Content uploaded by Stephanie Masters
Author content
All content in this area was uploaded by Stephanie Masters on Apr 24, 2020
Content may be subject to copyright.
This journal uses Open Journal Systems 2.4.8.1, which is open
source journal management and publishing software developed,
supported, and freely distributed by the Public Knowledge Project
under the GNU General Public License.
Changes in Math and Science Interest over School
Transitions: Relations to Classroom Quality, Gender
Stereotypes, and Efficacy
Joan M Barth, Stephanie Masters
University of Alabama, USA
ABSTRACT
Significant declines in STEM ability beliefs and interest are often found during the
transitions to middle school and high school. Girls generally report lower self-
concept and interest in STEM compared to boys. Some children remain interested in
math and science over these transitions, but we know little about the school and
social factors that contribute to their continued interest and if these factors differ by
gender. This study examines changes in math and science self-efficacy and interest
over two school transitions and the final two years in high school. It further
examines if changes and gender discrepancies in math and science interest, can be
accounted for by self-efficacy, classroom qualities and the gender stereotypical
beliefs about the usefulness of math and science. Student in grades 5, 8, and 11
(N= 595) completed surveys on their math and science interest, self-efficacy,
stereotypes, and classroom quality prior to transitioning to the next grade, and
then one year post-transition. Although there were declines in interest and efficacy
over school transitions they were not as substantial or pervasive compared to
previous research. Gender differences were more apparent in high school than in
earlier grades. Regression analyses indicated that changes in interest over time
were explained by self-efficacy, classroom quality, and gender stereotype beliefs,
although gender stereotypes were only predictive of science interest
KEYWORDS
STEM interest; STEM Efficacy, STEM Instruction; Gender Stereotypes; School
Transitions
International Journal of Gender, Science and Technology, Vol.12, No.1
5
Changes in Math and Science Interest over School
Transitions: Relations to Classroom Quality, Gender
Stereotypes, and Efficacy
INTRODUCTION
Research on women’s underrepresentation in STEM (science, technology,
engineering, and mathematics) indicates that girls in high school achieve at about
the same level as their male counterparts in math and science (National Science
Board, 2018), but fail to continue on to college majors in these fields. Recruitment
of students into STEM undergraduate programs depends to a large degree on
maintaining interest in math and science over the school years prior to college.
Previous work indicates that college students in a range of STEM majors became
interested in STEM prior to sixth grade (Maltese, Melki, & Wiebke, 2014). Students’
interest in science is related to their career aspirations (for a review, see Regan &
DeWitt, 2015), suggesting that early school experiences may be important for
recruiting STEM majors.
Theoretical models such as expectancy value (Eccles, 1994, 2011) and social
cognitive theory (Bandura, Barbaranelli, Caprara, & Pastorelli, 2001) point to self-
perceptions of math and science abilities (e.g., self-concept and self-efficacy) as
being critical to continuing academic engagement in these fields. Although
expectancy value theory and social cognitive theory use different terminology, both
theories address the importance of competence-related beliefs on academic
motivation and outcomes (Hyde & Durik, 2005). In expectancy-value theory,
expectancy belief (e.g., ability self-concept; expectation for success) refers to
individuals’ evaluations of their competence in different tasks and how well they will
do on upcoming tasks. Self-efficacy beliefs refer to persons’ judgments of their
confidence to learn, perform or succeed in academic domains and are also strong
predictors of adaptation and change, as well as academic aspirations, level of
motivation, and resilience (Bandura et al., 2001). Because these constructs are
highly correlated (Wigfield & Eccles, 2000), we will use the term preferred by the
author in discussing a particular study and refer to both terms generically as
perceived abilities throughout the introduction. The measure used in the current
research more closely aligns with self-efficacy.
Unfortunately, females often rate themselves lower than males on self-report
measures of competence and interest in math and science (Baird & Keane, 2019;
Cunningham, Hoyer, & Sparks, 2015; Watt, 2004; Watt, Eccles, & Durik, 2006).
Declines in these and related motivational and social support factors over the
course of schooling are well documented, especially during the transitions to middle
school and high school (Eccles, Midgely, & Adler, 1984; Gutman, 2006; Isakson &
Jarvis, 1999; Ma & Wilms, 1999; Pintrich, 2000; Watt, 2004; Wilkins & Ma, 2003).
However, some research has found that global self-esteem increases for boys and
girls when the school configuration has only one transition (8th to 9th grade; Blyth,
Simmons, & Carlton-Ford, 1983). Using stage-environment fit theory (Eccles et al.,
1993), social cognitive theory (Bandura et al., 2001), and expectancy value theory
(Eccles, 2011) for theoretical grounding, this study examines changes in math and
science self-efficacy and interest over two school transitions and the final two years
International Journal of Gender, Science and Technology, Vol.12, No.1
6
of high school. The research design is cross-sectional between grade levels (5th,
8th, and 11th grades), but longitudinal over the transitions. It further examines
factors derived from these theories that might account for changes and gender
discrepancies in interest, including math and science self-efficacy, classroom
qualities, and gender stereotypical beliefs about the usefulness of math and
science. Most of the research cited on school transitions is based on data collected
over two decades ago when there were larger, more substantial gender
discrepancies in high school math and science course taking (Cunningham et al.,
2015), as well as in some STEM college majors (e.g., Biology; National Science
Foundation [NSF], 2017). Thus, this study serves to update the trends observed in
earlier studies.
BACKGROUND
School Transitions
The Eccles et al. (1993) stage-environment fit model proposes that school
transitions negatively affect academic outcomes because there is an inconsistency
between children’s developmental needs and post-elementary school environments
(Eccles et al., 1993; Gutman & Eccles, 2007). Each school transition marks a
significant decline in teacher support, a change to a more peer-competitive school
setting, and a change in the academic context of math and science course-taking
(e.g., required vs. elective courses; regular vs. advanced classes). Together these
factors may account for some of the declines in student self-concept and attitudes.
Specifically, in contrast to elementary schools, middle schools are characterized as
having less supportive teachers, more ability-grouped classes, and more
competition (Eccles et al., 1993). Prior to the transition to middle school, students
have generally all received the same instruction in math and science. Both middle
and high school students have more specialized experiences in science and math
based on abilities and interests. Compared to middle school, the high school setting
is even less personal, more competitive, and more grade oriented. The last two
years of high school mark a significant period of preparation for post-secondary
education and careers as students select courses based on post-high school plans.
At this level, students are more likely to be taking advanced level math or science
classes beyond the required courses for the high school diploma (e.g., physics, pre-
calculus, and calculus). Thus, ability differences are even more emphasized as the
differences between college bound and other students are more salient (Gutman,
2006).
Several studies have identified self-concept, academic self-efficacy, and attitudes as
factors that might explain why some students’ interest and academic performance
are higher than others’ during school transitions (Bandura et al., 2001; Wigfield,
Eccles, MacIver, Reuman, & Midgley, 1991; Wilkins & Ma, 2003). For example, Ma
and Wilms (1999) found that adolescents drop from advanced math courses during
two transitions: from eighth grade to high school and from eleventh to twelfth
grade. During the transition from eighth to ninth grade, dropping out was attributed
primarily to prior achievement, while dropping out over the transition from eleventh
to twelfth grade was attributed to negative attitudes toward math. Gender
differences in changes in these factors are evident across school transitions, as girls
manifest a steeper decline in math interest over the course of adolescence
International Journal of Gender, Science and Technology, Vol.12, No.1
7
compared to boys (Fredricks & Eccles, 2002).
Prior work indicates that perceived competence better predicts academic pursuits
and occupational interest than previous achievement (Bandura et al., 2001; Brown
et al., 2008; Ferry, Fouad, & Smith, 2000; Frome & Eccles, 1998). Empirical work
also reveals that boys have higher general self-esteem, compared to girls, both
before and after school transitions (Blyth et al., 1983; Booth & Gerard, 2014).
However, the timing of children’s first school transition (6th to 7th grade vs. 8th to 9th
grade) and the number of transitions over the middle and high school years may
make a difference in whether girls’ self-esteem declines (earlier grade transition) or
increases (later grade transition; Blyth et al., 1983). Gender differences in math
self-efficacy have been found to mediate gender differences in the decisions to
enter some STEM fields (Correll, 2001; Parker, Nagy, Trautwein, & Ludtke, 2014).
Based on prior research, it is expected that academic self-efficacy and interest will
decline at each transition. In addition, it is predicted that there will be gender
differences in interest and efficacy favoring males. However, expectancy value
theory, suggests that these outcomes are affected by other factors besides gender,
including classroom quality and socialized gendered beliefs about math and science
(Eccles, 2011).
Classroom Quality
Stage-environment fit theory (Eccles et al., 1993) proposes that the new classroom
environment after a school transition accounts for general declines in students’
motivation, and more specifically, that such experiences influence students’
perceived abilities and interests. The classroom environment involves the personal
relationships between students and teachers as noted above (less personal and less
positive interactions and greater emphasis on control) and teachers’ instructional
style. Positive student-teacher relationships have immediate (Marchant, Paulson, &
Rothlisberg, 2001; Wang & Eccles, 2013) and long-term impacts on academic self-
concept (Murdock, Anderman, & Hodge, 2000). Teachers who support and
encourage students to achieve in math and science positively affect children’s
attitudes and interest toward math and science coursework (George, 2003),
promote higher levels of expectation for academic success (Goodenow, 1993;
Midgley, Feldlaufer, & Eccles, 1989), and more positive academic self-efficacy
(Ryan & Patrick, 2001).
Positive instructional methods include opportunities to engage in achievement-
related activities, which in turn provide students with information about their
competence and interest in academic subjects that eventually lead to the
development of ability self-concepts (Wang & Eccles, 2013). Teaching for relevance,
or meaningful instruction, refers to the extent to which class instruction is related
to students’ personal interests and goals (Wang & Eccles, 2013). For example,
Wang (2012) found that teaching for meaning and promoting cooperative learning
significantly predicted students’ subjective task values for math (e.g., interest,
usefulness), which in turn predicted the number of mathematics classes taken and
mathematics-related career plans. Indeed, there is evidence to suggest that
meaningful instruction can increase academic interests (National Research Council,
International Journal of Gender, Science and Technology, Vol.12, No.1
8
2004) and lead to stronger beliefs in one’s academic abilities (Stipek, 2001). For
example, students report greater motivation, interest, and future orientation
towards science when exposed to interactive, hands-on activities, and science
applications in classrooms (Hampden-Thomson & Bennett, 2013). This is consistent
with recommendations from the National Research Council’s framework for K-12
Science ducation and science instruction (2012; also see 2004), which include an
emphasis on students practicing science (e.g., participating in experiments and
demonstrations). The National Research Council also notes the importance of
students developing an awareness of science careers and the stories of individual
scientists to promote science identity.
Together, a number of studies suggest that the optimal classroom environment
creates a comfort zone with classmates and teachers, but also provides a degree of
challenge. Based on this research, we propose that classroom quality (teacher
support and meaningful instruction) may partially explain changes in students’
interest in science and math after a school transition. Consistent with expectancy
value theory, we further propose that gender stereotypical beliefs of the importance
of math and science also may account for declines in interest, especially for girls.
Gender Stereotypes
The expectancy value model of academic achievement formally proposes that
children’s perceptions of gender stereotypes should impact children’s ability beliefs
(expectations for success) and subjective task values (interest, utility value,
attainment value, and relative costs), which in turn impact academic choices and
behaviors (Eccles, 1994; 2011; Lane, Goh, & Driver-Linn, 2012). More traditional
gender stereotypical beliefs in childhood are associated with more gender
stereotypical vocations in early adulthood (Lawson, Lee, Crouter, & McHale, 2018).
Gender stereotype knowledge related to academic subjects and occupations
increases over schooling and may adversely affect both girls’ and boys’ interest in
math and science (Barth, Kim, Eno, & Guadgno, 2018; Chatard, Guimond, &
Selimbegovic, 2007; Evans, Copping, Rowly, & Kurtz-Costes, 2010; Katz &
Ksansnak, 1994; Kurtz-Costes, Rowley, Harris-Britt, & Woods, 2008; Wigfield &
Eccles, 2002). Blažev, Karabegović, Burušić, & Selimbegović (2017) found that
primary school students with stereotype-consistent interests are more prone to hold
stereotypical beliefs than those who have less gender-stereotypical interests in
school subjects. Similarly, female college students who endorse math gender
stereotypes have more negative self-perceptions related to their math abilities than
women who reject these stereotypes (Schmader, Johns, & Barquissau, 2004).
Additionally, there may be gender differences in the effect of stereotypes on STEM
attitudes. For instance, Nosek, Banaji, and Greenwald (2002) observed that strong
gender stereotypes were correlated with negative math attitudes for women, while
the opposite was true for men. Further, stereotypes that boys are more suited for
math or science than girls may influence girls’ motivation and interest in math and
science (Cvencek, Meltzoff, & Greenwald, 2011; Master, Cheryan, & Meltzoff, 2016;
Steffens, Jelenec, & Noack, 2010). However, Barth et al. (2018) found that holding
traditional STEM occupation gender stereotypes was related to lower STEM career
interests for both boys and girls.
International Journal of Gender, Science and Technology, Vol.12, No.1
9
There are conflicting views as to whether gender stereotypes have a stronger
influence on older or younger children (e.g., Gottfredson, 1981; Gottfredson &
Lapan, 1997 vs. Garrett, Ein, & Tremaine, 1977; Katz & Ksansnak, 1994). Cross-
sectional research comparing gender stereotypes in younger and older children
suggests that younger children do not endorse traditional gender stereotypes
because they tend to hold an “own gender” bias when comparing boys and girls
(Kurtz-Costes et al., 2008). However, other evidence suggests that children as
young as seven years of age endorse traditional gender stereotypes (Cvencek et
al., 2011).
In a recent study, Barth et al. (2018) found that across a wide range of ages from
elementary school to college, ability beliefs, whether for oneself or others, were
strong predictors of career interests, and that stereotypical beliefs about
occupations played a secondary role, but still a significant one. Girls and boys
appeared to become less stereotypical in their own STEM career interests and
efficacy over schooling, but the expectation that others would hold gender
stereotypical career interests did not change accordingly over the same time
period. To extend upon this work, we focus on the gender stereotyped belief that
math and science are more useful for males than females. This is consistent with
the Eccles et al. (2011) hypothesis that gender stereotypes are related to
subjective task values, such as the utility value.
Current Study
This study examines factors that affect change in children’s interest in math and
science over three transition periods: from fifth grade to middle school, from eighth
grade to high school, and from eleventh to twelfth grade in high school. The
research design is cross-sectional between grade levels (5th, 8th, and 11th grades),
but longitudinal over the transitions. An important contribution of this study is the
consideration of both math and science, as most previous research has primarily
included only math. There are two primary objectives. First, changes in interest and
self-efficacy for math and science over the three transition periods are examined,
partially replicating previous work (Ma & Wilms, 1999). Although the transition to
middle school and junior high has been extensively studied, there is less research
on the transition to high school and the last two years of high school that focuses
on math and science. Consequently, this objective seeks to validate and extend
previously reported declines in interest and efficacy over different transition
periods, as well as previously reported gender differences before and after the
transitions. It is expected that interest and efficacy will decline over each transition
and be greater for younger students compared to older students (i.e., fifth graders
> middle school students > high school students). Additionally, based on past
research, we expect that girls’ interest will drop more than boys’ over transitions.
The second objective examines if self-efficacy, perceived classroom quality, and
math/science usefulness stereotypes can account for changes in interest over
transitions, especially for gender differences in change in interest. Theorizing and
research based on expectancy value and social cognitive theory (Bandura et al.,
2001; Eccles, 2011) led to the prediction that self-efficacy will be the strongest
predictor of interest in math and science, and this study will assess if classroom
International Journal of Gender, Science and Technology, Vol.12, No.1
10
quality and gender stereotypes add explanatory power. Thus, we examine if
changes in interest over school transitions are accounted for by self-efficacy,
classroom quality, and gender stereotypes about the usefulness of math and
science after the school transition. Although grade level changes are examined in
the analyses, it is expected that these factors are similarly important across
schooling.
METHOD
Participants
Participants were recruited from fifth, eighth, and eleventh grade classrooms in
nine U.S. public schools (three each for elementary, middle, and high school), all of
which were part of the same county school system that included both urban and
non-urban schools. The nine schools from which the students were recruited were
predominantly non-Hispanic White (school average of 72%, range 42% to 94%),
but had a significant percentage of Black students (school average 24%, range 4%
to 51%). The average free/reduced lunch rate was 45% (range 27% to 73%). The
elementary, middle, and high schools were vertically aligned (i.e., in the same
school zone within the school system), such that the elementary school children
were expected to transition into one of the participating middle schools, and the
middle school students were expected to transition into one of the participating high
schools. The vertical alignment made it unlikely that differences between grade
levels were due to factors related to school demographic characteristics (e.g., SES)
since students were drawn from the same school zones. In middle and elementary
schools, we were able to recruit from all students at a particular grade level. At the
high school level, we recruited from math and science courses for 11th graders.
In the last half of the spring semester, parents received a letter that explained the
purpose of the project and asked them to return a consent form to school if they
were interested in allowing their child to participate. Parents were told that a $5
donation would be made to their school for each returned consent form and that
there was an opportunity for children to participate in a second career survey (not
included in this research report) and earn $15 (fifth graders) or $20 (eighth graders
and high school students). The initial response rate was 47% of the 1511 potential
students with 704 students completing the survey. This included 290 fifth graders,
207 eighth graders, and 207 high school students. Fifty-six percent of the students
were female, 74% were Caucasian (0.8% Hispanic), 22% Black, and the remaining
4% were Asian, mixed race, other, or did not specify a race.
Approximately one year later, schools and students were re-contacted and surveys
were re-administered for the post-transition time point. The retention rate was 84%
at the second time point (N = 595 students). The primary factor affecting retention
was that students had moved out of the school district or were absent on the days
that the survey was administered. This group of students was 55% female and had
a racial make-up similar to the original sample: 75% Caucasian (1% Hispanic),
22% Black, and the remaining 3% were all other races or unspecified. Comparisons
between students who participated at both time points to those who only
participated at the pre-transition time point on the pre-transition measures
described below revealed three significant differences. Compared to those who
International Journal of Gender, Science and Technology, Vol.12, No.1
11
dropped out after the first time point, those who were retained rated themselves
higher in math Efficacy (Mdroped = 3.66, SDdropped = 0.87; Mretained = 3.99, SD retained =
0.72, p < .001), science Efficacy (Mdropped = 3.83, SDdropped = 0.68; Mretained = 3.99,
SDretained = 0.68, p = .027), and math Interest (Mdropped = 3.66, SDdropped = 0.81;
Mretained = 4.03, SDretained = 0.73, p < .001). Despite this difference, the pattern of
relations among the measures between the two time points was very similar. (See
correlations in Table 2.)
Procedure and Measures
At each time point, students completed the 172 item Math, Science, and
Technology (MST) questionnaire in a group setting at their schools during regular
school hours. Assent information and directions were read aloud by the research
staff, and then students were allowed to proceed at their own pace. Research staff
was available to answer questions and provide assistance if necessary. Younger
children averaged 20 to 30 minutes to complete the questionnaire, and older
students were able to finish in 15 to 30 minutes.
The MST assessed students’ efficacy, interest, classroom experiences, attitudes,
goals, and gender stereotypes in the areas of math, science, and technology. We
report on four sets of measures from this battery that focus on math and science.
Items on the scales were primarily taken from previously published measures.
Some items were edited slightly to increase clarity for the younger children in the
study and/or make the response format consistent with a 5-point scale. It should
be noted that sample sizes vary slightly for different measures due to some
participants not fully completing the questionnaire. Unless otherwise specified, the
rating scales for Self-Efficacy, Interest, Classroom Quality and Usefulness
Stereotype were strongly disagree = 1 to strongly agree = 5. All scale scores were
calculated as averages over responses to items so that the range of scores for all
measures was 1 to 5. Participants had to have answered 75% of the items on a
scale to receive a score for that scale. Table 1 presents descriptive statistics for
each measure pre- and post-transition by gender and grade.
Measures for Self-Efficacy, Interest, and Usefulness Stereotypes were based on
measures used in the Michigan Study of Adolescent and Adult Life Transitions
(MSALT; http://garp.education.uci.edu/msalt.html).
Math and Science Self-Efficacy. Questions were adapted from MSALT and included
items related to performance in school (e.g., self-ranking of ability from 1 = near
the bottom to 5 = near the top, performance in math and science in comparison to
other subjects from 1 = much worse to 5 = much better), and items related to the
ability to learn math or science (“When taking a math/science test I’ve studied for,
I do very well;” “I could learn to do any type of math/science problem if I wanted
to”). Higher scores indicated better self-efficacy. Internal consistency for these four
Efficacy scales (Math pre- and post-transition; science pre- and post-transition) was
acceptable at each time point (Cronbach’s alpha range = .68 – .74, Median = .72).
International Journal of Gender, Science and Technology, Vol.12, No.1
12
Table 1
Descriptive Statistics for Measures
5th Grade
8th Grade
11th Grade
Field
Male
Female
Male
Female
Male
Female
M
SD
M
SD
M
SD
M
SD
M
SD
M
SD
Math
Useful Stereo.
Pre
2.49
1.26
1.97
1.23
2.14
1.17
1.76
0.97
2.21
1.27
1.81
1.17
Post
2.47
1.22
1.96
1.25
2.42
1.02
1.85
1.04
2.16
1.11
1.56
0.99
Class Quality
Pre
3.70
0.62
3.79
0.59
3.63
0.85
3.56
0.82
3.42
0.81
3.29
0.88
Post
3.37
0.94
3.70
0.81
3.42
0.70
3.49
0.90
3.61
0.77
3.61
0.65
Efficacy
Pre
3.96
0.70
3.94
0.65
3.97
0.77
3.88
0.71
4.08
0.87
3.88
0.89
Post
3.89
0.78
3.99
0.65
4.03
0.89
3.95
0.79
4.18
0.61
4.01
0.73
Interest
Pre
4.06
0.61
4.08
0.69
4.02
0.70
3.86
0.74
3.99
0.85
3.80
0.93
Post
3.79
0.78
3.96
0.74
3.96
0.75
3.86
0.74
4.11
0.67
3.88
0.85
Science
Useful Stereo.
Pre
2.42
1.24
2.03
1.26
2.27
1.18
1.71
1.03
2.40
1.33
1.79
1.14
Post
2.48
1.33
2.04
1.28
2.33
1.12
1.92
1.18
2.12
1.09
1.62
1.06
Class Quality
Pre
3.88
0.62
3.96
0.57
3.63
0.85
3.66
0.81
3.77
0.78
3.80
0.75
Post
3.55
1.02
3.87
0.76
3.71
0.69
3.55
0.86
3.98
0.69
4.02
0.63
Efficacy
Pre
3.97
0.69
3.96
0.68
3.98
0.72
3.91
0.72
4.14
0.60
3.90
0.65
Post
3.89
0.89
3.84
0.72
4.04
0.67
3.76
0.76
4.21
0.57
3.90
0.71
Interest
Pre
4.03
0.76
3.88
0.78
3.88
0.74
3.71
0.79
3.95
0.81
3.89
0.81
Post
3.75
0.85
3.76
0.83
3.84
0.79
3.75
0.82
4.15
0.68
3.84
0.82
N range
118-142
132-147
79-93
87-114
58-67
101-118
International Journal of Gender, Science and Technology, Vol.12, No.1
13
Math and Science Interest. Two separate scales were created for math and science,
one for each time point (four measures altogether). Items were adapted from the
MSALT. The scales combined five items related to attitudes (liking, interest in taking
more math or science) and the perceived importance and benefit of math or science
for the future. Higher scores indicated greater interest. The scales had acceptable
internal consistency (Cronbach’s alpha range = .70 – .77, Median = .72).
Student Perceptions of Classroom Quality. Two separate scales were created for
each time point, one for science and one for math (four scales altogether). Items
on the scales included assessments of teacher support or “push” (e.g., teacher
expectations for students to work hard, teacher encouragement in math and
science), and teacher-student relations (e.g., the teacher cares how students feel)
adapted from Midgley et al. (1989) and Wilkins and Ma (2003). Student experience
with different classroom activities associated with greater learning and interest in
science and math were also assessed by investigator–developed questions related
to teaching for relevance (e.g., use of hands-on activities, use of real world
examples, including information on careers) that were derived from
recommendations from various educational guides, such as the National Research
Council, 2004, 2012). These items were averaged to form Classroom Quality scales
for math (12 items) and science (17 items) such that 1 = low Classroom Quality
and 5 = high Classroom Quality. Internal consistency for these scales was high
(Cronbach’s alpha range = .85 – .91, Median = .87).
Math and Science Usefulness Stereotype. These questions were adapted from the
MSALT question “Who finds math more useful” (1 = women much more useful to 5
= men much more useful) to fit with the agreement response scale used for the
other items: “Math [science] is more useful for boys than girls” (1 = strongly
disagree to 5 = strongly agree). Higher scores on Usefulness Stereotype indicated
more stereotypical beliefs about math and science, specifically, that they are more
useful for boys than girls. These measures were assessed pre- and post-transition.
RESULTS
First, grade related changes in Efficacy and Interest in math and science were
examined to see how this sample compared with previously reported changes in
these constructs over school transitions and gender differences. Next, we examined
if Efficacy, Classroom Quality, and Usefulness Stereotypes could account for
Interest in math and science prior to the transition (including gender differences
therein) and then examined if these factors could account for change in Interest
over the transition. Table 1 presents descriptive statistics for all measures. It
should be noted that although participants are nested within schools, the small
number of schools (N = 9) does not meet the threshold recommended for
hierarchical or multi-level modeling (Maas & Hox, 2005; O’Dwyer & Parker, 2014;
Scherbaum & Ferreter, 2009).
Transition-Related Changes in Efficacy and Interest
Two repeated measures MANOVAs were conducted, one each for Interest and
Efficacy. The statistical design was a multi-level factorial design: 2(Gender) x
International Journal of Gender, Science and Technology, Vol.12, No.1
14
3(Starting Grade: 5th, 8th, or 11th) x 2(Transition: before or after) x 2(Field: math or
science). A Bonferroni correction was used for all follow-up comparisons.
Analyses examining change in Interest over time revealed three significant main
effects and two interactions. The main effect for Gender, F(1, 568) = 5.36, p =
.005,
p2 = .01, revealed that males reported greater Interest (for math and science
combined) than females, Mboys = 3.99, SEboys = .035; Mgirls = 3.88, SEgirls = .03. The
significant interaction between Gender and Grade, F(2, 568) = 2.97, p < .05,
p2 =
.01 sheds further light on this effect. As Figure 1 illustrates, follow-up comparisons
indicated that the gender difference was only significant for the oldest students.
There was a significant effect for Transition, F(1, 568) = 7.97, p = .005,
p2 = .01,
indicating that Interest scores (for math and science combined) declined from pre-
to post-transition, Mpre = 3.98, SEpre = .03; Mpost = 3.90, SEpost = .03. This effect
should be interpreted in the context of the significant Transition x Grade
interaction, F(1, 568) = 6.72, p = .001,
p2 = .023. For the transition to middle
school (5th to 6th grade) there was a significant decline in Interest, Mpre = 4.03, SEpre
= .04; Mpost = 3.83, SEpost = .04, p < .001; however for the other two grade levels,
the change from pre- to post-transition did not reach significance, for 8th to 9th
grade, Mpre = 3.94, SEpre = .05; Mpost = 3.86, SEpost = .05, p = .150; for 11th to 12th
grade, Mpre = 3.96, SEpre = .05; Mpost = 4.00, SEpost = .05, p = .47.
Finally, the main effect for Field was also significant, F(1, 568) = 8.86, p = .003,
p2 = .02, indicating that interest was higher for math than science, Mmath = 3.99,
SEmath = .03; Mscience = 3.89, SEscience = .03.
The results for Efficacy indicated a significant effect for Gender, F(1, 556) = 9.90, p
< .002,
p2 = .02, revealing that boys reported higher levels of Efficacy than girls,
Mboys = 4.07, SEboys = .03; Mgirls = 3.93, SEgirls = .03. Although only marginally
3.92 3.96
4.09*
3.95
3.84 3.87
3.50
3.70
3.90
4.10
4.30
5th-6th 8th-9th 11th-12th
Interest
Grade
Male
Female
Figure 1. Interest: Gender x Grade Interaction. Interest scores are for math and science
combined. Scores are also averaged between the pre- and post-transition scores for each of
the three grade levels. *p = .01
International Journal of Gender, Science and Technology, Vol.12, No.1
15
significant, F(1, 556) = 2.86, p = .058,
p2 = .01, the Gender x Grade interaction
effect is illustrated in Figure 2 because this interaction effect was significant for
Interest and the pattern over grade levels was similar. As indicated in Figure 2,
gender differences were not significant for the younger students, but were
significant in older grade levels. Additionally, the three-way interaction between
Transition, Field, and Gender was significant, F(1, 556) = 5.14, p = .024,
p2 = .01.
Follow-up comparisons indicated that at pre-transition gender differences were
evident for both fields, although only marginally significant for science (for science
p = .067, Mboys = 4.07, SEboys = .05; Mgirls = 3.96, SEgirls = .04 vs. for math p =
.015, Mboys = 4.10, SEboys = .05; Mgirls = 3.95, SEgirls = .04). However, post-
transition, gender differences were only evident for science, p = .001, Mboys = 4.05,
SEboys = .05; Mgirls = 3.83, SEgirls = .04. Additional comparisons indicated that only
girls had a significant drop in science efficacy over the transition, p = .005, Mpre =
3.96, SEpre = .04; Mpost = 3.83, SEpost = .04.
To summarize, Interest in math and science decreased in the transition from 5th to
6th grade, but not in other grades. Overall, boys reported higher Interest in math
and science than girls, but this difference was only significant for high school
students. Both boys and girls had greater Interest in math than science. For
Efficacy, gender differences favoring boys were more prominent in upper grades.
Boys reported higher math and science Efficacy than girls prior to the transition,
but after the transition the gender difference was only evident for science. Girls’
science Efficacy declined over the transition, but this pattern was not evident for
boys.
Predicting Interest from Efficacy, Classroom Quality, and Usefulness
Stereotype
3.95
4.06m
4.19*
3.94 3.92 3.93
3.50
3.70
3.90
4.10
4.30
5th-6th 8th-9th 11th-12th
Efficacy
Grade
Male
Female
Figure 2. Efficacy: Gender x Grade Interaction. Efficacy scores are for math and
science combined. Scores are also averaged between the pre- and post-transition
scores for each of the three grade levels.
mp= .070; *p = .002
International Journal of Gender, Science and Technology, Vol.12, No.1
16
As a preliminary step, math and science Interest scores were correlated with
comparable subject-specific Efficacy, Classroom Quality, and Usefulness Stereotype
scores (Table 2). All correlations were significant or marginally significant and in the
expected direction. It should be noted that negative correlations with Usefulness
Stereotype indicate that more stereotypical beliefs are associated with less Interest.
Correlations calculated separately for each gender generally showed the same
pattern of correlations. Specifically, pre- and post-transition Efficacy and Classroom
Quality were positively correlated with Math and Science Interest. Usefulness
Stereotype was negatively correlated with Math and Science Interest for both boys
and girls and similar to the results presented in Table 2, were weaker and at times
not significant for the correlation between Math Interest and Usefulness Stereotype.
Table 2
Correlations between Interest and Efficacy, Classroom Quality, and Stereotypes
Math Interest
Science Interest
Transition/
Pre
Post
Pre
Post
Measure
N = 671-677
N = 571-580
N = 674-678
N = 565-575
Pre
Efficacy
.69***
.47***
.62***
.36***
Classroom Quality
.45***
.27***
.34***
.25***
Usefulness Stereotype
-.09*
-.07m
-.17***
-.09*
Post
Efficacy
.68***
.66***
Classroom Quality
.40***
.48***
Usefulness Stereotype
-.07m
-.17***
Note. Higher scores on Efficacy and Classroom Quality are associated with better
ratings on these constructs. Higher scores on Usefulness Stereotype indicate more
traditional stereotypical beliefs.
m p < .10; * p < .05; **p < .01; ***p < .001
Next, regressions were calculated to examine the additive predictive effects of
Efficacy, Classroom Quality, and Usefulness Stereotype for math and science
Interest pre-transition. These analyses lay the foundation for the second set of
regressions which examined how these factors predicted change in Interest. It
should be noted that the second objective included an accounting of gender
differences in change in Interest over the school transitions. However, the ANOVA
results for Interest did not reveal a significant Gender X Transition interaction,
suggesting that males and females showed similar changes over time. As a result,
this part of the second objective was not considered in the analyses.
Predicting pre-transition Interest. A step-wise approach was used to assess the
impact of three sets of variables: 1) demographics, consisting of Gender (0 =
International Journal of Gender, Science and Technology, Vol.12, No.1
17
female, 1 = male), Grade, and the Gender x Grade interaction; 2) Efficacy; and 3)
Classroom Quality and Usefulness Stereotypes. If the third step of the regression
were significant, then that would suggest that Classroom Quality and Stereotypes
make unique contributions to explaining Interest above that of Efficacy. Table 3
presents the results of these regressions. In the second step of the model, Efficacy
significantly increased the amount of variance explained above Gender and Grade
for both math and science Interest. Classroom Quality and Usefulness Stereotypes
were added in the third step of the model and resulted in a significant increase in
variance explained for both math and science Interest. (The negative coefficient for
Usefulness Stereotype indicates that more traditional beliefs are associated with
less interest.) In the final models for both math and science Interest, Efficacy was
the strongest predictor, followed by Classroom Quality and Usefulness Stereotype.
For Science, in the final model, Gender was also a significant predictor, indicating
that boys had higher interest ratings than girls, as previously reported. (It should
be noted that the removal of non-significant Gender x Grade interaction effect from
the regression for Math Interest yields F(5, 667) = 147.09, p < .001, R2 = .53.)
Table 3
Regressions Predicting Pre-transition Interest from Demographics, Efficacy,
Classroom Quality, and Stereotypes
Math
Science
Step
Predictor
Beta
R2 Change
Beta
R2 Change
1
Gender
-.07
.02**
.14
.01
Grade
-.15**
.002
Gender x Grade
.14
-.06
2
Gender
-.03
.47***
.18*
.38***
Grade
-.13***
.02
Gender x Grade
.05
-.15*
Pre-Tran. Efficacy
.69***
.62***
3
Gender
.01
.04***
.20**
.04***
Grade
-.08*
.03
Gender x Grade
.03
-.14m
Pre-Tran Efficacy
.61***
.55***
Pre-Tran. Classroom
Quality
.20***
.18***
Pre-Tran Stereotype
-.06*
-.09**
Full Model
F(6, 666) = 122.45***;
R2 = .53
F(6, 666) = 82.68***;
R2 = .43
Note. Negative coefficients for Usefulness Stereotype indicate that more
traditional stereotypical beliefs are associated with lower Interest.
m p < .10; * p < .05; **p < .01; ***p < .001
To summarize, these findings are consistent with the hypothesis that Efficacy,
Classroom Quality, and Usefulness Stereotype would each predict Interest and that
International Journal of Gender, Science and Technology, Vol.12, No.1
18
Efficacy would have a stronger relationship to Interest than either Classroom
Quality or Usefulness Stereotype. Gender differences were still evident for science
Interest, even after these variables were considered. Otherwise, there were no
significant gender or grade level effects. It should be noted that including additional
grade and gender interaction terms with Efficacy, Classroom Quality, and
Usefulness Stereotype did not yield significant effects. For the sake of parsimony,
these effects were not presented.
Predicting change in Interest. The second set of analyses examined if Gender,
Grade, Efficacy, Classroom Quality, and Usefulness Stereotypes post-transition,
explained changes in interest over time. In these regression analyses post-
transition Efficacy, Classroom Quality, and Usefulness Stereotype measures were
used as predictors of change in Interest since stage-environment theory proposes
that declines in Interest are due to the new school environment. Before considering
the issue of mediation, regressions were first calculated following the same step-
wise procedure described previously, except that pre-transition Interest was
entered on the first step, followed by a) Grade and Gender, b) Efficacy, and then c)
Classroom Quality and Usefulness Stereotype to evaluate if each set of predictors
contributed to the variance explained, regardless of a mediational role. Significant
effects after the first step can be interpreted as accounting for change in interest.
The results in Table 4 indicate that each step of the model produced a significant
change in R2. Efficacy (step 3) produced the greatest increase in variance,
explaining 25% for math and 27% for science. Classroom Quality and Usefulness
Stereotype contributed smaller, although significant, explanatory power (2% for
math and 5% for science); however, Usefulness Stereotype was only significant for
science Interest. (The negative coefficient for Usefulness Stereotype indicates that
more traditional beliefs are associated with less interest.) It is important to note
that Gender and Grade effects were not significant in the final models. For this
reason, Gender and Grade were not included in the mediation analysis. (It should
be noted that removing the non-significant Gender x Grade interaction term yields
F(6, 554) = 107.94, p < .001, R2= .54 for Math Interest and F(6, 548) = 100.74, p
< .001, R2 = .52 for science Interest.)
International Journal of Gender, Science and Technology, Vol.12, No.1
19
Table 4
Regressions Predicting Change in Interest from Pre-Transition Interest
Demographics, Post-Transition Efficacy, Post-Transition Classroom Quality, and
Post-Transition Stereotypes
Math
Science
Step
Predictor
Beta
R2 Change
Beta
R2 Change
1
Pre-Tran. Interest
.51
.25***
.43***
.18***
2
Pre-Tran. Interest
.51***
.01**
.42***
.02**
Gender
-.21*
-.14
Grade
.02
.06
Gender x Grade
.19*
.18m
3
Pre-Tran. Interest
.27***
.25***
.21***
.27***
Gender
-.15*
-.11
Grade
-.02
.03
Gender x Grade
.13m
.10
Post-Tran. Efficacy
.55***
.57***
4
Pre-Tran. Interest
.26***
.02***
.19***
.05***
Gender
-.11
-.06
Grade
-.02
.01
Gender x Grade
.12
.08
Post-Tran. Efficacy
.50***
.46***
Post-Tran Classroom
Quality
.16***
.24***
Post-Tran. Stereotype
-.04
-.09**
Full Model
F(7, 553) = 92.44***;
R2 = .54
F(7, 547) = 86.55***;
R2 = .53
Note. Negative coefficients for the Usefulness Stereotype indicate that more
traditional stereotypical beliefs are associated with lower Interest.
mp < .10; *p < .05; **p < .01; ***p < .001
The next set of analyses examined whether post-transition Efficacy, Classroom
Quality, and Usefulness Stereotype accounted for the change in math and science
Interest over the transition. Using the Process Macro, a parallel mediation analysis
was conducted (Hayes, 2018). A parallel mediation analysis estimates the total and
specific effects of each mediator on the relation between pre-transition interest and
post-transition interest. As illustrated in Figures 3 and 4, the indirect paths are
indicated by the paths from pre-Interest to each of the proposed mediators and
from each mediator to post-Interest. The direct effects of pre-Interest on post-
Interest after accounting for Efficacy, Classroom Quality, and Usefulness Stereotype
appear above the arrow connecting the two; whereas the total direct effect is
presented below the arrow.
International Journal of Gender, Science and Technology, Vol.12, No.1
20
The total effect and direct effect of pre-transition Interest on post-transition
Interest was significant for both math and science. Pre-transition Interest in science
had a significant indirect effect on post-transition Interest in science through
Efficacy ( = .18, SE = .02, 95% CI [.14 to .22]), Classroom Quality ( = .05, SE =
.01, 95% CI [.02 to .07]), and Usefulness Stereotype ( = .01, SE = .01, 95% CI
[.001 to .02]). While all were significant, planned contrasts revealed that Efficacy
was a stronger mediator than Classroom Quality ( = .13, SE = .02, 95% CI [.08 to
.18]) and Usefulness Stereotype ( = .17, SE = .02, 95% CI [.13 to .21]).
Contrasts also revealed that Classroom Quality was a stronger mediator than
Usefulness Stereotype ( = .04, SE = .01, 95% CI [01 to .06]). Pre-transition
Interest in math had a significant indirect effect on post-transition Interest in math
through Efficacy ( = .22, SE = .02, 95% CI [.17 to .26]) and Classroom Quality
( = .03, SE = .01, 95% CI [.02 to .06]), but not Usefulness Stereotype ( = .001,
SE = .002, 95% CI [-.004 to .01]). Contrasts revealed that Efficacy was a stronger
mediator than Classroom Quality ( = .18, SE = .03, 95% CI [.13 to .23]). For both
mediation models, pre-transition Interest still had a significant direct effect on post-
Efficacy
Classroom
Quality
Usefulness
Stereotype
Post-Transition
Science Interest
Pre-Transition
Science Interest
.21 [.14 to .27]***
.46 [.37 to .54]***
Figure 3. Accounting for change in science Interest with post-transition Efficacy,
Classroom Quality, and Usefulness Stereotype. Results are for all grade levels and
genders combined. For the path between pre- and post-transition Interest, the
coefficient above the line is the remaining direct effect after taking into account the
three other indirect paths. The coefficient below the line is the total effect of pre-
transition Interest on post-transition Interest. 95% CIs are in brackets. Negative
coefficients for Usefulness Stereotype indicate more traditional stereotypical beliefs are
associated with lower Interest.
*p < .05; **p < .01; ***p < .001.
Post-Transition
International Journal of Gender, Science and Technology, Vol.12, No.1
21
transition Interest after accounting for the effects of Efficacy, Classroom Quality,
and Usefulness Stereotype (Figures 3 and 4), indicating that only partial mediation
was achieved.
DISCUSSION
This study examined factors that affect change in children’s interest in math and
science during key educational transitions. Prior research shows that students’
academic motivation and self-concept are vulnerable to decline during educational
transitions (e.g., Blyth et al., 1983; Eccles et al., 1993; 1984; Wilkins & Ma, 2003).
However, much of the research surrounding this topic is dated, and relative to the
transition to middle school or junior high, there is less research on the transition to
high school and experiences in the last two years of high school, especially as they
relate to math and science interest. Consequently, the first objective sought to
validate previously reported declines in interest and self-efficacy over different
Efficacy
Classroom
Quality
Usefulness
Stereotype
Post-Transition
Math Interest
Pre-Transition
Math Interest
0.27 [.20 to .33]***
0.53 [.45 to .60]***
Figure 4. Accounting for change in math Interest with post-transition Efficacy,
Classroom Quality and, Usefulness Stereotype. Results are for all grade levels
and genders combined. For the path between pre- and post-transition
Interest, the coefficient above the line is the remaining direct effect after
taking into account the three other indirect paths. The coefficient below the
line is the total effect of pre-transition Interest on post-transition Interest.
95% CIs are in brackets. Negative coefficients for Usefulness Stereotype
indicate more traditional stereotypical beliefs are associated with lower
Interest.
*p < .05; **p < .01; ***p < .001.
Post-Transition
International Journal of Gender, Science and Technology, Vol.12, No.1
22
transition periods, as well as previously reported gender differences before and
after the transitions. The second objective was to examine how self-efficacy,
perceived classroom quality, and stereotypes associated with the utility of math and
science are related to changes in interest over transitions, as well as gender
differences therein.
With respect to the first objective, although there were declines in Interest and
Efficacy over school transitions, they were not as substantial or pervasive compared
to previous research. Declines in interest were significant for students making the
transition to middle school, consistent with past research, but not for the other
grades. For Efficacy, results indicated a decline for girls’ science Efficacy, but not
boys’, which is generally consistent with past research that showed declines in
global self-esteem (Blyth et al., 1983). Our finding that math Efficacy did not
change over time is similar to some prior work (Friedel, Cortina, Turner, & Midgley,
2010), although self-efficacy did not vary across grade level, as other work might
suggest (Anderman & Midgley, 1997). Moreover, the effect sizes were generally
small (although still significant), suggesting that on average declines were perhaps
less substantial than reported in previous research.
We were especially interested in examining gender differences because college
enrollment in some STEM fields remains lower for girls than boys, despite similar
academic preparation. Our results suggest that gender differences in self-efficacy
and interest become more salient as children get older. Girls’ Interest and Efficacy
in math and science were comparable across grade levels, while older boys had
higher Interest and Efficacy than younger boys (Figures 1 and 2). Similar to this
study, previous research has documented an increase in academic interests and
self-concept for boys over school transitions (Blyth et al., 1983). However, our
findings for the stability in girls’ self-efficacy over transitions is inconsistent with
some previous research that suggests girls are more adversely affected by school
transitions as a whole (Blyth et al., 1983; Crockett, Petersen, Graber, Schulenberg,
& Ebata, 1989; Watt, 2004). They are also at odds with research suggesting that
gender differences in math self-competence beliefs decline over schooling
(Fredricks & Eccles, 2002).
Our results highlight the importance of considering changes in interests and
attitudes during late high school. Importantly, in this study, boys and girls were
recruited from the same math and science courses in high school, so gender
differences cannot be explained by course-taking. Unfortunately, this study did not
have the resources to follow the students who entered high school through their
senior year. Based on the cross-sectional data available in this study, gender
differences in science and math interest and self-efficacy may strengthen over high
school.
To summarize, we did not find strong support for the hypothesis that interest and
self-efficacy decline over each transition or across grade levels, and there was
limited support for gender differences favoring males across grade levels and
transitions. Nevertheless, gender differences in self-perceived abilities, efficacy, and
interest are well documented (e.g., Cunningham et al., 2015; Watt, 2004; Watt et
International Journal of Gender, Science and Technology, Vol.12, No.1
23
al., 2006), and the challenge before researchers is explaining why methodologically
sound studies find different results. Several factors might explain the differences
between our findings and those of previous research. Gender role norms and
educational practices are highly influenced by secular and historical changes, which
in turn impact children’s socialization and subsequent behaviors (Bronfenbrenner &
Evans, 2000). Significantly, the data in the current study were collected at a time
(2008-2010) when national educational statistics indicated that the gender gap in
math course taking was minimal, the gap for science course taking was on the
decline (Cunningham et al., 2015), and the gender gap in some STEM fields, most
notably Biology, had weakened (NSF, 2017). These historical shifts may be due to
many factors, for example, efforts by educators to directly intervene, the
availability of more female role models in STEM occupations, especially in popular
media, or changes in the larger society’s expectations for girls. Conjecture based on
stage-environment fit theory points to qualities of the post-transition school setting,
such as greater teacher support and engaging, age-appropriate classroom
practices, both of which were found to mediate interest outcomes in this study.
A second factor accounting for inconsistencies across studies may be due to
regional differences in educational approaches or gender role norms. Participants in
this study attended public schools in the Southeastern U.S., but much of the
previous research was conducted in the upper Midwest (e.g., studies using the
MSALT database or Blyth and colleagues’ research based in Wisconsin). Southern
states consistently lag behind Northern and Midwestern states on most academic
indicators, and this could affect how math and science are taught (e.g., focusing on
basics, lower expectations for achievement, fewer opportunities for taking advanced
math and science classes). It is possible that the emphasis on gender role norms
also varies across regional contexts throughout the United States, again with the
expectation that residents in Southern states might hold more traditional beliefs.
Both of these explanations point to the need to revisit educational outcomes and
gender differences related to school transitions periodically and to include samples
that represent different sectors of the U.S. The Bronfenbrenner and Evans (2000)
social-ecological model provides a framework to interpret differences over time and
across geographic regions.
The second objective examined how self-efficacy, perceived classroom quality, and
gender stereotypes for the utility of math and science explained change in interest
over transitions. Although each of the three factors were correlated with math and
science interest, analyses both within time point and across time points confirmed
that Efficacy was the strongest predictor of math and science Interest, followed by
Classroom Quality. Our regression analyses predicting pre- and post-transition
Interest in math and science are similar, attesting to the reliability of the
relationships within our sample and increasing our confidence in the predictive
ability of Efficacy, Perceived Classroom Quality, and gender stereotypes. Gender
stereotypes added explanatory power only for science Interest. Importantly,
Efficacy, Classroom Quality, and Usefulness Stereotypes partially mediated changes
in science interest; whereas Efficacy and Classroom Quality showed partial
mediation effects for change in math Interest. Despite mean level gender and grade
differences, Efficacy, Classroom Quality and Usefulness Stereotype were related to
International Journal of Gender, Science and Technology, Vol.12, No.1
24
changes in interest similarly across these groups. Moreover, the findings for
Classroom Quality support the tenets of the stage-environment fit theory (Eccles et
al., 1993) and suggest that positive teacher-student relationships and engaging
instructional approaches can help maintain interest in STEM at all grade levels.
The results for Usefulness Stereotype are interesting because of the differences in
its relation to math and science Interest and because the measures were similarly
related to Interest for boys and girls. Prior research has found a substantial amount
of shared variance between math-related gender stereotypes and math self-
concept, which could explain why gender stereotypes were not predictive for math
interest (e.g., Kurtz-Costes et al., 2008; Schmader et al., 2004). Additionally, the
utility value for math may have become more equivalent across genders in recent
years, compared to science. For example, in high school, girls and boys take
calculus at the same rate, but boys are more likely to take physics than girls
(Cunningham et al., 2015). Indeed, mathematics may serve as a basis for many
academic subjects and careers (e.g., accounting, STEM, social science research),
while science is narrower and leads to fewer non-science careers. Additionally, the
findings for gender stereotypes contradict some theorizing that boys benefit from
endorsing STEM gender stereotypes (Nosek et al., 2002; Walton & Cohen, 2003).
However, our results are consistent with some past research on occupation
stereotypes (Barth et al., 2018). Overall, our finding that the effects of Classroom
Quality and Usefulness Stereotype did not differ for males and females suggests
that both boys and girls would benefit from interventions that target these factors.
To summarize, a major conclusion of this study is that self-efficacy is a powerful
predictor of math and science interest, while classroom quality and stereotypes play
a secondary role. This conclusion held across gender and grade levels and is
consistent with expectancy value theory (Eccles, 2011) and social cognitive theory
(Bandura et al., 2001). A significant additional contribution of this study is that it
also supports other aspects of these models that propose that classroom
experiences and gender stereotypes are associated with interest across a range of
grade levels and for both males and females. Together, these findings contribute to
our understanding of the stage-environment fit theory (Eccles et al., 1993) and
have implications for educators. Prior interventions have focused on improving
elementary classroom quality (Gershenson, Lyon, & Budd, 2010; Spilt, Koomen,
Thijs, & van der Leij, 2012). Our work suggests that interventions should focus on
improving classroom quality across grade levels.
Limitations and Future Directions
We acknowledge some limitations of our study and suggest directions for future
research. The cross-sectional nature of the grade comparisons does not allow us to
make inferences on what happens to individuals over multiple school transitions.
Thus, there is a need for longitudinal data to support grade-related changes found
in this study, similar to the research conducted in past decades (Blyth et al., 1983;
Crockett et al., 1989; Eccles et al., 1984). Recent longitudinal studies have focused
on academic engagement (Wang & Eccles, 2013), self-esteem (Booth & Gerard,
2014), and family relations (Gutman & Eccles, 2007), but not math and science
self-efficacy and interest. Additionally, it would be valuable for future research to
International Journal of Gender, Science and Technology, Vol.12, No.1
25
include indicators of academic achievement as potential moderators of change over
transitions.
Additionally, stereotyping was assessed by the use of a single straightforward item
(“Math/Science is more useful for boys than girls”). This measure was selected
because of its direct mapping onto utility values associated with expectancy value
theory. It is not an uncommon practice to use a single-item scale to measure
constructs, such as gender stereotypes (e.g., Tiedemann, 2000). However, the use
of a single-item scale provides only a conservative test of the effects of
stereotyping. Future studies should replicate our findings with multi-item measures
that explore other aspects of gender stereotypes. Finally, we only found support for
a partial mediation, suggesting that there is still a lot to learn about why interest
changes over time. Future work might explore factors represented in other research
paradigms, such as peer support (Robnett, 2013), belonging (Cheryan, Plaut,
Davies, & Steele, 2009), values, and perceptions of discrimination (Hayes & Bigler,
2013).
Despite these limitations, this study makes an important contribution to the
research on school transitions. First, this study updates the increasingly dated
school transition literature. The present study provided support for stage-
environment fit theory (Eccles et al., 1993) and social cognitive theory (Bandura et
al., 2001). The findings highlighted above underscore the importance of self-
efficacy, classroom quality, and stereotypes on changes in math and science
interest across ages and gender.
International Journal of Gender, Science and Technology, Vol.12, No.1
26
REFERENCES
Anderman, E. M., & Midgley, C. (1997). Changes in achievement goal orientations,
perceived academic competence, and grades across the transition to middle-level
schools. Contemporary Educational Psychology, 22, 269-298.
Baird, C., & Keen, J. R. (2018). Closing the gender gap in math confidence: Gender
and race/ethnic similarities and differences in the effects of academic achievements
among high math achievers. International Journal of Gender, Science and
Technology, 10, 379-410.
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (2001). Self-efficacy
beliefs as shapers of children’s aspirations and career trajectories. Child
Development, 72, 187-206.
Barth, J. M., Kim, H., Eno, C. A., & Guadagno, R. E. (2018). Matching abilities to
careers for others and self: Do gender stereotypes matter to students in advanced
math and science classes? Sex Roles, 79, 83-97.
Blažev, M., Karabegović, M., Burušić, J., & Selimbegović, L. (2017). Predicting
gender-STEM stereotyped beliefs among boys and girls from prior school
achievement and interest in STEM school subjects. Social Psychology of Education,
20, 831–847.
Blyth, D. A., Simmons, R. G., & Carlton-Ford, S. (1983). The adjustment of early
adolescents to school transitions. Journal of Early Adolescence, 3, 105-120.
Booth, M. Z., & Gerard, J. M. (2014). Adolescents’ stage-environment fit in middle
and high school: The relationship between students’ perceptions of their schools
and themselves. Youth & Society, 46(6), 735–755.
Bronfenbrenner, U., & Evans, G. W. (2000). Developmental science in the 21st
century: Emerging questions, theoretical models, research designs, and empirical
findings. Social Development, 9, 115-125.
Brown, S. D., Tramayne, S., Hoxha, D., Telander, K., Fan, X., & Lent, R. W. (2008).
Social cognitive predictors of college students’ academic performance and
persistence: A meta-analytic path analysis. Journal of Vocational Behavior, 72,
298–308.
Chatard, A., Guimond, S., & Selimbegović, L. (2007). "How good are you in math?"
The effect of gender stereotypes on students' recollection of their school marks.
Journal of Experimental Social Psychology, 43, 1017-1024.
Cheryan, S., Plaut, V. C., Davies, P. G., & Steele, C. M. (2009). Ambient belonging:
How stereotypical cues impact gender participation in computer science. Journal of
Personality and Social Psychology, 97, 1045–1060.
Crockett, L. J., Petersen, A. C., Graber, J. A., Schulenberg, J. E., & Ebata, A.
(1989). School transition and adjustment during adolescence. Journal of Early
Adolescence, 9, 181-210.
Correll, S. J. (2001). Gender and the career choice process: The role of biased self-
assessments. American Journal of Sociology, 106, 1691–1730.
International Journal of Gender, Science and Technology, Vol.12, No.1
27
Cvencek, D., Meltzoff, A. N., & Greenwald, A. G. (2011). Math-gender stereotypes
in elementary school children. Child Development, 82, 766–779.
Cunningham, B. C., Hoyer, K. M., & Sparks, D. (2015). Gender differences in
Science, Technology, Engineering, and Mathematics (STEM) interest, credits
earned, and NAEP performance in the 12th grade. Stats in Brief. National Center for
Education Statistics. 2015-075. Retrieved from
https://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2015075
Eccles, J. S. (1994). Understanding women’s educational and occupational choices:
Applying the Eccles et al. model of achievement-related choices. Psychology of
Women Quarterly, 18, 585–610.
Eccles, J. S. (2011). Gendered educational and occupational choices: Applying the
Eccles et al. model of achievement-related choices. International Journal of
Behavioral Development, 35(3), 195–201.
Eccles, J. S., Midgley, C., & Adler, T. (1984). Grade-related changes in the school
environment: Effects on achievement motivation. In J. G. Nicholls (Ed.), The
development of achievement motivation (pp. 283-331). Greenwich, XT: JAI.
Eccles, J. S., Midgley, C., Wigfield, A., Buchanan, C.M., Reuman, D., Flanagan, C.,
& Mac Iver, D. (1993). Development during adolescence: The impact of stage-
environment fit on young adolescents’ experiences in schools and in families.
American Psychologist, 48, 90-101.
Evans, A. B., Copping, K. E., Rowley, S. J., & Kurtz-Costes, B. (2010). Academic
self-concept in black adolescents: Do race and gender stereotypes matter? Self &
Identity, 10, 263–277.
Ferry, T. R., Fouad, N. A., & Smith, P. L. (2000). The role of family context in a
social cognitive model for career-related choice behavior: A math and science
perspective. Journal of Vocational Behavior, 57, 384-364.
Fredricks, J. A. & Eccles J. S. (2002). Children’s competence and value beliefs from
childhood through adolescence: Growth curve trajectories in two male sex-typed
domains. Developmental Psychology, 38, 519-533.
Friedel, J. M., Cortina, K. S., Turner, J. C., & Midgley, C. (2010). Changes in
efficacy beliefs in mathematics across the transition to middle school: Examining
the effects of perceived teacher and parent goal emphases. Journal of Educational
Psychology, 102, 102–114.
Frome, P. M., & Eccles, J. S. (1998). Parents’ influence on children’s achievement-
related perceptions. Journal of Personality and Social Psychology, 74, 435-452.
Garrett, C., Ein, P., & Tremaine, L. (1977). The development of gender stereotyping
of adult occupations in elementary school children. Child Development, 48(2), 507-
512.
George, R. (2003). Growth in students’ attitudes about the utility of science over
the middle and high school years: Evidence from the Longitudinal Study of
American Youth. Journal of Science Education and Technology, 12, 439-448.
International Journal of Gender, Science and Technology, Vol.12, No.1
28
Gershenson, R. A., Lyon, A. R., & Budd, K. S. (2010). Promoting positive
interactions in the classroom: Adapting parent–child interaction therapy as a
universal prevention program. Education and Treatment of Children, 33, 261–287.
Goodenow, C. (1993). Classroom belonging among early adolescent students:
Relationships to motivation and achievement. The Journal of Early Adolescence, 13,
21–43.
Gottfredson, L. S. (1981). Circumscription and compromise: A developmental
theory of occupational aspirations. Journal of Counseling Psychology, 28, 545–579.
Gottfredson, L. S., & Lapan, R. T. (1997). Assessing gender-based circumscription
of occupational aspirations. Journal of Career Assessment, 5, 419-441.
Gutman, L. M. (2006). How student and parent goal orientations and classroom
goal structures influence the math achievement of African Americans during the
high school transition. Contemporary Educational Psychology, 31, 44-63.
Gutman, L. M., & Eccles, J. S. (2007). Stage–Environment fit during adolescence:
Trajectories of family relations and adolescent outcomes. Developmental
Psychology, 43, 522-537.
Hampden-Thompson, G., & Bennett, J. (2013). Science teaching and learning
activities and students’ engagement in science. International Journal of Science
Education, 35, 1325–1343.
Hayes, A. F. (2018). Introduction to mediation, moderation, and conditional process
analysis, 2nd Edition. New York: The Guilford Press.
Hayes, A. R., & Bigler, R. S. (2013). Values, perceptions of discrimination, and
graduate training in STEM fields, International Journal of Gender, Science and
Technology, 5, 255-280.
Hyde, J. S., & Durik, A. M. (2005). Gender, competence, and motivation. In A. J.
Elliot & C. S. Dweck (Eds.), Handbook of competence and motivation (pp. 375–
391). New York: Guilford Press.
Isakson, K., & Jarvis, P. (1999). The adjustment of adolescents during the
transition into high school: A short-term longitudinal study. Journal of Youth &
Adolescence, 28, 1-26.
Katz, P. A., & Ksansnak, K. R. (1994). Developmental aspects of gender role
flexibility and traditionality in middle school childhood and adolescence.
Developmental Psychology, 3, 272-282.
Kurtz-Costes, B., Rowley, S. J., Harris-Britt, A., & Woods, T. A. (2008). Gender
stereotypes about mathematics and science and self-perceptions of ability in late
childhood and early adolescence. Merrill-Palmer Quarterly, 54, 386–409.
Lane, K. A., Goh, J. X., & Driver-Linn, E. (2012). Implicit science stereotypes
mediate the relationship between gender and academic participation. Sex Roles,
66, 220-234.
Lawson, K. M., Lee, B., Crouter, A. C., & McHale, S. M. (2018). Correlates of
gendered vocational development from middle childhood to young adulthood.
Journal of Vocational Behavior, 107, 209–221.
International Journal of Gender, Science and Technology, Vol.12, No.1
29
Ma, X., & Wilms, J. D. (1999). Dropping out of advanced mathematics: How much
do students and schools contribute to the problem? Educational Evaluation and
Policy Analysis, 21, 365-383.
Maas, C. J., & Hox, J. J. (2005). Sufficient sample sizes for multilevel modeling.
Methodology, 1, 86-92.
Maltese, A., Melki, C., & Wiebke, H. (2014). The nature of experiences responsible
for the generation and maintenance of interest in STEM. Science Education, 98,
937–962.
Marchant, G., Paulson, S., & Rothlisberg, B. (2001). Relations of middle school
students’ perceptions of family and school contexts with academic achievement.
Psychology in the Schools, 38, 505-519.
Master, A., Cheryan, S., & Meltzoff, A. N. (2016). Computing whether she belongs:
Stereotypes undermine girls’ interest and sense of belonging in computer science.
Journal of Educational Psychology, 108, 424–437.
Michigan Study of Adolescent and Adult Life Transitions (MSALT) Retrieved from
http://garp.education.uci.edu/msalt.html.
Midgley, C., Feldlaufer, H., & Eccles, J. S. (1989). Student/teacher relations and
attitudes toward mathematics before and after the transition to junior high school.
Child Development, 60, 981-902.
Murdock, T. B., Anderman, L. H., & Hodge, S. A. (2000). Middle-grade predictors of
students’ motivation and behavior in high school. Journal of Adolescent Research,
15, 327-351.
National Research Council. (2004). Engaging schools: Fostering high school
students’ motivation to learn. Washington, DC: National Academies Press. Retrieved
from https://www.nap.edu/catalog/10421/engaging-schools-fostering-high-school-
students-motivation-to-learn
National Research Council. (2012). A Framework for K-12 Science Education:
Practices, Crosscutting Concepts, and Core Ideas. Washington, DC: The National
Academies Press. Retrieved from https://www.nap.edu/catalog/13165/a-
framework-for-k-12-science-education-practices-crosscutting-concepts
National Science Board. (2018). Science and Engineering Indicators. Retrieved from
https://www.nsf.gov/statistics/2018/nsb20181/assets/nsb20181.pdf
National Science Foundation. (2017). Women, Minorities, and persons with
Disabilities in science and engineering. National Center for Science and Engineering
Statistics Directorate for Social, Behavioral and Economic Sciences. Retrieved from
https://www.nsf.gov/statistics/2017/nsf17310/static/downloads/nsf17310-
digest.pdf
Nosek, B. A., Banaji, M. R., & Greenwald, A. G. (2002). Mathematics = male, me =
female, therefore mathematics ≠ me. Journal of Personality and Social Psychology,
83, 44–59.
O'Dwyer, L. M., & Parker, C. E. (2014). A primer for analyzing nested data:
Multilevel modeling in SPSS using an example from a REL study. The National
International Journal of Gender, Science and Technology, Vol.12, No.1
30
Center for Education Evaluation and Regional Assistance, REL 2015-046. Retrieved
from https://ies.ed.gov/ncee/edlabs/regions/northeast/pdf/REL_2015046.pdf
Parker, P., Nagy, G., Trautwein, U., & Ludtke, O. (2014). Predicting career
aspirations and university majors from academic ability and self-concept: A
longitudinal application of the internal–external frame of reference model. In I.
Schoon & J. S. Eccles (Eds.), Gender differences in aspirations and attainment: A
life course perspective (pp. 224–246). Cambridge, UK: Cambridge University Press.
Pintrich, P. R. (2000). Multiple goals, multiple pathways: The role of goal
orientation in learning and achievement. Journal of Educational Psychology, 92,
544- 555.
Regan, E., & DeWitt, J. (2015). Attitudes, interest and factors influencing STEM
enrolment behaviour: An overview of relevant literature. In E. K. Henriksen, J.
Dillon, & J. Ryder (Eds.), Understanding student participation and choice in science
and technology education (pp. 63–88). Dordrecht: Springer.
Robnett, R. (2013). The role of peer support for girls and women in the STEM
pipeline: Implications for identity and anticipated retention. International Journal of
Gender, Science and Technology, 5(3), 232-253.
Ryan, A. M., & Patrick, H. (2001). The classroom social environment and changes in
adolescents’ motivation and engagement during middle school. American
Educational Research Journal, 38, 437–460.
Scherbaum, C. A., & Ferreter, J. M. (2009). Estimating statistical power and
required sample sizes for organizational research using multilevel modeling.
Organizational Research Methods, 12, 347-367.
Schmader, T., Johns, M., & Barquissau, M. (2004). The costs of accepting gender
differences: The role of stereotype endorsement in women’s experience in the
math domain. Sex Roles, 50, 835–850.
Spilt, J. L., Koomen, H. M. Y., Thijs, J. T., & van der Leij, A. (2012). Supporting
teachers’ relationships with disruptive children: The potential of relationship-
focused reflection. Attachment and Human Development, 14, 305-318.
Steffens, M. C., Jelenec, P., & Noack, P. (2010). On the leaky math pipeline:
Comparing implicit math gender stereotypes and math withdrawal in female and
male children and adolescents. Journal of Educational Psychology, 102, 947–963.
Stipek, D. (2001). Classroom context effects on young children's motivation. In F.
Salili, C. Chiu, & Y. Hong (Eds.), Student motivation: The culture and context of
learning (pp. 273–292). New York, NY: Kluwer Academic.
Tiedemann, J. (2000). Parents’ gender stereotypes and teachers’ beliefs as
predictors of children’s concept of their mathematical ability in elementary school.
Journal of Educational Psychology, 92, 144–151.
Walton, G. M., & Cohen, G. L. (2003). Stereotype lift. Journal of Experimental
Social Psychology, 39, 456–467.
International Journal of Gender, Science and Technology, Vol.12, No.1
31
Wang, M. T. (2012). Educational and career interests in math: A longitudinal
examination of the links between perceived classroom environment, motivational
beliefs, and interests. Developmental Psychology, 48, 1643-1657.
Wang, M. T., & Eccles, J. S. (2013). School context, achievement motivation, and
academic engagement. Learning and Instruction, 28, 12-23.
Watt, H. (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.
Watt, H. M. G., & Eccles, J. S. (Eds.). (2006). Understanding women’s choice of
mathematics- and science-related careers: Longitudinal studies from four
countries. Educational Research and Evaluation, 1, 295-296.
Watt, H. M. G., Eccles, J. S., & Durik, A. M. (2006). The leaky mathematics
pipeline for girls: A motivational analysis of high school enrollments in Australia
and the USA. Equal Opportunities International, 25, 642–659.
Wigfield, A., & Eccles, J. S. (2000). Expectancy-value theory of motivation.
Contemporary Educational Psychology, 25, 68–81.
Wigfield, A., & Eccles, J. S. (2002). The development of competence beliefs,
expectancies for success, and achievement values from childhood through
adolescence. In A. Wigfield & J. S. Eccles (Eds.), Development of achievement
motivation (pp. 91-120). San Diego, CA, US: Academic Press.
Wigfield, A., Eccles, J. S., MacIver, D., Reuman, D. A., & Midgley, C. (1991).
Transitions during early adolescence: Changes in children’s domain-specific self-
perceptions and general self-esteem across the transition to junior high school.
Developmental Psychology, 27, 552–565.
Wilkins J. L. M., & Ma, X. (2003). Modeling change in student attitude toward and
beliefs about mathematics. Journal of Educational Research, 97, 52-63.
