Fact or Fiction? A Mixed-Method Investigation of the

Fact or Fiction? A Mixed-Method Investigation of the
Hunchback Hypothesis of Emotional Expressivity among
Members of Low Status Groups
Chuma Owuamalam1; Mishaari Weerabangsa1; Minnallah Mamdouh1; Jaya Kumar
Karunagharan1; Mark Rubin2
University of Nottingham, Malaysia Campus1
University of Newcastle, Australia2
A lay belief in parts of Western Africa is that “mkpu-mkpu azu” (an Igbo expression for
a hunched back) is an accumulation of negative experiences that one has encountered in a
previous life. This burden of negativity is thought to make individuals with a hunched back
more prone to emotional outbursts (e.g., anger). Surprising, as it might seem, this lay belief is
not far removed from a widespread assumption in many societies: that members of
disadvantaged groups are more emotionally ‘leaky’ than their higher status counterparts. In
the West, for example, members of disadvantaged communities are thought to hold a grudge
over the negative experiences they face (i.e., “chip on the shoulder”) and are portrayed as too
emotional and violent when they seek redress to social inequity (Owuamalam, Issmer,
Zagefka, Klaβen, & Wegner, 2013). Members of high status groups, on the other hand, are
generally portrayed as calm and civilized - relying more on intellect than raw aggression
(Loughnan, Haslam, Sutton, & Spencer, 2013). We refer to this tendency to attribute greater
emotional displays to members of low status groups than to their higher status counterparts as
the hunchback stereotype. An aim of this investigation is to test the existence of this
hunchback stereotype and to then examine how accurately it matches reality.
Relevance and Impact
The question of whether or not members of privileged and underprivileged groups are
more/less expressive of excitable emotional states (e.g., anger) is important, particularly
because of the societal sanction against the expression of such emotions. Previous research
has shown that people who express anger and frustration generally face negative evaluation
and social isolation (Kaiser & Miller, 2001), and this can have far-reaching consequences for
members of underprivileged social groups not only in terms of their psychological wellbeing, but also in terms of their life outcomes. Within the judicial system, for example, it is
conceivable that activating the hunchback stereotype may cloud jury decisions and this could
cause disproportionately higher conviction rates for low (rather than high) status members on
violent crimes. Similarly, activating the hunchback stereotype can result in greater numbers
of arrests or, in extreme cases, may result in fatalities during police apprehension of suspects
from disadvantaged communities (e.g., the recent police shooting of an unarmed Black
teenager in Missouri, USA - CNN, 2014). Despite these potentially deleterious outcomes, we
are far from understanding how these ‘emotional’ stereotypes come about in the first place
and whether or not they are based on facts or fiction.
Previous Research
The evidence so far, for the existence of the hunchback stereotype is mixed. Some have
shown that people expect members of low status groups to display less anger than their
higher status counterparts (Tiedens, Ellsworth, & Mesquita, 2000). Other correlational
evidence has shown the opposite: that members of low status groups display anger more than
their higher status counterparts (Park et al., 2013). The confusion in the evidence so far often
is because previous studies on status-based emotional displays have relied mainly on: (a)
retrospective accounts of emotional expressivity of high and low status groups members that
may be prone to mis-recollection (e.g., Park et al., 2013); (b) judgments of emotional
expressivity of members of high and low status groups by third parties, and have not directly
tapped the emotional experience of members of low and high status groups in situ, when
exposed to the same emotion arousing conditions (e.g., in Plant, Hyde, Keltner, & Devine,
2000); and (c) extensive use of self-reports that might itself be vulnerable to existing
stereotypes concerning the emotional expectations of people who are high or low in status.
Hence, in the current investigation we aimed to go beyond the traditional third-party
judgments of emotional displays and self-expressed emotions to also include novel:
psychophysical measures - in terms of judgments of the intensity of the voices of
members of high and low status groups. We reasoned that people’s voices provide signals to
others’ emotional states and the intensity of peoples’ voices might be one physical property
from which judgments about others’ emotional states can be reached.
IAT measure - to uncover the implicit manifestations of emotional stereotyping of
members of high and low status groups.
Theoretical Underpinnings of this Investigation.
Extant theoretical perspectives agree that underprivileged groups are more emotionally
leaky due to the chronic frustration they face in their daily lives (cf. Berkowitz, 1984). In
contrast, their privileged counterparts are presumed to be calm, even in the face of adversity,
and this is thought to be a state that allows their supposed competence to manifest (Fiske,
Cuddy, Glick, & Xu, 2002). Also in line with the hunchback hypothesis, Keltner, Gruenfeld,
and Anderson’s (2003) power, approach and inhibition theory of emotional displays assumes
that high status is associated with the experience and expression of ‘positive’ emotions (e.g.,
calm) and low status is associated with the experience and expression of ‘negative’ emotions
(e.g., anger). However, Keltner et al. (2003) also proposed that members of low status groups
are generally more inhibited in their behavior - although, it is somewhat unclear under this
framework how the ‘showy’ negative emotion of anger is observed in low status individuals
when such emotions are presumably suppressed.
Is the hunchback hypothesis a fact or fiction?
An anti-thesis to the hunchback stereotype is that it is a common social myth that has no
basis in fact. In line with the social identity theory (SIT; Tajfel & Turner, 1986), we reasoned
that members of underprivileged groups have the strongest motivation to disconfirm negative
ingroup stigma surrounding emotional outbursts, and should deliberately suppress their
display of anger particularly before an audience comprised of higher status group members.
We further reasoned that members of privileged groups may not be as constrained by such
emotional stigma in the manner that their underprivileged counterparts are and, as a result,
may be more disinhibited in their emotional expressions than the latter, provided the emotion
trigger is perceived to be illegitimate and the source of this trigger is relatively inferior to
their ingroup. That is, high status members should have the greatest motivation to re-assert
their superiority when this is illegitimately challenged, by displaying the very emotion known
to convey authority – i.e., anger (Tiedens, 2001).
Rationale For the Proposed Replication
In a previous study we showed that people generally perceive members of low status
groups to be more expressive of anger and less of calm compared to their higher status
counterparts. In that study we operationalised status in terms of race and perceivers were
asked to rate the extent to which they expected either a low status (African targets) or high
status (Caucasians targets) to be expressive of anger and calm. Although the results
confirmed our theorising in terms of the hunchback stereotype, the specific racial context we
used may be conceived by some as having pre-existing stereotypes regarding emotional
expressions. That is, Black Africans are generally perceived to be more aggressive than
Caucasians and this racial bias introduces another explanation to the patterns we found, over
and above the theorised hunchback effect. For this reason, we plan to replicate the same
experiment using the same racial context, but this time introducing another status variable weight. Weight is generally negatively regarded in most societies and people often attach low
social status to those who are ‘large’ compared to those who are thin. Note that individuals
who are large generally score highly on warmth (Durante et al., 2014) and this physical
feature provides a stringent test of the hunchback stereotype rather than a biased test of it,
because the positive warmth perceptions should work against a negative anger perception.
That is, if the hunchback stereotype is true, then beyond an effect of status based on race, one
should also see a similar effect based on size – such that large people of any of the two racial
groups would be expected to be (a) more expressive of anger, and (b) less expressive of calm,
compared to thin targets. For applied reasons, we would also include clothing type (formal
vs. casual) to explore whether or not members of low status group (race and size) are
buffered against the hunchback stereotype if they are dressed in clothing that is presumed to
accord high status in society (formal attire).
The experiment will consist of two parts. Part one will utilize an adapted computer-based
classic IAT paradigm - designed to replicate the effect of status that is based on race and to
extend this by including another status dimension based on attire. Part two of the
investigation will be an extended computer-based self-report questionnaire similar to the
previous study we conducted – except that this time participants will be exposed to carefully
selected targets of either Caucasian or African decent that varied in size (large vs. thin). The
IAT aspect of the experiment will precede the explicit self-reported aspect in order to reduce
familiarity effects in the IAT. For this reason too, we will be introducing the size
manipulation in the latter explicit tasks but not the IAT.
Part One
The IAT will use a 2 (Racial status: African vs. Caucasian) x 3(Attire: formal [high] vs.
‘street hoodie’ [low] vs. plain t-shirt [neutral condition]) x 2(Attribute type: consistent vs.
inconsistent word-category pairing). Attire-based status will be manipulated as a betweensubjects factor by creating three versions of the task (for high status, low status and neutral
status respectively), while the racial status will be manipulated as a within-subject factor. The
attributes will be ‘anger’ and ‘calm’, and these words will be paired with either an African vs
Caucasian male target. In the attribute consistent condition, pictures of a black target paired
with anger, and white target paired with calm will be presented at the top left and top right
hand of the screen while participants are responding to a probe word. In the attribute
inconsistent condition, pictures of a black target will be paired with calm, and white target
will be paired with anger. The probe word will be words associated with anger (e.g.,
explosive) or calm (e.g., silent). The dependent variable will be participants’ reaction latency
– time taken to decide whether or not a word type (e.g., explosive) was consistent with either
anger or calm.
Hypotheses Part One
If the hunchback stereotype is true, then one should see an interference effect in the
inconsistent condition such that participants are taking more time to decide that the word
(e.g., explosive) matches the emotion compared to the consistent condition. This effect
should be evident in the race-based status and also for attire-based status contexts.
Part Two
The self-report aspect of the experiment will use a 2(Racial status: White [high] vs. Black
[low]) ×2(Size: large vs. thin) x 3(Attire: formal [high] vs. street hoodie [low] vs. plain t-shirt
[neutral]) mixed design. The dependent variables will be participants’ responses to items
measuring perceived anger and calm of the targets in the pictures they will be exposed to. We
will also include known correlates of anger (aggressiveness, and dangerous) to capture threat
perceptions, and then measures of attractiveness to statistically control for ‘halo’ effects. To
maximize statistical sensitivity and to enhance test reliability we will have 24 trials (6 large
blacks, 6 thin blacks, vs. 6 large whites and 6 thin whites) crossed by attire.
Hypothesis Part Two
Assuming the hunchback hypothesis is robust, and not dependent on the specific racial
status used in our previous investigation, then one should expect people to overestimate the
anger ratings of low status groups (Black, large people, and street attire) compared to high
status people (Whites, thin people, and formal attire).
Participants and Sample Size Estimate
We determined that a mixed ANOVA with 12 cells, an alpha level of .05, a power value of
.95, an effect size of f=.20, and a correlation of within-subject measures of r=.16 (Cafri et al.,
2010) and 24 trials, would require 84 subjects to power it using the G*Power calculator (Faul
et al.’s (2007). We will round this number up to 100 to account for data exclusion that might
arise due to non-completion of both parts of the experimental protocol. Participants will be
recruited from the psychology undergraduates programme at the lead author’s institution and
will be compensated for their participation in the form of course credit (for first year recruits)
or a payment of 5 Malaysian Ringgit in lieu – for roughly half hour worth of their time.
Stimuli and Apparatus
Face stimuli will be selected from an existing bank: the Park Aging Mind Laboratory Face
Database (8 expression-neutral male faces, ranging in age from 18 to 30, 4 African descents,
4 Caucasian/European). These faces will be matched for age and attractiveness. These images
will each then be photo-edited into formal attire, ‘street’ clothes, and plain black t-shirts, and
cropped to a uniform size and a uniform background. Attribute-associated word stimuli will
be chosen from The University of South Florida free association, rhyme, and word fragment
norms (Nelson, McEvoy, & Schreiber, 2004).
Participants will be randomly allocated to one of the three conditions (Formal/high status
attire, ‘Street’/low status attire, Plain t-shirt/neutral status attire). Procedure from this point
onwards will be identical for all conditions.
Part One
Participants will first read basic instructions about the task before completing a consent
form. The task will consist of eight blocks of trials, each prefaced by an instruction screen
detailing what is required of the participant in each block. Participants are instructed to
respond as quickly and as accurately as possible.
Blocks 1 and 5 will be FACE blocks, where participants will be required to match names
to faces. A face will be displayed at each upper corner of the screen, one African, one
Caucasian/European. The position of each face (right or left) is counterbalanced between the
two blocks. In the centre of the screen, a fixation point will be displayed for 250 ms, after
which a human name (African, Caucasian/European, or Cantonese) will be displayed.
Participants must match the name to a face by pressing the left or the right arrow key, to
match to the face on the left or the face on the right respectively – they will be allowed
unlimited time to make this response.
Blocks 2 and 6 will be EMOTION blocks, where participants will be required to match
words to emotions. An emotion word will be displayed at each upper corner of the screen,
one reading ANGER, the other reading CALM. The position of each word (right or left) is
counterbalanced between the two blocks. In the centre of the screen, a fixation point will be
displayed for 250 ms, after which a noun or verb (anger-associated, calm-associated, or nonassociated) will be displayed. Participants must match the word to an emotion by pressing the
left or the right arrow key as before – they are allowed unlimited time to make this response.
Blocks 3, 4, 7 and 8 will be COMBINED blocks. In these blocks, each upper corner
displays one of the two faces previously encountered, paired with one of the emotion words
previously encountered. The position and pairing of words and faces is counterbalanced
across the four blocks. As before, a fixation point will be displayed at the centre of the screen
for 250 ms, after which participants will be presented with either a word stimulus (noun or
verb) or a name stimulus. Participants must pair the name stimuli with the appropriate face,
and the word stimuli with the appropriate emotion, using the left and right arrow keys as
before. They are allowed unlimited time to make this response.
Block 1, 2, 5, and 6 are in effect practice trials – only response times from the
COMBINED blocks (3, 4, 7, and 8) will be analysed. Once participants have completed all 8
blocks a screen instructing them to seek further instruction from the experimenter will be
displayed, at which point the experimenter will direct them to begin Part Two of the
Part Two
Participants will then proceed to the main questionnaire, where they will be required to
rate 8 expression-neutral male faces, 4 of African descent and 4 of Caucasian/European
descent, on perceived aggressiveness, angriness, calmness, attractiveness and social status
using a 7-point Likert scale. Each face will be rated on each of the six scales once. When
participants have completed both parts of the study, they will be debriefed and compensated
for their participation.
ANOVAs will be performed on the RTs from the IAT (Part One) and also the self-report
measures of emotional stereotyping (Part Two).
Summary of Studies to be Presented in Manuscript
Study One
Initial study testing the hnchback stereotype – in the context of race-based status.
Study Two
The proposed replication of the hunchback stereotype – mixed method using IAT and
explicit self-reports of 3 status contexts – race (same as in Study 1), size and attire.
Study Three
Both Study 1 and the proposed replication (Study 2) examine the ‘existence’ of the
hunchback stereotype. In this already conductd study we used novel acoustic signaling
techniques to examine the ‘reality’ and social constraints of the hunchback stereotyp. Details
are presented in the supplementary material.
Ethics approval to conduct this has already been secured from the Faculty of Science
Ethics Committee, University of Nottingham, Malaysia Campus (UNMC).
Expertise of Project Group Members
The proposed replications are already underway within the newly formed Social and
Affective Neuroscience Group (SANG) here at UNMC. In terms of expertise of the research
team, Dr. Chuma Owuamalam is an experimental social psychologist and has published work
on group dynamics. Chuma is overseeing the theoretical and design aspects of the work. Dr.
Mark Rubin provides additional theoretical support for our work, having published
extensively on social identity processes. The other named authors are students in our
laboratory who are either working on the project as part of their dissertation or involved in a
learning capacity and helping out with setting-up the experimental protocols and running the
experiments. Further details about SANG, including a week-by-week account of what we
have done so far can be seen on our SANG Facebook page – We are happy to provide access
to this exclusive page for review purposes. However, please note that access will require a
Facebook account. We will grant access when we receive this detail.
We are grateful to EASP for supporting our research with a Seedcorn grant to the first
author. We are also grateful to UNMC’s Faculty of Science Research Committee for
supporting this program of research with a top-up funding to the first author.
Cafri, G., Kromrey, J. D., Brannick, M. T. (2010). A meta-meta-analysis: Empirical
review of statistical power, type I error rates, effect sizes, and model selection of metaanalyses published in psychology. Multivariate Behavioral Research, 45, 239-270.
CNN News Report (August, 2014). Missouri governor imposes curfew in Ferguson,
Durante, F., Fasolo, M., Mari, S., & Mazzola, A.F. (2014). Children's Attitudes and
Stereotype Content Toward Thin, Average-Weight, and Overweight Peers. SAGE Open, 4(2).
Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G*Power 3: A flexible
statistical power analysis program for the social, behavioral, and biomedical sciences.
Behavior Research Methods, 39, 175-191.
Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of (often mixed)
stereotype content: Competence and warmth respectively follow from perceived status and
competition. Journal of Personality and Social Psychology, 82, 878–902.
Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition.
Psychological Review, 110, 265-84.
Loughnan, S., Haslam, N., Sutton, R. M., & Spencer, B. (2013). Dehumanization and
social class: animality in the stereotypes of “White Trash,” “Chavs,” and “Bogans”. Social
Psychology (early view).
Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University of South Florida
word association, rhyme, and word fragment norms. Retrieved 20th December 2014, from
Owuamalam, C., Issmer, C., Zagefka, H., Klaßen, M. & Wagner, U. (2014). Why do
members of disadvantaged groups strike back at perceived negativity towards the ingroup?
Journal of Community & Applied Social Psychology, 24, 249–264.
Park, J., Kitayama, S., Markus, H. R., Coe, C. L., Miyamoto, Y., Karasawa, M., [...] &
Ryff, C. D. (2013). Social status and anger expression: The cultural moderation hypothesis.
Emotion (early view).
Plant, A., Hyde, J. S., Keltner, D., & Devine, P. D. (2000). The gender stereotyping of
emotions. Psychology of Women Quarterly, 24, 80-91.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behaviour. In S.
Worchel, & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago:
Nelson Hall.
Tiedens, L. Z., Ellsworth, P. C., & Mesquita, B. (2000). Sentimental stereotypes:
Emotional expectations for high-and low-status group members. Personality and Social
Psychology Bulletin, 26, 560-575.