Girls and mathematics —A “hopeless” issue? A control-value approach to gender differences in emotions towards mathematics

TL;DR: In this article, the structural relationships between prior achievement, control and value beliefs, and emotions were assumed to be invariant across girls and boys in spite of hypothesized mean level differences of beliefs and emotions across genders.

Abstract: This study analyzed gender differences in achievement emotions in the domain of mathematics. Based on Pekrun’s (2000, 2006) controlvalue theory of achievement emotions, we hypothesized that there are gender differences in mathematics emotions due to the students’ different levels of control and value beliefs in mathematics, even when controlling for prior achievement. The structural relationships between prior achievement, control and value beliefs, and emotions were assumed to be invariant across girls and boys in spite of hypothesized mean level differences of beliefs and emotions across genders. The emotions and beliefs of 1,036 male and 1,017 female 5th grade students were assessed by self-report measures, and their prior mathematics achievement was assessed by academic grades. Even though girls and boys had received similar grades in mathematics, girls reported significantly less enjoyment and pride than boys, but more anxiety, hopelessness and shame. Findings suggested that the female emotional pattern was due to the girls’ low competence beliefs and domain value of mathematics, combined with their high subjective values of achievement in mathematics. Multiple-group comparisons confirmed that the structural relationships between variables were largely invariant across the genders.

Summary

Gender differences in mathematics-related affect

• Gender differences in affective variables relating to mathematics moved to the center of research interest since the Fennema-Sherman studies in the 1970s (Fennema & Sherman, 1977 , 1978) .
• In her study, she did not separate different emotions.
• Concerning potential gender differences in discrete mathematics emotions, the only variable which has been researched extensively is mathematics anxiety.
• Stipek and Gralinsky's (1991) data indicated that girls showed a "female" attribution bias by attributing failure to low ability, and not attributing success to high ability.

The control-value theory of achievement emotions

• The authors theorizing on emotions is grounded in an appraisal theoretical approach which posits that interindividual differences in emotions are due to differences in the cognitive interpretations of situations and events (e.g., Scherer, Schorr, & Johnstone, 2001) .
• Another student may be more eoneemed with performing well in mathematies in order to inerease eareer options, to meet parents' expeetations, or to attain personal aehievement standards.
• That is, even though these two types of values may be closely eonneeted in many individuals, there may also be students who have high values of a domain, but regard aehievement in the same domain as less important, or viee versa.
• Regarding the effeets of these appraisals on achievement emotions, the control-value theory would generally prediet that high eompetenee beliefs will be assoeiated with higher levels of the positive emotions enjoyment and pride.
• Anxiety, hopelessness, and shame, on the other hand, should result from low eompetenee beliefs, eombined with high values of aehievement (Abela & Seligman, 2000; Hembree, 1988; Pekrun, 1992) .

Assumptions concerning the interplay of gender, prior achievement, beliefs, and emotions in mathematics

• Gender differences in beliefs and emotions and a mediation assumption.
• Therefore, the authors assumed that girls would be less eonfident in their eompetenee judgments in mathematics in grade five, the age group under study here.
• Applying the eontrol-value theory, the authors predieted that this female pattem of eompetenee and value appraisals would lead to a debilitating emotional profile.
• Finally, in spite of expected similar levels of achievement values which should be important for the formation of pride, the authors assumed that the lower competence beliefs in girls reduce their experienced pride in relation to mathematics.

Control and value appraisals Emotions

• In contrast, evidence suggesting that gender can function as a moderator of relations between cognitive-affective variables is limited (see Meyer & Koehler, 1990; Zeidner & Safir, 1989) .
• Therefore, even though the authors did expect mean level differences in beliefs and emotions between girls and boys, they assumed that the relationships between prior achievement, students' beliefs, and their emotions would be universal, that is, they should fiinction similarly across genders.
• To summarize, the following hypotheses were tested in the present study.
• Hypothesis 2. Girls have lower eompetenee beliefs and domain values in mathematics than boys, but there are no gender differenees in the value of achievement in mathematies.

Sample and procedures

• Sampling included all three sehool types of the German state sehool system ("Hauptschule", "Realsehule", and "Gymnasium").
• Emotions, eompetenee beliefs, and value beliefs were assessed by a questionnaire at the end of the sehool year.
• This assessment was administered by external, trained testing personnel.

Measures

• The competence belief scale asked students to rate their competence in mathematics (10 items with 5-point Likert scales from 1, "strongly disagree" to 5, "strongly agree"; a=.92; sample item: "I am a good student in mathematics").
• Both the domain value and the achievement value scales consisted of five items (5-point Likert scales from "strongly agree" to "strongly disagree"; a=.84/.8O; sample item for domain value: "Mathematics is my favorite subject;" sample item for achievement value: "It is very important for me to get good grades in mathematics").
• Five discrete emotions (enjoyment, pride, anxiety, hopelessness, and shame) were assessed in the present study.
• Items are answered on the same 5-point Likert scale as described above.
• These mid-term grades represent the teacher-based judgment of the students' oral and written mathematics performance in the first half of the year.

Analysis Procedures

• First, the authors performed a multivariate analysis of variance to analyze gender differences in the mean levels of mathematics emotions and competence and value beliefs (Hypotheses 1 and 2).
• Second, the authors applied regression analyses to test Hypothesis 3, which stated that gender differences in mathematics emotions were mediated by students' competence and value beliefs.
• All regression analyses were eomputed with AMOS 5.0 (Arbuckle, 2003) using maximum likelihood estimation and ftiU information maximization imputation of missing data.
• Compared to the simple regression models which were just identified (zero degrees of freedom), the model whieh set the regression eoeffieients to equality had 4 degrees of freedom, making it possible to eompute a j^-statistie.
• Therefore, the authors were able to test whether the eonstraint of equality of regression eoeffieients across the male and female subgroups resulted in a loss of fit which was statistically signifieant in terms ofthe likelihood ratio test.

Gender differences in mean levels of mathematics emotions, beliefs, and achievement

• Based on eumulative empirical evidenee of gender differenees in mathematies eompetenee beliefs, values, and anxiety, and seattered findings eoneeming other mathematies emotions, their first two hypotheses eonsidered gender level differences among these variables.
• Given the large size ofthe present sample, effect sizes (Cohen's d) of differences were calculated to analyze the relevanee of gender differenees in emotions, beliefs and aehievement, in addition to f-values of the MANOVA (see Table 2 ).
• In line with Hypothesis 1, boys reported eonsiderably more enjoyment and pride in mathematies, and less anxiety and hopelessness than girls (see Table 2 , left part).
• This effeet size was noticeably smaller and therefore, this gender differenee eannot be regarded as substantial.
• The gender differences virtually remained the same (see Table 2 , right part).

Mediation and moderation effects in the interplay of gender, achievement, beliefs, and emotions

• Based on the assumptions of the control-value theory, the authors had assumed that competence and value beliefs mediated the gender differences in emotions (Hypothesis 3), but that gender would not act as a moderator of the relationships between prior achievement, beliefs, and emotions (Hypothesis 4).
• To test these two hypotheses, the authors conducted multiple regression analyses, and they compared the regression weights for the prediction of the emotions across girls and boys by means of a multiple-group comparison.
• Furthermore, the regression analyses revealed that prior achievement also had significantly positive effects on enjoyment and pride, and negative effects on anxiety, hopelessness, and shame when beliefs were not taken into account (Table 3 , left coefficients within emotions).
• The eonstrained models had four degrees of freedom, and a likelihood ratio test based on the ;(^-statistie was performed to judge whether the assumption of gender universality was tenable.
• The slight discrepancy between models was eaused by the fact that the regression eoefficient for aehievement value was higher for boys than for girls, thus suggesting that the value of achievement might be more important for boys' than for girls' pride in mathematics.

Discussion

• First, the authors aimed at examining girls' and boys' emotional experiences in mathematics, including the emotions enjoyment, pride, anxiety, hopelessness and shame.
• Pronounced gender differences were found among the 5th graders in their sample, with girls having a considerably more negative emotional pattem than boys.
• The authors second aim was to test the assumption derived from control-value theory (Pekrun, 2000 (Pekrun, , 2006 ) that these differences are mediated by girls' and boys' appraisals of competence and values in mathematics.
• The authors data largely confirmed this mediational hypothesis.
• The pattern of relations between these variables proved to be very similar across the genders, thus supporting universality assumptions.

Gender differences in mean levels of experienced emotions

• Conceming the gender differences observed in discrete emotions in mathematics, the authors found that girls experienced significantly less enjoyment and less pride than boys.
• Further, girls reported slightly more shame than boys.
• This suggests that there are factors other than achievement that bolster boys' emotional experiences in mathematics, or harm girls' feelings conceming this domain.
• Nevertheless, it would likely be misleading to interpret the gender differenees found in this study as being due to no more than the faet that "females are just more emotional", or to differenees in response behavior when dealing with self-report emotion items.

Mediation of gender differences in emotions by competence and value beliefs

• The eontrol-value theory of aehievement emotions (Pekmn, 2000 (Pekmn, , 2006) ) implies that eompetenee-related beliefs, as well as subjeetive domain-related and aehievement-related values in mathematies, function as mediators of students' emotional experienees.
• The findings ofthe present study are in line with this assumption.
• As expeeted, the different emotions assessed in this study proved to be differentially related to these beliefs.
• While the authors had hypothesized that high values of achievement, along with eompetenee beliefs, would be important for outome-related pride to occur, their data suggested that both domain and achievement value are positively related to pride.
• That is, the girls' laek of eompetenee beliefs and domain value pertaining to mathematics, paired with the need to do well in mathematies, seems to be responsible for their higher levels of negative emotions in this domain.

Cross-gender universality in the interplay of achievement, beliefs, and emotions

• The pattern of relationships between prior aehievement, beliefs, and emotions in mathematies was very similar for girls and boys in the present study.
• A test of invarianee between the regression models for the two genders corroborated the universality assumption for pride, anxiety, and shame, and deteeted minor diserepancies for enjoyment and hopelessness.
• In contrast, the eompetenee belief that one ean do well in mathematies was a stronger predictor for enjoyment in girls than in boys.
• The predictive power of any variable may have been rendered lower in the male as compared to the female subsample in the case of hopelessness.
• In conclusion, the authors argue that the gender differences in emotions found in this study were likely due to differences in competence and value beliefs as discussed above, rather than to gender differences in the relationships between beliefs and emotions.

Summary and implications for future research

• Research on mathematics emotions has focused on students' anxiety in mathematics, but has largely neglected emotions other than anxiety.
• The findings of the present study support the assumption that girls, as opposed to boys, are characterized by a debilitating pattem of mathematics-related emotions, and of underlying competence beliefs and value beliefs which can be observed as early as at the age of eleven.
• Nous avons pose comme hypothese la presence de cette difference mime une fois I'accomplissement anterieur pris en consideration.

Full text

Girls and mathematics - A "hopeless" issue?
A control-value approach to gender differences
in emotions towards mathematics
Anne C. Frenzel
Reinhard Pekrun
University of Munich, Germany
Thomas Goetz
University of Konstanz, Germany / College of Teacher Education,
Thurgan, Switzerland
This
study
analyzed gender differences
in
achievement emotions
in
the
domain
of
mathematics.
Based on Pekrun's
(2000,
2006) control-
value theory of
achievement
emotions,
we
hypothesized
that there are
gender differences in mathematics emotions due to the students'
different levels
of
control
and value
beliefs
in
mathematics,
even when
controlling for
prior
achievement.
The structural relationships between
prior achievement, control and value beliefs, and emotions were
assumed
to be
invariant across
girls and
boys
in spite of
hypothesized
mean level differences of
beliefs
and emotions across genders. The
emotions
and
beliefs
of
1,036
male
and 1,017female
5th grade students
were assessed by self-report
measures,
and their prior mathematics
achievement
was
assessed
by
academic
grades.
Even
though
girls and
boys had received similar grades in mathematics, girls reported
significantly less enjoyment and pride than boys, but more anxiety,
hopelessness
and
shame.
Findings suggested
that the female
emotional
pattern was due to
the
girls' low
competence beliefs
and
domain
value
of mathematics, combined with their high subjective values of
achievement in
mathematics.
Multiple-group comparisons confirmed
that the structural relationships between variables were largely
invariant across the
genders.
It is a widespread belief among students, teachers, and parents alike that girls and
mathematics are a "bad fit". Researchers from a wide range of disciplines have long been
This research was funded by a grant from the German Research Foundation (PE 320/11-1) awarded to the second
author.
An earlier version of this article was presented at the annual meeting of the American Educational Research
Association, Montreal, Canada, April 2005.
Konstanzer Online-Publikations-System (KOPS)
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-76744
URL: http://kops.ub.uni-konstanz.de/volltexte/2009/7674
First publ. in: European Journal of Psychology of Education 22 (2007), 4, pp. 497-514

498
eoncerned with the potential psychological, biological, and soeial reasons for gender
differenees in mathematics (e.g., Gallagher & Kaufman, 2005). Reeent findings of gender
differenees display a striking pattern of small and declining differences in mathematies
aehievement, but eonsistent and considerable differences in the affeetive domain (e.g..
Organisation for Eeonomic Co-operation and Development (2004a). In the latter eontext, much
researeh attention has been direeted towards girls' and boys' judgments of eompetenee,
self-
effieaey expeetations, or expeetations for their future performanee in a given domain (e.g.,
Bandura, 1997; Covington, 1984; Eccles, Wigfield, & Schiefele, 1997; NiehoUs, 1990).
Less attention has been paid to emotional variables in the context of learning and
aehievement. However, we ean assume that emotions - our "hot" cognitions - are prevalent
and infiuential in sueh situations. Recent researeh indicates that students experienee a wide
range of emotions in the context of learning and achievement. It is acknowledged that
emotions are highly relevant and important in learning and aehievement situations, as
indieated by three reeent speeial issues on emotions and learning in the Educational
Psychologist (Sehutz & Lanehart, 2002), in Learning and Instruction (Efklides & Volet,
2005),
and in the Educational Psychology Review (Linnenbrink, 2006), and one edited
volume (Sehutz & Pekrun, 2007). In this eontext, the most extensively researched emotion has
been anxiety (Hembree, 1988; Zeidner, 1998). The subject of mathematies in partieular has
been found to elicit anxiety in students (Ashcraft, 2002; Ma, 1999; Wigfield & Meeee, 1988).
Pekmn, Goetz, Titz, and Perry (2002) reported that emotions typieally experienced by stu-
dents inelude task- or activity-related feeling states such as task enjoyment, and emotions
whieh relate to leaming outcomes, sueh as pride, shame, anxiety, and hopelessness. However,
emotions other than anxiety have reeeived eonspicuously little attention in researeh on
leaming and achievement, with the exeeption of researeh on the attributional anteeedents of
aehievement emotions (Weiner, 1985, 1994).
We believe that there are at least three reasons why studying emotions in leaming and
aehievement is important. First, emotions are important dependent variables in their own
right, sinee they are key eomponents of subjeetive well-being and psychological health (e.g.,
Diener, 2000).
Second,
emotions impaet students' leaming and achievement. High quality
leaming is time-eonsuming and effort-intensive. Leamers are more willing to invest sueh
effort if leaming aetivities are affeetively rewarding - that is, enjoyable and interesting rather
than anxiety-laden or boredom-indueing. Moreover, emotions infiuenee leaming by ehanging
dopamine levels in the brain, thereby affecting long-term memory (Ashby, Isen, & Turken,
1999);
by directing attentional proeesses and the use of eognitive resourees (Meinhardt &
Pekrun, 2003); by indueing and sustaining student interest in learning material (Ainley,
Corrigan, & Riehardson, 2005; Krapp, 2005); by triggering different modes of information
processing and problem solving (Isen, 1999); and by faeilitating or impeding students'
self-
regulation of learning and performance (Pekrun et al., 2002).
Third,
while a student's
cognitive eompetencies predict sueeess at leaming, affective variables better predict whether
or not the student aetually enrolls in eourses or pursues a eareer in a given domain
(Haraekiewitz, Barron, Tauer, Carter, & Elliot, 2000; Wigfield, Battle, Keller, & Eeeles,
2002).
That is, if students are emotionally attraeted by the content of
a
domain, they beeome
keen to leam more about this domain, and deeide to follow a eareer in this domain.
The importanee of emotions for educational and occupational career ehoices makes
emotion-related gender differenees in mathematies partieularly relevant. Significant progress
has been made in redueing the gender gap in formal edueation, as indieated by gender equality
in graduation rates in many of the OECD member eountries (OECD, 2004b). This progress
notwithstanding, gender differenees in tertiary qualifieations remain persistently high in
mathematies and computer seience. On average, in the OECD eountries only 30% of the
university graduates in mathematies and eomputer seienee are female (OECD, 2004b).
Because of
the
importanee of emotions for eareer goal ehoiees, inquiry on gender differences
in emotional experienees is of
emeial
relevance if
one
wants to design interventions aiming to
inerease the number of women entering careers in traditionally male, soeially reeognized and
eeonomically important
job
domains.

EMOTIONS, GENDER, AND MATHEMATICS 499
In the present study, we focused on gender differences in five distinct mathematics
emotions, namely enjoyment, pride, anxiety, hopelessness, and shame. In doing so, we
attempted to describe and explain girls' and boys' patterns of emotions in mathematics.
Furthermore, we wanted to identify appraisal antecedents of mathematics emotions which
might be responsible for gender differences in these emotions.
Gender differences in mathematics-related affect
Gender differences in affective variables relating to mathematics moved to the center of
research interest since the Fennema-Sherman studies in the 1970s (Fennema & Sherman,
1977,
1978). Variables that have been extensively researched include confidence, perceived
usefulness, math as a male domain, and attitude towards math success. All of these variables
reflect a pattem of self-related cognitions and affect which is more favorable for boys in terms
of leaming and practicing mathematics. A meta-analysis by Hyde, Fennema, Ryan, Frost, and
Hopp (1990) showed that gender-based discrepancies of affect in mathematics are prevalent
and of considerable size.
In the literature on gender differences in mathematics, emotions are typically discussed
rather generally in terms of the positive vs. negative valence of experiences related to leaming
mathematics. For example, Eccles, Adler, Futtermann,
Goff,
Kaczala, Meece, and Midgley
(1983) emphasized the importance of "affective experiences in mathematics" (p. 96).
Similarly, Brush (1985) spoke of a "feelings factor" in mathematics learning. She
operationalized this factor as the combination of students' assessments of mathematics as
being easy vs. difficult, enjoyable vs. anxiety-provoking, and creative vs. dull. According to
Brush, this "feelings factor" was highly predictive of the level of students' course preferences.
However, in her study, she did not separate different emotions. Consequently, it seems that
emotions have so far been studied as constituents of more global factors in the context of
gender differences in mathematics attitudes, but not as distinct constructs in their own right.
Concerning potential gender differences in discrete mathematics emotions, the only
variable which has been researched extensively is mathematics anxiety. Various studies
showed that girls tend to be more anxious than boys during mathematics tasks and in contexts
involving mathematical thinking. In Hyde, Fennema, Ryan, et al.'s (1990) meta-analysis on
gender differences in mathematics affect, the difference in math anxiety yielded an overall
Cohen's
c^
of.
15.
One study that explicitly addresses mathematics-related emotions other than anxiety is
Stipek and Gralinsky's (1991) study on gender differences in emotional responses to success and
failure in mathematics. This study was conceptually based on Weiner's (1985) attributional
theory which posits that discrete emotional experiences are elicited by the attributions for
success and failure endorsed by the student. Stipek and Gralinsky's (1991) data indicated that
girls showed a "female" attribution bias by attributing failure to low ability, and not attributing
success to high ability. Girls were further found to report less pride after success and a
stronger desire to hide their paper after failure, a behavior interpreted as representing feelings
of shame. This study thus suggested that girls tend to experience a more negative emotional
pattem in mathematics, beyond the well-researched emotion of anxiety.
The control-value theory of achievement emotions
Our theorizing on emotions is grounded in an appraisal theoretical approach which posits
that interindividual differences in emotions are due to differences in the cognitive interpretations
of situations and events (e.g., Scherer, Schorr, & Johnstone, 2001). Specifically, we base our
assumptions on gender differences on Pekrun's control-value theory of achievement emotions
(Pekmn, 2000, 2006; Pekrun et al., 2002). The theory postulates that appraisals pertaining to
achievement activities and their outcomes are of primary importance for the instigation of
achievement emotions. More specifically, control-related appraisals (such as, competence
beliefs, causal expectations, and causal attributions) and value appraisals are held to be most

500
A.C.
FRENZEL,
R.
PEKRUN,
&
T.
GOETZ
important. The theory integrates assumptions from expectancy-value (Pekmn, 1992; Tumer &
Sehallert, 2001) and attributional approaches (Weiner, 1985) to achievement-related emotions.
In the present study, eoneeming eontrol appraisals, we specifically concentrated on the
students' subjective eompetenee beliefs in mathematics, i.e., subjeetive generalized judgments
eoneerning how competent one is at performing demanded tasks. Apart from eausal
expeetations and attributions, sueh eompetenee beliefs should be of high importance in
aeademie eontexts sueh as sehool mathematies.
In regard to value appraisals, we considered two categories addressed by the control-
value theory: (1) beliefs regarding the intrinsie values ofthe domain under eonsideration; and
(2) beliefs regarding the value of achievement outcomes in this domain (see also Eeeles et al.,
1983,
for a eoneeptual diseussion of subjeetive values of aeademie tasks). For example, a student
may like the subject of mathematics because he or she appreeiates the formal nature of
mathematies and the aetion of dealing with figures. We label this the (intrinsie) domain value
of mathematics. Another student may be more eoneemed with performing well in mathematies
in order to inerease eareer options, to meet parents' expeetations, or to attain personal
aehievement standards. We eall this achievement value of mathematies.
We consider these two core value dimensions to be eoneeptually orthogonal. That is,
even though these two types of values may be closely eonneeted in many individuals, there
may also be students who have high values of a domain, but regard aehievement in the same
domain as less important, or viee versa.
Regarding the effeets of these appraisals on achievement emotions, the control-value
theory would generally prediet that high eompetenee beliefs will be assoeiated with higher
levels of the positive emotions enjoyment and pride. Conversely, the belief that one is
ineompetent should be related to the negative emotions anxiety, hopelessness, and shame.
Concerning the effeets of value beliefs on aehievement emotions, the theory implies that
domain values should be positively related to activity-related emotions sueh as task enjoyment.
In eontrast, aehievement values should be positively related to any outeome emotion, positive
or negative (sueh as, pride, anxiety, hopelessness or shame). That is, if aehievement is
important, both positive and negative feelings relating to sueeess and failure are enhanced.
In sum, we expeeted that enjoyment should result from a pattem of appraisals combining
favorable judgments of competence with high domain values, whereas pride should result
from a eombination of favorable judgments of eompetenee and high aehievement values.
Anxiety, hopelessness, and shame, on the other hand, should result from low eompetenee
beliefs, eombined with high values of aehievement (Abela & Seligman, 2000; Hembree, 1988;
Pekrun, 1992).
Assumptions concerning the interplay of gender, prior achievement, beliefs, and emotions
in mathematics
Gender differences in beliefs and emotions and a mediation assumption. There is
eumulative evidence that girls tend to have lower eompetenee-related beliefs in mathematies
by the end of primary sehool (Hyde, Fennema, Ryan, et al., 1990). Therefore, we assumed that
girls would be less eonfident in their eompetenee judgments in mathematics in grade five, the
age group under study here. Furthermore, in line with the findings of
Hyde,
Fennema, Ryan, et
al.,
we expeeted girls to pereeive the subject of mathematies as less valuable, that is, having a
lower domain value. However, in reeent years adoleseent girls seem to have adopted similar
pereeptions as boys eoneerning the importanee of performing well in mathematies (e.g.
Meeee, Wigfield, &. Eeeles, 1990; Mullis, Martin, Fierros, Goldberg, & Stemler, 2000;
Wigfield et al., 1997). Thus, we expected that girls would have the same subjeetive value of
aehievement in mathematies as boys. In sum, girls were expected to judge their mathematics
eompetenee as being relatively low and the subjeet of mathematics as being unattraetive,
while being aware ofthe importanee of attaining good grades in this subjeet.
Applying the eontrol-value theory, we predieted that this female pattem of eompetenee
and value appraisals would lead to a debilitating emotional profile. This profile should be
eharaeterized by more anxiety, hopelessness and shame due to the girls' eomparably lower

EMOTIONS, GENDER, AND MATHEMATICS
501
control beliefs, but equally high achievement values as compared to boys. Furthermore, the
theory would also imply that girls would experience less enjoyment in mathematics, due to
their low competence beliefs and low domain values. Finally, in spite of expected similar lev-
els of achievement values which should be important for the formation of pride, we assumed
that the lower competence beliefs in girls reduce their experienced pride in relation to mathe-
matics. In other words, when they do not believe they are doing well in mathematics, girls
should have less to be proud of in this subject as compared to boys.
Thus,
the control-value theory suggests that gender differences in mathematics emotions
are due to girls' and boys' differential pattems of appraisals relating to mathematics. By
implication, we assumed that the effects of gender on mathematics emotions are mediated by
these appraisals (see Figure la).
Control and value appraisals
Emotions
Gender
.y^
B.a..iiLesS..effe,<S.
•^.
\ V
\ /•
-^
• Competence belief
• Domain value
• Achievement value
/
/
' Enjoyment
•Pride
• Anxiety
• Hopelessness
• Shame
Figure la. Effects of gender on achievement emotions by control and value beliefs: Mediation.
Prior achievement as a covariate. Since gender differences in mathematics achievement
have declined in recent years and tend to remain substantial only at the high school and university
levels (for meta-analyses, see Hyde, Fennema, & Lamon, 1990; Leahey & Guo, 2001; Linn &
Hyde, 1989), we did not expect any large differences in mathematics achievement in our target
population (grade five). However, prior academic achievement can be considered as an important
antecedent of individual competence beliefs, and, therefore, of students' emotional experiences.
In order to enstire that gender differences in beliefs and emotions were not confounded by prior
achievement, we considered it important to include this variable as a covariate.
Gender universality of relationships between variables. In the control-value theory, it is
assumed that the functional relationships between emotions, appraisals, and achievement are
universal across genders and cultures, with few exceptions (see Scherer, 1999). In the domain
of mathematics, universality assumptions are supported by findings reported by Meece et al.
(1990),
as well as by Seegers and Boekaerts (1996). Using path analytical procedures, Meece
et al. (1990) showed that the pattem of relationships between student attitudes (including
expectancies for success, subjective values, and perceived ability), math anxiety, and
achievement were very similar for boys and girls. Similarly, Seegers and Boekaerts (1996)
found no gender differences in the relations between self-related cognitions, task-specific
beliefs, and mathematics performance. In contrast, evidence suggesting that gender can
function as a moderator of relations between cognitive-affective variables is limited (see
Meyer & Koehler, 1990; Zeidner & Safir, 1989). Therefore, even though we did expect mean
level differences in beliefs and emotions between girls and boys, we assumed that the
relationships between prior achievement, students' beliefs, and their emotions would be
universal, that is, they should fiinction similarly across genders. By implication, we did not
expect that gender would function as a moderator of these relationships (see Figure lb).