Content uploaded by Christopher Rozek
Author content
All content in this area was uploaded by Christopher Rozek on Oct 19, 2017
Content may be subject to copyright.
ORIGINAL ARTICLE
Mathematics—aCriticalFilterforSTEM-RelatedCareer
Choices? A Longitudinal Examination among Australian
and U.S. Adolescents
Helen M. G. Watt
1
&Janet S. Hyde
2
&Jennifer Petersen
3
&Zoe A. Morris
1
&
Christopher S. Rozek
4
&Judith M. Harackiewicz
2
Published online: 5 December 2016
Abstract Although women have made progress in entering
scientific careers in biology, they remain underrepresented in
mathematically intensive fields such as physics. We investi-
gated whether gender differences in mathematics motivation
and socialisers’perceptions impacted choices for diverse
STEM careers of varying mathematical intensity. Drawing
on expectancy-value theory, we tested structural equation
models in which adolescents’preferred careers related to each
of physics, biology, chemistry, and mathematics were predict-
ed by prior mathematical performance, motivations, and
mothers’perceptions. We explored potential differences in
gendered processes of influence using multigroup models.
Samples were 331 Australian adolescents followed from 9th
to 11th grade in 1998 and 277 U.S. adolescents from 9th to
12th grade in 2009–10. In both samples female adolescents
preferred biological careers more than males did; male ado-
lescents preferred physics-related careers and also mathemat-
ical careers in the Australian sample. Mothers’perceptions
were important to female and male adolescents’mathematics
motivations; gendered motivations were more evident in the
Australian sample. Mathematics interest played the strongest
role in male adolescents’preferred careers, whereas actual or
perceived mathematical achievements were most important
for females, demonstrating the impacts of mathematical moti-
vations on preferences for diverse STEM careers.
Keywords Gender .STEM .Mathematics .Critical filter .
Career choice .Expectancy-value theory .High school
The winners of Google’s first-ever science fair were announced
in 2011. All three were young women (Miller 2011). In the last
three years (2014–2016) all grand prize winners have been
young women. The 2016 winning project involved a way to
reduce drought with fruit, the 2015 project was aimed at fight-
ing ebola, and the 2014 winners were three young women who
collaborated on a project to address a bacteria that could fight
world hunger (www.googlesciencefair.com). This snapshot
captures much about current trends concerning gender and
STEM (science, technology, engineering, and mathematics).
Girls and young women are talented at, and interested in,
science. They are most represented in biology and, as these
examples illustrate, often they have a goal of helping people
(Diekman et al. 2011; Eccles and Vida 2003).
It is imperative to disaggregate discussions of different
fields of sciences rather than use an aggregated concept of
STEM. The entry of women and men into STEM careers
presents a mixed picture of gender similarities and gender
differences (Hyde 2005). In 2014 in the United States, 53%
of the doctoral degrees in biology went to women, 38% in
chemistry, 20% in physics, and 22% in engineering
(National Science Foundation 2016). In mathematics, women
earned about 30% of doctoral degrees, and about 41% at the
Bachelor’s level (Hill et al. 2010). In Bachelor-level sciences
*Helen M. G. Watt
helen.watt@monash.edu
*Janet S. Hyde
jshyde@wisc.edu
1
Faculty of Education, Monash University, Clayton Campus,
Wellington Road, Melbourne, VIC 3800, Australia
2
Department of Psychology, University of Wisconsin—Madison,
1202 W. Johnson St, Madison, WI 53726, USA
3
Department of Educational Foundations, University of
Wisconsin—Whitewater, 800 W. Main St, Whitewater, WI 53190,
USA
4
Department of Psychology, University of Chicago, 5848 S
University Ave, Chicago, IL 60637, USA
Sex Roles (2017) 77:254–271
DOI 10.1007/s11199-016-0711-1
#Springer Science+Business Media New York 2016
women earned 19% of engineering degrees, 38% physical
sciences, 58% biological and agricultural sciences, and 62%
of social sciences and psychology (National Student
Clearinghouse Research Center 2013). In Australia, outside
the biological sciences, women’s representation is even low-
er—with women earning 50% of the doctoral degrees in biol-
ogy but 11% in chemical science and 5% in physical sciences
(Dobson 2012). At the Bachelor level, 53% of degrees went to
women in the natural and physical sciences, 13% in informa-
tion technology, and 15% in engineering and related technol-
ogies (Australian Government Department of Education and
Training 2014).
The dearth of women in advanced mathematics and
associated fields (such as physics and engineering) has
been attributed to numerous contributing factors includ-
ing discrimination, societal gender inequality, women’s
purported lack of mathematical ability and spatial skills,
cultural stereotyping of mathematics as masculine, gen-
der differences in interests, and high-school academic
preparation (Ceci and Williams 2011; Cheryan et al.
2015; Else-Quest et al. 2010; Hyde et al. 2008;
Riegle-Crumb and King 2010; Sorby et al. 2013;Su
et al. 2009; Wai et al. 2009). Mathematics was first
identified as a critical filter by Lucy Sells in 1980
(see also Shapka et al. 2006). Researchers concerned
with gender equity have focused on mathematics partic-
ipation precisely because of its identified role in limit-
ing access to high-status careers. Crucial motivational
factors emphasised as impacting mathematical participa-
tion in expectancy-value theory (Eccles et al. 1983;
Eccles 2005,2009)arestudents’interest, perceived im-
portance of mathematics, self-concept of own ability,
actual performance, and key socialisers’perceptions. A
wealth of literature has collectively established their role
in predicting mathematics-related enrolments and career
choices.
Our study asks whether mathematical expectancies and
values could predict preferences for STEM-related careers
beyond mathematics, that is, in fields of physics, chemistry,
and biology. To test this idea, our study takes advantage of two
longitudinal datasets—one from Australia and another from
the United States—to examine mathematics-related influences
on mathematical and scientific dimensions of students’subse-
quent career aspirations near the end of high school. We con-
structed and tested models capitalising on two study designs.
In these models, earlier mathematics self-concepts and values,
as well as prior mathematics performance and mothers’per-
ceptions of the adolescents’mathematics ability, predicted ca-
reer preferences in 11th (Australian study) or 12th grade (U.S.
study). Although collected a decade apart (1998 and 2009–10,
respectively), similar design and measures allow us to test the
role of mathematics-related expectancy-value factors on di-
mensions of STEM aspirations among two separate
longitudinal datasets from culturally similar settings. Testing
the model with two samples from two cultures and somewhat
different times allows us to assess the robustness of the model
and whether the findings replicate across variations in setting
and design.
High school is the crucial time when most students make
choices whether to concentrate on STEM in the future (Maltese
and Tai 2011). Course selections can foreclose future educational
and career pathways (Watt 2006). In a U.S. decade-long study
(Farmer et al. 1999), students’subsequent participation in science
careers in 1990 was significantly predicted by high-school vari-
ables including academic performance and valuing mathematics
and science from 1980, although predicting considerably more
variance for men than for women. Occupational preference in
senior high school plays a key role in the choice of students’
college courses and adult occupational interests. As such,
Riegle-Crumb and King (2010) have argued persuasively for
the importance of the high-school years in determining disparities
in STEM fields. Using data from the U.S. Educational
Longitudinal Study (ELS) 2002, they demonstrated that much
of the discrepancy in both gender and ethnic representation in
college majors in the physical sciences and engineering was
accounted for by differences in high-school course preparation.
According to the Eccles and colleagues’expectancy-value
theory (EVT), a person chooses to take on a task—such as phys-
ics in high school or a mathematics major in college—if the
person values the task and expects that s/he can succeed at it
(Eccles 2005;Ecclesetal.1983). Different aspects of task value
include importance and intrinsic values. Importance value refers
to how useful or important the person perceives the task to be;
intrinsic value resembles interest. Expectancy-value theory also
outlines how values and ability expectancy beliefs (also referred
to as mathematics self-concept) are shaped by prior performance
in the domain and by socialisation experiences, such as mother’s
perceptions.
An abundance of research has demonstrated the crucial role
of self-concept and values in adolescents’decisions to pursue
challenging mathematics courses above and beyond measures
of mathematical ability (Crombie et al. 2005; Meece et al.
1990; Updegraff et al. 1996;Wattetal.2006;Wigfield
1993,1994). For advanced mathematics enrolments in high
school, Watt et al. (2006) found that intrinsic value was a
predictor for Australian students, in contrast to self-concept
in the U.S. sample, as well as importance value for female
adolescents. Those authors interpreted these cultural differ-
ences in terms of the greater choice available to Australian
students in upper-secondary school, which allowed them to
follow their interests, whereas the highly test-driven culture in
the United States may focus students’attention more on their
ability.
Closely related research emphasises the importance of inter-
est in promoting academic performance, course choices, and
career paths (Hidi 1990; Hidi and Harackiewicz 2000;
Sex Roles (2017) 77:254–271 255
Harackiewicz et al. 2016b). For example, interest in introduc-
tory college courses predicts subsequent course choices, even
controlling for prior measures of ability and performance
(Harackiewicz et al. 2000,2002,2008). Moreover, utility value
and interest are linked; perceiving utility value in college course
material predicted interest in the material assessed at the end of
semester (Hulleman et al. 2008). Experimental manipulation of
utility value increased interest in high-school and college
STEM courses (Hulleman and Harackiewicz 2009;
Harackiewicz et al. 2016a;Hullemanetal.2010) and STEM
course-taking in high school (Harackiewicz et al. 2012).
Gender Differences
Gender differences in mathematics-related self-concept and
values are well-established. A cross-national meta-analysis
of 2003 data from the Programme for International Student
Assessment (PISA) and Trends in International Mathematics
and Science Study (TIMSS) studies found gender differences
favouring boys and male adolescents for mathematics self-
concept and self-efficacy (Else-Quest et al. 2010). Based on
TIMSS data for mathematics self-concept, effect sizes for gen-
der were d= .19 for Australia and .26 for the U.S. For math-
ematics self-efficacy assessed by PISA, effect sizes were
d= .37 for Australia and .19 for the U.S. A major meta-
analysis of research on vocational interests found a substantial
gender difference in interest in engineering among North
American adolescents and young adults (d= 1.11) (Su et al.
2009). More modest gender differences were found for inter-
est in science (d= .36) and mathematics (d= .34) averaged
across all ages tested. Stereotype threat theory and research
may be relevant here (Steele 1997). Mathematics and physics
continue to be culturally stereotyped as masculine domains
(Steffens et al. 2010) in which boys and male adolescents
may receive a stereotype boost (Shih et al. 2012; Walton and
Cohen 2003); simply because they are male, they have higher
expectations. At the same time, girls and female adolescents
experience stereotype threat and may doubt whether they can
succeed in these male-stereotyped fields. It is heartening that
moderator analyses have indicated smaller gender differences
in STEM interests among more recent cohorts (Su et al. 2009).
Gender differences in STEM perceptions are unlikely to be
due to mathematical achievement differences. A meta-
analysis of 242 studies of mathematics performance, pub-
lished between 1990 and 2007 across numerous nations,
showed no gender differences (Lindberg et al. 2010). In the
United States, a meta-analysis of state assessments from the
National Assessment of Educational Progress (NAEP)
showed no gender differences in mathematics performance
through the end of high school (Hyde et al. 2008). Using
cross-national data from 14 to 16 year-olds tested by PISA
and TIMSS, another meta-analysis found very small gender
differences overall, but substantial variation across nations in
the magnitude and direction of the gender differences (Else-
Quest et al. 2010). For 14–16 year-olds’mathematical perfor-
mance, gender effect sizes were: Australian d=.15inTIMSS
and .06 in PISA; U.S. d= .06 in TIMSS and .07 for PISA.
Motivations for Mathematical Career Choice
Researchers have increasingly begun to examine whether there
are gender differences in the processes that link expectancies
and values to career preferences. For example, is interest in
mathematics equally important in predicting preferences for
mathematics-related careers for male compared with female ad-
olescents? A handful of previous studies has examined these
possibilities and found evidence of gender differences in pro-
cesses (Lazarides and Ittel 2013;Leaperetal.2012; Robnett and
Leaper 2013;Wattetal.2012). Watt et al. (2012) examined
whether mathematics expectancies and values predicted the
mathematics-relatedness of careers to which students aspired
in high school in the late 1990s, with samples from Australia,
the United States, and Canada, using an occupational coding
system for the extent to which aspired careers involved mathe-
matics (none/any/average/high; from O*NET 98; U.S.
Department of Labor Employment and Training
Administration 1998).
Importance value predicted mathematical career plans for
Australian and Canadian female adolescents, consistent with
previous evidence that young women are engaged by tasks
they regard as socially meaningful and important (Eccles
and Vida 2003; see also Watt 2016). Stereotypic gender dif-
ferences in educational and occupational outcomes occurred
in the Australian sample. Male adolescents displayed higher
mathematics intrinsic value in the Australian sample and
higher mathematics self-concept in both North American sam-
ples. Self-concept was a key predictor in the North American
samples, in contrast to intrinsic value among the Australian
sample. Australian students’greater opportunities for choice
in mathematics course-taking in high school should promote
greater focus on interests (Watt et al. 2006;Wattetal.2012).
The centrality of values to Australian participants’future plans
may not be surprising, given high cultural emphasis on indi-
vidual freedom, self-expression, and imagination. For exam-
ple, Australia is ranked third on the Survival/Self-expression
values dimension of the Inglehart-Welzel Cultural Map of the
World - a scatterplot locating countries according to dimen-
sional scores from the World Values Survey (Inglehart and
Welz el 2005,p.64).
Societal values that emphasise self-expressive value
systemshavebeenproposedbyothersasanexplanation
for gender gaps in STEM careers among individuals in
Western, middle class societies (Charles and Bradley
2009). These values are prominent in the United
256 Sex Roles (2017) 77:254–271
States, Australia, and other post-industrial nations in
which people seek to take courses and find an
occupation that fits their interests and values. Charles
and Bradley (2009) found that the gender gap in engi-
neering was largest in postindustrial/postmaterialist na-
tions, such as Finland, Germany, Switzerland, and Hong
Kong. The gender gap in engineering was smaller in
materialist nations with developing economies, such as
Bulgaria, Colombia, Latvia, and Romania, where the
goal is to acquire a stable job that pays well rather than
to follow one’s passions or interests. The emphasis on
finding an interesting career matching one’s abilities in
the United States and Australia may contribute to gen-
der gaps in fields such as physics and engineering—
unless we can find ways to enhance girls’and young
women’s interest.
Measures of the mathematics-relatedness of adoles-
cents’career intentions provide an important extension
to understanding achievement-related choices beyond the
high-school years. Although limited career opportunities
as a consequence of limited participation in mathematics
have been widely recognised, until fairly recently (see
Wat t 2006,2008;Wattetal.2012), there had been little
study of relationships between these two aspects of
mathematics participation in terms of empirical work
establishing relationships between expectancies and
values and mathematical career-relatedness. Those stud-
ies utilised the U.S. O*NET 98 database to
operationalise mathematics career relatedness (U.S.
Department of Labor Employment and Training
Administration 1998; for full details see Watt 2002).
That comprehensive database was based largely on data
supplied by occupational analysts from sources such as
the Dictionary of Occupational Titles (DOT) (see
Osipow and Fitzgerald 1996, for an overview), refined
and applied to the O*NET 98 content model in which
the definition of career mathematics-relatedness was
Busing mathematics to solve problems^, with classifica-
tion codes ranging from none through any, average, and
high.
Since O*NET 1998, the more recent O*NET 2011 model
optimises power to detect individual differences by providing
a continuous score (0–100) per STEM dimension, which cap-
tures the extent to which knowledge of each of mathematics,
physics, chemistry, and biology is required for particular ca-
reers using updated information (National Center for O*NET
Development 2011). The current study builds on the previous
work by examining mathematics-related expectancy-value
predictors of Australian and U.S. adolescents’aspired ca-
reer—not only related to mathematics, but also to biology,
chemistry, and physics. Antecedent measures of mathematical
achievement and mothers’views were added as important
background factors in the expectancy-value model.
The Role of Parents
Much research and theory on the underrepresentation of
women in STEM fields seeks explanations within the
individual, whether in terms of the individual’smathe-
matical ability, interests, or course preparation. Long-
term longitudinal research conducted by expectancy-
value researchers within the Parent Socialisation Model
demonstrates the important role of parents’beliefs in
their children’s later academic and occupational choices
(Chhin et al. 2008; Davis-Kean 2005; Eccles et al.
1990,1993a; Meece et al. 1982; Pomerantz and Dong
2006). Parents’general and child-specific beliefs shape
children’s future beliefs and choices (Chhin et al. 2008).
Examples of parental beliefs have included gender-role
attitudes, perceptions of the child’s ability, and specific
expectations for the prestige of the occupation the child
will have. We utilised available measures of student-
perceived (Australian study) or actual mothers’percep-
tion (U.S. study) of their child’s mathematical ability to
examine their role in the adolescents’developing math-
ematics self-concept, interest, and importance value—
over and above the adolescent’s prior mathematical
achievement.
Perceived mothers’mathematics ability beliefs would
more likely be predictive of adolescents’motivations
than mothers’own-reported perceptions would be.
Students’interpretations of environments have been ar-
gued to be more important than objective indexes be-
cause their perceptions of events amount to reality for
them (Goodnow 1988). There is a sizeable body of data
demonstrating that children’s views on several issues are
better predicted by their perceptions of parental posi-
tions than by the positions parents report for themselves
(Goodnow 1988). Whether perceived or actual measures
were available, however, the direction of influence is
not completely clear. In a study investigating maternal
influences on girls’and female adolescents’perceptions
of mathematical ability, the reciprocal nature of parent-
child influences over time has found the strength of
influence over time to be greater from parent-to-child
than from child-to-parent (Eccles et al. 1993a).
The Current Study
Recognising that gender differences in mathematics have
been used as an explanation for the underrepresentation
ofwomeninmathematics-intensive careers, we sought
to test whether, and the extent to which, mathematics
motivations could explain aspirations towards different
science-related careers—for female and male adoles-
cents—over and above prior mathematical achievements
Sex Roles (2017) 77:254–271 257
and socialisers’perceptions. Motivational variables were
mathematics self-concept and components of value (im-
portance and interest). We expected that mathematics
self-concept, interest, and importance value would all
predict preferences for careers involving mathematics
and physics, but it was an open question whether they
would predict choices for the less mathematical sciences
of biology and chemistry. We examined the possibility
of gender-differentiated processes using the technique of
multigroup structural equation modelling.
xThe two samples available to us were adolescents who
completed 11th grade in 1998 (Australian sample) or 12th
grade in 2009 or 2010 (U.S. sample). The design included
two points in time, with data from 9th and 11th grades
(Australian) and 9th and 12th grades (U.S.), thus making it
possible to test predictions over time. We investigated gender
differences in mathematics self-concept, interest, importance
value, and aspired careers related to mathematics, physics,
chemistry, and biology. Mothers’perceptions were included
as a predictor of adolescents’expectancies and values, and
students’prior mathematics achievement and mothers’educa-
tion level were included as controls. We did not expect to find
gender differences in mathematical performance, but instead
we predicted that gender differences would favour young men
for self-concepts, values, and STEM-related career inten-
tions—except for biology, where higher values for female
adolescents were anticipated. Based on the preceding review,
gender differences were expected more among the earlier
Australian than among the contemporary U.S. sample.
Finally, values were expected to play a greater role in
predicting career choices in Australia than in the United
States.
Study 1: Australia
Method
Participants
Australian data come from the Study of Transitions and
Education Pathways (STEPS 2016;Watt2004) which in-
volved three government secondary coeducational schools in
metropolitan Sydney that enrolled students of upper middle-
class socioeconomic status. The eldest STEPS cohort, who
completed surveys through 1996–1998 at the start of each of
grades 9 and 11, is included in the present study. Those present
at both points in time were identified (N= 358 of the 459
students initially present at grade 9), of whom 27 students
who wrote that they were undecided on their career choice
were excluded. Thus the final sample included 331 students
(146, 44%, young women and 185, 56%, young men). At
grade 9, the average age for these 331 students was 14.41 years
(SD = .53); 73.1 % (n= 242) nominated English as their home
language (other home language groupings were 71 (21.6%)
Asian, 7 (2.0%) European, 10 (3.1%) Middle Eastern, and 1
(.2%) South American).
Measures
At Time 1 (grade 9) students reported their mothers’
highest level of education on an ordinal scale from 1
(partly completed high school), 2 (completed high
school), 3 (technical college), to 4 (university degree).
At Time 1 students also responded to two items about
their mothers’perceptions of students’own mathemati-
cal abilities: BHow talented does your mother think you
are at maths?^and BHow well does your mother expect
you to do at maths at school?^As a measure of prior
mathematics performance, at grade 9 students were
asked to report as a percentage, BWhat mark did you
get for your final Year 8 maths result?^
xMathematics-related motivations were measured in grade
9 using Eccles and colleagues’expectancy-value measures
(Eccles 2005; Wigfield and Eccles 2000), with grammatical
and contextualising modifications for the Australian sample
(Watt 2004). These measures focused on perceived ability/
success expectancy (self-concept), intrinsic value (interest),
and attainment/utility value (importance). Each item was rated
on a 7-point scale from 1 (not at all)to7(very). All items and
Cronbach’s alpha measures of internal consistency are pre-
sented in Table 1.
At Time 2 in grade 11, Australian students were asked:
BWhat career are you mainly considering for your future?^
Each student’s career was then coded for the amount of math-
ematics and science knowledge (biology, chemistry, and phys-
ics) that was required for that occupation, using O*NET 2011
(National Center for O*NET Development 2011), a project of
the U.S. Department of Labor. This coding provided a quan-
tification of the STEM knowledge required for a particular
occupation on a scale from 0 to 100. For example, for biostat-
isticians, mathematics = 95, biology = 49, chemistry = 12, and
physics = 10. Scores in the O*NET database are based on data
from workers and occupation experts. O*NET has the substantial
advantage of yielding a continuous score for each dimension
rather than a dichotomous variable (e.g., plans to be a physicist
or not). If students listed more than one occupation, the one that
they listed first was coded. Occupational data could be coded for
286 of the 331 participants.
Data Analyses
Analyses were conducted within the multiple-group mean and
covariance structures framework using Amos 21 (emulisrel6
option selected) to include mean-level information in addition
to the covariance matrix. All measurements were specified as
258 Sex Roles (2017) 77:254–271
interval data, except for mother’s level of education which
was ordinal. Because popular approaches to missing data,
such as mean substitution as well as listwise and pairwise
deletion, can bias results (Allison 2001), Full Information
Maximum Likelihood (FIML; Arbuckle 1996) estimation
was used to include all observed data.
Measurement Models
xMeasurement equivalence indicates that constructs are
generaliseable to different groups, that sources of bias and
error are minimal, that gender differences have not differen-
tially affected the constructs’underlying measurement
Tabl e 1 Unconstrained CFA completely standardised factor loadings (LX) and Cronbach alpha reliabilities
Construct Item Item stem LX
males
LX
females
Australia
Math self-concept (α= .832; grade 9)
Abil1: Compared with other students in your class, how talented do you consider yourself to be at maths? .60 .60
Abil2: How talented do you think you are at maths? .70 .78
Exp1: How well do you expect to do in your next maths test? .86 .87
Exp2: How well do you expect to do in school maths tasks this term? .84 .85
Exp3: How well do you think you will do in your school maths exam this year? .76 .92
Math interest (α= .941; grade 9)
Intrin1: How much do you like maths, compared with your other subjects at school? .91 .85
Intrin2: How interesting do you find maths? .92 .93
Intrin3: How enjoyable do you find maths, compared with your other school subjects? .94 .92
Math Importance (α= .881; grade 9)
Att1: To what extent will you need maths in your future work/career? .74 .77
Att2: How important is doing well in maths to you? .69 .66
Util1: How useful do you believe maths is? .82 .92
Util2: How useful do you think maths is in the everyday world? .91 .77
Util3: How useful do you think mathematical skills are in the workplace? .87 .82
Mother’s per ception
a
(α= .707; grade 9)
Mq2: How talented does your mother think you are at maths? .69 .74
Mq4: How well does your mother expect you to do at maths in high school? .80 .73
Math score
b
What mark did you get for your final Year 8 maths result? (%) 1.00 1.00
Mother’s
education
What level of education does your mother have? (1 = part high school; 2 = high school; 3 = technical
college; 4 = university)
1.00 1.00
What career are you mainly considering for your future? (grade 11)
Math
b
mathematics O*NET score /100 1.00 1.00
Physics
b
physics O*NET score /100 1.00 1.00
Chemistry
b
chemistry O*NET score /100 1.00 1.00
Biology
b
biology O*NET score /100 1.00 1.00
United States
Math Self-concept (α= .882; grade 9)
Abil1: How good are you at math? .95 .95
Abil2: If you were torank all of the studentsin your most recent math class from the worst to the best, where would you put
yourself?
.82 .84
Abil3: Compared to most of your other school subjects, how good are you at math? .80 .81
Math interest
b
(grade 9) How interesting is math to you? 1.00 1.00
Math importance (α= .811; grade 9)
Util1: How important is it that you learn math? .89 .91
Util2: How important is math to your future? .80 .73
Mother’s per ception (α= .933; grade 7)
Mq1: How good is your child at math? .94 .92
Mq2: How good is your child at math, compared to other kids? .93 .92
Mq3: How well did your child do in math this year? .86 .81
Mq4: How much natural talent does your child have in math? .90 .85
Math score
b
(grade 10) WKCE math scores 1.00 1.00
Mother’sed.
b
How many years of education do you have? 1.00 1.00
What job or career would you like when you’re 30 years old? (grade 12)
Math
b
Mathematics O*NET score 1.00 1.00
Physics
b
Physics O*NET score 1.00 1.00
Chemistry
b
Chemistry O*NET score 1.00 1.00
Biology
b
Biology O*NET score 1.00 1.00
Not presented are the measurement errors (TD); latent intercorrelations shown in Table 5
a
items constrained to load equally for two-item subscales
b
Alphas are not applicable for single-item indicators
Sex Roles (2017) 77:254–271 259
characteristics, and that between-gender differences in con-
struct means, variances, and covariances are quantitative in
nature. Strong factorial invariance (Little 1997; Meredith
1993), or scalar invariance, implies that constructs are funda-
mentally the same across groups and consequently directly
comparable. This is tenable when sequential introduction of
equality constraints for factors’loading and intercept param-
eters does not produce substantial change in model fit. The
sequence of analyses involves: (a) a combined multiple-group
model with no cross-group equality constraints imposed on
latent constructs for young men and young women (Model
1), (b) the addition of the constraint that loadings are invariant
across samples (Model 2), and (c) the assumption that inter-
cepts are equivalent (Model 3: the Measurement Equivalent
Model; Little 1997). Nested models are compared according
to change in the Chi-square statistic relative to change in de-
grees of freedom; significant worsening of model fit indicates
that the imposed model constraints are not tenable. Because
the Chi-square comparison is highly stringent and sensitive to
sample size, Little recommended inspection of changes in
practical fit indices—initially with a margin of .05 (Little
1997) and later a more stringent margin of .01 (Little
2013)—indicating acceptable model similarity to proceed
with the introduced constraints. For factors that were mea-
sured by a single indicator, those item loadings were fixed to
unity and error variances to zero; uniquenesses for the four
STEM career preference outcomes were permitted to freely
covary.
Gender Differences
Gender differences in STEM career preferences and mathe-
matics score were compared using MANOVA. For latent con-
structs measured by multiple indicators, latent mean gender
differences were estimated. Young men were set as the refer-
ence group, such that the freely estimated latent means for
young women produced the latent effect sizes corrected for
measurement error.
Gendered Processes
Multigroup structural equation models (SEMs) examined the
processes by which prior grade 8 achievement, grade 9 moti-
vational factors, mothers’perceptions, and mothers’education
influenced male versus female adolescents’STEM career
preferences by grade 11. Structural paths that were non-
significant for both female and male adolescents (p> .05)
were sequentially deleted to achieve the final model. To iden-
tify where different gender processes occurred, structural
paths were sequentially constrained to be equal; when the
change in Chi-square value relative to the single degree of
freedom change exceeded the critical value (3.841, p<.05),
the assumption of equivalent relationship was not tenable,
indicating significantly different processes for female and
male adolescents.
Results
Measurement Models
Among the Australian STEPS sample, initial unconstrained
multigroup confirmatory factor analysis (Model 1) examined
construct validity for male and female adolescents, with items
specified as indicators only for their respective factors, error
variances and factor correlations freely estimated, and no error
covariances permitted. Factor loadings were all statistically
significant and are presented in Table 1. Model fits for sequen-
tial constrained Models 1 through 3 are shown in Table 2.In
each, fit statistics were acceptable, and although the change in
Chi-square was statistically significant, showed small changes
in practical fit indices across sequential models, well below
the .05 margin referred to by Little (1997) and all but one
(p= .012) below the stricter .01 margin (Little 2013). The
condition of scalar invariance was therefore met, indicating
that quantitative comparisons of factor scores could be mean-
ingfully undertaken across gender groups.
Gender Differences
Despite similar reported prior mathematical achievement, F(1,
217) = .034, p=.853,η
p
2
= .000, male adolescents preferred
careers that were related to mathematics and physics signifi-
cantly more than did female adolescents, F(1, 217) = 16.018,
p<.001,η
p
2
= .069, for mathematics; F(1, 217) = 17.008,
p<.001,η
p
2
= .073, for physics. Female adolescents signifi-
cantly preferred careers related to biology, F(1, 217) = 4.773,
p= .030, η
p
2
= .022; there was no gender difference for
chemistry-related careers, F(1, 217) = 1.901, p= .169,
η
p
2
= .009. Careers related to mathematics were preferred
more than careers in physics, chemistry or biology (see
Tab le 3for descriptive statistics). Female adolescents had sig-
nificantly lower interest than male adolescents did in mathe-
matics, lower self-concept, and considered their mothers to
believe them to be less talented (see Table 4); male and female
adolescents perceived similar mathematics importance.
Motivational Processes
The final structural equation model, including scalar invariance
constraints for female and male adolescents, exhibited satisfacto-
ry model fit across a range of frequently emphasised indices
(χ
2
= 599.171 df = 369, RMSEA = .044, TLI = .927,
CFI = .942; see Fig. 1). xFor male adolescents, grade 11 math-
ematics-, physics- and chemistry-related careers were directly
predicted by grade 9 mathematics interest; biology-related ca-
reers, by grade 9 mathematics importance. For female
260 Sex Roles (2017) 77:254–271
adolescents, all four career STEM dimensions were predicted
by grade 9 mathematics self-concept. Interest and self-concept
were strongly correlated (ϕs = .70 for male, .59 for female
adolescents), as were interest and importance (ϕs = .49 for
male, .44 for female adolescents). Self-concept and impor-
tance correlated only for male adolescents (ϕ= .47). Grade
9 student-reported mother perceptions correlated highly with
students’mathematics interest (ϕs = .52 for male, .63 for
female adolescents), self-concept (ϕs = .70 for male, .84 for
female adolescents), and importance (ϕs=.41formale,.30
for female adolescents). Grade 8 mathematics achievement
was highly correlated with mothers’perceptions for female
adolescents (ϕ= .78), less so for male (ϕ= .52).
Achievement was moderately related to interest (ϕs = .48
for male and female adolescents), self-concept (ϕs=.46for
male, .70 for female adolescents), and importance for male
adolescents (ϕ= .23). Latent correlations between all con-
structs are shown in Table 5. Parameter estimates for the
SEM are presented in Table 6.
Significant gender moderation occurred between each of
prior mathematics achievement with self-concept, and
achievement with mothers’perceptions. Female adolescents’
mathematics self-concept was significantly and substantially
associated more with their prior achievement than was the
case for male adolescents; reported mothers’perceptions
concerning their daughters’mathematical abilities were sig-
nificantly and substantially more closely tied to daughters’
achievement in mathematics than to sons’.
Study 2: United States
Method
Participants
Participants were 298 adolescents (158, 53%, female adoles-
cents) who participated in the longitudinal Wisconsin Study of
Tabl e 2 Fit statistics for sequential constrained models
Steps χ
2
df RMSEA CFI TLI Δχ
2
(df) ΔCFI ΔTLI
Australia
Model 1: Freely estimated 470.34 302 .041 .957 .935
Model 2: Loadings invariant 493.56 312 .042 .954 .932 23.22(10)* .003 .003
Model 3: Scalar invariance 550.35 327 .046 .943 .920 56.79(15)* .011 .012
United States
Model 1: Freely estimated 218.39 138 .065 .982 .969
Model 2: Loadings invariant 221.64 144 .062 .983 .971 3.25(6) .003 .002
Model 3: Scalar invariance 253.24 161 .064 .980 .970 31.60(17)* .003 .001
*
p<.05
Tabl e 3 Descriptive statistics by gender for observed study variables, Australia and U.S.
Variables Australia United States
Males Females d
a
Males Females d
a
M (SD) M (SD) M (SD) M (SD)
Career preferences:
Math 53.74 (19.15) 43.45 (17.89) .56* 54.46 (18.08) 51.58 (17.06) .16
Physics 24.51 (22.13) 14.41 (15.75) .54* 29.40 (21.81) 19.68 (18.62) .48*
Chemistry 21.50 (19.81) 18.48 (20.96) .15 29.88 (22.65) 31.43 (22.96) -.07
Biology 16.41 (23.55) 24.84 (29.62) -.32* 27.31 (28.47) 40.11 (30.04) -.44*
Previous math performance 75.27 (16.33) 75.82 (15.21) -.03 599.97 (37.11) 598.34 (36.02) .04
Math interest
b
3.79 (1.79) 3.79 (1.88) .00
The mathematics and science knowledge requirements of preferred careers are scored on a scale from 0 to 100. Previous mathematical performance is
reported as a percentage for Australian data and as a standardised test score for U.S. data
a
dis the effect size for the gender difference, equal to the mean for males minus the mean for females divided by the pooled within-groups SD
b
Measured by a single item for U.S. sample and by multiple item indicators for Australian sample. Values for Australia are therefore shown in Table 4,
which reports values for latent variables
*p <.05
Sex Roles (2017) 77:254–271 261
Tabl e 4 Gender differences in mathematics motivation variables: latent means (KA), item intercepts (TX) and loadings (LX)
Variables KA (females) Variance p(C.R.) Item TX LX
Australia
Mom perception -.370 .560 <.001
a
mq2_9 5.358 1.000
a
mq4_9 6.042 1.000
Self-concept -.611 .754 <.001 Abil1 4.810 .765
a
Abil2 5.025 1.000
Exp1 5.498 1.080
Exp2 5.470 .990
Exp3 5.452 1.035
Interest -.715 2.753 <.001 Intrin1 4.262 .968
Intrin2 4.514 1.008
a
Intrin3 4.196 1.000
Importance -.008 1.113 .948 Att1 5.342 1.141
Att2 5.866 .752
Util1 5.826 1.111
Util2 5.827 1.079
a
Util3 5.851 1.000
United States
Mom perception -.090 1.338 .723
a
Mq1 5.671 1.000
Mq2 5.336 1.020
Mq3 5.842 1.002
Mq4 5.195 1.071
Self-concept .093 1.551 .720
a
Abil1 5.167 1.000
Abil2 5.072 1.007
Abil3 4.575 .842
Importance .283 2.709 .322
a
Util1 5.512 1.000
Util2 5.728 1.210
Parameter estimates are presented in unstandardised form; latent means (KA) are relative to the male adolescents as reference group in the original metric;
item loadings and intercepts were constrained to be equal across gender groups
a
Indicators of each construct were fixed to 1 to establish the factor metric. Not included are the measurement errors
Fig. 1 Structural diagram for
Australian STEPS male and
female adolescents. Mom perc =
Moms’perceptions; Chem.
career = Chemistry career.
Paired parameters indicate
standardised structural paths
for male/female adolescents
for scalar invariant SEM;
grade measured is presented
in brackets per construct. Not
represented are measurement
errors, uniquenesses, inter-
cepts or variances (refer to
Tab le 6). All single-item in-
dicators specified to have
loadings of 1 and 0 measure-
ment error. * p< .05. **
p<.01.*** p < .001
262 Sex Roles (2017) 77:254–271
Tab l e 5 Latent correlations between constructs for male / female adolescents
Constructs Mom ed. Mom perc. Math score Self-concept Interest Importance Math career Physics career Chemistry career Biology career
Australia
Mom ed. - -.02/.17 -.04/.03 -.01 /.02 -.08/-.01 .01/.04 -.13/-.14 .05/.03 .21
*
/.03 .25
**
/.04
Mom perc. -.02/.16 - .52
**
/.78
**
.71
**
/.84
**
.52
**
/.63
**
.41
**
/.31
**
.24
*
/.40
**
.07/.25
*
.13/.26
*
.14/.23
*
Math score .04/.02 .51
**
/.77
**
-.46
**
/.70
**
.48
**
/.48
**
.24
**
/.15 .25
*
/.38
**
.10/.24
*
.02/.25
*
.07/.21
*
Self-concept -.01 /.01 .71
**
/.82
**
.45
**
/.68
**
-.70
**
/.59
**
.47
**
/.16 .30
**
/.39
**
.08/.28
**
.05/.30
**
.18
*
/.24
*
Interest -.08/-.02 .50
**
/.60
**
.47
**
/.44
**
.69
**
/.54
**
-.49
**
/.45 .36
**
/.31
**
.20
*
/.24
*
.19
*
/.22
*
.20
*
/.16
Importance .02/.04 .41
**
/.33
**
.23
**
/.16 .47
**
/.19
*
.49
**
/.47
**
-.18
*
/.14 -.02/.06 .03/.14 .18
*
/.17
Math career -.13/-.14 .23
*
/.38
**
.24
**
/.37
**
.28
**
/.38
**
.34
**
/.28
**
.18
*
/.15 - .62
**
/.56
**
.45
**
/.54
**
.29
**
/.37
**
Physics career .05/.03 .06/.23
*
.10/.22
*
.07/.26
**
.19
*
/.22
*
-.03/.07 .62
**
/.56
**
-.73
**
/.74
**
.46
**
/.54
**
Chem. career .21
*
/.03 .12/.24
*
.01 /.23
*
.03/.27
**
.19
*
/.19
*
.03/.15 .45
**
/.53
**
.73
**
/.73
**
-.77
**
/.88
**
Bio. career .25
**
/.04 .13/.21 .07/.20
*
.17
*
/.23
*
.20
*
/.15 .18
*
/.17 .29
**
/.37
**
.46
**
/.54
**
.77
**
/.88
**
-
United States
Mom ed. - .11/.14 .19/.28
**
.16/.15 .08/.12 .13/.08 .06 /.06 .11/.08 .15/-.03 .20
*
/.08
Mom perc. .11/.14 - .43
**
/.42
**
.38
**
/.18
*
.23
**
/.20
**
.25
**
/.25
**
.06 /.11 .05/.15 .01/.14 .00/.15
Math score .19/.28
**
.42
**
/.42
**
-.34
**
/.27
**
.36
**
/.26
**
.21
*
/.18
*
.17 /.11 .02/.28
**
.07/.10 .15/.04
Self-concept .16/.15 .37
**
/.18
*
.34
**
/.28
**
-.80
**
/.71
**
.88
**
/.86
**
.28
**
/-.05 .09/.08 .07/-.12 .00/-.10
Interest .08/.12 .23
**
/.19 .36
**
/.25
**
.80
**
/.70
**
-.81
**
/.68
**
.32
**
/-.02 .11/.05 .09/-.12 .05/-.09
Importance .13/.08 .26
**
/.24
**
.21
**
/.19
*
.88
**
/.86
**
.81
**
/.68
**
-.26
**
/-.09 .10 /.02 .03/-.15 -.03 /-.14
Math career .06/.06 .05 /.11 .17 /.11 .28
**
/-.05 .32
**
/-.02 .26
**
/-.08 - .63
**
/.55
**
.45
**
/.53
**
.08 /.43
**
Physics career .11/.08 .05/.15 .02 /.28
**
.10/.08 .11/.05 .09/.02 .63
**
/.55
**
-.54
**
/.53
**
.09/.33
**
Chem. career .15/-.03 .01/.14 .07 /.10 .07/-.13 .09/-.12 .03/-.15 .45
**
/.53
**
.54
**
/.53
**
-.69
**
/.84
**
Bio. Career .21
*
/.08 .00/.15 .15 /.04 .00/-.11 .05/-.08 -.03/-.14 .09/.43
**
.09 /.33
**
.69
**
/.83
**
-
Mom perc mom’s perceptions, Mom ed mom’s education. Correlations below diagonal from unconstrained CFA; above diagonal from scalar invariant CFA
*
p< .05.
**
p<.01
Sex Roles (2017) 77:254–271 263
Families and Work (Hyde et al. 1995; the sample excludes
families who were part of an intervention group;
Harackiewicz et al. 2012). The average age of students was
15.50 years (SD = .32) in 9th grade in 2006–2007 and
18.51 years (SD = .33) in 12th grade in 2009–2010. Those
present at both points in time were included (reduced N=277;
153, 55%, female adolescents). Ethnic distribution was 90.4%
White adolescents, 1.0% Black, and 8.6% biracial or
Tabl e 6 Scalar invariant SEM: Item loadings (LX/Y in original metric and standardised), intercepts (TX/Y), and correlated uniquenesses (Z)
Construct Item TX/Y LX/Y Standardised LX/Y: Correlated uniquenesses (Males/Females p<.01)
Males / Females Math Physics Chem.
Australia
Self-concept Abil1 4.65 .76 .57 .67
Abil2 4.82 1.00 .72 .80
Exp1 5.27 1.08 .85 .90
Exp2 5.26 .99 .83 .88
Exp3 5.23 1.04 .81 .91
Interest Intrin1 3.94 .97 .91 .87
Intrin2 4.17 1.01 .93 .92
Intrin3 3.86 1.00 .94 .93
Importance Att1 5.30 1.15 .76 .74
Att2 5.84 .76 .68 .67
Util1 5.79 1.12 .84 .90
Util2 5.78 1.08 .89 .81
Util3 5.81 1.00 .86 .82
Mom. perc mq2_9 5.24 1.00 .69 .75
mq4_9 5.92 1.00 .80 .75
Math score 72.96 1.00 1.00 1.00
Mom ed. 2.79 1.00 1.00 1.00
Math career 52.05 1.00 1.00 1.00
Physics career 23.87 1.00 1.00 1.00 .24 / .51 --
Chemistry career 21.11 1.00 1.00 1.00 .43 / .48 .72 / .71 --
Biology career 15.47 1.00 1.00 1.00 .24 / .31 .46 / .51 .78 / .88
United States
Mom perc. Mq1 5.78 1.00 .94 .92
Mq2 5.35 1.02 .93 .92
Mq3 5.65 1.00 .86 .81
Mq4 5.48 1.05 .90 .85
Self-concept Abil1 5.22 1.00 .95 .95
Abil2 5.05 .84 .82 .84
Abil3 4.52 .99 .80 .81
Importance Util1 5.60 1.00 .79 .73
Util2 5.73 1.14 .89 .91
Interest 3.76 1.00 1.00 1.00
Math score 598.92 1.00 1.00 1.00
Mom ed. 15.25 1.00 1.00 1.00
Math career 54.57 1.00 1.00 1.00
Physics career 29.67 1.00 1.00 1.00 .66 / .55 --
Chemistry career 29.62 1.00 1.00 1.00 .43 / .53 .52/ .54 --
Biology career 27.31 1.00 1.00 1.00 .04
a
/ .43 .07
a
/ .34 .69 / .83
Mom perc mom’s perceptions, Mom ed Mom’s education. Not presented are the measurement errors (TD/TE), variances/covariances; latent intercor-
relations (Φ) shown in Table 5; structural paths (γ)inFigs.1and 2
a
not significant (p> .05)
264 Sex Roles (2017) 77:254–271
multiracial, characteristic of the state of Wisconsin where 90%
of the population is White (U.S. Census Bureau 2006).
Participants attended 144 different high schools across
Wisconsin and 12 other states. Roughly half the sample com-
pleted high school in 2009; the other half in 2010. Mothers’
total years of education ranged from 10 to 20
(M= 15.31 years, SD = 2.04), where high school graduation
or GED (passing the General Educational Development test)
counted as 12 years.
Measures
At grade 9 (Time 1), as in Study 1, youth completed items
developed by Eccles and Wigfield (1995). Mathematics self-
concept was measured with three items (BHow good at math
are you?^;BIf you were to rank all of the students in your most
recent math class from the worst to best, where would you put
yourself?^;BCompared to most of your other school subjects,
how good are you at math?^); interest was tapped by a single
item (BHow interesting is math to you?^); and importance by
two items (BHow important is it that you learn math?^;BHow
important is math in your future?^), rated on 7-point scales
from 1 (not at all)to7(very).
The study was embedded within a larger ongoing longitu-
dinal study. At an earlier wave of data collection in grade 7,
mothers had rated their perceptions of their child’s mathemat-
ical ability using four items from the Eccles scales (Eccles and
Wigfield 1995; Eccles et al. 1983) rated on 5-point scales that
were also included (BHow good is your child at math?^;How
good is your child at math, compared to other kids?^;BHow
well did yourchild do in math this year?^;BHow much natural
talent does your child have in math?^).
School transcripts were obtained for 242 of the partici-
pants. Standardised mathematics test scores were taken from
the Wisconsin Knowledge and Concepts Exam (WKCE) ad-
ministered to all Wisconsin students in public schools in
October of 10th grade. Participants’WKCE scores ranged
from 473 to 723 (M=599.06,SD = 36.43). These scores were
missing for students who had moved to other states and those
who attended private schools or were home schooled. Finally,
at 12th grade (Time 2), students were asked, BWhat job or
career would you like to have when you’re 30 years old?^
Responses were coded using O*NET 2011 as described in
Study 1 such that a continuous score provided a quantification
of the STEM knowledge required for nominated occupations
on a scale from 0 to 100.
Results
Measurement Model
The same method used to determine scalar invariance was
repeated with the U.S. sample. As in Study 1, all factor
loadings for the unconstrained model were statistically signif-
icant and the model indicated adequate fit (see Table 1).
Model fit for the sequentially constrained models supported
scalar invariance (see Table 2).
Gender Differences
A MANOVA examined gender differences in mothers’
education, adolescents’mathematics interest, standardised
test scores, and occupational aspirations. There was a sig-
nificant multivariate effect of gender, F(7, 185) = 4.127,
p<.001,η
p
2
= .135. This was due to significant differ-
ences on biology-related career aspirations, which were
higher for female adolescents, F(1, 191) = 7.014,
p=.009,η
p
2
= .035, and physics-related careers, which
were higher for male adolescents, F(1, 191) = 13.426,
p< .001, η
p
2
= .066; both effect sizes were moderate
(see Table 3). Other variables were similar for male and
female adolescents, including the standardised mathematics
test (d= .04). As in Study 1, latent means were set to
zero for male adolescents for constructs measured by mul-
tiple item indicators such that latent means represent fe-
male adolescents’difference relative to males. No signifi-
cant gender differences were found for mathematics inter-
est, self-concept, or importance (see Table 4).
Motivational Processes
The final structural equation model, shown in Fig. 2
(including scalar invariance constraints), exhibited satis-
factory fit (χ
2
= 333.56, df = 201, RMSEA = .049,
TLI = .923, CFI = .943; see Table 2). For male adoles-
cents only, 9th grade mathematics interest predicted
12th grade career aspirations related to mathematics.
Physics career aspirations were predicted by importance
value for male, and mathematics achievement for female
adolescents. Mathematics interest and self-concept were
moderately correlated (ϕs = .52 for male, .43 for female
adolescents), as were interest and importance value
(ϕs = .45 for male, .24 for female adolescents).
Students with higher self-concept had higher scores on
the mathematics standardised test (ϕs=.56formale,
.41 for female adolescents); test score was correlated
with mathematics interest (ϕs=.37formale,.22for
female adolescents). Mothers’perceptions correlated
with students’mathematics interest (ϕs=.27formale,
.21 for female adolescents), self-concept (ϕs=.63for
male, .33 for female adolescents), and standardised test
scores (ϕs = .64 for male, .58 for female adolescents).
There were no significant gender moderated paths.
Latent correlations among all constructs are shown in
Table 5; parameter estimates for the SEM are in
Tab le 6.
Sex Roles (2017) 77:254–271 265
Discussion
Our study sought to understand whether mathematics-related
motivational factors predicted the mathematics, biology, chem-
istry, and physics requirements of careers indicated by female
and male adolescents at the end of high school among samples
from two similar settings, albeit collected a decade apart and
with slightly different methods. Did mathematics act as a crit-
ical filter across diverse STEM career aspirations?
Mathematics-related perceptions and achievement had implica-
tions for each examined STEM domain in the Australian sam-
ple and for half the STEM domains in the U.S. sample, with
different emphases for female and male adolescents. To our
knowledge, this is the first study to disaggregate high-school
students’preferred future occupations for the knowledge re-
quired for not only mathematics, but also biology, chemistry,
and physics. This allowed us to examine gender differences
across different STEM dimensions of career plans; whether
and how they were predicted by mathematics self-concepts,
values, achievement background and mother perceptions; and
whether predictions were moderated by students’gender.
Findings across the two samples showed many similarities.
Consistent with the patterns noted in the introduction, male
adolescents’aspired careers required more knowledge of
physics than did females’—in both samples. In contrast, fe-
male adolescents were more likely than were male, to prefer
careers that involved knowledge of biology. In Australia, male
adolescents’preferred careers also required more mathemati-
cal knowledge than did females’; in the U.S. sample, the gen-
der difference was not significant (consistent with findings for
the Eccles et al. U.S. CAB sample collected a decade earlier,
reported in Watt et al. 2012). Female adolescents’lesser pref-
erence for physics-related careers (and mathematics-related
careers in the Australian sample) did not lie in deficits in
mathematical performance. Gender similarities occurred for
mathematical performance in both samples, consistent with
the well-sampled TIMSS and PISA studies of both nations
(see Else-Quest et al. 2010). Despite this performance similar-
ity, male adolescents in the Australian sample had higher
mathematics self-concept and interest than female adolescents
did. For chemistry, there was no significant difference in
career-related intentions in either sample, consistent with na-
tional patterns in the United States (National Student
Clearinghouse Research Center 2013)andAustralia
(Dobson 2012).
Limitations and Future Research Implications
The two studies reported here were not designed together,
limiting direct comparisons. Nonetheless, the core motivation
measures were based on Eccles and colleagues’(1995)mea-
sures in both, with slight variations. Both samples were pre-
dominantly White (73% for Australia, 90% for U.S.) and
overrepresented middle-class families, limiting the study’s
generalisability to ethnic minority and lower socioeconomic
background youth.
The longitudinal designs with two points in time add
weight to inferences about influences on students’STEM-
related career aspirations by the end of high school, although
caution should be exercised in regard to causal inferences. It is
tempting to interpret the finding that U.S. mothers’ratings of
their adolescents’mathematics ability in 7th grade predicted
the adolescents’self-concept of mathematics ability in 9th
grade (see Fig. 2) as an indicator of the influence of maternal
beliefs on children’s self-concept. However, third-variable ex-
planations are also possible. For example, the child’s mathematics
Fig. 2 Structural diagram for
WSFW United States male and
female adolescents. Mom perc =
Moms’perceptions; Chem.
career = Chemistry career.
Paired parameters indicate
standardised structural paths
for male/female adolescents
for scalar invariant SEM;
grade measured is presented
in brackets for each construct.
Not represented are measure-
ment errors, uniquenesses,
intercepts or variances (refer
to Table 6). All single-item
indicators were specified to
have loadings of 1 and 0
measurement error. * p<.05.
** p< .01. *** p <.001
266 Sex Roles (2017) 77:254–271
ability, demonstrated earlier through standardised tests or grades,
may account for the mother’s rating of her child’s ability and the
child’s mathematics self-concept. This objection was addressed
substantially by the inclusion in the model of measured mathe-
matical performance, which means that other paths represent the
significance of those relationships controlling for background
mathematics achievement.
An important innovation was our use of continuous mea-
sures of the amount of mathematics, biology, chemistry, and
physics knowledge required for the adult occupation preferred
by the adolescent by the end of high school using O*NET
2011 scores (National Center for O*NET Development
2011). These scores allowed us to disaggregate patterns of
prediction for mathematics-, physics-, chemistry-, and
biology-related aspects of aspired careers. Such distinctions
are particularly important when trying to understand gendered
participation in diverse scientific fields. We believe this ap-
proach will prove useful to future research as we continue to
work to understand why adolescents do, and do not, aspire to
STEM-related careers.
In the current study, mathematics self-concept or perfor-
mance was more important for female adolescents, and math-
ematics interest for male, in their STEM-related career
choices. These findings contrast with previous research
(Eccles and Vida 2003), including previous findings with this
same Australian sample in relation to mathematics career
plans (Watt et al. 2012). However, those analyses were based
on an earlier O*NET coding (U.S. Department of Labor
Employment and Training Administration 1998), which pro-
vided four less differentiated categories that could have atten-
uated relationships compared with the continuous scale of
O*NET 2011. Moreover, O*NET 1998 focused on the
amount of mathematics required in careers rather than the
knowledge prerequisite to careers (as done in the current cod-
ing). Career characteristics and knowledge are also likely to
have changed in the intervening 13 years. For example, the
field of law increased on its mathematics O*NET code from
1998 to 2011, perhaps due to the increased reliance on scien-
tific forms of evidence in litigation, and it was a career to
which more of the Australian female than male adolescents
had aspired. In contrast, the field of finance decreased in its
mathematics requirements, possibly due to technological de-
velopments in the form of helpful software, and it was a career
preferred more by male adolescents. The mathematics career
results reported here are therefore not directly comparable to
those reported by Watt et al. (2012).
Practice Implications
It was not the case that female adolescents showed a deficit in
mathematics achievement as assessed by school grades or
standardised tests. Despite this equivalence, female adoles-
cents in the Australian sample held lower mathematics self-
concept, which was more highly based on their prior achieve-
ments than males’,suggestingthatmaleadolescents’ability
beliefs may be inflated. Interventions developed to counter
stereotype threat effects show great promise for closing such
gender gaps (Miyake et al. 2010). Another strategy is to work
on the discrepancy between the gender difference in prefer-
ences for careers involving physics and the gender similarity
in actual mathematical performance. Female adolescents need
to know that they are scoring as well as male adolescents on
these objective measures and that they can expect to be suc-
cessful in mathematics-intensive careers. More pronounced
gender differences in aspirations among the Australian sample
raise the question of just how much choice should be offered
to students in terms of participation in core STEM disciplines
during high school.
Actual, or perceived, mathematical abilities were im-
portant for female adolescents’mathematics-related ca-
reerplansinbothsamples.IntheUnitedStates,for
female but not for male adolescents, standardised math-
ematics test score in 10th grade predicted 12th grade
preference for careers involving physics. In the
Australian sample, perceived mathematical abilities pre-
dicted each of mathematics-, physics-, chemistry-, and
biology-related career plans for Australian female ado-
lescents, but not for males. In addition, female adoles-
cents’self-concepts were significantly more closely tied
to their previously demonstrated mathematical achieve-
ments than were males’, supporting our anticipated gen-
der differences in processes. The fact that female ado-
lescents seemingly will not plan to pursue a STEM-
related career unless they think they are good at it
may stem from gender stereotypes, which magnify un-
certainty or doubts about their success.
The role of interest was also affirmed in our study. Interest
in mathematics was significantly linked to career physics, ca-
reer mathematics, and career chemistry dimensions for male
adolescents in the Australian sample. In the United States,
mathematics interest linked to career mathematics for male
adolescents, although measurement by a single item may have
reduced precision and the ability to predict other variables
(Cole and Preacher 2014). Why would mathematics interest
be an important predictor of mathematics-intensive careers for
male but not for female adolescents? The answer may lie in
the pattern for actual or perceived abilities noted earlier and
may be consistent with the original premise of the Eccles
et al.’s(1983) expectancy-value model, which posits that only
if individuals expect they can succeed and value mathematics
will they enrol in advanced mathematics.
Did mathematics-related predictors impact career out-
comes for biology, which is not a mathematics-intensive
field in the way that physics is? Mathematics interest was
unrelated to biology careers, for either female or male
adolescents, in either sample. In the U.S. sample, none
Sex Roles (2017) 77:254–271 267
of the mathematics variables (self-concept, interest, impor-
tance, test scores, or mothers’perceptions) predicted
biology-related career plans. However, mathematics self-
concept did predict all STEM-related career dimensions
for female adolescents (but not for males) in the
Australian sample, and mathematics importance was pre-
dictive of biological career plans for male adolescents.
Patterns of very high uniqueness between chemistry and
biology career plans in both samples (especially for female
adolescents) and between physics and chemistry career
plans in the Australian sample implied that students (espe-
cially female adolescents) who had chemistry interests
were more likely to plan to pursue biochemistry careers
(both samples), and students with physics interests were
more likely to plan to pursue physics/chemistry careers
(see Table 6). Therefore, mathematics-related motivations
are important for a range of STEM-related career plans,
beyond mathematics-intensive careers, towards which pro-
moting students’self-concepts and values is key.
According to expectancy-value theory, and the Parent
Socialisation Model embedded within it, the beliefs of
parents as socialisers exert an influence on their chil-
dren’s ability self-concept, values, and long-term goals
(Chhin et al. 2008; Eccles et al. 1993b;Fromeand
Eccles 1998; Jacobs and Eccles 1992). Our results are
consistent with this reasoning. The effects were stronger
with the Australian sample, likely because students re-
ported on their mothers’beliefs, whereas with the U.S.
data, mothers reported their own beliefs. Additionally,
mothers’perception data were collected earlier in the
U.S.study(grade7)thanintheAustralianstudy(grade
9) so that the former were a more distal influence. In
both samples, mothers’perceptions of their child’s
mathematics ability correlated with the child’smathe-
matics self-concept and interest. None of these correla-
tions was gender-differentiated; that is, mothers’beliefs
in their children’s mathematics ability related similarly
to female and male adolescents’mathematics self-
concept and interest (as well as importance in the
Australian sample).
Intriguingly, in the Australian sample, mothers’per-
ceptions were significantly and substantially higher for
sons than for daughters, at the same time as these were
significantly and substantially more tied to students’ac-
tual mathematical achievements for daughters than for
sons. These findings point to a gender bias in these
mothers’judgments about sons versus daughters, or they
reflect male and female adolescents’own differences in
ability self-concepts. Stronger associations among the
Australian sample reflect that students’perception of
how the parent feels is more important to their motiva-
tions, perceptions, and decisions than parents’own re-
ported perceptions (Goodnow 1988). These findings
point to the importance of parents in encouraging or
discouraging their adolescents’STEM career aspirations.
Therefore, interventions to increase STEM participation
may be able to target parents successfully (e.g.,
Harackiewicz et al. 2012).
Conclusions
In tune with the critical filter hypothesis, gender differences in
mathematics-related motivations had consequences beyond
mathematical career plans to diverse scientific career aspira-
tions. Consequently, educators’and policymakers’attention
to optimise positive student perceptions in mathematics
should yield benefits for STEM participation more broadly.
Patterns of gender differences and predictive variables varied
considerably for mathematics and different sciences. In future
research on gender and STEM, it will be important to differ-
entiate biology, chemistry, and physics, rather than relying on
aggregated measures of attitudes about science or STEM. At
the same time, we need to understand the choices into which
young women opt, instead of singularly focusing on why they
opt out of particular STEM domains (Eccles 2013). This ap-
proach entails taking a dual focus to encompass the career
choices that girls and young women actually make.
Acknowledgments The STEPS Study (www.stepsstudy.org)was
supported by Australian Research Council Discovery grant
DP110100472 and ARF awarded to Watt. The U.S. study was
supported by the National Science Foundation DRL 0814750 to Hyde
and Harackiewicz; and the Institute of Education Sciences, U.S.
Department of Education, Award #R305B090009 to the University of
Wisconsin—Madison. The opinions expressed are those of the authors
and do not represent views of the U.S. Department of Education or the
National Science Foundation; we thank Dan Lamanna, Maria Mens,
Stefan Slater, and Ryan Svoboda for assistance with career coding; and
Carlie Allison and Corinne Boldt for assistance with data collection.
Compliance with Ethical Standards We attest that all work conforms
with Australian and U.S. required ethical bodies and procedures.
Helen M. G. Watt, Janet S. Hyde, Jennifer Petersen, Zoe A. Morris,
Christopher S. Rozek & Judith M. Harackiewicz
References
Allison, P. D. (2001). Missing data.Sage University papers series on
quantitative applications in the social sciences, 07–136. Thousand
Oaks, CA: Sage.
Arbuckle, J. L. (1996). Full information estimation in the presence of
incomplete data. In G. A. Marcoulides & R. E. Schumacker
(Eds.), Advanced structural equation modelling: Issues and
techniques (pp. 243–277). Mahwah, NJ: Erlbaum.
Australian Government Department of Education and Training. (2014).
UCube higher education data. Retrieved from
http://highereducationstatistics.education.gov.au/. Accessed 26
August 2016.
268 Sex Roles (2017) 77:254–271
Ceci, S. J., & Williams, W. M. (2011). Understanding current causes of
women’s underrepresentation in science. PNAS, 108,3157–3162.
doi:10.1073/pnas.1014871108.
Charles, M., & Bradley, K. (2009). Indulging our gendered selves? Sex
segregation by field of study in 44 countries. American Journal of
Sociology, 114,924–976. doi:10.1086/595942.
Cheryan, S., Master, A., & Meltzoff, A. N. (2015). Cultural stereotypes as
gatekeepers: Increasing girls’interest in computer science and engi-
neering by diversifying stereotypes. Frontiers in Psychology, 6(49),
2–8. doi:10.3389/fpsyg.2015.00049.
Chhin,C.S.,Bleeker,M.M.,&Jacobs,J.E.(2008).Gender-
typed occupational choices: The long-term impact of parents’
beliefs and expectations. In H. M. G. Watt & J. S. Eccles
(Eds.), Gender and occupational outcomes (pp. 215–234).
Washington, DC: American Psychological Association.
Cole, D. A., & Preacher, K. J. (2014). Manifest variable path analysis:
Potentially serious and misleading consequences due to uncorrected
measurement error. Psychological Methods, 19(2), 300–315.
doi:10.1037/a0033805.
Crombie, G., Sinclair, N., Silverthorn, N., Byrne, B. M., DuBois, D. L., &
Trinneer, A. (2005). Predictors of young adolescents’math grades
and course enrollment intentions: Gender similarities and differ-
ences. Sex Roles, 52,351–367. doi:10.1007/s11199-005-2678-1.
Davis-Kean, P. E. (2005). The influence of parent education and family
income on child achievement: The indirect role of parental expecta-
tions and the home environment. Journal of Family Psychology, 19,
294–304. doi:10.1037/0893-3200.19.2.294.
Diekman, A. B., Clark, E. K., Johnson,A. M., Brown, E. R., & Steinberg,
M. (2011). Malleability in communal goals and beliefs influences
attraction to STEM careers: Evidence for a goal congruity perspec-
tive. Journal of Personality and Social Psychology, 101,902–918.
doi:10.1037/a0025199.
Dobson, I. R. (2012). Unhealthy science? University natural and physical
sciences 2002 to 2009/10. Retrieved from http://www.
chiefscientist.gov.au/wp-content/uploads/Unhealthy-Science-
Report-Ian-R-Dobson.pdf. Accessed 26 August 2016.
Eccles, J. S. (2005). Subjective task value and the Eccles et al. model of
achievement-related choices. In A. J. Elliot & C. S. Dweck (Eds.),
Handbook of competence and motivation (pp. 105–121). New York:
Guilford.
Eccles, J. S. (2009). Who am I and what am I going to do with my life?
Personal and collective identities as motivators of action.
Educational Psychologist, 44,78–89. doi:10.1080
/00461520902832368.
Eccles, J. S. (2013). Gender and STEM: Opting in versus dropping out.
International Journal of Gender, Science and Technology, 5(3), 2–3.
Retrieved from http://www.vhto.nl/fileadmin/user_
upload/documents/publicaties/Gender_and_STEM_Opting_in_
versus_dropping_out.pdf. Accessed 26 August 2016.
Eccles, J. S., & Vida, M. (2003, April). Predicting mathematics-related
career aspirations and choices. Paper presented at the Society for
Research in child development (SRCD) biennial conference,
Tampa, FL.
Eccles, J. S., & Wigfield, A. (1995). In the mind of the actor: The struc-
ture of adolescents' achievement task values and expectancy-related
beliefs. Personality and Social Psychology Bulletin, 21,215–225.
doi:10.1177/0146167295213003.
Eccles, (. P.). J., Adler, T. F., Futterman, R., Goff, S. B., Kaczala, C. M.,
Meece, J. L., & Midgley, C. (1983). Expectancies, values, and aca-
demic behaviors. In J. T. Spence (Ed.), Achievement and achieve-
ment motivation (pp. 75–146). San Francisco: W. H. Freeman.
Eccles, J. S., Jacobs, J. E., & Harold, R. D. (1990). Gender role stereo-
types, expectancy effects, and parents’socialization of gender dif-
ferences. Journal of Social Issues, 46,183–201. doi:10.1111/j.1540-
4560.1990.tb01929.x.
Eccles, J. S., Jacobs, J. E., Harold, R. D., Yoon, K. S., Abreton, A., &
Freedman-Doan, C. (1993a). Parents and gender-role socialization
during the middle childhood and adolescent years. In S. Oskamp &
M. Costanzo (Eds.), Gender issues in contemporary society,
Claremont symposium on applied social psychology, 6 (pp. 59–
83). Thousand Oaks, CA: Sage.
Eccles, J., Wigfield, A., Harold, R. D., & Blumenfeld, P. (1993b). Age
and gender differences in children’s self- and task perceptions during
elementary school. Child Development, 64,830–847. doi:10.1111
/j.1467-8624.1993.tb02946.x.
Else-Quest, N. M., Hyde, J. S., & Linn, M. C. (2010). Cross-national
patterns of gender differences in mathematics: A meta-analysis.
Psychological Bulletin, 136,103–127. doi:10.1037/a0018053.
Farmer, H. S., Wardrop, J. L., & Rotella, S. C. (1999). Antecedent factors
differentiating women and men in science/nonscience careers.
Psychology of Women Quarterly, 23,763–780. doi:10.1111
/j.1471-6402.1999.tb00396.x.
Frome, P. M., & Eccles, J. S. (1998). Parents’influence on children’s
achievement-related perceptions. Journal of Personality and
Social Psychology, 74,435–452. doi:10.1037/0022-3514.74.2.435.
Goodnow, J. J. (1988). Parents’ideas, actions, and feelings: Models and
methods from developmental and social psychology. Child
Development, 59,286–320. doi:10.2307/1130312.
Harackiewicz, J. M., Barron, K. E., Tauer, J. M., Carter, S. M., & Elliot,
A. J. (2000). Short-term and long-term consequences of achieve-
ment goals: Predicting interest and performance over time. Journal
of Educational Psychology, 92,316–330. doi:10.1037/0022-
0663.92.2.316.
Harackiewicz, J. M., Barron, K. E., Tauer, J. M., & Elliot, A. J. (2002).
Predicting success in college: A longitudinal study of achievement
goals and ability measures as predictors of interest and performance
from freshman year through graduation. Journal of Educational
Psychology, 94,562–575. doi:10.1037/0022-0663.94.3.562.
Harackiewicz, J. M., Durik, A. M., Barron, K. E., Linnenbrink-Garcia, L.,
& Tauer, J. M. (2008). The role of achievement goals in the devel-
opment of interest: Reciprocal relations between achievement goals,
interest, and performance. Journal of Educational Psychology, 100,
105–122. doi:10.1037/0022-0663.100.1.105.
Harackiewicz, J. M., Rozek, C. S., Hulleman, C. S., & Hyde, J. S. (2012).
Helping parents motivate their adolescents in mathematics and sci-
ence: An experimental test. Psychological Science, 23,899–906.
doi:10.1177/0956797611435530.
Harackiewicz, J. M., Canning, E. A., Tibbetts, Y., Priniski, S. J., & Hyde,
J. S. (2016a). Closing achievement gaps with a utility-value inter-
vention: Disentangling race and social class. Journal of Personality
and Social Psychology, 111,745–765. doi:10.1037/pspp0000075.
Harackiewicz, J. M., Smith, J. L., & Priniski, S. J. (2016b). Interest
matters: The importance of promoting interest in education. Policy
Insights from the Behavioral and Brain Sciences, 3, 220–227.
doi:10.1177/2372732216655542.
Hidi, S. (1990). Interest and its contribution as a mental resource for
learning. Review of Educational Research, 60, 549–571.
doi:10.3102/00346543060004549.
Hidi, S., & Harackiewicz, J. M. (2000). Motivating the academi-
cally unmotivated: A critical issue for the twenty-first centu-
ry. Review of Educational Research, 70, 151–179.
doi:10.3102/00346543070002151.
Hill, C., Corbett, C., & St. Rose, A. (2010). Why so few? Women in
science, technology, engineering, and mathematics. American
Association of University Women. Retrieved from https://www.
aauw.org/files/2013/02/Why-So-Few-Women-in-Science-
Technology-Engineering-and-Mathematics.pdf. Accessed 26
August 2016.
Hulleman, C. S., & Harackiewicz, J. M. (2009). Promoting interest and
performance in high school science classes. Science, 326,1410–
1412. doi:10.1126/science.1177067.
Sex Roles (2017) 77:254–271 269
Hulleman, C. S., Durik, A. M., Schweigert, S. A., & Harackiewicz, J. M.
(2008). Task values, achievement goals, and interest: An integrative
analysis. Journal of Educational Psychology, 100,398–416.
doi:10.1037/0022-0663.100.2.398.
Hulleman, C. S., Godes, O., Hendricks, B. L., & Harackiewicz, J. M.
(2010). Enhancing interest and performance with a utility value
intervention. Journal of Educational Psychology, 102(4), 880–895.
doi:10.1037/a0019506.
Hyde, J. S. (2005). The gender similarities hypothesis. American
Psychologist, 60,581–592. doi:10.1037/0003-066X.60.6.581.
Hyde, J. S., Klein, M. H., Essex, M. J., & Clark, R. (1995). Maternity
leave and women's mental health. Psychology of Women Quarterly,
19,257–285. doi:10.1111/j.1471-6402.1995.tb00291.x.
Hyde, J. S., Lindberg, S. M., Linn, M. C., Ellis, A., & Williams, C.
(2008). Gender similarities characterize math performance.
Science, 321,494–495. doi:10.1126/science.1160364.
Inglehart, R., & Welzel, C. (2005). Modernization, cultural change and
democracy. New York: Cambridge University Press.
Jacobs, J. E., & Eccles, J. S. (1992). The impact of mothers’gender-role
stereotypic beliefs on mothers’and children’s ability perceptions.
Journal of Personality and Social Psychology, 63, 932–944.
doi:10.1037/0022-3514.63.6.932.
Lazarides, R., & Ittel, A. (2013). Mathematics interest and achievement:
What role do perceived parent and teacher support play? A longitu-
dinal analysis. International Journal of Gender, Science and
Technology, 5(3), 207–231. Retrieved fromhttp://genderandset.
open.ac.uk/index.php/genderandset/article/viewFile/301/526.
Accessed 26 August 2016.
Leaper, C., Farkas, T., & Brown, C. S. (2012). Adolescent girls’experi-
ences and gender-related beliefs in relation to their motivation in
math/science and English. Journal of Youth and Adolescence, 41(3),
268–282. doi:10.1007/s10964-011-9693-z.
Lindberg, S. M., Hyde, J. S., Petersen, J., & Linn, M. C. (2010). New
trends in gender and mathematics performance: A meta-analysis.
Psychological Bulletin, 136, 1123–1135. doi:10.1037/a0021276.
Little, T. D. (1997). Mean and covariance structures (MACS) analyses of
cross-cultural data: Practical and theoretical issues. Multivariate
Behavioral Research, 32,53–76. doi:10.1207/s15327906
mbr3201_3.
Little, T. D. (2013). Longitudinal structural equation modeling. New
York, NY: The Guildford Press.
Maltese, A. V., & Tai, R. H. (2011). Pipeline persistence: Examining the
association of educational experiences with earned degrees in
STEM among U.S. students. Science Education, 95, 877–907.
doi:10.1002/sce.20441.
Meece, J. L.,Eccles, (. P.). J., Kaczala, C. M., Goff, S. B., & Putterman, R.
(1982). Sex differences in math achievement: Toward a model of
academic choice. Psychological Bulletin, 91,324–348. doi:10.1037
/0033-2909.91.2.324.
Meece, J. L., Wigfield, A., & Eccles, J. S. (1990). Predictors of math
anxiety and its influence on young adolescents’course enrollment
intentions and performance in mathematics. Journal of Educational
Psychology, 82,60–70. doi:10.1037/0022-0663.82.1.60.
Meredith, W. (1993). Measurement invariance, factor analysis and facto-
rial invariance. Psychometrika, 58(4), 525–543. doi:10.1007
/BF02294825.
Miller, C. C. (2011, July 13). Girl power wins at Google’s first science
fair. New York Times. Retrieved from http://www.nytimes.
com/2011/07/19/science/19google.html. Accessed 26 August 2016.
Miyake, A., Kost-Smith, L., Finkelstein, N. D., Pollock, S. J., Cohen, G.
L., & Ito, T. A. (2010). Reducing the gender achievement gap in
college science: A classroom study of values affirmation. Science,
330,1234–1237. doi:10.1126/science.1195996.
National Center for O*NET Development. (2011). O*NET products at
work. Retrieved from https://www.onetcenter.org/dl_
files/paw/Products_at_Work.pdf. Accessed 26 August 2016.
National Science Foundation. (2016). Doctorate recipients by sex and
subfield. Retrieved from https://www.nsf.gov/statistics/2016
/nsf16300/data-tables.cfm. Accessed 26 August 2016.
National Student Clearinghouse Research Center. (2013). Snapshot re-
port –degree attainment. Retrieved from https://nscresearchcenter.
org/snapshotreport-degreeattainment3/. Accessed 26 August 2016.
Osipow, S. H., & Fitzgerald, L. F. (1996). Theories of career development
(4th ed). London: Allyn & Bacon.
Pomerantz, E. M., & Dong, W. (2006). Effects of mothers’perceptions of
children’s competence: The moderating role of mothers’theories of
competence. Developmental Psychology, 42,950–961. doi:10.1037
/0012-1649.42.5.950.
Riegle-Crumb, C., & King, B. (2010). Questioning a white male advan-
tage in STEM: Examining disparities in college major by gender and
race-ethnicity. Educational Researcher, 39,656–664. doi:10.3102
/0013189X10391657.
Robnett, R. D., & Leaper, C. (2013). Friendship groups, personal moti-
vation, and gender in relation to high school students’STEM career
interest. Journal of Research on Adolescence, 23,652–664.
doi:10.1111/jora.12013.
Sells, L. W. (1980). Mathematics: The invisible filter. Engineering
Education, 70,340–341.
Shapka, J. D., Domene, J. F., & Keating, D. P. (2006). Trajectories of
career aspirations through adolescence and young adulthood: Early
math achievement as a critical filter. Educational Research and
Evaluation, 12,347–358. doi:10.1080/13803610600765752.
Shih, M. J., Pittinsky, T. L., & Ho, G. C. (2012). Stereotype boost:
Positive outcomes from the activation of positive stereotypes. In
M. Inzlicht & T. Schmader (Eds.), Stereotype threat: Theory, pro-
cess, and application (pp. 141–156). New York: Oxford University
Press.
Sorby, S., Casey, B. M., Veurink, N., & Dulaney, A. (2013). The role of
spatial training in improving spatial and calculus performance in
engineering students. Learning and Individual Differences, 26,
20–29. doi:10.1016/j.lindif.2013.03.010.
Steele, C. (1997). A threat in the air: How stereotypes shape intellectual
identity and performance. American Psychologist, 52, 613–629.
doi:10.1037/0003-066X.52.6.613.
Steffens, J. C., Jelenec, P., & Noack, P. (2010). On the leaky math pipe-
line: Comparing implicit math-gender stereotypes and math with-
drawal in female and male children and adolescents. Journal of
Educational Psychology, 102,947–963. doi:10.1037/a0019920.
STEPS. (2016). STEPS: Study of Transitions and Educational Pathways.
Retrieved from http://www.stepsstudy.org/. Accessed 26 August
2016.
Su, R., Rounds, J., & Armstrong, P. I. (2009). Men and things, women
and people: A meta-analysis of sex differences in interests.
Psychological Bulletin, 135,859–884. doi:10.1037/a0017364.
U.S. Department of Labor Employment and Training Administration.
(1998). O*NET: The occupational information network.
Washington, DC: U.S. Government Printing Office.
United States Census Bureau. (2006). State and county quickfacts.
Retrieved from http://quickfacts.census.gov/qfd/states/55000.html.
Accessed 26 August 2016.
Updegraff, K. A., Eccles, J. S., Barber, B. L., & O’Brien, K. M. (1996).
Course enrollment as self-regulatory behavior: Who takes optional
high school math courses? Learning and Individual Differences, 8,
239–259. doi:10.1016/S1041-6080(96)90016-3.
Wai, J., Lubinksy, D., & Benbow, C. P. (2009). Spatial ability for
STEM domains: Aligning over 50 years of cumulative psy-
chological knowledge solidifies its importance. Journal of
Educational Psychology, 101,817–835. doi:10.1037
/a0016127.
Walton, G. M., & Cohen, G. L. (2003). Stereotype lift. Journal of
Experimental Social Psychology, 39, 456–467. doi:10.1016
/S0022-1031(03)00019-2.
270 Sex Roles (2017) 77:254–271
Watt,H.M.G.(2002).Gendered achievement-related choices and be-
haviours in mathematics and English: The nature and influence of
self-, task- and value perceptions (Unpublished PhD thesis).
University of Sydney, Sydney, NSW, Australia.
Watt, H. M. G. (2004). Development of adolescents’self-perceptions,
values and task perceptions according to gender and domain in 7th
through 11th-grade Australian students. Child Development, 75,
1556–1574. doi:10.1111/j.1467-8624.2004.00757.x.
Watt, H. M. G. (2006). The role of motivation in gendered educational
and occupational trajectories related to maths. Educational Research
and Evaluation, 12,305–322. doi:10.1080/13803610600765562.
Watt, H. M. G. (2008). What motivates females and males to pursue sex-
stereotyped careers? In H. M. G. Watt & J. S. Eccles (Eds.), Gender
and occupational outcomes: Longitudinal assessments of individu-
al, social, and cultural influences (pp. 87–113). Washington, DC:
American Psychological Association. doi:10.1037/11706-003.
Watt, H. M. G. (2016). Gender and motivation. In K. R. Wentzel & D. B.
Miele (Eds.), Handbook of motivation at school (2nd ed., pp. 320–
339). New York: Routledge.
Watt, H. M. G., Eccles, J. S., & Durik, A. M. (2006). The leaky mathe-
matics pipeline for girls: A motivational analysis of high school
enrolments in Australia and the USA. Equal Opportunities
International, 25(8), 642–659. doi:10.1108/02610150610719119.
Watt, H. M. G., Shapka, J. D., Morris, Z. A., Durik, A. M., Keating,
D. P., & Eccles, J. S. (2012). Gendered motivational processes
affecting high school mathematics participation, educational as-
pirations, and career plans: A comparison of samples from
Australia, Canada, and the United States. Developmental
Psychology, 48,1594–1611. doi:10.1037/a0027838.
Wigfield, A. (1993). Why should I learn this? Adolescents’achievement
values for different activities. In M. L. Maehr & P. R. Pintrich (Eds.),
Advances in motivation and achievement (Vol. 8, pp. 99–138).
Greenwich, CT: JAI Press.
Wigfield, A. (1994). Expectancy-value theory of achievement motiva-
tion: A developmental perspective. Educational Psychology
Review, 6,49–78. doi:10.1007/BF02209024.
Wigfield, A., & Eccles, J. S. (2000). Expectancy-value theory of achieve-
ment motivation. Contemporary Educational Psychology, 25,68–
81. doi:10.1006/ceps.1999.1015.
Sex Roles (2017) 77:254–271 271
A preview of this full-text is provided by Springer Nature.
Content available from Sex Roles
This content is subject to copyright. Terms and conditions apply.
