Anda di halaman 1dari 13

Group Processes & Intergroup

Relations
http://gpi.sagepub.com/

A naturalistic study of stereotype threat in young female chess players


Hank Rothgerber and Katie Wolsiefer
Group Processes Intergroup Relations 2014 17: 79 originally published online 25 June 2013
DOI: 10.1177/1368430213490212

The online version of this article can be found at:


http://gpi.sagepub.com/content/17/1/79

Published by:

http://www.sagepublications.com

Additional services and information for Group Processes & Intergroup Relations can be found at:

Email Alerts: http://gpi.sagepub.com/cgi/alerts

Subscriptions: http://gpi.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://gpi.sagepub.com/content/17/1/79.refs.html

>> Version of Record - Dec 25, 2013


OnlineFirst Version of Record - Jun 25, 2013

What is This?

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


490212
2013
17110.1177/1368430213490212<italic>Group Processes & Intergroup Relations</italic>Rothgerber and Wolsiefer

G
Group Processes & P
Intergroup Relations I
Article R

Group Processes & Intergroup Relations

A naturalistic study of stereotype


2014, Vol 17(1) 79­–90
© The Author(s) 2013
Reprints and permissions:
threat in young female sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1368430213490212
chess players gpir.sagepub.com

Hank Rothgerber1 and Katie Wolsiefer2

Abstract
The present research sought to determine whether young female chess players would demonstrate
stereotype threat susceptibility in a naturalistic environment. Data from 12 scholastic chess
tournaments indicated that females performed worse than expected when playing against a male
opponent, achieving 83% of the expected success based on their own and their opponent’s prerating.
These effects were strongest for the youngest players in lower elementary school but also present for
those in upper elementary. Stereotype threat susceptibility was most pronounced in contexts that could
be considered challenging: when playing a strong or moderate opponent and when playing someone
in a higher or the same grade. As evidence of disengagement, those most vulnerable to stereotype
threat were less likely to continue playing in future chess tournaments. These results were not found in
a matched comparison male group suggesting the outcomes were unique to stereotype threat and not
universal to young chess players.

Keywords
chess, stereotype threat, stereotype threat development
Paper received 18 September 2012; revised version accepted 9 April 2013.

What has made stereotype threat such a fruitful Perhaps in recognition that chronic stereotype
area for researchers is also what makes it such threat may become self-perpetuating and highly
a stubborn social problem. Stereotype threat resistant to change, some researchers have sought
is wide-reaching: It is indiscriminate in curs- to determine when individuals initially experience
ing any group for which a negative stereotype stereotype threat. We share the belief that
applies, and it does so across a range of domains
from intellectual (Cadinu, Maass, Rosabianca, 1
Department of Psychology, Bellarmine University, USA
& Kisener, 2005; Steele & Aronson, 1995) to 2
Department of Psychology & Neuroscience, University of
physical (Chalabaev, Sarrazin, Stone, & Curry, Colorado, USA
2008; Stone, Lynch, Sjomeling, & Darley, 1999).
Corresponding author:
What is also so striking and debilitating about the Hank Rothgerber, Department of Psychology, Bellarmine
phenomenon is how seemingly easily stereotype University, 2001 Newberg Rd, Louisville, KY 40205, USA.
threat can be activated. Email: hrothgerber@bellarmine.edu

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


80 Group Processes & Intergroup Relations 17(1)

successful intervention may be predicated upon criticism of stereotype threat research focuses on
understanding the developmental aspects of stere- whether the mostly lab-based findings are gener-
otype threat susceptibility in children. The current alizable to real-world settings. For example,
research specifically examines whether stereotype Cullen, Hardison, and Sackett (2004) used archi-
threat susceptibility occurs in young populations val data to assess the degree to which the
and distinct from the literature, examines them not Scholastic Assessment Test (SAT) predicted aca-
in a lab but in natural environments. demic performance of women and Blacks in
A few researchers have moved away from math and the Armed Services Vocational Battery
studying stereotype threat in the traditional popu- predicted military performance among Blacks in
lation of college students and begun investigating the domain of technical proficiency. Contrary to
younger participants. Experimental manipula- stereotype threat predictions, only linear test–
tions in these studies have included diagnostic performance relationships were found as were
versus nondiagnostic testing conditions for eth- equivalent effects for women, men, and Blacks.
nic stereotypes (McKown &Weinstein, 2003), In a second study, Cullen, Waters, and Sackett
priming gender identity before a math test (2006) also examined stereotype threat suscepti-
(Ambady, Shih, Kim, & Pittinsky, 2001; Neuville bility with a large dataset in a natural setting. SAT
& Croizet, 2007), informing females that a geo- math scores of women intending to major in
metrical problem measured ability in geometry as math or a related field did not underpredict per-
opposed to ability in drawing (Huguet & Regner, formance in a separate area (i.e., English grades)
2007), exposing females to vignettes about compared to women not identified with math, at
famous mathematicians (mostly male) before a least any more so than they did with men. This
math test (Muzzatti & Agnoli, 2007), and leading runs counter to predictions that individuals most
females to believe that their chess opponent was highly identified with a domain are more suscep-
male as opposed to female (Wolsiefer & tible to stereotype threat.
Rothgerber, 2013). One of the reasons that stereotype threat
Overall, the limited research on stereotype effects may be absent or less prominent in these
threat susceptibility in children is not entirely real-world examples is that there may be stronger
consistent, but there are a few generalizations incentives for success in the natural environment
that can be drawn. Children in lower elementary than in laboratory examples, heightening effort
school—as young as 5—seem vulnerable to ste- and motivation. It is known that when monetary
reotype threat (Ambady et al., 2001; McKown & incentives are present, stereotype threat suscepti-
Weinstein, 2003; Wolsiefer & Rothgerber, 2013). bility dissipates, whether on a cognitive ability test
Among females living in the USA anyway, those (McFarland, Lev-Arey, & Ziegert, 2003; Nguyen,
in upper elementary school seem less vulnerable O’Neal, & Ryan, 2003) or a retail management
to stereotype threat (Wolsiefer & Rothgerber, selection test (Ployhart, Ziegert, & McFarland,
2013) and in some cases, there actually appears to 2003).
be a rebound effect whereby exposure to the ste- If external validity concerns are present in
reotype boosts performance (Ambady et al., studies of stereotype threat among adult college
2001). As postulated by Steele (1997), personal students, they certainly are not diminished when
endorsement of the stereotype does not seem to studying younger populations. None of the stud-
predict stereotype threat susceptibility (e.g., ies assessing stereotype threat susceptibility in
Huguet & Regner, 2007), but awareness of the younger participants have been conducted in real-
stereotype seems necessary to produce underper- world settings. Rather, the typical situation con-
formance effects (McKown &Weinstein, 2003; sists of a researcher capturing participants’
Wolsiefer & Rothgerber, 2013). attention, controlling all factors except the threat
The present research sought to examine some manipulation, and introducing a test of perhaps
of these issues in a more realistic context. One limited importance to the children.

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


Rothgerber and Wolsiefer 81

Elaborating on these concerns in reverse to eliminate or unsettle the effects of stereotype


order, there are simply few, if any, negative exter- threat.
nal consequences for poor performance or posi- In the present research, we attempted to clar-
tive external consequences for good performance ify some of these issues by studying the develop-
by these young participants. Although in the real ment of stereotype threat in young female chess
world positive math performance relates to players in a natural setting. Chess is widely
higher grades, scholarships, recognition, parental believed to be a male game, most commonly
praise, etcetera, and chess performance results in played and mastered by men. For example, in a
an increased rating, trophies, recognition, etcet- 1963 interview, future world champion Bobby
era, in these studies such rewards were absent. In Fisher remarked that women were “terrible chess
one study (McKown & Weinstein, 2003), the players” and thought they lacked the intellectual
tasks were probably so unusual to participants capacity for the game. In one study, when experi-
that it is difficult to evaluate how seriously they enced, expert female chess players were told their
considered their performance. Even if partici- Internet opponent was male as opposed to
pants were highly motivated, these studies may female, they were much more likely to lose
overestimate the impact of stereotype threat sus- (Maass, D’Ettole, & Cadinu, 2008). Stereotype
ceptibility by creating an unrealistic environment threat may offer one explanation for why females
in which it was highly likely that participants tend to be less prominent among chess elites. In
would attend to the threat activation. In Ambady fact, of the top 100 chess players in the world,
et al. (2001), for example, older participants expe- only one is female (www.fide.com). It is possible
rienced threat by answering questions about the that the disidentification that results from nega-
relevant identity, and in Wolsiefer and Rothgerber tive stereotype threat effects may cause female
(2013), the threat manipulation involved either chess players to no longer identify themselves as
displaying a face of a male or female picture to competent chess players. As a result, females may
participants. In the real world of course, many be less likely to play chess or to stay with it at
other factors are present that may distract young- higher levels. From our perspective, there are cer-
sters from noticing or thinking about these threat tain benefits to studying chess players (see
concerns. That is, these laboratory studies may be Wolsiefer & Rothgerber, 2013). One advantage is
creating conditions where gender or race is made that there are officially rated United States Chess
particular salient, and stereotype threat may be Federation (USCF) tournaments occurring regu-
less likely in the absence of such explicit manipu- larly throughout the USA allowing a rich data set
lations. In short, in the real world, situational fac- of performance outcomes.
tors promoting stereotype threat may be less
salient, or even if attended to, may be remedied
by incentives provided in a natural context. Stereotype Awareness
It is possible, then, that experimental research Before data collection for the main study, we
examining when stereotype threat susceptibility sought to assess whether young females were
develops may be distorting how readily such aware of chess-related gender stereotypes. To
effects are actually evoked in the real world and measure stereotype awareness, an 11-item verbal
may be encouraging researchers to conclude that questionnaire was administered as part of a larger
these effects are more common than they really study to 77 female chess players enrolled in after-
are in children as young as 5. Besides the theo- school chess programs in metropolitan Louisville,
retical importance of addressing how stereotype Kentucky, USA. Participants were distributed
threat susceptibility develops in a natural con- evenly from ages 6–11. Sample items included: (a)
text, practically, such information is important in “Have you heard that good chess players are usu-
helping concerned individuals better understand ally boys?” (b) “Have you heard that girls are sup-
when the critical age is for interventions designed posed to be good at chess?” (reverse scored).

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


82 Group Processes & Intergroup Relations 17(1)

Responses of Yes were coded as 2, Maybe as 1, and January 2008. The tournaments were restricted to
No as 0. Stereotype awareness was operationally elementary, middle, and high school participants,
defined as the total sum of all response scores on with individuals competing in one of four sepa-
the questionnaire; thus the range of scores was rate sections based on their grade. The tourna-
0–22 with 22 indicating a high level of stereotype ments were all officially rated by the USCF,
awareness. The reliability of the items was rea- meaning that each participant had a prerating and
sonable (α = .76). Overall, the average female a postrating. Participants competed in five games
reported stereotype awareness above the mid- during each tournament under G/30 time con-
point of the scale, M = 13.03, SD = 4.05; stereo- trols (i.e., each player had 30 minutes for the
type awareness was uncorrelated with age, r = .11, game), typical for scholastic tournaments in the
ns. Thus, it appears that even young females have USA. For each tournament, several local tourna-
been moderately exposed to ideas that chess is a ment directors provided information about each
male activity. player’s grade and sex. Initially, there were a few
Having shown that young females have some discrepancies, but further inquiry resolved these
awareness of chess-related gender stereotypes, cases. In the end, 219 females ranging from kin-
we next moved on to examine whether they dergarten age to ninth grade participated in the
would be affected by such stereotypes. To assess selected tournaments. The specific breakdown
whether stereotype threat susceptibility was pre- was as follows: 13 were enrolled in kindergarten;
sent among a young population in a real-world 47 in first grade; 33 in second grade; 45 in third
setting, the researchers examined USCF tourna- grade; 28 in fourth grade; 20 in fifth grade; 13 in
ment records of female chess players from sixth grade; 9 in seventh grade; 8 in eighth grade;
Kindergarten to ninth grade. We measured their and 2 in ninth grade. Participants’ mean USCF
general performance against males along with rating was 357.7 (SD = 280.3), with grade break-
their performance in certain conditions that downs as follows: kindergarten M = 123, SD =
should provoke greater threat, such as playing 44; first grade M = 187, SD = 116; second grade
against an older or higher rated opponent. M = 332, SD = 225; third grade M = 305, SD =
Because the research on the presence of stereo- 181; fourth grade M = 417, SD = 215; fifth grade
type threat among adults in natural settings is M = 300, SD = 250; sixth grade M = 716, SD =
mixed, it was unclear whether previous findings 216; seventh grade M = 798, SD = 420; eight
showing stereotype threat susceptibility in lower grade M = 901, SD = 105; ninth grade M = 900,
elementary school females and stereotype lift SD = 320.
among upper elementary school females would To control the possibility that all young chess
hold. Additionally, we measured the extent to players generally perform worse than expected,
which these female chess players continued to worse under certain difficult circumstances such
play in chess tournaments following our study of as playing a strong opponent, or worse when spe-
them. To the extent that our female sample would cifically playing against boys, we included a male
show evidence of stereotype threat, we expected comparison group whose performance against
that those experiencing it the most would disen- other males was assessed. Every female partici-
gage from chess and play in fewer future pant in the study was matched with a male of
tournaments. similar chess rating and grade who played in the
same tournament. The decision rule in choosing
a male counterpart was to select the male com-
Method petitor of the same grade who was closest in rat-
ing to the female participant. If a male participant
Participants and Procedure within 100 rating points of a female participant
Data were collected from 12 chess tournaments could not be located, a similarly rated male one
held in Louisville, KY from November 2006 to grade higher or lower than the female participant

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


Rothgerber and Wolsiefer 83

was chosen. In some cases, no appropriate coun- Expected/obtained winning percentage difference. Rat-
terpart could be found. In the end, 195 males ing differences may not always be the most sensi-
were chosen for the study. Their average rating tive measure of performance because when one
(M = 356.3; SD = 274.5) was nearly identical to plays a much lower or much higher rated oppo-
that of the female participants as was the specific nent and performs as expected, there will be little
rating breakdown by grade. Additionally, the aver- rating fluctuation. From information provided by
age grade of the male participants (M = 2.96; the USCF, we also calculated an expected winning
SD = 2.03) was highly similar to that of female percentage for each game played by the partici-
participants (M = 3.13; SD = 2.12). pants in the study and compared it to their
In addition to determining if there was an obtained winning percentage. The expected win-
overall stereotype threat effect when playing ning percentage was obtained through the follow-
against males, the study examined whether more ing formula: We (R, R1) = 1/1 + 10-(R-R1)/400,
challenging circumstances would activate where We is the standard winning expectancy
stronger threat. To assess this, a continuous between a player rated R and their opponent
measure of relative opponent strength was cal- rated R1. The obtained winning percentage was
culated for each opponent by subtracting the defined as 100% for a win, 50% for a draw, and
opponent’s rating from the participant’s prerat- 0% for a loss. Similar to pre/post rating differ-
ing. Opponents 1 SD above the mean on relative ences, the expected winning percentage is based
strength were labeled strong, at the mean level of on the preratings of the two competitors. A low-
opponent strength as moderate, and 1 SD below ered obtained relative to expected winning per-
the mean on opponent strength as weak. The centage could not be explained simply as an
male opponent’s grade level relative to the par- outcome of playing stronger opponents, for the
ticipant’s was also recorded. In this case, divi- formula takes this into account. Rather, a lower
sions were made for opponents of a higher grade, obtained winning percentage would be evidence
opponents of the same grade, and opponents of a lower of something disrupting an individual’s normal,
grade than participants. baseline chess ability.

Continued engagement with chess.  Several measures


Measures were used to determine how much participants
Pre/post rating difference.  For each game that the stayed engaged with chess following the tourna-
participants played against a male, their prerating ment in which we examined them. Using the
was compared with their postrating, as calculated United States Chess Federation website (http://
by the researchers. To calculate postratings, we uschess.org/) which provides participation data
used a rating estimator worksheet provided by the for every USCF-rated tournament and allows
USCF (available at http://www.glicko.net/rat- searches to be made by player, the researchers
ings/approx.pdf). Details on the complex algo- computed the number of tournaments played in the
rithm used to calculate ratings can be found in a year following the tournament in which we stud-
report by Glickman and Doan (2011), “The ied them. We also computed how many months (0–
USCF Rating System” available on the USCF 12) passed until they played in their next
website (Glickman & Doan, 2011). Simply stated, tournament that year from the time we studied
an individual’s postrating is based on the outcome them. These two items were negatively correlated
of the match (i.e., win, loss, or a draw), their pre- (r = −.33, p < .001).
rating, and their opponent’s prerating. Beating a
stronger opponent will increase one’s rating more
Results
than beating a weaker opponent, and losing to a
weaker opponent will decrease one’s rating more To account for nonindependence due to par-
than losing to a stronger opponent. ticipants who played multiple games, multilevel

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


84 Group Processes & Intergroup Relations 17(1)

models allowing the intercept and effect of Table 1.  Expected/obtained winning percentage
within-subjects predictors (i.e. opponent differences by gender and opponent type.
strength and opponent grade level) to vary were
Gender Exp/obt win% T
used to analyze these data1 (for further details
see Singer, 1998). The analysis followed the fol- Females
lowing form: first, the interactions between   Strong opponent .09/.05 −4.47***
gender and overall measures (i.e., pre/post rat-   Moderate opponent .39/.32 −4.70***
  Weak opponent .71/.59 −0.77
ing and expected/obtained winning percentage   Older opponent .37/.27 −3.78***
difference) were calculated; then, overall per-   Same opponent .41/.34 −5.63***
formance effects were calculated for each gen-   Younger opponent .42/.36 −2.00+
der separately. Following the overall analysis, we  Overall .41/.34 −5.52***
examined whether participants were affected by Males
the contextual variables (i.e., rating level and   Strong opponent .09/.19 4.93***
grade level of opponent) that may activate ste-   Moderate opponent .38/.46 4.89***
  Weak opponent .70/.73 2.40*
reotype threat. After assessing the interaction in   Older opponent .32/.36 1.21
each condition, analysis was conducted on each   Same opponent .37/.44 4.89***
gender. Once the overall and contextual effects   Younger opponent .38/.50 3.73***
were determined for all participants, the analy-  Overall .37/.44 4.88***
sis explored these effects in different age Note. Because opponent strength was treated as a continuous
groups. In trying to approximate divisions in predictor, performance versus a strong, moderate, and weak
U.S. schools, participants were divided into opponent refers to the expected and obtained winning per-
three groups based on grade level: lower ele- centages at 1 SD above the mean relative opponent strength,
at mean relative opponent strength and at 1 SD below the
mentary, K–2; upper elementary, 3–5; and mid- mean relative opponent strength respectively.
dle school, 6–8. Ninth graders were included +p<.10, *p <.05 **p<.01 ***p<.001.
with the middle school grouping. Results were
highly similar between pre/post rating differ- differences between males and females when
ences and expected/obtained winning percent- opponent strength and grade level were taken into
ages for performance effects and correlations account2 (stronger: b = −.07, t(790) = −5.26, p <
between performance and continued engage- .001; moderate: b = −.07, t(370) = −7.33, p < .001;
ment with chess; for brevity, only winning per- weaker: b = −.08, t(754) = −5.45, p < .001; higher:
centage will be further discussed. b = −.08, t(243) = −3.59, p < .001; same: b = −.08,
t(398) = −7.43, p < .001; lower: b = −.08, t(305) =
−4.13, p < .001. Female failure to achieve expected
Overall Effects outcomes was true for a variety of contexts that
The Gender x Winning Percentage interaction could be perceived as particularly threatening:
reached statistical significance, t(394) = −7.34, p when playing a strong and moderate opponent,
< .001. Table 1 presents means and t/p values for and when competing against an opponent from a
the overall study effects. Overall, the obtained higher grade, the same grade and marginally, even
winning percentage for female participants in the from a younger grade. Males exceeded expecta-
study failed to measure up to their expected win- tions in all contexts except when playing an oppo-
ning percentage. Male participants actually per- nent from a higher grade level, and in this case,
formed better than expected. there was no effect.

Opponent Strength and Grade Level Effects Grade Effects


Every two-way interaction was significant when Table 2 presents stereotype threat susceptibility
comparing expected/obtained winning percentage by grade level with and without consideration of

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


Rothgerber and Wolsiefer 85

Table 2.  Expected/obtained winning percentage


differences by grade, gender, and opponent type. Gender Exp/obt T
win%
Gender Exp/obt T
win%   Same opponent .46/.46 0.14
  Younger opponent .70/.72 0.27
Females   Overall .45/.46 0.07
 K–2
  Strong opponent .05/.02 −2.43* Note. Because opponent strength was treated as a continuous
  Moderate opponent .37/.28 −4.80*** predictor, performance versus a strong, moderate, and weak
opponent refers to the expected and obtained winning per-
  Weak opponent .72/.57 −3.96*
centages at 1 SD above the mean relative opponent strength,
  Older opponent .43/.27 −3.88*** at mean relative opponent strength, and at 1 SD below the
  Same opponent .43/.32 −4.81*** mean relative opponent strength respectively.
  Younger opponent .42/.27 −2.85** +p < .10; *p < .05; **p < .01; ***p < .001.
  Overall .42/.32 −4.96***
 3–5
  Strong opponent .10/.08 −0.76*** context. Female participants in both lower and
  Moderate opponent .40/.34 −3.02**
  Weak opponent .72/.60 −3.60***
upper elementary grades did not obtain a winning
  Older opponent .38/.26 −2.01* percentage as high as would have been expected.
  Same opponent .37/.33 −3.16** The youngest participants displayed performance
  Younger opponent .37/.33 −1.01 deficits under all conditions: when playing strong,
  Overall .39/.33 −2.84** moderate, and weak opponents, and opponents
 6–9 from a higher, same, and lower grade, their
  Strong opponent .11/09 −0.83 obtained winning percentage was significantly
  Moderate opponent .39/.36 −1.03
  Weak opponent .67/.64 −0.82
lower than expected. Those in upper elementary
  Older opponent .30/.28 −0.30 school obtained a lower than expected winning
  Same opponent .46/.43 −1.01 percentage when confronting all opponents
  Younger opponent .64/.64 0.38 except those from a lower grade. Older partici-
  Overall .46/.42 −1.03 pants were not vulnerable to underperforming
Males expectations. That is, their obtained winning per-
 K–2 centage matched expectations. For males, the
  Strong opponent .06/.11 0.78
  Moderate opponent .37/.43 1.57
obtained winning percentage was generally simi-
  Weak opponent .72/.74 1.39 lar to the expected winning percentage, except
  Older opponent .34/.38 1.13 for one condition: males in grades 3–5 performed
  Same opponent .40/.44 1.43 better than expected in all cases except when
  Younger opponent .33/.36 0.13 playing an opponent from a higher grade.
  Overall .39/.43 0.88
 3–5
  Strong opponent .10/.27 6.15*** Correlations Between Stereotype
  Moderate opponent .39/52 5.38***
  Weak opponent .70/.77 2.32*
Threat and Continued Engagement in
  Older opponent .28/.38 1.47 Chess
  Same opponent .32/.45 5.14***
  Younger opponent .33/.50 4.47***
A clear gender difference occurred in how per-
  Overall .32/.46 5.75*** formance related to continued engagement in
 6–9 chess. Table 3 presents these correlations. For
  Strong opponent .13/.14 0.13 females, nearly all correlations were significant
  Moderate opponent .38/.39 0.09 between performance impairment associated
  Weak opponent .65/.65 0.01 with stereotype threat and continued engage-
  Older opponent .33/.26 −1.25 ment in chess. Overall, females with depressed
(Continued) performance (i.e., a larger expected than obtained

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


86 Group Processes & Intergroup Relations 17(1)

Table 3.  Correlations between expected/obtained winning percentage difference, tournaments played during
the following year, and months since the last tournament.

Female expected/obtained % Male expected/obtained %


(n = 219) (n = 193)
Overall
 Tournaments −.29*** .02
 Months .29*** .03
K–2
 Tournaments −.38*** −.16
 Months .32** .20+
3–5
 Tournaments −.28** .04
 Months .28** −.14
6–9
 Tournaments −.43* .25
 Months .12 −.22
Note. +p < .10; *p < .05; **p < .01; ***p < .001.

winning percentage) played in fewer tourna- beliefs emerge to the detriment of females when
ments the next year and allowed a greater num- they compete with boys. The current results
ber of months to elapse from their last extend those of Wolsiefer and Rothgerber (2013)
tournament. This was true for all grade levels by showing that this form of threat is strong
except older females, in which case months from enough to persist in real-world situations when
their last tournament did not show this effect. attention is not as controlled, with many other
For males however, performance was virtually competing stimuli in the environment, and
unrelated to future chess participation. None of stronger motivation to succeed present among
the correlations were significant, except in the individuals. Evidence of stereotype threat among
case of the youngest males. For K–2 boys, young children, then, cannot be dismissed merely
months from the last tournament were margin- as an artifact of, or limited to experimental
ally correlated in the same direction as the female paradigms.
data. That is, poor performance only predicted Performance deficits triggered by stereotype
less future chess participation for the youngest threat were more likely to appear when the young
of boys. female sample was playing males in more chal-
lenging situations. Specifically, females did worse
than expected when competing against a strong
Discussion opponent (achieving 56% of what would be pre-
In this examination of data from a realistic set- dicted based on prior ratings), a moderate one
ting, females playing against males performed at (82% of expected performance), an opponent
about 83% (.34 / .41) of expectations based upon from a higher grade (73% of expected perfor-
their own and their opponents’ preratings. mance), and an opponent of the same grade
Because these measures adjust for rating strength, (83% of expected performance). This supports
the present results do not simply demonstrate the contention that anxiety, arousal, and other
that girls lose to boys more often than vice versa, processes related to stereotype threat only impair
but rather that girls lose to boys at a rate that cannot be performance on difficult, challenging tasks
explained in terms of initial rating strength. Because (Steele, Spencer, & Aronson, 2002). To beat
young girls are aware of negative stereotypes someone in chess who is better than or equal to
about their chess-playing ability, it seems these oneself requires complete concentration and

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


Rothgerber and Wolsiefer 87

focus; it may be that playing against a boy diverts about 85% of what was expected. This finding is
some of these resources toward worrying about inconsistent with experimental evidence (Ambady
one’s skills and how one will be perceived by oth- et al., 2001) who noted that females in upper ele-
ers, thus lowering performance. When competing mentary school experience a lift in performance
against weaker opponents and opponents from when stereotype threat is activated. The afore-
lower grades, young females performed about as mentioned researchers have speculated that
expected or only marginally worse so. This sug- increased chauvinism may lead upper elementary-
gests that playing someone perceived as weaker aged girls to excel when working under stereo-
or younger may buffer the negative impact of type threat. It may be that this chauvinism is
playing against a male chess player. Especially especially likely to be triggered when participants
when playing a weaker opponent there may be are aware that they are being studied in an experi-
less frustration that is critical for the activation of ment or when placed under more artificial situa-
stereotype threat (Steele et al., 2002). tions; these contexts may increase reactance
Another major outcome was that the youngest against the stereotype. Secondly, in the experi-
female chess players we studied were the most ments individuals are studied in isolation, but
susceptible to stereotype threat activated by com- because tournaments involve social interaction
peting against a male. Those in lower elementary (especially between rounds) and the visible pres-
achieved an obtained winning percentage slightly ence of many others, social dynamics may have
less than 76% of their expected winning percent- caused these upper elementary girls to be less
age when playing against a male. That stereotype successful at fending off stereotype threat.
threat would be found in females at such a young Perhaps seeing themselves as a relative numerical
age supports prior research (Ambady et al., 2001; minority in the tournament activated threat in
McKown & Weinstein, 2003; Neuville & Croizet, these girls to a greater degree than would have
2007; Wolsiefer & Rothgerber, 2013) showing occurred in more artificial settings. In general,
performance deficits in young participants. It also these females were less prone to contextual varia-
suggests that even when young children face mul- tion, although they did significantly underachieve
tiple stimuli and have strong incentive for good (two thirds of expected performance) when com-
performance, they still notice and are affected by peting against stronger males.
stereotype threat. This can be seen in the espe- For the oldest participants in the sample,
cially poor performance outcomes under what playing against a male opponent did not signifi-
could be construed as more challenging circum- cantly lower overall performance. One possible
stances. For example, when playing a stronger explanation is that these older players represent a
opponent, lower elementary-aged girls only select sample that has been immune to the attri-
achieved 40% of the success that would have tion common to female chess players. Selection
been expected based on their own and their pressures have been shown to impact older
opponents’ past performances (again, this pro- female chess players (Vollstadt-Klein, Grimm,
found underachievement cannot be framed Kirsch, & Bilalic, 2010), but it is unclear whether
merely as individuals not doing well when playing such pressures would have exerted influence
stronger opponents because the winning percent- over the younger participants in the current
age formula adjusts for rating strength). When research. It may well be that the youngest chess
playing a boy from an upper grade, they achieved players who are most susceptible to stereotype
success at 63% of what would have been threat experience disidentification (Steele, 1997)
expected. and thus may be less likely to continue playing
Those females in upper elementary school chess in later childhood. The smaller number of
also showed evidence of stereotype threat middle school females in the tournaments we
although not as strong as with the younger par- studied may reflect this as well. On the surface,
ticipants. Third, fourth, and fifth graders achieved the lack of overall effects for middle-school

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


88 Group Processes & Intergroup Relations 17(1)

participants seems difficult to reconcile with stronger by the fact that for the most part, males
recent evidence of stereotype threat susceptibil- who did worse when playing other males did not
ity among adult chess players (Maass et al., 2008). disengage from chess any more than males whose
Because this latter study examined expert players performance exceeded expectations.
who are heavily identified with chess, it may be Finally, the fact that stereotype threat effects
that they are more vulnerable to stereotype threat are evident—indeed even strongest—at such a
associated with it. young age in a natural setting indicates that inter-
It should be noted that all of the aforemen- vention or prevention efforts must occur as early
tioned female results cannot be attributed to as possible. Borrowing from the existing litera-
young players’ general underperformance in ture on reducing stereotype threat, we suggest
chess tournaments, underperformance against that programs teaching chess to young girls
strong opponents, or underperformance when include one or more of the following: providing
competing against male opponents. For male female role models whether it be as instructors or
participants in the study did not show evidence showcasing top female players and their games;
of underperformance as females did. In the reframing the task as a problem-solving activity
absence of stereotype threat, male participants or concentration game and away from chess; de-
were likely better able to concentrate and focus emphasizing female identity in lieu of other iden-
on the game itself, especially important when tities such as being a member of the school chess
playing stronger opponents. In many cases, males team, grade level, etcetera; emphasizing high
exceeded rating performance and expectations. standards for all participants with assurances
This reinforces our interpretation that there is about the capability of meeting them; emphasiz-
something specific to the interaction between ing an incremental view of chess performance
female and male competitors that produced (i.e., that one’s ability is not constant but can be
these performance deficits in females. Although improved by practice and study); and providing
the retrospective nature of data collection pro- external attributions for difficulty (i.e., telling
hibited manipulation checks and measurement young chess students that mistakes are common
of the mediating mechanisms, the stereotype in new players and can be overcome with experi-
threat explanation is made stronger by examin- ence and practice). Whether some of these inter-
ing the relationship between performance and ventions would be more appropriate for a
future chess participation for girls and boys. younger population and for the domain of chess
That is, one unique aspect of the present requires further empirical verification. Whatever
research is its ability to explore possible future the method of intervention, the findings indicate
consequences of stereotype threat susceptibility. that for females to fully experience the cognitive
Obviously, in one-time experimental research it is and emotional benefits of chess, the earlier the
impossible to determine if those most vulnerable intervention, the better.
to stereotype threat continue participation in the
domain or if they eventually disengage and disi- Funding
dentify with it. In the tournaments we examined, This research received no specific grant from any fund-
those females who demonstrated the largest per- ing agency in the public, commercial, or not-for-profit
formance deficits were the least likely to play in sectors.
future tournaments and waited longer between Notes
tournaments if they did play. This fits with the
1. The analyses reported utilized maximum likeli-
notion that stereotype threat can produce disen- hood estimation. Therefore, all degrees of free-
gagement from the threatened domain (Steele, dom are estimated using the Satterthwaite method
1997). Our evidence suggests that this process of (Satterthwaite, 1946). Estimates of degrees of
disengagement can occur even with participants freedom using this method may result in degrees
as young as grades K–2. This argument is made of freedom with decimals.

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


Rothgerber and Wolsiefer 89

2. In addition to the interactions there were several McFarland, L. A., Lev-Arey, D. M., & Ziegert, J. C.
significant main effects as well. First, there was (2003). An examination of stereotype threat in a
a significant effect of opponent strength (b = motivational context. Human Performance, 16, 181–
.001, t(1540) = 2.54, p < .01), such that partici- 205. doi:10.1207/S15327043HUP1603_2
pants performed somewhat better when playing a McKown, C., & Weinstein, R. (2003). The develop-
stronger opponent. An effect of opponent grade ment and consequences of stereotype conscious-
was also demonstrated (b = −.04, t(1152) = −2.41, ness in middle childhood. Child Development, 74,
p < .05) indicating that participants’ performance 498–515. doi:10.1111/1467-8624.7402012
decreased as the relative grade level of their Muzzati, B., & Agnoli, F. (2007). Gender and math-
opponent increased. ematics: Attitudes and stereotype threat suscepti-
bility in Italian children. Developmental Psychology, 43,
747–759. doi:10.1037/0012-1649.43.3.747
References Neuville, E., & Croizet, J. (2007). Can salience of gen-
Ambady, N., Shih, M., Kim, A., & Pittinsky, T. (2001). der identity impair math performance among 7–8
Stereotype susceptibility in children: Effects of year old girls? The moderating role of task diffi-
identity activation on quantitative performance. culty. European Journal of Psychology of Education, 22,
Psychological Science, 12, 385–390. doi:10.1111/1467- 307–316. doi:10.1007/BF03173428
9280.00371 Nguyen, H.-H. D., O’Neal, A., & Ryan, A. M. (2003).
Cadinu, M., Maass, A., Rosabianca, A., & Kiesner, J. Relating test-taking attitudes and skills and stereo-
(2005). Why do women underperform under ste- type threat effects to the racial gap in cognitive abil-
reotype threat? Evidence for the role of nega- ity test performance. Human Performance, 16, 261–
tive thinking. Psychological Science, 16, 572–578. 293. doi:10.1207/S15327043HUP1603_5
doi:10.1111/j.0956-7976.2005.01577.x Ployhart, R. E., Ziegert, J. C., & McFarland, L. A. (2003).
Chalabaev, A., Sarrazin, P., Stone, J. & Curry, F. (2008). Understanding racial differences on cognitive abil-
Do achievement goals mediate stereotype threat? ity tests in selection contexts: An integration of
An investigation on females’ soccer performance. stereotype threat and applicant reactions research.
Journal of Exercise and Sports Psychology, 30, 143–158. Human Performance, 16, 231–259. doi:10.1207/
Retrieved from http://hal.archives-ouvertes.fr/ S15327043HUP1603_4
docs/00/38/98/57/PDF/Chalabaev_et_al_ Satterthwaite, F. E. (1946). An approximate distribu-
JSEP_2008.pdf tion of estimates of variance components. Biom-
Cullen, M. J., Hardison, C. M., & Sackett, P. R. (2004). etrics Bulletin, 2, 110–114. Retrieved from http://
Using SAT-grade and ability-job performance rela- www.jstor.org/discover/10.2307/3002019?uid=37
tionships to test predictions derived from stereo- 38664&uid=2&uid=4&sid=21102178857171
type threat theory. Journal of Applied Psychology, 89, Singer, J. D. (1998). Using SAS PROC MIXED to fit
220–230. doi:10.1037/0021-9010.89.2.220 multilevel models, hierarchical models and individ-
Cullen, M. J., Waters, S. D., & Sackett, P. R. (2006). Test- ual growth models. Journal of Educational and Behav-
ing stereotype threat theory predictions for math- ioral Statistics, 24, 323–355. Retrieved from http://
identified and non-math-identified students by gen- www.jstor.org/discover/10.2307/1165280?uid=37
der. Human Performance, 19, 421–440. doi:10.1207/ 38664&uid=2&uid=4&sid=21102178857171
s15327043hup1904_6 Steele, C. M. (1997). A threat in the air: How ste-
Glickman, M., & Doan, T. (2011). The USCF rating reotypes shape the intellectual identities and
system. Retrieved from http://www.glicko.net/rat- performance of women and African-Americans.
ings/rating.system.pdf American Psychologist, 52, 613–629. doi:10.1037/
Huguet, P., & Regner, I. (2007). Stereotype threat 0003-066X.52.6.613
among schoolgirls in quasi-ordinary classroom Steele, C. M, & Aronson, J. (1995). Stereotype threat
circumstances. Journal of Educational Psychology, 99, and the intellectual test performance of African
545–560. doi:10.1037/0022-0663.99.3545 Americans. Journal of Personality and Social Psychology,
Maass, A., D’Ettole, C., & Cadinu, M. (2008). Check- 69, 797–811. doi:10.1037/0022-3514.69.5.797
mate? The role of gender stereotypes in the ulti- Steele, C. M., Spencer, S. J., & Aronson, J. (2002).
mate intellectual sport. European Journal of Social Contending with images of one’s group: The psy-
Psychology, 38, 231–245. doi:10.1002/ejsp.440 chology of stereotype and social identity threat. In

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014


90 Group Processes & Intergroup Relations 17(1)

M. Zanna (Ed.), Advances in experimental social psy- chess players and its relation to chess skill. Learn-
chology (379–440). San Diego, CA: Academic Press. ing and Individual Differences, 20, 517–521. Retrieved
Stone, J., Lynch, C., Sjomeling, M., & Darley, J. (1999). from http://www.sciencedirect.com/science/arti-
Stereotype threat effects on Black and White athletic cle/pii/S1041608010000403
performance. Journal of Personality and Social Psychology, Wolsiefer, K., & Rothgerber, H. (2013). The influence
77, 1213–1227. doi:10.1037/0022-3514.77.6.1213 of stereotype awareness an endorsement on the development
Vollstadt-Klein, S., Grimm, O., Kirsch, P., & Bilalic, of stereotype threat in young female chess players. Unpub-
M. (2010). Personality of elite male and female lished manuscript.

Downloaded from gpi.sagepub.com at Ondokuz Mayis Universitesi on November 6, 2014

Anda mungkin juga menyukai