Sartorial Symbols of Social Class 1
Sartorial symbols of social class elicit class-consistent behavioral and physiological
responses: A dyadic approach
Michael W. Kraus
University of Illinois, Urbana-Champaign
Wendy Berry Mendes
University of California, San Francisco
Correspondence concerning this article should be addressed to Michael W. Kraus, Department of
Psychology, University of Illinois, Urbana-Champaign, 603 East Daniel Street, Champaign, IL,
61820. Email: firstname.lastname@example.org
In Press: Journal of Experimental Psychology: General
Sartorial Symbols of Social Class 2
Social rank in human and non-human animals is signaled by a variety of behaviors and
phenotypes. In this research, we examined whether a sartorial manipulation of social class would
engender class-consistent behavior and physiology during dyadic interactions. Male participants
donned clothing that signaled either upper-class (business-suit) or lower-class (sweats) rank prior
to engaging in a modified negotiation task with another participant unaware of the clothing
manipulation. Wearing upper-class, compared to lower-class, clothing induced dominance in
participants—measured in terms of negotiation profits and concessions, and testosterone levels.
Upper-class clothing also elicited increased vigilance in perceivers of these symbols: Relative to
lower-class symbols, perceiving upper-class symbols increased vagal withdrawal, reduced
perceptions of social power, and catalyzed physiological contagion such that perceivers’
sympathetic nervous system activation followed that of the upper -class target. Discussion
focused on the dyadic process of social class signaling within social interactions.
Sartorial Symbols of Social Class 3
Sartorial symbols of social class elicit class-consistent behavioral and physiological
responses: A dyadic approach
In mammalian social life, social rank is a powerful influence on a wide range of life
outcomes (Sapolsky, 2004). As such, the capacity to signal one’s rank in social encounters with
others is beneficial across social domains: Accurate communication of social rank helps
individuals to predict others’ behavior, find desirable mates, and avoid potential costly
aggressive encounters (Krebs, Davies, & Parr, 1993). In human societies, social rank is defined ,
at least in part, by one’s position in the social class hierarchy (Adler et al., 1994; Kraus, Tan, &
Tannenbaum, 2013; Marmot et al., 1991). Symbols of social class—expressed in a variety of
ways including in one’s manners, tastes, and preferences—communicate the social rank of
individuals during everyday interactions (Gillath, Bahns, Ge, & Crandall, 2014; Kraus &
Keltner, 2009). The present research examines the extent that sartorial symbols of social class
shape social interactions by changing the behavioral and physiological responses of both the
wearers and the perceivers of these symbols.
Social Class as Rank vis-à-vis Others
Researchers typically define social class as contrasting levels of material and social
resources that individuals possess, and measure the construct using indices of annual income,
educational attainment, and occupation status (Kraus & Stephens, 2012; Oaks & Rossi, 2003).
Together, these measures make up the objective material substance of social class.
Social class is more than simply one’s level of available material resources: Social class
environments, defined by varying levels of material and social resources, socialize and produce
unique conceptions of the self that are expressed in class-specific behavioral profiles (see Fiske
& Markus, 2012; Markus & Kitayama, 2010; Stephens, Markus, & Fryberg, 2012; Weininger &
Sartorial Symbols of Social Class 4
Lareau, 2009). When expressed during public life, these behaviors become signals of social class
that are used by perceivers to infer relative position in the social class hierarchy—by virtue of
comparing the class-based behaviors of the self to that of others. Thus, every time a person in
America engages in behavior consistent with relatively lower-class models of the self (e.g.,
attends a Nascar event) or with the relatively upper-class self (e.g., drinks a local craft beer), the
information carried in those behaviors can be used to accurately discern a target’s position in the
social class hierarchy. It is through this social class signaling process that individuals lea rn their
position on the social ladder of society at a chronic level, in comparison to society as a whole,
and specific to a particular situation or context (for a review, see Kraus et al., 2013).
One implication of this dynamic social class signaling process is that symbols of social
class allow individuals to dynamically judge others’ social class rank at levels above chance
accuracy. In support of this perspective, viewing 60s slices of a social interaction between two
University students led a sample of naïve observers to accurately predict the social class position
of the students on a ten rung ladder representing ascending levels of social class , based solely on
behavior during the interaction (Kraus & Keltner, 2009). In other work, a similar sample of naïve
observers was able to accurately discern the social class of participants after viewing a selection
of profile photographs from Facebook.com (Rheinschmidt, Kraus, & Keltner, 2014; Kraus et al.,
2013). In a study examining sartorial symbols, a sample of naïve judges were able to accurately
discern a person’s income and a host of other personality characteristics based only on a
standardized photograph of their shoes (Gillath et al., 2014).
A second implication of this class signaling process is that merely expressing symbols of
social class, regardless of the objective social class environment in which a person developed,
will shape an individual’s own experience of their rank in society vis-à-vis others. Data in
Sartorial Symbols of Social Class 5
support of this overarching hypothesis is limited but suggestive: For instance, nonverbal
behaviors that communicate socially valued success (e.g., dominance, victory) have been shown
to elicit behavioral changes on the part of targets expressing these behaviors. In one experiment,
people randomly placed in a dominant body position, expanding the chest and body, tended to be
more focused on gambling rewards—they were more likely to wager a sure $2 for the chance to
double their money—relative to individuals positioned submissively b y constricting the arms and
torso (Carney, Cuddy, & Yap, 2010). In another study using sartorial symbols, wearing a lab
coat—a form of clothing presumably associated with attention to detail and precision —induced
increased performance on attention-related tasks whereas just seeing a lab coat or wearing the
same coat labeled as a “painter’s coat” did not improve performance (Adam & Galinsky, 2013).
Based on the above conceptual analysis, we tested the overarching hypothesis that merely
wearing sartorial symbols associated with a particular social class will be enough to elicit
changes in the class-consistent behavior and physiology of both the targets who display these
symbols as well as the perceivers who view these symbols expressed by others.
Evidence suggesting that sartorial symbols of social class elicit changes in behavior and
physiology is theoretically important for two reasons: First, though mounting evidence suggests
that social class environments influence patterns of behavior, that evidence is co rrelational in
nature and subject to several alternative causal explanations (e.g., neighborhood effects, political
and economic trends). That an experimental manipulation of symbols of a person’s social class
can shift patterns of behavior establishes these symbols as a causal force in shaping a person’s
experience of their own social class rank in society relative to others (see Kraus, Côté, &
Keltner, 2010). Moreover, examining this process in a dyadic social interaction adds to our
understanding of the origins of class-based behavioral profiles observed in prior research: These
Sartorial Symbols of Social Class 6
profiles do not solely arise from socialization processes; rather, they also occur based on relative
comparisons of the symbols of social class one expresses in interactions and p erceives on others
(Kraus et al., 2013).
Social Class, Dominance, and Threat Vigilance
We make two theoretical predictions with respect to how symbols of social class shape
behavior and physiology of targets and perceivers: First, we predict that wearing upper-class
symbols will activate concepts in memory and behavioral scripts that are consistent with
expectations for how people wearing upper-class clothing feel and behave. Specifically, we
expect that wearing upper-class symbols will elicit behavior and physiology associated with
elevated dominance. Dominance includes a variety of social behaviors that involve tendencies to
value the self, or one’s in-group, over others (Sidanius, Pratto, & Bobo, 1994), engaging in self-
benefitting actions or actions to gain or maintain social influence (Goodwin, Operario, & Fiske,
1998), and in men, elevated levels of testosterone (Mazur & Booth, 1998). Dominance elicits
similar influences on social behavior, to other rank-related constructs like social power
(Goodwin et al., 1998; Keltner, Gruenfeld, & Anderson, 2003), but is a theoretically distinct
construct because, unlike social power, it does not necessarily include control over others’
rewards and punishments.
A wealth of research indicates that people from relatively upper-class backgrounds tend
to engage in dominance, specifically related to behaviors and perceptions that benefit the self:
For instance, prior research indicates that people from relatively upper -class backgrounds
evaluate the self more positively than their lower-class counterparts (Twenge & Campbell,
2002), tend to think that high status groups in society obtain their posit ions legitimately (e.g.,
Brandt, 2013), and are also less likely to engage in pro-social behaviors to help others in need
Sartorial Symbols of Social Class 7
relative to their lower-class counterparts (Piff, Kraus, Côté, Cheng, & Keltner, 2010). Thus, we
predict that wearing upper-class sartorial symbols will elicit increased self-benefitting patterns of
behavior relative to lower-class symbols.
Given that sartorial symbols of social class communicate one’s relative position in
society in comparison to others, wearing these symbols, we predict, will influence hormone
responses related to dominance (i.e., testosterone). Specifically, wearing lower-class sartorial
symbols will elicit decreases in testosterone levels (Akinola & Mendes, 2013; Mazur & Booth,
1998; Mehta, Jones, & Josephs, 2008; c.f. Mehta & Josephs, 2006) relative to wearing upper-
class sartorial symbols. We make this prediction based on prior research indicating that low
status contexts reduce testosterone levels whereas high status contexts maintain those levels: For
instance, research on chess tournament winners and losers found that losing chess matches
decreased testosterone whereas winning chess matches maintained testosterone levels (Mazur,
Booth, & Dabbs, 1992).
For our second prediction, we expect that upper-class sartorial symbols will elicit
increased vigilance of threats in perceivers of these symbols. Upper -class symbols observed
within an interaction, we predict, will increase the likelihood that perceivers will scan their
external environments for potential social- and survival-related threats. This will occur, we
reason, because having subordinate status relative to an interaction partner elevates expectations
that one may have lower standing in society and reduced resources to cope with demands of the
social context (for a review, see Kraus et al., 2012).
Several studies suggest that, within social interactions, perceiving the elevated s ocial
class of interaction partners elicits heightened vigilance: An eye tracking study revealed that
individuals focus more of their visual attention on targets who behave dominantly during social
Sartorial Symbols of Social Class 8
interactions (e.g., aggressively stating their opinion; Cheng, Tracy, Foulsham, Kingstone, &
Henrich, 2013). In related research in which two friends engaged in an interaction where they
were required to tease each other, participants exhibited heightened perceptions of hostile
emotions when interacting with an upper-class interaction partner (Kraus, Horberg, Goetz, &
In the present study, we predicted that perceivers of sartorial symbols of high social class
would exhibit increased threat vigilance—indexed in terms of cardiac vagal withdrawal. Cardiac
vagal withdrawal occurs during active tasks when the vagal brake on the heart is released and as
a result, the heart presents with reduced inter-beat variability (Grossman & Taylor, 2007). Vagal
withdrawal is indexed by examining changes in heart rate variability (HRV), which is the
variability between heart beats (Porges, 2007). In prior research, vagal withdrawal has been
linked to a variety of psychological states including greater conscious control and effort, reports
of psychological distress, and threat vigilance (e.g., Grossman & Taylor, 2007).
Increased threat vigilance is also likely to manifest in the form of affect contagion.
Affect contagion is the extent that people pass on the emotions they experience to others
(Barsade, 2002; Hatfield, Cacioppo, & Rapson, 1994). We predict that perceivers of upper-class
symbols will be more likely to catch others’ affective states than perceivers of lower -class
symbols. Given that upper-class symbols elicit vigilance and increased attentional focus, targets
wearing these symbols are likely to be used as informational guides for appropriate behavioral
and emotional responses during social interactions. Thus, the affect and physiological responses
of targets wearing upper-class symbols are more likely to influence their interaction partner than
Sartorial Symbols of Social Class 9
Several correlational studies are suggestive that symbols of high status result in greater
attention from low status partners, which will elicit contagious affect: Low -status individuals
tended to modulate their voices to become more similar in tone to a high -status partner over the
course of a live televised interview (Gregory & Webster, 1996), and low -status individuals are
more likely to engage neural circuitry involved in mentalizing—thinking about others thoughts
and feelings—than individuals higher in social status (Muscatell, et al., 2012). In an example
specific to social class, during the aforementioned laboratory interaction between close friends,
individuals were more likely to shift their self-reported emotions over the course of a social
interaction to become more similar to the emotions of their interaction partner, if that partner was
upper-class (Kraus et al., 2011).
Affect contagion can be assessed in multiple ways including coordinated ch anges in
facial expressions and behavior (Chartrand & Bargh, 1999), autonomic physiology (Levenson &
Reuf, 1992; Waters, Mendes, & West, 2014), neuroendocrine responses (e.g., Saxbe & Repetti,
2010), and vocal frequency (Gregory & Webster, 1996). We have chosen to examine
physiological changes, specifically changes in sympathetic nervous system (SNS) activation, as
our measure of affect contagion. We view SNS responses as an ideal way to examine affect
contagion because physiological responses vary as a function of arousal states, a key component
of affective experiences (Barrett & Russell, 1999; Mendes, 2009; Waters, Mendes, & West,
2014). Moreover, unlike self-reports of affect, SNS responses capture affective states as they
occur during interactions thereby allowing us to examine moment-to-moment changes in affect
(e.g., Blascovich, Mendes, Vanman, & Dickerson, 2010; Kassam & Mendes, 2013; Waters et al.,
Sartorial Symbols of Social Class 10
Based on the above analysis, we expected that wearing sartorial symbols of higher social
class would elicit dominance-related behavior and physiology whereas perceiving these upper-
class symbols would elicit increased threat vigilance—indexed in terms of vagal withdrawal and
affect contagion. We tested these two predictions in an experiment wherein one participant (the
target) donned clothing of upper-, neutral, or lower-class rank prior to a negotiation exercise.
In this study, male participants arrived at the experiment separately. One of the
participants (i.e., the target) was randomly assigned to a condition where he would wear lower-
class, neutral, or upper-class clothing. The target then joined his interaction partner (i.e., the
perceiver) in a large experiment room where the two would engage in a negotiation exercise
from prior research (Galinsky & Mussweiler, 2001). The perceiver was always unaware of the
target’s clothing change. Autonomic physiology was recorded prior to, during, and immediately
following the negotiation. The experiment was completed in two hours.
Our sample was recruited through Craisglist.org, and consisted of 134 healthy adult
males from the San Francisco Bay Area ranging in age from 18 to 34 (Mage=24.14). Participants
completed the study in dyads and we confirmed, prior to the study, that they did not know each
other. Two experimental sessions encountered errors and so the data for the four participants in
those sessions could not be used in analyses. In one other session, a participant expressed
suspicion about the bogus physiological sensors and was excluded from analyses along with his
partner. The final sample consists of the remaining 128 participants (64 dyads). Participants
(allowed to check multiple ethnic categories) were European American (n=88), Latino (n=22),
Sartorial Symbols of Social Class 11
Asian American (n=21), African American (n=15), Native American (n=9), or other (n=2) and
were from diverse social class backgrounds: Seventy-one participants were, at most, high school
educated, and the median family income was between $50,001 and $75,000 annually.
Participants arrived, scheduled 10 minutes apart and were seated in different rooms to
avoid meeting until the designated time. Disposable sensors were applied to the participants’
torso and limbs and they sat quietly for a 5-minute baseline/resting period (Mendes, 2009).
Following baseline, participants provided a saliva sample. After baseline, half of participants
(i.e., perceivers) were instructed that they would move to another room where they would meet
The other half of participants, (i.e., targets), were instructed after baseline that they were
taking part in a test of ambulatory physiological equipment embedded in clothes that were to be
validated against the stationary equipment. This “cover story” distracted participants’ attention
from the purpose of the clothing manipulation in an attempt to reduce demand characteristics.
One of two types of clothing, which had visible sensors that were sewn into the lining, was then
presented to participants. Participants in the neutral condition (ndyad =24) were allowed to remain
in their current clothing and were instructed that the laboratory needed to run some additional
tests on the stationary physiological sensors. Photos of the upper- (ndyad =20) and lower-class
(ndyad =20) sartorial symbols are displayed in Figure 1.
Following this clothing change, targets and perceivers were placed in the same
experiment room, initially separated by a portable dividing wall, where they would sit for the
remainder of the study. We then removed the portable wall allowing the participants to see each
other for the first time. We confirmed that the participants did not know each other, and had
Sartorial Symbols of Social Class 12
them get acquainted. Following this interaction, we provided instructions for a negotiation task.
After the negotiation, a second saliva sample was collected, and questionnaire s were completed.
The target then changed back into his original clothing and all participants were probed for
suspicion, debriefed, and paid for their participation.
Materials and Measures
Clothing manipulation. The upper-class clothing consisted of a black suit, a white long-
sleeve button-down collared shirt, black socks, and a pair of black leather dress shoes all
purchased at Macy’s. The lower-class clothing consisted of a white short-sleeve t-shirt, blue
sweat pants, and plastic sandals all purchased at Walgreens.
The experimenter described to targets that the laboratory was attempting “to test a new
version of advanced ambulatory physiological sensors that can record a person’s physiological
responses while they conduct normal daily activities.” The experimenter then showed
participants the “experimental physiological sensors,” which were bogus sensors that were sewn
into the clothing inside one leg, under one arm, and inside the shoes. The experimenter described
the sensors while showing them to participants. Participants were then disconnected from the
physiological monitoring station and went to an adjacent room to put on the clothing.
After the clothing change, participants viewed themselves in a full length mirror before
returning to the laboratory room. Upon return, the experimenter re -attached the leads to the
monitoring equipment and then placed an IPhone in the pocket of the suit or the sweats with the
explanation that the sensors would “transmit the participants’ physiological responses wirelessly
to the IPhone.” To enhance the realism of the procedures, the experimenter spent time with the
participant, pretending to validate the lab sensors against the sensors in the clothing.
Sartorial Symbols of Social Class 13
Negotiation task. In the negotiation task, participants role play as the chief financial
officer of a biotechnology company—Synertech or Dosagen (see Galinsky & Mussweiler, 2001).
As in prior research, participants were given written and verbal instructions for the negotiation
that included shared information about the negotiation (e.g., estimates of real estate market
changes) and confidential information about their own company (e.g., the cost of building a new
plant). The confidential information included a best alternative to a negotiated agreement
(BATNA) as in prior research—for the seller, the profit from stripping the plant and selling the
parts, and for the buyer, the cost of building a new plant. With this information, a fair sale price
for the manufacturing plant, where both parties benefit equally is $20.5 million, but because the
buyer and seller have confidential information, neither party knew this true compromise value.
Given that Masters in Business Administration (MBA) students are the typical
participants for this negotiation task, we modified the procedures for our community sample
who, based on pilot testing, had far less negotiation experience: To give participants more
information about successful negotiation behavior, we provided several examples (e.g., not
settling for an unfair offer, taking the full allotted time to mull over an agreement, holding out
for more than just an opponent’s first offer). To encourage competition between our participants,
the experimenter informed participants that they could win up to an additional $5 for engaging in
these types of successful negotiation behaviors. We also imposed a 6-minute limit on the
negotiation because pilot testing revealed this time limit was sufficient for reaching agreement in
most dyads. Specifically, during experimenter training, all but one of our practice dyads (made
up of research assistants) reached agreement prior to 6-minutes of negotiation. Participants were
instructed that they did not need to reach an agreement in the 6-minute time period. If
participants did not agree on a sale price (which occurred in 18 dyads), the last offer a participant
Sartorial Symbols of Social Class 14
made during the negotiation was used as the final price offer in the analyses. A chi -square
analysis revealed no differences in reaching agreement during the negotiation by condition
Cardiac Vagal Withdrawal. Mendes (2009) provides a comprehensive and detailed
review of methods to collect and score autonomic physiology. Inter-beat variability between
heart beats, i.e., cardiac vagal tone, was measured using electrocardiography (ECG, Biopac,
Goleta, CA). Two pre-gelled snap electrodes were placed directly on the skin of participants in a
Lead II configuration (right arm, left leg). Signals were integrated with Biopac MP 150 hardware
and ECG responses were edited and scored offline by the first author. ECG responses were
visually examined and HRV scores were calculated using Mindware software (Mindware
Technologies, Gahanna, OH; HRV 3.0), averaged in 30s intervals, and computed using a Fast
Fourier Transformation (Mendes, 2009). Heart rate variability (HRV) reactivity scores were
calculated by subtracting HRV scores during the last 30 seconds of the 5-minute baseline
(presumably the most relaxed interval) from the HRV scores collected during the negotiation
task (Mendes, 2009). Rates of breathing (i.e., respiration rates), measured from this same time
interval, were assessed from impedance cardiography (see below). We controlled for
participant’s respiration rate in all analyses of HRV data given that respiration influences inter-
beat variability (Grossman & Taylor, 2007).
Sympathetic Nervous System. Pre-ejection period (PEP) is a measure of sympathetic
nervous system activation representing the time interval between the contraction of the left
ventricle and the opening of the aortic valve. We used PEP scores to calculate affect contagion,
as in prior research (Waters et al., 2014). The benefits of this measure include that PEP reactivity
Sartorial Symbols of Social Class 15
is responsive to affective states, responds in a short-time frame (3 to 5 seconds), can be measured
unobtrusively and continuously, and has extant literature linking the responses to psycholo gical
states, primarily general arousal, that are directly related to a specific underlying biological
system—the sympathetic nervous system.
PEP is derived from both ECG and impedance cardiography. Cardiac impedance was
collected using four strips of disposable tetrapolar aluminum/mylar electrodes that are placed
directly against the skin and completely encircled participants’ neck and torso (Mendes, 2009).
PEP also was scored in 30s intervals using Mindware (IMP 3.0) and is calculated as the duration
between the Q-point on the ECG waveform, and the b-point on the Δz/Δt waveform obtained
from impedance cardiography. Data were visually inspected by the authors for any recording
artifacts and following numerous published studies the b-point was manually adjusted rather than
relying on algorithms (e.g., Blascovich & Mendes, 2010; Mendes, 2009; c.f., Lozano et al.,
Testosterone assays. Participants passively drooled 1ml of saliva into a 2ml vial (IBL
salicap) using plastic straws (Mbaseline=111.57pg/ml, SDbaseline=51.79pg/ml). These samples were
stored in a -80°C freezer and then shipped to be assayed at Kirschbaum’s laboratory in Dresden,
Germany. The testosterone assay used 25ul of saliva per determination, has a lower limit of
sensitivity of 1 pg/mL, and average intra-and inter-assay coefficients were below 11% (IBL
International, Hamburg, Germany).
We used two common outcome measures from the negotiation exercise as indices of
performance during the negotiation: profit earned and concessions offered (e.g., Galinsky &
Mussweiler, 2001). To determine the amount of profit earned and concessions offered during the
Sartorial Symbols of Social Class 16
negotiation, two coders watched video recordings of the negotiation and recorded the first and
last price that each participant offered. First (M=$1.93m, SD=$4.12m) and final (M=$0.25m,
SD=$3.39m) offers were coded to indicate amount of profit buyers and sellers would earn over
and above the $20.5m fair price. These two price values were also subtracted from each other to
indicate the total amount of concessions, in millions, during the negotiation (M=$2.02m,
Sense of Power
Prior to the start of the social interaction, participants indicated their baseline sense of
power on 7-point Likert scales (1 = strongly disagree, 7 = strongly agree). A sample item from
the eight-item scale is “I can get people to listen to what I say.” (M=4.95, SD=0.86; Anderson,
Keltner, & John, 2012). Participants indicated their sense of power right after the negotiation and
at the end of the experiment by answering the question: “During the negotiation/experiment, I
felt I had power and influence.” These two items were highly correlated ( r=.54, p<.05) and so a
composite measure was created indicating sense of power during the experiment ( M=5.00,
SD=1.09). We then computed changes in sense of power by subtracting standardized scores of
the mean for power in the experiment from the baseline mean (M=0.00, SD=1.16).
To determine the success of our sartorial manipulation, a team of three coders rated
photographs of targets in the upper- and lower-class clothing conditions taken prior to the
clothing change using a 7-point Likert scale indicating ascending levels of sartorial formality (1
=“sweats or clothing for exercising”, 7 = “dress shirt and slacks or business suit”). The three
coders’ ratings of participants clothing correlated highly (rs between .71 and .98, ps < .01; α =
Sartorial Symbols of Social Class 17
.88). In general, participants wore moderately informal clothing when arriving at the experiment
(M=3.25, SD=0.82) and this clothing was uncorrelated with the assigned social class of
participants (coded “-1” for lower-class, “0” for neutral, and “1” for upper-class) during the
experiment (r=.03, p=.70).
Our manipulation of sartorial symbols was effective in shifting participant clothing
formality: In the lower-class symbols condition, we determined if coder judgments of target
clothing differed from a value of 1 (“sweats or clothing for exercising”). This analysis revealed
that targets in the lower-class condition had significantly more formal clothing when they arrived
at the experiment (M= 3.43) than the clothing they were assigned by the manipulation
t(19)=12.15, p<.01. We then compared upper-class targets’ rated clothing to a value of 7 (“dress
shirt and slacks, or business suit”). Again, this analysis revealed that targets in the upper-class
condition had significantly less formal clothing (M= 3.53) than they were assigned by the
manipulation t(19)=-15.06, p<.01.
We also determined if the sartorial manipulation was successful in manipulating
perceptions of social status using a separate sample of 200 observers who were collected online
through Mechanical Turk (www.mturk.com). The online observers were instructed to view still
photographs of the first 10 of our laboratory participants assigned to the upper- and lower-class
symbols conditions wearing either their assigned clothing or their own neutral clothing prior to
the clothing change. The online observers were instructed to make various judgments about the
person based solely on the photographs. These stimuli were randomly presented such that each
of the observers had an equal chance of seeing the 10 laboratory participants in the manipulated
clothing or their own neutral clothing. In addition to a number of filler measures, participants
Sartorial Symbols of Social Class 18
responded to the single item “This person has high social status.” using a 7 -point Likert scale (1
=disagree strongly, 7 = agree strongly).
A repeated measures Analysis of Variance (ANOVA) comparing observer ratings of
social status for laboratory participants in the upper-, neutral, and lower-class symbols conditions
yielded significant mean differences in observer-rated social status between participant clothing
conditions F(2,380)=112.40,p<.01, µp2=.37. Examination of the specific means reveals a pattern
aligning with our expectations: Laboratory participants wearing a business suit were judged to be
highest in social status (M= 4.65, 95% CI [4.51 to 4.78]) with participants wearing the t-shirt
and sweats rated the lowest in status (M= 3.42, 95% CI [3.24 to 3.59]) and neutral participants in
between (M= 3.61, 95% CI [3.48 to 3.73]). These results indicate that the sartorial manipulation
was successful in shifting participant social status, as rated by observers viewing static
Dominance in the Negotiation
Profits and concessions. Our overarching hypothesis was that sartorial symbols of social
class would shift the behavior and physiology of both the wearers and perceivers of these
symbols in class-consistent ways. To test this prediction, we first examined negotiation outcomes
with the expectation that targets wearing upper-class symbols would optimize their own self-
interests during the negotiation relative to their partner, which we operationalized as obtaining a
higher relative profit beyond the $20.5m fair compromise value. Given that the negotiation was a
competitive interaction between the target and perceiver, we conducted an Analysis of Variance
(ANOVA) predicting negotiation profit with dyad as the unit of analysis, target versus perceiver
profit as the within subjects factor, and dyad social class as the between subjects factor (Kenny,
Kashy, & Cook, 2006). This ANOVA technique allows us to compare differences in profit
Sartorial Symbols of Social Class 19
between members of the same dyad, which is the most appropriate comparison for our relative
profit hypothesis. For all subsequent analyses, we use a hierarchical linear model (HLM)
controlling for dyad-level dependence (Kenny et al., 2006). Unlike the ANOVA technique which
compares outcomes within the same dyad, the HLM technique allows for the independent testing
of target and perceiver effects on behavior and physiology.1
The ANOVA revealed no main effects for target vs. perceiver role F(1,59)=1.14, p=.29
or dyad social class F(2,59)<1, but as predicted, a significant interaction emerged between dyad
social class and target versus perceiver role F(2,59)=3.39, p<.05, µp2=.06. Examination of 95%
confidence intervals surrounding the calculated profit mean revealed that targets wearing upper-
class symbols obtained significantly higher profits (M=$2.06m, 95% CI [$0.47m to $3.65m])
than did perceivers (M=-$1.20m, 95% CI [-$2.63m to $0.24m]) beyond the $20.5m fair
compromise value. Targets and perceivers did not differ in profits in the lower-class
(Mtarget=$0.68m, Mperceiver=-$0.50m) or neutral dyads (Mtarget=-$0.66m, Mperceiver=$1.15m).
We also expected sartorial symbols to influence concessions in the negotiation, with
lower-class targets offering more concessions than upper-class targets. To conduct this analysis,
we predicted negotiation concessions with target and perceiver clothing condition (coded “-1”
for lower-class symbols, “0” for neutral, and “1” for upper-class symbols) in a HLM analysis
controlling for dyad-level dependence (Kenny et al., 2006).
As expected, lower-class targets made significantly more concessions than did upper-
class targets b=-1.01, t(105.79)=-2.95, p<.01, d=.41 (see Table 1 for means). For perceivers, no
significant differences in concessions emerged by condition b= -0.56, t(105.79)=-1.63, p=.11.
Target (t<1) and perceiver (t=-1.04, ns) social class did not influence first offers in the
Sartorial Symbols of Social Class 20
negotiation. This latter result suggests that differences in the amount that participants offered in
concessions were not driven by differences in starting offers.
Testosterone. Next, we examined the influence of sartorial symbols on testosterone (T)
change. To the extent that lower status reduces feelings of dominance relative to elevated rank,
we expected that participants wearing lower-class sartorial symbols would show lower levels of
salivary T after the negotiation relative to upper-class targets (Mazur & Booth, 1998). The HLM
analysis revealed that lower-class targets showed a significant reduction in percent of T relative
to upper-class targets b=9.87, t(122.23)=2.25, p<.05, d=.15. Perceivers showed no differences in
T change based on their partner’s social class symbols b=1.22, t(122.39)=0.27, p=.79. The
sartorial symbols manipulation effect on T change held when controlling for target and perceiver
age b=11.05, t(106.78)=2.49, p<.05 (Table 1). Importantly, targets wearing lower-class sartorial
symbols also had significantly lower T levels relative to neutral targets b=8.82, t(98.82)=2.43,
p<.05. This latter finding suggests that lower-class sartorial symbols, and not upper-class
symbols, are the catalyst of T change. This finding is consistent with prior research examining
associations between T change and social status (e.g., Mazur et al., 1992).
Threat Vigilance in the Negotiation
Cardiac vagal withdrawal. Having observed evidence suggesting that sartorial symbols
elicit class-consistent changes in self-benefitting behavior and T, we then examined predictions
related to physiological reactivity that is correlated with threat vigilance within dyadic
interactions. We expected that upper-class sartorial symbols would elicit cardiac vagal
withdrawal in perceivers. The HLM analysis did not yield a target effect: Participants wearing
upper- and lower-class clothing did not show differences in HRV reactivity b=0.20,
t(116.70)=1.03, p=.30, however perceivers were affected by their partner’s sartorial symbols.
Sartorial Symbols of Social Class 21
Specifically, perceivers interacting with targets in upper-class clothing showed greater vagal
withdrawal from baseline relative to perceivers interacting with a target in lower-class clothing
b=-0.38, t(116.70)=-1.97, p=.05, d=.12 (Table 1). These results suggest that perceivers may have
engaged in greater vigilance of their upper-class interaction partner.
Subjective sense of power. Prior research indicates that low-power individuals tend to
exhibit enhanced vigilance relative to high-power individuals (e.g., Galinsky, Magee, Inesi, &
Gruenfeld, 2006). Given these data, we tested whether perceivers of upper-class targets
experienced increased powerlessness during the competitive negotiation with their partner . We
tested this possibility by conducting an analysis predicting self-reported changes in sense of
power from baseline with target and perceiver social class symbols while controlling for target
and perceiver baseline power. Consistent with the HRV analysis, we observed no effect of target
social class b=0.03, t(121.65)=0.19, ns, but perceivers showed a change in sense of power b=-
0.32, t(121.65)=-2.17, p<.05, d=.29. That is, perceivers of upper-class targets felt reduced power
during the negotiation relative to perceivers of lower-class targets.
Affect contagion. Finally, we hypothesized the upper-class sartorial symbols would
catalyze affect contagion in perceivers. To assess affect contagion, we tested whether upper-class
targets’ physiological responses would influence their partner’s in a time-lag design: We
examined changes in PEP during the negotiation; where shorter intervals between the ventricle
contraction and the aortic valve opening indicate greater sympathetic nervous system (SNS)
activation. We computed average correlations representing the association between a
participant’s PEP at time X and their partner’s PEP at time X+1 ( M=0.001, SD=0.34). We chose
a time lag of one unit (equaling 30 seconds) given the rapid responsiveness of SNS activation.
Sartorial Symbols of Social Class 22
Higher average positive correlations indicate that the participants’ time X PEP predicted his
partner’s change in PEP at time X+1.
Consistent with our affect contagion hypothesis, when targets wore upper-class clothing
their time-X PEP was more positively associated with their partner’s time X+1 PEP —indicating
physiological contagion—than it was for targets who wore lower-class clothing b=0.11,
t(120.78)=2.03, p<.05, d=.13 (Table 1). No perceiver effects for PEP contagion on target’s
responses were observed b=-.02, t(120.78)=-0.45, p=.65.
Raw correlations between variables reported across analyses are displayed in Table 2.2
When controlling for respiration rates, heart rate variability (HRV) during the negotiation was
significantly positively associated with offering more concessions. One possible explanation of
this effect is that activation of the vagus nerve has been associated with pro-social emotional
states, such as the experience of compassion, in prior research (Goetz, Keltner, & Simon -
Thomas, 2010). Alternatively, lower levels of HRV reactivity have been linked to better
decision-making, effort and conscious control, which might explain lower concessions (Kassam,
Koslov, & Mendes, 2009). No other correlations were significant. The lack of significant
correlations between physiological assessments and self-reports or behaviors potentially reflects
the differing psychometric properties of each measure, or that physiological responses are
differentially impacted by unconscious processes (Mauss et al., 2005; Mendes, 2013). Self-
reports of annual income and educational attainment did not significantly interact with the
sartorial manipulation (ts < 1), and controlling for these indices did not change the relationship of
the sartorial manipulation to any of the outcome measures.3
Sartorial Symbols of Social Class 23
Symbols of rank are communicated across social living animals and organize social life
by managing the expectations of those perceiving these symbols (Krebs et al., 1993). In humans,
symbols of social class communicate hierarchical position in dyadic interactions, but up to this
point, the capacity of these symbols to actually elicit class-consistent behavior and physiology
had never been tested. In the present research, we examined the influence of manipulated
sartorial symbols of social class within dyads to determine their unique influence on the behavior
and physiology of targets who wear the symbols and on perceivers of these symbols.
Consistent with our hypotheses, results suggest that wearing sartorial symbols influence
both self-benefitting behavior during negotiations and hormone levels related to dominance.
With respect to behavior, wearing an upper-class business suit increased profits within a
competitive negotiation and decreased concessions offered relative to wearing lower-class
sartorial symbols. Wearing sartorial symbols also shifted neuroendocrine responses: wearing
lower-class clothing resulted in significantly lower testosterone relative to wearing upper-class
Perceiving symbols of social class worn by others influenced participants’ threat
vigilance during the negotiation: Specifically, perceivers of upper-class targets experienced
significant cardiac-vagal withdrawal during the negotiation —a physiological state associated
with increased vigilance and attention (Grossman & Taylor, 2007)—relative to perceivers of
lower-class targets. Importantly, perceivers of upper-class targets also reported reductions in
sense of power following the negotiation in comparison to lower-class perceivers. This result
dovetails with the HRV findings because, as prior research indicates, low power is also
associated with increased vigilance (Keltner et al., 2003). Given that sartorial symbols of social
Sartorial Symbols of Social Class 24
class were manipulated in the present research, these results represent the first causal evidence
suggesting that perceiving others’ upper-class symbols in interactions elicits vagal withdrawal.
Finally, upper-class targets catalyzed affect contagion in their partner during the
negotiation. That is, the SNS responses of individuals wearing upper-class sartorial symbols
were more likely to be caught by their partner 30s later in the interaction than were the SNS
responses of lower-class individuals. This effect aligns with past correlational evidence
suggesting that affect contagion occurs in the direction of upper-class individuals’ affect (Kraus
et al., 2011) and is, in our estimation, the first evidence suggesting that social class symbols are
catalysts in this contagion process.
The current research examined sartorial symbols of social class and we differentiate these
social class symbols from other symbols of dominance or status (e.g., social power). We make
this distinction because, although measures of social power are correlat ed with measures of
social class in prior research (Bullock & Lott, 2010; Keltner et al., 2003; Kraus et al., 2012), the
correlations are only moderate in size and so these rank-based constructs are empirically
separate. As well, whereas social power represents direct control over others’ resources and
ability to administer punishments, symbols of social class are defined within social comparisons
between one’s own resources and that of others—elevated social class does not necessarily
indicate control over others’ outcomes. Moreover, research indicates that status and power vary
independently of each other: In prior research, high power individuals perceived as low in status
tended to be perceived less warmly (Fragale, Overbeck, & Neale, 2011) and behave d more anti-
socially than their high power/high status counterparts (Fast, Halevy, & Galinsky, 2012). These
results underscore the empirical distinctions between power and status.
Sartorial Symbols of Social Class 25
Related to this point, it is important to acknowledge that our manipulation of sartorial
symbols of social class, though it elicited a higher sense of power in perceivers of lower -class
targets, is not a true manipulation of social power. Manipulations of social power typically
involve participants thinking of a time when they had control over others or actually give
participants control over some resource (e.g., Côté et al., 2011). Participants in the present
research had equal control over rewards and punishments during the experiment tasks across the
conditions of the sartorial manipulation. Despite these theoretical and methodological
distinctions between class and power, it is clear from these results that symbols of social class
exert influence on social cognition and behavior in ways that are similar to that of social power
(Keltner et al., 2003).
One noteworthy implication of the findings of the current experiment is that class -based
patterns of behavior and physiology, typically thought of as culturally learned and socialized
during early environment experiences, are more malleable than researchers have previously
argued (e.g., Weininger & Lareau, 2009; Fiske & Markus, 2012). That these patterns change as a
function of the symbols of social class that people express in everyday social interactions is
indicative of the process by which relative social class is communicated in society more broadly
(Kraus et al., 2013). Specifically, individuals enter into social interactions with others, bringing
with them symbols of social class that, in turn, both change perceptions of their own position in
the social hierarchy and the way they are perceived by others. Understanding the ways in which
social class symbols elicit class-consistent behavior in dyadic settings is an exciting area of
future research. Importantly, future research on nonverbal status symbols would benefit from
considering the simultaneous and dynamic influence of these symbols on both targets and
perceivers within dyadic interactions.
Sartorial Symbols of Social Class 26
Several important limitations should be noted with respect to this study and its
conclusions. The negotiation task we used in the present research was modified in order to
accommodate differences in negotiation experience between the typical MBA student
participants and our community sample (e.g., Blader & Chen, 2012; Galinsky & Mussweiler,
2001). Thus, we advocate some caution when interpreting the findings from this research with
respect to the broader negotiation literature.
Our sartorial manipulation was easily visible for participants, and as such, it is possible
that demand characteristics played a role in some of the patterns we observed in our results. Our
experiment went to great effort to reduce the demand characteristics inherent in donning suits
versus sweats by directing the attention of participants to the goal of validating ambulatory
physiological equipment. These steps help us minimize the extent that our results can be
explained by participant expectations (Boot, Simons, Stothart, & Stutts, 2013). As well, our
examination of funnel debriefing questions asked of participants at the end of the study
suggested a lack of awareness of the study hypotheses.
In our results we found some differences between upper- and lower-class sartorial
symbols on behavior and physiology. For T change, we also found that participants in the lower -
class condition differed from neutral participants, indicating that lower -class sartorial symbols
decrease T levels. This finding is sensible given that most research on social status and T change
reveals that low status reduces T rather than high status increasing it (Mazur, Booth, & Dabbs,
1992; Mehta & Josephs, 2006). For our other outcome measures, however, neutral participant
responses were directly in between upper- and lower-class participants—thereby, limiting the
extent we can conclude that our effects were driven by either lower - or upper-class sartorial
Sartorial Symbols of Social Class 27
symbols. Future research examining the effects of symbols of social class on behavior and
physiology, using larger samples, may more definitively answer the question of whether lower -
or upper-class symbols drive dominance or threat vigilance.
We made the decision to focus on male dyads rather than female or mixed -gender dyads
because we wanted to constrain participants’ sartorial options for the clothing manipulation.
Because of this, we are limited in what we can conclude about the sensitivity of female dyads to
sartorial class symbols.
Additionally, we chose a negotiation context that conceptually matched the upper-class
sartorial symbols we used in our manipulation. Because of this, it is unclear how much the “fit”
between clothing and context accounts for the effects we observe here. Some of our own
exploratory analyses (see Footnote 2) suggest that clothing condition did not influence
participant feelings of comfort with the negotiation task. Still, a more definitive test of our
hypotheses might occur in an experiment where the sartorial symbols completely misfit the
context of the interaction: For example, if we had participants engage in a coaching exercise
would the lower-class sartorial symbols have been more effective in terms of affect contagion
because the clothes matched this context? Clearly, our arguments regarding social class symbols
suggest that, in most cases, we expect social class symbols to elicit a similar pattern of results to
what we observed in this study. Future research is necessary in this regard, to completely rule out
situational fit as an alternative explanatory variable.
With respect to our affect contagion findings, research indicates that the spread of affect
involves both the perception of others’ affective states as well as target expressions of affect
(Zaki, Bolger, & Ochsner, 2008). In the present research, we cannot completely rule out the
possibility that expressivity, and not vigilance, accounts for the affect contagion findings.
Sartorial Symbols of Social Class 28
However, that overall changes in cardiac output did not differ by experimental condition (see
Footnote 2) indicates that targets were not different in activity levels during the negotiation based
on the sartorial manipulation. This result further raises the possibility that the enhanced vigilance
of perceivers of upper-class targets played a role in the contagious spread of affe ct.
Future Directions and Conclusions
Our findings generate several promising lines of future inquiry. A first concerns whether
there are other symbols that might engender immediate changes in social class (e.g., Gillath et
al., 2014). We chose to focus on clothing because of its clear use as symbols of social class in
social groups (Bourdieu, 1979). Since the current investigation involved men, it would also be
important to determine if similar patterns would be observed both for women and in mixed -
gender interactions. For the latter case, perhaps the social class signals inherent in clothing are
harder to discern across genders.
Notably, our study involved a context that asked participants to engage in a competitive
interaction. Would a similar pattern emerge if participants were instructed to help each other
complete a cooperative task? Perhaps incentives for cooperation would lead upper-class
individuals to enhanced cooperative patterns. Research showing enhanced cooperation among
upper-class individuals induced to experience pro-social emotional states suggests this possibility
(Kraus et al., 2012). It was also interesting that mean levels of concessions offered were highest
among partners of lower-class targets (see Footnote 1). We speculate that this pattern may have
been due to either assumptive helping or disengagement from an “unworthy” opponent, but
future research should test these explanations more systematically.
Finally, it is interesting to speculate about how sartorial symbols of social class influence
behavior outside the laboratory. For instance, school uniforms have helped schools improve
Sartorial Symbols of Social Class 29
student conduct in a variety of studies, and this might be because of the way these policies
dampen naturally occurring sartorial symbols (Bodine, 2003). Importantly, our research suggests
that one of the benefits of “dressing the part” at a job might be that it helps individuals more
easily shift their behavior to match their desired position in society.
Sartorial Symbols of Social Class 30
Adam, H., & Galinsky, A. D. (2012). Enclothed cognition. Journal of Experimental Social
Psychology,48 (4), 918-925.
Adler, N. E., Boyce, T., Chesney, M. A., Cohen, S., Folkman, S., Kahn, R. L., & Syme, S. L.
(1994). Socioeconomic status and health: The challenge of the gradient. American
Psychologist, 49, 15-24.
Akinola, M. & Mendes, W. B (in press). It’s good to be the king: Neurobiological
benefits of higher social standing. Social and Personality Psychological Science.
Anderson, C., John, O. P., & Keltner, D. (2012). The personal sense of power. Journal of
personality,80 (2), 313-344.
Barrett, L. F., & Russell, J. A. (1999). The structure of current affect controversies and emerging
consensus. Current Directions in Psychological Science,8(1), 10-14.
Barsade, S. G. (2002). The ripple effect: Emotional contagion and its influence on group
behavior. Administrative Science Quarterly,47(4), 644-675.
Blader, S. L., & Chen, Y. R. (2012). Differentiating the Effects of Status and Power: A Justice
Perspective. Journal of personality and social psychology,102(5), 994-1014.
Blascovich, J., & Mendes, W. B. (2010). Social psychophysiology and embodiment. Handbook
of social psychology. Wiley Online.
Blascovich, J., Vanman, E., Mendes, W. B., & Dickerson, S. (2011). Social psychophysiology for
social and personality psychology. SAGE Publications Limited.
Boot, W. R., Simons, D. J., Stothart, C., & Stutts, C. (2013). The Pervasive Problem With
Placebos in Psychology Why Active Control Groups Are Not Sufficient to Rule Out
Placebo Effects. Perspectives on Psychological Science,8(4), 445-454.
Sartorial Symbols of Social Class 31
Bourdieu, P. (1979). Distinction. (Routledge, London).
Bodine, A. (2003). School uniforms, academic achievement, and uses of research. Journal of
educational research,97 (2), 67-71.
Brandt, M. J. (2013). Do the disadvantaged legitimize the social system? A large -scale test of the
status–legitimacy hypothesis. Journal of personality and social psychology,104(5), 765-
Bullock, H. E., & Lott, B. (2010). Social class and power. The social psychology of power, 408-
Carney, D. R., Cuddy, A. J., & Yap, A. J. (2010). Power Posing: Brief Nonverbal Displays
Affect Neuroendocrine Levels and Risk Tolerance. Psychological science,21 (10), 1363-
Chartrand, T. L., & Bargh, J. A. (1999). The Chameleon Effect: The Perception-Behavior Link
and Social Interaction. Journal of Personality and Social Psychology ,76(6), 893-910.
Cheng, J. T., Tracy, J. L., Foulsham, T., Kingstone, A., & Henrich, J. (2013). Two ways to the
top: Evidence that dominance and prestige are distinct yet viable avenues to social rank
and influence. Journal of personality and social psychology,104 (1), 103-125.
Côté, S., Kraus, M. W., Cheng, B. H., Oveis, C., Van der Löwe, I., Lian, H., & Keltner, D.
(2011). Social power facilitates the effect of prosocial orientation on empathic
accuracy. Journal of personality and social psychology,101(2), 217-232.
Fast, N. J., Halevy, N., & Galinsky, A. D. (2012). The destructive nature of power without
status. Journal of Experimental Social Psychology,48(1), 391-394.
Fiske, S. T. & Markus, H. R. (Eds.). (2012). Facing Social Class. New York: Russell Sage.
Sartorial Symbols of Social Class 32
Fragale, A. R., Overbeck, J. R., & Neale, M. A. (2011). Resources versus respect: Social
judgments based on targets' power and status positions. Journal of Experimental Social
Galinsky, A. D., Magee, J. C., Inesi, M. E., & Gruenfeld, D. H. (2006). Power and perspectives
not taken. Psychological science,17 (12), 1068-1074.
Galinsky, A. D., & Mussweiler, T. (2001). First offers as anchors: The role of perspective -taking
and negotiator focus. Journal of personality and social psychology,81 (4), 657-669.
Gillath, O., Bahns, A. J., Ge, F., & Crandall, C. S. (2012). Shoes as a source of first
impressions. Journal of Research in Personality,46(4), 423-430.
Goodwin, S. A., Operario, D., & Fiske, S. T. (1998). Situational power and interpersonal
dominance facilitate bias and inequality. Journal of Social Issues,54(4), 677-698.
Gregory Jr, S. W., & Webster, S. (1996). A nonverbal signal in voices of interview partners
effectively predicts communication accommodation and social status
perceptions. Journal of personality and social psychology, 70 (6), 1231-1240.
Grossman, P., & Taylor, E. W. (2007). Toward understanding respiratory sinus arrhythmia:
relations to cardiac vagal tone, evolution and biobehavioral functions. Biological
psychology,74 (2), 263-285.
Hatfield, E., Cacioppo, J. T., Rapson, R. (1994). Emotional contagion. Cambridge, UK:
Cambridge university press.
Kassam, K. S., & Mendes, W. B. (2012). The Effects of Measuring Emotion: Physiological
Reactions to Emotional Situations Depend on whether Someone Is Asking. PloS one,8
Sartorial Symbols of Social Class 33
Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and
inhibition. Psychological review,110 (2), 265-284.
Kenny, D., Kashy, D., Cook, W. (2006). Dyadic data analysis. (Guilford, New York).
Kraus, M. W., Côté, S., & Keltner, D. (2010). Social class, contextualism, and empathic
accuracy. Psychological science,21 (11), 1716-1723.
Kraus, M. W., Horberg, E. J., Goetz, J. L., & Keltner, D. (2011). Social class rank, threat
vigilance, and hostile reactivity. Personality and Social Psychology Bulletin,37 (10),
Kraus, M. W., & Keltner, D. (2009). Signs of Socioeconomic Status A Thin -Slicing
Approach. Psychological science,20 (1), 99-106.
Kraus, M. W., Piff, P. K., Mendoza-Denton, R., Rheinschmidt, M. L., & Keltner, D. (2012).
Social class, solipsism, and contextualism: How the rich are different from the
poor. Psychological review,119 (3), 546-572.
Kraus, M. W. & Stephens, N. M. (2012). A road map for an emerging psychology of social class.
Social and Personality Psychology Compass, 6, 642-656.
Kraus, M. W., Tan, J. J. X., & Tannenbaum, M. B. (2013). The Social Ladder: A Rank -Based
Perspective on Social Class. Psychological Inquiry,24(2), 81-96.
Krebs, J., Davies, N., Parr, J. (1993). An introduction to behavioural ecology (3rd ed.).
(Blackwell, Cambridge, MA).
Levenson, R. W., & Ruef, A. M. (1992). Empathy: a physiological substrate. Journal of
personality and social psychology,63(2), 234-246.
Markus, H. R., & Kitayama, S. (2010). Cultures and Selves: A Cycle of Mutual
Constitution. Perspectives on Psychological Science,5(4), 420-430.
Sartorial Symbols of Social Class 34
Marmot, M. G., Stansfeld, S., Patel, C., North, F., Head, J., White, I., ... & Smith, G. D. (1991).
Health inequalities among British civil servants: the Whitehall II st udy. The
Mauss, I. B., Levenson, R. W., McCarter, L., Wilhelm, F. H., & Gross, J. J. (2005). The tie that
binds? Coherence among emotion experience, behavior, and physiology. Emotion,5(2),
Mazur, A., & Booth, A. (1998). Testosterone and dominance in men. Behavioral and Brain
Sciences,21 (3), 353-363.
Mazur, A., Booth, A., & Dabbs Jr, J. M. (1992). Testosterone and chess competition. Social
Psychology Quarterly, 70-77.
Mehta, P. H., Jones, A. C., & Josephs, R. A. (2008). The Social Endocrinology of Dominance:
Basal Testosterone Predicts Cortisol Changes and Behavior Following Victory and
Defeat. Journal of Personality and Social Psychology,94(6), 1078-1093.
Mehta, P. H. & Josephs, R. A. (2006). Testosterone change after losing predicts the decision to
compete again. Hormones and Behavior, 50, 684-692.
Mendes, W. B. (2009). Assessing autonomic nervous system activity. Methods in social
Mendes, W. B. (2013). Inspired by the question not the measure: Exploiting neurobiological
responses in the service of intergroup research. In N. Ellemers, D. Scheepers, and B.
Derks (Eds.). Social Neuroscience of Prejudice.
Muscatell, K. A., Morelli, S. A., Falk, E. B., Way, B. M., Pfeifer, J. H., Galinsky, A . D., ... &
Eisenberger, N. I. (2012). Social status modulates neural activity in the mentalizing
network. Neuroimage,60(3), 1771-1777.
Sartorial Symbols of Social Class 35
Oakes, J. M., & Rossi, R. H., (2003). The measurement of SES in health research: Current
practice and steps toward a new approach. Social Science and Medicine, 56, 769-784.
Oyserman, D. (2011). Culture as situated cognition: Cultural mindsets, cultural fluency, and
meaning making. European Review of Social Psychology,22(1), 164-214.
Piff, P. K., Kraus, M. W., Côté, S., Cheng, B., & Keltner, D. (2010). Having less, giving more:
The influence of social class on prosocial behavior, Journal of Personality and Social
Psychology, 99, 771-784.
Porges, S. W. (2007). The polyvagal perspective. Biological psychology,74(2), 116-143.
Rheinschmidt, M. L., Kraus, M. W., & Keltner, D. (2014). Signaling social class online.
Manuscript in preparation.
Sapolsky, R. M. (2004). Social status and health in humans and other animals. Annual Review of
Anthropology, 33, 393-418.
Saxbe, D., & Repetti, R. L. (2010). For better or worse? Coregulation of couples' cortisol levels
and mood states. Journal of personality and social psychology,98(1), 92-103.
Sidanius, J., Pratto, F., & Bobo, L. (1996). Racism, conservatism, affirmative action, and
intellectual sophistication: A matter of principled conservatism or group
dominance?. Journal of Personality and Social Psychology,70(3), 476-490.
Stephens, N. M., Markus, H. R., & Fryberg, S. A. (2012). Social class disparities in health and
education: Reducing inequality by applying a sociocultural self model of behavior.
Psychological Review,119, 723-744.
Twenge, J. M., & Campbell, W. K. (2002). Self-esteem and socioeconomic status: A meta-
analytic review. Personality and Social Psychology Review,6(1), 59-71.
Sartorial Symbols of Social Class 36
Waters, S. F., West, T. V., & Mendes, W. B. (2014). Stress contagion: Physiological covariation
between mothers and infants. Psychological Science. Advanced online publication.
Weininger, E. B. & Lareau, A. (2009). Paradoxical Pathways: An Ethnographic
Extension of Kohn’s Findings on Class and Childrearing. Journal of Marriage and
Family, 71, 680-695.
Zaki, J., Bolger, N., & Ochsner, K. (2008). It takes two the interpersonal nature of empathic
accuracy. Psychological Science,19(4), 399-404.
Sartorial Symbols of Social Class 37
1The alternative analyses using the ANOVA method are reported below: In the analysis
of participant concessions offered during the negotiation, the ANOVA revealed a significant
main effect of target vs. perceiver concessions F(1,53)=4.08, p<.05, with targets offering more
concessions than perceivers. There was also a significant main effect of social class
F(2,53)=5.79, p<.05, with lower-class dyads showing more concessions than neutral or upper -
class dyads. The interaction was not significant F(2,53)=1.10, p=.34. These results indicate that
concessions were more likely in the lower-class dyads, and that it was particularly likely among
targets that had changed clothing. In the analysis of testosterone change, the ANOVA revealed
no main effects of social class F(2,61)<1 or target vs. perceiver testosterone change
F(1,61)=2.11, p=.15. However, the analysis did yield a significant interaction F(2,61)=3.91,
p<.05. Examination of the 95% confidence interval surrounding the means revealed that targe ts
(95% CI [-31.79pg/ml to -8.51pg/ml]) had significantly reduced testosterone relative to
perceivers (95% CI [-7.89pg/ml to 29.77pg/ml]) in the lower-class condition, whereas no such
differences in testosterone emerged for the neutral or upper-class dyads. For HRV reactivity, the
Analysis of Covariance (controlling for respiration rate) revealed no main effects for social class
symbols F(2,57)<1, target v. perceiver HRV reactivity F(1,57)<1, or a significant interaction
F(2,57)=2.59, p=.08. Examination of 95% confidence intervals surrounding the means revealed
that in upper-class dyads, targets (95% CI [-0.33 to 0.66]) did not exhibit significantly higher
HRV reactivity than did perceivers (95% CI [-1.31 to -0.07]), although the pattern was in line
with our predictions. No such differences emerged in the neutral or lower -class dyads. For the
affect contagion analysis, no main effects for social class F(2,60)<1, or target v. perceiver affect
contagion emerged F(1,60)<1. The interaction was also not significant F(2,60)=1.75, p=.18.
Sartorial Symbols of Social Class 38
Examination of 95% confidence intervals with this analytic strategy revealed no mean
differences across conditions.
2We also examined correlations between partner responses on each of the critical
outcome measures. Partner profits were significantly negatively associated r(60) = -.91, p< .05,
as were power and profits r(57) = -.36, p< .05—the latter indicating that as one participant felt
more powerful, their partner tended to earn fewer profits. Partner concessions were positively
correlated r(54) = .26, p= .05, although this latter result did not reach conventional levels of
statistical significance. No other significant correlations were observed between partner outcome
3It is possible that the behavioral and physiological responses observed in the experiment
were driven by a willingness of participants to act in a manner that is consistent with their style
of dress, or that dressing in a business suit increased fluency on a negotiation task where this
attire is more appropriate (e.g., Oyserman, 2011). A few analyses we conducted suggest that
fluency is unlikely to be a mechanism explaining the effects of social class symbols on behavior
and physiology: When we examined self-reports of how challenging participants felt the
negotiation was using a 6-item scale (e.g., “The previous task was very demanding.” M= 4.64,
SD = 0.86, ɑ= .62) we found no significant influence of target [t(122.71) = 1.07, ns] or perceiver
[t(122.71) = -0.98, ns] status on challenge judgments. A single item measure of self-rated
performance yielded similar results (i.e., “I think I performed well during the negotiation task.”
M= 5.09, SD = 1.34). Participants wearing the business suit also did not enjoy the task more
than other participants (i.e., “I enjoyed the negotiation task.” M= 5.94, SD = 1.10) [t’s < 1].
Examining physiological responses reveals a similar pattern: If participants in the high class
symbols condition felt more like they were dressing the part in the negotiation, one might expect
Sartorial Symbols of Social Class 39
these participants to exhibit physiological responses suggesting their comfort with meeting the
demands of the negotiation task—indexed by increases in cardiac output from baseline (M=
0.33, SD = 0.99; Blascovich & Mendes, 2008). Contrary to this prediction, targets show no
condition differences in cardiac output during the negotiation [t< 1]. Overall, these results cast
doubt on the notion that participants in the upper- class symbols condition felt more comfortable
than their lower-class symbols or neutral counterparts.
Sartorial Symbols of Social Class 40
Table 1. Mean differences in hierarchical linear model analyses comparing upper-, neutral, and lower-class
symbols of targets and perceivers on key outcome variables. Numbers in parenthesis indicate standard errors of
the mean. Significant (p < .05) mean differences between upper- and lower-class targets and perceivers are
indicated using different subscript letters.
T = testosterone; HRV = heart rate variability; Δ = change from baseline; PEP = pre-ejection period
+Analysis reports results controlling for target and perceiver respiration rates.
Sartorial Symbols of Social Class 41
Table 2. Raw correlations between behavioral and physiological responses during the
3. T change
5. Δ Power
6. Average r (PEP)
* p < .05, +p < .10; T = testosterone; HRV = heart rate variability; PEP = pre-ejection period;
Δ Power = change in power from baseline; aAnalysis reports results controlling for respiration
Sartorial Symbols of Social Class 42
Figure 1. Lower-class (left panel) and upper-class (right panel) clothing used in the sartorial
manipulation. Photos of M. Halverson taken with permission by M. W. Kraus.