Content uploaded by Nour S. Kteily
Author content
All content in this area was uploaded by Nour S. Kteily on Dec 13, 2015
Content may be subject to copyright.
They See Us As Less Than Human: Meta-Dehumanization Predicts Intergroup
Conflict Via Reciprocal Dehumanization
Nour Kteilya, Gordon Hodsonb, Emile Bruneauc
la Department of Management and Organizations, Northwestern University, Evanston, IL
60208
b Department of Psychology, Brock University, St Catharines, ON, Canada L2S 3A1
c Annenberg School for Communication, University of Pennsylvania, Philadelphia, PA
19104
Corresponding Author:
Nour Kteily (n-kteily@kellogg.northwestern.edu)
Title page with All Author Information
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
1
Abstract
Although the act of dehumanizing an outgroup is a pervasive and potent intergroup
process that drives discrimination and conflict, no formal research has examined the
consequences of being dehumanized by an outgroup – i.e. ‘meta-dehumanization’. Across
ten studies (N = 3,440) involving several real-world conflicts spanning three continents,
we provide the first empirical evidence that meta-dehumanization (a) plays a central role
in outgroup aggression that is (b) mediated by outgroup dehumanization, and (c) distinct
from meta-prejudice. Studies 1a and 1b demonstrate experimentally that Americans
receiving information that Arabs (Study 1a) or Muslims (Study 1b) blatantly dehumanize
Americans are more likely to dehumanize that outgroup in return; by contrast,
experimentally increasing outgroup dehumanization did not increase meta-
dehumanization (Study 1c). Using correlational data, Study 2 documents indirect effects
of meta-dehumanization on Americans’ support for aggressive policies towards Arabs
(e.g., torture) via Arab dehumanization. In the context of Hungarians and ethnic minority
Roma, Study 3 shows that the pathway for Hungarians from meta-dehumanization to
aggression through outgroup dehumanization holds controlling for outgroup prejudice.
Study 4 examines Israelis’ meta-perceptions with respect to Palestinians, showing that:
(a) feeling dehumanized (i.e., meta-dehumanization) is distinct from feeling disliked (i.e.,
meta-prejudice), and (b) meta-dehumanization uniquely influences aggression through
outgroup dehumanization, controlling for meta-prejudice. Studies 5a and 5b explore
Americans’ meta-perceptions regarding ISIS and Iran. We document a dehumanization-
specific pathway from meta-dehumanization to aggressive attitudes and behavior that is
distinct from the path from meta-prejudice through prejudice to aggression. In Study 6,
Masked Manuscript without Author Information
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
2
American participants learning that Muslims humanize Americans (i.e., meta-
humanization) humanize Muslims in turn. Finally, Study 7 experimentally contrasts
meta-dehumanization and meta-humanization primes, and shows that resulting
differences in outgroup dehumanization are mediated by (1) a general desire to
reciprocate the outgroup’s perceptions of the ingroup, and (2) perceived identity threat. In
sum, our research outlines how and why meta-dehumanization contributes to cycles of
ongoing violence and animosity, thus providing direction for future research and policy.
Keywords: Dehumanization; Meta-Dehumanization; Conflict Resolution; Intergroup
Relations; Meta-perceptions; Stigma; Prejudice
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
3
They See Us As Less Than Human: Meta-Dehumanization Predicts Intergroup
Conflict Via Reciprocal Dehumanization
In early September 2012, a 14-minute video titled Innocence of Muslims, dubbed
in Arabic, appeared on YouTube. It soon emerged that this video had been shot in the
United States by an American resident of Egyptian origin. The video depicted the Prophet
Muhammad, an act viewed throughout the Muslim world as an offence. Worse, the
Prophet was portrayed not just in a negative light, but specifically in animalistic terms: as
an incompetent buffoon, a hyper-sexual womanizer lacking in self-control, and a
“bloodthirsty” leader of a savage, “ragtag group of men who enjoy killing” (BBC News,
2012). The response to this depiction was swift and violent. Within days of the film’s
release, anti-American protests erupted in Egypt and quickly spread to American
embassies in a number of cities around the world, leaving scores dead and hundreds of
others injured, and causing outrage among the American populace (BBC News, 2012).
Capturing the mood at the time, Rifaei Taha, a leader of a political party in Egypt,
reciprocated the dehumanization he perceived, calling on then-President Mohammad
Morsi to “cut relations with those [American] monkeys and pigs” (CBS News, 2012).
How might the more powerful group (here, Americans), react to being dehumanized by
the lower status ‘other’? Might such meta-perceptions predict increasingly hostile
intergroup responses, including reciprocal dehumanization and collective aggression, thus
contributing to the vicious cycles that have marred so many longstanding conflicts?
We sought to specifically explore the effects of this perception that one’s own
group is perceived by another as less than fully human – i.e., ‘meta-dehumanization’. A
large and growing literature has detailed the pervasiveness of dehumanization, its
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
4
uniqueness from prejudice (e.g., Andrighetto, Baldissarri, Lattanzio, Loughnan, &
Volpato, 2014; Haslam, 2006; Leyens et al., 2000, 2007) and its potency in legitimizing
intergroup aggression (Kteily, Bruneau, Waytz, & Cotterill, 2015; Struch & Schwarz,
1989; Viki, Osgood, & Phillips, 2013; see also Bandura, Barbaranelli, Caprara, &
Pastorelli, 1996). At the same time, despite the fact that negative meta-perceptions more
broadly construed (e.g., meta-stereotypes; meta-prejudice) are known to have damaging
effects on intergroup relations (Frey & Tropp, 2006; Kamans et al., 2009; Owuamalam,
Issmer, Zagefka, Klaben, & Wagner, 2014; Vorauer, Main, & O’Connell, 1998), no prior
work has examined how individuals respond to the (meta)perception that their group is
dehumanized. If blatant dehumanization is such a strong and unique predictor of negative
intergroup outcomes, might meta-dehumanization also galvanize hostile intergroup
processes over and above other negative meta-perceptions? Focusing on intensive real-
world intergroup conflicts such as the Israeli-Palestinian conflict and American-Muslim
relations, the present work considers the effects of perceiving that one’s group is not just
disliked, but actively and blatantly dehumanized by the outgroup.
The central hypothesis of this work is that perceiving that one’s ingroup is
blatantly dehumanized by an outgroup can provoke individuals to reciprocate by
dehumanizing that group in turn, thus increasing support for violent and aggressive
collective actions against it. We argue further that this will occur over and above any
effects of (a) meta-perceptions of prejudice, and (b) any outgroup prejudice that meta-
prejudice might provoke. That is, we propose a novel dehumanization-specific path from
meta-dehumanization to outgroup aggression through reciprocal dehumanization.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
5
The question of how group members respond to perceived outgroup
dehumanization of their group is both theoretically important and practically relevant.
Our work combines insights from the largely separate literatures on dehumanization and
meta-perceptions, extending theory in each. We advance the literature on intergroup
dehumanization by considering how meta-perceptions can lead to intergroup hostility
through entrenching dehumanization, and broaden the scope of research on meta-
perceptions by identifying a unique and consequential type of meta-perception.
Practically, if meta-dehumanization indeed promotes outgroup dehumanization and thus
more aggressive outgroup attitudes and behavior, it has the potential to contribute
importantly to vicious cycles of intergroup conflict (e.g., Bar-Tal, 2000; Kelman, 1987).
Therefore, understanding the role of meta-dehumanization, the mechanisms underlying
its effects, and the ways in which it can be attenuated has important implications for the
prospects of intergroup harmony.
Below, we briefly review the existing literatures on each of negative intergroup
meta-perceptions and dehumanization, and then consider how meta-dehumanization
might uniquely influence outgroup attitudes and behavior.
Responses to Intergroup Meta-Perceptions
Vorauer et al. (1998) introduced the term meta-stereotyping to describe the
content of individuals’ cognitions about how they are perceived by an outgroup. This
research focused on the dominant group’s (e.g., White Canadians) sense of how they are
perceived by a subordinate group (e.g., Aboriginal Canadians) in the context of a cross-
group interpersonal encounter. These authors showed that White Canadians expected
Aboriginal Canadians to stereotype them with a mixture of positive and negative traits: as
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
6
high status and ambitious, but also as egocentric, unfeeling, selfish, and prejudiced.
Importantly, the more White Canadians expected to be stereotyped, the less they
anticipated enjoying contact with an outgroup member, and the more prejudice they
exhibited towards the outgroup. Moreover, in an actual interaction with an outgroup
member, feeling meta-stereotyped was associated with threat to individuals’ self-concept.
This research illustrated the potency of meta-perceptions by showing that they played an
even bigger role in affecting the intergroup interaction than the stereotypes participants
themselves held about the outgroup (e.g., seeing Aboriginal Canadians as lazy; see also
Vorauer & Kumhyr, 2001). Consistent with this perspective, other research has
documented the aversive effects of engaging in or anticipating intergroup encounters,
attributable in part to expectations of (and anxiety about) being negatively evaluated by
an outgroup individual (Richeson & Shelton, 2007; Vorauer, Hunter, Main, & Roy, 2000;
see also MacInnis & Hodson, 2012).
Expecting to be viewed negatively by another group can lead not only to
avoidance of and discomfort with intergroup interactions, but also to increased intergroup
bias (Paolini, Hewstone, Voci, Harwood, & Cairns, 2006) and hostility (Kamans et al.,
2009). Why might individuals respond to negative meta-perceptions by reciprocating that
negativity in kind? Research on social identity threat (Branscombe, Ellemers, Spears, &
Doosje, 1999) suggests that because individuals derive esteem from their membership in
groups (Tajfel & Turner, 1986), they experience devaluation of their group as an aversive
threat that they seek to remedy (see also Hornsey, 2008). One means by which
individuals might restore ingroup status is by derogating the outgroup responsible for the
status threat (e.g., Bourhis, Giles, Leyens, & Tajfel, 1979; Branscombe, Schmitt, &
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
7
Harvey, 1999; Branscombe & Wann, 1994), especially when the status threat is perceived
as illegitimate or offensive, and thus rejected out of hand (Ellemers, Wilke, & Van
Knippenberg, 1993; Vorauer & Sakamoto, 2008; see also Jetten, Schmitt, & Branscombe,
2013).
Indeed, there is good evidence that individuals reciprocate negative evaluations
they perceive from others, and become more willing to endorse hostile behavior towards
them (Bourhis et al., 1979; Devine, Evett, & Vasquez-Suson, 1996; Kamans et al., 2009).
For example, Belgian Flemish speakers experimentally exposed (vs. not exposed) to a
French-speaking Belgian confederate insulting the Flemish language were more likely to
retaliate by directing obscenities at the offender (Bourhis et al., 1979). Moroccans in the
Netherlands who harbored negative attitudes towards the majority-Dutch and expected
them to perceive Moroccans negatively (e.g., as ‘criminal’ or ‘aggressive’) were more
likely to endorse aggressive behavior against them (Kamans et al., 2009). Other research
has also shown that individuals who activate negative group meta-perceptions are more
likely to respond with anger and reciprocal negativity towards the offending outgroup
(Owuamalam, Tarrant, Farrow, & Zagefka, 2013), and to support collective action to
seek redress (Owuamalam et al., 2014). Finally, examining the Eurovision song contest,
Doosje and A. Haslam (2005) showed that reciprocation of outgroup actions and
perceptions extends to behavior: members of European nations punished nations that had
voted for the ingroup less in previous years by voting for them less; they also rewarded
nations that had previously favored the ingroup with more votes.
In sum, people think about how they are perceived by other groups, and these
meta-perceptions are frequently negative in content (Frey & Tropp, 2006). Because they
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
8
impact the ingroup’s social identity, negative meta-perceptions are experienced as
aversive and threatening. This promotes desires to restore ingroup status, a goal that can
be achieved by derogating the offending outgroup in kind. Consistent with this,
individuals who perceive that their group is viewed negatively oftentimes reciprocate:
they respond with anger, hostile outgroup evaluations, and even support for aggression
and collective action against the outgroup, particularly when the meta-perception is
viewed as inaccurate and offensive. Despite this prior research, no work has examined
the (meta)-perception that another group perceives the ingroup as less than fully human.
As we develop further below, perceiving that another group blatantly dehumanizes the
ingroup represents a stark and harsh devaluation of the ingroup that we hypothesize
would be viewed as particularly offensive, and is likely to be reciprocated in kind.
Intergroup Dehumanization
In parallel to work on negative meta-perceptions, a growing body of research has
examined dehumanization and its effects on intergroup (and interpersonal) relations. This
research has examined both blatant and subtle forms of dehumanization. Early research
focused on the role of blatant dehumanization in licensing aggression. For example,
Bandura, Underwood, and Fromson (1975) showed that participants who ‘mistakenly’
overheard an experimenter describe subjects using dehumanizing (vs. humanizing, or
neutral) language provided more intense shocks to these (purported) participants when
they erred (see also Struch & Schwarz, 1989). The facilitating effect of blatant
dehumanization on aggression was explained by the fact that dehumanization places the
target outside the scope of moral consideration (see also Kelman, 1976; Opotow, 1990).
Speaking to the unique power of dehumanization, Kelman (1976) noted that fear or
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
9
intense dislike do not preclude seeing an enemy as a human, and indeed may even afford
that enemy a certain level of respect; in contrast, when another is viewed as less than
human moral restraints are removed and violence is condoned (or even encouraged).
Contemporary research has additionally explored the variety of indirect ways in
which we engage in ‘everyday’ dehumanization. Whereas blatant dehumanization
involves the active and deliberate denial of others’ humanity, subtle dehumanization may
occur even outside conscious awareness (Haslam, 2014; Leyens et al., 2000; Leyens,
Demoulin, Vaes, Gaunt, & Paladino, 2007). The process of viewing athletes as statistics,
women as objects, inmates as numbers, and sick individuals as ‘patients’ may involve
attributing them fewer traits unique to humans and central to our nature, reflecting a
subtle denial of what it means to be fully human (e.g., Haslam, 2006; Waytz, Schroeder,
& Epley, 2014; see also Bain, Vaes, & Leyens, 2014). In the intergroup domain,
pioneering work on infrahumanization (Leyens et al., 2000) showed that individuals
attribute complex secondary emotions (more closely associated with humans than
animals) to members of their own group than an outgroup (see also Demoulin et al.,
2004). Building on this work, Haslam (2006) posited two bases of dehumanization:
Animalistic dehumanization, akin to infrahumanization, involves the relative under-
attribution to others of characteristics (e.g., cognitive aptitude, refinement, civility)
considered unique to humans; mechanistic dehumanization, on the other hand, involves
denying others traits (e.g., warmth, emotionality) that, while central to being human, may
also be shared with animals (see Haslam, Bain, Douge, Lee, & Bastian, 2005).
Importantly, by showing that individuals preferentially attribute both positive and
negative secondary emotions (and characteristics) to the ingroup, the research on
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
10
infrahumanization and animalistic/mechanistic dehumanization differentiates these
phenomena from simple outgroup negativity (Albarello & Rubini, 2012; Andrighetto et
al., 2014; Leyens et al., 2000).
A number of studies have now illustrated the intergroup consequences of subtly
dehumanizing outgroups and shown that they can occur in parallel with (or in addition to)
outgroup prejudice. For example, research has shown that infrahumanization is associated
with decreased helping intentions after Hurricane Katrina (Cuddy, Rock, & Norton,
2007), and that it reduces acceptance of responsibility for past ingroup wrongdoings
(Castano & Giner-Sorolla, 2006). Similarly, Andrighetto and colleagues (2014) showed
that Italians’ unwillingness to help outgroup members (Japanese and Haitians) after a
natural disaster is predicted by their animalistic (Haitians) and mechanistic (Japanese)
dehumanization of these groups, controlling statistically for outgroup prejudice.
Although this past dehumanization research indicates that both blatant and subtle
dehumanization are relevant to intergroup processes and distinct from prejudice, these
forms of dehumanization had not been directly contrasted until very recently. Kteily et al.
(2015) provided the first formal comparison of the effects of blatant dehumanization
relative to subtle dehumanization (i.e., infrahumanization; animalistic/mechanistic
dehumanization) on intergroup attitudes and behavior, examining contexts marked by
open hostility and intergroup conflict (e.g., American-ISIS relations). Using their newly
developed ‘Ascent scale of blatant dehumanization’, these authors argued that blatant
dehumanization—so central to past intergroup conflicts— remains a feature of
contemporary society. Across a range of contexts they showed that blatant (vs. subtle)
dehumanization is a stronger predictor of extreme intergroup attitudes (such as support
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
11
for torture, and drone strikes) and behavior (such as signing online petitions in support of
militarism). The effects of blatant dehumanization held controlling for outgroup
prejudice, confirming a distinction between dehumanization and dislike (see also Goff,
Eberhardt, Williams, & Jackson, 2008; Goff et al., 2014) that also receives neuroimaging
support (Bruneau, Jacoby, Kteily, Ligouri, & Saxe, under review).
In sum, considerable progress has been made in understanding when, why, and
how we dehumanize other targets. This research suggests that conflictual intergroup
contexts continue to be marked by blatant outgroup dehumanization, a potent perception
distinguishable from both subtle dehumanization and outgroup prejudice. Yet the vast
majority of this research has focused on the dehumanization of others (e.g., ‘they are
animal-like’), with little known about the consequences of feeling dehumanized by others
(i.e., ‘they think we are animal-like’), a perception that may well feature in the context of
intergroup conflict. This lacuna is surprising, given both the prevalence of
dehumanization research and the recognized importance of meta-perceptions to
intergroup relations (e.g., Frey & Tropp, 2006; Yzerbyt, Judd, & Muller, 2009).
Integrating Theorizing on Meta-perceptions and Dehumanization
Here we integrate insights from research on negative meta-perceptions and on
blatant dehumanization. Specifically, combining the knowledge that individuals often
reciprocate negativity they perceive from outgroups, and the fact that blatant
dehumanization is distinct from outgroup prejudice, we ask whether feeling dehumanized
(vs. disliked or negatively stereotyped) by another group can uniquely contribute to
intergroup hostilities by increasing outgroup dehumanization. There are a number of
theoretical arguments in favor of this possibility. First, previous research demonstrates
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
12
that outgroup dehumanization is conceptually and empirically distinct from outgroup
prejudice, and also that it is uniquely associated with outgroup aggression (Kteily et al.,
2015). Second, given that (blatant) dehumanization involves extreme devaluation, and
outright exclusion from the moral domain (Kelman, 1976) and the most basic shared
superordinate identity of ‘human’, meta-dehumanization should provoke particularly
strong threats to ingroup identity. In line with theorizing suggesting that individuals
respond in kind to outgroups’ negative perceptions of the ingroup (Branscombe et al.,
1999a), meta-dehumanization may provoke very harsh responses, including reciprocal
dehumanization of— and aggression towards— the offending outgroup.
Some recent research in the interpersonal domain provides support for our notion
that individuals dehumanize those who they perceive to dehumanize them. Bastian and
Haslam (2010) showed that people socially excluded (vs. included) in a Cyberball game
were significantly more likely to report that that target treated them as less human
(Bastian & Haslam, 2010, Study 2). This led participants to dehumanize the ostracizer,
primarily by attributing them fewer traits considered central to human nature, such as
emotionality and warmth (see Haslam, 2006). Relatedly, Bastian and Haslam (2011)
found that participants who were asked to vividly recall a time when they were treated as
irrelevant or unimportant (i.e., mechanistically dehumanized) reported feeling more
numbness, lethargy, and sadness (presumably reflecting the pain of the meta-perception),
but also more anger (presumably towards the perpetrator). Thus, existing research on
interpersonal processes suggests that feeling excluded or undervalued by others can be
painful, and we may reciprocate that negativity.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
13
In the present research, we examine blatant (vs. subtle) forms of meta-
dehumanization, and focus on conflictual intergroup (vs. interpersonal) contexts, which
tend to involve greater competitiveness and aggression (Wildschut, Pinter, Vevea, Insko,
& Schopler, 2003). In this context, we reason that individuals will be highly likely to feel
threatened by the outgroup’s perception and reciprocate any dehumanization they
perceive. Extending prior work, we distinguish for the first time between meta-
dehumanization and other negative meta-perceptions.
Hypotheses
In line with prior research on responses to negative meta-perceptions, we
hypothesized that perceiving that the ingroup is blatantly dehumanized by an outgroup
would predict greater blatant dehumanization of that outgroup (see Figure 1 for a diagram
of our overall conceptual model). Consistent with previous work (Kteily et al., 2015), we
hypothesized that outgroup dehumanization would itself be associated with support for
aggressive intergroup attitudes and behaviors, such as support for torture and collective
aggression, independent of outgroup prejudice (i.e., dislike). Thus, we expected that
meta-dehumanization would indirectly affect aggressive outgroup attitudes and behavior
via outgroup dehumanization (i.e., a significant a*b path; see Figure 1). Consistent with
previous research highlighting the importance of meta-perceptions beyond outgroup
perceptions (e.g., Vorauer et al., 1998), we also considered the possibility that meta-
dehumanization might exert direct effects (i.e., significant c path) on the outcome
variables. We hypothesized that effects of meta-dehumanization would be theoretically
independent from meta-prejudice, or the perception that the outgroup dislikes the
ingroup. Finally, we hypothesized that the effects of meta-dehumanization would be
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
14
unique from political ideologies such as social dominance orientation, right-wing
authoritarianism, and conservatism, previously shown to be associated with
dehumanization (e.g., Costello & Hodson, 2010, 2014; Hodson & Costello, 2007; Kteily
et al., 2015) and aggressive intergroup attitudes (e.g., Hetherington & Suhay, 2011;
Kteily, Ho, & Sidanius, 2012; Kteily, Cotterill, Sidanius, Sheehy-Skeffington, & Bergh,
2014).
Overview of Studies
We examined our hypotheses across ten samples. In Studies 1a and 1b we used
two separate experimental manipulations to examine whether meta-dehumanization
causes outgroup dehumanization among community samples of Americans in the context
of American-Arab relations (Study 1a; N = 210) and American-Muslim relations (Study
1b; N =214). In Study 1c, we experimentally examined whether there was any evidence
for the reverse causal pathway (i.e., from dehumanization of Muslims to meta-
dehumanization; N = 213 Americans). In Study 2, we cross-sectionally examined whether
meta-dehumanization in American-Arab relations was associated with hostile intergroup
attitudes and policy support through greater outgroup dehumanization (N = 270). In
Study 3, we again tested the meta-dehumanization dehumanization hostile attitudes
pathway, but among a large sample of ethnic Hungarians (N = 906), further controlling
for outgroup prejudice toward the Roma. In Study 4, we considered Israelis’ (N = 493)
meta-perceptions about Palestinians, examining not only meta-dehumanization but also
meta-prejudice (“they dislike us”), to address whether meta-dehumanization effects
predict beyond meta-perceptions that are simply negative in valence. In Studies 5a (N =
366) and 5b (N =310), we explored the extent to which Americans feel dehumanized by
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
15
ISIS (in the aftermath of the Charlie Hebdo attacks; Study 5a) and Iran (shortly after the
announcement of the nuclear deal; Study 5b), and examined how this was uniquely
associated with hostile attitudes and behavior. In Study 6 (N = 211), we experimentally
examined whether learning that Muslims humanized Americans (i.e., meta-humanization)
could reduce outgroup dehumanization. Finally, in Study 7 (N = 231), we examined the
mechanism underlying the relationship between meta-perceptions about outgroup ratings
of ingroup humanity and outgroup dehumanization.
Study 1a
In Study 1a, we sought to examine the experimental effect of meta-
dehumanization. Specifically, we tested our prediction that Americans who learn that
they are dehumanized by Arabs will be more likely to dehumanize Arabs in turn.
Method
Participants. We collected data from 216 participants through Amazon
Mechanical Turk, a reliable and high-quality platform for recruiting diverse samples
(Buhrmester, Kwang, & Gosling, 2011) in June 2015. We excluded 6 participants not
native to the U.S. leaving 210 participants (M age = 31.68, SD = 11.55; 51.9% male;
74.8% White American, 9.0% Black American; 7.6% Asian American; 5.2%
Latino/Hispanic American; 1.0% Native American; 2.4% Biracial/mixed race).
Procedure. Participants filled out demographic information and items assessing
patriotism and nationalism, and were then told that we were interested in “people’s
social, economic, and political perceptions and how they compare across cultures”.
Participants read that we had previously conducted “an online survey very similar to the
one you are now completing among large samples of Arabs living in each of 5 different
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
16
and diverse countries: Lebanon, Egypt, Saudi Arabia, Jordan, and Yemen”. The survey
was purported to measure Arabs’ perceptions of how the political system in their
societies functions, their sense of corruption in their social system, their attitudes towards
religion, and their perception of Americans. Participants were then told that they would
see some of the questions that Arab participants answered and their responses.
Participants were then provided with the Arab responses. Specifically,
participants received demographic information about the purported Arab sample,
including country of origin, average age, and religion. They also received (filler)
information about their supposed social networks, and their attitudes relating to political
transparency in their country. Next, participants saw the results of Arabs’ (purported)
perception of Americans, with all participants learning that Arabs felt warmer towards
their own group (Arabs) than towards Americans (i.e., Arab prejudice towards Americans
was held constant).
After seeing Arab warmth towards Americans, participants were provided with
the information central to our manipulation: Arab ratings of Arabs and Americans on the
Ascent Dehumanization scale (Kteily et al., 2015; see Figure 2 for depiction of the
Ascent scale). In the control condition, participants learned that Arabs had rated Arabs
and Americans to be highly and equally evolved (i.e., around 96 on the unmarked 0-100
Ascent scale). In the experimental condition (‘high meta-dehumanization’), participants
learned that Arabs had rated Arabs as highly evolved (i.e., 96 on the Ascent scale), but
rated Americans substantially lower (i.e., around 67 on the Ascent scale). After reading
this information, participants saw a final item indicating Arabs’ purported feelings about
their family’s economic situation (constant across conditions).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
17
Consistent with the cover story that we were interested in cross-cultural
comparisons between Arabs and Americans on a range of metrics, we next gave
participants a series of filler questions that matched the types of questions we had
supposedly asked Arabs (e.g., perceptions of corruption in the U.S.; questions about their
social network). Subsequently, we moved on to the question of central interest: blatant
dehumanization of Arabs.
Outcome Measures.
We first assessed meta-dehumanization and meta-prejudice, presented in
randomized order.
Meta-dehumanization. As a manipulation check, we assessed the extent to which
participants perceived that Arabs dehumanized Americans. Specifically, we asked
participants to indicate their agreement with each of the following items on a 1 (Strongly
Disagree) to 7 (Strongly Agree) scale: “Arabs perceive Americans to be sub-human”,
“Arabs think of Americans as animal-like”, “Arabs see Americans as less evolved than
other groups”, “Arabs think Americans are beasts”, and “Arabs consider Americans to
belong to a lower form of civilization” (α = .95).
Meta-prejudice. We assessed the extent to which participants felt that they were
disliked by Arabs on a 1 (Strongly Disagree) to 7 (Strongly Agree) scale, asking them to
indicate their agreement with the following two items: “Arabs feel cold towards
Americans”, and “Arabs do not have positive attitudes towards Americans” (r = .78, p <
.001).
Next, we assessed dehumanization and prejudice, presented in randomized order.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
18
Dehumanization. We assessed participants’ dehumanization of Arabs by using
the Ascent scale of blatant dehumanization (Kteily et al., 2015). Specifically, participants
rated the average ‘evolvedness’ of members of a series of groups, including Arabs and
Americans, using an unmarked slider bar. The scale is scored from 0-100, with 100
indicating full humanity. In order to assess dehumanization of Arabs, we reversed scores
such that higher scores indicate more dehumanization.
Prejudice. We assessed prejudice towards Arabs using the widely-used feeling
thermometer (Haddock, Zanna, & Esses, 1993). Specifically, participants rated how
warm they felt towards members of a series of groups, including Arabs, using an
unmarked slider bar anchored at 0 (‘very cold and unfavorable’) and 100 (‘very warm
and favorable). Scores were reversed such that higher scores indicate more prejudice
towards Arabs.
Attention check. Finally, we asked participants two questions designed to assess
their level of attention to the experimental materials that they had been provided. Each of
the two questions tested whether they could correctly recall which topics had not been
mentioned in the survey report purportedly completed by Arabs. We excluded
participants (n = 55) who incorrectly answered one of these two questions, leaving a total
sample of 155 participants for the remaining analyses (M age = 32.06, SD = 11.45; 51.0%
male; 78.1% White American; 7.1% Black American; 6.5% Asian American; 5.8%
Latino/Hispanic American; 2.6% Biracial/mixed race).
1
Results
We first assessed whether our manipulation had successfully influenced
participants’ perception of the extent to which Arabs dehumanized Americans. Indeed,
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
19
those participants who saw the survey results suggesting that Arabs dehumanize
Americans were significantly more likely to report that they were dehumanized by Arabs
(M = 4.05, SD = 1.65) than were those who reported that Arabs perceived both Arabs and
Americans as highly (and equally) evolved (M = 2.36, SD = 1.29), F (1, 153) = 49.20, p <
.001, partial η2 = .24. There was also a significant but smaller effect of the meta-
dehumanization manipulation on participants’ sense that they were disliked by Arabs
(high meta-dehumanization condition: meta-prejudice M = 4.88, SD = 1.25; low meta-
dehumanization condition: meta-prejudice M = 4.29, SD = 1.32), F (1, 153) = 8.18, p =
.005, partial η2 = .05.
We next examined how the experimental manipulation influenced participants’
own dehumanization of, and prejudice towards, Arabs. As predicted, American
participants in the high meta-dehumanization condition were significantly more likely to
dehumanize Arabs in turn (M = 28.22, SD = 27.92) than those in the control condition (M
= 16.73, SD = 22.74; F (1, 153) = 7.81, p = .006, partial η2 = .05)
2
. Those in the high
meta-dehumanization condition also reported greater prejudice towards Arabs (M =
48.93, SD = 26.52) than those in the control condition (M = 38.82, SD = 25.49), F (1,
153) = 5.82, p = .02, partial η2 = .04.
3
In sum, the results of Study 1a showed support for our hypothesis that being
exposed to information that one’s ingroup is dehumanized by an outgroup can, in turn,
cause dehumanization of that outgroup. One potential limitation of Study 1a, however, is
that we manipulated meta-dehumanization by giving participants information about how
they were perceived by Arabs using the same scale (Ascent) which was used to assess
outgroup dehumanization. This may have resulted in participants in the meta-
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
20
dehumanization condition rating Arabs lower in part simply because seeing lower ratings
of another group on the Ascent scale licensed (or shifted norms about) outgroup
dehumanization. Furthermore, Study 1a compared meta-dehumanization to a condition in
which participants learned that they were perceived by the outgroup as equally human,
making it possible that participants in that condition felt humanized by the outgroup. In
Study 1b, we sought to examine the effect of meta-dehumanization through a different
experimental manipulation. Specifically, using a text-based prime, we again tested our
prediction that Americans who learn that they are dehumanized by an outgroup (here,
Muslims) would be more likely to dehumanize that group in turn. Here, we compared
individuals in the meta-dehumanization condition to a control condition in which
participants received no information about how they were perceived by the outgroup.
Study 1b
Method
Participants. We collected data from 225 participants through Amazon
Mechanical Turk in October 2015. Three Muslim participants, six non-native U.S.
participants and two participants who asked that their data be excluded from the study
were removed from analysis; we focused our analyses on the 214 remaining participants
(M age = 33.79, SD = 10.22; 56.1% male; 72.4% White American, 9.8% Black
American; 7.0% Asian American; 5.6% Latino/Hispanic American; 2.8% biracial; 0.9%
Native American; 0.5% Middle Eastern/Arab American; 0.9% Other)
4
.
Procedure. After completing demographic information and items assessing
conservatism, patriotism and nationalism, participants were told that we were interested
in “people’s social, economic, and political perceptions and how they might be similar or
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
21
different across cultures”. In the control condition, participants were simply asked to
click ahead to the survey questions. In the experimental condition, participants read an
article purportedly published in the Boston Globe describing the results of a (fictional)
report by the United Nations’ Commission on Global Relations, examining public
perceptions of Americans in the Muslim world. The news article was entitled “In large
parts of Muslim world, Americans perceived as ‘animals’” (see Supplementary Materials
for full text). In order to invoke meta-dehumanization, the article noted that Muslims
across a number of Muslim-majority countries perceived Americans as “brutes, lacking in
self-control and sophistication”, and highlighted quotations from Muslim respondents
describing Americans in animalistic terms. The report noted that these perceptions of
Americans were highly normative, held by a majority of Muslims.
5
Participants’ then completed two outcome measures – dehumanization and
prejudice – in randomized order.
Outcome Measures.
Dehumanization. We assessed participants’ dehumanization of Muslims in two
ways. First, participants rated several groups on the Ascent scale of blatant
dehumanization (Kteily et al., 2015), including Muslims and Americans; as in Study 1a,
we focused on (reverse-scored) ratings of the outgroup (here, Muslims). Second, we
assessed blatant dehumanization of Muslims by asking participants to rate the extent to
which a series of animalistic traits described Muslims, adapted from Bastian, Haslam,
and Denson (2013) and expanded on in Kteily et al. (2015, Study 5): “backward”,
“savage”, “lacking morals”, “cold-hearted”, “scientifically/technologically advanced”,
“primitive”, “aggressive” “barbaric”, “refined” (reverse-scored), “rational” (reverse-
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
22
scored), “capable of self-control” (reverse-scored), “mature” (reverse-scored), “cultured”
(reverse-scored), “logical” (reverse-scored), and “responsible” (reverse-scored). Ratings
were made for these traits on a scale anchored at 1 (‘Not at all’) and 7 (‘Extremely so’; α
= .96). As in Kteily et al. (2015, Study 5), we also assessed blatant dehumanization as a
composite by standardizing the two dehumanization scores (i.e., Ascent and the
animalistic trait composite) and averaging them together (r = .60, p < .001).
Prejudice. We assessed prejudice towards Muslims as in Study 1a.
After presenting all outcome measures, we provided all participants in the
experimental condition with a second text describing Muslims’ perceptions of Americans
in a humanizing light, in an effort to help reverse the negative effects of the manipulation.
Participants were then thoroughly debriefed.
Results
Analyses indicated a chance failure of randomization with respect to
conservatism: specifically, although conservatism was assessed prior to the experimental
manipulation, and despite the large sample size, participants in the meta-dehumanization
condition (vs. control) were significantly more conservative, F (1, 212) = 10.01, p =
.002.
6
Thus, we included conservatism as a covariate in all analyses reported below.
We examined whether the experimental manipulation influenced participants’
dehumanization of, and prejudice towards, Muslims. We began by examining the
(standardized) blatant dehumanization composite. As predicted, Americans who read the
article suggesting that Muslims dehumanized Americans (M = .13, SE = .08) were
significantly more likely to themselves dehumanize Muslims than participants in the
control condition (M = -.12, SE = .08; F (1, 211) = 4.68, p = .03, partial η2= .02).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
23
Examining each of the two dehumanization measures in isolation, we observed a
significant effect on the animalistic trait ratings (meta-dehumanization condition: M =
4.08, SE = .13; control condition: M = 3.65, SE = .13; F (1, 211) = 5.56, p = .02, partial
η2= .03, and a trend using the Ascent scale (meta-dehumanization condition: M = 31.38,
SE = 3.10; control condition: M = 24.71, SE = 3.01; F (1, 211) = 2.34, p = .13, partial η2=
.01.
7
,
8
In addition to the effects on dehumanization, we also observed that participants in
the meta-dehumanization condition reported greater prejudice towards Muslims (M =
55.35, SE = 2.80) than those in the control condition (M = 45.18, SE = 2.71; F (1, 210) =
6.68, p = .01, partial η2= .03.
In sum, the results of Study 1b converged with those of Study 1a: receiving
information that one’s ingroup is dehumanized by an outgroup, either through a
graphical/visual presentation (Study 1a) or through a text-based manipulation (Study 1b),
caused dehumanization of (and prejudice towards) that outgroup in turn. Aside from
showing that the effects of meta-dehumanization on dehumanization held across distinct
manipulations, Study 1b showed that meta-dehumanization increased outgroup
dehumanization relative to a control condition in which participants received no
information about how they were perceived by the outgroup.
Despite our evidence that meta-dehumanization can cause outgroup
dehumanization, it remains possible that the reverse causal pathway also holds. That is,
when individuals dehumanize another group, they may become more likely to feel
dehumanized by that group: this might occur, for example, simply because an animalistic
outgroup seems more likely to itself dehumanize others, because we infer a norm of
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
24
reciprocal dehumanization among those we dehumanize, or as a motivated rationalization
for one’s own outgroup dehumanization. In Study 1c, we set out to test the reverse causal
pathway from outgroup dehumanization to meta-dehumanization.
Study 1c
Method
Participants. We collected data from 220 participants through Amazon
Mechanical Turk in October 2015. Six participants non-native to the U.S. and one
participant who asked that their data be excluded from the study were removed from
analysis, leaving 213 participants (M age = 33.51, SD = 10.56; 50.2% female; 79.3%
White American; 5.6% Black American; 5.6% Asian American; 5.6% Latino/Hispanic
American; 2.8% Biracial; 0.9% Native American).
Procedure. The methodology of Study 1c was similar to Study 1b. Specifically,
after filling out the same demographics and ideological measures as in Study 1b,
participants were randomly assigned into a control or experimental condition. In the
control condition, participants were asked to click ahead to the survey questions. In the
experimental condition, participants were asked to “read the following newspaper article
describing the conclusions of a recently released report about Muslim-majority societies.”
We used a text prime very similar to that used in Study 1b (but here manipulating
outgroup dehumanization). Specifically, we presented participants with a (purported)
Boston Globe article on the results of a (fictional) report from the United Nations’
Commission on Human Rights. This article was entitled “In large parts of Muslim world,
violence harkens to Dark Age” (see Supplementary Materials for full text). The report
purportedly examined the use of violence as a means of punishment and dispute in the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
25
Muslim world. It detailed the use of practices such as public whippings and hand
amputations as punishment for stealing, and death by stoning and public beheadings as
punishment for adultery and drug dealing, which were described using animalistic terms
such as “brutal”. The report also noted that Muslim survey respondents reported using
(and approving of) a number of aggressive actions (e.g., slapping, punching, or biting) in
response to personal disputes, which were (purportedly) responsible for a high number of
deaths in the Muslim world. Finally, the report noted that these violent actions had deep
cultural roots, and were highly normative.
9
Participants then completed two outcome measures – meta-dehumanization and
meta-prejudice – in randomized order.
Outcome Measures.
Meta-Dehumanization. We assessed meta-dehumanization as in Study 1a (but
with respect to Muslims), with the exception that the item referring to ‘beasts’ was
replaced with “Muslims consider Americans to be uncivilized”. Items were assessed on a
1 (‘Strongly Disagree’) to 7 (‘Strongly Agree’) scale (α = .95).
Meta-Prejudice. We assessed meta-prejudice using the following items:
“Muslims feel cold toward Americans”, “Muslims do not have positive attitudes towards
Americans”, “Muslims don’t like Americans much”, “Muslims don’t think of Americans
in a friendly light”, “Americans are not the favorite people of Muslims”, “Muslims are
very fond of Americans” (reverse-scored) and “Muslims feel warm toward Americans”
(reverse-scored). The same scale anchors as for meta-dehumanization was used (α = .94).
Results
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
26
Studies 1a and 1b examined the effect of meta-dehumanization primes on
dehumanization; here, we examined the opposite: whether increasing dehumanization of
Muslims influenced participants’ sense that Muslims dehumanized Americans. We found
that participants primed to dehumanize Muslims did not report significantly higher levels
of meta-dehumanization (M = 3.68, SD = 1.43) than those in the control condition (M =
3.43, SD = 1.58; F (1, 211) = 1.53, p = .22, partial η2 = .007). The same was true for
meta-prejudice: participants primed to dehumanize Muslims showed slightly higher
levels of meta-prejudice (M = 4.95, SD = 1.19) than those in the control condition (M =
4.70, SD = 1.24), but this trend was not significant (F (1, 211) = 2.26, p = .13, partial η2=
.01).
The results of Studies 1a-1c illustrate that whereas meta-dehumanization caused
outgroup dehumanization, the reverse causal pathway— examined using a very similar
sample size and power to detect a comparable effect— was not reliable. Although the
existence of the reverse causal pathway cannot be definitively ruled out on the basis of
these studies (a point we develop further in the General Discussion), our results are more
consistent with the notion that feeling dehumanized by another group induces
dehumanization of that group, rather than the reverse.
In Studies 2-5b, we assessed individual variation in (measured) meta-
dehumanization perceptions across a series of consequential real-world intergroup
conflicts, examining how these perceptions might be uniquely associated with a range of
hostile intergroup attitudes and behaviors via outgroup dehumanization.
Study 2
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
27
In Study 2 we examined the relationship between Americans’ meta-
dehumanization with respect to Arabs, their dehumanization of Arabs, and measures such
as torture support and support for drone strikes. Specifically, we examined whether any
effects of meta-dehumanization on these variables might be explained in part by meta-
dehumanization’s effects on outgroup dehumanization. In order to ensure that any
association between meta-dehumanization and intergroup outcomes was unique from any
potential effects of political ideology, we controlled for a series of ideological
variables— social dominance orientation, right-wing authoritarianism, and political
conservatism— known to be strongly associated with aggressive intergroup outcomes
generally, and dehumanization in particular (e.g., Kteily et al., 2012, 2014, 2015;
Leidner, Castano, Zaiser, & Giner-Sorolla, 2010).
Method
Participants. American residents (N = 286) completed the study online through
Amazon Mechanical Turk in February 2014. As in Studies 1a-c, we focused on the 271
native-born U.S. participants and further excluded one Arab-American participant (M age
= 33.18, SD = 11.78; 50.2% female; 80.3% White American; 5.2% Black American;
4.8% Asian American; 4.5% Latino/Hispanic American; 3.3% Biracial; 1.5% Native
American; 0.4% Other).
Measures. The following constructs were assessed in fixed order.
Social dominance orientation. We used the 16-item SDO-6 scale (Pratto,
Sidanius, Stallworth, & Malle, 1994; α = .89) to assess participants’ support for hierarchy
between groups.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
28
Right-Wing authoritarianism. We used a 12-item version of the RWA scale
(Altemeyer, 1988; α = .87) to tap conventionality, submission to authority, and
aggressiveness against norm violators.
Political conservatism. We assessed political conservatism with three items. Two
items assessed the extent to which participants rated their social and economic views,
respectively, on a continuum from ‘Very Liberal’ (0) to ‘Very Conservative’ (100). One
item assessed political party preference on a scale from 0 (‘Strong Democrat’) to 100
(‘Strong Republican’); α = .88.
Meta-Dehumanization. We assessed meta-dehumanization with six items: “Arabs
perceive Americans to be sub-human”, “Arabs think of Americans as animal-like”,
“Arabs see Americans as less evolved than other groups”, “Arabs think Americans are
beasts”, “Arabs consider Americans to belong to a lower form of civilization”, and
“Arabs think of Americans as vermin” (α = .97). All responses were made using
unmarked sliders anchored at 0 (‘Strongly Disagree’) and 100 (‘Strongly Agree’).
Dehumanization. We assessed dehumanization as in Study 1a, focusing on
Ascent ratings of Arabs.
Emotional hostility. We assessed emotions towards Arabs by providing
participants with seven emotions (anger, disgust, contempt, respect, and sympathy, fear,
and envy), from which we used the first five emotions towards Arabs for our assessment
of emotional hostility (respect and sympathy were reverse-scored; α = .80).
Drone support. We assessed support for drone strikes using five items (e.g., “I
support America’s use of drone attacks against suspected militant targets in Yemen”; α =
.87; see Kteily et al., 2014, 2015).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
29
Surveillance of Arabs. Support for surveillance of Arabs was assessed using a 4-
item scale (sample item: “I think American intelligence services should place extra effort
on the surveillance of Arab immigrants to the U.S.”; α = .91; see Supplemental Materials
for full scale). All responses were made using unmarked sliders anchored at 0 (‘Strongly
Disagree’) and 100 (‘Strongly Agree’); the same scale was used for all other constructs
assessed below (unless otherwise specified).
Arab distancing. We assessed Americans’ support for distancing Arabs using six
items tapping into a broad set of attitudes and social policies reflecting social rejection of
Arabs and resistance to Arab integration into U.S. society (e.g., “It would bother me if
my son or daughter ended up marrying an Arab”; “The U.S. government should set up
programs to help Arab immigrants integrate into U.S. society (reverse-scored); α = .79;
see Supplementary Materials).
Opposition to Arab immigration. We assessed opposition to Arab immigration by
asking participants to assign a limited number of immigrant visas to various groups (e.g.,
East Asians, Arabs, Eastern Europeans), and then taking the reverse-scored proportion of
visas assigned to Arabs (Kteily et al., 2015).
Torture support. We assessed support for torture by averaging participant
responses to five items taken from Kteily et al., 2014, 2015 (e.g., “To put an end to the
war on terror in the Middle East, I think it is OK to use enhanced interrogation
techniques; α = .91; see Supplementary Materials).
We also included items about perceived American and Arab power, perceptions
of American foreign policy, items assessing patriotism and nationalism, and items about
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
30
support for intervention in Syria for exploratory purposes. We did not use these variables
for our primary analyses and they are not discussed further.
Results
Descriptive statistics and variable inter-correlations are presented in Table 1. As
can be seen in the table, meta-dehumanization was significantly associated with
dehumanization of Arabs, r = .38, p < .001. Moreover, both meta-dehumanization and
Arab dehumanization were significantly correlated with each of the outcome measures.
We were primarily interested in examining whether meta-dehumanization was
associated with anti-Arab attitudes and policy support through outgroup dehumanization.
We included political ideology (i.e., each of SDO, RWA, and political conservatism) as
control variables, and examined the full model using Hayes’ (2013) PROCESS macro
(Model 4), with 1,000 bootstrap resamples. We modeled each outcome measure
separately (see Figure 3 for an example outcome, ‘torture of Arabs’).
In support of our predictions, the indirect effect from meta-dehumanization to
anti-Arab attitudes and policy support through Arab dehumanization was significant for
each of the criterion variables (see Table 2). Indeed, in line with the experimental
manipulations in Studies 1a and 1b, meta-dehumanization was strongly associated with
outgroup dehumanization (β= .32, b = .30, p < .001, 95% CI [.19, .41])
10
. Outgroup
dehumanization was itself uniquely associated with each of the outcome variables
(Surveillance of Arabs: β = .24, b = .25, p < .001, 95% CI [.15, .35]; Emotional Hostility:
β =.42, b = .32, p < .001, 95% CI [.25, .39]; Torture support: β =.16, b = .14, p = .004,
95% CI [.05, .24]; Drone support: β =.14, b = .13, p = .004, 95% CI [.04, .23]; Arab
distancing: β =.16, b = .13, p = .006, 95% CI [.04, .22]; Opposition to Arab Immigration:
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
31
β =.36, b = .09, p < .001, 95% CI [.06, .12]).
After accounting for its indirect effects via dehumanization (as well as political
ideology), meta-dehumanization had direct effects on each of the outcome variables, with
the exception of opposition to immigration. Across all outcome measures, meta-
dehumanization exhibited significant total effects. In sum, these data illustrate the
potency of meta-dehumanization as a unique contributor to intergroup hostility, and
demonstrate that a part of its role can be accounted for by its effect on outgroup
dehumanization.
Study 3
In Study 3, we extended the examination of meta-dehumanization to a different
cultural context, specifically Hungarians’ responses to perceived dehumanization of the
ingroup by the Roma population. Although the majority Hungarian population is
advantaged relative to the minority Roma population, the discourse surrounding the
Roma describes them as self-segregating from society and expressing disdain for the
majority population by targeting them for theft and other ‘parasitic’ actions, implying a
(perceived) disregard for the majority Hungarians and their suffering. Therefore, we
reasoned that majority Hungarians could perceive that they themselves are dehumanized
by the Roma, a perception we hypothesized would be associated with dehumanizing the
Roma and aggressive intergroup attitudes. Study 3 also extended Study 2 by further
controlling for outgroup prejudice in testing our theoretical model.
Method
Participants. We used a collection service (Solid Data SIA) to obtain a large
sample online from Hungary in August 2014 for an omnibus study (we focus on the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
32
variables relevant to the current work)
11
. The survey was translated into Hungarian by a
native-speaking social psychologist. Of the 1,002 respondents, 12 were excluded for
being Roma, and 84 for answering at least one of two attention check questions
incorrectly, leaving 906 participants (M age = 42.04, SD = 12.75; 50.7% male).
Measures. For all measures not already assessed on a 0-100 scale, scores were
converted from their original scales (reported below) to a 0-100 scale for ease of
comparison with previous studies. Variables were presented in fixed order.
Conservatism. Political conservatism was assessed using two items asking for
self-placement on the left-right political spectrum (1 = Left; 7 = Right) and the liberal-
conservative (1 = Liberal; 7 = Conservative) spectrum (r = .47, p < .001).
SDO-D. We assessed the dominance sub-dimension of the SDO scale (α = .86;
Ho et al., 2012; see also Ho et al., in press). Responses were made on scales anchored at
1 (‘Strongly Disagree’) and 6 (‘Strongly Agree’).
Prejudice. We assessed prejudice against the Roma similarly to prior studies,
using feeling thermometer ratings towards the Roma on a scale anchored at 0 (‘Cold,
negative feelings’) and 10 (‘Warm, positive feelings’). Scores were reversed such that
higher scores indicated greater outgroup prejudice.
Dehumanization. As in Studies 1 and 2, we assessed dehumanization using the
Ascent scale. We assessed responses to a range of groups, including the Roma and ethnic
Hungarians. Roma dehumanization was assessed by using ratings of Roma humanity on
the Ascent measure, using a scale anchored at 0 (least ‘evolved’) and 10 (most
‘evolved’). As in earlier studies, the scores were reversed such that higher scores indicate
dehumanization.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
33
Emotional hostility. We assessed hostile emotions towards the Roma by indexing
each of the following emotions: anger, hatred, contempt, compassion (reverse-coded) and
pity (reverse-coded); α = .83. Responses were made on scales anchored at 1 (‘Not at all’)
and 6 (‘Very much so’).
Funding to Roma integration. We assessed support for providing funding to
Roma integration by asking participants to indicate the proportion of an EU fund they
thought should be spent on “Roma integration and support” versus “Urban
beautification” (0= ‘None of the budget; 100 = ‘All of the budget’)
12
.
Support for discrimination. Support for discrimination was assessed using 14
items that indexed the extent to which individuals agreed with policies that discriminated
against the Roma in domains spanning education, employment, and housing (e.g.,
“Decrease the number of Roma teachers”; “Cancel currently operating scholarships for
Roma students”; see Supplementary Materials; α = .90)
13
. Responses were made on
scales anchored at 1 (‘Completely Disagree’) and 6 (‘Completely Agree’), and recoded
such that higher scores indicated more discriminatory attitudes.
Perceptions of Roma homogeneity. We assessed perceptions of Roma
homogeneity by asking participants to indicate how similar they thought the Roma are to
each other across several dimensions (e.g., ‘intellect’, ‘values’, ‘morality’; α = .91; see
Kteily et al., 2015). Responses were made on scales anchored at 1 (‘Very different from
one another’) and 6 (‘Very similar to one another’).
Meta-dehumanization. Meta-dehumanization was assessed as in Study 1a (α =
.94), but with Roma as the target group. Scores were assessed on a 1-6 scale.
Responses to injustice. We assessed responses to injustices committed by the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
34
ingroup towards the Roma by asking participants to read a real newspaper story about
Hungarian hooligans who threatened and shouted vulgarities at Roma children and
urinated around their school campus. We asked participants how angry, guilty and
ashamed they felt as Hungarians, and how compassionate they felt towards the Roma
children in response to the story (α = .86). Responses were made on scales anchored at 1
(‘Not at all’) and 6 (‘Very much so’).
Results
Descriptive statistics and variable inter-correlations are presented in Table 3.
As in Study 2, perceived meta-dehumanization by the Roma was positively associated
with dehumanization of the Roma, r = .35, p < .001. Meta-dehumanization was also
associated with prejudice towards the Roma, r = .35, p < .001.
As with Study 2, in Study 3 we examined our proposed model using PROCESS
(see Figure 4 for an example outcome, ‘support for discrimination’). Specifically, we
considered the extent to which meta-dehumanization was associated with hostile
intergroup attitudes through dehumanization of the outgroup, controlling for political
ideology (here, SDO and conservatism) throughout. We extended Study 2 by including
prejudice (i.e., dislike) in the model. Meta-dehumanization was associated with greater
outgroup dehumanization (β =.19, b = .27, p < .001, 95% CI [.19, .35]) and outgroup
prejudice (β = .22, b = .21, p < .001, 95% CI [.15, .26]). Roma dehumanization was
uniquely associated with each of the outcome measures (all bs > .08, ps < .003), with the
exception of responses to injustice (b = -.05, p = .10). Similarly, Roma prejudice was
uniquely associated with each of the outcome measures (all bs > .19, ps < .001), with the
exception of perceived homogeneity (b = .04, p = .37).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
35
We next examined the direct effects of meta-dehumanization on the outcome
measures, and its indirect effects via each of dehumanization and prejudice (entered
simultaneously as predictors; see Table 4). Consistent with the results of Study 2, meta-
dehumanization had significant indirect effects, via dehumanization, on each of: reduced
funding for Roma integration, support for discrimination against the Roma, perceptions
of Roma homogeneity, and emotional hostility towards the Roma. Beyond these indirect
effects via outgroup dehumanization, meta-dehumanization had significant indirect
effects through the mediator of outgroup prejudice on funding to Roma integration,
support for discrimination, less contrite responses to injustices committed towards the
Roma, and emotional hostility towards the Roma. With the exception of responses to
injustice and emotional hostility, meta-dehumanization had significant direct effects on
each of the outcome measures after its relationships with outgroup dehumanization and
outgroup prejudice were taken into account. Moreover, meta-dehumanization had
significant total effects on all outcome measures.
In sum, among a large sample of ethnic Hungarians, the perception that the
outgroup (i.e., Roma) dehumanizes the ingroup (i.e., majority Hungarians) was strongly
associated with hostile outgroup perceptions and policy support. This was in part
channeled through both dehumanization of and prejudice towards the ‘offending’
outgroup, even after controlling for political ideology. Consistent with the previous
studies, these findings suggest the importance of meta-dehumanization as a predictor of
hostile and aggressive intergroup perceptions, attitudes, and policy support.
Study 4
In Study 4, we sought to examine the role of meta-dehumanization in a highly
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
36
consequential social context involving very conflictual intergroup relations and a vicious
ongoing cycle of intractable conflict (Bar-Tal, 2000): the Israeli-Palestinian conflict.
Given the overt animosity frequently expressed and experienced in this context, we
expected that meta-dehumanization would be highly associated with intergroup
aggression. Another important goal of this study was to examine the uniqueness of meta-
dehumanization as a meta-perception: specifically, we considered whether the perception
that the ingroup is dehumanized by an outgroup contributes to the perpetuation of
intergroup hostility beyond the perception that the ingroup is disliked by an outgroup
(i.e., meta-prejudice).
Method
In May 2015, we collected data from 547 Israelis who responded to an online
questionnaire about attitudes and perceptions towards Palestinians. We excluded 54
participants who answered at least one of two attention check questions incorrectly,
resulting in a final sample of 493 (M age = 40.96, SD = 13.07; 52.9% male; 97.4%
Jewish; 0.8% Christian; 0.4% Other; 1.4% No religion)
14
. In addition to the variables of
interest, the questionnaire included an experimental manipulation and associated
measures for purposes unrelated to the current study.
15
Measures. All variables were assessed on scales anchored at 1 (‘Strongly
Disagree’) and 7 (‘Strongly agree’), but were converted to a 0-100 scale for ease of
comparison across studies. Variables were presented in fixed order.
Emotional hostility. We assessed hostile emotions towards Palestinians by
indexing each of the following emotions: hatred, anger, hostility, empathy (reverse-
coded), shame (reverse-coded), and guilt (reverse-coded; α =.76).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
37
Support for negotiations. Support for negotiation was assessed using two items:
“How willing would you be for Israel to enter direct negotiations with the Palestinians?”,
and “Israel should make a concerted effort to negotiate a resolution with the Palestinians”
(r =.89, p < .001).
Expulsion of Palestinians. We assessed support for the expulsion of Palestinians
as a potential ‘solution’ to the Israeli-Palestinian conflict. Specifically, we asked
participants to indicate their agreement with the following item: “Absorbing the West
Bank and Gaza into Israel and forcing the Palestinians to go to Jordan”.
16
Support for aggressive Policies. We assessed support for aggressive policies
towards the Palestinians, using seven items (e.g., “We should torture any Palestinian
suspected of building tunnels in Gaza”; “Israel should use live fire to disperse Palestinian
protests, even at the cost of hurting civilians and bystanders”; α = .84; see Supplementary
Materials).
Meta-dehumanization. Meta-dehumanization was assessed using the same 5
items as in Study 3 (α = .92), here with respect to Palestinians.
Meta-prejudice. Meta-prejudice was assessed using the following two items:
“Palestinians feel cold towards Israelis”, and “Palestinians do not hold positive attitudes
towards Israelis” (r = .63, p < .001).
Dehumanization. Dehumanization was assessed using the Ascent scale of blatant
dehumanization, assessed with respect to a number of groups, including Palestinians and
Israelis. We assessed outgroup dehumanization as in previous studies, by reverse scoring
ratings of Palestinians.
Results
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
38
Descriptive statistics and variable inter-correlations can be found in Table 5. We
began by exploring the dimensionality of the items assessing meta-dehumanization and
meta-prejudice. Submitting these seven items to a principal component factor analysis
with oblique rotation yielded two distinct factors. Consistent with our theorizing, the first
factor (eigenvalue = 4.34, 62.04% variance explained) reflected meta-dehumanization (all
factor pattern loadings > .74), and the second factor (eigenvalue = 1.17, 16.68% variance
explained) reflected meta-prejudice (factor pattern loadings > .85). There were no cross-
loadings across factors (i.e., no variable had a factor pattern loading > .30 on the other
factor). We created two composites, one for each of these constructs (meta-prejudice: M
= 72.32, SD = 25.31; meta-dehumanization: M = 41.34, SD = 28.00), which were inter-
correlated, r =.48, p < .001.
We next examined the relationship between meta-dehumanization, meta-
prejudice, and dehumanization of Palestinians. Both meta-dehumanization (r = .30, p <
.001) and meta-prejudice (r = .25, p < .001) were associated with greater dehumanization
of Palestinians. When we entered these two variables into a simultaneous regression
predicting dehumanization, each explained unique variance (meta-dehumanization: β =
.23, b = .22, p < .001, 95% CI [.12, .32]; meta-prejudice: β = .15, b = .16, p = .003, 95%
CI [.05, .26]).
We next examined the indirect effects of meta-dehumanization on the outcome
variables via dehumanization of Palestinians, as well as its direct and total effects. As can
be seen in Table 6, meta-dehumanization had a significant indirect effect on all outcome
variables via outgroup dehumanization, controlling for meta-prejudice (see Figure 5 for
an example outcome, ‘support for aggressive policies’). Meta-dehumanization further had
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
39
significant direct and total effects on all variables, again suggesting its unique role in
predicting hostile intergroup attitudes and policies. In parallel to meta-dehumanization,
meta-prejudice was also uniquely associated with several of the outcome variables (see
Supplementary Table 1).
Study 4 extended our research in several important ways: First, it documented the
importance of meta-dehumanization in another national context marked by hostile
intergroup relations. Second, it showed this consistently across a range of highly
meaningful and consequential outcome measures, including efforts for peaceful conflict
resolution (i.e., support for negotiation) as well as highly belligerent actions, such as
population transfer and support for collective aggression, likely to contribute to an
ongoing cycle of conflict. Finally, and importantly, we established that perceiving that
another group dehumanizes the ingroup is distinct from perceiving that they dislike the
ingroup. Interestingly, this meta-perceptual distinction between being (perceived to be)
dehumanized and disliked parallels a similar distinction between dehumanizing and
disliking, which have been shown to exert independent effects on intergroup outcomes
(e.g., Andrighetto et al., 2014; Kteily et al., 2015).
In Studies 5a and 5b, we sought to further our research in two primary ways. First,
we aimed to re-examine the unique effects of meta-dehumanization and meta-prejudice
while simultaneously examining both outgroup dehumanization and outgroup prejudice.
This allowed us to test an important and heretofore unanswered question: whether there is
a dehumanization-specific pathway— from meta-dehumanization through
dehumanization to aggressive intergroup attitudes and policy support— that is
independent of both meta-prejudice and prejudice (as per Figure 1). Second, in addition
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
40
to attitudes and policy support, we included behavioral criterion measures.
Study 5a focused on Americans’ perceptions and meta-perceptions with respect to
ISIS members. This study was conducted shortly after the attacks in Paris, France in
January 2015 in which Muslim extremists associated with ISIS killed 11 individuals at
the satirical magazine Charlie Hebdo, following the publication of what were perceived
as demeaning and highly offensive images of the prophet Mohammed. Study 5b
examined Americans’ perceptions and meta-perceptions with respect to Iran in the
summer of 2015, shortly after the Iran nuclear deal was announced and while its benefits
and risks were being hotly debated.
Study 5a
Method
Participants. We collected data from 423 participants on Amazon’s mTurk
platform a few days after the Charlie Hebdo attacks in January 2015. We excluded one
Arab participant and 56 non-native born participants, leaving 366 participants (M age =
32.72, SD = 11.01; 57.9% female; 83.1% White American, 4.9% Asian American, 4.4%
Black American; 4.6% Latino/Hispanic American; 1.9% Biracial; 0.8% Native
American; 0.3% Other).
Measures. We began by assessing meta-dehumanization and meta-prejudice,
which were presented in randomized order.
Meta-dehumanization. Meta-dehumanization was assessed (with respect to how
ISIS perceive Westerners), using the same six-item scale as in Study 2 (α= .88). We
focused here on meta-perceptions relating to Westerners given that the Charlie Hebdo
attacks occurred in France (rather than the U.S.), and reflected a larger conflict between
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
41
ISIS and Western nations (and values). We reasoned that in this context, Americans’
membership in the broader category of ‘Westerner’ would be most relevant with respect
to their meta-perceptions.
Meta-prejudice. Meta-prejudice was assessed (with respect to how ISIS perceive
Westerners) using the two-item scale in Study 4 (r = .71, p < .001).
Next participants reported their dehumanization and prejudice towards ISIS.
Dehumanization. Dehumanization of ISIS was assessed as in previous studies
(i.e., by taking the reverse score of the humanity attributed to ISIS on the 0-100 Ascent
scale of blatant dehumanization).
Prejudice. Prejudice towards ISIS was assessed using reverse scored ratings of
ISIS on the feeling thermometer, as in Study 3 (on a 0-100 scale).
We next assessed a range of outgroup attitudes and behavior. Variables were
presented in randomized order.
17
Drone support. Support for drone strikes was assessed using the same five-item
scale as in Study 2, with slight modifications (e.g., “I support Western countries using
drone attacks against suspected militant targets in Yemen”; α = .84).
Opposition to Muslim immigration. Opposition to Muslim immigration to the
U.S. was assessed as in Study 2.
Militaristic counter-terrorism. Support for militaristic counter-terrorism was
assessed using a 13-item scale adapted from Kteily et al. (2014, 2015; e.g., “To put an
end to terrorist acts by ISIS, I think it is OK to use enhanced interrogation techniques”;
“We should strike back with brutal force against any members of ISIS who seek to
intimidate us”; α = .93; see Supplementary Materials). Responses were made on scales
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
42
anchored at 1 (‘Strongly Disagree’) and 7 (‘Strongly Agree’), and then converted to a 0-
100 scale.
Signing anti-ISIS petitions. We examined whether participants chose to sign in
support or opposition of six different petitions, taken from Kteily et al. (2015, Study 5),
about taking various measures to combat ISIS (e.g., “Increase the military budget allotted
to combating the ISIS threat”; “Forcibly deport all Islamic clerics in the U.S. who preach
in favor of ISIS.” Participants were told that the petition sponsors had agreed to use
mTurk IDs as proxies for names because they are uniquely assigned to individuals. For
each petition, participants could indicate whether they would like to add their mTurk ID
in support (coded as +100), in opposition (coded as -100), or choose not to add their
mTurk ID at all (coded as 0). Responses were coded such that higher scores reflected
more hostility towards ISIS (α = .78).
Anti-Islamic extremism fund disbursement. We assessed anti-Islamic extremism
fund disbursement by asking participants to distribute funds between two programs
aimed at decreasing extremism among Islamic communities in the U.S.: one centered on
policing and surveillance of Muslims, and the other on providing them with education
and opportunities for learning (see Kteily et al., 2015, Study 5). We used the percentage
of funds allocated to policing and surveillance of Muslims as our measure of
punitiveness.
Encouragement of U.S. troops fighting ISIS. We gave participants the
opportunity to write messages in support of American troops combatting ISIS: we
assigned a score of 100 to participants who chose to write a message, and a score of 0 to
those who did not (Kteily et al., 2015, Study 5).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
43
Supportive Messages to Families of Hebdo Victims. Subsequent to being given
the opportunity to write in support of U.S. troops, participants received the same prompt
about whether or not they would like to write a message in “support of the families of the
French journalists at Charlie Hebdo targeted in the past several hours”. We assigned a
score of 100 to participants who chose to write a message, and a score of 0 to those who
did not.
18
Punitiveness towards Hebdo Attackers. After selecting whether or not they
wanted to write a message to the families of the victims of the Charlie Hebdo attacks, we
also asked participants to report their agreement with each of the following items
assessing punitiveness towards the perpetrators (Kteily et al., 2014): “The perpetrators of
the Charlie Hebdo attacks deserve to die a slow, painful death”, “If found guilty of the
attack, the perpetrators of the Charlie Hebdo attack should be subjected to the death
penalty”, and “I hope the perpetrators of the Charlie Hebdo attack rot in hell” (α = .80).
Responses were made on scales anchored at 1 (‘Strongly Disagree’) and 7 (‘Strongly
Agree’), and transformed to a 0-100 scale, with higher scores indicate greater agreement
for ease of comparison with previous studies.
Conservatism. We assessed conservatism with 3 items as in Study 2 (α = .89).
Results
Descriptive statistics and variable inter-correlations can be found in Table 7. We
submitted the eight items assessing meta-dehumanization and meta-prejudice to a
principal components analysis with oblique rotation. Consistent with Study 4, two factors
emerged: the first factor (eigenvalue = 3.56, 50.84% variance explained) reflected meta-
dehumanization (all factor pattern loadings > .70), and the second factor (eigenvalue =
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
44
1.33, 19.01% variance explained, all factor pattern loadings > .90) reflected meta-
prejudice. Using a factor loading of .30 as a cutoff, there were once more no cross-
loadings across the two factors, which were correlated at r = .38, p < .001.
In our main analysis, we tested our full theoretical model (see Figure 1): we
examined whether meta-dehumanization was uniquely associated with the outcome
measures via dehumanization, controlling for both meta-prejudice and outgroup prejudice
(as well as conservatism). That is, we sought to identify, for the first time, a
dehumanization-specific pathway from meta-perceptions to outgroup attitudes and policy
support.
The first part of our model (i.e., ‘a’ path in Figure 1) links meta-dehumanization
to outgroup dehumanization. As expected, meta-dehumanization (r = .26, p < .001), but
not meta-prejudice (r = .04, p = .47), was significantly correlated with dehumanization of
ISIS. We observed this same pattern when we entered these two variables into a
simultaneous regression (controlling for conservatism): meta-dehumanization predicted
dehumanization of ISIS (β = .29, b = .50, p < .001, 95% CI [.31, .68]), whereas meta-
prejudice did not (β = -.07, b = -.16, p = .19, 95% CI [-.38, .08]). Beyond its association
with outgroup dehumanization, we also observed that meta-dehumanization was
associated with anti-ISIS prejudice (r = .33, p < .001), as was meta-prejudice (r = .38, p <
.001). When simultaneously entered into a regression, both variables predicted unique
variance in anti-ISIS prejudice (meta-dehumanization: β = .21, b = .24, p < .001, 95% CI
[.11, .38]; meta-prejudice: β = .30, b = .42, p < .001, 95% CI [.23, .64].
The second part of our model (i.e., ‘b’ path in Figure 1) posits that
dehumanization will be uniquely associated with outcome measures, independent of
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
45
outgroup prejudice (and all other predictors; i.e., meta-dehumanization, meta-prejudice,
and conservatism). With the exception of sending messages of support to American
troops fighting ISIS (b = .01, p = .07, 95% CI [-.00, .02]) and messages of support to the
families of the Hebdo attack victims (b = .001, p = .73, 95% CI [-.01, .01]),
dehumanization of ISIS was itself uniquely associated with each of the outcome variables
(all bs > .05, all ps < .045). Beyond the role of dehumanization, prejudice towards ISIS
itself was significantly associated with punitiveness towards the Hebdo attackers (β = .19,
b = .27, p = .001, 95% CI [.11, .44], support for punishing and controlling (vs. educating)
Muslims to prevent extremism (β = .13, b = .18, p = .02, 95% CI [.04, .33]), signing anti-
ISIS petitions (b = .17, p = .04, 95% CI [.01, .33]), militaristic counter-terrorism (β = .20,
b = .21, p < .001, 95% CI [.11, .30]), and drone strike support (β = .21, b = .22, p < .001,
95% CI [.11, .33]).
Finally, our model posits that meta-dehumanization will be associated with
intergroup outcomes in part via its effects on outgroup dehumanization. As in prior
studies we examined the indirect, direct, and total effects of meta-dehumanization on the
outcome variables, controlling for ideology (here, conservatism; see Figure 6 for an
example outcome, ‘signing anti-ISIS petitions’). We began by focusing on the pathway of
particular interest: from meta-dehumanization to the outcome variables via
dehumanization of ISIS. Consistent with predictions, this indirect pathway was
significant for drone support, militaristic counter-terrorism, opposition to immigration,
signing anti-ISIS petitions, distributing funds to the control and punishment of Muslims
(vs. educating them), and punitiveness towards the Hebdo attackers (but not for sending
messages to U.S. troops fighting ISIS or families of the Hebdo victims; see Table 8).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
46
Meta-dehumanization also had significant indirect effects on six of the eight outcome
measures via prejudice towards ISIS. In fact, meta-dehumanization was associated with
intergroup attitudes and behavior in this study largely through its indirect effects: once
these were accounted for, it had significant direct effects only on signing anti-ISIS
petitions and punitiveness towards the Hebdo attackers. Exploring the total effects, there
was evidence of meta-dehumanization contributing uniquely to the explanation of
militaristic counter-terrorism, opposition to immigration, signing anti-ISIS petitions, and
punitiveness towards the Hebdo attackers.
We observed less evidence, on the other hand, for unique effects of meta-
prejudice on the outcome measures (see Supplementary Table 2). There was some
support for a prejudice-specific pathway from meta-perceptions to outcomes: meta-
prejudice had significant indirect effects via prejudice on five of the outcome measures.
On the other hand, it exhibited no indirect effects via dehumanization, and had no
significant direct effects on any of the outcome variables. When total effects were
examined, there was no evidence that meta-prejudice provided additional utility in
predicting the outcome variables in this context.
In sum, the results of Study 5a provide further evidence for the unique role of
meta-dehumanization in intergroup conflict settings. Among a large community sample
of Americans, and specifically focusing on meta-perceptions regarding ISIS (the group
widely thought to be associated with the Charlie Hebdo attacks), we observed that: (a)
meta-dehumanization is distinct from meta-prejudice; (b) meta-dehumanization is
associated with intergroup outcomes via outgroup dehumanization, independent of
prejudice and political ideology; and (c) meta-dehumanization is associated not only with
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
47
intergroup attitudes, but also with behavior.
In Study 5b, we re-examined the role of meta-dehumanization and meta-prejudice
among Americans in a separate context and focusing on a different outgroup.
Specifically, we assessed meta-perceptions relating to Iran shortly after the
announcement of the hotly debated nuclear deal that President Obama announced in July
2015. We also expanded our measurement of meta-prejudice (i.e., using a 5-item scale, as
in Study 1b) and of blatant dehumanization (including, as in Study 1b, outgroup ratings a
series of blatant animalistic traits in addition to Ascent scale ratings).
Study 5b
Participants. We collected data from 312 U.S. residents on Amazon’s mTurk
platform in July 2015. We excluded one Arab participant and one who did not report
ethnicity, leaving 310 participants (M age = 31.80, SD = 9.86; 53.6% female; 79.7%
White American; 8.1% Hispanic American; 5.8% African American; 3.9% Asian
American; 1.9% Native American; 0.6% Other).
Measures.
Conservatism. Conservatism was assessed using two items assessing self-
placement on 1 (‘Very liberal’) to 7 (‘Very conservative’) scales assessing economic and
social conservatism, respectively. Scores were transformed to a 0-100 scale (r = .52, p <
.001).
Next, participants responded to items about prejudice and dehumanization, in
randomized order.
Prejudice. Prejudice towards Iranians was assessed using reverse scored ratings
of Iranians on the feeling thermometer, as in previous studies (on a 0-100 scale).
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
48
Dehumanization. As in Study 1b, we computed a dehumanization composite by
taking the average of (a) reverse-scored ratings of Iranians on the Ascent scale (as in
previous studies reported here; M = 22.53, SD = 25.48) and (b) ratings of Iranians on a
series of nine animalistic traits adapted from Bastian et al. (2013); e.g., “savage,
aggressive”, “barbaric, cold-hearted”, “capable of self control” (reverse-scored), and
“rational and logical” (reverse-scored). Participants rated the extent to which these traits
descried Iranians on a 1 (‘Not at all’) to 7 (‘Extremely so’) scale (α= .90; M = 4.01, SD =
1.20).
19
,
20
Ascent dehumanization scores and animalistic trait ratings were highly
correlated (r = .57, p < .001), and they were thus standardized and averaged to comprise a
blatant dehumanization composite.
Meta-dehumanization. Meta-dehumanization was assessed (with respect to how
Iranians perceive Americans) using the same six-item scale as in Study 2, with the last
item from that scale (referring to ‘vermin’) replaced with “Iranians would happily step on
Americans like cockroaches” (α= .95).
Meta-prejudice. Meta-prejudice was assessed using the following items: “Iranians
feels cold towards Americans”, “Iranians do not have positive attitudes towards
Americans”, “Iranians don’t like Americans much”, “Iranians don’t think of Americans
in a friendly light”, and “Americans are not Iranians’ favorite people” (α = .97).
We next assessed a range of outgroup attitudes and behavior, presented in fixed
order:
Opposition to the Iran Nuclear Deal was assessed by asking participants to
indicate their agreement or disagreement with six statements on a scale anchored at 1
(Strongly Disagree) and 7 (Strongly Agree): “I am embarrassed that the U.S. negotiated
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
49
with the Iranians rather than enforcing our will militarily”, “Iran is dedicated to the
destruction of the U.S.”, “The Iranians will use the money we give them to buy
conventional weapons, and the moment they get a chance, they will build a nuclear
weapon”, “This treaty has the potential to heal old wounds between the U.S. and Iran,
and bring our two countries closer together” (reverse-scored), “I am in favor of the U.S.
nuclear deal with Iran” (reverse-coded), and “I think Iran is just as entitled to a nuclear
program as any other nation.” (reverse-scored; α = .87)
21
. We transformed scores to a 0-
100 scale for ease of comparison.
Aggressive Actions towards Iranians was assessed by asking participants to
indicate their agreement or disagreement with each of twelve actions towards Iranians
(e.g., “I think it’s acceptable to assassinate Iranian nuclear scientists”, “Anyone caught in
America spying on the U.S. for Iran should be sentenced to prison for life without
parole”; see Supplemental Materials for full scale). Items were assessed on a scale
anchored at 1 (‘Strongly Disagree’) and 7 (‘Strongly Agree’), and scores were
transformed to a 0-100 scale (α = .93).
Signing Anti-Nuclear Deal Petitions. Using the same methodology as in Study
5a, we assessed whether participants chose to sign in support or opposition to five
different petitions relating to the Iran nuclear deal (e.g., “Urge congressional members to
examine military options against Iran”; α = .85; see Supplemental Materials for full
scale).
We also included items outside the scope of the current study (e.g., items on
intergroup contact), which were not used and are not discussed further.
Results
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
50
Descriptive statistics and variable inter-correlations can be found in Table 9. We
submitted the items assessing meta-dehumanization and meta-prejudice to a principal
components analysis with oblique rotation. Consistent with Studies 4 and 5a, two factors
emerged: the first factor (eigenvalue = 71.44, 71.44% variance explained) reflected meta-
dehumanization (all factor pattern loadings > .74), and the second factor (eigenvalue =
1.45, 13.15% variance explained, all factor pattern loadings > .86) reflected meta-
prejudice. Using a factor loading of .30 as a cutoff, there were once more no cross-
loadings across the two factors. The meta-dehumanization and meta-prejudice
composites were correlated at r = .70.
We examined our full theoretical model as in Study 5a. Meta-dehumanization (r
= .64, p < .001) and meta-prejudice (r = .50, p < .001) were both significantly correlated
with dehumanization of Iranians. Replicating the results of Study 5a, when we entered
these two variables into a simultaneous regression (controlling for conservatism), meta-
dehumanization was uniquely associated with dehumanization of Iranians (β = .55, b =
.02, p < .001, 95% CI [.01, .02]), whereas meta-prejudice was not (β = .09, b = .00, p =
.15, 95% CI [-.00, .01]). Beyond outgroup dehumanization, meta-dehumanization (r =
.48, p < .001) and meta-prejudice (r = .46, p < .001) were each correlated with anti-Iran
prejudice. Consistent with Study 5a, when simultaneously entered into a regression, both
variables were uniquely associated with anti-Iran prejudice (meta-dehumanization: β =
.28, b = .30, p < .001, 95% CI [.15, .45]; meta-prejudice: β = .23, b = .25, p = .001, 95%
CI [.10, .41]), controlling for conservatism.
Next, we examined the effects of dehumanization on the outcome measures.
Consistent with our theoretical model, outgroup dehumanization predicted all of the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
51
outcome variables, controlling for all other predictors (i.e., meta-dehumanization, meta-
prejudice, outgroup prejudice, and conservatism): opposition to the Iran nuclear deal (β =
.32, b = 8.42, p < .001, 95% CI [4.52, 12.21]), aggressive actions towards Iranians (β =
.30, b = 7.52, p < .001, 95% CI [3.91, 11.13]), and signing petitions against the nuclear
deal (β = .16, b =7.19, p = .04, 95% CI [.77, 13.64]). In contrast to dehumanization,
although prejudice towards Iran significantly predicted aggressive actions (β = .13, b =
.09, p = .02, 95% CI [.004, .18]), it was not uniquely associated with opposition to the
Iran nuclear deal (β = .00, b = .00, p = .97, 95% CI [-.09, .10]) or signing petitions against
the nuclear deal (β = .00, b = .00, p = 1.00, 95% CI [-.17, .17].
Finally, we examined whether the dehumanization-specific indirect pathway from
meta-perceptions to outgroup attitudes and behavior held as in Study 5a. We examined
the indirect, direct, and total effects of meta-dehumanization on the outcome variables,
controlling for ideology (here, conservatism; see Figure 7 for an example outcome,
‘aggressive actions towards Iran’). As with Study 5a, the indirect pathway from meta-
dehumanization to outcomes via outgroup dehumanization was significant across
outcome measures, including our behavioral measure of signing anti-Nuclear deal
petitions (see Table 10). In addition to its indirect effects via outgroup dehumanization,
meta-dehumanization also had indirect effects on aggressive actions towards Iranians via
outgroup prejudice (but not for the other outcome measures). Beyond these indirect
effects, meta-dehumanization’s direct and total effects were significant across the
outcome measures.
Also consistent with Study 5a, meta-prejudice exhibited less of a unique
association with the outcome measures (see Supplementary Table 3). There was some
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
52
support for a prejudice-specific pathway: there were significant indirect effects from
meta-prejudice on aggressive actions towards Iranians (though not for the other two
outcome measures) through outgroup prejudice. On the other hand, meta-prejudice had
no indirect effects via dehumanization, and exhibited no significant direct or total effects
on any of the outcome variables.
In sum, the results of Study 5b replicate the results of Study 5a, and provide
further evidence for the unique role of meta-dehumanization in intergroup conflict
settings. Among a large community sample of Americans, and focusing on meta-
perceptions regarding Iranians during a period where the intergroup relationship between
the two sides came under close scrutiny, we again observed that: (a) meta-
dehumanization is distinct from meta-prejudice, (b) meta-dehumanization is associated
with intergroup outcomes via outgroup dehumanization, independent of prejudice and
political ideology, and (c) meta-dehumanization is associated not only with intergroup
attitudes, but also with behavior.
The set of studies described thus far have documented, for the first time, that
perceiving that one’s own group is blatantly dehumanized by an outgroup increases
blatant dehumanization of that group in turn. These studies have also provided further
empirical support for the notion that blatant outgroup dehumanization itself contributes to
a range of aggressive outcomes that may foment cycles of intergroup conflict and
violence. If perceiving that another group dehumanizes the ingroup increases outgroup
dehumanization, thereby helping to ignite a process contributing to intergroup strife,
might providing information that the outgroup humanizes the ingroup serve as an
effective intervention to reduce outgroup dehumanization? We examined this question in
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
53
Study 6.
Study 6
In Studies 1a and 1b, we established that priming Arab meta-dehumanization of
Americans resulted in the reciprocal dehumanization of Arabs. In Study 6, we examined
whether priming meta-humanization could similarly humanize a typically dehumanized
target group?
Method
Participants. We collected data from 220 participants through Amazon
Mechanical Turk in October 2015
22
. Data from three Muslim participants, four
participants non-native to the U.S. and two participants who asked that their data be
excluded from the study were removed from analysis, leaving 211 participants (M age =
36.77, SD = 12.85; 50.7% male; 80.1% White American; 7.6% Asian American; 5.7%
African American; 3.8% Hispanic American; 1.9% Biracial; 0.9% Other).
Procedure. Participants followed the same general procedure as in Study 1b: in
the experimental condition, participants read an article purportedly published in the
Boston Globe describing the results of a report by the United Nations’ Commission on
Global Relations examining public perceptions of Americans in the Muslim world. The
article was entitled “In large parts of Muslim world, American achievements greatly
admired” (see Supplementary Materials for full text). The article was modeled closely
after that used in Study 1b, but described Muslims’ perceptions of Americans in
humanizing (rather than dehumanizing) terms, such as “technologically advanced”,
“sophisticated”, “culturally advanced”, and as having “enlightened” principles. The
article also quoted Muslims respondents extolling the American educational system, and
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
54
U.S. efforts to avoid civilian casualties during wartime. As in Study 1b, the report noted
that these perceptions of Americans were highly normative, held by a majority of
Muslims.
23
We examined the effect of this meta-humanizing prime on Muslim
dehumanization and prejudice, presented in random order.
Outcome Measures.
Dehumanization. As in previous studies, participants rated a number of groups,
including Muslims and Americans, on the Ascent scale. We assessed participants’
dehumanization of Muslims, as in Study 1b, using both the (reverse-scored) Ascent scale
rating of Muslims (M = 25.01, SD = 29.49), and ratings of Muslims on the animalistic
traits adapted from Bastian et al. (2013; α = .96; M = 3.69, SD = 1.35). We again also
examined blatant dehumanization as a composite by standardizing the two
dehumanization scores (i.e., Ascent and the animalistic trait composite) and averaging
them together (r = .59, p < .001).
Prejudice. We assessed prejudice towards Muslims as in Study 1b, using
reverse-scored feeling thermometer ratings of Muslims (M = 54.84, SD = 30.46).
Results
In our primary analyses, we examined whether the experimental manipulation
influenced participants’ dehumanization of, and prejudice towards, Muslims. We began
by examining the (standardized) blatant dehumanization composite. As predicted,
Americans who read the article suggesting that Muslims humanized Americans were
significantly less likely to dehumanize Muslims (M = -.18, SD = .78) than participants in
the control condition (M = .18, SD = .96), F (1, 208) = 7.11, p = .004, partial η2 = .04. We
also examined each of the dehumanization scales separately. Ascent dehumanization was
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
55
significantly lower in the meta-humanization condition (M = 19.89, SD = 24.34) than it
was in the control condition (M = 30.09, SD = 33.17; F (1, 209) = 6.49, p = .01, partial η2
= .03).
24
,
25
,
26
Similarly, animalistic trait ratings of Muslims were also lower in the meta-
humanization condition (M = 3.44, SD = 1.29) relative to the control condition (M = 3.93,
SD = 1.36; F (1, 209) = 7.10, p = .008, partial η2 = .03).
In contrast to dehumanization, prejudice towards Muslims was similar in the
meta-humanization condition (M = 42.37, SD = 28.77) and the control condition (M =
47.92, SD = 31.95; F (1, 209) = 1.76, p = .19, partial η2 = .01).
Thus, these results demonstrated that individuals who learn that the outgroup
perceives them in humanizing ways in turn humanize the outgroup. In this way, meta-
humanization may serve as an effective intervention in decreasing the tendency to
dehumanize outgroups.
Studies 1-6 support the notion that (meta-)perceptions about the extent to which
the ingroup is seen as human influence individuals’ own attributions of outgroup
humanity (with implications for intergroup outcomes), but do not directly provide
empirical support for the mechanisms that we posit to underlie this association. In Study
7, we tested the proposed mediating role of identity threat and desires for reciprocity.
Study 7
In the introduction, we reasoned that perceiving that the ingroup is dehumanized
by an outgroup would provoke social identity threat (Branscombe et al., 1999a),
generating a desire to reciprocate that hostility towards the offending outgroup (Bourhis
et al., 1979; Doosje & A. Haslam, 2005). Here, we examined whether individuals learned
that they were dehumanized (vs. humanized) by an outgroup would be more likely to feel
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
56
identity threat and a desire to reciprocate the outgroup’s perception, which might account
for individuals’ own ratings of the outgroup’s humanity.
Method
Participants. We collected data from 259 participants through Amazon
Mechanical Turk in October 2015. Data from one Muslim participant, nine participants
non-native to the U.S. and one participant who asked for data exclusion were removed
from analysis; an additional sixteen subjects failed one of the attention check questions
(see below), leaving 231 participants (M age = 36.51, SD = 11.86; 52.8% female; 82.3%
White American; 6.5% African American; 4.8% Hispanic American; 3.0% Asian
American; 1.3% Native American; 1.3% Biracial; 0.4% Arab American; 0.4% Other).
Procedure. Participants followed the same procedure as in Study 6, with the
exception that participants were randomly assigned to receive either the meta-
dehumanization prime (from Study 1b) or the meta-humanization prime (from Study 6).
After reading the primes, participants were asked to respond to items assessing their
sense of threat and desire to reciprocate Muslims’ perceptions of Americans, before being
presented with items assessing dehumanization. We further used two attention checks at
the end of the study, asking participants to report (a) whether the article they read
suggested that Muslims had a positive, negative, or undetermined view of Americans,
and (b) correctly recall the organization that authored the report detailed in the newspaper
article.
27
Measures.
Identity threat. We assessed participants’ sense of threat to their American
identity by asking them to answer the following items, all preceded by the stem, “When I
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
57
think about the way that Muslims perceive Americans”: “I find it offensive”, “I find it
illegitimate”, “I find it threatening”, “I think it is reasonable” (reverse-coded), and “I find
it appropriate” (reverse-coded). Participants responded on a scale anchored at 1 (‘Not at
all’) and 7 (‘Very much so’; α = .86).
Reciprocity. We assessed participants’ desire to reciprocate Muslims’ perceptions
of Americans by asking them to indicate their agreement with each of the following items
(preceded by the same stem as above): “It makes me want to respond back negatively”,
and “It makes me want to respond back positively” (reverse-coded). Participants again
responded on a scale anchored at 1 (‘Not at all’) and 7 (‘Very much so’; r = .58, p <
.001).
28
Dehumanization. As in Study 6, dehumanization of Muslims was assessed by
Ascent dehumanization, animalistic trait ratings, and their composite (r = .56, p < .001).
Results
We first examined the effect of experimental condition on each of the proposed
mediators. Participants in the meta-dehumanization condition reported higher levels of
identity threat (M = 4.22, SD = 1.39) than those in the meta-humanization condition (M =
2.50, SD = .99; F (1, 229) = 115.46, p < .001, partial η2 = .34). Similarly, participants in
the meta-dehumanization condition reported a higher desire to reciprocate hostility (M =
3.84, SD = 1.66) than those in the meta-humanization condition (M = 2.30, SD = 1.12; F
(1, 229) = 67.54, p < .001, partial η2 = .23). These two mediators were themselves quite
correlated, r = .71, p < .001.
We next examined the effect of experimental condition on outgroup
dehumanization. We observed that those in the meta-dehumanization condition indicated
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
58
higher levels of outgroup dehumanization on the composite dehumanization measure (M
= .21, SD = .96) than those in the meta-humanization condition (M = -.23, SD = .73; F (1,
229) = 15.04, p < .001, partial η2 = .06). We observed similar results for the individual
scales: participants in the meta-dehumanization condition indicated greater Ascent
dehumanization (M = 27.22, SD = 32.57) than those in the meta-humanization condition
(M = 15.27, SD = 21.71; F (1, 229) = 10.63, p = .001, partial η2 = .04), and participants in
in the meta-dehumanization condition attributed more animalistic traits to Muslims (M =
3.89, SD = 1.34) than did those in the meta-humanization condition (M = 3.29, SD =
1.25; F (1, 229) = 12.52, p < .001, partial η2 = .05).
29
Finally, we examined whether the effects of experimental condition were
mediated by identity threat and reciprocity. Theorizing suggests that identity threat and
desires to reciprocate are intimately related (Branscombe et al., 1999a), supported by the
high correlation we observed between these two constructs. Moreover, it follows from
this theorizing that identity threat precedes desires to reciprocate a hostile perception.
Thus, using the PROCESS Macro (Hayes, 2013; Model 6) we examined a serial
mediation model, in which meta-dehumanization led to increased identity threat, leading
to a desire to reciprocate, and thus, outgroup dehumanization. This analysis revealed a
significant indirect effect, .19, 95% CI [.07, .35], suggesting the plausibility of this
model. When these mediators were taken into account, the main effect of experimental
condition on outgroup dehumanization was no longer significant, b = .02, p = .90, 95%
CI: -.24, .27.
30
Thus, individuals who received a meta-dehumanizing (vs. meta-
humanizing) prime were significantly more likely to feel threatened and to feel a desire to
reciprocate that hostile perception, together accounting for their tendency to dehumanize
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
59
the outgroup in kind.
In sum, the results of Study 7 provided empirical support for our theorizing about
the reasons that meta-perceptions about the humanity attributed to the ingroup by an
outgroup and individuals’ own attributions of humanity to that group are linked.
Specifically, our results suggested that when individuals learn that their group is
dehumanized (vs. humanized) by another group, they dehumanize that group in turn
because they become more likely to feel that their social identity is threatened and thus
show a greater desire to reciprocate that perception with hostility.
General Discussion
Ten studies document the importance of intergroup meta-dehumanization – that
is, perceiving that one’s group is dehumanized by another group. Whereas previous
research has examined a range of negative meta-perceptions (e.g., Barlow, Sibley, &
Hornsey, 2012; Kamans et al., 2009; Owuamalam et al., 2013; Vorauer et al., 1998;
Vorauer & Kumhyr, 2001), no prior research has examined individuals’ perception that
their group is perceived by another group as less than fully human. Combining past
research suggesting that individuals feel a threat to their social identity (and a desire to
reciprocate) when their ingroup is derogated (e.g., Branscombe et al., 1999a; Doosje &
A. Haslam, 2005), with research documenting the unique role of dehumanization in
shaping intergroup hostility (e.g., Andrighetto et al., 2014; Haslam, 2006; Kteily et al.,
2015), we theorized that meta-dehumanization would uniquely predict aggressive
outgroup attitudes and behavior via reciprocated outgroup dehumanization. Using large
samples across a range of cultural contexts and significant real-world conflicts (i.e., the
Israeli-Palestinian conflict, the Charlie Hebdo attacks, tensions between ethnic
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
60
Hungarians and the Roma population, and the U.S.-Iranian nuclear deal), we found
consistent support for our predictions.
Examining the intergroup relationship between Americans and each of Arabs
(Study 1a) and Muslims (Studies 1b and 1c), we began by providing evidence suggesting
that meta-dehumanization causes outgroup dehumanization rather than the reverse. In a
series of studies, we next documented the unique role of meta-dehumanization in
predicting aggressive intergroup outcomes via its effects on outgroup dehumanization:
among Americans who perceive they are dehumanized by Arabs (Study 2), ISIS (Study
5a), and Iranians (5b), among ethnic Hungarians who perceive they are dehumanized by
the Roma (Study 3), and among Israelis who perceive they are dehumanized by
Palestinians (Study 4). This relationship could not be accounted for by mere outgroup
prejudice (Study 3) or meta-prejudice (Study 4). Indeed, by controlling for both these
constructs in Studies 5a and 5b, we were able to document a novel dehumanization-
specific pathway from meta-perception to action (including aggressive intergroup
attitudes and behavior). We further showed (Study 6) that meta-humanization can
decrease outgroup dehumanization just as meta-dehumanization increases it, and
provided evidence (Study 7) for the roles of identity threat and reciprocity desires in
linking meta-perceptions about the humanity attributed to the ingroup by the outgroup on
the one hand, and the humanity ratings ascribed to the outgroup on the other. Across
studies, our effects could not be accounted for by political ideology (assessed via SDO,
RWA and conservatism in Study 2, and conservatism in Studies 3, 5a, and 5b).
31
In sum, we obtained strong evidence that meta-dehumanization is a unique meta-
perception heretofore unexamined, and one that likely contributes importantly to the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
61
perpetuation of cycles of intergroup conflict and violence (Bar-Tal, 2000; Kelman, 1987).
Indeed, one troubling implication of the dehumanization-specific pathway from meta-
perception to action identified here is that aggressive responses on the part of group
members who feel dehumanized by another group could subsequently increase meta-
dehumanization perceptions on the other side, increasing the probability of escalating
conflict. For example, Americans who think that Arabs dehumanize them are willing to
behave more aggressively towards Arabs, which could in turn drive perceptions among
Arabs that Americans view them as ‘beasts’, perpetuating a vicious cycle.
By examining dehumanization as a dynamic, interactive process involving both
perceptions and meta-perceptions, our work importantly extends dehumanization
research, which has typically examined outgroup perceptions in isolation (see Bastian &
Haslam, 2010, 2011 for exceptions in the interpersonal domain). By documenting a
novel and consequential type of meta-perception, we also extend the scope of previous
work on the harmful effects of negative meta-perceptions on intergroup interactions and
encounters (e.g., Richeson & Shelton, 2007;Vorauer et al., 1998) to the realm of
dehumanization and aggressive intergroup relations. Just as recent research highlights the
need to return to a consideration of blatant intergroup attitudes (Forscher, Cox, Graetz, &
Devine, 2015; Kteily et al., 2015), the present work suggests the importance of extending
the examination of meta-perceptions from concerns about being disliked or subtly
rejected in cross-group interactions, to a specific consideration of expectations about
being openly dehumanized. Given, for example, the tenor of race relations in the U.S.
today— exemplified by claims among members of both the African American and
policing communities that the other side openly diminishes their humanity— the time is
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
62
ripe to give meta-dehumanization greater attention.
Our research also calls for greater efforts towards identifying interventions
capable of attenuating the link between meta-dehumanization and outgroup
dehumanization. One possibility is that asking individuals to engage in perspective-taking
exercises (e.g., Hodson, Choma, & Costello, 2009) could reduce the potency of meta-
dehumanization. If individuals were tasked with putting themselves in the shoes of an
outgroup individual who was on the receiving end of a hostile action previously
committed by the ingroup (e.g., a drone strike that killed several of their family
members), they might come to understand that even ‘reasonable’ people on the other side
could, under certain circumstances, come to see one’s group as savage. This
understanding might help steer individuals who feel dehumanized by an outgroup away
from reciprocal dehumanization and aggression, and towards more productive means of
engaging with their counterpart.
Another potential approach could be to target identity threat. Study 7 suggested
that people respond to meta-dehumanization with reciprocal outgroup dehumanization in
part because meta-dehumanization causes them to feel that their ingroup identity is
derogated. Given this, the use of group affirmation interventions may prove fruitful in
reducing the link between identity threat and reciprocal dehumanization: If individuals
faced with meta-dehumanization are able to affirm the ingroup’s identity in other ways
(for example, by focusing on its high standing among third party groups), they may be
buffered from devaluation resulting from the outgroup’s perceptions (see Bendersky,
2014, for a related approach).
Beyond influencing the way in which individuals respond to meta-
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
63
dehumanization, it would be interesting to examine the extent to which individuals’ meta-
dehumanization perceptions may be biased to begin with. Research suggests that
individuals frequently overestimate the extent to which other groups’ views differ from
their own views (e.g., Keltner & Robinson, 1997); therefore, individuals may perceive
that they are more dehumanized by the outgroup than is objectively true. In Study 6, we
observed that providing Americans with information that Muslims perceive Americans in
highly humanized terms reduced Americans’ dehumanization of Muslims.
Encouragingly, our intervention was based in part on actual perceptions that Muslims
have of Americans (Esposito & Mogahed, 2007). If individuals’ baseline meta-
dehumanization perceptions are more pessimistic than is warranted by reality, providing
them with disconfirming humanizing information may be one effective route towards
improving intergroup relations.
Despite the advances made by the current work, there are also a number of open
theoretical questions that would benefit from further examination. For example, we
considered here intergroup contexts relatively high in conflict, where we hypothesized
that blatant meta-dehumanization would be particularly relevant and potent; the extent to
which our findings would generalize to less hostile intergroup relations remains
unknown. In more peaceful contexts, there may be a greater inclination among group
members to acknowledge and repair dehumanization they perceive from respected
outgroups. Moreover, there may be contexts in which meta-dehumanization is altogether
less relevant than in those considered here. We think it less likely that meta-
dehumanization of this nature would be as prevalent or relevant when considering, for
example, U.S.-German relations, despite any tensions that may exist between them. In
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
64
such contexts, where tensions may exist in parallel with a mutual regard for the other
side’s humanity, we would expect meta-prejudice to play a more prominent role. In
examining the relative roles of meta-dehumanization and meta-prejudice, future work
could experimentally manipulate both constructs in one study. In Study 1a, we
manipulated meta-dehumanization and held meta-prejudice constant. It would be
interesting to examine how individuals respond to different combinations of learning that
they are (or are not) dehumanized and disliked. In particular, it would be interesting to
examine whether learning that one’s ingroup is humanized improves intergroup relations
beyond learning that one’s ingroup is liked.
Another aspect of our work worth investigating further is the extent to which our
findings extend to more subtle forms of meta-dehumanization. In this work we
manipulated and assessed blatant forms of meta-dehumanization (and outgroup
dehumanization)
32
. These measures were particularly appropriate and relevant given the
tenor of intergroup relations in the contexts we examined. Nevertheless, contemporary
dehumanization research has also highlighted the importance of more subtle forms of
dehumanization in ‘everyday’ contexts, such as denying outgroups complex secondary
emotions, or traits central to human nature (e.g., Haslam, 2006, 2013; Leyens et al., 2000,
2007). If the outgroup does not explicitly perceive the ingroup as less evolved, but rather
subtly conveys a sense that it is less relevant (e.g., Bastian & Haslam, 2010) or that it has
fewer complex emotions (Leyens et al., 2000), it is possible that individuals will respond
primarily by feeling disheartened or saddened, rather than aggressive (see Bastian &
Haslam, 2010, 2011).
It will also be important for future work to consider how the effects of meta-
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
65
dehumanization might operate across the power spectrum. As with much research on
intergroup meta-perceptions, we focused on relatively high power groups (i.e.,
Americans, Israelis, and ethnic Hungarians) and how they feel they are perceived by their
lower-power/status counterparts. It remains to be seen how relatively low power groups
(e.g., Arabs, Roma, African Americans) respond to the perception that they are seen as
less than human by the high power outgroup. There are a number of reasons to suspect
that the processes we identified might operate similarly— and perhaps even more
powerfully— among low power groups. For one, as with high power groups,
disadvantaged groups should experience derogation of the ingroup as aversive and seek
to rectify it (Branscombe et al., 1999a), particularly with respect to meta-dehumanization,
which has deep implications for the very worth of the group. Moreover, previous work
among low power or disadvantaged groups has shown that they too can reciprocate
negative meta-perceptions by expressing negativity towards, or endorsing collective
action against, the outgroup (Branscombe et al., 1999b; Kamans et al., 2009;
Owuamalam et al., 2013). This is likely to be especially true in highly conflictual
contexts, where low power group members (e.g., Palestinians) may be un-inclined to take
on board the other side’s (e.g., Israel) hostile perceptions of them. The idea that low
power groups may be highly reactive to meta-dehumanization is also in line with research
on divergent goals in inter-racial interactions. This research suggests that disadvantaged
groups care more than advantaged groups about being respected and perceived as
competent, whereas advantaged groups care more about being seen as moral (Bergsieker,
Shelton, & Richeson, 2010; see also Schnabel, Nadler, Ullrich, Dovidio, & Carmi, 2009).
It is highly plausible that meta-dehumanization maps particularly well on to the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
66
perception that the ingroup is disrespected, which suggest that it may be especially
impactful among low power groups.
On the other hand, there are also some reasons to predict that the link between
meta-dehumanization and outgroup dehumanization might be attenuated among
disadvantaged groups. For example, it is possible that those low power group members
higher on the motivation to justify the system (Jost, Banaji, & Nosek, 2004) may
internalize even blatantly hostile perceptions that the outgroup holds of them. It is also
possible that meta-dehumanization might come to be accepted by members of low power
groups who lack collective efficacy (see Van Zomeren, Postmes & Spears, 2008). The
possibility that low power groups will be less likely than high power groups to respond to
meta-dehumanization with outgroup dehumanization is bolstered by research showing
that some disadvantaged groups (e.g., the Roma, or recent immigrants to Europe) do not
(subtly) dehumanize the outgroup, despite clear evidence of their marginalization and
maltreatment (Miranda, Gouveia-Pereira, & Vaes, 2014; see also Capozza, Andrighetto,
Di Bernardo, & Falvo, 2012; Iatridis, 2013). Future work should examine both the link
between meta-dehumanization and outgroup dehumanization as a function of group
power, as well as likely moderators such as perceptions of system illegitimacy and group
efficacy.
Future work would also benefit from further exploration of the nature of the
causal relationship between meta-dehumanization and outgroup dehumanization. In the
present work, we provided causal evidence that meta-dehumanization increases
dehumanization (Studies 1a-1b; see also Studies 6-7), but did not find evidence for the
reverse causal pathway (Study 1c). Nevertheless, despite the results we obtained in Study
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
67
1c, it would be premature to definitively rule out the possibility that outgroup
dehumanization may also cause meta-dehumanization. For example, it is possible that
individuals who dehumanize an outgroup may come to interpret that outgroup’s actions
in ways that confirm the idea that they dehumanize the ingroup, perhaps in order to
legitimize pre-meditated aggression (see Castano & Giner-Sorolla, 2006, for similar
reasoning). Alternatively, individuals who dehumanize another group may simply infer
that that group similarly dehumanizes their own through a perceived norm of reciprocal
hostility (perhaps especially when two groups are already embroiled in violent conflict
with one another). Longitudinal designs that examine the developmental course of meta-
dehumanization and dehumanization— perhaps across an episode of intense intergroup
conflict— would be well suited to continued examination of this question.
33
Beyond further examining the causal relationship between meta-dehumanization
and outgroup dehumanization, it will be useful for future research to consider the
conditions under which meta-dehumanization versus outgroup dehumanization may play
a relatively more important role in predicting intergroup aggression. For example, we
observed that the correlation between dehumanization and intergroup attitudes tended to
be somewhat stronger for ethnic Hungarians (with respect to Roma; Study 3) and Israelis
(with respect to Palestinians; Study 4) than the correlations between meta-
dehumanization and outcomes. In our American samples, on the other hand, these
associations were generally closer in magnitude (with respect to Arabs and Iranians in
particular; Studies 2 and 5b)
34
. Although purely speculative, it is possible that this may
have something to do with Americans’ status as a global hegemon. Perhaps feeling
dehumanized is particularly shocking to members of group possessing very high power,
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
68
directly provoking especially strong intergroup responses. Another difference between
these samples that may account for this pattern is the fact that the Roma and Palestinians
belong to the same social system as Hungarians and Israelis, respectively, whereas the
group examined among Americans represent distant outgroups. Feeling dehumanized by
a relatively unknown quantity may be particularly threatening, calling for especially
aggressive responses.
Finally, it will be important for future work to consider the role that ingroup
humanization may play in responses to meta-dehumanization. Vaes, Leyens, Paladino,
and Miranda (2012) suggest that outgroup dehumanization and ingroup humanization
may be independent (paralleling the distinction between ingroup love and outgroup hate;
Brewer, 1999), and may arise for different reasons. For example, these authors argue that
whereas ingroup humanization may be fairly common and is rooted in variables such as
ingroup identification (see Demoulin et al., 2009), outgroup dehumanization may depend
on contextual features such as the present of threat, and is more closely associated with
variables such as ingroup glorification (see Leidner et al., 2010).
Given the conflictual contexts we examined, and because we were examining how
“we see them” based on how “they see us”, we focused here on outgroup dehumanization
rather than ingroup humanization. Because ingroup and outgroup humanity attributions
may be correlated, however, it is theoretically possible that our results might have been
driven more by ingroup humanization in response to meta-dehumanization than by
outgroup dehumanization per se. In supplemental analyses, we controlled for ingroup
humanity attributions in order to test this possibility and obtained the same pattern of
results throughout (see Supplementary Materials). This suggests that our observed
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
69
relationship between meta-dehumanization and reciprocal outgroup dehumanization is
distinct from ingroup humanization.
On the other hand, when we examined the relationship between meta-
dehumanization and ingroup humanity attributions controlling for outgroup ratings, we
found inconsistent results: in most samples, meta-dehumanization was unassociated with
ingroup humanization. In two cases (among Israelis and Hungarians), meta-
dehumanization was surprisingly (but weakly) associated with lower ingroup humanity
ratings, suggesting that participants’ own perceptions may have been somewhat
influenced by the outgroup’s view (see also Bastian and Haslam, 2011). Consistent with
this notion, in Studies 6 and 7, we observed that Americans who learned that they were
humanized by Muslims were also more likely to humanize the ingroup (as well as
humanizing the outgroup). Nevertheless, it is important to note that association between
meta-dehumanization and dehumanization was consistent throughout when we assessed
dehumanization in relative terms (i.e., as a difference score between ingroup and
outgroup ratings). This suggests that even where meta-dehumanization might have been
associated with less ingroup humanization, this was outweighed by its association with
greater outgroup dehumanization. Future work should examine how individuals respond
to meta-dehumanization in terms of their perception of the ingroup more directly.
35
Conclusion
Our work provides clear evidence that meta-dehumanization is a novel and
important meta-perception that can contribute to the perpetuation of highly consequential
forms of intergroup conflict and hostility. Particularly given the recent spread of
movements such as ISIS, the persistence of the Israeli/Palestinian conflict, and ongoing
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
70
racial and ethnic hostilities in societies around the world, determining how to blunt meta-
dehumanization, or its effects on reciprocal dehumanization, may be critical to reducing
vicious cycles of intergroup conflict.
References
Albarello, F., & Rubini, M. (2012). Reducing dehumanisation outcomes towards Blacks: The
role of multiple categorisation and of human identity. European Journal of Social
Psychology, 42(7), 875–882. http://doi.org/10.1002/ejsp.1902
Altemeyer, B. (1988). Enemies of freedom: Understanding right-wing authoritarianism. San
Francisco, CA, US: Jossey-Bass.
Andrighetto, L., Baldissarri, C., Lattanzio, S., Loughnan, S., & Volpato, C. (2014). Human-
itarian aid? Two forms of dehumanization and willingness to help after natural disasters.
British Journal of Social Psychology, 53(3), 573–584. http://doi.org/10.1111/bjso.12066
Bain, P. G., Vaes, J., & Leyens, J. P. (2014). Humanness and dehumanization. New York:
Psychology Press.
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Mechanisms of moral
disengagement in the exercise of moral agency. Journal of Personality and Social
Psychology, 71(2), 364–374. http://doi.org/10.1037/0022-3514.71.2.364
Bandura, A., Underwood, B., & Fromson, M. E. (1975). Disinhibition of aggression through
diffusion of responsibility and dehumanization of victims. Journal of Research in
Personality, 9(4), 253–269. http://doi.org/10.1016/0092-6566(75)90001-X
Barlow, F. K., Sibley, C. G., & Hornsey, M. J. (2012). Rejection as a call to arms: Inter-racial
hostility and support for political action as outcomes of race-based rejection in majority
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
71
and minority groups. British Journal of Social Psychology, 51(1), 167–177.
http://doi.org/10.1111/j.2044-8309.2011.02040.x
Bar-Tal, D. (2000). From Intractable Conflict Through Conflict Resolution To Reconciliation:
Psychological Analysis. Political Psychology, 21(2), 351–365.
http://doi.org/10.1111/0162-895X.00192
Bastian, B., Denson, T. F., & Haslam, N. (2013). The Roles of Dehumanization and Moral
Outrage in Retributive Justice. PLoS ONE, 8(4), e61842.
http://doi.org/10.1371/journal.pone.0061842
Bastian, B., & Haslam, N. (2010). Excluded from humanity: The dehumanizing effects of
social ostracism. Journal of Experimental Social Psychology, 46(1), 107–113.
http://doi.org/10.1016/j.jesp.2009.06.022
Bastian, B., & Haslam, N. (2011). Experiencing Dehumanization: Cognitive and Emotional
Effects of Everyday Dehumanization. Basic and Applied Social Psychology, 33(4), 295–
303. http://doi.org/10.1080/01973533.2011.614132
BBC News (2012, September 14). US missions braced for protests over anti-Islam film. BBC.
Retrieved from http://www.bbc.com/news/world-middle-east-19596026.
Bendersky, C. (2014). Resolving ideological conflicts by affirming opponents’ status: The Tea
Party, Obamacare and the 2013 government shutdown. Journal of Experimental Social
Psychology, 53, 163-168.
Bergsieker, H. B., Shelton, J.N., & Richeson, J. A. (2010). To be liked versus respected:
Divergent goals in interracial interactions. Journal of Personality and Social Psychology,
99 (2), 248-264.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
72
Bourhis, R. Y., Giles, H., Leyens, J. P., & Tajfel, H. (1979). Psycholinguistic distinctiveness:
Language divergence in Belgium. Language and social psychology, 158-185.
Branscombe, N. R., Ellemers, N., Spears, R., & Doosje, B. (1999a). The context and content
of social identity threat. In N. Ellemers, R. Spears, & B. Doosje (Eds.), Social identity:
Context, commitment, content (pp. 35–58). Oxford, England: Blackwell Science.
Branscombe, N. R., Schmitt, M. T., & Harvey, R. D. (1999b). Perceiving pervasive
discrimination among African Americans: Implications for group identification and well-
being. Journal of Personality and Social Psychology, 77(1), 135–149.
http://doi.org/10.1037/0022-3514.77.1.135
Branscombe, N. R., & Wann, D. L. (1994). Collective self-esteem consequences of outgroup
derogation when a valued social identity is on trial. European Journal of Social
Psychology, 24(6), 641–657. http://doi.org/10.1002/ejsp.2420240603
Brewer, M. B. (1999). The psychology of prejudice: Ingroup love or outgroup hate? Journal
of Social Issues, 55, 429-444.
Bruneau, E., Jacoby, N., Kteily, N., Ligouri, L., & Saxe, R. (under review). Denying
humanity: Denying humanity: The distinct neural correlates of blatant dehumanization.
Manuscript submitted for publication.
Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). A new source of inexpensive, yet high-
quality, data? Perspectives on Psychological Science, 6, 3–5.
http://doi.org/10.1177/1745691610393980
Capozza, D., Andrighetto, L., Di Bernardo, G. A., & Falvo, R. (2012). Does status affect
intergroup perceptions of humanity? Group Processes and Intergroup Relations, 15 (3),
363-377.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
73
Castano, E., & Giner-Sorolla, R. (2006). Not quite human: Infrahumanization in response to
collective responsibility for intergroup killing. Journal of Personality and Social
Psychology, 90(5), 804–818. http://doi.org/10.1037/0022-3514.90.5.804
CBS News (2012, September 15). Widespread protests against U.S. over anti-Muslim film.
CBS. Retrieved from http://www.cbsnews.com/news/widespread-protests-against-us-
over-anti-muslim-film/
Costello, K., & Hodson, G. (2010). Exploring the roots of dehumanization: The role of
animal—human similarity in promoting immigrant humanization. Group Processes &
Intergroup Relations, 13(1), 3–22. http://doi.org/10.1177/1368430209347725
Cuddy, A. J., Rock, M. S., & Norton, M. I. (2007). Aid in the aftermath of Hurricane
Katrina: Inferences of secondary emotions and intergroup helping. Group Processes &
Intergroup Relations, 10(1), 107-118. http://doi.org/10.1177/1368430207071344
Demoulin, S., Leyens, J.-P., Paladino, M.-P., Rodriguez‐Torres, R., Rodriguez‐Perez, A., &
Dovidio, J. (2004). Dimensions of “uniquely” and “non‐uniquely” human emotions.
Cognition and Emotion, 18(1), 71–96. http://doi.org/10.1080/02699930244000444
Demoulin, S., Cortes, B. P., Viki, T. G., Rodriguez-Perez, A., Rodriguez-Torres, R., Paladino,
M. P., et al. (2009). The role of ingroup identification in infrahumanization. International
Journal of Psychology, 44, 4-11. doi: 10.1080/00207590802057654
Devine, P. G., Evett, S. R., & Vasquez-Suson, K. A. (1996). Exploring the interpersonal
dynamics of intergroup contact. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of
motivation and cognition, Vol. 3: The interpersonal context (pp. 423–464). New York,
NY, US: Guilford Press.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
74
Doosje, B., & Haslam, S. A. (2005). What have they done for us lately? The dynamics of
reciprocity in intergroup contexts. Journal of Applied Social Psychology, 35(3), 508–535.
http://doi.org/10.1111/j.1559-1816.2005.tb02133.x
Ellemers, N., Wilke, H., & Van Knippenberg, A. (1993). Effects of the legitimacy of low
group or individual status on individual and collective status-enhancement strategies.
Journal of Personality and Social Psychology, 64(5), 766–778.
http://doi.org/10.1037/0022-3514.64.5.766
Enyedi Zs., Ers F. & Fábián Z. (2001). Tekintélyelvség és elítélet: Újabb magyarországi
vizsgálatok. In Hunyadi Gy. & Nguyen Luu Lan Anh (Eds.) Sztereotípia-kutatás.
Hagyományok és irányok (391-402). Budapest: Eötvös Kiadó.
Esposito, J. L., & Mogahed, D. (2007). Who speaks for Islam? What a billion Muslims really
think. Gallup Press.
Forscher, P. S., Cox, W. T., Graetz, N., & Devine, P. G. (2015). The motivation to express
prejudice. Journal of Personality and Social Psychology, 109 (5), 791-812.
Frey, F. E., & Tropp, L. R. (2006). Being Seen As Individuals Versus As Group Members:
Extending Research on Metaperception to Intergroup Contexts. Personality and Social
Psychology Review, 10(3), 265–280. http://doi.org/10.1207/s15327957pspr1003_5
Goff, P. A., Eberhardt, J.L., Williams, M.J., & Jackson, M. C. (2008). Not yet human: implicit
knowledge, historical dehumanization, and contemporary consequences. Journal of
Personality and Social Psychology, 94(2), 292- 306.
Goff, P.A. Jackson, M. C., Di Leone, B.A. L., Culotta, C.M., & DiTomasso, N. A. (2014). The
essence of innocence: Consequences of dehumanizing Black children. Journal of
Personality and Social Psychology, 106 (4), 526-545.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
75
Haddock, G., Zanna, M. P., & Esses, V. M. (1993). Assessing the structure of prejudicial
attitudes: The case of attitudes toward homosexuals. Journal of Personality and Social
Psychology, 65(6), 1105–1118. http://doi.org/10.1037/0022-3514.65.6.1105
Haslam, N. (2006). Dehumanization: An Integrative Review. Personality and Social
Psychology Review, 10(3), 252–264. http://doi.org/10.1207/s15327957pspr1003_4
Haslam, N. (2014). What is dehumanization? In P. Bain, J. Vaes, & J. P. Leyens (Eds.),
Humanness and dehumanization (pp. 34–48). London, England: Psychology Press.
Haslam, N., Bain, P., Douge, L., Lee, M., & Bastian, B. (2005). More human than you:
Attributing humanness to self and others. Journal of Personality and Social Psychology,
89(6), 937–950. http://doi.org/10.1037/0022-3514.89.6.937
Haslam, N., & Loughnan, S. (2014). Dehumanization and infrahumanization. Annual Review
of Psychology, 65, 399-423.
Hayes, A. F. (2013). Introduction to Mediation, Moderation, and Conditional Process
Analysis: A Regression-Based Approach. Guilford Press.
Hetherington, M., & Suhay, E. (2011). Authoritarianism, Threat, and Americans’ Support for
the War on Terror. American Journal of Political Science, 55(3), 546–560.
http://doi.org/10.1111/j.1540-5907.2011.00514.x
Ho, A. K., Sidanius, J., Pratto, F., Levin, S., Thomsen, L., Kteily, N., & Sheehy-Skeffington,
J. (2012). Social Dominance Orientation Revisiting the Structure and Function of a
Variable Predicting Social and Political Attitudes. Personality and Social Psychology
Bulletin, 38(5), 583-606.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
76
Ho, A.K., Sidanius, J., Kteily, N., Sheehy-Skeffington, J., Pratto, F., Henkel, K., Foels, R., &
Stewart, A. (in press). The Nature of Social Dominance Orientation: Introducing the
Social Dominance Orientation-7 Scale. Journal of Personality and Social Psychology.
Hodson, G., Choma, B.L., & Costello, K (2009). Experiencing Alien-Nation: Effects of a
simulation intervention on attitudes toward homosexuals. Journal of Experimental Social
Psychology, 45, 974-978. http://doi.org/10.1016/j.jesp.2009.02.010
Hodson, G., & Costello, K. (2007). Interpersonal disgust, ideological orientations, and
dehumanization as predictors of intergroup attitudes. Psychological Science, 18(8), 691-
698. http://doi.org/10.1111/j.1467-9280.2007.01962.x
Hodson, G. & MacInnis, C.C., & Costello, K. (2014). (Over)Valuing “humanness” as an
aggravator of intergroup prejudices and discrimination. In P.G. Bain, J. Vaes, & J.-Ph.
Leyens (Eds.), Humanness and dehumanization (pp. 86-110). London: Psychology Press.
Hornsey, M. J. (2008). Social Identity Theory and Self-categorization Theory: A historical
review. Social and Personality Psychology Compass, 2(1), 204–222.
http://doi.org/10.1111/j.1751-9004.2007.00066.x
Iatridis, T. (2013). Occupational status differences in attributions of uniquely human emotions.
British Journal of Social Psychology, 52 (3), 431-449.
Jetten, J., Schmitt, M. T., & Branscombe, N. R. (2013). Rebels without a cause:
Discrimination appraised as legitimate harms group commitment. Group Processes and
Intergroup Relations, 16(2), 159-172. http://doi.org/10.1177/1368430212445075
Jost, J. T., Banaji, M. R., & Nosek, B. A. (2004). A Decade of System Justification Theory:
Accumulated Evidence of Conscious and Unconscious Bolstering of the Status Quo.
Political Psychology, 25(6), 881–919.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
77
Kamans, E., Gordijn, E. H., Oldenhuis, H., & Otten, S. (2009). What I think you see is what
you get: Influence of prejudice on assimilation to negative meta-stereotypes among
Dutch Moroccan teenagers. European Journal of Social Psychology, 39(5), 842–851.
http://doi.org/10.1002/ejsp.593
Kelman, H. C. (1987). The Political Psychology of the Israeli-Palestinian Conflict: How Can
We Overcome the Barriers to a Negotiated Solution? Political Psychology, 8(3), 347.
http://doi.org/10.2307/3791039
Keltner, D., & Robinson, R. J. (1997). Defending the status quo: Power and bias in social
conflict. Personality and Social Psychology Bulletin, 23(10), 1066-1077.
Kteily, N., Ho, A. K., & Sidanius, J. (2012). Hierarchy in the mind: The predictive power of
social dominance orientation across social contexts and domains. Journal of
Experimental Social Psychology, 48(2), 543-549.
http://doi.oorg/10.1016/j.jesp.2011.11.007
Kteily, N., Cotterill, S., Sidanius, J., Sheehy-Skeffington, J., & Bergh, R. (2014). " Not One of
Us": Predictors and Consequences of Denying Ingroup Characteristics to Ambiguous
Targets. Personality and Social Psychology Bulletin, 40(10), 1231-1247.
http://doi.org/10.1177/0146167214539708
Kteily, N., Bruneau, E., Waytz, A., & Cotterill, S. (2015). ‘The Ascent of Man’: Theoretical
and Empirical Evidence for Blatant Dehumanization. Journal of Personality and Social
Psychology, 109 (5), 901-931. doi: 10.1037/pspp0000048
Leidner, B., Castano, E., Zaiser, E., & Giner-Sorolla, R. (2010). Ingroup glorification, moral
disengagement, and justice in the context of collective violence. Personality and Social
Psychology Bulletin, 36(8), 1115-1129. http://doi.org/10.1177/0146167210376391
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
78
Leyens, J.-P., Demoulin, S., Vaes, J., Gaunt, R., & Paladino, M. P. (2007). Infra-
humanization: The Wall of Group Differences. Social Issues and Policy Review, 1(1),
139–172. http://doi.org/10.1111/j.1751-2409.2007.00006.x
Leyens, J.-P., Paladino, P. M., Rodriguez-Torres, R., Vaes, J., Demoulin, S., Rodriguez-Perez,
A., & Gaunt, R. (2000). The Emotional Side of Prejudice: The Attribution of Secondary
Emotions to Ingroups and Outgroups. Personality and Social Psychology Review, 4(2),
186–197. http://doi.org/10.1207/S15327957PSPR0402_06
MacInnis, C. C., & Hodson, G. (2012). Intergroup bias toward “Group X”: Evidence of
prejudice, dehumanization, avoidance, and discrimination against asexuals. Group
Processes & Intergroup Relations, 15(6), 725-743.
http://doi.org/10.1177/1368430212442419
Miranda, M., Gourveia-Pereira, M., & Vaes, J. (2014). When in Rome…Identification and
acculturation strategies among minority members moderate the dehumanization of the
majority outgroup. European Journal of Social Psychology, 44 (4), 327-336.
Opotow, S. (1990). Moral Exclusion and Injustice: An Introduction. Journal of Social Issues,
46(1), 1–20. http://doi.org/10.1111/j.1540-4560.1990.tb00268.x
Owuamalam, C., Issmer, C., Zagefka, H., Klaßen, M., & Wagner, U. (2014). Why Do
Members of Disadvantaged Groups Strike Back at Perceived Negativity Towards the In-
group? Journal of Community & Applied Social Psychology, 24(3), 249–264.
http://doi.org/10.1002/casp.2165
Owuamalam, C. K., Tarrant, M., Farrow, C. V., & Zagefka, H. (2013). The effect of
metastereotyping on judgements of higher-status outgroups when reciprocity and social
image improvement motives collide. Canadian Journal of Behavioural Science/Revue
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
79
Canadienne Des Sciences Du Comportement, 45(1), 12–23.
http://doi.org/10.1037/a0030012
Paolini, S., Hewstone, M., Voci, A., Harwood, J., & Cairns, E. (2006). Intergroup contact and
the promotion of intergroup harmony: The influence of intergroup emotions. In R. Brown
& D. Capozza (Eds.), Social Identities: Motivational, Emotional and Cultural Influences
(pp. 209–238). Hove, England: Psychology Press/Taylor & Francis (UK).
Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social dominance orientation:
A personality variable predicting social and political attitudes. Journal of Personality and
Social Psychology, 67(4), 741.
Richeson, J. A., & Shelton, J. N. (2007). Negotiating interracial interactions costs,
consequences, and possibilities. Current Directions in Psychological Science, 16(6), 316-
320. http://doi.org/10.1111/j.1467-8721.2007.00528.x
Schnabel, N., Nadler, A., Ullrich, J., Dovidio, J. F., & Carmi, D. (2009). Promoting
reconciliation through the satisfaction of the emotional needs of victimized and
perpetrating group members: The needs-based model of reconciliation. Personality and
Social Psychology Bulletin, 35(8), 1021-1030.
Struch, N., & Schwartz, S. H. (1989). Intergroup aggression: Its predictors and distinctness
from in-group bias. Journal of Personality and Social Psychology, 56(3), 364–373.
http://doi.org/10.1037/0022-3514.56.3.364
Tajfel, H., & Turner, J. (2001). An integrative theory of intergroup conflict. In M. A. Hogg &
D. Abrams (Eds.), Intergroup relations: Essential readings (pp. 94–109). New York, NY,
US: Psychology Press.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
80
Tessler, M. (2003). Arab and Muslim Political Attitudes: Stereotypes and Evidence from
Survey Research. International Studies Perspectives, 4(2), 175–181.
http://doi.org/10.1111/1528-3577.402005
Waytz, A., Schroeder, J., & Epley, N. (2014). The Lesser Minds Problem. In P. Bain, J. Vaes,
& J. P. Leyens (Eds.), Humanness and dehumanization (pp. 49–67). London, England:
Psychology Press
Van Zomeren, M., Postmes, T., & Spears, R. (2008). Toward an integrative social identity
model of collective action: a quantitative research synthesis of three socio-psychological
perspectives. Psychological Bulletin, 134 (4), 504-535.
Vaes, J., Leyens, J. P., Paladino, M. P., & Miranda, P. (2012). We are human, they are not:
Driving forces behind outgroup dehumanisation and the humanization of the ingroup.
European Review of Social Psychology, 23 (1), 64-106.
Viki, G., Osgood, D., & Phillips, S. (2013). Dehumanization and self-reported proclivity to
torture prisoners of war. Journal of Experimental Social Psychology, 49, 325–328.
http://dx.doi.org/10 .1016/j.jesp.2012.11.006
Vorauer, J. D., & Kumhyr, S. M. (2001). Is this about you or me? Self-versus other-directed
judgments and feelings in response to intergroup interaction. Personality and Social
Psychology Bulletin, 27(6), 706-719. http://doi.10.1177/0146167201276006
Vorauer, J. D., Hunter, A. J., Main, K. J., & Roy, S. A. (2000). Meta-stereotype activation:
Evidence from indirect measures for specific evaluative concerns experienced by
members of dominant groups in intergroup interaction. Journal of Personality and Social
Psychology, 78(4), 690–707. http://doi.org/10.1037/0022-3514.78.4.690
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
81
Vorauer, J. D., Main, K. J., & O’Connell, G. B. (1998). How do individuals expect to be
viewed by members of lower status groups? Content and implications of meta-
stereotypes. Journal of Personality and Social Psychology, 75(4), 917–937.
http://doi.org/10.1037/0022-3514.75.4.917
Vorauer, J. D., & Sakamoto, Y. (2008). Who cares what the outgroup thinks? Testing an
information search model of the importance individuals accord to an outgroup member’s
view of them during intergroup interaction. Journal of Personality and Social
Psychology, 95(6), 1467–1480. http://doi.org/10.1037/a0012631
Wildschut, T., Pinter, B., Vevea, J. L., Insko, C. A., & Schopler, J. (2003). Beyond the group
mind: a quantitative review of the interindividual-intergroup discontinuity
effect. Psychological Bulletin, 129(5), 698-722.
Yzerbyt, V. Y., Judd, C. M., & Muller, D. How do they see us? The Vicissitudes of
Metaperception in Intergroup Relations. (2009). In Demoulin, S., Leyens, J. P., &
Dovidio, J. F. (Eds.), Intergroup misunderstandings: Impact of divergent social realities
(pp. 63-100). Psychology Press.
1
Including the participants who incorrectly answered the attention check questions resulted in the same
significant pattern on our primary outcome measure (dehumanization), F (1, 208) = 4.90, p = .03, η2= .02.
2
We note that the experimental effect of the meta-dehumanization manipulation on Arab dehumanization
was unaffected by including meta-prejudice ratings as a covariate (i.e., after its association with meta-
dehumanization ratings was partialed out), F (1, 152 = 7.89, p = .006, partial η2= .05.
3
We also examined difference scores (i.e., ratings of American vs. Arab humanity) rather than absolute
dehumanization scores. We obtained similar results (F = 5.56, p = .02, η2= .035). We focus here on
absolute outgroup ratings rather than difference scores because the majority of our studies examine the
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
82
perception that the outgroup dehumanizes the ingroup in absolute (rather than relative) terms. Nevertheless,
we obtain similar patterns across studies when a relative dehumanization score is used.
4
One participant reported an age of 0, which was recoded as a missing value.
5
Results from a pre-test among a separate sample of non-Muslim native Americans on mTurk (n = 209; M
age = 36.73, SD = 12.09; 51.2% female) indicated that this prime was successful in increasing meta-
dehumanization (assessed on a 1-7 scale) relative to control (experimental condition: M = 4.95, SD = 1.68;
control condition: M = 3.34, SD = 1.58; F (1, 133) = 32.28, p < .001, partial η2 =.20). This pretest included
a third condition testing whether a meta-humanization prime (see Study 6) decreased meta-dehumanization
relative to control.
6
As in the other studies with American participants, conservatism was assessed using 3 items, reflecting
political party preference (1 = Strong Democrat; 7=Strong Republican), and economic and social
conservatism (1= Very liberal; 7= Very conservative; M = 3.29, SD = 1.56; α = .91)
7
All analyses were significant when conservatism was not a covariate.
8
Two of the animalistic traits were mentioned in the meta-dehumanization prime (i.e., ‘savage’ and ‘self-
control’). Results were the same if these two items were removed from the animalistic scale and the blatant
dehumanization composite.
9
Results from a pretest using an independent sample of non-Muslim Americans on mTurk (N = 159; M age
= 34.14, SD = 11.50; 58.5% male) indicated that this prime was successful in increasing dehumanization of
Muslims, assessed using the blatant dehumanization composite described in Study 1b. Control condition: M
= -.24, SD = .79; Experimental condition: M =.23, SD = .94; F (1, 157) = 11.56, p < .001, partial η2 = .07
(effects were also significant using each of the Ascent scale and the animalistic trait ratings alone).
10
Confidence intervals here and throughout the manuscript refer to the unstandardized coefficient.
11
The dataset included here uses the same dataset collected from this sample and reported in Kteily et al.
(2015, Study 4). However, that paper examined the effects of blatant vs. subtle dehumanization of the
Roma among Hungarians, and did not consider the role of meta-dehumanization, as it was beyond the
scope of that research.
12
One participant entered a score of -100 on this item. This response was recoded as a missing value.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
83
13
We also included the Roma Attitudes Scale (Enyedi, Ers and Fábián, 2001). Because of its substantial
overlap with the items about discriminatory policy, we did not include it as a separate outcome measure;
we note that we observe the same results using this variable.
14
Results were consistent when the entire sample was used.
15
The experimental manipulation involved reading a text that either suggested that the U.S.’ view of Israel
was declining or continued to be positive. A MANOVA examining the effect of condition across the
variables examined showed no significant effect of condition (Wilks’ Lambda = .96, F (21, 1375.98) =
1.05, p = .40), though there was a small effect on support for aggressive policies, F (3, 485) = 2.70, p =
.045, partial η2 = .016. To account for any effects of experimental condition, we residualized all variables on
condition. We further note that the patterns of interest were similar across experimental condition.
16
We also assessed support for four other potential solutions. Although we focused on the most aggressive
‘solution’ (i.e., Expulsion of Palestinians), we note that we observed very similar results on other hostile
solutions (e.g., “Absorbing the West Bank and Gaza into Israel without giving Palestinians a right to
vote”). Meta-dehumanization also predicted decreased support for a two-state solution to the conflict.
17
Due to a programming error, the set of items including militaristic counter-terrorism and beyond were
presented in the fixed order in which they appear in text.
18
62 participants indicated that they were unfamiliar with the Charlie Hebdo incident. These participants
did not receive the item asking about sending a message to families of the victims, nor the subsequent items
assessing punitiveness towards the Hebdo attackers.
19
For exploratory purposes, we also assessed four traits associated with mechanistic dehumanization (e.g.,
“mechanical and cold, like robots”). Since Ascent dehumanization is more closely related to animalistic (vs.
mechanistic) dehumanization (Kteily et al., 2015, Study 5), we excluded these items from our blatant
dehumanization composite. Nevertheless, we obtained comparable results when these items were included.
20
Participants also rated the ingroup (i.e., Americans) on these same animalistic traits (α= .85; M = 3.28,
SD = .94).
21
We assessed two further items: “This nuclear deal is like giving your dog a treat after it pees on the rug”,
and “If you think you can tame a gorilla, it’s only your fault if it returns to its true nature and bites”.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
84
Because of the conceptual overlap with animalistic dehumanization, we removed these items from our
composite. Results are unaffected when these items are included.
22
Using TurkPrime’s functionalities, we ensured that the samples in Studies 1b, 6, and 7 did not contain
repeat participants.
23
Results from the pre-test reported in Study 1b indicated that this prime was successful in decreasing
meta-dehumanization (assessed on a 1-7 scale) relative to control (experimental condition: M = 2.04, SD =
1.27; control condition: M = 3.34, SD = 1.58; F (1, 132) = 28.08, p < .001, partial η2 =.18).
24
Participants in the control condition happened to be marginally higher in political conservatism, F (1,
209) = 3.21, p = .08 (assessed prior to our manipulation). Including conservatism as a covariate did not
affect the significance of any of our analyses.
25
Two of the animalistic traits were mentioned in the meta-humanization prime (i.e., ‘cultured and
‘advanced’). Analyses excluding these two items from both the animalistic scale and the blatant
dehumanization composite yielded equivalent results.
26
Interestingly, we also observed that reading that Arabs perceived Americans as highly human increased
Americans’ ratings of their own humanity (M = 92.69, SD = 11.72) on the Ascent scale relative to control
(M = 88.81, SD = 16.04), F (1, 209) = 4.23, p = .04, partial η2 = .02, suggesting that participants’ own
ingroup perceptions were in line with the outgroup’s view. We return to the topic of ingroup humanization
in the general discussion.
27
Results were consistent when these participants were included in the analyses.
28
In a pilot correlational study (n = 178; separate from Study 5b) conducted prior to this experiment
(focusing on meta-dehumanization with respect to Iran), we further examined the potential mediating role
of perceiving a norm of conflict between the ingroup and outgroup (e.g., “It makes me think that hostility
between us is just the norm”). We observed that whereas both identity threat and reciprocity significantly
mediated the relationship between meta-dehumanization and dehumanization, perceived norm of conflict
did not. Thus, we included only identity threat and reciprocity in this study.
29
Analyses excluding traits mentioned in the primes yielded the same conclusions.
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
85
30
Because serial mediation models involve more assumptions, we also examined a simultaneous mediation
model in which identity threat and desires for reciprocity were modeled in parallel. Examining these
variables simultaneously (PROCESS Macro, Model 4), we observed that desires for reciprocity
significantly mediated the effect of experimental condition on outgroup dehumanization (indirect effect:
.24, 95% CI: .08, .44), whereas identity threat was a marginally significant mediator (indirect effect: .18,
95% CI: -.01, .40; i.e., just including 0).
31
We note that we obtained the same (or stronger) results throughout when these control variables were
not included.
32
We note that our use of the term ‘blatant’ refers to the type (and not extent) of dehumanization
perceptions (see also Kteily et al., 2015). Across our correlational samples, dehumanization levels on the
(reverse-scored) Ascent scale ranged from a score of 18 (Study 2; Americans rating Arabs) to 50 (Study 3;
Hungarians rating Roma). Although this certainly implies differences in the extent of outgroup
dehumanization across these contexts, we consider both to reflect blatant dehumanization, because in each
case participants consciously and explicitly lessen the full humanity of the outgroup on a metric that
unambiguously captures animalistic associations. We note further that in all studies reported here,
individuals rate the outgroup as significantly lower in humanity than the ingroup. For a broader theoretical
and empirical discussion of the differences between blatant dehumanization and subtler, potentially
unconscious, forms (such as the denial of secondary emotions), see Kteily et al., 2015.
33
We note that the cross-sectional models we examine in Studies 2-5b assume, in line with our
experimental findings, that meta-dehumanization causes outgroup dehumanization. Although the causal
model we propose is derived from our experimental findings and supported in our correlational data, this
does not definitively rule out the plausibility of the reverse causal order (i.e., with dehumanization causing
meta-dehumanization).
34
We thank an anonymous reviewer for highlighting this pattern of results.
35
We note that the Ascent scale may not be ideal for assessing ingroup humanization, because a sizeable
proportion of participants rate the ingroup at ceiling on the scale. Thus, an individual who feels
dehumanized by the outgroup and who may otherwise have sought to humanize the group beyond the scale
RUNNING HEAD: THEY SEE US AS LESS THAN HUMAN
86
maximum will not be able to register that response. In Study 5b, we examined animalistic trait attributions
to the outgroup and ingroup in addition to the Ascent scale. For these traits, a smaller proportion of
participants responded at floor when rating the ingroup, providing more ‘room to move’. Thus, the use of
these trait attributions may be more appropriate for those interested in directly examining ingroup
humanization. Using these trait attributions in Study 5b, we found that meta-dehumanization was
associated with outgroup dehumanization but not ingroup humanization.
Table 1. Descriptive statistics and variable intercorrelations in Study 2.
1
2
3
4
5
6
7
8
9
10
11
1. Meta-Dehumanization
-
2. Dehumanization
.38***
-
3. Surveillance of Arabs
.43***
.42***
-
4. Emotional Hostility
.51***
.59***
.48***
-
5. Torture support
.36***
.34***
.59***
.48***
-
6. Drone support
.37***
.30***
.64***
.47***
.59***
-
7. Arab distancing
.32***
.27***
.36***
.52***
.28***
.44***
-
8. Opposition to Arab
immigration
.29***
.43***
.45***
.47***
.31***
.39***
.44***
-
9. SDO
.26***
.19**
.32***
.35***
.44***
.42***
.44***
.18**
-
10. RWA
.37***
.24***
.57***
.41***
.48***
.63***
.36***
.27***
.38***
-
11. Political Conservatism
.26***
.12
.40***
.29***
.34***
.42***
.39***
.22***
.50***
.57***
-
M
43.44
18.23
31.61
34.61
19.67
40.85
55.00
84.68
26.78
43.20
39.47
Table 1
Note. *** p < .001 ** p < .01 * p < .05
SD
29.40
27.68
28.00
20.59
24.41
25.96
21.93
6.82
20.98
19.70
26.18
Table 2. Unstandardized Indirect, Direct, and Total Effects of Meta-Dehumanization on Anti-Arab Attitudes and Policy Support
via Dehumanization of Arabs in Study 2, Controlling for Political Ideology.
Surveillance of Arabs
Emotional
Hostility
Torture
Support
Drone
Support
Arab
Distancing
Opposition to Arab
Immigration
Indirect Effect
.07 [.03, .13]
.10 [.06, .16]
.04 [.01, .09]
.04 [.01, .08]
.04 [.01, .09]
.03 [.01, .05]
Direct Effect
.15 [.06, .25]
.18 [.11, .25]
.10 [.01, .19]
.09 [.002, .18]
.09 [.001, .18]
.02 [-.01, .05]
Total Effect
.23 [.13, .33]
.27 [.20, .35]
.14 [.06, .23]
.13 [.05, .21]
.13 [.04, .21]
.05 [.02, .08]
Table 2
Table 3. Descriptive statistics and variable intercorrelations in Study 3.
1
2
3
4
5
6
7
8
9
10
1. Meta-Dehumanization
-
2. Dehumanization
.35***
-
3. Prejudice
.35***
.57***
-
4. Funding to Roma
Integration
-.31***
-.48***
-.49***
-
5. Support for discrimination
.37***
.51***
.59***
-.63***
-
6. Perceptions of Roma
homogeneity
.22***
.23***
.18***
-.19***
.20***
-
7. Responses to Injustice
-.23***
-.32***
-.35***
.43***
-.56***
-.17***
-
8. Emotional Hostility
.34***
.56***
.64***
-.54***
.65***
.25***
-.44***
-
9. SDO
.35***
.50***
.41***
-.37***
.48***
.19***
-.38***
.51***
-
10. Conservatism
.17***
.23***
.27***
-.26***
.27***
.10**
-.32***
.22***
.31***
-
M
32.41
50.44
72.49
29.06
41.51
39.56
63.86
45.77
23.55
50.00
SD
27.82
37.83
25.75
25.30
21.69
26.76
30.22
24.55
21.86
25.13
Table 3
Note. *** p < .001 ** p < .01 * p < .05
Table 4. Unstandardized Indirect, Direct, and Total Effects of Meta-Dehumanization on Anti-Roma Attitudes and
Policy Support via (a) Dehumanization and (b) Prejudice towards Roma in Study 3, Controlling for Political Ideology.
Funding to Roma
Integration
Support for
Discrimination
Perceptions of
Roma
homogeneity
Responses to
Injustice
Emotional
Hostility
Indirect Effect
(Dehumanization)
-.04 [-.06, -.02]
.02 [.01, .04]
.02 [.01, .04]
-.01 [-.03,
.00]
.03 [.02, .05]
Indirect Effect
(Prejudice)
-.05 [-.08, -.03]
.06 [.04, .09]
.01 [-.01, .03]
-.04 [-.06, -
.02]
.08 [.05, .11]
Indirect Effect
(Total)
-.10 [-.13, -.07]
.09 [.06, .11]
.03 [.01, .05]
-.05 [-.08, -
.03]
.11 [.08, .15]
Direct Effect
-.08 [-.13, -.03]
.09 [.05, .13]
.13 [.06, .20]
-.05 [-.12,
.02]
.04 [-.00, .09]
Total Effect
-.17 [-.23, -.12]
.17 [.13, .22]
.16 [.09, .22]
-.10 [-.17, -
.03]
.16 [.11, .21]
Table 4
Table 5. Descriptive statistics and variable intercorrelations in Study 4.
Note. *** p < .001 ** p < .01 * p < .05
1
2
3
4
5
6
7
1. Meta-Dehumanization
-
2. Meta-Prejudice
.48***
-
3. Dehumanization
.30***
.25***
-
4. Support for Negotiations
-.27***
-.19***
-.37***
-
5. Expulsion of Palestinians
.25***
.19***
.27***
-.48***
-
6. Support for Aggressive Policies
.30***
.23***
.48***
-.56***
.55***
-
7. Emotional Hostility
.27***
.36***
.40***
-.55***
.47***
.57***
-
M
41.34
72.32
55.31
58.91
36.46
41.09
65.71
SD
28.00
25.31
27.10
33.30
36.24
25.33
19.12
Table 5
Table 6. Unstandardized Indirect, Direct, and Total Effects of Meta-Dehumanization on Anti-Palestinian Attitudes and
Policy Support via (a) Dehumanization in Study 4, Controlling for Meta-Prejudice.
Support for Negotiations
Expulsion of
Palestinians
Support for
Aggressive
Policies
Emotional
Hostility
Indirect Effect
(Dehumanization)
-.08 [-.13, -.05]
.06 [.03, .11]
.09 [.05, .13]
.05 [.03, .08]
Direct Effect
-.19 [-.30, -.07]
.21 [.08, .33]
.14 [.06, .22]
.04 [-.03, .10]
Total Effect
-.27 [-.39, -.15]
.27 [.14, .39]
.22 [.14, .31]
.09 [.02, .15]
Table 6
Table 9. Descriptive statistics and variable intercorrelations in Study 5a.
1
2
3
4
5
6
7
8
9
10
11
12
13
1. Meta-Dehumanization
-
2. Meta-Prejudice
.38***
-
3. Dehumanization
.26***
.04
-
4. Prejudice
.33***
.38***
.33***
-
5. Drone support
.10
.07
.27***
.25***
-
6. Militaristic Counter-
Terrorism
.24***
.15**
.39***
.31***
.69***
-
7. Opposition to Muslim
Immigration
.11*
-.03
.31***
.14**
.40***
.39***
-
8. Signing Anti-ISIS Petitions
.22***
.09
.24***
.19***
.46***
.55***
.27***
-
9. Anti-Islamic Extremism
Fund Disbursement
.11*
.03
.29***
.19***
.54***
.61***
.46***
.42***
-
10. Encouragement of US
soldiers fighting ISIS
.11*
.08
.12*
.01
.23***
.23***
.11*
.24***
.24***
-
11. Supportive messages to
families of Hebdo victims
.07
.12*
.05
.14*
.01
.02
-.06
.14*
.04
.40***
-
12. Punitiveness towards
Hebdo attackers
.29***
.14*
.39***
.33***
.52***
.73***
.26***
.46***
.43***
.23***
.12
-
Table 7
Note. *** p < .001 ** p < .01 * p < .05
13. Conservatism
.08
-.02
.18***
.01
.42***
.48***
.21***
.40***
.39***
.16**
-.01
.38***
-
M
70.52
90.53
38.17
85.87
45.75
49.98
85.05
5.92
32.95
19.73
25.33
61.90
43.45
SD
21.06
17.02
36.55
23.70
24.55
24.65
7.49
34.85
32.97
39.85
43.56
30.34
26.80
Table 8. Unstandardized Indirect, Direct, and Total Effects of Meta-Dehumanization on Anti-ISIS Attitudes and Behavior via (a) Dehumanization and (b) Prejudice in Study
5a, Controlling for Meta-Prejudice and Political Ideology.
Drone Support
Militaristic
Counter-
Terrorism
Opposition to
Muslim
Immigration
Signing
Anti-ISIS
Petitions
Anti-Islamic
Extremism
Fund
Disbursement
Encouragement
of US soldiers
fighting ISIS
Supportive
Messages
to families
of Hebdo
victims
Punitiveness
towards
Hebdo
attackers
Indirect Effect
(Dehumanization)
.04 [.01, .09]
.07 [.04, .13]
.02 [.01, .04]
.05 [.002,
.12]
.08 [.03, .15]
.003 [.00, .01]
.001 [-
.003, .004]
.10 [.05, .17]
Indirect Effect
(Prejudice)
.05 [.02, .10]
.05 [.02, .09]
.01 [-.002,
.04]
.04 [.01,
.09]
.05 [.01, .09]
-.002 [-.006,
.002]
.004 [.001,
.01]
.05 [.05, .17]
Indirect Effect
(Total)
.10 [.04, .16]
.12 [.07, .18]
.03 [.02, .05]
.09 [.04,
.17]
.12 [.07, .21]
.002 [-.003,
.007]
.004 [-
.001, .01]
.14 [.08, .23]
Direct Effect
-.05 [-.17, .08]
.08 [-.03, .19]
.02 [-.02, .06]
.21 [.03,
.38]
.01 [-.15, .18]
.009 [-.006,
.02]
-.001 [-.01,
.01]
.20 [.04, .37]
Total Effect
.05 [-.07, .17]
.20 [.09, .32]
.05 [.006,
.08]
.29 [.13,
.46]
.14 [-.03, .30]
.01 [-.004, .03]
.002 [-.01,
.02]
.35 [.18, .51]
Table 8
Table 9. Descriptive statistics and variable intercorrelations in Study 5b.
Note. *** p < .001 ** p < .01 * p < .05
1
2
3
4
5
6
7
8
1. Meta-Dehumanization
-
2. Meta-Prejudice
.70***
-
3. Blatant Dehumanization
Composite
.64***
.50***
-
4. Prejudice
.48***
.46***
.67***
-
5. Opposition to the Iran
Nuclear Deal
.60***
.50***
.62***
.46***
-
6. Aggressive Actions towards
Iranians
.60***
.49***
.63***
.54***
.73***
-
7. Signing Anti-Nuclear Deal
Petitions
.35***
.25***
.36***
.26***
.57***
.43***
-
8. Conservatism
.26***
.24***
.28***
.27***
.50***
.43***
.38***
-
M
44.10
66.49
.00
47.12
45.19
29.46
-7.10
39.73
SD
28.03
26.98
.89
30.00
23.59
22.19
39.59
24.22
Table 9
Table 10. Unstandardized Indirect, Direct, and Total Effects of Meta-Dehumanization on Anti-Iranian Attitudes
and Behavior via (a) Dehumanization and (b) Prejudice in Study 5b, Controlling for Meta-Prejudice and Political Ideology.
Opposition to Iranian
Nuclear Deal
Aggressive Actions
towards Iranians
Signing anti-Nuclear
deal Petitions
Indirect Effect
(Dehumanization)
.15 [.09, .22]
.13 [.07, .21]
.13 [.01, .24]
Indirect Effect (Prejudice)
.00 [-.03, .03]
.03 [.002, .07]
.00 [-.05, .06]
Indirect Effect
(Total)
.15 [.10, .21]
.16 [.10, .23]
.13 [.03, .22]
Direct Effect
.23 [.13, .33]
.21 [.11, .31]
.29 [.07, .51]
Total Effect
.38 [.28, .47]
.37 [.28, .46]
.42 [.22, .62]
Table 10
Figure 1. Conceptual diagram of the proposed theoretical model, in which Meta-
dehumanization influences aggressive attitudes and behavior via outgroup dehumanization,
controlling for Meta-Prejudice, outgroup prejudice, and political ideology (not shown).
Primary variables (blue) and paths (black) of interest appear highlighted.
Figure 1
“Using the image below, indicate using the sliders how evolved you consider the
average members of each group to be: ”
Figure 2. The ‘Ascent of Man’ measure of blatant dehumanization. This figure was
originally published in Kteily, Bruneau, Waytz, & Cotterill, 2015, JPSP, Figure 1.
Figure 2
Figure 3. Path model showing effects of Americans’ meta-dehumanization
perceptions on support for torture of Arabs via outgroup dehumanization in Study
2, controlling for political ideology (not shown). Numbers reflect unstandardized
beta coefficients. *** p <.001; ** p < .01
Figure 3
Figure 4. Path model showing effects of majority Hungarians’ meta-dehumanization
on support for aggressive policies towards the Roma in Study 3 via outgroup
dehumanization and prejudice, controlling for political ideology (not shown).
Numbers reflect unstandardized beta coefficients. *** p < .001
Figure 4
Figure 5. Path model showing effects of Israelis’ meta-dehumanization on support
for Aggressive policies towards Palestinians via dehumanization of Palestinians in
Study 4, controlling for meta-prejudice. Numbers reflect unstandardized beta
coefficients. *** p < .001; ** p < .01; * p < .05
Figure 5
Figure 6. Path model showing effects of Americans’ meta-dehumanization
perceptions on signing anti-ISIS petitions via dehumanization of and prejudice
towards ISIS in Study 5a, controlling for meta-prejudice and political ideology (not
shown). Numbers reflect unstandardized beta coefficients. *** p < .001; * p < .05
Figure 6
Supplemental Material - Integral
Click here to access/download
Supplemental Material - Integral
Supplementary Table 3 for Metadehum Revision Nov
2015.docx
Supplemental Material - Integral
Click here to access/download
Supplemental Material - Integral
Supplemental Materials (Prime for Study 1b).docx
Supplemental Material - Integral
Click here to access/download
Supplemental Material - Integral
Supplemental Materials (Prime for Study 1c).docx
Supplemental Material - Integral
Click here to access/download
Supplemental Material - Integral
Supplemental Materials (Prime for Study 6).docx
Supplemental Material - Integral
Click here to access/download
Supplemental Material - Integral
Supplemental Analyses.docx
Supplemental Material - Integral
Click here to access/download
Supplemental Material - Integral
Supplemental Appendix (items).docx
