information to users - library and archives...
TRANSCRIPT
INFORMATION TO USERS
This manu-pt has bœn mpmâwd frwn the rnbdilrn msrter. UMI film the
text directiy from the original or copy wknitted. Thus, $orne thesis and dissemtion oogies are in typenintsr face. whik others rnay be from any type of
cornputer pnnder.
Ovenize materials (e-g.. maps, dramngs. charb) are tsproalcsd by sectionin9 the original. beginning at the uppw left- corner m d contiming frwn left do
nght in equal sections with small ov-.
Photographs induded in the original have bal mpmduced
xerographically in this wpy. HQher quîlity 6" x W #adr and w h i phaaoomphic
prints are avaiîable for any M s or ilhrrtnlior# appearing in this wpy for
an additional charpe. Contad UMI d i do order.
Ml& H a A l Inl9rmatiocr and Lwming 300 North Zeeb Road, Ann Arbor, MI 181-1346 USA
THE INTERPERSONAL CONSEQUENCES OF CONFRONTING THE
NONPREJUDICED SELF
Abigail G. Poore
Department of Psychology McGill University
Montreal, Quebec, Canada
July 1998
A thesis submitted to the Faculty of Graduate Stuâies and Research in partial hùnllment of the requitements of the degree ofMaster of Science.
O Abigail G. Poore 1998
National Library Bibliothèque nationale NI dCmada du Canada
Acquisitions and Acquisiins et Bibliographie Services services bibliographiques
395 Wellington Street 395. nie Wdlingtori OttawaON K1AON4 OttawaON K t A W Canada CaMda
The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, dism'bute or seii copies of this thesis in microform, paper or electronic formats.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fkom it may be printed or otherwise reproduced without the author's permission.
L'auteur a accordé une licence non exclusive permettant a la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/nlm, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thése ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.
The thesis describes a pmgram of research aimed at exploring the consequences of
confionting potential perpetrators of discrimination with their own prejudice. An
experimental paradigm was introduced that involveci confionting advantaged group
members with their own potential prejudices in a series of two studies. In the fht study,
White participants who admitted to some prejudice were, surprisingly, dso more ükely to
genuinely reflect on and chailenge their prejudice. A second experiment investigated the
interpersonal consequences of stereotype threat in advantaged group members. Male
participants were confionted with the possibility of c o r b i n g the negative stereotype
that "men are sexist". The results indïcated that male participants atternpted to avoid
confirmation of the negative stereotype by makuig a deliberate effort to appear nonsexist
on a gender task. These fïndings contrast the results of stereotype threat theory found for
disadvantaged group members and are discussed in ternis of their implications for
advantaged group members.
Cette thèse décrit un programme de recherche visant à explorer les conséquences reliées
au fait de confkonter i leurs propres préjugés les individus susceptibles de pratiquer la
discrimination. La première &ude démontre que les participants de race blanche qui
admettent entretenir des préjugés sont ceux qui les remettent davantage en question. La
seconde étude avait pour but d'explorer les conséquences interpersonnelles reliées au
sfereorype threat chez les membres de groupes avantagés. Des participants de sexe
masculin ont été confrontés à la possibilité de confirmer le stéréotype négatif voulant que
les hommes soient sexistes. Les résultats indiquent que les participants, pour éviter de
confirmer le stéréotype, se sont efforcés de ne pas paraître sexistes lors d'une décision
reliée au genre. Ces observations contredisent les résultats de la théorie du stereotype
threat obtenus auprès de membres de groupes désavantagés, et sont discutées par rapport
à leurs implications pour les membres de groupes avantagés.
ACKNOWLEDGEMENTS
I would like to thank the foiiowing people for their contributions:
Dr. Donald M. Taylor for both his genemus insights and guidance throughout, not
to mention his humour which contributeci to my enthusiasm for this research.
The participants of both studies who were rudely confiontexi with the possibility
of their own prejudice and yet graciously offered to come back for more.
Tamarha Pierce and Winnifked Louis who helped contribute to my statistical
knowledge and understanding.
Evelyne Bougie for her wiilingness to translate the abstract.
Elyssia, Danny, Phil, Angela and Lauren for their help as codederates and
experimenters during Study II.
Ryan, Cathy, Mat, Patricia and KKaz for many hours of data coding.
TABLE OF CONTENTS
Abstract
Résumé
................................. Acknowledgements ... ... .. ................
Table of contents ...............................................................
List of Tables
List of Figures
......... CHAPTER ONE: An Andysis of Contemporary Prejudice
Contemporary theories of racisrn ....................................
Methodological challenges in the Study of prejudice
Program of research
CaAPTER TWO: Confionthg Conlliet: What Peopk Think ...... .........* ... About their Own Prejadice ... ....
...................................... Method
Results and Discussion ......................................
Discussion ......................................
Page
1
. - ll
.*. I l l
TABLE OF CONTENTS (continued)
Page
CHAPTER THREE:The Interpcrsonal Consequences of Confrontbg One's Own Prejudice -.... --.---....---.
S tereotype threat and Victims of discrimination
S tereotype threat as Applied to perpetrators of Discrimination
A rnethodological Framework
Method
Results and Discussion
Discussion
Conclusions
References
Appendices
LIST OF TABLES
Table 1 Mean Ratings of Positive and Negative Mect for
............................. Low and High Discriminators 31
Table 2 Mean Ratings of Qualitative Items for Low and High Discrllniaators ............................. 32
Table 3 Rotated Factor Maîrk Loadings for Democratic Racism Scde Items ............................. 34
Table 4 Mean Ratings for High and Low Personal Discrimination Subgroups on Individual Items of the Democratic Racism Scale ............................. 36
Table 5 Mean Ratings of Positive and Negative Emotions for Experimental And Control Conditions ..............................
TabIe 6 Summary of Results For Experimental and Control Conditions in Private and Joint Task
Table 7 Mean Ratings for Behavioural Interaction Categones in Rototypical Control and Experimental Participants ...............................
LIST OF FIGURES
Page
CHAPTER THREE
Figure 1.1 Stereotype Threat and ................................. V i c h s of Discrimination 49
Figure 1.2 Steteotype Tbueat and Pexpetraton of Discrimination ................................. 50
CHAPTER ONE
An Anilysis of Contemporary Prejudice
Prejudice and discrimination have long been the focus of social psychoiogical
theoiy and research. This is understandable given the destructive coasequences of
discrimination for members of disadvantaged groups specificaliy, and society as a whole
more generally. Indeed the expression of prejudice in many daily contexts has motivated
social scientists to make a concerted effort to address issues such as racisrn and sexism in
socieiy today. The shared similarities and the worrying persistence of these social
problems underscores the need for m e r attention from researchers in tenns of the basic
underlying processes associateci with prejudice and discrimination. This chapter aims to
address the main themes in contemporary tneories of prejudice as well as to discuss how
the cornmonalties inheient in these themes may be used to examine the underlyhg
psychologicai processes involved in prejudice.
Both racism and sexism are historical and cultural phenornena that involve a
fundamental bias in the way in which people process information. These biases rnay arise
at both an individuai and a group level. Theories of prejudice and discrimination take hto
account this distinction between group and individual level processes when describing the
potential causes of prejudice. Universal processes, for example, emphasise the inherent
human potential for pmjudice, while social and intergroup processes focus on the socidy
shared patterns of prejudice in p u p s . Individual-difference dimensions, on the other
hand, ascertain individuai susceptibility to prejudice (Duckitt, 1994). These different
orientations to prejudice have commonly been used to interpret blatant discrimination,
such as that found in the United States in the 50's and 60's (McConahay, 1986) or in
South Africa during apartheid (Pettigrew, 1958), as well as discrimination involving not
only people of colour, but also wornen and other disadvantaged groups including
aboriginal people or the disabled.
A fùndamental and challenging dilemma has become increasingly apparent to
those tackluig problems of discrimination in North American society. On the one han&
opinion polls during the 70's and 80's clearly document declining empirical measures of
racism. On the other han& the persistence of racial and gender conflict reflects the
possible persistence of discriminatory attitudes and behaviours (Dovidio & Gaertner,
1986). Most importanly, the paradox has left researchers wondering whether people are
genuineiy less prejudiced and more favourable towards either f i c a n Americans or
wornen, or whether prejudice has simply assumed a more disguised and l e s overtly
negative form? The present thesis describes a program of research that aims to explore
this question in the context of contemporary theones of both racial and gender prejudice.
While most theories of prejudice focus on racial discrimination, these processes
could be argued as generalisable to ail intergroup contexts involving prejudice, thereby
including gender discrimination as well as racial discrimination. Indeed, Dovidio (1989)
points out that many of the critical elements of racism also relate to sexism. National
surveys on women's equality have been shown to support the possibility of structural
similarities between modem racism and modern sexism (Swim, Aikin, Hall & Hunter,
1995). Moreover, opinion polis suggest that fewer people endorse old-fashioned beliefs
about AEncan-Americans and also that fewer people disapprove of nontraditional roles
for women. Thus, despite the preponderance of research focussing on racism, there are a
nurnber of paraUels between sexism and racism.
This generalisability is iliustrated by the fact that recent theories of gender
discriminationy including Modem Sexism, have drawn on a range of theories on racism
3
and perpetrators of racial discrimination. The psychologicai fiterature on race emphasises
that while it does appear that the old fashioned 'rd-necked' form of bigotry is less
prevalent today (Dovidio & Gaertner, 1986), many researchers would c l ah that racism
has, in fact, not disappeared. I n d e current appmaches to studying discrimination have
argued that Whites' more liberal attitudes toward Blacks in recent years betrays a more
"modem" form of racism (McConahay, 1986). Outright and blatant foms of racism have,
it is argued, given way to more subtle and rationalisable forms of discrimination. The
detection of such contemporary, pervasive fonns of racism within more egalitarian and
educated samples is alanning because of its potential influence on social policies, the
legal process and society in general. Furthemore, in this unrecognised form, racism
becomes difficult to detect, and even more dinicult to eradicate. Thus, the study of
prejudice reduction presents a formidable challenge requiring theory that can adequately
expiain its seemingly "invisible" nature and how people "explain away" their own
discrimination towards others.
The apparent profound changes in racial attitudes have led researchers to believe
that the nature of Whites' attitudes towards M c a n Americans is more complex and
conflicted than was once the case. Thus, theories of prejudice need to address two
principal challenges to better investigate the underground character of prejudice. Firstly,
researchers need to uncover and make explicit the hidden psychological processes
underlying prejudice. Secondly, they need to overcome a number of methodological
challenges that are involved in the process of making obvious the hidden psychological
processes of prejudice. In order to set the stage for the present prograrn of research it will
be necessary to k t , review contemporary theories of prejudice. Second, the themes
emerging fiom contemporary theoria of prejudice need to be specined and fhaily, it wiU
be necessary to address the methodological challenges arising h m the mergent themes.
The challenge of explaining the hidden and conflicted nature of racial attitudes has
spawned a numt er of contemporary theories of prejudice, including ambivalent racism,
modem racism, aversive racism, and automatic and contrulled processes involved in
stereotype activation. A review of each of these theories aims to reveal some converging
theoretical themes.
: Ambivalent racism suggests a conflict between two core
Amencan value orientations: - I . - . and (Katz &
Hass, 1988). The humanitarian-egalitarim value reflects concems for dernomtic ideals
and concern for othem' weii-being. It is believed to evoke sympathy reactions and pro-
Black sentiment among Whites who identiS Blacks as members of a group that has been
the target of prejudice and discrimination. However, Americans are also taught to value
elements of the Protestant Ethic iocluding personal fieedom, self-reliance, discipline,
dedication to work, and acbievement. This individualist orientation is believed to result in
a negative anti-Bladc attitude because negative behaviours associated with Blacks (crime,
dmg addiction and unemployment) are attributed to personal weaknesses rather than
situational factors (Katz & Hass, 1988).
The resulting ahbivalence originates fkom the simultaneous holding of pro- and
anti-Black attitudes. For example, White colIege students presented with opposing
fiiendly and hostile opinions about BLack people and their civil rights experienced a
conflicted and negative mood state (Hass, Katz, Rizzo, Bailey Br Moore, 1992)- Thus,
5
making the ambivalence salient creates emotionai tension. This tension implies that the
pro- and anti- components are experienced as incompatible and that making a person
aware of the coexistence of the components is threatening to favourable selfregard (Hass
et ai., 1992). Furthexmore, ambivalences create a tendency towards behaviourai instability
in which a positive or negative response toward the object of ambivalence may occur. For
examp!e, Bell & Esses (1997) have found evidence for this behavioural instability in
regard to people's reactions to abonginai people. They refer to this instability as
"response amplification". Specificaily, ambivalence toward stigmatised p u p s in society
can lead to polariseci, or ampüfied, responses (Katz, 1981). That is, an ambivalent
peson's response to a stigmatised member may be amplifieci in a positive or negative
direction compared to those directed at a nonstigmatiseci group. For example, a Black
person who displays sociaiiy desirable characteristics wodd be evaluated more
favourably than a White person with the same characteristics. Alternatively, a Black
person who displays socially undesirable charactenstics would be evaluated less
favourably than a comparable White person (Katz, Wackenhut, & Glass, 1986, cited in
Bell & Esses, 1997).
According to the theory of ambivalent racism, then, the potential for denial of
one's own prejudicial thoughts or actions is a consequence of holding humanitarian-
egaiitarian thoughts that act to suppress negative anti-Black attitudes. In fact, denial rnay
be one means of avoiding the ambivalence and associated negative affect that a person
may feel.
: Modem racism involves a confiict between the endorsement of
abstract principles of justice (equality, faimess and fkeedom) and the negative affect
toward Blacks that is acquired through socialisation (McConahay, 1986; Kinder & Sem,
6
198 1). Like the theory of racial ambivalence, modem racism maintains that the Protestant
ethic contributes to the negative component of the confiict- However, the nature of the
ambivalence charactensed by modem racism d i f f a h m that of ambivalent racism in
that ambivalent racism involves a genuine positive pro-Black component while the theory
of modem racism does not (Katz & Hass, 1988).
According to modern racism, an attempt is made to maintain a nonprejudiced image
despite negative feelings towards Blacks (Monteith, 1996). In this theory then, the denial
of prejudice is motivated by an attempt to "appear" nonprejudiced. In order to avoid
attributions of prejudice, modem racists supposedly develop negative attitudes toward
Blacks that can be justified with reference to nonprejudiced explanations. For example,
when asked why she is voting against a Black candidate, a racist voter may use the
explanation that the candidate is ''tm liberal" rather than because she is Black.
McConahay (1986) contends that people are largely unaware of their prejudices.
Sirnilarly, modem racists may rationalise their negative feelings in terrns of more abstract
political issues. For example they may claim that 'Black people have gotten more
economicaIly thao they deserve" (McConahay, 1986).
versive m: Aversive racism represents a particuiar type of ambivalence in
which there is a conflict between feelings and beliefs associateci with a sincerely
egalitarian value system and unacloowleâged negative feelings. In this way it is very
similar to models of modem racism and ambivalent racisme
Like modem racism, Gaertner and Dovidio's (1986) aversive racism does not
assume the widespread existence of genuineiy pro-Black, favourable components of
Whites' racial attitudes that are independent of egalitarian values. Instead, aversive
racists' rejection of racial stereotypes and their sympathetic feelings towards victims of
7
injustice convince them that theh attitudes are largely positive, and not prejudiced.
Aversive racists are not consciously aware of their conflicteci restions and express their
prejudice only in subtle or covert ways. Thus racial discrimination may typicaily go
unrecognised because it usuaîîy occurs in situations in which there is a lack of normative
structure defUluig appropriate action, or where unfavourable responses can tte rationalised
by attributing its cause to some factor other than race (Gaertner and Dovidio, 1986)- For
example, in the case of an emergency invohing a Black victim, a White bystander would
help only if there were no other bystanders available to heip the Black victim. When the
White is the only witness, then there is strong normative pressure to intervene and help
the Black victirn so as to appear nonracist.
ut- andm Proc-: Patricia Devine's (1 989) research on
stereotyping and racism presents a more optimistic picture in which people are able to
control the impulsc to act on racial stereotypes. Devine (1 989) maintains that al1 people
within a culture have access to the same social information and resulting stereotypes
about minority groups. In support of this she reports empirical evidence for the fact that
low, moderate, and high prejudice individuals were found to possess equivalent
kmwledge regarding the negative social stereotype of blacks. For example, Devine
(1989) found that both high and low prejudice participants were equaily Wrely to report
negative stereotypes regarding Blacks when asked to List the content of the cultural
stereotype of Blacks regardlas of their own personal beliefs. The content of such
stereotypes included category's such as "poor, aggressive, criminal, low intelligence and
uneducated".
Stereotypes are a weli-learned set of associations that are automatically activated
in the presence of a member of a target group. It follows, therefore, that in the presence of
8
an outgroup member, a person's f h t automatic tendency is to behave in a stereotype
consistent rnanner. Devine (1989) provides evidence to suggest that this activation is
equally strong for both high and low prejudiced individuals. In spite of this, high and low
prejudice individuals do, however, differ with respect to their personal beliefs about
Blacks. Unlike the high prejudice person whose beliefs reflect the cuiturai stereotype, low
prejudice people have decided not to base their behaviour or evaluation on these
stereotypes. Thus, the dtura l stereoiype about Blacks confiicts with their nonprejudiced,
egalitarïan values. Overt nonprejudiced responses, therefore require inhibition of the
automatically activateci stereotype and activation of a newer personal belief structure
(Devine, 1 989). This inhibition of prejudiced stereotypes and the activation of newer
nonprejudiced beliefs and responses requires "controlied processes". Devine (1989)
provides ernpiricai evidence for the assumption that controlled processes can inhibit the
effects of automatic processing to express and maintain a nonprejudiced identity.
Devine's research has important implications for our understanding of how people deal
with their own prejudiced impulses. Her theory and research suggests that most people
are aware of the negative stereotypes regarding blacks, but that not al1 people are driven
to base their behaviour on such stereotypes. Instead, people c m choose to control their
racist impulses in an attempt to appear nonracist. It therefore follows that people may be
able to ''unieam" their racist behaviours and thereaAer leam to become less racist in
interaction with people of colour.
In cornparison to traditional blatant forms of prejudice, the process of wrestling
with ones owa prejudice appears to be an anxiety-pmvokhg one. The theories of
ambivalent, modern, and aversive racism as weU as Devine's automatic and controiied
processes illustrate how inner psychological confiict is a central feature of present day
interpretations of reactions to racial minority groups. In addition, each of the fou. theories
points out an affective consequence to the conflict- Firstly, ambivalence theory daims
that activation of the confîicted attitude will lead to psychological tension and discodort.
Indeed, ambivalent participants (ie. high pro-Black and high anti-Black scores) have been
found to report greater discodort d e r conside~g an interracial scenario than did
nonambivalent participants (Hass et al., 1992). Katz & Hass (1988) have shown how, in a
situation where race is made salient, high ambivalent subjects experienced greater
negative mood, while mood was unrelated to ambivalence in the low-salience condition.
According to aversive racism theory, people experience conflict and ambivalence
that may Iead to the subtle expression of negative racial affect. This negative affect is not
describeci as hostiiity or hatred, but rather as discornfort, uneasiness, disgust, and
sometimes fear, which tends to motivate avoidance (Gaertner & Dovidio, 1986).
Furthexmore, attempts to maintain a nonprejudice self-image can increase disaffection for
Blacks because interracial interactions lead to anxiety and uneasiness. Given this fear of
acting inappropriately in interracial contexts, it is not surprising that much of the research
on aversive racism has focused on the influence of normative contexts (Gaertner &
Dovidio, 1986) and the extent to which it afTects how racial bias is, or is not,
demonstrated.
In contrast, modem racists avoid any conflict-related feelings through justimg
their negative attitudes with nonprejudiced explanations. McConahay (1986) claims that
modem racists are unaware of the codict because they believe their attitudes to be
nonprejudiced. Monteith (19%) has indeed conkned the general absence of affective
consequences in relation to modem racism,
To summarise, both ambivalent and aversive racism involve conflict that results in
psychological tension and discodort. In contrast, modem racism argues that the
ps ychological tension and discomfort is kept h m awareness.
In the final theory, Devine (1991) found that her participants were aware of the
codlict involved in violating one's own seif-reported nonprejudiced standards- According
to Devine's (1991) mode1 of conflict, the change nom prejudice to nonprejudice is not
viewed as an aii-or-nothing event, but as a process during which the low prejudice person
is especially minerable to contlict between his or her habitua1 negative responses and
nonprejudiced beliefs. Devine (1991) h d s evidence to suggest that people have
internaiised standards regarding how they "should" behave in the presence of Blacks. A
trangression of these standards can lead to a discrepancy between societal and personal
standards, and this could have affective consequences. Devine (1991) found that low
prejudiced participants with large should-would discrepancies experienced compunction
(guilt and self-criticism). High prejudiced participants with large discrepancies
experienced global discomfort but not compunction
Each of the four theones of racism describe some f o m of a "two-factor" mode1 of
prejudice; in each case, one factor tends to promote prejudice, the other to suppress it
(Biemat, Vescio, Theno & Crandall, 1996). Factors that promote racism include anti-
Black affect that is common to al1 models, conventional values toward individuaikm in
modem and ambivalent racism, and the perception that Blacks violate important values as
in symbolic racism. Moreover, negative normative or contextual cues, such as a value
violating Black target, or a situation that condones the expression of prejudice, as
11
explained by ambivalence and aversive racism, is also argued to result in racism. Finally,
in Devine's (1989) model, behaviourai and evaluative responses to automatic negative
stereotypes about Blacks would be considered as promoting racism in high prejudice
uidividuals.
Counteracting these prejudice-promotion factors are variables that prevent the
expression of racisrn. These include pro-Black affect which is present only in the
ambivalence model, egalitarian and/or humanitarian values that are evident in ail the
models and positive normative or contextual cues - a value supporting a Black target, or a
situation in which anti-Black responding can readily be interpreted as prejudice, as is
explained in ambivalent and aversive racism (Biernat et al., 1996). In Devine's ( 1 989)
theory, nonprejudiced responses are a hc t ion of controlled processes that actively
inhibit prejudiced behaviour and evaluations.
Both prejudice-promothg and prejudice-inhi'biting factors are likely to enhance
the core conflict experienced by a White who is interacting with a Black, or in 0 t h
cases, a man who is interacting with a woman. The present program of research aims to
explore the hidden aspects of this codict by attempting to codiont people with both
sides of the confiict and then capturing their responses. The b t study wiil present a
methodology that will expose people to the possibility of their own prejudice, and is
primarily exploratory in nature. It aims to uncover people's spontaneous defenses and
rationalisations regarding their own prejudice against people of colour. A second, more
experimental, study aims to explore people's reactions f i e r they are explicitly confronted
with the possibility that they are prejudiced This experiment has the benefits of
experimental control and manipulation that will hopefully yield greater clarity of
interpretation regarding the consequences of being exposed to one's own prejudice.
Contemporary theones of racism aiso highlight how pressures to appear
nonprejudiced present researchers with a number of methodological diniculties.
BelieWig racism to be mostly 'underground' as opposed to overt in expression,
researchers have developed a number of indirect methods for studying this apparently
'disguised' fom of prejudice. These methods include the bogus pipeline (Fazio, Jackson,
Dmton & Williams, 1999, priming methods (Crosby, Bromley & Saxe, 1980) and the
use of apparently nomeactive scales such as the Modem Racism Scale (McConahay,
1986) or the more recent Motivation to Control Prejudice Scale @unton & Fazio, 1997).
In summary, ambivalent, modem, and aversive racism, and Devine's automatic and
controiied processes al1 describe a contlict between anti-Black feehgs and egalitarian-
humanitanau values that suppras the expression of these anti-Black feelings, thoughts
and actions. Rejudice is therefore not readily apparent because egalitarian values and the
pressure to appear nonprejudiced or open-mindeci inhibits the impulse to act on one's
prejudice. In spite of such inhibition of prejudice, it has aiso been argued that prejudice is
far nom nonexistent and that it is only hidden at the surface level. Contemporary theories
of prejudice, therefore, offa an explmation as to how racism is paradoxicaily 'hidden'
yet also maintaineci. It has also been argued that the pressure to appear nonprejudiced and
the role of social desirabiiity is crucial to understanding not only the disguised form of
racism but also the potential problws associated with the difficulties of studying
prejudicial thoughts and feelings indirectly.
The traditional methodological approach to studying racism has been to keep the
'me' purpose of the study concealed by a 'cover story'. These are designed to prevent the
arousd of social desirability concems and defenses on the part of the participant. This is
13
because social desirability concerns are thought to prevent the expression of any existuig
'real' prejudice that needs to be measured by the researcher. The researcher does not
challenge the participant's defenses, but rather colludes with the participant's
nonprejudiced presentation in order to access the ' r d prejudice the participant may be
attempting to conceal. In doing so, the researcher attempts to conceal the very issue of
racisrn that they are interested in. One problematical consequence of the covert nature of
this method of investigation is that it makes racism very difficult to measure empuicaily
as well as makes interpretation problematic. For example, the Modem Racism Scale
(McConahay, 1986) differs fiom the Old Fashioned Racism Scale because it is less direct
and blatant about the kind of racism that it is measuruig. The Modem Racism Scale
permits the expression of negative attitudes and prejudice but in a way that allows the
respondent to feel nonprejudiced. This is achieved because prejudiced responses can be
explained by racially neutral ideology or nonprejudiced race-relevant attributions. For
example, an item asserting "is not discrimination a problem faceci by al1 of us fiom time
to time?'allows an indirect expression of prejudice that could easily be justified as
nonprejudiced.
If researchers continue to avoid the issue of racism in a manner similar to the way
in which their participants are managing to do so, it is questionable whether they will
ultimately penetrate the cornfortable d a c e of nonprejudice and access the psychological
processes that result in the core conflict. To address the problem of inroidance of
prejudice, it is argued here that racism needs to be investigated through directly
codkonting participants with the possibility of their own prejudice. As Biemat et al.
(1996) suggest the next step toward testing models of racism is not merely to lead
participants to think about egalitarian values, but rather to directly threaten their
egalitarian and humanitarian ~e~ images .
The chailenge for the present program of research is to develop a method for
confkonting people with their own prejudices that has three important featines. Firstly, it
needs to address discrimination directly, but in a way that is credible for the participaut.
Secondly, it needs to have the potential to provoke the emotionai conflict that is
suggested in al1 the theories. Thirdly, since the present stedy is a f k t exploratory attempt
to address the hidden conflicts of prejudice, it ne& to lend itself to both a quantitative
and qualitative evaluation of the participants' spontaneous expressions of confïïct-
One method that addresses these three criteria draws on recent theory and research
conducted, not on perpetrators, but on their potential victims. Mïnority group members
who are potential victims of discrimination have been found to deny, or minunise, their
experience of the discrimination that they face (Crosby, 1984; Ruggiero & Taylor, 1997).
This has been iiiustrated by a phenomenon that has been labeled the personal/group
discn'rnination dhmepancy (Taylor, Wright, Moghaddam & Lalonde, 1990). Here,
members of disadvantaged groups consistently rate discrimination directed at their group
as a whole to be higher then discrimination aimed at themselves, personally, as a member
of that group. The personaügroup discrimination discrepancy has proven to be robust.
The phenomenon has been reported for women, and for a range of luiguistic, religious
and racial minority groups including Anglophones and Francophones, Jewish
Montrealers, huer city Aûican Americans (Taylor, Wright & Porter, 1993) and a sample
of Inuit in the remote Arctic (Poore, Gagne, Barlow, Lydon, Taylor & Wright, 1997).
The personal/group discrimination discrepancy has appeal for the present context
because of its paradoXical implications. Specificaily, it would not be logical for al1
participants to claim that the dimimination directed at their group as a whole is greater
than that directed at them personally. Instead, the discrepancy behueen personal and
group is irrational because each participant must be considered as being part of their
"group" and thus for most, if not ail, their personal responses of discrimination should
match those of the group's perceived discrimination. The difference between persmal and
group accounts of discrimination would, therefore, seem to be a false perception. It is the
irratiodty of the false discrepancy that is challenging for researchers.
The present study aims to use similar perceptions of personal and group
discrimination fiom the perspective of the discriminator as opposed to the victixn. Thus,
individuals would be required to rate the extent to which they personaiiy, as a member of
their advantaged group. either W t e s or males, engage in discrimination against Blacks
or women, and separately, how much their group as a whole discriminates against Blacks
or women. According to the theones of ambivalent, modem and aversive racism, people
are well aware of the undesirability of appearuig opedy prejudice. Thus, any admission
of discrimination would be viewed as socially undesirable compared to the
judgement that one's group discriminates. Therefore, it would be expected that
participants would rate the extent to which they, -, as a White, discriminate
against Blacks as lower than the extent to which theu B, as a whole, discrirninates
against Blacks. Similariy, male phcipants would not want to appear sexist and they
would therefor? argue that they, persondly, discrimlliate against women less than men, as
a group, discriminate against women. Thus, a discrepancy is expected to Prise between
personal and group ratings, not uniike that found in the personallgroup discrimination
discrepancy. In the case of the of discrimination, however, the phenornenon
needs to be labeled the personaVgroup discrimina& discrepancy.
16
The discriminator discrepancy is usefid because it raises the possibility for an
advantaged group member to express persona1 d i a t i o n directly. It does this
through juxtaposing the ratings of personai and p u p discrimination each of which seem
logical when considered separately, but which are bgically impossible if taken together.
Specifically, a low rating of pemnal discrimination against Blacks is logical to an
individual who considers himself to be liberal and nonprejudiced. At a group level, it wiU
also be logical to assume that Whites, as a whole, discriminate agaiast BIaçks, as a whole.
When juxtaposing the two, however, it becomes Iogically impossible for every individuai
to perceive him or herseIf to be not discriminating while simuitaneously considering the
group of which they are a member, as a whole, to be discriminathg. Making the
discrepancy explicitly obvious to the participant, potentiaiiy requires them to consider the
conflict between demands to appear nonprejudiced and the reality of discrimination by
the group, of which they are a member. In other words, a man may personally want to
appear nonsexist and egalitarïan while at the same t h e he is aware that men as a group
are perceived to be discriminatory towards women.
ProeramsfResearch
A review of contemporary theories of prejudice, with a focus on theories of racism,
leads to certain wnclusions. Contemporary foms of both racial and gender prejudice
appear to be in a more subtle and rationalisable fom compared with the overt and direct
expressions of prejudice of the 50's. Some suggest that contemporary prejudice is not
only more hidden, but also more coaflicted. The aim of the present chapter was to review
the conflicts outhed by four modern theories of prejudice, namely ambivalent, modem.
aversive racism and Devine's theory of controlled and automatic processes. Furthemore,
the characteristics of the inner c o d c t outiined by each of the theories is highlighted. The
source of the confiict is viewed as either pro- or anti-Black sentiment as in ambivalent
racism, or the tension between non-racist, egalitanan values and negative racial aflèct as
suggested by modern and aversive racism. Finally, the conflict may be between negative
stereotypes about Blacks and nonprejudiced, egalitarian values that suppress anti-Black
beiiefs as in Devine's theory. In ail cases, it appears that the confiict may result in
negative affect such as psychological tension, discodort, or even guilt and seIfcriticism.
To date, methods of studying prejudice have IargeIy relied on indirect or 'hidden'
methods of assessment. Such methods have aimed to study the disguised forms of
contemporary prejudice in a nomeactive and discrete manner that respects concerns about
social desirability. By emphasising such a strategy, many researchers attempt to conceal
the very issue of prejudice in which they are interested- The present cbapter has argued
that indirect methods of studying prejudice may not uncover core psychological processes
or conflicts that may be driving the prejudice. What is needed is a direct form of
exploring the core conflicts in a way that uncovers what is driving the prejudice as
opposed to what is suppressing it.
The present thesis, therefore, descnbes two studies, both of which make use of
methods that directly c o h n t people with their own potential prejudice. The £kst study is
explorative. It aims to understand the nature of the potentid conflict underlying
contemporary forms of racial and gender prejudice. This study aims to directly confiont
people with their experiences of prejudice by pointing out to them the discrepancy
between their pzrceived personal and group discrimination ratings; that is, their
personaI/group discriminator dacrepancy. This direct confrontation is aimed at evoking
people's spontaneous rationalisations regarding their prejudice, thereby allowing the
exploration of any conflict underlying their prejudice.
The results of the Grst study will be used as a guide for a second experimental
study. The methodological strengths of the first study, wherein participants will be
directly cofionted with theV prejudice, will also be applied to the second study. The
second study will be conducteci in an experimental setting, and wili involve actuai
intergroup interaction.
CaAPTERTWO
Confronting Conflict: What Peopk Think About Their Owo Prejudice
The purpose of this nrst study was to explore the potential conflict inherent in
Whites' attitudes towards Blacks. The aim was to confkont a sample of Whites with the
possibility of their own prejudice and to assess theu spontaneous reactions to the
provocation.
Because of the exploratory nahue of the study, hypotheses, need to be made
cautiously. It is first hypothesised that participants wiîl rate their personal discrimination
as lower than group discrimination because of the social undesirability of expressing
one's personal prejudices. Group discrllnioation wodd be rated as consistently higher
because there is usualiy a degree of consensus regarding how the group behaves (Taylor,
Ruggiero & Louis, 1996), and because there is less need to be socially desirable. Thus,
advantaged group members are expected to express the "shared stereotype" that Whites,
as a group, do discriminate against BIacks, as a group, without feehg defensive about it.
hdeed, the expected discrepancy between personal and group ratings of discrimination
has arisen in pilot research for this study.
It is m e r hypothesised that negative feelings and uncertainty will be associated
with cons ide~g one's personal discrimination but less so with group discrimination. This
is predicted because admitting personal discrimination shouId be associated with concems
about appearing sociaily desirable in a society that does generally not value expressions
of personal prejudice. In contrast, admitting to group discrimination is more distanced
f?om the self and is therefore less likely to evoke anxiety-provoking concerns.
When combiollig the two hypotheses, an interesting paradox is expected to arise.
Penonal ratings, which are expected to be Iower than group ratings, are nevertheless,
hypothesised to be associated with higher rathgs of uncertainty and negative affect than
group ratings. The paradox arises because there is more Wrely to be psychologicd conflict
conceming social desirability when personal, rather then group ratings, are made. When
group ratings are made, participants can employ a cornmon social perception regarding
the prevalence of group discrimination. This process does not evoke much conflict and is
not likely to be associated with uncertainty or negative a i t .
Special attention to ratings of personal discrimination is required because this is the
question that is pivotal to the psychologicai mechanisms underlyîng discrimînation.
Participants who may be regarded as 'high' in their personal rating are admitting they
discriminate. Thus they are also expected to rate more highly on negative Hect than
those who score 'low' on their personal rating.
'High' personal discriminators are also expected to express theu views on
discrimination more clearly and to wrestle with their discrimination less because they are
less concemed about inhibiting socially undesirable responses associated with arimitting
personal discrimination. Since they are more willing to admit to personal discrimination,
they are also expected to score higher than 'low' personal discriminators on overt
measures of discrimination but lower than 'low' personal discriminators on more covert
measures of discrimination
Findy, it is hypothesised that the spontaneous rationalisations provided by high
personal discriminators wiii be judged as less socially desirable because they are expected
to be more overt in theu spontaneous expression of discrimination. ûvert expression of
personal discrimination is considered undesirable because it Molates acceptable
egalitarian values.
Method
The participants were 52 white male students who were recniited h m a university
library. Participation was voluntary. Ages ranged fiom 18-30 years with the mean age of
1 8-24 years.
Procedm
Each individual was approached separateiy by the experimenter and asked to
participate in a study on discrimination and social attitudes. They were told that it would
take about half an hour and that they would be required to complete a questionnaire.
Permission to have their response to one question tape recorded verbaily was also
requested. Data was obtained in the following four stages:
In the h t stage, the participant was asked to fXi out an initial questionnaire
consisting of demographic questions conceming age and number of years of education.
This was foilowed by two neutral questions on general topics of discrimination that were
unrelated to the focus of the study. These were aimed at aquainting the participant with
the rating scales. Both neutral questions were each foilowed by four ratings scales that
asked the participant to indicate how they felt when anmering these initial neutrai
questions. The scales ranged h m one (Not at all) to ten (Very much) and asked about
four feelings: anxiety, discornfort, anger and uncertainty.
In stage two, two fiirther questions were posed that focused directly on the central
topic of perceived personal and group discrimination. The two questions were
counterbalanced and used the same format as the neutrai questions. The f k t question
obtained a rneasure of perceived personal discrimination and was phrased, 'Wease rate
the extent to which you, penonally, as a White, discriminate against Blacks?". The
participant read the question and was requùed to answers on a 10-point scale ranging
fiom one (Never) to ten (Ofien). They were then asked to rate how they felt using the four
negative etnotion scaies when answering the question on personal discrimination.
The next question was intended as a meastue of perceivecl group discfimitlation.
Participants read the foilowing question before making a rating, "As a White, to what
extent does your group discriminate against Blacks?". This was followed by ratings of the
four negative emotions.
Based on the results of a pilot study, it was expected that virtually every participant
would rate the extent to which they pemaliy discriminate against blacks as lower than
the extent to which Whites as a group discriminate against Blacks. In the third stage of
the procedure, the aim was to confkont the participant with his ratings of persona1 and
group discrimination and to ask him to explain the expected discrepancy. In order to
gather explanations that were as spontaneous and uninbibited as possible, verbal (as
opposed to written explanations) were gathered. To minimise self conscioumess on the
part of the participant, the interviewer asked the following question after glancing at the
questionnaire, "Oh, 1 see that you have given yourself an 'Y' for how much you consider
yourself to discriminate against Blacks, and a "Y' for how much Whites as a group
discriminate against Blacks as a group. So there is a difference between the two scores. In
your own words, how would you explain this difference?". No prompts were used,
although the question "anything else?'was asked when the participants indicated that
24
they had finished what they wanted to Say. The participant was then given the negative
emotion rating scale asking about the extent to which they felt anxious, uncornfortable
and uncertain while giving their explanation of the discrepancy.
Participants' explmations were transmibed verbath h m a tape recording. The
transcriptions were then rated on a five-point scale by the experirnenter h m one (none)
to five (very much) on the foilowing categories:
The extent to which the respondent's perceived peisonal disahination was
verbalised. This is a score fkom 1 (none) to 5 (lots) representing the number of
sentences or phrases mentioning anything to do with personal discrimination;
The extent to which their own discrimination is questioned. This is a score from 1
(none) to 5 (iots) representing the amount of persona1 'wrestling' with discrimination
which is expresseci in the form of phrases, words, questions and so forth.
The extent to which their own discrimination is denied. This is a score fiom 1 (none)
to 5 (lots) which represents how much the person denies or defends against their own
prejudice e.g- '7 don't discriminate at dl" would rate 5 whereas "1 try not to
discriminate" would rate 3.
The clarity of stance regarding discrimination. This is scored on a scale of 1 (none) to
5 (lots) and concems whether or not the person is able to clearly express their views
and ideas on discrimination e.g. '7 don? discriminate" and "1 do discriminate" would
both rate 5, while "1 don't know if 1 discriminate'' would rate 3. The issue is not
whether the person expresses or denies prejudice, but rather how confident they are on
their views.
The extent to which the rater perceived the participant as likeable, based on reading of
their transcription. Here 1 represents 'hot at dl" and 5 represents 'tery much".
four:
In the fouah stage of data collection, the person was asked to complete the
Democratic Racism Scale. This scale consisteci of twelve items (see Appendix A) that
were denved from a set of statements outiined by Henry & Tator (1994) that describe the
underlying values and sentiments of "dmocratic racism". Similar in essence to
ambivalent, aversive, and modern racism, democratic racism addresses the confiict
between egalitarian notions of justice and equalîty, and negative feelings towards
minority group members. The twelve statements were used as individual items with a
rating scale of 1 (do not agree) to IO (agree). Examples of items include "It is wondemil
that young people have the fieedorn to protest against things they don? like and to "do
thei. own thhg'"' and 'In these troubled times laws have to be enforceci without mercy,
especially when deahg with the agitators and revolutionaries who are stimng things up".
Al1 items are reversed scored, except for item H. The items have not been used as a
measure of racism before and do not yet constitute a validated scale.
Two items were exchded fkom the analysis because they failed to correlate with
any of the other items. The remaining ten items evidenced a moderate intemal reliability
with a Cronbach's alpha of .70.
Results and Discussion
Ratings of personal and group discrimination as well as the motions that
participants felt while m a b g their ratings were exploreci. The ratings of personal and
group discrimiliation wili be presented fkst followed by a discussion of the feelings
associated with ratings of personal and group discrimination. Subsequently, the sarnple
was divided into those who had low personal discrimination and those whose ratiugs of
personal discrimination were higher.
. - of-
The mean rating for personai discrimination was 2.37, which may be regarded as
very low. In contrast, the mean for gmup discrimination was 5.57 and very much higher
than the personal discrimination mean rating but still not extremely hi&- The mean
ratings of the penonallgroup discriminator discrepancy rnay also be compared with
ratings of the personaVgroup discrimination discrepancy found in previous research for
potential victims of discrimination. The ratings of personal and group discrimination for
perpetrators compare with ratings made by innercity Black men who are considering
discrimination in the domain of housing, and with non-university women (Taylor, Wright
& Porter, 1994). However, samples of South Asian and Haitian wornen tend to have
elevated scores of both personai and group discrimination compared to the personal
discrimination scores made by perpetraton in the present sample. This could be
interpreted as rneaning that the participants in the present sample are denying the
perpetration of their own discrimination and their group's discrimination to the sarne
extent as inner-city Black men and non-university wornen are denying their experiences
as victims of personal and group discrimination.
A t-test for paired merences revealed a signifïcant merence between group and
personal ratings of discrimination, 0(5 1) = 1 1,M, c .001). The highly significant
difference between the means for the penonai and gmup ratings connmis the existence of
a persondgroup discriminatpf discrepancy. The Low personal rathgs are assumed to be
the result of strong social desirability pressure to appear nonprejudiced. In this way, the
27
penonaVgroup discriminator discrepancy is similar to the personaVgroup discrimination
discrepancy where people report low personal expenence with discrimination in part
because they want to distance themselves h m the stigma of being discrimiaated against.
The hypothesis that group ratings of discrimination will be higher than ratings of personal
discrimination was, therefore, confhned.
- . . - of f e e u e d w w
Three emotions were assessed, including anxiety, discornfort and uncertainty. It
was found that anxiety and discornfort were highly correlated, (1 = -76, p c -01). Thus,
these emotions were not treated as separate constructs, but rather as a composite that is
considered to represent some fom of general negative afïkct.
Ratings of group discrimination correlated negatively and marginally signincantly
with uncertainty, @ = -.24, p = -08) and evidenced a positive but nonsignincant
correlation with negative affect, (1 = -17, p > -05). Ratings for personal discrimination
correlated positively and signincantly with both negative affect, (1 = -42, p < -01) and
uncertainty, (l= .45, p c .O 1 ). Ratings of group discrimination are, therefore, correlated
with a lack of uncertainty. This is in contrast to ratings of personal discrimination that are
associated with negative feelings and uncertauity. The hypothesis that personal
discrimination, but not group discrimination, would be associated with negative affect
and uncertainty was, therefore, supported. Group discrùniaation was not hypothesised to
correlate with uncertainty and negative affect. While this was codïmed for negative
affect, it was not codïmed for uncertainty. Group discrimination was found to relate to a
lack of uncertainty, suggesting that people are fairly certain about theù perception of the
extent of group discrimination.
When asked to rate personal discrimination, the mean rathg for negative e t was
2.8, for group discrimination, the mean rating for negative affect was 2.4. There was a
margindy sigdicant difference between negative affect for personal and poup
discrimination, (f(5 1) = 1 -9, p = -07). The mean for uncertainty was rated at 3.6 when
asked about personai discrimination, and at 4.1 when asked about group discrimination.
There was no significant Merence between unccrtainty for personai and group
discrimination, (N51) = 1 .O, p > -05). Each of these means is on a 7-point scale thus
indicating that scores of uncertainty and negative affect were not very hi@.
In summary, mean ratings for uncertainty and negative affect associated with
personai discrimination were higher than mean ratings for uncertainty and negative affect
associated with group discrimination, as was hypothesised. Thus, personai discrimination
is related to both negative mect and uncertainty while ratings of group discrimination are
related only to a marginal sense of certainty. This suggests that ratings of persona1
discrimination may be associated with more psychological conflict than ratings of group
discrimination. It also suggests that people are more certain about the autostereotype that
"Whites as a group" discriminate against Blacks and that consideration of this
autostereotype does not evoke anxiety or discornfort. Ratings of personal discrimination,
however, possibly involve confiicts of social desirability that in turn evoke negative affect
and uncertainty when attempts at social desirability are threatened. Such conflicts of
social desirability are central to the theory of modem racism, and to a lesser extent,
ambivalent, aversive and Devine's theory of controiied and automatic processes.
* * - - S of 1 0 ~ of D-
A median split was used to divide the sample into those who scored relatively
"hi@ on the discrimination question (rathgs > 1; n = 27) a d those who scored
relatively ''iow" on the discrimination question (rating c and = 1; Q = 21). The
tenn low discnntinutors wiii be used to refer to those whose scores on the persona1
discrimination question, according to standard m e m e s of prejudice, are nonprejudiced
(-1). High discriminators are those who score moderately high (BI) on the sarne
question thus suggesting that they hold beliefs that allow for slightly greater levels of
prejudice. AU these scores are, however, very low and there may well be a floor effect for
this variable thus making it difficult to obtain meaningfid effects.
w t i v e affeçt and for low 1 " . *
Table 1 shows mean ratings of negative affect and uncertainty for those who scored
low and high on personal discrimination.
For those in the high personal discrimination group, ratings of both negative affect
and uncertainty were higher than for those who scored low on personal discrimination.
High personal discriminators scored higher on negative &ect than did low personai
discriminators and the difference was significant, (l(49) = 2.47, p c .01). High personal
discriminators score signincantly higher on uncertahty than do low personal
discriminators, (i(49) = 3.83, p c -00 1).
Low discriminators, therefore, appear to expenence less negative a e c t and are
more certain about their ratings than high personal discriminators who experïence more
negative affect and greater uncertainty. This supports the hypothesis that those who report
engaging in more personal discrimination also experience more emotional coafiïct when
thinking about it.
Table 1
Negative affect
Uncertainty
Mean
Low discriminators 2.27
High discriminators 3.86
Low discriminators 1.96
High discriminators 3.30
The subgroups of hi& and low personal discriminaton were used to explore the
spontaneous responses when participants were asked to explain the discrepancy between
their personal and gmup ratings. Qualitative data was rated using two raters and the
interrater reliability was calculated at an acceptable -70 using the Speaman-Brown
Reliability Coefficient (Rosenthai & Rosnow, 1991).
Table 2 compares the subgroups of high and low on personal discrimination with
respect to a) how much the person defends or denies his own personal discrimination, for
example, "1 don't discriminate at all" would rate higher on denial than "I try not to
discriminate" ; b) how much he mentions or talks about his own personal discrimination,
31 .
c) how much he questions or w~estles with his own discrimination, d) how clear he is
when expressing his views on discrimination and finally, e) a rating of the social
desirability of the participant.
Table 2
-
Low discriminators High discriminators
-
Defend or deny personal discrimination 3.29
Clarity of position regarding discrimination 2.89
Mentioning of personal discrimination 2.57
Questioning/wrestling with own discrimination 1.48
Desirability rating of respondent ' s account 2.45
High discnminators are shown to defend their own discrimination less than low
discriminators, l(42) = 2.4, p < .OS, to mention their discrimination more than low
discriminators, l(42) = 2.4, p < .OS, to question their discrimination more, l(42) = 2.2, p c
.O5 and were judged as more sociaiiy desirable than low discriminators, l(42) = .2.1, p <
.OS. High discriminators expressed their discrimination more clearly than Iow
discriminators, but the Merence was not signincant, $(42) = 1.4, p > -05.
In summary, there is a tendency for high discriminators to mention their own
personal discrimination more and to defend it less than low discriminators. They also
wrestle with and question their tendency to engage in discrunination more than low
discriminators but their expression of discrimination is no clearer than low discriminators.
Surprishgly high discriminators are judged as more socially desirable than low
discriminators. One possibility is that high discriminators are more sewaware. They
admit to personal discrimination and therefore to grapple with the confîïct provoked by
such social undesirable behaviours. Indeed their willingness to con.fiont the conflict is
reflected in their elevated negative motion and uncertainty scores. This is interesting
because it does not support the original hypothesis that high discriminators would be
uninhibited about expressing their personal discrimination largely because they do not
wrestle with their own prejudicial thoughts or actions. The present results suggest that
high discriminators are more concemed about social desirability than low discriminators,
even though the extent of their discrimination is stiîl relatively low.
~ O W DI of- . . . .
The democratic racism scale was fïrst examined using principal components
analysis. This yielded four factors with Eigenvalues over 1 that together accounted for
approximately 66% of the total variance. A varimax rotation was performed and factor
loadings were forced into a three factor solution. The f h t factor accounted for 3 1% of the
variance, the second factor 14% of the variance and the third factor 1 1 % of the variance-
The factor loadings, following varimax rotation. are displayed in Table 3.
Table 3
-
Factor 1 Factor 2 Factor 3
Racism is a consequence of immigration; racial conflict cornes about because there are too many -42 -62 -09 different people here.
Minority groups refiise to fit in and adapt to .60 -32 - . I l
Canadian society.
The problems experienced by minority groups are -.20 .32 -.l I fiequently cultural problems and have nothing to do with race.
Non-whites corne to Canada to benefit fkom the social advantages of living in this country.
The policies of multiculturalism sufficiently address the problems of racism.
Racism comes nom ignorance and what is needed is education about other people which focuses on changing attitudes.
The over-sensitivity of people of colour makes them look for racism where it is not intended.
Strategies such as employment equity interfie with the basic tenets of fkeedom of individual thought, choice and action on which notions of western dernocracy are founded.
Anti-racism initiatives are racism in reverse. .O6 -64 .22
The items that load most highly on the first factor appear to involve overt racism.
These include items such as 'kninority groups refuse to fit in and adapt to Canadian
society" and "the over-sensitivity of people of colour makes them look for racism where
it is not intended". The second factor was interpreted as a measure of covert, or subtie,
racism where a potentialiy nonprejudiced explanation is offered in place of an overt
explanation. Items include "the problems experienced by minority groups are fiequently
culturai probtems and have nothing to do with race7' and "anti-racism initiatives are
racism in reverse". The third factor consists of only two items and is therefore barder to
interpret. Both items could be interpreted as a denial of any present day racial problerns.
It includes items such as 'tacism camot exist withui a democratic society" and 'the
policies of multiculturalism sdficiently addresses the problems of racism".
Scores of highand low qJIJ&ce on 0 . .
Table 4 shows the means for high and low discriminators on individual items.
The low discriminators a = 4.7) tended to score more highly than high
discriminators (M = 4.3) on the items that together form factor 1, the overt racism factor.
There was, however, no significant difference between high and low discriminators for
the overt factor, (l(47) = 1 .O, p > .OS). In contrat, high discriminators (M = 4.0) tended
to score more highly than low discriminators (M = 3.3) on items that together form factor
2, the subtle racism factor. Again, however, the difference between low and high
discriminators, (t(47) = 1.5, p > .OS) was not significant.
Table 4
of the D
Racism cannot exist within a democratic society 3 -2 2.9
Racism is a consequence of immigration; racial conflict coma about because there are too many different people here.
Minority groups refûse to fit in and adapt to Canadian society.
The problems experienced by minority groups are fkequently cultural problems and have nothing to do with race.
Non-whites corne to Canada to benefit fiom the social advantages of Iiving in this country.
The policies of multiculturalism sufficiently address the problems of racism.
Raçism cornes fiom ignorance and what is needed is education about other people which focuses on changing attitudes.
The over-sensitivity of people of colour makes them look for racism where it is not intended.
Strategies such as employment equity interfere with the basic tenets of fieedom of individuai thought, choice and action on which notions of western democracy are founded.
Anti-racism initiatives are racism in reverse.
It is possible that low discriminators, who do not wrestle with their prejudice as
much as high discriminators, could be in denial of their prejudice and are therefore less
likely to be aware of their prejudice. Their expression of prejudice may in fact be more
overt and less hidden than high discriminators. On the other hand, high discriminators are
not only more iikely to wrestle with their prejudice, but are also more Wely to admit to it.
Their awareness of their own prejudice may lead them to express their prejudice in a
more subtle and complex way.
Discussion
Using contemporary theones of prejudice, the present study addressed the
psychological con£licts experienced by advantaged group members who are c o h n t e d
with the possibility of their own racial prejudice. As expected, perceived persona1
discrimination was rated significantly lower than group discrimination. The present study
has argued that admitting to personal discrimination is likely to be rated as low because it
arouses concems about social acceptability that, in tum, are iikely to make the participant
hide any anti-Black sentiment. According to contemporary theones of racism, this is
because admitting to racial prejudice is regarded as violating contemporary values of
fairness, justice and equality (Katz & Hass, 1988; Biemat et al., 1996).
While personal discrimination was rated as less than group discrimination, it was
associated with more negative affect than group discrimination. This was expected
because admitting personal discrimination against Blacks is likely to enhance intemal
conflicts about expressing prejudice and thereby increase one's anxiety about social
desirability . In contrast, admitting that Whites as a group discriminate is a more
normative concept that is less emotionally provocative because it allows the respondent to
37
distance hirnself h m his group. Admitting to group discrimination did, nonetheless,
provoke higher mean ratings of uncertainty than did personal discrimination. This could
be explained if a higher rating of uncertainty is considered more a form of cognitive
rather than an emotional discodort. In this case, admitting to group discrimination does
not arouse emotional discornfort associatecl with being socially inappropnate because it is
not directed at the 'self' in the same way that personal discrimination is. It may, however,
amuse cognitive uncertahty because it challenges the participant to admit that the
participant identifies and 'belongs' to a gmup that neveaheless engages in discrimination
which is socially davourable.
It is not clear whether those who rate themselves as low on persona1 discrimination
are genuinely non-prejudiced. According to Devine's (1989) fiamework those who daim
to be nonprejudiced are in fact nonprejudiced. In the present study low discriminators are
probably genuinely low prejudiced. On the other han& those who rate themselves low in
personal discrimination could be denying their own discrimination. The theones of
modern, ambivalent and aversive racism would argue that low discriminators are more
likely to be denying their discrimination than presenting themselves as truly low
prejudiced.
The uncertainty regarding the interpretation of Low discriminators' scores
highlights the importance of fausing on those scores that may be regarded as
comparatively 'high' . Unlike the low discriminators, those who were higher in their
ratings of personal discrimination are easier to interpret. Hi& discriminators are
arimitting to at least some discrimination so there is les arnbiguity regarding whether or
not they may be regarded as discriminatov. The high discriminators are interesting
because they are admitting to a certain amount of personal discrimination and this is less
38
normative than rating low personal discrimination. By admitting to more personal
discrimination, they also admit to more anxiety and negative affect.
Contrary to what was hypothesised, those who rate high on personal discrimination
were also found to be signincantly more w i h g to grapple, or psychologically wrestle,
with the possibility of their own racial discrimination. They mentioned their personai
discrimination more and also defended their discrimination less than low discriminators-
These stiategies suggest that those who are a little more "high" in discrimination put more
psychological effort into dealing with their prejudices than do low discriminators, if we
assume that low discriminators do in fact discruninate.
The hding that high discriminators psychologicaliy grapple more with their
prejudice was unexpected, especially when one considers that they cannot be regarded as
genuinely high in discrimination. High discriminators were expected to show less
psychological grappling and defensiveness of their own prejudice bec ause they were
believed to have fewer social desirability concerns. The reduced social desirability
concerns in high discriminaators scores could be expected because high discriminators are
probably more cornfortable with expressing their discrimination.
Furthemore, it was expected that high personal discriminators' explanations of the
personaVgroup discriminator discrepancy would be rated as less socially desirable
because they were more likely to express prejudice openly. It was found, however, that
high discnminator 's explanations were rated as significantly more socially desirable than
low discriminators.
This hding that high discriminators are socially more desirable in spite of the fact
that they express more discrimination is counterintuitive. According to contemporary
theories of racism, revealing ones' prejudices does not convey a favourable self-image or
39
enhance social desirability. The findings in the present study reveal a difierent pattern
where those who are more sociaiiy desirable are those who are outwardly prejudice, but
also more willing to deal wiîh, and challenge, their own prejudice, than those who appear
outwardly nonprejudiced, or low prejudiced. High discriminator's are apparently more
open to confionting their prejudice and thereafter more willhg to grapple and talk about
it. In contrast to the openness to confiont discrimination, low discriminators showed l e s
openness and awareness of the possibility of their own prejudice. They were ais0 rated as
less socially desirable. Thus, low discriminators may 'cover up' discrimination in order
to suit the noms of a nonprejudiced self-image and therefore may be construed as more
hypocntical, l a s genuine and hence, less socially desirable.
While high discriminators were open about their persona1 discrimination, they
scored more highly than iows on the covert rather than overt scale of racism. This was
contrary to what was expected. The contradiction of being high on an overt measure of
persona1 discrimination and yet simultaneously scoring relatively more highly on a covert
measure of discrimination deserves explanation. One simple explanation is to consider the
personal discrimination rating as a rating of awareness of discrimination rather than of
actual discrimination itseK In this case, ifhigh discriminators are indeed more aware of
their discrimination, they may put more effort into attempting to conceal their prejudices
because they are al1 the more aware of its social undesirability. This would explain why
they score highly on the covert discrimination rating.
A second explanation for the fbding that high discriminators score more highly on
covert measures of racism involves the theory of modem racism. Modem racist theonsts
would argue that people attempt to disguise their racism and would deny any personal
discrimination because of the unfavourability of appearing racist (McConahay, 1986).
40
Modem racists would therefore score high on the covert scde of racism and low on the
personal discrimination rating. According to modem racism the conflict here is between
appearing nonprejudiced and feeling negative e t towards Blacks. In the present study,
it was not the low discriminators but the high discriminators who scored high on the
covert scale. One explanation for the high discruninators endorsing the covert racism
items is that the nature of the conflict illustrateci by modem racism has changed. On the
one hand, as outüned by modem racinn, people still avoid overtly generic racist
expressions and prefer to endorse covert expressions of racism. On the other hand, unlike
modem racism, the present study also reveals how it is socially desirable to express
insight into and 'conféss' to one's own potential racial intoleiance. This openness and
awareness of one's own racist thoughts and behaviour is valued because it shows that one
has psychological insight and a flexibility to adapt to the changing requirements of
'political correctness' and tolerance. This interpretation, with its support for open and
honest awareness of one's own intolerance without actually directing ones prejudice at
anyone, may represent a 'new' form of modern racism. Such 'ultramodern' racism arises
as expressions of intolerance continue to change as a result of the changed values attached
to racism as new social issues emerge in North America.
What are the affective consequences to the 'ultramodem' conflict? The demands of
maintaining a nonprejudice self-image, as well as admitting one's racist flaws, rnay well
result in a contradictory double bind of expectations on behaviour. The double bind could
augment any existing conflicting feelings of anxiety or uncertainty. Indeed, in the present
study, hi& discriminators who could be experiencing this double bind because they
endoned both covert racist expression and high personal discrimination also experienced
higher indices of uncertainty and negative affect compareci to low discriminators.
41
The 'ultramodem' confiïct described here could represent a new form of racism
characterised by a contradictory tension between admitting personal intolerances while
also behg covertly racist. The nature of this conflict is different to the general formula
proposed by contemporary theones of prejudice where there is a conflict between
negative Black affect and social deskbility concem. This has implications for both
theoretical and applïed perspectives. At a theoretical level, it highlights the changing
nature of social psychobgicd phenornena in accord with changing values and intellectual
climate, and this in tum creates a climate for creativity. At an applied level, the more
direct expression of tolerance and the willingness to wrestle with personal discrimination
underlying any new fonn of 'ultramodern' racism, is viewed positively because
awareness of intolerance gained fiom this process sets into motion to change underlying
feelings of antipathy.
The present study aimed to explore the confiïct underlying the expression of
prejudice. This was achieved by having participants rate their perceived personal and
group discrimination as weii as their emotional responses to each rating. Participants were
conûonted with the discrepmcy between the ratings and asked to provide a spontaneous
verbal explmation for the discrepancy.
Ratings of personal discrimination were found to be very iow, while ratings of
group discrimination were found to be signincmtly higher, thus indicating a
personal/group discrimination discrepancy. The ratings were used to detemiuie the
presence of any emotional discornfort or uncertainty evoked by reporthg personal and
group discrimination. Persona1 discrimination was associated with feelings of negative
42
affect and uncertainty. Group discrimination was related to a sense of certainty but not
with negative affiect.
It appears that advantaged group members are reluctant to report thei. own
personal potential for discrimination, but are clearly convinceci that discrimination against
Blacks does occur by Whites as a group. Participants' reluctance to report personal
discrimination could be related to experiences of negative affect and uncertainty when
making personal ratings. Sime they are less Wcely to experience these feelings when
making group discrimination ratings, they may also be more cornfortable with atûibuting
higher scores for group discrimination. Participants are clearly anxîous with the stigma
attached to admitting personal prejudice as well as equally aware of the autostereotype
that whites, as a p u p , do discriminate against Blacks, as a group.
Furthemore, high discriminators experienceâ significantly higher levels of negative
affect and uncertainty than did low discriminators. High discriminaton were found to
question their own discrimination more. It appears that those who are more likely to
admit personal discrimination are also more Wtely to wrestle with it more, and expenence
more emotional confïict regarding theu actions.
The study therefore confirmed the existence of the personaVgroup discriminator
discrepancy in White male majonty group members. Evidence suggests that admitting to
one's own personai discrimination is associateci with more emotional conflict than is
admitting to one's group discrimination. Furthemore, people who report more personal
discrimination also appear to experïence more emotional conflict than those who report
iittle or no discrimination.
However, a number of basic questions about the nature of the ezmtional conflict
and its implications remain. Firstiy, what are the consequences of the conflict for
43
interactions between people of different race or difterent gender? Secondly, does
confionhg people with the possibility of their own prejudice evoke an emotionai confiict
that ùifluences the behaviour of the actor? Thirdly? does threatening people with the
possibility that they are pnjudiced or that they may confirm the stereotype of a "sexist
man" or a "racist White" have consequences for how people behave with members of
disadvantaged groups?
The present exploratory study did not provide the necessary expezimental control to
clearly delineate the psychologicai mechanisms underlying the conflict, or the
consequences of such conflict. In order to explore the questions above, an experimental
setting with features of experimental control would be needed. Furthermore, the tape-
recorded responses and low ratings of negative affect reveal that the present method of
coni?onting people with the discrepancy between their personal and group discrimination
ratings may not SufZiciently evoke the intensity of intemal ps ychological codict required
to potentiaüy ifluence the interactional style or behaviour of an advantaged group
member.
Thus a second study would require an intensification of the psychological confZict
in order to observe its effects on intergmup interactions. Finally, the dependent measures
in the present study made use of self-report measures of emotions and verbal responses to
the discrepancy in personal discrimination. A second study would require more explicitly
defined behavioral measures to better differentiate between responses.
ClriAPTERTaREE
The Iaterpersonil Consequences of Confkonthg One's Ona Prejrdice
One aim of the present program of research was to devise a methodology that
would adequately challenge people's '%idden" conflicts regarding their own prejudices.
In order to develop a methodology that allows for a more direct means of accessing the
complexity of people's responses, an attempt was made in Study 1 to develop a
methodology that was not only direct, but that also prompted respondents to discuss their
reactions in a spontaneous, upfbnt and immediate manner. In Shidy 1, the personai/group
discriminator discrepancy was used to prompt people to discuss their personal
discrimination or reported lack theteof.
While üiis methodology was partially successfid, the moderate negative emotions
obtained in Study 1 suggest a need in the present experiment to introduce a more
provocative codbntation with the participant. In particular, the codkontation ne& to be
more direct, rather than simply duded to. That is, participants need to be told directly
that they are more prejudiced than they believe they are, rather than being asked to
discuss their prejudices in a non-directive mamer.
It is anticipated that the unapologetic directness of telling a participant that they
are prejudiced will be a signincant challenge to the seif-image of the participant. This is
likely because most undergraduates are likely to see themselves as either nonprejudiced
or low in prejudice just as the undergraduate participants in Study 1 proved to be. The
participants will no doubt experience a contlict between their nonprejudiced self-image
and the experimenter's evidence suggesting that they are in fact prejudiced. The present
experiment aims to examine how this conflict influences actual behavioural interaction
between individuals representing groups that are potentid perpetrators and victims of
discrimination.
The need for direct conhntation thst threatens the self-image of advantaged
group members and for genuine behavioural measures places the experirnent in the
domain of ment theory and research in the area of (Steele and Ammon,
1995). This impactfiil and novel theory has sought to understand the social psychological
underpinnings to the underachievemmt of disadvantaged groups.
Although the theory of stereotype threat has focussed exclusively on the
experiences of victims of disakination, it dso fias the potential to contribute to
knowledge about m, the aim of the present research. The following section will
outline the theory of stereotype threat as it relates to victims to discrimination. In an
attempt to highlight the relevance of stereotype threat to the present research, the theory
of stereotype threat wil1 then be applied to perpetrators of discrimination.
Stereotype threat is a situational threat that c m affect the members of any group
about whom a negative stereotype exists, for example, older adults or dnig addicts
(Steele, 1997). An advantaged group member is in the 'field' of stereotype threat when
that person can be judged or treated in terms of the negative stereotype. For example,
when African Americaas perfiorm an intellectual task, they face the possibiiity of
connmiùig or being judged by a negative societal stereotype about that group's
intellectual cornpetence (Steele and Aroiwn, 1995). Anxiety about the possibility of
confirming a negative pup-based stereotype is predicted to interfere with intellectual
functioning, especialiy when the task is enough to activate and make plausible
a threatening stereotype. An inteiIectual test, in the case of Anican Americans, would be
considered fhstrating if it was challeaging and presented as diagnostic of intellechial
abiliîy.
Steele and Aronson (1995) have eaipirical support for the negative consequences
of stereotype threat for disadvantaged groups. They focusseci on a sample of high
achieving and academicaiiy successful Black and White students, and asked them to unite
an intelligence test. The sample was divided into a diagnostic condition where
participants were required to reveai their race to the testers, and a nondiagnostic condition
where testers were blind to the race of the participant. Black participants greatly
underperformed compared to White participants in the diagnostic condition for
intelligence but equaled Whites in the nondiagnostic condition. This is ïnteresting because
it would be expected that Black participants in the diagnostic condition would work that
much harder to belie the negative stereotype, yet this appears not to be the case.
Spencer, Quinn & Steele (1997) (cited in Steele, 1997) found evidence for the
impairhg effects of stereotype threat in women. Women students who were competent in
mathematics significantly underperformed in relation to equaily qualifieci men in a
difficult math test. Steele (1997) interprets their underachievement as the result of a
stereotype threat regarciing the wmmon perception that women are not as good as men at
mathematics. The test was presented as diagnostic of mathematical skilis and thus
heightened the stereotype threat by m a h g it more relevant. On tests of advanced
literature, however, women performed equaiiy as well as men, presumably because the
women were not threatened with a negative sterentype in this area
Apart fiom Steele's work, a number of other studies have explored elements of
stereotype b a t , incluâing its effects on self-esfeem (Osborne, 1995) or social identity
(Lee & Ottati, 1995). In a similar vein to Steele's (1997) research and theory, this
48
literature seeks to understand the experiences of potential victims of discrimination. The
focus is on the experiences of successfbi disadvanfaged group members nich as Blacks or
women who are confkonted with negative stereotypes that are relevant to their group
membership. Stereotype threat is experienced when minority members are in a situation
where the negative stereotype may be confirme& The process of stereotype threat
therefore offm important insights into the underachievement of disadvantageâ groups.
By drawing on the Uuiovative theorising on victims of discrimination, the present
study aims to apply the concept of stereotype threat in a novel rnanner. In keeping with
the theme of Study 1, it aims to expbre whether similar processes of stereotype threat
may arise in bernetrators of discrimination in a manner that pardels victims of
discrimination (se Figure 1 and Figure 2).
NQahE Individual -: -: m: Black who Black reveals Underperforms Blacks is racial relative to underachieve 1) academically L.$ membership on a Whites and relative to Whites comptent intellectual test confïrms negative
& f- stereotype
Stereotype Threat and Victims of Discrimination
mai= Individual -: GQllm==: Stweotvbe man who Male is told he is sexist Sexist Men are sexist a has & f- codrmhg behaviour
liberai views sexist stereotype
E@re 7& Stereotype b a t and Perpetraton of Discrimination
A starting point for exploring the stereotype threat process for perpetrators is to
examine the parallels between stemtype threat for disadvantaged group mernbers and
possible stereotype threat afEecting advantaged groups, such as Whites or males. Since
stereotype threat revolves around the knowledge of negative stereotypes peaainuig to
one's group, it is necessary to consider the negative stereotypes associated with
advantaged groups. One clear negative stereotype associated with advantaged group
members is their potential to be intepreted as racist or sexist. For example, Whites may be
labelled as racist, and men may be stereotyped as sexist. In the Study 1, white males were
presented with the possibility that they may be discriminatory towards Blacks and this
appeared to be threatening to the participants because the experience was associated with
negative affîxt and uncertainty.
The stereotype that Whites are racist, or that men are sexist M e r suggests that
advantaged group members are discriminatory towards meanbers of other groups. Thus,
uniike Steele and Ammon's (1995) notion of stereotype threat for Blacks or women,
negative stereotypes for Whites and men do not revolve around issues of intellectual
cornpetence, but rather issues of --. Ifparallels with Steele's
hdings are drawn, it foliows that threatening advantaged group members with a negative
stereotype in the domain of interpersonal behaviour and attitudes will have an efféct on
their behaviour in the context of a disadvmtaged member, just as stereotype threat
affected the intellectual cornpetence of disadvantaged group members. Steele and
Aronson (1995) found that victims confirriied the very stereotype with which they were
threatened and academicaiiy underperfomed relative to contmls. A similar pmess would
be predicted to occur with perpetrators of discrimination who wiU be expected to confirm
the negative stereotype of prejudiced behaviour. T'us, Whites threatened with the
negative stereotype that Whites are racist, may actually corne to behave in a racist
fashion, while men threatened with the stereotype that men are sexist, may behave in an
offensive manner towards women.
Steele and Ammon's (1995) theory, therefore, anticipates that the threat of a
negative stereotype will influence the behaviour of a disadvantaged group member such
that they will behave in a maMer that ultimately c o ~ s the stereotype. The logic of
Steele and Ammon's (1 995) theory would therefore predict that thmatenecl
with being sexist, for example, wW be affected by the threat so as to behave in a way thar
could be construed as sexist towards a woman. Similarly, perpetrators threatened with
being racist would behave in a racist manner. The act of behaving in a potentially sexist
manner following the threat of king laôelied as sexist illustrates how stereotype threat
could result in a display of the very behaviour that is descri'bed in the negative stereotype.
Similarly, whites who are accused of being racist may ultimately behave in a marner that
is congruent with the racist stereotype rather thaa opposed to it. In both cases, the effect
of the manipulation is in the direction of the stereotype threat.
However, a completely different prediction than that arising h m stereotype threat
follows h m the pmliminary d t s of Study 1. The prediction arising h m Study 1 that
. . perpetmtors of discnmmation will behave in the direction of the negative stereotype
but instead in a direction away h m the stereotype. For example, a man who is threatened
with the stereotype that "most men consider women as sexuai ~bjects '~ may go make extra
effort to present himself as a man who relates to women on a number of other dimensions
other than a semial one.
In Study 1, participants were confionteci with a mild threat in the form of behg
asked to explain their ratings of their own personal discrimination. The rating of personal
discrimination was associated with negative e t and unccrtaintyy possibly because of
conflicts regarding social desirability. According to contemporary theones of racism,
such as Modern Racism (McConahay & Hough, 1976), conflicts of social desirability
tend to suppress overt expressions of prejudice in favour of more subtle forms. It is
therefore predicted that men who are confronteci with a negative stereotype in a direct and
provocative way may experience high levels of negative affect and uncextainty. These
emotions are related to concerns about appearhg sexist or racist where the n o m is to
appear nonprejudiced. He may attempt to reduce his psychologicai distress by behaving
in a more valueangruent manner and attempting to overcompensate so as to assert his
nonsexist seIfIfimage with certaine, or notvacist self-image in order to disconfirm the
stereotype threat. For exampley S a man is waiting for a bus very late at night and a single
woman arrives at the bus stop but approaches with hesitation, the man may take steps to
appear nonthreatening and vulnerable. He will behave in such a manner so as to try and
52
disconfim the stereotype that men are threatening to women in potentiaily dangerous and
isolateci situations,
The pradiction arising f h n Study 1 is, therefore, completely different h m Steele
and Aronson's (1995) hdings whem disadvantaged p u p members were found to
behave in a direction that is congruent with stemmtype threat. In contrast, the prediction
arîsing h m Study 1 is that paiticipants wiii attempt to avoid c o ~ a t i o n of the negative
stereotype such as king sexist or racist and wilî take the steps necessary to do so. This
differs fiom the prediction based on Steele and Aronson's (1995) finduigs that men will
behave in a more sexist manner or Whites in a more racist rnanner.
Any experimental paradigm designed to investigate stefeotype threat for
advantaged group members requires that the fwus be not on cornpetence, but on
interpersonal behaviour. The present study aims to do this by exploring the efféct of
stereotype threat ditected towards male participants and their subsequent interactions with
a woman. Male participants need to be confionted with the stereotype that men are sexist.
As Steele (1997) points out in his research on disadvantaged group memben, the
stereotype also needs to be aimed at hstrating the participant. This wiil be accomplished
in the present experiment using the personaVgroup discriminator discrepancy as in Study
I to S o m the male participant directiy that he is more sexist than he thinks he is. It is the
expected contradiction between his own image of himseifas a nonsexist male, versus his
fears of confirming the experimenter's evaluations that he is indeed sexist that are likely
to fiutrate the participant. Fioaiiy, the efféct of stereotype threat needs to be explored in
ternis of its effects on the participant's actuai behaviour toward a woman. Interacting with
53
a woman maices the negative stereotype that "Men are sexist" relevant to the participant's
behaviour. It is the cdiapstic' nature of these tasks that lead to a more impacthil and
challenging stereotype threat.
The task needs to be plausible, mal, and relevant to the stereotype. For example, a
man and a woman may be rquired to choose between two candidates for a job, a suitable
male and an equivalent although slightly less well quaiifïed woman. Such a decision task,
therefore, gïves the male participant the opporhmity to choose the female candidate over
the male candidate and to explain this decision reasonabiy and rationaMy. Yet, in doing so
he is also choosing the slightly less q W e d candidate, which suggests that he must have
had a secondary reason for choosing her. The diagnostic nature of the task, wherein the
participant can choose h m either a competent male or fernale candidate, wiU therefore
allow him the opportunity to reveal his true attitudes towards women.
Two sets of competing hypotheses can be proposed. On the one hand, and
according to the theory of stereotype threat, hyporhe~is Ipredicts that men who have been
threatened will be influenced by the threat and experience greater negative affect than
those who are not threatened. Furthemore, it would be predicted that men who are
threatened with the sexist stereotype wiil behave in accordance with the stereotype. Thus,
men in the threatened gmup would choose a male candidate over a female candidate on
job selection task more ofkm than men who are not threatened. Moreover, in the context
of an actual interaction with a woman threatened men should behave in a more dominant
manner towards the female than those in a non-threatened group. These hypotheses are
based on the rationale that when people are threatened by negative stereotypes of their
group, they will behave in such a way as to confhn the stereotype, even if they desire not
to do so.
54
The second set of hypotheses is based on the predictious that would foliow nom
Study 1. Fintly, it would be predicted that those mde participants in the threatened
condition would experience more negative afZect than those not threatened. This is
because, as found in the nrst study, the proces of being directly confkonted with one's
own prejudice evokes social desllribility concems. Furthermore, it would be expected that
participants in the threatcned condition would attempt to avoid c o ~ a t i o n of the
negative stereotype that he is sexist because of social desirabifity concems and wouId
behave in such a way as to distance himself h m the label of sexist. Thus, it would be
predicted that he would endorse the female candidate in the decision-making task.
Finally, it would be predicted that he is Wrely to behave in a less dominant manner
towards a wornan than those men in the unthreatened condition. The hypottheses are based
on the assumption that a threatened male wiii attempt to disconfimi the negative
stereotype that they are sexist by behaving in a nonsexist way towards women.
98 undergraduate males were d t e d h m a range of classes other than
psychology. 76% were aged between 17 and 20 years, 20% were aged between 21 and 24
years, 3% between 28 and 35 years and 1% was older than 36 years. The mean age was
between 17 and 20 years. Participants were primarily White, although no fomal method
of evaiuating ethnicity was used since the act ofreporthg ethnicity could confound the
study. Participants were told that they would be participating in a laboratory experiment
on 'Decision-making and Gender". Each participant gave signed consent for participation
and each was compensated for his time afta the eXpenment.
Procedwe
The experiment involved four stages:
The testing period was approximatcly 30 minutes. Each participant was invited
into the Iaboratory hdividualIy. He was met by the experhenter and a fernale
confederate who was introduced as another participant in the study. For purposes of
counterbalancing, half the experimenters were h a l e and half were male. Both the
participant and the confederate were told about the presence of a video carnera and what
they would be required to do in the experiment.
Both confederate and participant were then separated and asked to fiil out a
questionnaire at their respective desks. They were separateci such that they could not see
or hear each other as titis would interfere with the experiment. Both were led to believe
that they were f i h g out the same questiomaire. In fact the participant filled out a
questionnaire containing the two questions pertaining to the personaVgroup discriminator
discrepancy. The f h t question asked, "To what extent do you, as a man, discriminate
against women?" and the second question was phrased, "To what extent do men, as a
group, discriminate against women". Participants' were required to make a rating on a 7-
point scale, ranging h m 1 (not at aii) to 7 (very much).
When the participant indicated that he had finished his questionnaire, the
experimenter approached his desk and picked up his questionnaire while saying, cc Let me
check that you have fiiled everything out". In the case of those male participants who
were in the experimental group, the experimenter looked through the questionnaire before
paushg and then tuming to the participant. At this point the experimenter says in a quiet
voice, '1 shouldn't actuaUy teli you this, but 1 see you filied out an "X' for how much you
penonaily disçriminate against women and a 'Y" for how much men as a p u p
discriminate against women as a group. So there is a difference between the two.. . what
this suggests, dong with your other responses, is that you are not aware of your
discrimination against women and that you are actualfy more sexist than you think you
are." If the participant had been assigned to the control group the experimenter sirnply
picked up the questio~aire and said "Let me check that you have fïiied everything out".
If the participant asks any questions or says anything, the experimenter only answers,
'We'll get a chance to speak about this later" and then moved on to the next stage of the
experiment.
After their questionnaires had been collected, the participant and the confederate
were then asked to move their chairs into the center of the room. The confederate
positioned her chair on a series of markings in order to ensure that her chair was always
in the same position. A carnera was placed such that the interaction between the
confederate and the participant could be video taped. The participant had a chair with
wheels that can be fieely moved around. This allowed him to choose a position at a
chosen distance h m the codederate.
The participant and the confederate were then presented with a decision-making
task. They were both given a job description and the Curriculum Vitae of three candidates
(see Appendix B). One was a male candidate who was very suitably qualified for the job
and had lots of technical expenence, but lacked interpersonal skills. Pilot testhg revealed
57
that he was most Wrely to be rated as the best person for the job. The second candidate
was a female who also had technicd sic& but was stronger in tenns of interpersonal
skills than technical skiUs. Pilot testing revealed that she was most likely to be rated as
second best to the first candidate. Finally a "fiilei' candidate was aiso included. He was
consistently rated as least suitable for the job during pilot testing and was not expected to
be chosen by the participant.
The participant and confederate were nrst required to choose one best petson for
the job. This decision was made in silence while the participant and confederate worked
pnvately. In the second stage, the participant and confederate condted with each other in
an attempt to fmd one best person for the job. The confederate was instmcted before hand
to choose the female job candidate. If the participant also originally chose the f d e
candidate, then they both agree on a choice. if the participant originaiiy chose the male
candidate, he may either change his choice to the female candidate or he may pursue his
original choice of the male candidate and not reach a joint compromise with the
confederate.
During the interaction, the experimenter was out of the room. Both the private
choice and the joint choice was M t e d to four minutes so as to standardise the time spent
making the decision. The codederate foliowed a script when she was tequired to explain
her choice of the female candidate to the male participant. She was instnicted not to
influence his decision by providing overly convincing arguments to the male confederate.
Finally, an unobtnisive camera was used to record the interaction between the pair.
Once the ta& was completed, the participant and confederate are once again
separated and required to each fiil out a questionnaire. The participant £iiied out a sheet
58
coocerning his feeling of amie@, discornfort, positivity, uncertainty7 relaxation,
frustration and interest at the point when he was told by the experimenter that he was
more sexist than he thought he was. These scales had two hmctions. Firstly their aim was
to assess the level of negative affkct and c o d c t of the participant in both conditions.
Secondly, they hction as a manipulation check because it was hypothesised that those in
the treatment condition would experience greater d e t y and negative mood than those in
the controt condition. Conf.hnation of this diffefence would prove that the stereotype
threat had an effect on the participants in the treatment condition. The scala were placed
at the end of the study and the participant was asked to think back to the point of the
stereotype threat This was aimed to ensure that the task did not give away the purpose of
the study.
Finally, the participant was debriefed and paid.
Results and Discussion
Results will be presented in three sections. The first two sections are preiiminary
and include presentation of the personaVgroup discriminator discrepancy and a
manipulation check, The thud section focusses on r d t s pertainin . . g to the main
hypotheses. Controt and experimental groups will be presented with regard to the primary
dependent variable, the choice of a male or fernale candidate for a job. No effect of
experimenter's gemder was found in any of the three sections and this variable was
therefore ornitted h m the analysis.
The decision as to whether or not to choose the male and female candidate is
designed such that the male candidate is the more favourable candidate with the female
candidate being only slightly less suitable. The dependent variable is therefore a powerful
59
one because it involves a Me-like situation where a person is requifed to choose the b a t
candidate for a job. The choice of candidates does not involve rating scales where
participants are aUowed to express the neutraîity of their opinions, but rather a mutuaily
exclusive choice of two candidates that binds the participant to the decision he has made.
Participants' are required to choose the male or fernale candidate on two separate
occasions. The first occasion requires the participant to make his decision after being
threatened but without consulting either the experimenter or the confederate. TbÏs £kt
decision is labelled the prira;tetask. During the second decisiou task, the participant is
required to actually interact with, and discuss the same two candidates, with a h a l e
conféderate who is instructed to always choose the femaie candidate and follow a
standardisai script. The confederate is required to be noncommittal about her choice and
she is told not to be persuasive. This dual decision-making process is referred to as the
&& task.
Both control and expe-enta1 groups will be presented in ternis of the results of
the private and joint decision tasks. A cornparison of both the pnvate and joint decision
tasks within each condition will be made, followed by a cornparison of the control and
experimental group.
The mean rating of pasonal discrimitlsltion was 2.1 on the 7-point xaie and this
may be regarded as predictably low. The mean rating for group discrimination was 4.4
indicating that participant's believe that, as a group, men do discriminate agaiost women
moderately ofien, but not extremely o h . Ratings of personal and group discrimination
differed significantly h m one another, l(83) = 14.0, p < .001. Thus the existence of a
60
personaUgroup discriminator discrepancy in the context of gender discrimination is
supported.
Measures of personal and p u p discrimination in the present study on sexual
discrimination are comparable to the personaVgroup disahkator discrepancy fomd in
Study 1 in the context of racial discrimination. Study 1 reporteci the mean scores for
personal discrimination and group discrimination on a 10-point sa le as 2.37 and 5.57,
respectively. Once converteâ to the 10-point %ale useà in Study I, the present study
reveals personal discrimination as 2.95 and group discrimination as 6.35. Thus mesures
of personal and group discrimination appear to be similar in regard to both the sexism and
racism assessed in Study 1 and II. That is, potential perpetrators have a tendency to
express low scores of personal discrimination and higher scores of p u p discrimination
in the case of both racism and sexism.
Seven different feelings were assessed, including anxiety, discornfort, uncertainty,
htration, feeling positive, relaxed and interested. A principal component factor analysis
using varimax rotation revealed two factors. The first factor may be interpreted as a
tive factor that includes feeling anxious, uncornfortable, uncertain and
fnistrated. The second factor may be interpreted as a ppSitive emotions factor that
includes feeling positive and relaxcd. Feeling "interested" did not correlate with any of
the other emotions, possibly because it is tao cognitive in na-. 'Tnterest" was therefore
not included in either of the factors.
The means for both positive and negative emotion factors were compareci for
those in the expeximental and control condition, and are presented in Table 5. Participants
61
who were threatened in the enpaimental condition were sign&antly less positive than
those in the control condition, l(82) = 4.35, p < -001. Moreover, those in the experimental
condition were significantly more negative than those in the control condition, g82) =
4.68, p < .001.
Table 5
Positive ernotions
Negative emotions
The analysis of emotions across experimental and control conditions reveals that
those participants who were threateued experienced significantly more negative emotioos
and significantly less positive emotions than those who were not threatened.
The analysis of emotions pmves to be a successfiil manipulation check. The
manipulation involved participants king told that they are sexist and this manipulation
was effective d f a r as it aroused more powerful conflict in participants in the treatment
group more than those in the controi group.
The finding that those participants in the experimental condition experienced
significantly more negative emotions and significantly l a s positive motions than those
62
in the control leu& support to the hypothesis that the process of threatening participants
with a negative stereotype about their group Urneases emotional confîict. This is
expected because stereotype threat chalienges a person's positive self-image with a
relevant and plausible negative chanictenstic. The question to be addresseci in the main
body of results is how this conflict influences the behaviour of men as they attempt to
CO- or disconfirm the negative stereotype.
Coagol c o r n - -
: There were 45 participants in the control
group. Of these, six showed they had some iiuight into the purpose of the experiment
during debriehag and were excluded fkom the anaiysis? leaving a total of 39 participants
in the analysis. In the private decision task where the participant was required to make a
decision on his own and in the absence of any stereotype threat, a sizeable proportion
(71.8%) chose the male candidate while the remairing 28.2% chose the female candidate.
No participants chose the third unsuitable candidate.
A X2 gwdness of fit analysis reveaied that in the contcol condition, participants
were significantly more likely to choose the male candidate than the fernale candidate, X2
(1) = 7.4, g < -01.
The sigdicantly large proportion of control participants in favour of the male
candidate confirmeci the intentions of the pilot test, which was to create a profile of a
male candidate who was more favourable than the female candidate, although not
substantially so. The small, but not insubstantial amount of support available for the
female candidate suggests that she was also a potentiaily favourable and competitive
candidate, ody not to the extent that the male candidate was.
The perceived dinerence in favourability betwem the two candidates is important
to the central hypothesis of the present study. This hypothesis aims to ascertain the extent
to which participants who were stereotype threatened not only endorse the f aa l e
candidate, but do so realising that there is a more suitable caadidate who happens to be
male. Clearly, if people are mon willing to support the female candidate who is less
suitable than the male candidate, it can be presumed that there is an underlying motive for
this decision and it is possible that stereotype threat may explain this.
: When asked to make a decision together with
the female conMerate who endorsed the female candidate, 56.4% of the participants
chose the male candidate and 41% chose the femde candidate. Candidates were therefore
no more likely to chwse the male h m the female candidate in this condition, k (1) =
.95, p > .OS, even though the codederate endorsed the female candidate.
When comparing participants' choices across the private and the joint condition,
participants did not choose the female candidate any more often than they chose the male
candidate in the j ~ & decision task than in the priYate decision task, X2(1) = 3.63, p = .M.
The finding that participants were no more iikely to choose the female candidate in the
joint task and in the absence of stereotype threat, suggests that male participants felt their
nonsexist seif-image was unthreatened. They did not wed to prove theu nonsexist beliefs
by altering their behaviour or conforming to the codederate's choice.
S 5 . .
. . Exaen'mefital: There were a total of 53 participants
in the experimental condition and each participant experienced the stereotype tbreat- Of
the 53 participants, during debriehg 8 were excluded afkr revealing insight into the
experiment and the remaining 45 participants formed the experimental group. Of these,
55.6% chose the male candidate while 42.2% chose the femaie candidate. One person
(2.2%) chose the "filler" candidate and was excluded h m M e r analysis, leaving a total
of 44 participants.
In the experimental condition, participants in the pnvate task were therefore no
more iikely than chance to choose the male candidate than they were to choose the €male
candidate, X2(1) = 32, p > .OS. The finding that participauts were equaiiy likely to choose
either the male or female candidate is important because it suggests that participants were
a choosing the better quPLified candidate signincantly more than the female candidate
as would be expected. It is possible that in the experimentai condition, the preseace of the
stereotype threat leâ participants to chose the female candidate as often as they would
choose the male candidate.
In the joint decision task, participants in the experimental condition chose the
male candidate 15.6% of the time and the female candidate 82.2% of the tirne. A 9 test
of goodness of fit reveald that participants were signincantly more likely to choose the
female candidate in the joint task than the male candidate, X2(1) = 20.46, p c -001.
Therefore, participants were highly motivated to choose the female candidate over the
male candidate in the presence of k& stereotype threat and an interaction with a fernale
codéderate. This decision was made despite the fact that the f d e cadidate was
considered to be less qualified.
Significantly more participants chose the fernale candidate in the joint task than in
the pnvate task, X2(1) = 30.01, p c -001. The joint tasir, thetefore, hhad the effect of
increasing participants' willingness to chwse the female candidate- Thus, candidates
whose nonsexïst self-image was threateneà appeared compe1Ied to behave so as to restore
their nonsexist self-image, especially when they were in the presence of a woman.
A k test of goodncss of fit was performed to compare the experimental and
control groups in terms of their choice of male and fernale candidates in the private task.
Participants in the experimental condition chose the female candidate over the male
candidate significantly more often that those in the control grop, XZ(l) = 4.87, p c .OS.
These results suggest that participants contionted with the negative stereotype that
they are sexist behaved in such a way as to appear more favourable to women. in other
words, it appears that they attempted to disconfirm the negative stereotype that they are
sexist. Their attempt ta disconfirm the nonsexist label and present a nonsexist self-image
parallels results found in Study 1, rather than Steele and Aronson (1995) who wodd
predict men would behave in accordance with the sexist stereotype.
Experimental and control participant's choices in the joint decision task were
compared. Results revealed that participants in the experimental condition were
significantly more likely than those in the control group to choose the f d e candidate
over the male candidate in the joint tas& %'(1) = 3 1.82, p c -001.
These results indicate that those exposed to stereotype threat, and then asked to
interact with a woman in a gender relevant task are ultimately fa more likely to choose
the female candidate over the mde candidate compared to those who are not threatened.
In doing so, they are disconfirming the negative stereotype that they are sexist and
promoting a seif-image of themselves as nonsexist and favourable towards women. Given
that men in the control group were significantly less ükely than the experimental group to
choose the femaie candidate, it is clearly the process of stereotype threat that motivated
the enhancement of a nonsexist seKimage and not simply the process of working with a
female confederate alone.
Table 6 summarises the results of the control and experimental conditions m o u
the private and joint decision tasks. In the control condition, participants in the private
task were more k e l y to choose the male candidate. This was anticipated, given that the
two candidates were designed such that the male was süghtly better than the female
candidate. In the joint condition, however, candidates were equally likely to choose either
the male or female candidate. A cornparison of the private and joint decisions for the
control group reveals that participants in the joint task were no more likely to choose the
female candidate over the male than they were in the private task. The results for the
control group therefore reveal that, in the absence of threat, participants were not
motivated to admit that the f d e candidate was any better than the male candidate.
They therefore opt for both the maie and fanale candidate as theù most favourable
choice.
Table 6
Private decision task Joint decision task with a
fernale confiderate
Control More likely to choose Equaljy likely to chwse
male candidate male or fernale candidate
Equally likely to choose More likely to choose
Male or female candidate femaie candidate
Experimental participants more likely to choose female candidate than controls in both
private and joint conditions
Participants in the experimental group, on the other haad, were confkonted with a
negative stereotype about their group, specifically, the notion that men are sexist. In the
decision-making task they were reminded of this negative stereotype insofar as the choice
between a male and female job candidate was relevant to gender discrimination. In this
sense, the task put them "in the field" (Steele, 1997) of the stereotype threat where they
68
were codonted with the possibility of codirmhg the stereotype that as a man, they too
are sexist-
In the private task, participants who were threatened weie equaliy likely to choose
either the male or the f d e candidate. It appears that the stereotype threat prompted
concerns about appearing sexist and participants were therefore motivated to endorse the
female candidate as often as the male candidate so as to reduce concerns about appearing
nonsexist-
In the joint task where the participant was in the presence of a woman who
endorsed the female candidate, he was signincantly more likely to chwse the female
candidate over the male candidate. Concems about connrming the negative stereotype
appear to have strengthened the participants' efforts to appear nonsexist and this was
especially so in the presence of a woman where self-consciousness about sexism rnay be
To f i d e r support the fïnding that participants faced with confirming the
stereotype threat are more motivated to present the female candidate as the most
favourable, qualitative data h m videotaped recordings of behavioural interactions were
analysed. Nonverbai behaviour is considered a rich source of social information as weil as
a genuine reflection regarding how the person is feeling (Hart & Morry, 1996).
Furthermore, behavioural indices have been used in research m e a s h g intergroup
discrimination, including interracial interactions (Word, Zanna & Cooper, 1974).
The aim of the analysis was to rate the male participants' behaviour toward the
woman confederate. Six categories of behavioural measures were used. These included
69
how far the participant chose to sit h m the confederate, an estimation of the number of
words he used to explain his choice to the confederate, amount of eye contact he initiates
with the confederate, whether his interactive style would desmie him as a "foliowei' or
a "leader" in the conversation, his level of assertiveness and hally, the extent to which
he attempts to control the mood, pace and topic of the interaction.
Four raters, consisting of two men and two women, were required to evaluate the
video-taped interactions in terms of the six categories. The raters were b h d to the
condition of the participant. Interrater reliability was calcuiated as an acceptable -70
according to the Spearman-Brown Reliability Coefficient (Rasenthal & Rosnow, 199 1).
The aim of this analysis was to compare individds in the control group who do
not feel the need to present themselves as nonsexist, with those male participants who are
chaiienged with the threat that they are sexist and respond by presenting themselves as
nonsexist. A decision was made to compare the most extrerne prototypes of the threatened
and nonthreatened groups. In the experimental group, a prototypicaily threatened
participant is likely to choose the female candidate in the private task as well as in the
joint task so as to portray hümelfas nonsexist. In the control group, the prototypically
nonthreatened participant is likely to chaase the male candidate in both the private and
the joint tasks because he has no need to protect bis nonsexist selsimage.
A subsample of participants was selected, including 20 control participants who
chose the male candidate in the both pnvate and joint tasks, and 14 experimental
participants who chose the fernale participant in both private and joint tasks.
Table 7
for -via-n C m Proto- C m . .
B ehavioural category Prototype Mean
Seating distance (1 -5) Control
Experimental
Number o f words in explanation for choice (1-1 1) Control
Experimentai
Amount o f eye contact (1-1 1)
Leader or follower (1 - 1 1)
ControL
Experimental
Control
Experimental
Control
Experimental
Control of mood, pace and topic (1-1 1) Conîml
Experimeatal
T-tests performed on each category revealed that prototypical expetimental
participants used significdy fewer words to explain their choice of the fernale candidate
than prototypical wntrol participants who chose the male candidate, l(17.7) = 2.91, p <
.05. Prototypical experimentd participants who consistently chose the female candidate
also made siguficantly less eye contact, t(29) = 2.21,a c .OS, were more assertive, l(32)
= 2.73, p < -05 and were sign.cantly l e s controlling of mood, pace and topic of the
interaction thaa were control participants who chose the male candidate, l(32) = 2.07, p c
-05. There was no signiscant ciifference between control and experimental prototypes'
seating distance fhm the fernale codederate, l(32) = -27, p > -05. Furthermore control
and experimental participants did not ciiffer in terms of whether or not they showed
leadership behaviour, t(32) = 1.48, p > .OS.
To summarise, it appears that participants exposed to the threat that they may be
sexist and choose the fernale candidate on both occasions are significantly less
controlling, make less eye contact and use fewer words to jus- their choice than
participants who are not threatened and who choose the male candidate at both occasions.
Paradoxically experimentd prototypes were judged as more assertive although it is
possible that the threatened participants behaved in a marner that reflected their
cornmitted efforts to appear nonsexist.
It is assurned that the participants who were threatened and chose the female
candidate on both occasions are the participants who are most affected by the experience
of stereotype threat. In an attempt not to wnfhn the stereotype that they are sexist, it is
possible that these men present themselves in the Ieast dominant manner. They are,
therefore, likely to d u c e nonverbal behaviours associated with dominance and
influence, such as fiequent eye contact (Hart & Morry, 1996). This is particularly
important in a situation where the fcmPle codederate is Wrely to remind the participant of
the threat that he is sexist. Thus participants' are Likely to avoid the label of sexist and any
behaviours associateci with it by being less controllhg, by avoiding eye contact and by
using fewer words so as to give the confederate more time to talk
Another explmation for nondominant behaviour associateci with the experimental
prototype is deriveâ h m the fact that these men have aixeady chosen the fwale
candidate and believe thaî by doing so, they have successfidly relieved themselva of the
negative stereotype that they are sexist. Once they have achieved their goal of appearing
nonsexist, their behaviour reflects a more relaxed disposition and they appear l e s
controllin& and use less effortfid behaviour such as speaking or eye contact-
Discussion
The present experiment investigated how male participants respond to a gender
decision-making task d e r being challengeci with the negative stereotype that "men are
sexist". Results reveal that male participants who were chaüenged with the stereotype that
"men are sexist" were more likely to deliberately choose a slightly less qualified female
job candidate than a M y qualified male candidate compareci to a control group who were
not exposed to the stereotype. These results suggest that male participants who are in a
situation where they may CO- the negative stereotype that "men are sexist" make
efforts to behave in a manner that counteracts the sexist stereotype so as to avoid the label
of "sexist". This is especiaiiy ükely to happen when the male is interacting with a female
who has made the choice of the fernale candidate, probably because the presence of the
woman augments the threat of negative stereotype.
Finaliy, men who make the effort to avoid the sexist stereotype by chcoshg the
female candidate when doing the task both on their own and in the presence of a woman
were also more iikely to engage in interpersonaiiy non-dominant interpersonal behaviour
73
that is least likely to characterise them as sexist. That is, when compareci to men who
were not threatmed and who chose the male candidate both on their own and in the
presence of a woman, threatened men made less eye contact, were l e s controlling and
less takative. They were, however, more assertive than nonthreatened participants who
chose the male candidate, aithough this is interpreted as reflecting participants'
committed behaviour to present themselves as nonsexist, rather tha. any attempt to be
dominant.
These results, therefore, support the hypothesis that participants who are
threatened with a relevant stereotype about men and sexism will experience greater
negative &ect because of their concem about being judged as prejudiced and therefore
socially undesirable. In order to discodinn the negative stereotype that he, as a man, is
sexist, the participant makes every attempt to present himselfas nonsexist. When aven
the opportunity, he chooses the female job candidate to strengthen his nonsexist self-
image, particularly in the presence of a woman. He also engages in nondominant
behaviour so as to relieve himself of a sexist label. Participants therefore successfilly
maintain theu positive self-image and avoid fulnlling the negative stereotype.
In contrast, these fïndùig do not support predictions that might arise h m Steele
and Aronson's (1995) theory of stereotype threat when applied to advantaged group
members. Their theory of stereotype threat predicts that participants will react to
stereotype threat with a fear of confimllng the threat and in doing so, will ultimately
CO- the negative stereotype they are trying to avoid Steele and hnson's (1995)
theory and research, however, is based on the experiences of disadvantaged group
memben, including blacks, and women. In contrast the a h of the present study was to
explore the phenornenon of stereotype threat in the context of advantaged group
74
members. Advantaged group participants in Study II responded to a negative and
therefore undesirable stereotype threat about men and sexism by behaving so as to
distance themselves h m the negative stemtype. The advantaged group participants in
Study II may be regardeci as successfiil in their attempts to avoid the unfavourable
stereotype. Advantaged group members therefom behaved differently to the
disadvantaged group members in Steele aud Aronson's (1995) midy. In Steele and
.4ronsons7 (1995) experiments7 disadvantaged gmup members attempted to disconfirm
the negative stereotype about their group, but ultimately behaved so as to c0niï.m the
negative stereotype threat.
Conclusions
The present program of research aimed to explore the potential contlict inherent in
advantaged group members' attitudes towards disadvantaged group memben, including
Blacks and women. An exploratory study and a laboratory experiment were conducted
and a novel and direct means of codbnting participants with their own prejudice was
used.
In Study 1, participants' spontaneous reactions to being provoked were assessed. It
was found that participants expressed less personal than group discrimination.
Furthemore, personal discrimination was associated with negative affect and uncertaintiy
unlike group discrimination, which was unrelated to negative affect and associated oniy
with a lack of uncertainty. The paradoxical low ratings of petsonal discrimination
combined with the increased ratings of negative a@ct and uncertainty were assumed to
be the result of concems about appearing socially desirable in a society that does not
value the expression of prejudice. Finally, participants who expressed slightly higher
ratings of personal discrimination were found to question and wrestie with their
discrimination more and to actively defend it less than participants who may be regarded
as expressing Little or no personal discrimination. High discrimi~tors also expenenced
more negative affect and uncertain@ than low discriminators. Counter to what was
hypothesised, high discriminators were uninhi'bited about their personal discrimination,
not because they have no concems or conflict about their own discrimination, but because
they experience hcreased psychologid connia and at the same thne were also more
willing to express and challenge their own prejudice. The observation that participants are
willing to discuss their prejudice openly in spite of the fact that prejudice is considered
socially undesirable, suggests that participants' value insight into one's own potential
intolerances or that it is seen as desirable to do so. This new form of 'ultramodem' racism
différs h m modem racism, which considers racism as largely covert and disguised.
Indeed, ultramodern racism appears to better describe the new form of racism apparent
torfav. Future tesearch should therefore focus on racism in the context of the political
correctness that is essential to the expression of ultramodern racism.
Study II aimed to f k t challenge participants' nonprejudiced self-image and then
to investigate participants' responses in an interpersonal interaction. Drawing on the
theory of stereotype threat (Steele and Armson, 1995)' male participants were threatened
with the possibility of fulnlling the negative stereotype that "men are sexist". It was
found that participants' who were threatened with the negative stereotype atternpted to
avoid the sexist stereotype by taking the opportunity to deliberately choose a less
qualifieci female candidate for a job on a decision task. In the presence of a woman, where
it is assumed that the impact of the threat is increased, participants' also engaged in fewer
dominant behaviours. by decreasing eye contact and being l e s talkative and less
controliing.
In the face of the threat that they may be sexist, participants' therefore
successfully managed to maintain a sociaiiy desirable nonsexist seK-image. This &ding
does not support the fitadings of Steele and Arotlson's (1995) stereotype threat theory.
Steele and Aronson (1995) found that when disadvantaged group memben were faeed
with a negative stereotype about th& academic competence, their behaviour ultmiately
confirmed the negative stereotype. Using advantaged group members instead of
disadvantaged group members, the hdings of Study II challenge Steele and Aronson's
(1995) findings by reveahg how advantaged gmup members are able to successfûlly
avoid confirming the negative stereotype about theu group.
Both Study 1 and II emphasise how participants who are faced with a threat to
their nonprejudiced selfexperience psychologicaI conflict and may even behave so as to
deliberately overemphasise their nonprejudiced sense of self. The desire to present
oneself as nonprejudiced highlights how sùnilar the findings of Study II are to those of
contemporary theories of prejudice, including modem racisrn, ambivalent racism,
aversive racism and Devine's automatic and controlled processes. Contemporary theories
of prejudice highlight how the denial of prejudice is motivated by an attempt to present
oneself as nonprejudiced. In Study II. however, participants did not so much deny their
own prejudice as they actively behaved in a manner that may be judged as nonprejudiced.
This could be explained by ultramodem racism, which places the emphasis on a person's
awareness and willingness to challenge their prejudiced reactious, rather then the desire to
deny them.
Clearly stereotype threat bas an effect on the behaviour of advantaged group
mernbers. What are the behaviourai consequences of stereotype threat in advantaged
group members? Whîîe the CeSultant behaaviur may be described as 'overcompemation'
or extreme in its attempt to appear nonpnjudiced, it is also possible that it may have a
positive long-tem efféct on interpersonal behaviour. It is hoped tbat such effortful
attempts at egalitarian behaviour wiîl becorne less focussed on protecting one's self-
image and more focussed on genuinely less prejndiceà attitudes and responding.
One limitation of the present research concerns the conf?ontation used to threaten
the participants. While the threat was partially successfid insofar as it ïutpacted on the
behaviour of the male participants, it could be argued that the codhntation was not
powemil enough to evoke a range of potential responses towads the woman. For
example, a man may rwpond to a threat by either behaving in an appropriately nonsexist
manner or by behaving in a deliberately discriminatory and sexist manner because he has
devalued the relevance of the threat, or perceived it as biased or illegitimate.
The second limitation of the present research concerns the homogenous sample
used in both Studies I and II. Participants included only undergraduate, White, male
students who are probably best described as politically liberal. In order to address both
limitations, fùture research should focus fïrstly, on mll ing a less homogenous sample of
participants and secondly, on increasing the impact of the stereotype threat. h is proposeci
that using a less homogenous sample and a more p o w d threat will result in a range of
behavioural respomes, h m sexist to nonsexist. Future research e x p i o ~ g the range of
potential responses is needed before concluding that stereotype threat in advantaged
members always leads to participants' behaving in a manner that counteracts the negative
stereotype.
Finally, the consequences of stereotype threat in both disadvantaged and
advantaged groups may be considered- According to Steele (1997) victims of
discrimination may respond to stereotype threat by confïrming a negative stereotype
about their p u p , for example, the mderachievement of Blacks as a group. On the other
hand, potential perpetrators may respond to stereotype threat by behaving in a manner
that protects them b m attributions of pmjudice associated with the negative stereofypes
that they rnay confkont. For example, they wil l protect themselves h m amibutions of
prejudiced behaviour. Clearly, such an unfortunate situation is iikely to encourage the
persistence of discrimination and to work to the benefit of advantaged group members
and to the detriment of disadvantaged group members.
REFERENCES
Biemat, M., Vescio, TX,, Theno, S.A. & Crandall, C.S. (1996). Values and
Prejudice: Toward understanding the impact of American values on outgroup attitudes. In
e P- of V&es: a, Svmwsium,, 153-189. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Bell, D.W. & Esses7 V.M. (1997). Ambivalence anci response amplification toward
native peoples. Journal, z, 11063-1084.
Crosby, F. Bromley, S. & Saxe, L. (1980). Recent unobtnisive studies of Black and
White discrimination and prejudice: A literature review. -, 82,546-
563.
Devine, P .G. (1 989). S tereotypes and Prejudice: Theu automatic and controlled
components. 5, andSocial 5-1 8.
Devine, P.G. (1991). Rejudice with and without compunction.
pers- S-, a, 81 7-830.
. . . . . . Dovidio, J.F. & Gaertner, S.L. (1 986). m. New
York: Academic Press.
Dovidio, J.F., Mann, I. & Gaertner, S.L. (1989). Resistance to m a t i v e action:;
The implications of aversive racism. In F.A. Blanchard & F. J. Crosby (Eds.), Affimiative S .
(pp 83-103). New York: Springer-Verlag.
Duckitt, J.H. (1994). S- of Pr-. New York: Raeger.
Dunton, B.C. & Fazio, R.H. (1997). An individual difference measure of
motivation to control pmjudiced reactions. pasMinlihr S s ,
Fazio, RH., Jackson, J E , Dunton, B.C., Br Williams,C.J. (1995). Vaiability in
automatic activation as an unobtnisive measure of racial attitudes: A bona fide pipeline?
of P- S- @, 1013-1027.
Gaertner, S.L. & Dovidio, J. (1986). The aversive fom of racism. In J. Dovidîo &
S.L. Gaertner (Eds.), 1 . @p 6 1-89). New York:
Academic Press.
Hass, R.G., Katz, I., Rizzo, N., Bailey, J. & Moore, L. (1 992). When racial
ambivalence evokes negative affect, using a disguiseci meanire of mood
B m , 18,786-797.
Hart, A.J. & ~ o r r y , M.M. (19%). Nonverbai behaviour, race and attitude
attributions. Journal SfExaaimental, 2,165-1 79.
Henry, F. & Tator, C. (1994). The ideology of racism - 'Democratic Racism'.
c s - , m .
Katz, 1. (1981). Stigma: A social psychological analysis. Hilldale, NJ: Erlbaum.
Katz, 1. & Hass, R-G. (1988). Racial ambivaIence and Amencan value conflict:
Correlational and priming studies of dual cognitive structures, JO- of P-
Social, I, 893-905.
Katz, I., Wackenhut, J. & Hass, R.G. (1986). An ambivalence-amplincation theory
of behaviour toward the stigmatised. In S. Worchel & W.G. Austin (Eds.),
(pp 103-1 17). Chicago: Nelson-Hall.
Kinder, D R & Sears, D.O. (1981). Symbolic racism versus racial threats to the
good iife. Joiunat SfPersonalitv, 4,414-43 1.
Lee, T.-Y. & Ottati, V. (1995). Perceived in-group homogeneity as a hc t ion if
group membership d e n c e and stereotype threat. S-, a, 610-619.
McConahay, J.B. (1986). Modem racism, ambivalence, and the modern racism
scale. In J. Dovidio & S.L. Gaertner (Eds.), P r e p - . . . (pp 91-
125). New York: Academic Press.
McConahay, J B - & Hough, J-C. Jr. (1976). Symbolic racism. of S a
Issues, 2 , 2 3 4 5 .
Monteith, M. (1996). Contemporary forms of prejudice-related connict: In search
of a nutsheU. Per- Soc--, 2,461 -473.
Osborne, J.W. (1995). Academics, seIf-esteem and race: A look at the underiying
assurnptions of the disidentification hypothesis. fers- SOC-
-. a, 449-455.
Pettigrew, T.F. (1958). Personaiity and sociocultural factors in intergroup attitudes:
A cross-national cornparison. Jounial, Conflict, 29-42.
Poore, AG., Gagne, F. Barlow, K., Lydon, J.E., Taylor, D.M. & Wright, SC.
(1 997, June). Implications of the personaVgroup discrimination discrepancy in an Inuit
community. Poster presented at the Canadian Psychological Association, Toronto, ON.
Rosenthal, R. & Rosmw, R (1991). of e v i o ~
d D m (2nd Edition). New York: McGraw-Hill.
Ruggiero, K.M. & Taylor, DM. (1 997). Why minority group mernbers perceive or
do not perceive the discrimination that confiants thern: The role of self esteem and
perceived control. Journal S f P e n o n a l i t v , n, 373-389.
Steele, CM. (1997). A threat in the air. How stemtypes shape inteilectual identity
and performance. Psvcholonist,, 613-629.
Steele, CM, & Aronson, J. (1995). Steteotype threat and the intellectual
performance o f Afiican Am~cans . Jounial S a n d P&&gy, a, 797-8 1 1.
Swim, J.K., Aikin, K.J., Hall, W .S. & Hunter, B A (1995). Sexism and racism: OId
fashioned and modern prejudices. S a n d S o c i a L P S y E b p j q W . 68,199-
214.
Taylor, DM., Ruggiero, KM., Louis, W.R (1996). personaVgmup discrimination
discrepancy: Towards a two-factor explanation. of -,
a, 193-202.
Taylor, D.M., Wright, SC., Moghaddam, F.M. & Lalonde, R.N. (1 990). The
personal/group discrimination discrepancy: Perceiving my group, but not myself, to be a
Target of discrimination. and-, 16,254-262.
Taylor, D.M., Wright, S C . & Porter, L.E. (1994). Dimensions of perceived
discrimination: The personal/group diMimination discrepancy. In M.P. Zama & J.M.
. * Olson (Eds.), The of f , 2,233-255-
Killsdale, NJ, USA: Laurence Eslbaum Associates.
Word, CO., Zanna, M.P. & Cooper, J. (1974). The nonverbal mediation of self-
fulnlling prophecies in interracial interaction. Journal of F - m i - a I ,
u, 109-120.
1. Racism cannot exist within a democratic society
2. Racism is a consequence of immigration; racial confiict cornes about because there are
too many different people here.
3. Minority groups rehw to fit in and adapt to Canadian society-
4. The problems experienced by rninority groups are fkequently cultural problems and
have nothing to do with race.
5. Non-whites corne to Canada to benefit from the social advantages of living in this
country.
6. The policies of multicuituralism sufnciently address the problems of racism.
7. Racism cornes h m ignorance and what is needed is education about other people
which focuses on changing attitudes.
8. The over-sensitivity of people of colour makes them look for racisrn where it is not
intended.
9. Strategies such as employment equity interfere with the basic tenets of fieedom of
individual thought, choice and action on wbich notions of western democracy are
founded,
10. Anti-racism initiatives are racism in reverse.
Instructions: please read the following and answer the questions.
Job description:
Marketing Research Company seeh to nII the position o f an Administrator. Applicants should have leadership and people skiiis. M u t be highly proficient in Microsoft sofhvare, particuiarly Microsoft Word. 5 years experience in computing necessary. Must be
[ bilingual (reading and writing). Salas) negotiable. 1
PROFILE ONE
Born 1963
1992- 1997 Media designer for Consumer Health Board. Some experience in Microsofi soffware.
1 984- 1988 Worked as a secretary for Dees, West & Thompson. Training and experience in corporate and commercial law with (Macintosh) cornputers.
1980-1983 Certincate in Business Management.
Assets: communication skiils Bilingual Responsible
Comment: Sue Lessing is a responsible worker with a range of work expenence. She has the skiils to motivate and encourage people in the work environment. Her present job provides her with some experience in Microsoft software.
PROFILE TWO
Born 1962
1992- 1997
l989-lW2
1983-1988
1980-1 983
Assets:
Comment:
Technical support person with Jemdye Design. Experience with Microsofl products.
Computer technician for Desig Inc.
Sales person for Downtown Publishiog Company
Certificate in Marketing & Computer Engineering.
Computer skills in many languages B ilingual Outgoing personality.
Matthew Morgan is a skilied & responsible worker with good knowledge & experience in various software packages, inctuding Microsott He is hard working.
PROFILE THREE
Born 1965
1994- l997 Computer support for Maidiame system at Hardy & Smiths Architects computer lab. Experience in Pascal, Delphi & some experience in Microsoft.
1 986 - 1993 Computer technician in factory setting.
1984-1985 Part-time one year course at Dawson College in computers. Obtained certificate in cornputm.
1982- 1983 Traveled.
Assets: Knowledge of compter skills & languages Willing to l e m new skills Bilingual
Comment: Mark Lacey has a technical background in computers. He is a willuig worker with potential to learn more skiUs. He is also willing to work over- tune. He has some experience in Microsoft, but would require more training. His leadership skills are not clear.