images, dehumanization and south koreans' attitudes towards...
TRANSCRIPT
-
Images, Dehumanization, and South Koreans’ Attitudes
towards Unification
In KINU 통일의식조사 2018: 남북평화 시대의 통일의식. The KINU Unification Survey2018: New Era of Peace in Korean Peninsular and Unification Attitudes. Ed. Sang Sin
Lee. Seoul: Korea Institute for National Unification, 2019.
Pre-publication version: September 3, 2018
Joshua D. Kertzer
Paul Sack Associate Professor of Political EconomyDepartment of Government, Harvard University.1737 Cambridge St, Cambridge MA, 02138, USA.
Email: [email protected]: http:/people.fas.harvard.edu/˜jkertzer/
mailto:[email protected]://people.fas.harvard.edu/~jkertzer/
-
Introduction
At both the elite and mass level, international politics is driven by the perceptions that
actors have about one another.1 Just as the images leaders have about other countries
shape the types of foreign policies they pursue, the images that citizens have about other
countries affects the types of foreign policies they support.2 One of the key barriers to conflict
resolution, for example, involves changing the images that citizens on each side of the conflict
have of one another; when citizens on one side view their opponents as inferior or less than
human, it legitimizes the use of violence and retaliatory aggression.3 Yet just as images can
shape the likelihood of intergroup conflict, they can also affect the likelihood of cooperation.
For example, the key causal mechanism in classic theories of economic integration — and
central rationale behind economic and cultural exchanges — is that as individuals from
different groups interact with one another, their images of each other change, making future
conflict less likely.4
This chapter borrows from psychological research on images, stereotypes, and dehu-
manization to ask three questions.5 First, what images do South Koreans have about North
Koreans? Second, what demographic factors and political orientations explain variation in
these perceptions? Third, how do these perceptions relate to South Koreans’ attitudes about
unification? These questions have clear substantive importance, but are also of interest for
theoretical reasons, since some of these theoretical frameworks have yet to be systematically
tested in the context of inter-Korean relations.
The discussion below has four parts. It begins by reviewing how political psychologists
1. Robert Jervis, Perception and Misperception in International Politics (Princeton, NJ: Princeton Uni-versity Press, 1976); Richard K. Herrmann, Perceptions and Behavior in Soviet Foreign Policy (Pittsburgh:University of Pittsburgh Press, 1985).
2. Richard K. Herrmann and Michael P. Fischerkeller, “Beyond the Enemy Image and Spiral Model:Cognitive-Strategic Research After the Cold War,” International Organization 49, no. 3 (1995): 415–50.
3. Ifat Maoz and Clark McCauley, “Threat, dehumanization, and support for retaliatory aggressive poli-cies in asymmetric conflict,” Journal of Conflict Resolution 52, no. 1 (2008): 93–116; Emile Bruneau andNour Kteily, “The enemy as animal: Symmetric dehumanization during asymmetric warfare,” PLOS ONE12, no. 7 (July 2017): 1–20.
4. Karl W. Deutsch, Nationalism and Social Communication (Cambridge, MA: MIT Press, 1953);Thomas F. Pettigrew, “Intergroup Contact Theory,” Annual Review of Psychology 49 (1998): 65–85.
5. Susan T. Fiske et al., “A Model of (Often Mixed) Stereotype Content: Competence and WarmthRespectively Follow From Perceived Status and Competition,” Journal of Personality and Social Psychology82, no. 6 (2002): 878–902; Nick Haslam, “Dehumanization: An Integrative Review,” Personality and SocialPsychology Review 10, no. 3 (2006): 252–64.
1
-
have studied images in international relations, focusing on two theoretical frameworks in par-
ticular: stereotypes, and dehumanization. It then investigates the content of South Koreans’
images of North Koreans, finding that South Koreans perceive their Northern counterparts
as being not particularly warm, but also not especially cold, and significantly less compe-
tent. It also finds evidence of South Korean dehumanization of North Koreans, specifically
with respect to what the dehumanization literature refers to as “human nature” traits,
rather than “human uniqueness” traits. South Koreans don’t dehumanize North Koreans
by viewing them animalistically, but by viewing them mechanistically instead; in as much
as North Koreans are perceived as somehow less than human, it is that they are perceived
as nonhuman, rather than subhuman.
Third, it investigates the demographic correlates of these perceptions in a multivariate
context, identifying the types of respondents more or less likely to embrace each stereotype.
It finds in particular the existence of stark generational differences, with younger South
Koreans displaying more negative stereotypes about North Koreans than older generations,
as do the less wealthy, and more conservative. Younger South Koreans are also more likely
to dehumanize the North with respect to human nature traits, as do the less educated.
Finally, it explores the political consequences of these perceptions, showing that vari-
ation in perceived warmth and human nature dehumanization is significantly associated
both with attitudes towards reunification, operationalized both in terms of general support
for reunification, and as willingness to personally sacrifice for unification. These effects hold
even when controlling for a variety of demographic and political characteristics, demonstrat-
ing the importance of examining perceptions of outgroups in the context of inter-Korean
relations.
Images and perceptions in international politics
One of the central insights of psychological theories of international politics is their emphasis
on actors’ perceptions.6 In privileging the ideational over the material, these theories can be
6. Robert Jervis, Richard Ned Lebow, and Janice Gross Stein, Psychology and Deterrence (Baltimore:John Hopkins University Press, 1985); Keren Yarhi-Milo, Knowing the Adversary: Leaders, Intelligence,
2
-
contrasted with realist theories of international politics that focus more on the structure of
the international system as a whole, and less on the perceptions of the units that populate
it.7 Theories of international politics that emphasize perceptions argue that state behavior
(at the elite level) or policy preferences (at the mass public level) cannot always be strictly
deduced from structural factors like the balance of power, since actors can define situations
in myriad ways, such that actors’ perceptions are not simply reducible to their environment.
If perceptions were completely idiosyncratic, they would be difficult to study systemat-
ically. Image theory, however, argues that even if the specific content of actors’ perceptions
vary, the structure of the perceptions remain the same; even though we have a large number
of perceptions and beliefs about our relationships with other actors, we tend to categorize
them along a small number of common dimensions that work together to form a holistic
judgment of who the target is and what they want. Early work on image theory argued that
actors’ images of others in international politics essentially reduced to a single dimension:
whether the target’s intentions were understood to be hostile or friendly, and thus, whether
it poses a threat, or an opportunity.8
More recent work on images in international politics has gone beyond the simple dis-
tinction between friends and foes to argue that images of other countries are structured
along three dimensions. First, we analyze others based upon the degree of perceived threat
or opportunity they pose, such as whether they pose a threat, a chance for mutual gain,
or an opportunity for exploitation; this is the valence dimension central to earlier work on
images. Importantly, however, we also categorize targets on two additional dimensions: their
relative power, and perceived status. Along the second dimension, a target can be evaluated
as less powerful, equally powerful, or more powerful than the observer; along the third, the
target can be perceived as of lower status, equal status, or higher status.9 Thus, an enemy
and Assessment of Intentions in International Relations (Princeton, NJ: Princeton University Press, 2014);Joshua D. Kertzer, Resolve in International Politics (Princeton, NJ: Princeton University Press, 2016).
7. Kenneth N. Waltz, Theory of International Politics (Boston: McGraw-Hill, 1979).8. Kenneth E Boulding, “National images and international systems,” Journal of Conflict Resolution 3,
no. 2 (1959): 120–131.9. Herrmann and Fischerkeller, “Beyond the Enemy Image and Spiral Model: Cognitive-Strategic Research
After the Cold War”; Richard K. Herrmann, “Perceptions and Image Theory in International Relations,”in Oxford Handbook of Political Psychology, 2nd Ed., ed. Leonie Huddy, David O. Sears, and Jack S. Levy(Oxford University Press, 2013), 334–363.
3
-
image, for example, is attributed to an actor perceived of being comparable in power, with
threatening intentions, of equal status, whereas a colony image is attributed to an actor
weaker in power, lower in status, and who offers an opportunity for exploitation.
The power of images is that they are linked to behavioral tendencies. If we think of
another group as our enemy (as the Americans did towards the Soviets during the second
half of the twentieth century), we treat it very differently than if we think of it as a colony
(as the Japanese did towards the Koreans in the first half of twentieth century). Images thus
not only shape how we wish the target to be treated, but function as cognitive schema, with
implications for how we process information.10 They also have an affective component: it is
because of the emotions that images generate that they shape behavior.
Stereotypes: warmth and competence
In this sense, how we understand images of others in international politics bears a strong
resemblance to how psychologists understand social stereotypes more generally.11 Whereas
earlier psychological work on person perception argued that we form impressions of others
along a single dimension, current research on stereotype structure, most notably Fiske et
al.’s stereotype content model, argues our perceptions of other actors are clustered along
two dimensions: perceived warmth (a function of perceived intentions, thus corresponding
to image theory’s interest in whether the target poses a threat or an opportunity), and
perceived competence (a function of perceived capabilities, thus corresponding to image
theory’s interest in whether the target is more or less powerful than the observer).12 The
cognitive processes through which we think about other countries is thus similar to those
through which we think about other people.13
10. Richard K. Herrmann et al., “Images in International Relations: An Experimental Test of CognitiveSchemata,” International Studies Quarterly 41, no. 3 (1997): 403–433; Emanuele Castano, Alain Bonacossa,and Peter Gries, “National Images as Integrated Schemas: Subliminal Primes of Image Attributes ShapeForeign Policy Preferences,” Political Psychology 37, no. 3 (2016): 351–366.
11. Gordon W. Allport, The nature of prejudice (Cambridge, MA: Addison-Wesley, 1954); Susan T. Fiske,Amy J. C. Cuddy, and Peter Glick, “Universal dimensions of social perception: Warmth and competence,”Trends in Cognitive Sciences 11 (2007): 77–83.
12. Solomon E. Asch, “Forming impressions of personality,” Journal of Abnormal and Social Psychology41 (1946): 1230–1240; Fiske et al., “A Model of (Often Mixed) Stereotype Content: Competence and WarmthRespectively Follow From Perceived Status and Competition.”
13. Kathleen M. McGraw and Thomas M. Dolan, “Personifying the state: Consequences for attitude for-mation,” Political Psychology 28, no. 3 (2007): 299–327.
4
-
How do South Koreans perceive North Koreans in this framework? To answer this ques-
tion, the analysis below studies the first two dimensions of South Koreans’ images of North
Koreans based on instrumentation from the stereotype content model, asking respondents
the extent to which they characterize North Korean people as warm, and as competent;
to provide a reference point, respondents were also asked the same questions about South
Koreans, as well as the Japanese. This theoretical model has been tested in Korea before,
most notably as part of a large cross-national study by Cuddy et al., though the approach
here differs both through the use of a nationally representative sample, and in focusing on
stereotypes of citizens in other countries, rather than groups within domestic society.14
Dehumanization: human uniqueness and human nature
A key related construct in intergroup relations is dehumanization, or the tendency for in-
dividuals to deny human characteristics to members of other groups.15 Dehumanization
ranges in severity. Its most extreme form, blatant dehumanization, involves literally com-
paring members of outgroups to animals or insects, thereby weakening moral constraints
against violence by denying the target’s humanity.16 Its more subtle form involves being
relatively less likely to attribute characteristics we associate with humanness to members of
outgroups.17 Dual models of dehumanization argue that there are two broad types of char-
acteristics we tend to associate with human essence.18 The first cluster, human uniqueness,
concerns the attributes we perceive as distinguishing humans from other animals, such as
intelligence, self-control, moral sensibility, and secondary emotions.19 The second cluster,
human nature, involves those traits we perceive as distinguishing humans from machines,
robots and inanimate objects: warmth, agency and individuality, cognitive openness, and
14. Amy JC Cuddy et al., “Stereotype content model across cultures: Towards universal similarities andsome differences,” British Journal of Social Psychology 48, no. 1 (2009): 1–33.
15. Haslam, “Dehumanization: An Integrative Review.”16. Nour Kteily et al., “The ascent of man: Theoretical and empirical evidence for blatant dehumanization,”
Journal of personality and social psychology 109, no. 5 (2015): 901.17. Jacques-Philippe Leyens et al., “Psychological essentialism and the differential attribution of uniquely
human emotions to ingroups and outgroups,” European Journal of Social Psychology 31, no. 4 (2001): 395–411.
18. Nick Haslam and Steve Loughnan, “Dehumanization and Infrahumanization,” Annual Review of Psy-chology 65, no. 1 (2014): 399–423.
19. Whereas we attribute primary emotions like anger and anxiety to both humans and non-human animals,we tend to attribute secondary emotions like embarassment or regret only to humans.
5
-
emotionality. Denying human uniqueness to an outgroup is an animalistic form of dehuman-
ization — viewing others as subhuman — whereas denying human nature to an outgroup is
a mechanistic form of dehumanization — viewing others as nonhuman.
Since denying human uniqueness traits to outgroups involves perceiving others as low in
competence, and denying human nature traits to outgroups involves perceiving others as low
in warmth, dehumanization and stereotype content models are interrelated.20 Dehumaniza-
tion is related to the relative status dimension of image theory as well; imperial powers often
justified their interventions in their colonies with dehumanizing rhetoric, viewing colonial
inhabitants of being backwards and incapable of governing themselves.21
Do South Koreans dehumanize North Koreans? And if so, does it involve denying human
nature traits, or human uniqueness traits? In the analysis below, we address these questions
by including instrumentation from the dual model of dehumanization.22 Respondents were
presented with a list of eight traits previous research has found to vary in human nature,
human uniqueness (and, to avoid valence effects, desirability), and asked to indicate the
extent to which each trait described the people of South Korea, North Korea, and Japan.
By calculating the respondent-level differences in trait ratings between the ingroup and each
of the two outgroups, and averaging these differences for the traits high in human uniqueness,
versus those high in human nature, we obtain measures of dehumanization along each of
the two dimensions, for each of the two outgroups.
Results
What are South Koreans’ perceptions of North Koreans?
We begin by looking at the structure of South Koreans’ perceptions of North Koreans,
in Figure 1. For purposes of comparability, the left-hand panel displays South Koreans’
perceptions of the South Korean people as a whole, while the right-hand panel displays
20. Lasana T. Harris and Susan T. Fiske, “Dehumanizing the lowest of the low: neuroimaging responsesto extreme out-groups,” Psychological Science 17, no. 10 (2006): 847–853.
21. Herrmann and Fischerkeller, “Beyond the Enemy Image and Spiral Model: Cognitive-Strategic ResearchAfter the Cold War.”
22. Nick Haslam and Paul Bain, “Humanizing the Self: Moderators of the Attribution of Lesser Humannessto Others,” Personality and Social Psychology Bulletin 33, no. 1 (2007): 57–68.
6
-
Fig
ure
1:
South
Kore
an
s’im
ages
1. S
outh
Kor
eans
2. N
orth
Kor
eans
3. J
apan
ese
025
5075
100
025
5075
100
025
5075
100
0255075100
Competence
Warmth
Fig
ure
1p
lots
surv
eyre
spon
den
ts’
per
cep
tion
sof
citi
zen
sfr
om
the
Rep
ub
lic
of
Kore
a,
Nort
hK
ore
a,
an
dJap
an
.T
he
y-a
xis
mea
sure
sth
ep
erce
ived
warm
thof
each
ster
eoty
pe
targ
et,
wh
ile
the
x-a
xis
mea
sure
sth
ep
erce
ived
com
pet
ence
;ea
chd
ot
dep
icts
the
resp
on
seof
an
ind
ivid
ual
surv
eyre
spon
den
t,ji
tter
edh
ere
top
rod
uce
legib
ilit
y.T
he
solid
blu
elin
esco
rres
pon
dto
the
aver
age
level
of
warm
than
dco
mp
eten
ceatt
rib
ute
dto
agiv
enst
ereo
typ
eta
rget
.T
hu
s,th
efi
rst
pan
elsh
ow
sth
at
most
Kore
an
sh
ave
ap
osi
tive
image
of
thei
rfe
llow
citi
zen
s,ra
tin
gth
emh
igh
inb
oth
warm
than
dco
mp
eten
ce.
Inco
ntr
ast
,th
en
ext
two
pan
els
show
that
Kore
an
sh
ave
more
neg
ati
ve
images
of
citi
zen
sfr
om
Nort
hK
ore
aan
dJap
an
.N
ort
hK
ore
an
sare
per
ceiv
edas
bei
ng
neu
tral
inw
arm
th,
an
dsl
ightl
yb
elow
the
scale
mid
poin
tin
com
pet
ence
;th
eJap
an
ese
are
per
ceiv
edas
bei
ng
alm
ost
as
com
pet
ent
as
the
Kore
an
s,b
ut
less
warm
.
7
-
South Koreans’ perceptions of the Japanese. The y-axis on each plot measures the perceived
warmth of each stereotype target, while the x-axis measures the perceived competence. Each
grey dot depicts the response of each individual survey respondent, while the solid blue lines
correspond to the average values attributed to a given target on each dimension. The left-
hand panel shows that most Koreans have a positive image of their fellow citizens, perceiving
them as being relatively high in both warmth and competence — as is expected given that
respondents were being asked to evaluate their ingroup. In the stereotype content model, the
emotion elicited by this particular combination of warmth and competence is admiration.23
In contrast, the other two panels show that South Koreans have more negative images of
both North Koreans and Japanese, with a much higher variance. The right-hand panel shows
that South Koreans perceive the Japanese to be almost as competent as the South Koreans
are, but significantly less warm. In the stereotype content model, this particular combination
is associated with envy, the result of competition with a high-status outgroup. The middle
panel shows that South Koreans perceive the North Koreans to be similarly cold as the
Japanese, but to be significantly less competent, indicative of a perception of lower status.
While the emotion this particular combination is associated with is contempt, on the whole
the averages on each dimension are relatively close to the midpoint of each scale, such that
the stereotype is not particularly extreme. Additionally, it is noteworthy that the dispersion
of responses is greater for North Korea than for the other stereotype targets; this indicates
a relative degree of dissensus amongst the respondents in their images of North Koreans
compared to South Koreans or Japanese. This raises the question of which types of South
Korean respondents — younger or older, richer or poorer, more liberal or more conservative
— espouse which types of stereotypes, a question we return to in the following section.
Figure 2 displays the results of the dehumanization measures. Here, since the measures
are all calculated relative to the ingroup, there are two panels; the left-hand panel depicts
South Koreans’ dehumanization scores for the Japanese, and the right-hand panel shows
South Koreans’ dehumanization scores for North Koreans. The x-axis measures human
uniqueness dehumanization: the extent to which outgroup members are attributed rela-
23. Fiske et al., “A Model of (Often Mixed) Stereotype Content: Competence and Warmth RespectivelyFollow From Perceived Status and Competition.”
8
-
Fig
ure
2:
Dis
trib
uti
on
of
deh
um
an
izati
on
score
s
Japanese
Nor
th K
orea
ns
-25
025
50-25
025
50-40
-20020406080
Deh
uman
izat
ion:
hum
an u
niqu
enes
s
Dehumanization: human nature
Fig
ure
2p
lots
surv
eyre
spon
den
ts’d
ehu
man
izati
on
score
sfo
rth
eJap
an
ese
an
dN
ort
hK
ore
an
s.T
he
x-a
xis
mea
sure
sth
ed
egre
eof
deh
um
an
izati
on
inte
rms
of
ass
essm
ents
of
hu
man
un
iqu
enes
s—
the
trait
san
dm
enta
lst
ate
sse
enas
sep
ara
ting
hu
man
sfr
om
an
imals
.T
he
y-a
xis
mea
sure
sth
ed
egre
eof
deh
um
an
izati
on
inte
rms
of
ass
essm
ents
of
hu
man
natu
re—
the
trait
sse
enas
dis
tin
gu
ish
ing
hu
man
sfr
om
mach
ines
.T
hu
s,th
ex-a
xis
corr
esp
on
ds
toan
imalist
icfo
rms
of
deh
um
an
izati
on
,w
hile
the
y-a
xis
corr
esp
on
ds
tom
ech
an
isti
cfo
rms
of
deh
um
an
izati
on
;in
both
case
s,h
igh
ervalu
esco
rres
pon
dw
ith
hig
her
level
sof
deh
um
an
izati
on
.T
he
plo
tsh
ow
sth
at
nei
ther
the
Jap
an
ese
nor
the
Nort
hK
ore
an
sare
deh
um
an
ized
signifi
cantl
yin
term
sof
hu
man
un
iqu
enes
s,b
ut
that
each
pop
ula
tion
isd
ehu
man
ized
sub
stanti
ally
inte
rms
of
hu
man
natu
re,
or
mec
han
isti
cd
ehu
man
izati
on
.
9
-
tively fewer traits that distinguish humans from animals. In contrast, the y-axis measures
human nature dehumanization — the extent to which outgroup members are attributed
relatively fewer traits that distinguish humans from machines. As before, each dot repre-
sents the scores of a different respondent. Interestingly, we find similar patterns for both
outgroups. Neither the Japanese nor the North Koreans are dehumanized animalistically,
and are attributed human uniqueness traits like conscientiousness and politeness at roughly
the same rate as South Koreans attribute to their fellow citizens. However, South Korean
respondents dehumanize both the Japanese and the North Koreans with respect to human
nature traits, the characteristics that distinguish humans from robots or machines, such as
curiosity or friendliness. As was the case with the stereotype content results in Figure 1,
however, there is considerable variation in dehumanization scores in Figure 2: some South
Koreans dehumanize North Koreans significantly more than others. It is explaining this type
of variation that we turn to next.
What shapes South Koreans’ perceptions of North Koreans?
The analysis has thus far shown that South Koreans have relatively neutral stereotypes
of North Koreans, while viewing them as significantly less competent than either South
Koreans or Japanese, and significantly colder than South Koreans. We have also shown that
South Koreans dehumanize North Koreans mechanistically, denying them in relative terms
some of the traits that we associate with human nature. At the same time, we found evidence
of significant variation across both respondents’ stereotype and dehumanization scores.
Table 1 attempts to explain this variation systematically, presenting the results of a se-
ries of linear regression models that examine respondents’ warmth and competence scores for
North Koreans as a function of a series of demographic variables and political orientations.
Models 1-3 present results for the warmth scores, while models 4-6 present the results for the
competence scores. For purposes of comparability, all variables are scaled from 0-1 except
for age (where a one-unit increase corresponds to one additional year of age). The first three
models show that male respondents perceive North Koreans as significantly colder than fe-
male North Koreans; depending on the other covariates in the model, the effect of gender
10
-
Tab
le1:
Pre
dic
tors
of
ster
eoty
pe
conte
nt
Warm
thC
om
pet
ence
(1)
(2)
(3)
(4)
(5)
(6)
Mal
e−
2.168∗
−1.9
48∗
−1.8
90∗
−0.9
19
−0.7
66
−0.7
88
(1.1
32)
(1.1
36)
(1.1
35)
(1.1
06)
(1.1
11)
(1.1
12)
Age
0.088∗
∗0.0
90∗∗
0.0
75∗
0.1
10∗∗
∗0.1
09∗
∗∗0.1
10∗∗
∗
(0.0
41)
(0.0
41)
(0.0
42)
(0.0
40)
(0.0
40)
(0.0
41)
Ed
uca
tion
−0.
963
−0.4
24
−0.2
92
0.1
72
0.6
87
0.2
19
(3.6
23)
(3.6
43)
(3.6
24)
(3.5
39)
(3.5
62)
(3.5
52)
Inco
me
28.8
97∗∗
28.5
33∗∗
29.2
36∗
∗11.9
15
11.7
40
12.0
41
(14.
524)
(14.5
10)
(14.4
95)
(14.
185)
(14.1
86)
(14.2
08)
Ideo
logy
−8.
140∗
∗∗−
8.1
32∗
∗∗−
8.3
85∗∗
∗−
5.7
01∗
−5.7
55∗∗
−5.6
75∗
(2.9
81)
(2.9
79)
(2.9
76)
(2.9
11)
(2.9
13)
(2.9
17)
Cla
ss0.
912
1.5
87
0.6
87
4.387
4.8
61
4.5
63
(3.1
47)
(3.1
63)
(3.1
78)
(3.0
74)
(3.0
92)
(3.1
15)
Seo
ul
3.75
2∗∗
∗3.7
72∗∗
∗3.4
05∗∗
4.0
30∗∗
∗4.0
21∗
∗∗3.9
96∗∗
∗
(1.4
17)
(1.4
16)
(1.4
26)
(1.3
84)
(1.3
84)
(1.3
98)
Soci
ald
omin
ance
orie
nta
tion
−6.2
92∗
−4.7
69
−3.4
66
−3.6
78
(3.2
40)
(3.3
18)
(3.1
67)
(3.2
52)
Rig
ht
win
gau
thor
itar
ian
ism
2.2
93
3.3
89
(3.7
52)
(3.6
68)
Nat
ion
alat
tach
men
t6.5
77∗
∗−
0.8
45
(3.1
54)
(3.0
91)
Nat
ion
alch
auvin
ism
0.6
91
1.9
61
(2.8
45)
(2.7
88)
Inte
rcep
t48
.374∗∗
∗48.9
43∗
∗∗46.7
06∗
∗∗41.2
46∗
∗∗40.3
13∗
∗∗41.9
75∗∗
∗
(4.0
13)
(4.6
76)
(4.4
49)
(3.9
19)
(4.5
71)
(4.3
61)
N1,
001
1,0
01
1,0
01
1,0
01
1,0
01
1,0
01
R2
0.025
0.0
29
0.0
34
0.0
23
0.0
25
0.0
25
∗ p<
.1;∗∗
p<
.05;
∗∗∗ p
<.0
1.A
llco
effici
ents
an
dst
an
dard
erro
rsre
port
edfr
om
OL
Sm
od
els.
11
-
ranges from −1.9 to −2.2 percentage points. In contrast, the next three sets of models show
that gender has no statistically significant association with perceptions of competence. Age
is significantly related with perceptions of both warmth and competence. Since potential
generational effects may be nonlinear, we test for potential nonlinearities by estimating a
generalized additive model with a nonparametric smoothing spline on the effects of age, the
results for which are shown in Figure 3.24 Despite modeling the effects of age more flexibly,
it nonetheless depicts a straightforward linear effect: younger respondents perceive North
Koreans as relatively colder than respondents from older generations, as well as relatively
less competent. On the whole, then, the relatively neutral average stereotypes shown in Fig-
ure 2 are partially a function of generational divides, with younger respondents displaying
more negative stereotypes about North Koreans than older respondents.
Figure 3: Effects of age on stereotypes about North Koreans
20 30 40 50 60 70 80
-15
-10
-50
510
15
Age
Warmth
20 30 40 50 60 70 80
-15
-10
-50
510
15
Age
Competence
Figure 3 plots the effects of respondents’ age on their perceptions of North Koreans’ warmth and compe-tence; to account for potential nonlinear effects, the results here come from a generalized additive modelwith a nonparametric smoothing spline on the effects of age. Nonetheless, age displays a straightforwardlinear effect. Younger respondents perceive North Koreans as relatively colder and less competent, whileolder respondents perceive North Koreans as relatively warmer and more competent.
Interestingly, education is not significantly associated with stereotypes about North
Koreans, but income is. The wealthiest category of respondents are significantly more likely
24. Luke Keele, Semiparametric Regression for the Social Sciences (Chichester, UK: John Wiley & Sons,2008).
12
-
to perceive North Koreans as warmer than South Koreans are; a shift from the lowest to
highest income category in the survey is associated with a 29 percentage point increase
in perceptions of warmth. Political ideology is significantly associated with stereotypes of
North Koreans, for both warmth and competence: more conservative respondents display
significantly more negative stereotypes about North Koreans than more liberal respondents,
with the effects being larger for warmth (roughly an 8.0 percentage point difference) than for
competence (roughly a 5.7 percentage point difference). Once income is controlled for, social
class is not significantly associated with stereotypes of North Koreans, while respondents
living in Seoul display 3.7−4.0 percentage point increase in positive stereotypes with respect
to both warmth and competence.
The next set of predictor variables estimate the effects of a variety of political orienta-
tions. Social dominance orientation is a psychological trait that measures beliefs in social
and economic inequality.25 Individuals high in social dominance believe in the inevitability
of hierarchies, and of the powerful dominating the weak, and thus tend to be economically
conservative and opposed to social welfare programs.26 As one might expect, respondents
high in social dominance orientation demonstrate more negative stereotypes about North
Koreans, particularly with respect to perceptions of warmth. In contrast, right wing author-
itarianism — a psychological construct measuring beliefs in respect for authority, tradition,
and social control — is not significantly associated with respondents’ stereotypes.27 The final
two covariates refer to two different facets of national identity. National attachment refers to
the extent to which respondents identify with their nation, whereas national superiority or
chauvinism refers to the extent to which they view it as superior to others.28 Interestingly,
national superiority is never significantly associated with stereotypes about North Kore-
ans, but national attachment is: the more being South Korean forms an important part of
25. Felicia Pratto et al., “Social Dominance Orientation: A Personality Variable Predicting Social and Po-litical Attitudes,” Journal of Personality and Social Psychology 67, no. 4 (1994): 741–763; Jim Sidanius andFelicia Pratto, Social Dominance: An Intergroup Theory of Social Hierarchy and Oppression (Cambridge:Cambridge University Press, 2001).
26. John Duckitt and Chris G. Sibley, “A Dual-Process Motivational Model of Ideology, Politics, andPrejudice,” Psychological Inquiry 20 (2009): 98–109.
27. Bob Altemeyer, Right-wing Authoritarianism (Winnipeg: University of Manitoba Press, 1981).28. Richard K. Herrmann, Pierangelo Isernia, and Paolo Segatti, “Attachment to the Nation and Interna-
tional Relations: Probing the Dimensions of Identity and their Relationship to War and Peace,” PoliticalPsychology 30, no. 5 (2009): 721–754.
13
-
respondents’ identities, the less warm they perceive North Koreans to be.
Figure 4: Effects of age on dehumanization of North Koreans
20 30 40 50 60 70 80
-15
-10
-50
510
15
Age
Hum
an u
niqu
enes
s
20 30 40 50 60 70 80
-15
-10
-50
510
15
AgeH
uman
nat
ure
Figure 4 plots the effects of respondents’ age on human uniqueness and human nature dehumanization ofNorth Koreans. As in Figure 3, to account for potential nonlinear effects, the results here come from ageneralized additive model with a nonparametric smoothing spline on the effects of age. The left-hand panelshows no significant effect of age on human uniqueness dehumanization, while the right-hand panel displaysa straightforward linear effect for human nature dehumanization. Just as younger respondents in Figure 3perceived North Koreans as relatively colder, here we find that younger respondents deny human naturetraits like warmth and curiosity to North Koreans at relatively greater rates, whereas older respondentsdehumanize North Koreans less.
Table 2 conducts a similar analysis as Table 1, but regressing dehumanization scores
on demographic variables and political orientations. Models 1-3 focus on the effects of these
covariates on human uniqueness judgments, while models 4-6 turn to human nature de-
humanization. The results show that although stereotype content and dehumanization are
related concepts, they have different patterns of relationships with the covariates. The only
significant predictors of human uniqueness dehumanization in models 1-3 are social dom-
inance orientation and national attachment. Respondents with the highest levels of social
dominance orientation are 4.0−5.5 points higher in human uniqueness dehumanization than
respondents with the lowest levels of social dominance orientation. And, respondents higher
in national attachment dehumanize along the human uniqueness dimension significantly less.
In contrast, we find more significant results in models 4-6 that examine the predictors of
human nature dehumanization, the dimension we saw the greatest levels of dehumanization
14
-
Tab
le2:
Pre
dic
tors
of
deh
um
an
izati
on
Hu
man
un
iqu
enes
sH
um
an
natu
re
(1)
(2)
(3)
(4)
(5)
(6)
Mal
e0.
865
0.7
52
0.6
69
0.9
35
0.9
30
1.0
72
(0.7
11)
(0.7
12)
(0.7
09)
(0.9
34)
(0.9
38)
(0.9
36)
Age
−0.
008
−0.0
14
−0.0
01
−0.0
77∗∗
−0.0
73∗∗
−0.0
92∗
∗∗
(0.0
26)
(0.0
26)
(0.0
26)
(0.0
34)
(0.0
34)
(0.0
34)
Ed
uca
tion
2.55
22.6
11
1.8
33
−10.4
22∗
∗∗−
10.6
73∗
∗∗−
9.9
95∗
∗∗
(2.2
76)
(2.2
83)
(2.2
66)
(2.9
88)
(3.0
08)
(2.9
90)
Inco
me
−12
.795
−12.4
14
−12.5
81
−3.9
84
−4.1
13
−3.3
43
(9.1
24)
(9.0
92)
(9.0
64)
(11.9
76)
(11.9
81)
(11.9
61)
Ideo
logy
−2.
538
−2.6
87
−2.3
65
−0.5
88−
0.4
86
−0.8
83
(1.8
72)
(1.8
67)
(1.8
61)
(2.4
58)
(2.4
60)
(2.4
55)
Cla
ss2.
208
1.8
65
2.1
63
−1.6
98
−1.7
17
−2.3
48
(1.9
77)
(1.9
82)
(1.9
87)
(2.5
95)
(2.6
12)
(2.6
22)
Seo
ul
0.558
0.4
90
0.7
54
−3.1
55∗∗
∗−
3.1
15∗∗
∗−
3.5
31∗
∗∗
(0.8
90)
(0.8
87)
(0.8
92)
(1.1
68)
(1.1
69)
(1.1
77)
Soci
ald
omin
ance
orie
nta
tion
5.5
21∗∗
∗4.0
27∗
−1.4
65
0.0
06
(2.0
30)
(2.0
75)
(2.6
75)
(2.7
38)
Rig
ht
win
gau
thor
itar
ian
ism
3.1
69
−3.1
36
(2.3
51)
(3.0
98)
Nat
ion
alat
tach
men
t−
6.3
64∗∗
∗6.4
39∗∗
(1.9
72)
(2.6
02)
Nat
ion
alch
auvin
ism
2.0
48
0.7
66
(1.7
79)
(2.3
47)
Inte
rcep
t0.
304
−3.2
32
1.0
50
24.6
06∗∗
∗26.8
90∗
∗∗21.5
17∗
∗∗
(2.5
21)
(2.9
30)
(2.7
82)
(3.3
09)
(3.8
61)
(3.6
71)
N1,
001
1,0
01
1,0
01
1,0
01
1,0
01
1,0
01
R2
0.009
0.0
18
0.0
27
0.0
290.0
30
0.0
36
∗ p<
.1;∗∗
p<
.05;
∗∗∗ p
<.0
1.A
llco
effici
ents
an
dst
an
dard
erro
rsre
port
edfr
om
OL
Sm
od
els.
15
-
in Figure 2. Here, older respondents dehumanize North Koreans less than their younger
counterparts. To test for nonlinear generational effects, we estimate the effects of age flexibly
using a nonparametric smoothing spline in Figure 4, which finds, as before, a relatively
straightforward linear effect. Interestingly, the pattern here corresponds with the effect of age
on perceived warmth: younger respondents dehumanize North Koreans more with respect to
human nature traits — traits like curiosity, emotion and agency that are typically associated
with warmth. Younger South Koreans thus are less likely to perceive North Koreans as
fellow humans, and more likely to perceive them as machine or robot-like. Similar patterns
are shown in models 4-6 of Table 2 with respect to education: the most educated South
Koreans attribute human nature traits to North Koreans 10−10.5 points less than the least
educated respondents; residents of Seoul dehumanize North Koreans 3.2 − 3.5 points less.
Finally, respondents high in national attachment dehumanized North Koreans significantly
more than their less nationalist counterparts.
How do South Koreans’ perceptions of North Koreans shape their
attitudes towards unification?
While understanding South Koreans’ perceptions of North Koreans is important in its own
right, it is particularly important if these stereotypes influence South Koreans’ policy pref-
erences. Table 3 explores the relationship between perceptions and policy attitudes towards
reunification. Survey respondents were asked a series of twelve items measuring their sup-
port for unification (including statements such as “It is my sincere wish that the two Koreas
be unified”, and “Even if we couldn’t witness the unification in our life time, we must endure
through all means to achieve the grand goal of unification.”); parallel axis factoring with
varimax rotation finds that that 10 of the items load neatly on a two-factor solution, where
one factor refers to respondents’ general levels of support for reunification (e.g. disagreeing
with statements like “It is better to stay divided than to unify two Koreas.”), and the other
to respondents’ self-described willingness to personally sacrifice in order to achieve reunifi-
cation (e.g. agreeing with statements like “I am in favor of raising taxes to cover the costs
16
-
of reunification.”)29
Table 3: Predictors of attitudes towards unification
Support for reunification Personal sacrifice for unification
(1) (2) (3) (4) (5) (6)
Male −0.023∗∗ −0.022∗∗ −0.023∗∗ −0.047∗∗∗ −0.044∗∗∗ −0.046∗∗∗(0.011) (0.011) (0.011) (0.011) (0.011) (0.011)
Age 0.002∗∗∗ 0.002∗∗∗ 0.002∗∗∗ 0.004∗∗∗ 0.003∗∗∗ 0.003∗∗∗
(0.0004) (0.0004) (0.0004) (0.0004) (0.0004) (0.0004)Education 0.080∗∗ 0.081∗∗ 0.070∗∗ 0.078∗∗ 0.079∗∗ 0.072∗∗
(0.035) (0.035) (0.035) (0.036) (0.035) (0.036)Income 0.151 0.130 0.151 0.339∗∗ 0.296∗∗ 0.327∗∗
(0.140) (0.140) (0.140) (0.144) (0.142) (0.144)Ideology −0.068∗∗ −0.061∗∗ −0.068∗∗ −0.125∗∗∗ −0.111∗∗∗ −0.127∗∗∗
(0.029) (0.029) (0.029) (0.030) (0.029) (0.030)Class −0.056∗ −0.057∗ −0.058∗ 0.005 0.001 0.005
(0.030) (0.030) (0.030) (0.031) (0.031) (0.031)Seoul 0.007 0.004 0.004 −0.010 −0.017 −0.012
(0.014) (0.014) (0.014) (0.014) (0.014) (0.014)Warmth 0.075∗∗ 0.130∗∗∗
(0.037) (0.038)Competence 0.0003 0.058
(0.038) (0.038)Human uniqueness 0.015 −0.080
(0.053) (0.055)Human nature −0.104∗∗ −0.089∗∗
(0.041) (0.042)Intercept 0.422∗∗∗ 0.386∗∗∗ 0.472∗∗∗ 0.368∗∗∗ 0.281∗∗∗ 0.448∗∗∗
(0.039) (0.042) (0.049) (0.040) (0.042) (0.050)N 995 995 995 995 995 995R2 0.046 0.052 0.052 0.094 0.121 0.101
∗p < .1; ∗∗p < .05; ∗∗∗p < .01. All coefficients and standard errors reported from OLS models.
Table 3 thus estimates the effect of stereotypes and dehumanization on both general
support for reunification and willingness to sacrifice for reunification (operationalized using
factor scores for more precise estimates),30 controlling for a set of demographic character-
istics. Models 1-3 focus on support for reunification more generally, while models 4-6 turn
29. The two factor solution produces a good fit according to standard model fit criteria: RMSEA: 0.042,TLI=0.976. Two items were dropped due to cross-loading. See Kristopher J. Preacher et al., “Choosing theOptimal Number of Factors in Exploratory Factor Analysis: A Model Selection Perspective,” MultivariateBehavioral Research 48, no. 1 (2013): 28–56
30. Joshua D. Kertzer et al., “Moral Support: How Moral Values Shape Foreign Policy Attitudes,” Journalof Politics 76, no. 3 (2014): 825–840.
17
-
to a willingness to personally sacrifice for unification. For ease of interpretability, all covari-
ates except for age have been rescaled from 0-1. Model 1 estimates the effects of a set of
basic demographic covariates on support for reunification, showing that male respondents,
younger respondents, the less educated, the upper-class, and the more conservative are less
supportive of unification. Model 2 adds covariates for our two stereotype content measures.
Importantly, the warmer the image South Korean respondents have towards North Korea,
the more supportive they are of unification: respondents with the warmest images of North
Koreans are 7.5 percentage points more supportive of reunification than respondents with
the coolest images, even when controlling for all of the demographic covariates listed above.
Perceived competence has no significant effect. Model 3 adds the dehumanization measures
instead, showing a significant effect for human nature dehumanization. Respondents who
deny human nature traits to the North Koreans are significantly less supportive of reunifi-
cation; in a multivariate regression context, the respondents who dehumanize the most are
10.4 percentage points less supportive than those who dehumanize the least.
Model 4 replicates model 1, but this time estimating the effects of the demographic
covariates on respondents’ willingness to personally sacrifice for unification. Similar to the
results in model 1, male respondents, younger respondents, the less educated, the lower-
income, and the more conservative are less willing to sacrifice for unification. Adding the
stereotype measures in model 5 yields similar results as in model 2, although stronger in
magnitude: respondents who perceive the North Korean people to be warmer express sig-
nificantly more willingness to sacrifice for unification; controlling for the other demographic
covariates, the respondents with the warmest images of North Koreans display a 13.0 per-
centage point increase in reported willingness to sacrifice than the respondents with the
coolest images. As before, images of competence are not significantly associated with ex-
pressed willingness to sacrifice. Turning to the dehumanization measures in model 6, we
gain find evidence of significant negative effects of human nature dehumanization: the more
respondents dehumanize North Korean respondents mechanistically, the less willing they
are to sacrifice for unification. Respondents who dehumanized the North Korean people
the most expressed 8.9 percentage points less willingness to sacrifice for unification than
18
-
respondents who dehumanized the North Korean people the least.
Conclusion
This chapter borrowed from a range of theoretical frameworks from political psychology —
including image theory, stereotypes, and dehumanization — to explore the nature, origins,
and consequences of South Korean perceptions about North Koreans. First, it looked at
the images that South Koreans have of North Koreans, showing that South Koreans have
a neutral image of North Koreans overall, but nonetheless perceive North Korean citizens
to be significantly less competent and significantly less warm than South Koreans. This is
somewhat different than how the South Koreans perceive the Japanese people, who are per-
ceived as being similarly competent as South Koreans, but less warm. In contrast, both the
North Koreans and the Japanese are dehumanized by South Koreans in similar ways, with
respect to human nature traits, rather than human uniqueness traits: they are dehumanized
mechanistically, rather than animalistically. With both stereotypes and dehumanization,
however, there is considerable variation in respondents’ perceptions, raising the question of
how this variation should be best explained theoretically.
Thus, the second set of analyses turned to the demographic and dispositional correlates
of respondents’ perceptions, finding evidence of stark generational divides, with younger
Koreans displaying more negative stereotypes and greater levels of dehumanization than
older Koreans. With stereotypes, similar patterns are also displayed by male respondents,
ideologically conservative respondents, and Korean residents outside of Seoul, while human
nature dehumanization is highest amongst the less educated, and respondents who do not
live in Seoul. The third set of analyses shows that these patterns are politically consequential:
even when controlling for a range of demographic variables, South Koreans who perceived
North Koreans as less warm, or who dehumanized them with respect to human nature traits,
were significantly less supportive of reunification, and expressed significantly less willingness
to make personal sacrifices in order for unification to occur. Taken together, these findings
highlight the applicability of these theoretical frameworks to the study of inter-Korean
19
-
relations, and showcase the importance of the study of perceptions in international relations
more broadly.
20