See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233207824

Development and Construct Validation of an Academic

Emotions Scale

Article  in  International Journal of Testing · February 2008

DOI: 10.1080/15305050701808649

CITATIONS READS

35 890

2 authors:

Sophie Govaerts Jacques Grégoire

Université Catholique de Louvain - UCLouvain Université Catholique de Louvain - UCLouvain
2 PUBLICATIONS   62 CITATIONS    101 PUBLICATIONS   2,057 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Adherence to treatment in depression View project

Test Standards View project

All content following this page was uploaded by Jacques Grégoire on 15 April 2015.

The user has requested enhancement of the downloaded file.

International Journal of Testing, 8: 34–54, 2008
Copyright
C Taylor & Francis Group, LLC
ISSN: 1530-5058 print / 1532-7574 online
DOI: 10.1080/15305050701808649

Development and Construct Validation

of an Academic Emotions Scale
Sophie Govaerts and Jacques Grégoire
Faculté de psychologie et des sciences de l’éducation
Université Catholique de Louvain, Louvaine-la-Neuve, Belgium

This article describes the development and two studies on the construct validity of
the Academic Emotions Scale (AES). The AES is a French self-report questionnaire
assessing six emotions in the context of school learning: enjoyment, hope, pride,
anxiety, shame and frustration. Its construct validity was studied through exploratory
and confirmatory factor analyses. A six-factor structure emerged from exploratory
factor analyses on two different samples (N = 188; N = 516). These six factors
correspond to six academic emotions. Confirmatory factor analysis was conducted
to test the factorial structure of the final version of the AES. A hierarchical model
with a second-order factor (valence) underlying the six first-order emotional factors
was the most suited to the data. The psychometric characteristics of the AES justify
the use of this questionnaire in future research.

Keywords: assessment, emotion, learning, motivation, validity

Introducing a special issue of the Educational Psychologist on emotions in edu-

cation, Schutz and Lanehart (2002) noted:

As motivation, cognitive, developmental, and educational psychologists have

continued to contextualize their inquiry within the schools, it has become clear
that emotions are an integral part of educational activity settings. In the 2000s,
researchers interested in teaching, learning, and motivational transactions within the
classroom context can no longer ignore emotional issues. Emotions are intimately
involved in virtually every aspect of the teaching and learning process and, therefore,
an understanding of the nature of emotions within the school context is essential.
(p. 67)

Correspondence should be sent to Sophie Govaerts, Université Catholique de Louvain, Fac-

ulté de psychologie, Place du Cardinal Mercier, 10, 1348-Louvain-la-Neuve, Belgium. E-mail:
[email protected]
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 35

To reach such a goal, it is critical to have valid and theoretically grounded

instruments for measuring emotions. This article describes the development and
the construct validation of a French scale focused on the assessment of emotions
in learning contexts.
Emotions have long been a neglected topic in educational research. Two major
exceptions should be mentioned (Pekrun, Goetz, Titz, & Perry, 2002). The first
one is the research on test anxiety (for a recent review, see Zeidner, 1998) that for a
long time has dominated the research on emotions in education, overshadowing the
study of other emotions, pleasant or unpleasant. The second exception is Weiner’s
attribution theory (1985), in which emotions are seen as outcomes of attributional
processes.
Despite these two exceptions, research has been slow to understand emotions
in education. This situation changed recently when some American, European,
and Australian researchers started investigating the concept of emotion in learn-
ing (e.g., Ainley, Corrigan, & Richardson, 2003; Goetz, Zirngibl, Pekrun, & Hall,
2003; Schutz & Lanehart, 2002). A new concept was born: “academic emo-
tions.” According to Pekrun et al. (2002), the term “academic” is associated with
“emotions” as it is in expressions similar to “academic motivation” or “academic
self-concept.” The current research on academic emotions has three main charac-
teristics. The first is the importance accorded to a large range of emotions (both
pleasant and unpleasant) experienced by students in academic settings. The sec-
ond characteristic is the independence from the old conception opposing emotions
and reason. For centuries, philosophers and psychologists considered emotions as
inferior to cognition. Current research is, on the contrary, more interested in the
reciprocal interactions between emotion and cognition. The third characteristic of
the current research is the interest in emotions in all learning situations, which is
no longer restricted to testing and exam situations. Current research on academic
emotions is contextualized in a large range of situations because emotions are
specific to each learning context.
The study of academic emotions is promising and should help to a better under-
standing of teaching, motivation, and self-regulated learning. However, research
on academic emotions faces several challenges (Schutz & DeCuir, 2002). Two
major interrelated challenges have to be emphasized. The first is the operational
definition of the concept of academic emotions. The second is, based on this con-
ceptual definition, developing and validating instruments to measure academic
emotions.
A crucial issue affecting research on emotions is the lack of broad agreement
on terms such as feelings, emotion, and mood (Forgas, 2000; Frijda, 1986). One
commonly accepted view in the field of psychology of emotions defines affect as
the general positive or negative tone of an event and distinguishes between mood
and emotion (e.g., Forgas, 2000; Mayer, 1986). Moods can be defined as relatively
low intensity, diffuse, and enduring affective states that have no salient antecedent
36 GOVAERTS, GRÉGOIRE

cause and therefore little cognitive content. In contrast, distinct emotions are more
short-lived, intense phenomena and usually have a highly accessible and salient
cause, as well as a specific contextual referent (e.g., anger, enjoyment, pride;
Forgas, 2000; Linnenbrink & Pintrich, 2002; Rosenberg, 1998; Schwarz & Clore,
1996). As emotions fade over time, they may change into general mood states.
Emotions are thus short and intense subjectively experienced feeling states related
to a specific context. These states are expressed in particular physiological patterns
(Smith & Lazarus, 1990). Our own research is based on this definition of emotion.
This definition does not include motivational components. From our viewpoint,
having a definition of emotion that does not integrate motivational components
is crucial in research on academic emotions. A growing number of researchers
(Govaerts & Grégoire, 2008a, 2008b; Meyer & Turner, 2002; Schutz & DeCuir,
2002) are working on a major issue in educational research, that is to build a
model linking emotional, motivational, and cognitive components in learning and
achievement. For reaching such a goal, researchers need distinct definitions of
these concepts that allow for the development of clearly separate measures of each
construct.
Another important discussion in the literature relates to the structure underlying
emotional diversity. Three theoretical models have been proposed. The first one,
grounded on a biological basis (Ekman, 1992; Izard, 1994), postulated the exis-
tence of discrete basic emotions, which are mutually exclusive. The number of cat-
egories varied from six to ten (fear, joy, anger, etc.). The second theoretical model
hypothesized that smaller numbers of dimensions accounted for emotions. Circum-
plex models were created (e.g., Feldman Barrett & Russell, 1998; Russell, 1980;
Russell & Feldman Barrett, 1999), but a controversy remained about the number
of dimensions organizing emotions. With self-reported data, researchers found
that emotions can be plotted in a two-dimensional space, with one axis repre-
senting the hedonic value (positive versus negative or pleasant versus unpleasant)
and the other axis representing the level of activation1 (high versus low). There is
considerable agreement on the first factor (e.g., Green & Salovey, 1999; Russell
& Feldman Barrett, 1999), but the existence of an independent general activa-
tion dimension is more controversial (Schimmack & Grob, 2000; Watson, Wiese,
Vaidya, & Tellegen, 1999). The extraction of this activation dimension varies
across samples. Terracciano, McCrae, Hageman, and Costa (2003) found that in
people unfamiliar with the experience and expression of their feelings, the activa-
tion dimension is difficult to assess. In the academic domain, Pekrun et al. (2002)

1 The term “activation” will be used in this study to label this second dimension of the emotional

structure. It refers, at the level of subjective experience, to a sense of mobilization or energy (Russell
& Feldman Barrett, 1999). Alternatively, this has also been labelled the ‘arousal’ dimension (Feldman
Barrett & Russell, 1998).
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 37

suggested the existence of a two dimensional space underlying academic emotions

(valence: positive and negative emotions; and activation: activating and deactivat-
ing emotions). These two dimensions are supposed to be basic determinants of
many effects of emotions on achievement. To our knowledge, this two-dimensional
model of emotion has never been empirically tested in the academic area. Recently,
a third theoretical model was developed that tried to bring together categorical and
dimensional models in a hierarchical structure of emotions (Church, Katigbak,
Reyes, & Jensen, 1999; Russell & Feldman Barrett, 1999). This model postulates
a hierarchical organization of emotions in which specific emotions, such as joy
or anger, are underlaid by broad dimensions. This model needs further empirical
support.
The development and validation of instruments designed to assess academic
emotions is the second challenge with which researchers in this area are faced.
With the exception of some studies using a qualitative assessment of emotions
by way of interviews (deMarrais & Tisdale, 2002; Op’t Eynde, De Corte,
& Verschaffel, 2001), previous studies on academic emotions used general
quantitative measures of emotions without any references to the academic
context or rating scales assessing specific emotions with a single item. These two
measures were not appropriate. General measures of emotions are less predictive
of academic achievement than more domain-specific scales (Pekrun et al., 2002).
Measurements of emotions based on a single item (e.g., “I feel angry”) are
economic, but have a weak reliability.
A first well-designed instrument in the area of academic emotional assessment is
the Academic Emotions Questionnaire (AEQ; Pekrun et al., 2002). This instrument
was developed to measure nine relevant academic emotions2 in different contexts
and according to different time perspectives. The three contexts are being inside
the classroom, studying outside the classroom, and taking tests and exams. The
emotional items could be administered in three different time perspectives: to
measure trait academic emotions, course-related emotions, and state academic
emotions. Multiple items are used for each emotion to represent the affective,
cognitive, physiological, and motivational components of emotions.
This instrument was first developed in German and then translated into
English. To our knowledge, no validity studies were published about the AEQ
in its English version, except in the area of tests/exams in which a validation
of the Test Emotion Questionnaire (TEQ) was conducted (Pekrun et al., 2004).
Nevertheless, a theoretical article about academic emotions (Pekrun et al., 2002)
presented good reliability indices for the AEQ’s nine academic emotions scales.
From our viewpoint, the theoretical background of the AEQ raises two questions.

2 These academic emotions were: enjoyment, pride, hope, relief, anxiety, boredom, shame, anger,

and hopelessness.
38 GOVAERTS, GRÉGOIRE

Due to the multidimensional model of academic emotions underlying this scale,

including motivational and cognitive aspects, some items of the AEQ measure
motivational and cognitive characteristics. This is a problem if we want to study
the relationship between emotional and motivational variables. The second issue
is related to the broad meaning the authors gave to the concept of emotions. For
them, the whole affective domain could be studied under the label “emotion”:
trait academic emotions, course-related emotions, and state academic emotions.
As for us, we selected a definition of the concept of emotion (see below) restricted
to what the authors of the AEQ called “state academic emotion.” We agree that
affective traits can have an influence on emotions individuals reported in specific
contexts. But, for us, these affective traits are not emotions in their restricted fluid
and context-dependent meaning (Rosenberg, 1998; Schwarz & Clore, 1996).
Moreover, this restricted view of emotion as state academic emotions allows us
to answer a methodological problem in the measurement of emotions raised by
Goetz et al. (2003). These authors emphasized that it is often

rather unclear what exactly is being measured in terms of emotions, with respect to
the intensity and/or frequency of emotional experiences (which are not necessarily
linked, cf. Larsen & Diener, 1987). For example, when assessing emotions in a
questionnaire (e.g. “during tests, my hands get shaky”),3 intensity and frequency
are most likely interweaved (very shaky or often shaky?). On the other hand, when
assessing explicit emotional states (“My hands are shaky at the moment”), one can
be rather sure of assessing emotional intensity. (Goetz et al., 2003, p. 20)

The objective of the current studies was to present an instrument, the Academic
Emotions Scale (AES), which was developed to address the aforementioned
limitations. We present two studies on the construct validity of the AES with
adolescents in secondary schools. The first study reports the initial development
of the subscales and initial exploratory factor analysis (EFA). The second study
focuses on the final version of the subscales. In this study, exploratory and confir-
matory factor analyses (CFA) were conducted, providing strong construct validity
evidence.
Due to the context-dependent nature of emotion (Frijda, 1993; Goetz et al.,
2003), the two studies were conducted in a specific context: the situation of a math
exam preparation. This situation was chosen for two reasons: (1) given the central
focus on exams in the majority of Western educational schools, studying the role
of pleasant and unpleasant academic emotions in this context seems a key issue;
(2) this situation was not assessed per se in the AEQ (Pekrun et al., 2002).

3 That is to say not a state academic emotion.

 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 39

STUDY 1: INITIAL VERSION OF THE SCALE

Scale Development
Eight4 academic emotions, which were identified as the most relevant in an aca-
demic context (Pekrun et al., 2002), were selected to have three pleasant emotions
(enjoyment, hope, and pride) and five unpleasant emotions (anxiety, shame, anger,
boredom, and hopelessness) measured:. A first set of items designed to measure
these emotions was developed based on two major sources.
The first source encompassed items from the AEQ (Pekrun et al., 2002). Two
criteria were considered for the selection of those items. The first criterion was
to discard items measuring what Pekrun and colleagues (2002) called the motiva-
tional and cognitive components of emotion. As mentioned previously, emotion
and motivation are two constructs that are linked but conceptually distinct. For this
reason, they have to be measured separately. In the remaining items of the AEQ,
a second selection criterion was used. The items had to be relevant to the context
of exam preparation. This situation is not assessed per se in the AEQ, but two
assessment domains of this scale are related: “Studying outside the classroom” and
“Taking tests and exams.” Items from these two domains were selected with regard
to their relevance to the context of preparing for an exam. Those items selected from
the AEQ represented about 40% of the total scale. There were based on exploratory
studies done by Pekrun et al. (2002) that analyzed how students said they expressed
their emotions. This methodology provided ecological validity. Items were trans-
lated from their initial English version into French following the International Test
Commission (ITC) guidelines (Hambleton, 1994). Three separate fluent bilingual
psychologists translated the items. The three French versions were compared for
discrepancies, which were resolved by discussion to produce a final form.
The 60% remaining items were created by the authors based on a detailed
review of the literature on emotions (e.g., Forgas, 2000; Smith & Lazarus, 1990),
and on a linguistic analysis of how people express their emotions (Rosenthal,
1998). These items incorporated a number of terms and expressions that articulate
the variety of the subjective experience of the different emotions (Fainsilber &
Ortony, 1987; Galati & Sini, 1998). From the perspective of content validity,
emotions should be measured with terms that have the best chance of actually
being used by individuals to describe their feelings. The items were worded to
match the terms generally used by individuals, and especially with adolescents,
to express their subjective experience about particular emotions. This provides
ecological validity for our instrument (Hoffman, Waggoner, & Palermo, 1991).

4 The emotion ‘relief ’, included in the AEQ (Pekrun et al., 2002), was not measured in our scale

because it was not appropriate in a situation of testing preparation.

40 GOVAERTS, GRÉGOIRE

Some items reflected the affective component and some the physiological aspect
of the emotional language.
Fifty-two items constituted the first version of the scale. They were grouped as
follows: eight items measured enjoyment (e.g., “Studying my math course makes
me feeling great”), five items measured hope (e.g., “I’m optimistic when I think
about my math exam”), five items measured pride (e.g., “I’m proud of the way I
prepare for my math exam”), eight items measured anxiety (e.g., “When I prepare
for my math exam, I have a lump in my throat”), eight items measured shame
(e.g., “I feel ashamed when I couldn’t manage the exercises for my math exam
preparation”), six items measured anger (e.g., “I feel irritated when I do math
again for my exam”), six items measured boredom (e.g., “I find it boring to study
the math course”) and six items measured hopelessness (e.g., “I feel hopeless when
I prepare my math exam”).
The aims of this first study were to examine the structure of the scale in a
sample of secondary school students and to provide construct validity data. It
was expected that the scale would have eight factors corresponding to the eight
emotional subscales.

Subjects and Procedure

This first version of the scale was administered to 188 upper-secondary school stu-
dents (grades 10 to 12) during the period of the math exam preparation. Students’
mean age was 16.8 years (SD = 1.1), including 55% boys and 45% girls.
Students were asked to rate each of 52 items on a 7-point Likert-type scale
ranging from 1 (“I don’t agree at all”) to 7 (“I totally agree”). To be sure to measure
state academic emotions, two precautions were taken: (1) students completed the
questionnaire at the time at which they were actually studying for their math exam,
and (2) the following sentence preceded the items: “Now that I am studying for
my math exam. . .”.

RESULTS

Two items were removed from the pool before validity analysis because of a high
nonresponse rate (more than 20% nonresponse) due to badly worded formulation.
To investigate the construct validity of this first version of the scale, sev-
eral EFAs were performed. An unweighted least squares (ULS) procedure was
used because several items showed a skewed score distribution and a ULS pro-
cedure was the most suited for non-normal data (Nunnally & Bernstein, 1994).
It was followed by an oblimin rotation as the eight subscales were hypothe-
sized to be correlated (Pekrun et al., 2002). At each step of the analysis, items
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 41

were excluded for subsequent analysis respecting the following criteria (Field,
2000):

1. Before running the first EFA, the items were screened. First, the items with a
very low variance were deleted because they failed to discriminate between
the students. Second, the correlation matrix was used to check the patterns of
relationships between items. We excluded any items for which the majority
of significance values were greater than 0.05. These items presented very
few correlations with other items and yet this is an absolute condition for
running EFA.
2. At each step of the EFA, we excluded items that presented communalities
below 0.02.
3. Items with a factor pattern coefficients < 0.4 on any of the factors were
discarded.
4. Finally, we excluded complex items, that is items with multiple high factor
pattern coefficients (> 0.4) on several factors.

A six-factor solution emerged from the final EFA with 37 items. The Eigenval-
ues were 11.99, 5.48, 2.63, 1.80, 1.33, 1.02. This six-factor solution accounted for
56% of the total variance and presented a very good Kaiser-Meyer-Olkin measure
of sampling adequacy (.909). Five emotions emerged clearly as factors: Shame
(five items), Hope (five items), Pride (four items), Enjoyment (five items), and
Anxiety (six items). The items of the boredom and anger subscales loaded on the
same factor (eight items). The items from the hopelessness scale loaded on two
unpleasant emotional factors: Anxiety and Boredom/Anger.

DISCUSSION

The six-factor solution that emerged did not match the eight-factor structure that
we postulated. In particular, three factors postulated within the AEQ (Pekrun
et al., 2002) did not emerge. The items of the anger and boredom subscales
did not appear as separate factors and the items of the hopelessness subscale
loaded on two other unpleasant emotional factors: Anxiety and Boredom-Anger.
But, these results did not really contradict the results of Pekrun et al. (2002).
Examining the interrelationships between different emotions by cluster analysis,
those authors reported a specific cluster bringing together boredom and anger.
They also observed that the items measuring hopelessness were embedded in a
cluster, including hopelessness and other unpleasant emotions. The within-cluster
correlation between anxiety and hopelessness was particularly high (r > .70) and
above the respective reliabilities of these two emotional scales. Nevertheless, the
authors decided to keep these scales separate, “because anxiety and hopelessness
may differentially relate to outcomes” (Pekrun et al., 2002, p. 96).
42 GOVAERTS, GRÉGOIRE

As the discrepancy between the observed structure in Study 1 and the structure
that we have postulated could be due to some poor items, having inappropriate
content or an unclear wording, we decided to conduct a second validity study after
modifying the weakest items of the original version.

STUDY 2: FINAL VERSION OF THE SCALE

Item Modifications
The hopelessness subscale that did not emerge from the EFA in Study 1 was
strengthened by the addition of new items. Some items from the boredom and
anger subscales were modified. New items were added to the three emotional
subscales (pride, enjoyment, and shame) that presented the weakest item factor
pattern coefficients on the EFA in the first study. The second version of the scale
included 56 items.

Subjects and Procedure

This second version of the scale was administered to 516 upper-secondary school
students5 (grades 10 to 12) during a class period. This sample included 53% boys
and 47% girls. Students’ average age was 16.8 years (SD = 1.1).
Students were asked to rate each of the 56 items on a 7-point Likert-type scale
ranging from 1 (“I don’t agree at all”) to 7 (“I totally agree”).

RESULTS

Exploratory Factor Analysis and Reliability Analysis

EFA was used to examine the factor structure of the scale. To produce the final
factorial solution from the initial 56 items, several EFAs were performed on the
data using ULS method followed by oblimin rotation.
The screen test of the final EFA suggested the extraction of six factors. The
Eigenvalues of the first six factors were 8.69, 3.72, 1.88, 1.35, 1.07, 1.02. This
solution accounted for 57% of the total variance and showed a very good Kaiser-
Meyer-Olkin measure of sampling adequacy (.924). Table 1 presents the factor
pattern coefficients of the 26 items retained after the analysis. These items are
presented in the Appendix. The first factor (Anxiety) encompassed five items of
the anxiety subscale. Three items from the boredom subscale and three items
from the anger subscale loaded on the same second factor (Boredom/Anger). The
third factor, labeled Enjoyment, consisted of four enjoyment items. Four items of

5 These students were different from the 188 students of the Study 1.
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 43

TABLE 1
Factor Pattern Coefficients of the Exploratory Factor Analysis (EFA) For the Six Factors of
the Final Version of The Academic Emotions Scale (26 Items) (N = 516)

Item Factor

Category No. Anxiety Boredom-Anger Enjoyment Hope Shame Pride

Anxiety 4 0.76 0.17 −0.08 −0.21 0.12 0.04

3 0.77 0.14 −0.17 −0.18 0.17 0.06
2 0.76 0.23 −0.05 −0.24 0.21 0.05
5 0.73 0.19 −0.01 −0.13 0.18 −0.02
1 0.72 0.24 −0.11 −0.18 0.22 0.05
Boredom 1 0.23 0.73 −0.16 −0.25 0.09 −0.02
3 0.07 0.71 −0.28 −0.14 0.08 −0.09
2 0.10 0.68 −0.22 −0.03 0.16 −0.06
Anger 4 0.23 0.74 −0.18 −0.10 0.11 −0.08
6 0.39 0.58 −0.07 −0.20 0.08 −0.04
5 0.35 0.63 −0.18 −0.11 0.06 0.06
Enjoyment 2 −0.10 −0.34 0.71 0.29 0.03 0.19
3 −0.08 −0.35 0.69 0.17 0.03 0.26
1 −0.01 −0.29 0.67 0.18 0.13 0.13
4 −0.12 −0.30 0.53 0.20 0.04 0.23
Hope 4 −0.33 −0.11 0.28 0.74 −0.12 0.09
1 −0.20 −0.18 0.31 0.70 −0.03 0.19
2 −0.36 −0.21 0.18 0.65 −0.15 0.09
3 −0.21 −0.21 0.05 0.49 0.02 0.28
Shame 3 0.10 0.18 0.02 −0.05 0.73 0.02
4 0.23 0.02 −0.04 −0.05 0.67 0.09
1 0.39 0.08 0.11 −0.03 0.45 0.15
2 0.28 0.13 0.22 −0.05 0.42 0.08
Pride 2 0.10 −0.02 0.05 0.08 0.24 0.68
3 0.09 −0.05 0.28 0.07 0.06 0.54
1 −0.08 −0.08 0.19 0.24 −0.06 0.50

Note. Exploratory factor analysis was performed using an Unweighted Least Square method and
an oblique rotation. The highest factor pattern coefficients are printed in bold.

the hope subscale formed the fourth factor (Hope). The fifth factor included four
items of the shame subscale (Shame). Finally, the sixth factor consisted of three
items from the pride subscale (Pride). These remaining 26 items had their highest
coefficients on their respective factors ranging from .42 to .76.
This six-factor solution confirmed the results of the factor solution that emerged
from the first version of the scale. The items of the boredom and anger subscales
loaded again on the same factor. The correlation between these two subscales was
particularly high (r = 0.74). As these results were consistent across different sam-
ples in Studies 1 and 2, we decided to create a single subscale, labeled “frustration,”
that gathered the boredom and anger items. Again, the hopelessness items loaded
44 GOVAERTS, GRÉGOIRE

TABLE 2
Correlations among the 6 Emotional Factors Yielded By the EFA (N = 516)

1 2 3 4 5 6

Enjoyment 1̧ – – – – –
Hope 0̧.54∗∗ 1̧ – – – –
Pride 0̧.40∗∗ 0̧.30∗∗ 1̧ – – –
Anxiety −0.29∗∗ −0.54∗∗ 0̧.06 1 – –
Boredom-Anger −0.57∗∗ −0.49∗∗ −0.11∗∗ 0.54∗∗ 1 –
Shame 0̧.04 −0.21∗∗ 0̧.21∗∗ 0.49∗∗ 0.29∗∗ 1

∗∗ p ≤ 0.01.

weakly on two unpleasant emotional subscales: anxiety and boredom-anger. As

EFA conducted in Study 1 and 2 did not confirm the existence of this subscale, we
decided to remove these items and to give up the measurement of hopelessness.
The six retained subscales showed a good internal consistency. The standardized
Cronbach’s alpha coefficients were respectively .91 for the anxiety subscale, .88
for the frustration subscale, .87 for the enjoyment subscale, .85 for the hope
subscale, .73 for the shame subscale, and .65 for the pride subscale. Such indices
are excellent, especially with rather small subscales comprised of three to six items
(Cortina, 1993). The lowest alpha was observed in the shortest subscale, pride,
which included only three items.
Table 2 shows the intercorrelations among the six emotional factors extracted
with the EFA. These correlations justified the use of an oblique rotation, rather than
an orthogonal one. Positive correlations between pleasant emotions and unpleasant
emotions, and negative correlations between pleasant emotions and unpleasant
emotions were observed. But the correlation pattern is more complex. Enjoyment
and Shame, and Pride and Anxiety showed null correlations, while Shame and
Pride showed a slight, but statistically significant, positive correlation.

Confirmatory Factor Analysis

Because the six emotions measured by the AES were correlated, the structure of
these relationships was analyzed through a CFA on the second sample data (N =
516). The CFA was conducted with the LISREL 8.3 program (Jöreskog & Sörbom,
1993) to compare the goodness-of-fit statistics for the three following models.
These models represented respectively the dimensional (model 1), categorical
(model 2), and hierarchical (model 3) approaches of the organization of emotions.

Model 1 (dimensional approach): One bipolar first-order factor (unpleasant-

ness/pleasantness) underlying the 26 items. We postulated that the 15 items
measuring pleasant emotions were loading positively on this factor and,
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 45

conversely, that the 11 items measuring unpleasant emotions were loading

negatively on the same factor.
Model 2 (categorical approach): Six first-order factors underlying the 26 items
and representing six independent emotions.
Model 3 (hierarchical approach): Six first-order factors (six emotions) underlying
the 26 items and a second-order factor (unpleasantness/pleasantness) underly-
ing the six first-order factors. We postulated that the three pleasant academic
emotions (enjoyment, hope, pride) were loading positively on this second-order
factor and, conversely, that the three unpleasant emotions (anxiety, frustration,
and shame) were loading negatively on the same factor.

In this study, testing the activation dimension (in model 1 as an additional

bipolar first-order factor and in model 3 as an additional bipolar second-order
factor) was impossible. Only one of the seven academic emotions pertained to the
negative pole of the activation dimension; i.e., a deactivating emotion (boredom).
The six other emotions were activating emotions (Pekrun et al., 2002). Testing
such an unbalanced model was impossible.
Because the item scores were non-normally distributed ordinal variables, the
CFA was conducted on the polychoric correlation coefficient matrix associated
with the corresponding asymptotic covariance matrix (Jöreskog & Sörbom, 1993).
Four fit statistics were considered: the chi-square, the adjusted goodness-of-fit
index (AGFI), the root mean square of approximation (RMSEA), and the
comparative fit index (CFI). The chi-square fit index measures the difference
between the sample covariance matrix and the predicted covariance matrix.
The chi-square is more a badness-of-fit measurement because the lower the
chi-square, the better the model fit. A no-statistically significant chi-square value
(p < .05) indicates a good model fit. Unfortunately, chi-square is very sensitive
to the sample size. Moreover, chi-square is not penalized for lack of parsimony:
the more complex the model, better the fit. The AGFI is a fit measure reducing
dependence on sample size. It is an adjustment of the goodness-of-fit index for
the degrees of freedom relative to the number of variables. Varying from 0 to 1, an
AGFI value ≥ .90 indicates a good model fit. The RMSEA introduces a correction
for lack of parsimony since, all other things being equal, more complex models
are penalized. An RMSEA value ≤ .08 is a reasonable error of approximation of
the population (Byrne, 1998). The CFI compares the model fit to a baseline model
where the latent variables are uncorrelated. The CFI index varies from 0 to 1. A
value between .92 and .95 is considered well-fitting (Byrne, 1998; Hu & Bentler,
1999).
Table 3 shows the goodness-of-fit statistics for the three models. Model 1
(dimensional emotional structure) and model 2 (categorical emotional structure)
were clearly the less suited to the data. None of their goodness-of-fit indices
(except CFI for model 2) reached a sufficient level. On the other hand, model 3
46 GOVAERTS, GRÉGOIRE

TABLE 3
Goodness-Of-Fit Indices for Confirmatory Factor Analysis (N = 516)

χ2 df AGFI RMSEA 90% C.I. CFI

Model 1 2541.53 299 0.89 0.120 0.120–0.130 0.83

Model 2 2372.04 299 0.69 0.086 0.082–0.091 0.95
Model 3 1236.47 293 0.94 0.079 0.075–0.084 0.93

AGFI = adjusted goodness-of-fit; RMSEA = root mean square error of approximation;

90% C.I. = 90% confidence interval for RMSEA; CFI = comparative fit index.

(hierarchical emotional structure) showed good fit indices. Its AGFI, RMSEA, and
CFI indices reached a good model fit level.
To determine which model fit the data significantly better than the other mod-
els, the chi-square difference was computed between model 1 and model 2, and
between model 2 and model 3. These chi-square differences have a chi-square dis-
tribution with degrees of freedom equal to the difference between the two models
being compared. The two differences are presented in Table 4. Both are statistically
significant (p < .01). From these results, we conclude that model 3 fit the data sig-
nificantly better than the two other models. For this model, the second-order factor
was clearly a bipolar unpleasant-pleasant latent dimension. Unpleasant emotions
covaried negatively with this factor (−0.83 for Anxiety, −0.90 for Frustration,
and −0.46 for Shame), while pleasant emotions covaried positively with it (0.88
for Enjoyment, 0.81 for Hope, and 0.48 for Pride). Model 3, representing a hier-
archical emotional structure, was the most theoretically meaningful model, taking
into account the correlations between emotions and organizing these relationships
according to a theoretically well-founded valence dimension.

DISCUSSION

In Studies 1 and 2, a clear six-factor structure consistently emerged from ex-

ploratory factor analysis on two different samples. This six-factor solution did not
confirm the eight-factor structure that we have postulated. Three factors postulated

TABLE 4
Chi-Square Difference Tests Contrasting the Models
from the CFA

Models Compared χ 2 Difference df Difference

1 vs. 2 1559.18∗∗ 15
2 vs. 3 254.12∗∗ 9

∗∗ p < .01.
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 47

within the literature, and included in the AEQ (Pekrun et al., 2002), did not emerge.
The items measuring anger and boredom did not appear as separate factors and the
items measuring hopelessness loaded on two other unpleasant emotional subscales,
anxiety and boredom-anger. These results agree with the observations made by
Pekrun et al. (2002). Cluster analysis performed by these authors showed that anger
and boredom were in the same cluster and that hopelessness was in a cluster with
others AFSEHP unpleasant emotions. Despite these results, these authors decided
to keep separate scales measuring anger, boredom, and hopelessness, “because
anxiety and hopelessness may differentially relate to outcomes” (Pekrun et al.,
2002, p. 96). We made the same choice at the end of Study 1, deciding to keep eight
emotional subscales and improving the content of the weakest subscales. Neverthe-
less, Study 2 confirmed a six-factor structure. It is very unlikely that these results
are a consequence of ambiguous and improperly written items. Most of the items
included the common name of the tagged emotion, or a very close name or expres-
sion (see Appendix). Consequently, we consider that the observed correlations
between boredom and anger reflect an in-depth relationship between these two
emotions. A phenomenological analysis of both emotions helps to clarify this rela-
tionship. Boredom is an unpleasant emotion experienced in a frustrating situation.
In the classroom setting, this emotion arises when the activity does not stimulate
any interest and frustrates any potential pleasure. The student has to wait until
the end of this activity before having the opportunity to experience any possible
pleasure. The student who is bored in the classroom, seems passive, but most often
expresses a low aggressiveness level: refusing to participate, disturbing the lesson,
bothering the other students, tagging the desk, etc. This behavior may turn into
more an aggressive behavior against the teacher, the material, or the other students.
Anger is also an unpleasant emotion experienced in a frustrating situation, but an
immediate expression of a high-aggressiveness level. In the classroom, anger is
usually caused by a sudden and high frustration: failure, punishment, etc. But its
outcomes can be close to some outcomes produced by boredom (e.g., vandalism).
Consequently, we consider boredom and anger as close emotional reactions to
frustration (e.g., Ellsworth & Smith, 1988; Frijda, 1993; Smith & Lazarus, 1990),
being only different in the level of aggression. They are at both ends of an aggres-
sion continuum, boredom being associated with a low level of aggressiveness,
and anger being associated with high aggressiveness. Other research has been
carried out to fully understand the gathering of these two emotions in an academic
context.
As in Studies 1 and 2 hopelessness items were inconsistently loaded on other
unpleasant subscales, we decided to remove this subscale from AES. In doing so,
we do not deny the existence of hopelessness, but we consider it is difficult, if
not impossible, to measure it as a short-lived emotion. Hopelessness seems to be
a mood more than an emotion. Students seem to feel hopelessness in relation to
general failure and not to a specific task.
48 GOVAERTS, GRÉGOIRE

A CFA was conducted to test the factorial structure of the six emotions mea-
sured by the final version of the AES. Three models were tested, representing
respectively a dimensional, categorical, and hierarchical emotional structure: (1) a
bipolar one-factor model underlying pleasant and unpleasant emotions; (2) a six-
factor model underlying the 26 items and representing six independent emotions;
and (3) a hierarchical model with a second-order factor underlying the six emo-
tional factors. According to the circumplex model of emotions (e.g., Green &
Salovey, 1999; Tellegen, Watson, & Clark, 1999), this second-order factor was
considered as a bipolar unpleasant/pleasant dimension related positively to enjoy-
ment, hope and pride, and negatively to anxiety, frustration and shame. None of the
goodness-of-fit indices of model 1 (dimensional emotional structure) and model
2 (categorical emotional structure) reached a sufficient level. The goodness-of-fit
statistics and the chi-square difference tests indicated otherwise good model fit
for model 3 (hierarchical emotional structure). These results show the superiority,
for a better understanding of the emotional structure of students, of a model that
brings together a categorical (model 2) and a dimensional (model 1) approach of
emotions. These categorical and dimensional approaches are linked in a hierarchi-
cal perspective (Church et al., 1999; Russell & Feldman Barrett, 1999), organizing
the correlations between emotions according to a theoretically well-founded va-
lence dimension. Consequently, results emphasize, on one hand, that a simple
dimensional distinction between pleasant and unpleasant emotions (model 1) does
not adequately convey the emotional structure of students. This is in line with
recent findings in academic emotions research that show a simple dimensional
distinction between pleasant and unpleasant emotions is too poor to account for
the effect of emotions on other variables (Goetz et al., 2003). On the other hand, a
strict categorical structure (model 2) that does not account for dimensional aspects
of emotions is not more adequate.
Regarding the dimensional structure of emotions, our data confirmed the va-
lence dimension of emotions, at least in a hierarchical perspective. The activation
dimension was not tested in this study. It should nevertheless be mentioned that
results revealed that one theoretically deactivating (boredom) and one activating
(anger) emotion formed a unique factor. Appraisal models gave a better explana-
tion of the gathering of these two emotions than a dimensional one. This empha-
sizes, as noted by Russell and Feldman Barrett (1999), the importance of taking
into account the appraisal component to understand the emotional structure.
The pride and the shame subscales could be improved. The results of the EFA
and CFA showed that they formed clearly separate factors, but their reliability
was lower than the other subscales, although reaching an acceptable level. The
reliability level of these two subscales could be partially due to a methodological
issue. As noted by Goetz et al. (2003), when assessing student’s emotions it is
not easy to know if they are able to verbalize their emotions or even if there are
conscious of them. Thus, the emotional measurement could reflect the student’s
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 49

implicit theories about emotions more than what they actually feel. This could be
particularly true for emotions such as pride and shame, which are more difficult
to recognize and express than anxiety or enjoyment.
The context of development of the AES was an exam preparation. We hypoth-
esized that the psychometric characteristics could be replicated in other academic
contexts. In another study (Govaerts & Grégoire, 2008a), the AES was contextual-
ized to measure academic emotions during a logical reasoning task. EFA conducted
on the data of 246 adolescents confirmed the six-factor structure that emerged in
the context of an exam preparation. Anger and boredom items loaded again on
the same factor. These results showed the possibility of using the AES in different
academic areas, but it should be confirmed in further studies in other academic
contexts.
The psychometric characteristics of the AES justify the use of this questionnaire
in future research. Because the AES specifically measures academic emotions,
without any reference to motivation, this instrument can be used to study the
relationships between emotional and motivational components of learning and
achievement.

REFERENCES

Ainley, M., Corrigan, M., & Richardson, N. (2003, August). Students, tasks and emotions: Identifying
the contribution of emotions in students’ learning. Paper presented at 10th Biennial Conference of
the European Association for Research on Learning and Instruction, Padova, Italy.
Byrne, B. (1998). Structural equation modeling with LISREL, PRELIS, and SIMPLIS: Basic applica-
tions and programming. Mahwah, NJ: Lawrence Erlbaum Associates.
Church, A. T., Katigbak, M. S., Reyes, J. A. S., & Jensen, S. M. (1999). The structure of affect
in a non-Western culture: Evidence for cross-cultural comparability. Journal of Personality, 67,
505–534.
Cortina, J. M. (1993). What is coefficient alpha? An examination of theory and applications. Journal
of Applied Psychology, 78, 98–104.
deMarrais, K., & Tisdale, K. (2002). What happens when researchers inquire into difficult emotions?
Reflections on studying women’s anger through qualitative interviews. Educational Psychologist,
37, 91–105.
Ekman, P. (1992). An argument for basic emotions. Cognition and Emotion, 6, 169–200.
Ellsworth, P. C., & Smith, C. A. (1988). From appraisal to emotion: Differences among unpleasant
feelings. Motivation and Emotion, 12, 271–302.
Fainsilber, L., & Ortony, A. (1987). Metaphorical uses of language in the expression of emotions.
Metaphor and Symbolic Activity, 2, 239–250.
Feldman Barrett, L. F., & Russell, J. A. (1998). Independence and bipolarity in the structure of current
affect. Journal of Personality and Social Psychology, 74, 967–984.
Field, A. (2000). Discovering statistics using SPSS for Windows. London: Sage.
Forgas, J. P. (2000). Feeling and thinking: The role of affect in social cognition. New York: Cambridge
University Press.
Frijda, N. H. (1986). The emotions. New York: Cambridge University Press.
50 GOVAERTS, GRÉGOIRE

Frijda, N. H. (1993). The place of appraisal in emotion. Cognition and Emotion, 7, 357–387.
Galati, D., & Sini, B. (1998). Echelonnement multidimensionnel de termes du lexique français
des émotions: Une comparaison entre trois procédés d’analyse [Multidimensional scaling of the
emotional lexicon in French language]. Cahiers Internationaux de Psychologie Sociale, 37, 76–
96.
Goetz, T., Zirngibl, A., Pekrun, R., & Hall, N. (2003). Emotions, learning, and achievement from
an educational psychological perspective. In P. Mayring, & C. Von Rhoeneck (Eds.), Learning
emotions: The influence of affective factors on classroom learning (pp. 9–28). Frankfurt: Peter Lang
Publishing.
Govaerts, S., & Grégoire, J. (2008a). Relationships between motivational and emotional variables at
a mathematical reasoning task and their influence on task performance. Manuscript submitted for
publication.
Govaerts, S., & Grégoire, J. (2008b). The effect of achievement goals on academic emotions and task
performance at a mathematical reasoning task. Manuscript submitted for publication.
Green, D. P., & Salovey, P. (1999). In what sense are positive and negative affect independent?
Psychological Science, 10, 304–306.
Hambleton, R. K. (1994). Guidelines for adapting educational and psychological tests: A progress
report. European Journal of Psychological Assessment, 10, 229–244.
Hoffman, R. R., Waggoner, J. E., & Palermo, D. S. (1991). Metaphor and context in the language of
emotion. In R. R. Hoffman, & D. S. Palermo (Eds.), Cognition and the symbolic processes: Applied
and ecological perspectives (pp. 163–185). Hillsdale, NJ: Lawrence Erlbaum Associates.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Con-
ventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.
Izard, C. E. (1994). Innate and universal facial expressions: Evidence from developmental and cross-
cultural research. Psychological Bulletin, 115, 288–299.
Jöreskog, K. G., & Sörbom, D. (1993). LISREL 8 user’s reference guide. Chicago: Scientific Software
International.
Linnenbrink, E. A., & Pintrich, P. R. (2002). Achievement goal theory and affect: An asymmetrical
bi-directional model. Educational Psychologist, 37, 69–78.
Mayer, J. D. (1986). How mood influences cognition. In N. E. Sharkey (Ed.), Advances in cognitive
science (Vol. 1, pp. 290–314). Chichester, UK: Ellis Horwood.
Meyer, D. K., & Turner, J. C. (2002). Discovering emotion in classroom motivation research. Educa-
tional Psychologist, 37, 107–114.
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory. New York: McGraw-Hill.
Op’t Eynde, P., De Corte, E., & Verschaffel, L. (2001). “What to learn from what we feel?”: The role
of students’ emotions in the mathematics classroom. In S. Volet, & S. Järvelä (Eds.), Motivation in
learning contexts: Theoretical advances and methodological implications (pp. 149–167). Elmsford,
NY: Pergamon Press.
Pekrun, R., Goetz, T., Perry, R. P., Kramer, K., Hochstadt, M., & Molfenter, S. (2004). Beyond test
anxiety: Development and validation of the Test Emotions Questionnaire (TEQ). Anxiety, Stress and
Coping, 17, 287–316.
Pekrun, R., Goetz, T., Titz, W., & Perry, R. P. (2002). Academic emotions in students’ self-regulated
learning and achievement: A program of qualitative and quantitative research. Educational Psychol-
ogist, 37, 91–105.
Rosenberg, E. L. (1998). Levels of analysis and the organization of affect. Review of General Psychol-
ogy, 2, 247–270.
Rosenthal, L. H. (1998). Talking about feelings: The language of emotion and its relationship to
physiology. Unpublished Ph.D. dissertation. University of California, Berkeley, California.
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39,
1161–1178.
 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 51

Russell, J. A., & Feldman Barrett, L. (1999). Core affect, prototypical emotional episodes, and other
things called emotion: Dissecting the elephant. Journal of Personality and Social Psychology, 76,
805–819.
Schimmack, U., & Grob, A. (2000). Dimensional models of core affect: A quantitative comparison by
means of structural equation modeling. European Journal of Personality, 14, 325–345.
Schutz, P. A., & DeCuir, J. T. (2002). Inquiry on emotions in education. Educational Psychologist, 37,
125–134.
Schutz, P. A., & Lanehart, S. L. (2002). Introduction: Emotions in education. Educational Psychologist,
37, 67–68.
Schwarz, N., & Clore, G. L. (1996). Feelings and phenomenal experiences. In E. T. Higgins, & A.
Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 433–465). New York:
Guilford Press.
Smith, C. A., & Lazarus, R. S. (1990). Emotion and adaptation. In L. A. Pervin (Ed.), Handbook of
personality: Theory and research (pp. 609–637). New York: Guilford Press.
Tellegen, A., Watson, D., & Clark, L. A. (1999). On the dimensional and hierarchical structure of
affect. Psychological Science, 10, 297–303.
Terracciano, A., McCrae, R. R., Hageman, D., & Costa, P. T. (2003). Individual differences vari-
ables, affective differentiation, and the structure of affect. Journal of Personality, 71, 669–
703.
Watson, D., Wiese, D., Vaidya, J., & Tellegen, A. (1999). The two general activation systems of
affect: Structural findings, evolutionary considerations, and psychobiological evidence. Journal of
Personality and Social Psychology, 76, 820–838.
Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological
Review, 92, 548–573.
Zeidner, M. (1998). Test anxiety: The state of the art. New York: Plenum Press.

APPENDIX

This appendix provides the items of the final French version of the Academic Emotions
Scale (26 items). An English translation of these items is proposed below. This translation
is given only for non-French speakers to understand of the French version of the items. The
English version was not validated and, consequently, cannot be used for research or clinical
purposes. Validity studies of this version are still needed.

Academic Emotions Scale–French version

Maintenant que je suis en train d’étudier mon examen de mathématiques. . .

FRUSTRATION

Ennui
1. Je trouve rasant d’étudier le cours de mathématiques
2. Quand j’étudie pour l’examen de mathématiques, j’éprouve un sentiment de lassitude
3. Je ressens de l’ennui quand je refais des exercices en vue de l’examen de
mathématiques
52 GOVAERTS, GRÉGOIRE

Colère
4. La matière à étudier en mathématiques m’énerve
5. Quand je suis assis(e) à mon bureau pour un long moment à étudier des maths, je
jetterais bien mon cours par la fenêtre
6. Je me sens agacé(e) quand je refais des exercices en vue de l’examen de
mathématiques

ANXIÉTÉ

1. Je me sens anxieux(se) quand j’étudie pour l’examen de mathématiques

2. Quand je pense à la matière que j’ai à étudier en mathématiques, je deviens stressé(e)
3. Quand je prépare mon examen de mathématiques, je me sens angoissé(e)
4. Quand je pense à l’examen de mathématiques, j’ai une boule dans la gorge
5. Je me sens anxieux(se) quand je refais des exercices en vue de l’examen de
mathématiques

HONTE

1. Je suis gêné(e) à l’idée de mal préparer l’examen de mathématiques

2. Je me sens honteux(se) quand je n’arrive pas à refaire des exercices en vue de
l’examen de mathématiques
3. Je me sens honteux(se) de la faible qualité de ma préparation de l’examen de
mathématiques
4. Je me sens honteux(se) de ne pas m’y être pris(e) plus tôt pour étudier mon cours de
mathématiques

PLAISIR

1. Etudier mon cours de mathématiques me donne la pêche

2. Je prends plaisir à travailler la matière de mathématiques en vue de l’examen
3. Je prends plaisir à refaire des exercices de math en vue de la préparation de l’examen
4. Je prends du plaisir à résoudre des exercices difficiles en préparant l’examen de
mathématiques

ESPOIR

1. Je me sens optimiste par rapport à l’étude de l’examen de mathématiques

2. Je ressens que tout marchera bien pour moi à l’examen de mathématiques
3. Au vu de la manière dont j’arrive à refaire les exercices de mathématiques, je me
sens confiant(e) pour l’examen
4. Je me sens optimiste quand je pense à l’examen de mathématiques

FIERTÉ

1. Je suis fier(ère) de la manière dont je prépare l’examen de mathématiques

 VALIDATION OF AN ACADEMIC EMOTIONS SCALE 53

2. J’éprouve de la fierté à l’idée que quelqu’un remarque la qualité de ma préparation

de l’examen de mathématiques
3. Je me sens fier(ère) de moi quand je refais des exercices en vue de l’examen de
mathématiques

Academic Emotions Scale–English version

Now that I am studying for my math exam . . .

FRUSTRATION

Boredom
1. I am bored studying for the math exam
2. When I am preparing the math exam, I feel weary
3. It is boring doing exercises for the math exam

Anger
4. The math subjects I have to study for the exam irritate me
5. When I am sitting on my desk for a long time preparing the math exam, I could
throw my notes through the window
6. I feel irritated when doing exercises for preparing the math exam

ANXIETY

1. I feel anxious when I study for the math exam

2. When I think about the math subjects I have to study, I become anxious
3. When I am preparing for the math exam, I become anxious
4. When I think about the math exam, I have a lump in my throat
5. I feel anxious when I am doing exercises for the math exam

SHAME

1. I feel ashamed thinking I might have not prepared the math exam properly
2. I am ashamed when I cannot do exercises for preparing the math exam
3. I am ashamed of the poor quality of my preparation for the math exam
4. I feel ashamed not having started preparing the math exam earlier

ENJOYMENT

1. I feel great when I study the math exam

2. I am pleased working on math subjects for the preparing the exam
3. I am pleased doing exercises for preparing the math exam
4. I am happy solving hard problems when I prepare the math exam
54 GOVAERTS, GRÉGOIRE

HOPE

1. I feel optimistic about the preparation of the math exam

2. I am sure that everything will go well at the math exam
3. I am confident for the math exam when I see how I solve the exercises
4. I am optimistic when I think about the math exam

PRIDE

1. I am proud of the way I am preparing the math exam

2. I feel proud thinking of somebody noticing the quality of my preparation for the
math exam
3. I feel proud when I do exercises with my math exam in mind
View publication stats