Content uploaded by Jennifer Petersen
Author content
All content in this area was uploaded by Jennifer Petersen on Oct 11, 2017
Content may be subject to copyright.
Gender differences in identification of gifted youth and in gifted program
participation: A meta-analysis
Jennifer Petersen
University of Wisconsin – Whitewater, 800 W. Main St, Whitewater, WI 53190, United States
article info
Article history:
Available online 23 July 2013
Keywords:
Gender
Gifted and talented
High ability
Meta-analysis
abstract
Although numerous studies have researched gender differences in gifted identification and program par-
ticipation, the results of these studies are largely mixed. The goal of the present study was to synthesize
data on gender differences in gifted identification and programming by combining data from multiple
studies into a single meta-analysis. The combined results from 130 studies published between 1975
and 2011 indicated that boys were 1.19 times more likely than girls to be identified as gifted and included
in gifted programs. Moderator analyses indicated that gender differences were particularly evident
among pre-adolescents, within gifted summer programs, and for students who were identified as gifted
using IQ scores and standardized tests. Recommendations for reducing gender bias include encouraging
pre-adolescent girls to participate in gifted programs and using multiple assessment criteria to identify
gifted students.
Ó2013 Elsevier Inc. All rights reserved.
1. Introduction
One of the many goals of education is to challenge students of
all levels to learn and to grow. Gifted education is important in or-
der to ensure that all students have an appropriately challenging
curriculum and teachers who are trained to meet their need
(NAGC, 2008). However, bias in identifying high ability students
and in programs designed to serve these students may results in
unequal opportunities. The purpose of the current study was to
combine the results on gender differences in identification of gifted
youth and in gifted programming across all available studies in or-
der to determine whether a gender bias exists.
The decades of research on gender differences in identification
of gifted youth reveal mixed results. Although some research sug-
gests that boys are more likely to be identified as gifted than girls
are (Fox, 1982; Lubinski, Benbow, Webb, & Bleske-Rechek, 2006;
Preckel, Goetz, Pekrun, & Keleine, 2008), other research suggests
the opposite (Read, 1991; Siegle & Reis, 1998), and still more re-
search provides evidence for no gender differences at all (Crombie,
Bouffard-Bouchard, & Schneider 1992). The inconclusive nature of
the existing research indicates a need for a meta-analysis.
1.1. Gender similarities hypothesis
The inconclusive nature of the existing research may be a result
of small gender differences in the gifted population. If the gender
difference in the population is small or non-existent then some
studies are likely to find small differences favoring girls whereas
others may find small differences favoring boys. A meta-analysis
combines the data from several studies to obtain a sample that
more closely reflects the population to present a more accurate
account of gender differences in gifted youth (Hedges & Nowell,
1995).
Hyde’s Gender Similarities Hypothesis proposes that gender dif-
ferences for most cognitive variables are small or non-existent
(Hyde, 2005). She reviewed 46 meta-analyses to discover that gen-
der differences were near zero for all measured cognitive variables
(Hyde, 2005) including math (Lindberg, Hyde, Petersen, & Linn,
2010), science (Hedges & Nowell, 1995), and vocabulary and read-
ing comprehension (Hyde & Linn, 1988).
If there are few gender differences in cognitive variables as indi-
cated by the Gender Similarities Hypothesis, then there should be
little difference in boys and girls identified as gifted and participat-
ing in gifted programs. Therefore, the Gender Similarities Hypoth-
esis suggests that there would be no gender difference, or at least a
small gender difference, in the current meta-analysis of the gender
differences in gifted identification and program participation.
1.2. Greater variability hypothesis
Although meta-analyses reported by Hyde generally indicate no
gender differences in cognitive ability (e.g., Hyde, 2005), it is
worthwhile to investigate moderators and the greater variability
hypothesis. The greater variability hypothesis suggests that boys
have more variability in cognitive performance than girls do,
regardless of gender differences in mean levels of academic
0361-476X/$ - see front matter Ó2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.cedpsych.2013.07.002
E-mail address: Petersej@uww.edu
Contemporary Educational Psychology 38 (2013) 342–348
Contents lists available at SciVerse ScienceDirect
Contemporary Educational Psychology
journal homepage: www.elsevier.com/locate/cedpsych
performance (Feingold, 1994; Hedges & Friedman, 1993). Accord-
ing to this hypothesis boys would be more likely than girls to be
identified as gifted because more boys than girls would be likely
to be found at the upper end of the cognitive distribution. A large
national sample in the United Kingdom and the United States indi-
cated greater male variability for verbal, quantitative, and nonver-
bal scales of the cognitive abilities test (Lohman & Lakin, 2009;
Strand, Dreary, & Smith, 2006). The IQ distribution for ‘‘almost
everyone born in Scotland in 1921’’ revealed greater male variabil-
ity, with boys more likely to be identified in the top of the IQ dis-
tribution than girls were (Deary, Thorpe, Wilson, Starr, & Whalley,
2003).
Some research, however, suggests that boys’ greater variability
in cognitive performance may be exaggerated. For example, a
cross-cultural study indicated that gender differences in variability
were inconsistent from one country to another. Some countries re-
ported greater male variability, whereas others reported greater fe-
male variability (Feingold, 1994). A recent meta-analysis on gender
differences in mathematical performance indicated that the boys
and girls had almost the same variance even in a stereotypically
male domain such as mathematics (Lindberg et al., 2010).
1.3. Moderators of gender differences in giftedness
Gender differences in giftedness may depend on a variety of
moderating factors. Selection criteria, method of gifted identifica-
tion, year of publication, country of origin, socioeconomic status,
and age of the sample were included as moderators of gender dif-
ferences in gifted identification and program participation.
1.3.1. Self-selection
If there is a gender difference in gifted programming there are
two possible explanations. First, one gender may be less likely than
the other to be identified as gifted. Second, boys and girls may be
equally likely to be identified as gifted, but one gender might be
more likely to self-select to not take part in gifted programs. There
are a number of reasons why gifted students may choose not to
participate in gifted programs. They may be intimidated to join a
gifted program; they may have few role models of their gender
who are gifted; or they may prefer to spend their time in other ex-
tra-curricular programs such as sports (Kerr, Vuyk, & Rea, 2012;
Schober, Reimann, & Wagner, 2004).
In order to determine whether gender differences in giftedness
are due to identification or self-selection, each study included in
the current meta-analysis was coded for whether the gifted sample
was identified directly from assessment criteria (such as IQ scores)
or whether the gifted sample was taken from an existing gifted
program. This variable was included as a moderator variable of
the gender differences in general giftedness. Many gifted programs
exist within the schools such as pullout programs or specific clas-
ses marked for gifted studies. However, some gifted programs were
outside of the school such as summer talent search programs in
which students attended a university to obtain special gifted clas-
ses (e.g., Putallaz, Baldwin, & Selph, 2005).
1.3.2. Gifted identification
Methods of identification vary immensely, but may include IQ
scores, standardized test scores, grades, teacher nominations and
any combination thereof. Different methods of gifted identification
may be more likely to identify boys or girls. For example, in the ele-
mentary grades girls tend to receive higher grade than boys do
(Corbett, Hill, & St. Rose, 2008); therefore girls may be more likely
to be identified as gifted when grades are used as the primary
method of identification. Teachers who were given identical stu-
dent profiles for boys and girls were more likely to nominate
the boys for a gifted program than the girls (Bianco, Harrison,
Garrison-Wade, & Leech, 2011). Therefore boys may be more likely
to be identified as gifted when teacher nominations are the pri-
mary method of identification.
1.3.3. Year of publication
Following the passage of US Title IX in 1972 girls gained
increasing opportunities in education and the women’s movement
created a culture that values the equality of women (Goodman,
Rodgers, & Albisetti, 2010). In accordance with these changes, gen-
der differences in giftedness are likely to change across time. In
particular, gender differences are likely to be smaller in studies
published in recent years in comparison to studies published
decades ago.
1.3.4. Country of origin
Giftedness is defined by the values of a particular society
(Subotnik, Olszewski-Kubilius, & Worrell, 2011). Some cultures
may view exceptional talent differently than others. Therefore,
gender differences or similarities in gifted identification and pro-
gram participation may differ depending on the country in which
giftedness is defined.
1.3.5. Race/ethnicity
Research within the United States has found larger academic
gender differences within ethnic minorities than within Whites.
For example, African American women are much more likely than
African American men to graduate from high school and to attend
college (United States Census, 2012). Race/ethnicity was includes
as a moderating variable to account for gender differences across
differences racial groups.
1.3.6. Socio-economic status
Children from low socio-economic backgrounds are largely
underrepresented in gifted programs and among children identi-
fied as gifted even when ethnicity is controlled (e.g., Carman &
Taylor, 2010; McBee, 2006). Low SES girls may be at a particular
disadvantage. These girls are more likely to be held responsible
for domestic work and may have less freedom to take on nontradi-
tional gender roles (Darity, 1995). The academic gender gap among
low SES children is typically larger than for middle or upper class
students (Freeman, 2003). Therefore, socioeconomic status may
moderate gender differences in giftedness.
1.3.7. Age
The age at which students are identified as gifted and partici-
pate in gifted programs may moderate gender differences in gifted-
ness. Giftedness is a developmental construct that may first appear
as potential and later develop into achievement (Subotnik et al.,
2011). This conception of giftedness at different ages may create
larger gender differences at different developmental periods. For
example, girls are more likely than boys to be enrolled in gifted
programs in grade school (Read, 1991; Reis, 2002) and grade school
girls who have been identified as gifted are more likely to be rated
by their teachers as having a higher work quality, effort, and ability
than their gifted male peers (Siegle and Reis, 1998). However, as
students grow older the gender difference reverses. Between
10th and 12th grade boys are more likely to be enrolled in gifted
programs than girls are (Read, 1991), and the majority of students
who ‘‘used to be gifted’’ are girls (Silverman, 1991).
1.4. The current study
This meta-analysis synthesized data from multiple studies on
gender differences in gifted youth. The purposes of the current
meta-analysis were:
J. Petersen / Contemporary Educational Psychology 38 (2013) 342–348 343
1. To identify the magnitude of gender differences or similarities
in gifted identification and program participation.
2. To determine whether there is a self-selection bias in gifted pro-
gramming by comparing gender differences from samples that
were selected directly from assessment criteria to samples that
were selected from existing gifted summer programs or in-
school gifted programs.
3. To examine gender difference among gifted identification crite-
ria including teacher nomination, grades, standardized tests
scores, IQ scores, and multiple assessment criteria.
4. To determine whether gender differences in gifted youth are
changing across time.
5. To compare gender differences in gifted identification and pro-
gram participation among nations across the world and among
racial/ethnic groups in the US.
6. To determine whether gender differences in giftedness differ
across socio-economic status.
7. To examine gender differences in giftedness across different age
groups.
2. Method
2.1. Search criteria
A literature search using the terms ‘‘gifted or talented’’ was per-
formed in ERIC, PsycINFO, Academic Search Elite, and PsycArticles
for all dissertations and peer-reviewed articles that were written in
English, used human subjects, and were published before June
2011. Including ‘‘gender’’ in the search terms may lead to studies
biased toward findings of gender differences, therefore this term
was not included in the search. However, studies that included
the word ‘‘gender’’ were uncovered in the search as long as the
words ‘‘gifted’’ or ‘‘talented’’ were in the title or abstract as well.
The result of the initial search revealed 6787 articles.
2.2. Study inclusion and exclusion criteria
The abstracts from all of the articles were screened for relevant
content. Studies were included in the meta-analysis if they met the
following criteria:
2.2.1. Gifted and non-gifted students
Studies that used both a sample of gifted students and a com-
parison sample of non-gifted students were included. A compari-
son group was necessary to determine whether any gender
differences that existed among gifted students were typical among
the general population of similar students. Both the gifted sample
and the non-gifted sample had to be of similar age and background
in order to be included. When the gifted and non-gifted samples
were matched by gender the study was excluded, because this arti-
ficially creates no gender differences in giftedness. However, if pre-
screening data was available before students were matched by
gender, that data was used. Studies were included when the com-
parison sample was comprised of students of average intelligence
or from general-ability classrooms. Studies were excluded when
the non-gifted sample included students of atypical intelligence
or students with special cognitive needs.
2.2.2. Gender
Usable samples included both boys and girls in the gifted sam-
ple and the non-gifted sample. Single-sex samples were excluded.
2.2.3. Statistics
Studies had to provide original, empirical, quantitative data in
order to be included. Studies were excluded if they reported a re-
view of only prior research or if they included qualitative data only.
The number of boys and girls in the gifted sample and the num-
ber of boys and girls in the non-gifted sample had to be provided in
order to calculate effect sizes. Several articles reported data from a
mixed-gender gifted sample and a mixed-gender non-gifted sam-
ple, but did not give the gender breakdown for each group. Authors
of these studies were contacted for the usable data if the study was
published within the past six years. Seventeen authors were con-
tacted and nine authors replied with usable data.
2.3. Coding the studies
After all of the abstracts were screened, 399 abstracts had
usable information or did not have sufficient information to justify
exclusion. The full article for each of these abstracts was obtained
and coded by the author. Of the 399 articles that were coded, 111
articles published usable information and nine authors provided
usable information that was not published. The majority of ex-
cluded studies did not include a comparison group of non-gifted
individuals or reported data for only a single sex. Some articles in-
cluded multiple independent studies or data from multiple inde-
pendent groups. Therefore, it was possible for some articles to
include more than one effect size. The resulting 120 usable articles
yielded 130 independent studies that were included in the meta-
analysis.
Studies were coded for the number of boys and girls who were
identified as gifted/participated in a gifted program and the num-
ber of boys and girls who were not identified as gifted/ participated
in non-gifted classes. In addition to coding studies for gender dif-
ferences in identification of gifted youth and in gifted program par-
ticipation, each study was coded for additional information to be
used as moderators.
2.3.1. Gifted selection
All studies selected gifted students either directly from assess-
ment criteria or from existing gifted programs. The overall effect
size reflected both groups of students combined. However, studies
were coded for whether students in the gifted sample were identi-
fied directly from assessment criteria, were selected based on their
participation in gifted programs within the school, or were selected
because they participated in a summer gifted program. The most
common summer programs included in the meta-analysis were
the Duke Talented Identification Program (TIP) and the John’s Hop-
kins Center for Talented Youth. This moderator variable described
whether the gender differences in general giftedness differed
depending on whether students were identified directly as gifted
or participated in in-school gifted programs or summer gifted
programs.
2.3.2. Method of gifted identification
The various studies included in the meta-analysis used a variety
of methods to identifying gifted individuals. After reviewing the
studies the method of gifted identification was coded as teacher
recommendation, IQ scores, standardized tests/achievement tests,
grades or multiple criteria. One study did not report how students
were identified.
2.3.3. Year
Year of publication was coded for each usable article.
2.3.4. Nationality
The usable studies represented a number of counties around the
world. As an exploratory analysis, nationality of the sample was
coded to determine whether gender differences were moderated
by country. Although some countries had enough studies to have
their own category, other countries were grouped by region or
continent to develop a category that included multiple studies.
344 J. Petersen / Contemporary Educational Psychology 38 (2013) 342–348
Nationality was coded as United States, Canada, Western Europe,
Eastern Europe/Russia, Australia/New Zealand, Asia, Middle East
and South America.
2.3.5. Race/ethnicity
Since studies conducted in the United States were more com-
mon than studies in any other country, race and/or ethnicity was
coded according to race and ethnic categories that are prevalent
in the US. Race and ethnicity were not coded for studies conducted
outside of the US. Race/ethnicity was coded as >75% White, >75%
African American, >75% Asian American, >75% Hispanic/Latino,
mixed ethnicity or unreported.
2.3.6. Socio-economic status
Socio-economic status (SES) was coded as >75% lower class,
>75% working class, >75% middle class, >75% upper class, mixed
SES or unreported.
2.3.7. Age
Primary age of the sample was coded as children (6–10.9 years),
pre-adolescents (11–13.9 years), adolescents (14–17.9 years) or
young adults (18–22 years).
2.4. Calculation of effect sizes
Since gifted identification/program participation and non-gifted
identification/program participation are both binary outcomes, an
odds ratio (OR) was the appropriate effect size to use for this meta-
analysis (Borenstein, Hedges, Higgins, & Rothstein, 2009; Lipsey &
Wilson, 2001). Odds ratios were calculated by multiplying the
number of gifted boys by the number of non-gifted girls and divid-
ing this product by the product of gifted girls and non-gifted boys.
OR ¼
Boys
gifted
Girls
non-gifted
Girls
gifted
Boys
non-gifted
Odds ratios greater than one indicated that boys were more
likely to be identified as gifted or participate in gifted programs
than girls were, whereas odds ratios less than one indicated that
girls were more likely to be identified as gifted or participate in
gifted programs than boys were. An odds ratio of one indicated
that there was no gender difference in gifted identification or pro-
gram participation. Odds ratios must be transformed with a natural
log in order to center them at zero and conduct further analyses
(Borenstein et al., 2009). Once analyses were completed, the natu-
ral log inverse transformed odds ratios back into their original
form.
2.5. Statistical analysis
Odds ratios, standard errors and inverse variance weights were
calculated for each usable study. The distribution of odds ratios
was examined for outliers. According to Tukey (1977), outliers in-
clude observations that fall three times the interquartile range
above the 75th percentile or below the 25th percentile in a distri-
bution of scores. One study was excluded from analyses because
the effect size it yielded fit this criterion.
The mixed-effects model was employed to determine an aver-
age odds ratio across all studies. The mixed effects model ac-
counted for random errors among studies and allowed for the
remaining variance to be accounted for by study characteristics.
This model provided a measure of homogeneity among studies. If
the studies were heterogeneous, moderator analyses were em-
ployed to account for additional variance among the studies. Lipsey
and Wilson (2001) macros were used to calculate the mixed effects
model and all moderator analyses.
3. Results
3.1. Summary of effect sizes
The mean effect size for all 130 studies was OR = 1.19, d= 0.09.
This indicates that boys were 1.19 times more likely than girls to
be identified as gifted or included in gifted programs. In other
words, the ratio of youth included in gifted programs or identified
as gifted was approximately six boys for every five girls. Although
this odds ratio was significantly different from zero, 95% CI [1.14,
1.25], Cohen (1988) suggested that odds ratios of 1.5 or lower
should be considered a small effect.
A homogeneity analysis determined that there was significant
variability across effect sizes in the current study,
Q(129) = 461.17, p< 0.0001. The random effects variance compo-
nent was 0.22. When significant heterogeneity was found, the
mixed effects model accounted for variability due to moderator
variables and random error among the studies (Lipsey & Wilson,
2001).
3.2. Moderator analyses
3.2.1. Year of publication
A weighted ordinary least squares regression examined the var-
iance in effect sizes due to year of publication. Usable studies were
published between 1975 and 2011. Results indicated that year of
publication did not significantly predict odds ratio for the gender
difference in gifted youth, b= 0.04, t(129) = 0.86, p= 0.39.
3.2.2. Gifted selection
An analog to ANOVA examined variability in effect sizes due to
sample selection criteria, gifted identification, nationality, ethnic-
ity of the sample, socio-economic status, and age of the sample. Ta-
ble 1 presents the results from these analyses. Method of gifted
selection significantly moderated the variability among effect
sizes, F(2,129) = 11.52, p= 0.003. There was no gender difference
for students who were selected directly from identification criteria
or from in-school programs, but there was a gender difference
among students selected from a gifted summer program. Boys
were 1.81 times more likely than girls to attend a summer program
for the gifted.
3.2.3. Gifted identification
Method of gifted identification accounted for variance among
the studies, F(4,128) = 15.68, p= 0.004. Boys were more likely than
girls to be identified as gifted when IQ scores or standardized/
achievement test scores were used for assessment. There was no
gender difference among youth who were nominated as gifted by
teachers or when multiple assessment criteria were used. Although
it appears as though girls were more likely than boys to be identi-
fied as gifted when grades were used as the assessment criteria,
this result should be interpreted with caution because it was based
on only one study.
3.2.4. Nationality, race/ethnicity and SES
Nationality, F(7,129) = 8.14, p= 0.32, ethnicity, F(4, 75) = 0.43,
p= 0.98, and socio-economic status, F(5, 129) = 3.26, p= 0.66, did
not account for significant variance among the studies.
3.2.5. Age
Age of the sample did account for significant variation among
the studies, F(3,129) = 11.71, p= 0.01. Although pre-adolescent
boys (age 11–13.9) were slightly more likely to be identified as
gifted or participate in gifted programs than pre-adolescent girls
J. Petersen / Contemporary Educational Psychology 38 (2013) 342–348 345
were, there was no gender difference for gifted children (ages 6–
10.9), adolescents (age 14–17.9), or young adults (18–22 years).
4. Discussion
The Gender Similarities Hypothesis suggested there would be
small gender difference in gifted identification and programming
because there are small gender differences in cognitive variables
(Hyde, 2005). In contrast, the greater variability hypothesis sug-
gested that boys were more likely to be identified as gifted than
girls because there is more variability in boys’ cognitive perfor-
mance (e.g., Shields, 1982). With an average odds ratio of 1.19, or
d= 0.09, the current meta-analysis supported the Gender Similari-
ties Hypothesis. This effect size was small according to Cohen’s
(1988) criteria. For every six boys who were identified as gifted
or included in gifted programs, there were approximately five girls.
This suggests that boys and girls are equally likely to be high-abil-
ity students. In accordance with the Gender Similarities Hypothe-
sis, this study provides evidence that boys and girls are more
alike in terms of cognitive variables than they are different, and
even the highest end of the cognitive distribution show little gen-
der differences.
Although there was no significant difference in identification of
gifted youth, there was a gender difference in representation in
gifted programs, particularly summer gifted programs. It was
promising to find that boys and girls are equally likely to be iden-
tified as gifted, yet concerning that gifted girls are selecting not to
participate in some of the programs. Talent search programs, such
as the Duke Talented Identification Program (TIP) or the John’s
Hopkin’s Center for Talented Youth Program, can offer leadership
opportunities and educational services such as educational and
career counseling (Lee, Matthews, & Olszewski-Kubilius, 2008).
Participation in talent search programs may even reverse under-
achieving behaviors of gifted youth in schools (Matthews & McBee,
2007). Girls who are accepted to these programs, but self-select not
to participate in them, may be doing themselves a disservice that
may have long-term outcomes for their academic and career
success.
There are a variety of explanations for a somewhat greater rep-
resentation of boys than girls in gifted summer programs. Talent
search programs may begin with children as young as 5th grade
and they often require children to leave their home and attend a
university for a week or more (Lee et al., 2008). These university
programs may be hundreds of miles away from the child’s home.
Parents may be more protective of their daughters than their sons
and may be hesitant to allow their daughters to leave home to at-
tend a summer program. In addition, girls may be more reluctant
than boys to leave their friends and family either because they fear
Table 1
Frequencies of studies and odds ratios within each moderator category.
kFrequency (%) OR d
Selection
F(2,129) = 11.52
*
Identified directly 47 37 1.02 0.01
School program 68 52 1.13 0.06
Summer program 15 12 1.81
*
0.33
*
Identified
F(4,128) = 15.68
*
Teacher 9 7 1.07 0.04
IQ 51 40 1.24
*
0.12
*
Grades 1 1 0.26
*
0.74
*
Achievement 31 24 1.38
*
0.18
*
Multiple criteria 37 28 0.96 0.02
Nationality
F(7,129) = 8.14, n.s. Asia 10 8 1.49
*
0.22
*
Australia 5 4 1.33 0.16
Canada 9 7 1.00 0.00
Eastern Europe/Russia 11 9 1.60
*
0.26
*
Western Europe 12 9 0.94 0.03
Middle East 6 5 1.36 0.17
South America 1 1 0.90 0.06
United States 76 59 1.11 0.06
Race/ethnicity
F(4,75) = 0.43, n.s. Mixed 13 17 1.07 0.04
White 21 28 1.19 0.10
African American 1 1 1.23 0.09
Asian American 0 0 –
Latino/Hispanic 2 3 1.17 0.09
Unreported 39 51 1.07 0.04
Socio-economic status
F(5,129) = 3.26, n.s. Mixed 8 6 1.49
*
0.22
*
Lower 1 1 1.04 0.02
Working 3 2 0.87 0.08
Middle 13 10 0.98 0.01
Upper 12 9 1.27 0.13
Unreported 93 72 1.16
*
0.08
*
Age
F(3,129) = 11.71
*
Children 40 31 0.99 0.01
Pre-adolescents 51 39 1.44
*
0.20
*
Adolescents 32 25 1.07 0.04
Young adults 7 5 0.74 0.17
n.s. Not significant at
a
= 0.05.
k= Number of effect sizes in each category.
OR = Odds ratio.
*
p< 0.05.
346 J. Petersen / Contemporary Educational Psychology 38 (2013) 342–348
separation from these close relationships or because they are
anxious about the intensity of talent search programs. Alterna-
tively, parents may send their underachieving gifted children
to these programs in order to encourage them to reach their po-
tential. Since gifted boys are more likely to underachieve than
gifted girls are (e.g., Peterson & Colangelo, 1996; Whitmore,
1980), parents may be more likely to send their sons than their
daughters.
Although there was a gender difference in participation in
gifted programs, there was no gender difference in identification
of gifted youth. However, the method of gifted identification did
appear to make a difference. Boys were more likely than girls to
be identified as gifted when IQ scores were used. This may pro-
vide evidence for the greater male variability hypothesis, which
suggests that boys are more likely to be at the upper end of
the IQ distribution (Feingold, 1994; Hedges & Friedman, 1993;
Shields, 1982). Boys were also more likely to be identified than
girls when standardized/achievement tests were used. Talent
search programs often use college entrance exams to identify eli-
gible students (Lee et al., 2008). Gender differences in standard-
ized test scores could be a product of studies that selected the
gifted sample from summer programs, which were shown to
have a substantial gender difference. Alternatively, gender biases
in standardized testing could lead to more boys and fewer girls
being accepted to summer talent search programs. There was no
gender difference in gifted youth identified with teacher nomina-
tions. Although secondary teachers may be more likely to favor
boys for gifted programs (Bianco et al., 2011), elementary teach-
ers may be more likely to favor girls (Read, 1991; Reis, 2002).
These differences may counteract, resulting in no gender differ-
ence when all studies using teacher nominations are combined.
There was no gender difference in gifted youth when multiple
assessment criteria were used. The use of multiple assessment
criteria may help to reduce bias associated with standardized
tests, IQ tests, and teacher nomination.
Results indicated that effect sizes did not change significantly
over time. Although it was expected that increasing gains in educa-
tional opportunities for women would decrease the gender gap,
other studies have found no gender difference in academic ability
since the 1960s (Nowell & Hedges, 1998).
Results indicated that neither race/ethnicity nor SES moderated
the gender difference in gifted youth. As Table 1 indicated there
were very few studies that focused on ethnic and racial minorities
or on children from low or working class environments. In order to
obtain a more accurate representation of gender differences in
gifted youth among ethno-racial minorities and low SES youth,
more studies focusing on these underrepresented groups must be
available. Similarly, there was no significant difference in gender
differences across nations. This may be because the majority of
the studies represented were from Western nations.
Although there was no gender difference in gifted identifica-
tion and programming for children and adolescents, there was
a small difference favoring boys among pre-adolescent youth.
The effect of age that was evident in this analysis could be a
product of the larger gender differences in gifted summer pro-
grams in comparison to in-school gifted programs available in
middle school. Many of the talent search programs such as the
Duke Talent Identification Program (TIP) and the Johns Hopkins
Center for Talented Youth (CTY) include students in grades 7
and 8 (Lee et al., 2008). Therefore, gender differences in age
may be masked by a greater representation of boys in these tal-
ent search programs during the pre-adolescent years. Alterna-
tively, boys may be more likely to be identified as gifted in
the middle grades because they are more often referred for
assessment for other academic concerns such as ADHD (Gaub
& Carlson, 1996).
4.1. Strengths and limitations
By combining data from multiple studies the current meta-
analysis gives a more accurate account of gender differences in
gifted identification and programming than any individual study
could (Borenstein et al., 2009). This research synthesizes the data
to determine what types of gifted programs and identification
are most frequent in the literature and the magnitude of gender
differences in gifted identification and programming.
As is often the case with meta-analyses, the detailed review of
the literature in any field often reveals areas for improvement.
The review of the research in gifted education for the current
meta-analysis revealed a severe concern for the lack of racial, eth-
nic, and socio-economic information provided in the gifted litera-
ture. As seen in Table 1, the vast majority of studies did not
provide information about the demographics of their sample. The
American Psychological Association’s publication manual requests
that all research studies describe the gender, racial, ethnic, and
socio-economic distribution among their participants (American
Psychological Association, 2009). This information is essential for
readers to interpret the findings of each study.
Among the studies that did report ethnicity or racial distribu-
tions a shortage of studies including ethnic and racial minorities
was clearly evident. Gifted education has frequently been criticized
for ethno-racial biases, however little has been done to alleviate
the problem (Ford, 1998). A variety of explanations have been of-
fered for the underrepresentation of ethnic minorities in gifted
education including problems with recruitment and identification,
personal issues such as teacher expectations, and problems with
retention (Ford, 1998). Just as more efforts must be made to recruit
and retain girls and women in gifted programs, more focus should
also be placed on increasing the representation of ethnic
minorities.
Defining giftedness is a difficult challenge because the defini-
tion of high-ability varies across contexts and according to the val-
ues of each culture. Therefore, there does not seem to be one
universally agreed upon definition for giftedness. Although the
methods of identifying students as gifted may partially align with
each study’s definition of giftedness, studies rarely provided a clear
explanation of what they believed to be ‘‘giftedness.’’ In addition,
many high-ability students may never even be assessed for gifted-
ness. There may be a gender bias, or ethno-racial biases, which pre-
vent students from even being assessed as gifted. Results of this
meta-analysis should be interpreted with caution and with the
understanding that the construct of giftedness is socially con-
structed and varies across the studies (Pfeiffer, 2012; Subotnik
et al., 2011).
5. Conclusions
The purpose of this meta-analysis was to provide the field of
gifted education with additional information synthesized from
several studies so that all gifted youth may be appropriately chal-
lenged. The current study indicated that boys are girls are equally
likely to be identified as gifted. However, IQ scores, achievement
test scores, and participation in summer gifted programs may favor
boys.
The current study suggested that there is very little gender bias
in gifted identification, but that gender biases in gifted program-
ming may benefit from the use of multiple assessment criteria
and emphasizing pre-adolescent girls’ participation in gifted sum-
mer programs. High ability students who are not identified as
gifted or who do not participate in gifted programs may not receive
an appropriately challenging curriculum or have teachers who are
trained to meet their educational needs (NAGC, 2008). Gifted
J. Petersen / Contemporary Educational Psychology 38 (2013) 342–348 347
programs must make increased efforts to recruit and retain girls
in their programs in order to provide equal opportunities for all
students to succeed.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.cedpsych.2013.
07.002.
References
American Psychological Association (2009). Publication manual for the American
Psychological Association (6th ed.). Washington, D.C.: American Psychological
Association (6th ed.).
Bianco, M., Harrison, B., Garrison-Wade, D., & Leech, N. (2011). Gifted girls: Gender
bias in gifted referrals. Roeper Review, 33, 170–181.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction
to meta-analysis. West Sussex, United Kingdom: Wiley.
Carman, C., & Taylor, D. K. (2010). Socio-economic status effects on using the
Naglieri Nonverbal Ability Test to identify the gifted/talented. Gifted Child
Quarterly, 54, 75–84.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).
Hilsdale, NJ: Erlbaum.
Corbett, C., Hill, C., & St.Rose, A. (2008). Where the girls are: The facts about gender
equity in education. Washington, D.C.: American Association of University
Women Educational Foundation.
Crombie, G., Bouffard-Bouchard, T., & Schneider, B. H. (1992). Gifted programs:
Gender differences in referral and enrollment. Gifted Child Quarterly, 36,
213–214.
Darity, W. (1995). The formal structure of a gender-segregated low-income
economy. World Development, 23, 1963–1968.
Deary, I. J., Thorpe, G., Wilson, V., Starr, J. M., & Whalley, L. J. (2003). Population sex
differences in IQ at age 11: The Scottish mental survey 1932. Intelligence, 31,
533–542.
Feingold, A. (1994). Gender differences in variability in intellectual abilities: A
cross-cultural perspective. Sex Roles, 30, 81–92.
Freeman, J. (2003). Gender differences in gifted achievement in Britain and the US.
Gifted Child Quarterly, 47, 202–211. http://dx.doi.org/10.1177/
001698620304700304.
Ford, D. (1998). The underrepresentation of minority students in gifted education:
Problems and promises in recruitment and retention. The Journal of Special
Education, 32, 4–14.
Fox, L. H. (1982). The study of social processes that inhibit or enhance the development
of competence and interest in mathematics among highly able young women. Final
Report to the National Institute of Education: Washington, D.C.
Gaub, M., & Carlson, C. (1996). Gender differences in ADHD: A meta-analysis and
critical review. Journal of the American Academy of Child and Adolescent
Psychiatry, 36, 1036–1045.
Goodman, J., Rodger, R., & Albisetti, J. C. (2010). Girl’s secondary education in the
Western world: From the 18th to the 20th century. New York: Palgrave Macmillan.
Hedges, L. V., & Nowell, A. (1995). Sex differences in mental test scores, variability,
and numbers of high-scoring individuals. Science, 269, 95–105.
Hedges, L. V., & Friedman, L. (1993). Sex differences in variability in intellectual
abilities: A reanalysis of Feingold’s results. Review of Educational Research, 63,
95–105.
Hyde, J. S. (2005). The gender similarities hypothesis. American Psychologist, 60,
581–592.
Hyde, J. S., & Linn, M. C. (1988). Gender differences in verbal ability: A meta-
analysis. Psychological Bulletin, 104, 53–69.
Kerr, B. A., Vuyk, M. A., & Rea, C. (2012). Gendered practices in the education of
gifted girls and boys. Psychology in the Schools, 49, 647–655.
Lee, S., Matthews, M. S., & Olszewski-Kubiliius, P. (2008). A national picture of talent
search and talent search educational programs. Gifted Child Quarterly, 52, 55–69.
Lindberg, S. M., Hyde, J. S., Petersen, J. L., & Linn, M. C. (2010). New trends in gender
and mathematics performance: A meta-analysis. Psychological Bulletin, 136,
1123–1135.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. London, England: Sage
Publications.
Lohman, D. F., & Lakin, J. M. (2009). Consistencies in sex differences on the Cognitive
Abilities Test across countries, grades, test forms and cohorts. British Journal of
Educational Psychology, 79, 389–407.
Lubinski, D., Benbow, C. P., Webb, R. M., & Bleske-Rechek, A. (2006). Tracking
exceptional human capitol over two decades. Psychological Science, 17, 194–199.
Matthews, M. S., & McBee, M. T. (2007). School factors and underachievement of
gifted students in a talent search program. Gifted Child Quarterly, 51, 167–181.
McBee, M. (2006). A descriptive analysis of referral sources for gifted identification
screening by race and socio-economic status. Journal of Secondary Gifted
Education, 17, 103–111.
National Association for Gifted Children (NAGC), (2008). Advanced students in
today’s classrooms: What do we know? In National Association for Gifted
Children.<http://www.nagc.org/index.aspx?id=533> Retrieved 20.05.13.
Nowell, A., & Hedges, L. V. (1998). Trends in gender differences in academic
achievement from 1960 to 1994: An analysis of differences in mean, variance,
and extreme scores. Sex Roles, 39, 21–43.
Peterson, J. S., & Colangelo, N. (1996). Gifted achievers and underachievers: A
comparison of patterns found in school files. Journal of Counseling and
Development, 74, 399–407.
Pfeiffer, S. I. (2012). Current perspectives on identification and assessment of gifted
students. Journal of Psychoeducational assessment, 1, 3–9.
Preckel, F., Goetz, T., Pekrun, R., & Keleine, M. (2008). Gender differences in gifted
and average-ability students: Comparing girls’ and boys’ achievement, self-
concept, interest, and motivation in mathematics. Gifted Child Quarterly, 52,
146–159. http://dx.doi.org/10.1177/0016986208315834.
Putallaz, M., Baldwin, J., & Selph, H. (2005). The Duke university talent identification
program. High Ability Studies, 16, 41–54.
Read, C. R. (1991). Gender distribution in programs for the gifted. Roeper Review, 13,
188–193.
Reis, S. M. (2002). Gifted females in elementary and secondary school. In M. Neihart,
S. M. Reis, N. M. Robinson, & S. M. Moon (Eds.), The social and emotional
development of gifted children: What do we know? (pp 13–18). Waco, TX:
Prufrock Press.
Schober, B., Reimann, R., & Wagner, P. (2004). Is research on gender-specific
underachievement in gifted girls and obsolete topic? New findings on an often
discussed issue. High Ability Studies, 15, 43–62.
Shields, S. A. (1982). The great variability hypothesis: The history of a biological
model of sex differences in intelligence. Journal of Women in Culture and Society,
7, 769–782.
Siegle, D., & Reis, S. M. (1998). Gender differences in teacher and student
perceptions of gifted students’ ability and effort. Gifted Child Quarterly, 42,
39–47.
Silverman, L. K. (1991). To be gifted or feminine: The forced choice of adolescence.
Journal of Secondary Gifted Education, 6, 141–156.
Strand, S., Dreary, I. J., & Smith, P. (2006). Sex difference in cognitive ability test
scores: A UK national picture. Journal of Educational Psychology, 76. 463-380.
Subotnik, R. F., Olszwski-Kubilius, P., & Worrell, F. C. (2011). Rethinking giftedness
and gifted education: A proposed direction forward based on psychological
science. Psychological Science in the Public Interest, 12, 3–54.
Tukey, J. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley.
United States Census (2012). Educational attainment by race, Hispanic origin, and
sex. In Census Bureau Homepage.<http://www.census.gov/compendia/statab/
2012/tables/12s0230.pdf> Retrieved 06.11.12.
Whitmore, J. R. (1980). Underachievement – empirical studies. Journal of
Adolescence, 3, 143–151.
348 J. Petersen / Contemporary Educational Psychology 38 (2013) 342–348
