response domly assigned to one of four conditions in a 2 (gender:
times (in seconds), showed a significant interaction of face male vs. female)  2 (ratio: high vs. low) between-subjects
gender and face ratio (F(1; 535) 5 4:66, p < :05, h2 5 :009). design, with political orientation used as a continuous fac-
Specifically, participants took longer to judge high-ratio fe- tor. The same set of four faces (all high-ratio or all low-
male faces (M 5 4:49, SD 5 4:46) than to judge high-ratio ratio) was presented as either male or female with the four
male faces (M 5 3:48, SD 5 1:82, p < :05, h2 5 :01). How- faces being treated as a within-subject factor. The faces
ever, the time to judge low-ratio female faces (M 5 3:82, were ostensibly potential candidates for a mayoral election
SD 5 2:58) and low-ratio male faces did not differ (M 5 in participants’ city. Participants were asked to assess each
4:18, SD 5 4:20; p 5 :46). Consistent with our expectations, candidate in this potential role.
however, political orientation did not significantly moderate All participants provided their voting intentions for each
these judgment times (B 5 2:04, SE 5 :28, p 5 :87). candidate using two items (can be a good leader, likely to
vote) along a scale from 1 (not at all) to 7 (extremely). Next,
Discussion participants responded to the two items assessing their
Study 1B clarified the reason why fWHR is associated with dominance ratings and the five-item political orientation
dominance judgments of male but not female faces. Al- scale, as in study 1B. Responses to each set of items were av-
though high fWHR signals dominance for both male and fe- eraged (in each case, a 5 :87).
male faces, the stereotype associated with females appar-
ently led final judgments for high-ratio female faces to be Pretest
corrected, resulting in higher response time due to this ad- To examine the gender neutrality of the computer-generated
ditional processing. However, because the stereotypes asso- faces, we conducted a separate pretest in which 256 participants
Volume 7 Number 3 2022 365

Figure 1. Study 2. Measuring points used for morphometric calculations and examples of high- and low-ratio faces. (A) Measure of f WHR:
a–b represents the maximum distance between the left and right facial boundaries (bizygomatic width); c–d represents the upper lip and
highest point of the eyelids (upper face height). The fWHR was calculated as width divided by height (a–b/c–d). (B) Example of high-ratio
gender-neutral face ( f WHR 5 2:07). (C) Example of low-ratio gender-neutral face (fWHR 5 1:55). The face images were created using
FaceGen Modeller 3.5 (Singular Inversions, www.facegen.com). Reproduced with permission.

were asked to indicate the gender of a number of faces along than low-ratio faces (M 5 3:41, SD 5 1:35, p < :001, h2 5
a scale from 1 (definitely male) to 7 (definitely female). We :06). However, when the same faces were presented as fe-
presented actual male and female faces (used in study 1B) male, there was no significant difference in the dominance
along with our target stimuli (gender neutral faces). The ratings of high-ratio (M 5 4:08, SD 5 1:52) and low-ratio
gender-neutral faces (M 5 4:30) were rated as less male faces (M 5 4:40, SD 5 1:33; p 5 :11).
than the male faces (M 5 1:21, p < :001, d 5 24:28) and Next, participants’ political orientation score was entered
less female than the female faces (M 5 6:79, p < :001, into a regression model using participants’ gender and its in-
d 5 3:64), confirming our assumption that these faces, con- teraction with face ratio, face gender, and political orienta-
sidered out of context, were neutral. tion as covariates. When the faces were presented as male,
high-ratio male faces were rated as more dominant than
Results low-ratio faces by both conservatives (1 SD above the mean:
Manipulation Check. All participants were asked to indi- B 5 1:14, SE 5 :29, p < :001, h2 5 :04) and liberals (1 SD
cate their perception of the gender of the presented (neu- below the mean: B 5 :76, SE 5 :28, p < :01, h2 5 :02); see
tral) faces. As expected, 170 of 193 (88%) participants as- figure 2. When the same faces were presented as females,
signed to the female condition indicated the gender of the however, fWHR had no effect on judgments by either con-
presented gender-neutral faces as female. Similarly, of the servatives (B 5 20:35, SE 5 :28, p 5 :21) or liberals (B 5
204 participants assigned to the male condition, 185 (91%) 2:25, SE 5 :30, p 5 :39). These results suggest that gender
indicated the identical gender-neutral faces to be male, stereotypes exert an influence at the perceptual level, inhib-
x2(1) 5 246:7, p < :001. Hence the gender manipulation was iting the effects of fWHR dominance cues in the case of fe-
successful. males and facilitating them in the case of males.

Dominance Judgments. A 2 (face ratio)  2 (face gender) Candidate’s Electability. An analogous covariance analysis
covariance analysis was performed on dominance judg- was performed on the index of voting intentions. The in-
ments using participants’ gender and its interaction with teraction of face ratio and face gender was significant
face gender as covariates. As expected, the interaction of (F(1; 391) 5 20:73, p < :001, h2 5 :05). The f WHR did
face gender and face ratio was significant (F(1; 391) 5 not influence voting intentions regardless of whether the
20:73, p < :001, h2 5 :05). Specifically, when gender- faces were presented as male (Mhigh 5 3:32, SD 5 1:36 vs.
neutral faces were presented as male, high-ratio faces were M low 5 3:39, SD 5 1:36, p 5 :72) or female (M high 5 3:66,
rated as significantly more dominant (M 5 4:37, SD 5 1:33) SD 5 1:40 vs. M low 5 3:72, SD 5 1:16, p 5 :68). However,
366 Face of Political Beliefs Maeng and Aggarwal

Figure 2. Study 2. (A) High fWHR male faces are judged as greater on dominance by both conservatives (political orientation measure
11 SD) and liberals (political orientation measure 21 SD). (B) High f WHR male faces were judged as more electable by conservatives
but not by liberals.

when political orientation was entered into the regression diated by dominance judgments only for those with conser-
model along with participants’ gender, the interaction of vative political ideology. Table S1 ( in the appendix, available
interaction of ratio, gender, and political orientation was online) shows the detailed estimates.
significant. (B 5 :36, SE 5 :14, p < :05, h2 5 :02). Figure 2
shows that when faces were presented as male, f WHR Discussion
was related positively to conservatives’ voting intentions The results of study 2 provide compelling evidence for top-
(B 5 :63, SE 5 :27, p < :05, h2 5 :01) but was related neg- down influence of gender stereotypes. When a set of faces
atively to liberals’ voting intentions (B 5 2:68, SE 5 :26, was presented as male, high-ratio faces were assigned
p < :01, h2 5 :02). When the exact same faces were pre- greater dominance than were low-ratio faces. When these
sented as female, however, f WHR did not influence the same faces were presented as female, however, perceptions
voting intentions of either conservatives (B 5 2:09, SE 5 of the dominance of high-ratio faces and low-ratio faces did
:26, p 5 :72) or liberals (B 5 2:08, SE 5 :27, p 5 :78; see not differ. This finding supports our hypothesis that the
fig. 2). However, regardless of the face ratio, as expected, lib- stereotypes associated with “female” suppress dominance
erals’ intentions to vote for female faces was greater than cues from the high-ratio faces.
their intentions to vote for male faces (B 5 2:54, p < :001), Our results also show that although conservatives judge
whereas this difference was not significant for conservatives male candidates with high-ratio faces to be more electable,
(B 5 2:15, p 5 :43). liberals judge high-ratio male faces to be less electable. Even
though liberals judge high-ratio male faces as highly domi-
Conditional Indirect Effect. For insight into the process nant, they do not link these judgments to electability. Be-
that underlies the effect of dominance on voting intentions, cause liberals are motivated to change the status quo to
we executed a moderated mediation model with face gender promote gender equality, they do not evaluate dominant-
and participants’ political orientation as moderators (Hayes looking male faces as more electable as doing so would up-
2012; PROCESS model 12; see fig. S2). There was a sig- hold the status quo and further strengthen the current so-
nificant index of moderated mediation (B 5 :3, SE 5 :08, cial system. In contrast, conservatives, who are motivated
p < :001), indicating that the effect of face ratio on voting to maintain the social hierarchies, do associate dominant-
intention is mediated by dominance ratings but the effect looking (high-ratio) male faces with greater electability,
of fWHR on dominance judgments is an interactive function thereby helping to keep current social system intact.
of the target’s gender and political ideology. In particular, Our results also show that low-ratio males were judged as
the effect of fWHR on voting intention is significantly me- less dominant than female faces. Why should this be true?
Volume 7 Number 3 2022 367

Prior research has suggested that people use different stan- male or four female faces with all four faces having either
dards when evaluating different social groups, especially high or low ratio (randomly chosen). Participants indicated
when rating stereotyped versus non-stereotyped targets their likelihood of voting for each of the candidates, donat-
(Biernat, Manis, and Nelson 1991). When evaluating a per- ing to the political campaign and volunteering for the polit-
son, the standard of judgment is usually the “average per- ical campaign along scales from 1 (not at all) to 7 (extremely);
son.” However, a different standard is evoked when a target a 5 :96. Each candidate’s dominance rating and political ori-
person exhibits an attribute that has its own standard asso- entation was assessed using measures employed in Study 2.
ciated with it. Thus, the standard of comparison that is evoked
when judging dominance might differ depending on the tar- Results
get’s gender because men are expected to be more dominant Manipulation Check. As indicated by the pretest con-
than females. These stereotypes could give rise to the use of ducted for study 2, male faces were perceived significantly
very different judgment standards for assessing men and more male (M 5 1:21) than female faces (M 5 6:79,
women, leading raters to “shift” the end anchors of their rat- p < :001, d 5 210:5).
ing scale according to the type of exemplar being evaluated.
Thus, what is considered “average” dominance for a man Dominance Judgments. A 2 (face ratio)  2 (face gender)
might be seen as “very” dominant for a woman. The evidence covariance analysis was performed on dominance judg-
that the average dominance rating of females was rated ments using participants’ gender and its interaction with
higher than that of males could be a result of this process. face gender as covariates. As expected, the gender by ratio
Although conservatives perceived high-ratio female faces interaction was significant (F(1; 723) 5 31:82, p < :001,
to be low in dominance, this difference was not significant. h2 5 :04). Specifically, high-ratio male faces were rated as
One possible reason for this might be that we employed gen- significantly more dominant (M 5 4:37, SD 5 1:08) than
der neutral faces in the study, presenting them as both male low-ratio male faces (M 5 3:72, SD 5 1:31, p < :001,
and female. Consequently, those faces did not have actual h2 5 :04). However, high-ratio female faces were rated as
sexually dimorphic facial characteristics (e.g., larger eyes, less dominant (M 5 4:01, SD 5 1:03) than low-ratio fe-
rounded facial shape, etc.) and hence may have attenuated male faces (M 5 4:29, SD 5 1:03, p < :01, h2 5 :008).
the expected effect. To examine this possibility, real male Participants’ political orientation score was then entered
versus real female human faces were used in study 3. into a regression model with face ratio, face gender, partici-
pants’ gender and its interaction terms as covariates. A sig-
ST UDY 3 : T H E EF FECT OF f W HR nificant three-way interaction effect of gender  ratio  po-
ON CANDIDA TE ELECTABILITY litical orientation (B 5 :16, SE 5 :08, p < :05, h2 5 :006)
WITH REAL HUMAN FACES on dominance (a 5 :86) was found. As figure 3 shows,
Study 3 evaluated the ecological validity of the effects high-ratio male faces were judged as more dominant than
found in study 2 by using real human face images instead low-ratio male faces by both conservatives (B 5 :94,
of computer-generated faces. SE 5 :17, p < :001, h2 5 :04) and liberals (B 5 :38, SE 5
:16, p < :05, h2 5 :0082), although conservatives demon-
Method strate such tendency more extremely than liberals do
Seven hundred twenty-nine respondents from an online (Bcons vs: lib 5 :55, SE 5 :23, p < :05). In contrast, face ratio
panel participated in this study for monetary compensation negatively influenced conservatives’ perception of the dom-
(388 male, 518 White; mean age 42.2; sensitivity analysis inance of female faces (B 5 2:34, SE 5 :16, p < :05, h2 5
using G-Power suggested small to medium effect size :006), whereas its effect on liberals’ perception was negative
(f 5 0:17)) and were randomly assigned to one of four con- but not significant (B 5 2:18, SE 5 :17, p 5 :29). These re-
ditions in a 2 (ratio: high vs. low)  2 (gender: male vs. fe- sults suggest that, when the actual faces with sexually dimor-
male) between-subjects design with political orientation phic facial characteristics rather than gender neutral faces
as an additional continuous measure. The face stimuli used are presented as female, the predicted effect of political ori-
in study 1B were used as experimental stimuli. The same entation becomes more pronounced.
cover story about the faces belonging to different candi-
dates in an election as used in study 2 was presented to Candidate’s Electability. An analogous covariance analysis
the participants. They were then presented with either four was performed on the three-item voting intention index. The
368 Face of Political Beliefs Maeng and Aggarwal

Figure 3. Study 3. (A) High f WHR male faces are judged as greater on dominance by both conservatives (political orientation measure
11 SD) and liberals (political orientation measure 21 SD). (B) Males with high f WHR faces are more likely to be elected by conservatives
but not by liberals.

interaction of ratio and gender was significant (F(1; 723) 5 ratio on voting intention is mediated by perceived domi-
32:82, p < :001, h2 5 :04). High-ratio male faces were judged nance with dependence on face gender and perceivers’ po-
as more electable (M 5 3:37, SD 5 1:53) than low-ratio male litical orientation. See table S1 in the appendix for the
faces (M 5 3:04, SD 5 1:65, p < :05, h2 5 :006). However, detailed estimates.
intentions to vote for high-ratio female faces (M 5 3:41,
SD 5 1:43) did not differ from that of low-ratio female faces GENERAL DISCUSSION
(M 5 3:67, SD 5 1:46, p 5 :08). Across four studies, our research demonstrates that the as-
A regression analysis was then performed on voting in- sociation of face ratio (fWHR) with perceptions of dominance
tention with face ratio, face gender, participants’ political differs for male versus female faces. A key contribution of our
orientation as predictors and participants’ gender and its in- research is the evidence for the top-down influence of gender
teraction terms as covariates. This analysis revealed a signif- stereotypes on dominance ratings. Because women are ste-
icant three-way interaction of gender, ratio and political ori- reotyped to be communal, it mutes the effect of high fWHR
entation (B 5 :21, SE 5 :10, p < :05, h2 5 :005). As shown female faces on their dominance ratings. The stereotype of
in figure 3, conservatives reported greater intention to vote men as aggressive, however, is consistent with the assess-
for high-ratio male faces than for low-ratio male faces ment that high fWHR male faces are high in dominance.
(B 5 :69, SE 5 :23, p < :001, h2 5 :01), but liberals did We further find that the effect of gender stereotypes on
not (B 5 2:03, SE 5 :22, p 5 :9). In contrast, female faces’ the dominance attributed to male versus female faces was ev-
fWHR did not influence voting intentions by either con- ident in judgments of their electability. Conservatives are
servatives (B 5 2:34, SE 5 :22, p 5 :12) or liberals (B 5 more likely to rate high fWHR male faces as more electable
2:17, SE 5 :23, p 5 :47). Regardless of the face ratio, how- than low fWHR male faces, whereas this same electability
ever, liberals’ intentions to vote for female faces was greater effect was not exhibited by liberals. In contrast, neither conser-
than their intentions to vote for male faces (B 5 2:44, vatives nor liberals exhibited any indication that high fWHR
p < :01), whereas this was not true of conservatives (B 5 affected females’ electability, although liberals, unlike con-
2:20, p 5 :20; see fig. 3). servatives, were generally more likely to attribute electability
to female faces than to male faces.
Conditional Indirect Effects. A moderated mediation
model (Hayes 2012; PROCESS model 12) revealed a signif- Theoretical Implications
icant index of moderated mediation effect (B 5 :84, SE 5 Previous research has shown that people make rapid person-
:08, p < :001; see fig. S3), indicating that the effect of face ality judgments based solely on other people’s appearance
Volume 7 Number 3 2022 369

and that this can predict their electoral success (Antonakis Alrajih, Shuaa, and Jamie Ward (2014), “Increased Facial Width-to-Height
Ratio and Perceived Dominance in the Faces of the UK’s Leading Busi-
and Dalgas 2009; Olivola and Todorov 2010). However,
ness Leaders,” British Journal of Psychology, 105 (2), 153–61.
these earlier studies relied primarily on subjective judgments Antonakis, John, and Olaf Dalgas (2009), “Predicting Elections: Child’s
of others’ personality based on their faces without determin- Play!” Science, 323 (5918), 1183.
ing the basis for these judgments. Our research eliminated Austin, Darren E. J., and Mervyn Jackson (2019), “Benevolent and Hostile
this ambiguity by highlighting the causal relationship be- Sexism Differentially Predicted by Facets of Right-Wing Authoritarian-
ism and Social Dominance Orientation,” Personality and Individual Dif-
tween face ratio, dominance ratings and judgments of lead- ferences, 139 (March), 34–38.
ership. It also examined differences between male and fe- Bauer, Nichole M., and Colleen Carpinella (2018), “Visual Information and
male faces driven by the associated gender stereotypes. Candidate Evaluations: The Influence of Feminine and Masculine Im-
Our research also contributes to the literature on motiva- ages on Support for Female Candidates,” Political Research Quarterly,
71 (2), 395–407.
tional influences and social perception. Prior studies show
Biernat, Monica, Melvin Manis, and Thomas E. Nelson (1991), “Stereo-
that motivational factors such as goals can bias perceptual types and Standards of Judgment,” Journal of Personality and Social
processes by enhancing or inhibiting the sensitivity to visual Psychology, 60 (4), 485–99.
cues that can help to achieve the activated goal. Assuming Carpinella, Colleen M., Eric Hehman, Jonathan B. Freeman, and Kerri L.
Johnson (2016), “The Gendered Face of Partisan Politics: Conse-
that political conservatism and liberalism reflect motivated
quences of Facial Sex Typicality for Vote Choice,” Political Communica-
social cognition (Jost et al. 2003), our research demonstrates tion, 33 (1), 21–38.
the central role of this motivation in face perceptions and ex- Carpinella, Colleen M., and Kerri L. Johnson (2013a), “Appearance-Based
pectations of political success. Finally, our findings might Politics: Sex-Typed Facial Cues Communicate Political Party Affilia-
help us better understand why women are still a minority tion,” Journal of Experimental Social Psychology, 49 (1), 156–60.
——— (2013b), “Politics of the Face: The Role of Sex-Typicality in Trait
in political leadership roles, and why the United States is still Assessments of Politicians,” Social Cognition, 31 (6), 770–79.
one of the few countries in the world that has not elected a Carré, Justin M., Cheryl M. McCormick, and Catherine J. Mondloch (2009),
woman as head of state. “Facial Structure Is a Reliable Cue of Aggressive Behavior,” Psychological
Science, 20 (10), 1194–98.
Future Research Chiao, Joan Y., Nicholas E. Bowman, and Harleen Gill (2008), “The Polit-
ical Gender Gap: Gender Bias in Facial Inferences That Predict Voting
Although we demonstrated the differential effect of fWHR Behavior,” PloS One, 3 (10), e3666.
on dominance judgments and electability of male versus fe- Eagly, Alice H., and Steven J. Karau (2002), “Role Congruity Theory of
male political candidates, our research is far from the final Prejudice toward Female Leaders,” Psychological Review, 109 (3), 573–98.
word on this topic. One question for future investigation Evans, Geoffrey, Anthony Heath, and Mansur Lalljee (1996), “Measuring
Left-Right and Libertarian-Authoritarian Values in the British Elector-
is what facial cues, if not fWHR, are linked to females’ elect-
ate,” British Journal of Sociology, 47 (1), 93–112.
ability. Some research suggests that masculinity of female Fiske, Susan T., Amy J. C. Cuddy, and Peter Glick (2007), “Universal Di-
faces are a cue to dominance perception (Quist et al. 2011) mensions of Social Cognition: Warmth and Competence,” Trends in
and leadership judgments (e.g., Rule and Ambady 2009), Cognitive Sciences, 11 (2), 77–83.
Freeman, Jonathan B., and Kerri L. Johnson (2016), “More Than Meets the
while other research shows that feminine-looking females
Eye: Split-Second Social Perception,” Trends in Cognitive Sciences, 20 (5),
(Hehman et al. 2014) as well as attractive-looking females 362–74.
(e.g., Chiao, Bowman and Gill 2008) are more likely to get Geary, David C. (1998), Male, Female: The Evolution of Human Sex Differences,
elected as leaders. Thus, it might be worthwhile to further in- Washington, DC: American Psychological Association.
vestigate whether perceptual congruence of female facial cues Geniole, Shawn N., Thomas F. Denson, Barnaby J. Dixson, Justin M. Carré,
and Cheryl M. McCormick (2015), “Evidence from Meta-Analyses of
with (a) the social stereotype of women as nurturing or the Facial Width-to-Height Ratio as an Evolved Cue of Threat,” PloS
(b) the traditional leader prototype (assertiveness and mas- One, 10 (7), e0132726.
culinity) leads to greater political leadership preferences. Hahn, Tim, Nils R. Winter, Christine Anderl, Karolien Notebaert, Alina Ma-
Another potentially rewarding avenue for future research rie Wuttke, Celina Chantal Clément, and Sabine Windmann (2017), “Fa-
cial Width-to-Height Ratio Differs by Social Rank across Organizations,
might be to explore if the electability perceptions of racial
Countries, and Value Systems,” PloS One, 12 (11), e0187957.
minority faces are also influenced by racial stereotypes, es- Hayes, Andrew F. (2012), PROCESS: A Versatile Computational Tool for Ob-
pecially with respect to people’s own political orientations. served Variable Mediation, Moderation, and Conditional Process Modeling,
2nd ed., New York: Guilford. https://www.afhayes.com/public/process
2012.pdf.
REF ERE NCES Hayes, Danny (2005), “Candidate Qualities through a Partisan Lens: A The-
Ahuja, Anjana, and Mark van Vugt (2010), Selected: Why Some People Lead, ory of Trait Ownership,” American Journal of Political Science, 49 (4),
Why Others Follow, and Why It Matters, London: Profile Books. 908–23.
370 Face of Political Beliefs Maeng and Aggarwal

Hehman, Eric, Colleen M. Carpinella, Kerri L. Johnson, Jordan B. Leitner, Maeng, Ahreum, and Pankaj Aggarwal (2018), “Facing Dominance: An-
and Jonathan B. Freeman (2014), “Early Processing of Gendered Facial thropomorphism and the Effect of Product Face Ratio on Consumer
Cues Predicts the Electoral Success of Female Politicians,” Social Psy- Preference,” Journal of Consumer Research, 44 (5), 1104–22.
chological and Personality Science, 5 (7), 815–24. Mileva, Viktoria R., Mary Louise Cowan, Kelly D. Cobey, K. K. Knowles, and
Hess, Ursula, Reginald B. Adams Jr., and Robert E. Kleck (2004), “Facial A. C. Little (2014), “In the Face of Dominance: Self-Perceived and
Appearance, Gender, and Emotion Expression,” Emotion, 4 (4), 378– Other-Perceived Dominance Are Positively Associated with Facial-
88. Width-to-Height Ratio in Men,” Personality and Individual Differences,
Ito, Tiffany A., and Geoffrey R. Urland (2003), “Race and Gender on the 69 (October), 115–18.
Brain: Electrocortical Measures of Attention to the Race and Gender Olivola, Christopher Y., and Alexander Todorov (2010), “Elected in
of Multiply Categorizable Individuals,” Journal of Personality and Social 100 Milliseconds: Appearance-Based Trait Inferences and Voting,”
Psychology, 85 (4), 616–26. Journal of Nonverbal Behavior, 34 (2), 83–110.
Johnson, Kerri L., Jonathan B. Freeman, and Kristin Pauker (2012), “Race Is Oosterhof, Nikolaas N., and Alexander Todorov (2008), “The Functional
Gendered: How Covarying Phenotypes and Stereotypes Bias Sex Cat- Basis of Face Evaluation,” Proceedings of the National Academy of Sci-
egorization,” Journal of Personality and Social Psychology, 102 (1), 116– ences, 105 (32), 11087–92.
31. Quist, Michelle C., Christopher D. Watkins, Finlay G. Smith, Lisa M. DeBruine,
Jost, John T., Jack Glaser, Arie W. Kruglanski, and Frank J. Sulloway and Benedict C. Jones (2011), “Facial Masculinity Is a Cue to Women’s
(2003), “Political Conservatism as Motivated Social Cognition,” Psycho- Dominance,” Personality and Individual Differences, 50 (7), 1089–93.
logical Bulletin, 129 (3), 339–75. Rule, Nicholas O., and Nalini Ambady (2009), “She’s Got the Look: Infer-
Koenig, Anne M., Alice H. Eagly, Abigail A. Mitchell, and Tiina Ristikari ences from Female Chief Executive Officers’ Faces Predict Their Suc-
(2011), “Are Leader Stereotypes Masculine? A Meta-Analysis of Three cess,” Sex Roles, 61 (9–10), 644–52.
Research Paradigms,” Psychological Bulletin, 137 (4), 616–42. ——— (2010), “Democrats and Republicans Can Be Differentiated from
Laustsen, Lasse, and Michael Bang Petersen (2016), “Winning Faces Vary Their Faces,” PloS One, 5 (1), e8733.
by Ideology: How Nonverbal Source Cues Influence Election and Com- Stirrat, Michael, and David I. Perrett (2010), “Valid Facial Cues to Co-
munication Success in Politics,” Political Communication, 33 (2), 188– operation and Trust: Male Facial Width and Trustworthiness,” Psycho-
211. logical Science, 21 (3), 349–54.
——— (2017), “Perceived Conflict and Leader Dominance: Individual and Tausch, Nicole, and Miles Hewstone (2010), “Intergroup Contact and Prej-
Contextual Factors behind Preferences for Dominant Leaders,” Politi- udice,” in Handbook of Prejudice, Stereotyping, and Discrimination, ed.
cal Psychology, 38 (6), 1083–101. J. F. Dovidio, M. Hewstone, P. Glick, and V. Esses, Thousand Oaks,
Laustsen, Lasse, Michael Bang Petersen, and Casey A. Klofstad (2015), CA: Sage, 544–60.
“Vote Choice, Ideology, and Social Dominance Orientation Influence Tsujimura, Hikaru, and Michael J. Banissy (2013), “Human Face Structure
Preferences for Lower Pitched Voices in Political Candidates,” Evolu- Correlates with Professional Baseball Performance: Insights from Pro-
tionary Psychology, 13 (3), 1–13. fessional Japanese Baseball Players,” Biology Letters, 9 (3), 20130140.
Lewis, Gary J., Carmen E. Lefevre, and Timothy C. Bates (2012), “Facial Van Vugt, Mark (2006), “Evolutionary Origins of Leadership and Follow-
Width-to-Height Ratio Predicts Achievement Drive in US Presidents,” ership,” Personality and Social Psychology Review, 10 (4), 354–71.
Personality and Individual Differences, 52 (7), 855–57. Wong, Elaine M., Margaret E. Ormiston, and Michael P. Haselhuhn (2011),
Ma, Debbie S., Joshua Correll, and Bernd Wittenbrink (2015), “The Chi- “A Face Only an Investor Could Love: CEOs’ Facial Structure Predicts
cago Face Database: A Free Stimulus Set of Faces and Norming Data,” Their Firms’ Financial Performance,” Psychological Science, 22 (12),
Behavior Research Methods, 47 (4), 1122–35. 1478–83.

You might also like