Sunteți pe pagina 1din 28

VALIDATION OF A SHORT FRENCH STATE TEST WORRY AND

EMOTIONALITY SCALE
Marine Beaudoin et Olivier Desrichard

Presses universitaires de Grenoble | « Revue internationale de psychologie sociale »

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
2009/1 Tome 22 | pages 79 à 105
ISSN 0992-986X
ISBN 9782706115356
Article disponible en ligne à l'adresse :
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

--------------------------------------------------------------------------------------------------------------------
https://www.cairn.info/revue-internationale-de-psychologie-
sociale-2009-1-page-79.htm
--------------------------------------------------------------------------------------------------------------------

Distribution électronique Cairn.info pour Presses universitaires de Grenoble.


© Presses universitaires de Grenoble. Tous droits réservés pour tous pays.

La reproduction ou représentation de cet article, notamment par photocopie, n'est autorisée que dans les
limites des conditions générales d'utilisation du site ou, le cas échéant, des conditions générales de la
licence souscrite par votre établissement. Toute autre reproduction ou représentation, en tout ou partie,
sous quelque forme et de quelque manière que ce soit, est interdite sauf accord préalable et écrit de
l'éditeur, en dehors des cas prévus par la législation en vigueur en France. Il est précisé que son stockage
dans une base de données est également interdit.

Powered by TCPDF (www.tcpdf.org)


MEP_1_2009 8/04/09 10:25 Page 79

RIPS / IRSP, 22 (1), 79-105 © 2009, Presses universitaires de Grenoble

Validation of a Short French State Test Worry and


Emotionality Scale
Validation d’une échelle courte d’anxiété d’évaluation état en français

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Marine Beaudoin*
Olivier Desrichard*
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

Abstract Résumé Key-words


Test anxiety, worry,
The objective of this research was L’objectif de cette recherche était emotionality, test
to create and validate a short French de construire et valider une échelle performance, scale
state test worry and emotionality courte d’anxiété d’évaluation état validation
scale that can be used in all test en français qui puisse être utilisée Mots-clés
settings: the “Echelle d’Anxiété dans toute situation d’évaluation : Anxiété d’évaluation,
d’Evaluation Etat” (EAEE; state test l’Échelle d’Anxiété d’Évaluation État anxiété cognitive,
anxiety scale). A pool of eighteen (EAEE). Un ensemble de dix-huit anxiété émotionnelle,
performance, test,
statements describing the presence items décrivant la présence ou l’ab- validation d’échelle
or absence of anxiety signs was sence de signes d’anxiété a été créé
created and submitted to et soumis à une analyse factorielle
exploratory factor analysis to reduce exploratoire dans le but de réduire
the pool to six items. Six validation le nombre d’items à six. Six études
studies were carried out, in which de validation ont été conduites dans
the EAEE was completed in various lesquelles l’EAEE était remplie dans
test settings. The results of the vali- des situations d’évaluation variées.
dation studies generally support the Les résultats des études de valida-
psychometric properties of the tion supportent globalement les
EAEE. First, its bifactorial structure propriétés psychométriques de
was confirmed in various test l’EAEE. Premièrement, sa structure
settings, as was the reliability and bifactorielle est confirmée dans des
sensitivity of its worry and emotion- situations d’évaluation variées, tout
ality subscales. Second, the two comme la fiabilité et la sensibilité
subscales appear valid as they des sous-échelles d’anxiété cogni-
converged with valid state, trait and tive et émotionnelle. Deuxiè-
test anxiety measures, they were mement, les deux sous-échelles
sensitive to the presence or absence apparaissent valides dans la mesure
of a test and, with few exceptions, où a) elles convergent avec des
they correlated with variables gener- mesures d’anxiété état, trait, et

* Laboratoire Inter-universitaire de Psychologie, Université de Savoie, UFR LLSH, 73000


Chambéry, France. Phone: +33 4.79.75.83.12.
E-mail : marine.beaudoin@univ-savoie.fr, olivier.desrichard@univ-savoie.fr

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


79
MEP_1_2009 8/04/09 10:25 Page 80

ally linked to test anxiety, such as d’évaluation validées, b) elles sont


gender, performance expectancy sensibles à la présence ou l’absence
and test performance. These find- d’évaluation et, c) à quelques excep-
ings suggest that the EAEE will be a tions près, elles corrèlent avec des
useful tool for researchers who variables généralement liées à l’an-
need a short scale in order to assess xiété d’évaluation, telles que le
state test anxiety in various acade- genre, les attentes de performance

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
mic test situations. The EAEE is also et la performance effectivement
designed to be used in non-acade- obtenue lors de l’évaluation. Ces
mics settings. However further vali- résultats suggèrent que l’EAEE est
dation studies are needed in order un outil utile aux chercheurs qui ont
to confirm its psychometric prop- besoin d’une échelle courte afin
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

erties in this context. d’évaluer l’anxiété d’évaluation état


dans des situations académiques
variées. L’EAEE est également cons-
truite de façon à pouvoir être utili-
sée dans des situations d’évaluation
non-académiques. Cependant, des
études de validation supplémen-
taires sont nécessaires pour confir-
mer ses propriétés psychomé-
triques dans ce contexte.

S tate test anxiety (STA) is the anxiety one experiences in evalu-


ative situations (Zeidner, 1998). Many researchers believe that
STA is involved in several psychosocial phenomena. For example,
the stereotype threat theory postulates that test anxiety may medi-
ate the deleterious effect of stereotype threat on performance
(Smith, 2004). Another example is self-efficacy theory, which consid-
ers that the impact of self-efficacy on performance is at least partially
mediated by state anxiety (Bandura, 1989).
Testing such hypotheses requires a tool for measuring STA.
Currently available questionnaires have a number of limitations,
one of the most important being the fact that they are often
based on outdated theories. Recent test anxiety research has
shown that test anxiety consists of two components: worry
(concerns about one’s performance) and emotionality (percep-
tion of autonomic arousal and unpleasant feeling states, see
Zeidner, 1998). However, several measures of STA, such as the
widely used State-Trait Anxiety Inventory (STAI; Spielberger,
1983) or the Comparative Anxiety subscale of the Test Attitude
Survey (TAS; Arvey, Strickland, Drauden, & Martin, 1990), do not

VALIDATION OF A FRENCH TEST ANXIETY SCALE


80
MEP_1_2009 8/04/09 10:25 Page 81

make a distinction between these two components. In another


scale, the Anxiety subscale of the Test Emotions Questionnaire
(TEQ; Pekrun, Goetz, Perry, Kramer, Hochstadt, & Molfenter,
2004), the emotional and cognitive components of STA are
conceptually distinguished but the subscale only provides a
single global STA score. This point is of importance since worry

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
and emotionality are not involved in the same phenomena. For
example, worry and emotionality are known to have different,
even opposite, effects on test results in that worry can impair
performance (e.g., Blair, O’Neil, & Price, 1999), whereas
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

emotionality can improve it (e.g., Hong, 1999). Thus, a global


measure of test anxiety may be inadequate for predicting perfor-
mance and may lead to falsely concluding a null effect.
Some recent questionnaires have been designed to measure
worry and emotionality in STA; however, they have practical limi-
tations that often conflict with organizational considerations in
experiments or field studies. First, the time available for
measuring STA before or after a performance, for example an
academic exam, is often limited. Second, the scale used must be
appropriate for the population studied (children, students,
adults, etc.) and suited to the test-situation (academic exam,
psychological testing, recruitment, etc.). To the best of our
knowledge, none of the existing scales meets these two criteria.
The Revised Worry-Emotionality Scale (RWES; Morris, Davis, &
Hutchings, 1981) and the Hong and Karstensson (2002) short
state version of the Test Anxiety Inventory (TAI; Spielberger,
1980) are specific to academic test anxiety; therefore, they cannot
be used in other situations (e.g., memory testing of the elderly).
Although it can be used in all test settings, the Cognitive
Interference Questionnaire (CIQ; Sarason, Sarason, Keefe, Hayes,
& Shearin, 1986) is a post-performance report, which can be
problematic as post-performance anxiety assessments can be
biased by perceived performance or self-handicapping (Smith,
Snyder, & Handelsman, 1982). The Endler Multidimensional
Anxiety Scale-State (EMAS-S; Endler, Edwards, & Vitelli, 1991) and
the Self- versus Other-Referenced Anxiety Questionnaire (SOAQ;
Proost, Derous, Schreurs, Hagtvet, & De Witte, 2008), which are
completed just before taking a test, contain a large number of
items (20 and 24, respectively), thereby increasing the risk of
disrupting manipulation effects.

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


81
MEP_1_2009 8/04/09 10:25 Page 82

Having repeatedly come up against these limitations during our


own researches, we realized that there is a need for a short state
worry and emotionality test anxiety scale that can be used in all
test situations. The aim of this research was to construct and vali-
date such a scale: the “Echelle d’Anxiété d’Evaluation Etat”
(EAEE; state test anxiety scale). The first section of the present

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
paper describes the construction of the scale; the second section
reports six studies that were carried out to test the factorial
validity, reliability and sensitivity of the EAEE and to assess its
convergent, criterion-related and construct validities in various
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

test settings.

Scale Construction

Overview
We began by creating an initial item pool, which was then orga-
nized into a questionnaire. This questionnaire was completed by
a sample of candidates for the written part of the French driving
test just before the test outset. The participants’responses to
each item were screened for floor and ceiling effects and were
then subjected to an exploratory factor analysis in order to select
those items that were the best indicators of their expected factor.

Development of an Initial Item Pool


Eighteen items were created or selected from existing scales or
from test anxiety definitions. Ten items described the presence or
absence of the emotional manifestations of STA, for example, the
presence of autonomic arousal (expressions such as “my hands
are moist”) and the absence of an unpleasant feeling state
(expressions such as “I feel relaxed”). Eight items described the
presence or absence of cognitive manifestations of STA: negative
expectations, concerns about oneself, the situation at hand, and
potential consequences (e.g., “I can’t stop myself thinking the
test will go wrong”, “My mind is untroubled”). These items were
randomly arranged in a questionnaire and presented with the
instruction to evaluate how well each of the statements describes
the participant’s present condition using a 6-point scale ranging
from 1 (“not at all”) to 6 (“very well”).

VALIDATION OF A FRENCH TEST ANXIETY SCALE


82
MEP_1_2009 8/04/09 10:25 Page 83

Item Selection
Methods
Participants. Two hundred French candidates for the written part
of the driving test voluntarily participated in the study. The
sample included 98 men and 100 women (two participants did

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
not report their gender on the questionnaire), aged from 15 to 68
years. Sixty-four of the participants were high school or university
students.
Procedure. Once the applications had been registered, the inter-
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

viewer invited the candidates to complete a short and


anonymous questionnaire. Before the test outset, the candidates
who were willing to participate in the study completed a ques-
tionnaire containing socio-demographic questions (age, gender
and activity status) and the 18 initial items of the EAEE. In addi-
tion, five items describing performance expectancies (e.g., “I
expect to perform well on this test”) were included with the
anxiety items. This was done for two reasons. First, it enabled us
to test the reliability of a performance expectancy scale that we
planned to use in later validity studies of the EAEE. Second,
although performance expectancy and test worry are two distinct
theoretical constructs, they are related and items measuring
them may partially overlap (e.g., “I’m afraid of doing poorly on
this test”). Thus, the inclusion of the expectancy items in the
EAEE factor analysis helped us to identify worry items that
measure worry more than they measure performance
expectancy.

Results
We excluded four anxiety items from the factor analysis because
of the floor effect. We then submitted the 19 remaining items to
an exploratory factor analysis, testing for a three factors solution.
The unweighted least squares parameters estimation method was
chosen because several items were not perfectly normally distrib-
uted. Sample adequacy for factorization was good according to
the Kaiser-Meyer-Olkin index (KMO =.90) and Bartlett’s
sphericity test (c2 (171) = 1531.62, p ⬍.01). Given the correlation
between worry and emotionality reported in the literature
(Hong, 1999; Hong & Karstensson, 2002; Morris et al., 1981;
O’Neil, Baker, & Matsuura, 1992; O’Neil & Fukumura, 1992;

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


83
MEP_1_2009 8/04/09 10:25 Page 84

Oostdam & Meijer, 2003) and between each of these two vari-
ables and performance expectancy (Mulkey & O’Neil, 1999;
Malpass, O’Neil, & Hocevar, 1999; Oostdam & Meijer, 2003; Kim
& Rocklin, 1994; Bandalos, Yates, & Thorndike-Christ, 1995;
Awang-Hashim, O’Neil, & Hocevar, 2002; Morris & Liebert, 1970),
we tested the three factors solution with an oblique promax rota-

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
tion. The first factor, Emotionality, accounted for 39% of the
variance before rotation and was essentially made up of the
emotionality items with an eigenvalue of 5.86 after rotation. The
second and third factors, which explained 9% and 6.7% of the
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

variance before rotation, respectively, were made up of perfor-


mance expectancy and worry items with eigenvalues of 5.39 and
3.20 after rotation, respectively. A closer look at these items
revealed that the second factor was essentially made up of items
describing concerns about the testing situation, including some
ambiguous performance expectancy items containing the idea of
fear of failure. The third factor consisted of classic performance
expectancy items and worry items that describe concentration
state more than concerns. Thus, the second and third factors
seemed to correspond to worry and performance expectancy,
respectively. As expected, the factor correlations were all signifi-
cant. The Worry and Emotionality factors were correlated
(r =.59), and they were negatively correlated with the
Performance expectancy factor (r = –.49 and r = –.38, respec-
tively).
The items that most strongly and most specifically loaded on
their theoretical factor were then selected. Hence, the items “I
feel relaxed” (je me sens détendu (e)), “I feel my heart beating
fast” (je sens que mon coeur bat vite) and “I feel calm” (je me
sens calme) were selected for the Emotionality (EAEE-E)
subscale, and the items “I’m thinking of things I’m preoccupied
with” (je pense à des choses qui me préoccupent), “I can’t
prevent myself from thinking the test will go wrong” (je ne peux
m’empêcher de penser que le test va peut-être mal se passer),
and “I’m afraid of losing my head during the test” (j’ai peur de
perdre mes moyens pendant le test) were selected for the Worry
(EAEE-C) subscale (Appendix). Each anxiety scale provides an
anxiety score ranging from 1 (low anxiety) to 6 (high anxiety),
which corresponds to the mean of its three item scores after
inversion of the items “I feel relaxed” and “I feel calm”.

VALIDATION OF A FRENCH TEST ANXIETY SCALE


84
MEP_1_2009 8/04/09 10:25 Page 85

Descriptive statistics and indices of internal consistency are


presented in Table 1.
TABLE 1:
Study Subscale Cronbach’s Average Score Mean Standard Kurtosis Skew Summary of Internal
Alpha interitem range score deviation Consistency and
correlation Sensitivity Indices of
the EAEE Subscales.

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Scale EAEE-C .64 .37 1-6 3.52 1.45 -0.99 -0.10
construction EAEE-E .77 .53 1-6 3.84 1.46 -0.83 -0.39
study
Study 1 EAEE-C .78 .55 1-6 4.10 1.38 -0.45 -0.48
EAEE-E .87 .70 1-6 4.49 1.29 0.23 -0.85
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

Study 2 EAEE-C .66 .39 1-6 3.00 1.33 -0.80 0.26


EAEE-E .70 .43 1-5.67 3.00 1.25 -0.81 0.18
Study 3a EAEE-C .67 .40 1-6 4.21 1.35 -0.98 -0.28
EAEE-E .77 .53 1-6 4.18 1.32 -0.83 -0.37
Study 4 EAEE-C .72 .45 1.33-6 3.84 1.43 -1.2 -0.24
EAEE-E .75 .50 1-6 3.87 1.33 -0.57 -0.45
Study 5 EAEE-C .58 .31 1-6 3.62 1.40 -0.91 -0.13
EAEE-E .80 .57 1-6 3.31 1.35 -0.93 -0.04
Study 6a EAEE-C .66 .41 1-6 2.27 1.17 -0.31 0.67
EAEE-E – – – – – – –
Note. Study 6a corresponds to the figure recall task situation of Study 6.
aData from the written end-of-year exam situation.

Scale Validation

Overview
We tested the psychometric properties of the EAEE in six studies
conducted on different samples and in different test situations
(Table 2). In each study, the participants had to complete the
EAEE just before the test outset. In addition, several variables
were measured or manipulated in order to test the external
validity of the EAEE.
Several variables associated with STA were measured to test the
construct validity of the EAEE. Depending on the study (see the
Methods section), the variables collected were participants’per-
formance expectancy for the test, their performance on the test,
and their gender. Several studies have shown that in test settings
women report more state worry and emotionality than men do
(Bors, Vigneau, & Kronlund, 2006; Hong & Karstensson, 2002;
Malpass et al., 1999; O’Neil & Abedi, 1992); therefore we hypoth-

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


85
MEP_1_2009 8/04/09 10:25 Page 86

TABLE 2:
Studies Participants Test setting
Brief Description of
the EAEE Validation Study 1 Undergraduate students Reading of a paper
Studies. Study 2 High school students Written examination
Study 3 Undergraduate students t1 (3 months before exam):
none (course)
t2: written end-of-year exam
Study 4 Undergraduate students Written end-of-year exam

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Study 5 Candidates Written part of the driving test
Study 6 Elderly Memory tests
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

esized that women would have higher EAEE-C and EAEE-E scores
than men. It has also been shown that the state worry experi-
enced in test settings consistently correlates negatively with both
the performance level the participant expects to attain during the
test (Blair et al., 1999; Mulkey & O’Neil, 1999; Malpass et al.,
1999; Oostdam & Meijer, 2003; Kim & Rocklin, 1994; Bandalos et
al., 1995; Awang-Hashim et al., 2002; Morris & Liebert, 1970) and
his or her actual test performance (Elliot & McGregor, 1999;
Hong, 1998, 1999; Kim & Rocklin, 1994; Morris & Liebert, 1970;
Malpass et al., 1999; Mulkey & O’Neil, 1999; O’Neil & Abedi, 1992;
O’Neil & Fukumura, 1992; Smith, Arnkoff, & Wright, 1990). Thus,
we expected participants’EAEE-C scores to correlate negatively
with both their performance expectancy scores and their actual
performance on the test. On the other hand, the state emotion-
ality-performance expectancy relationship is inconsistent, as
study results have shown no relation (Bandalos et al., 1995; Kim
& Rocklin, 1994; Morris & Liebert, 1970), a negative relation
(Morris & Liebert, 1970; Oostdam & Meijer, 2003) or a positive
relation (Hong, 1999). Similarly, although some studies have
found a negative correlation between state emotionality and
performance (Hong, 1998; Kim & Rocklin, 1994; Morris &
Liebert, 1970), some have found a positive relation (Hong, 1999),
and others have not found any significant relation (Elliot &
McGregor, 1999; Morris & Liebert, 1970; O’Neil & Abedi, 1992;
O’Neil & Fukumura, 1992). Because of the inconsistency
observed in the literature concerning the relationships between
emotionality and performance expectancy and between emotion-
ality and performance, the relationships between
participants’EAEE-E scores and their performance expectancy
scores or their actual performance on the test are reported with

VALIDATION OF A FRENCH TEST ANXIETY SCALE


86
MEP_1_2009 8/04/09 10:25 Page 87

the construct validity results, but they are not taken into account
when considering the construct validity.
In order to test the convergent validity of the EAEE, we adminis-
tered several other anxiety scales to the participants. The state
and trait parts of the French version of the STAI-State
(Spielberger, 1983; Ansseau, 1997) and a trait test worry and

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
emotionality scale (our French translation of the TAI; Spielberger,
1980) were each completed in one study. We expected both of
these measures to correlate positively with the EAEE-C scores
and with the EAEE-E scores.
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

The criterion validity of the EAEE was assessed in one study by


asking one group of participants to complete it on two occasions,
namely several months before an end-of-year examination and
just before the start of this examination.
Finally, the EAEE’s factorial structure, reliability and sensitivity
were assessed using confirmatory factor analysis (when allowed
by sample size), internal consistency and the examination of
score distributions, respectively.
For the sake of clarity, the method used in the six studies will be
reported first, followed by the psychometric property tests
results (factorial validity, reliability, sensitivity, convergent validity,
criterion validity and construct validity).

Methods
Study 1
Participants. One hundred and four French undergraduate
students, aged between 19 and 31 years old (M = 20.8,
SD = 1.93), including 90 women and 13 men (one participant did
not report her or his gender) participated voluntarily.
Procedure. The participants completed a questionnaire just
before starting an oral examination (reading a paper). The ques-
tionnaire included the EAEE, the STAI-State (a =.96) and a
performance expectancy scale (inter-item r =.77, p ⬍.01). For
the latter, two items were selected from the initial item pool
described in the Scale Construction section. We chose two
performance expectancy items (“I feel confident about my
performance on this test”, and “I expect to perform well on this
test”) that most strongly and more specifically loaded on the
Performance expectancy factor to form a performance

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


87
MEP_1_2009 8/04/09 10:25 Page 88

expectancy scale, with scores varying from 1 (low performance


expectancy) to 6 (high performance expectancy).

Study 2
Participants. One hundred and seven French high school
students (M = 17.3 years, SD = 0.92), including 78 girls, partici-

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
pated voluntarily.
Procedure. The participants completed a questionnaire just
before starting a written examination. The questionnaire
included the EAEE and the performance expectancy scale used in
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

Study 1 (inter-item r =.48, p ⬍.01). The participants’marks for


the examination were collected when they were made available
by the examiner.

Study 3
Participants. One hundred and fourteen French undergraduate
students aged between 19 and 45 years old (M = 22, SD = 3.9),
including 94 women and 16 men (four participants did not report
their gender), participated voluntarily.
Procedure. Fifty-one of the participants completed a question-
naire at the beginning of a lecture. The questionnaire included
the EAEE, which had to be completed with reference to an end-
of-year written examination that would occur tree months later,
and the STAI-Trait (a =.93). Three months later, just before the
start of the examination, the 114 participants completed a ques-
tionnaire including the EAEE, the performance expectancy scale
used in Study 1 (inter-item r =.72, p ⬍.01) and an item designed
to assess perceived preparedness for the exam (“How well did
you prepare for this examination?”, assessed according to a 6-
point scale from 1 “very badly” to 6 “very well”). This item was
included so its influence could be statistically controlled during
analyses, in order to ensure that the correlation expected
between performance and anxiety was not explained by this
common determinant. We collected the participants’marks for
the examination when they were made available by the exam-
iners.

VALIDATION OF A FRENCH TEST ANXIETY SCALE


88
MEP_1_2009 8/04/09 10:25 Page 89

Study 4
Participants. Thirty-six French undergraduate students aged
between 19 and 54 years old (M = 23.11, SD = 7.48), including
29 women, participated voluntarily.
Procedure. Several weeks before their end-of-year examinations,

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
the participants completed the TAI (a =.82 and a =.88 for the
worry and emotionality subscales, respectively). On the day of
the examination and just before the start of a written test they
completed a questionnaire including the EAEE and the perfor-
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

mance expectancy scale used in Study 1 (inter-item r =.89,


p ⬍.01). The participants’marks for the examination were
collected when they were made available.

Study 5
Participants. Two hundred and thirty-five French candidates for
the written part of the driving test participated voluntarily. The
participants were aged between 16 and 63 years old (M = 21,
SD = 7.6) and the sample included 102 women and 146 high
school or undergraduate students.
Procedure. Once the applications had been recorded and just
before the test began, the participants completed a questionnaire
that included the EAEE, the performance expectancy scale used
in Study 1 (inter-item r =.60, p ⬍.01) and the measure of
perceived preparedness used in Study 3. At the end of the test,
the participants’sums of correct answers were collected.

Study 6
Participants. We used data obtained from a sample of people
who had participated in a study of normal aging. The sample
included 228 French people aged between 55 and 85 years old
(M = 67, SD = 8.02), 123 of whom were women. The participants
were independent and none of them had neurological histories,
showed any signs of dementia or were undergoing any form of
treatment that could affect their cognitive or affective func-
tioning.
Procedure. For the purposes of the study of normal aging, a
trained examiner evaluated the participants individually during
interviews that lasted, on average, two hours. Only the measures

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


89
MEP_1_2009 8/04/09 10:25 Page 90

used for the validation of the EAEE are presented here. During
the interviews, participants were asked to undergo two memory
tests taken from the “Batterie d’Efficience Mnésique” (BEM 144,
Signoret, 1991): immediate recall of a complex figure composed
of several geometric shapes and oral recall of a 12-sentence story
read by the examiner. For each of the two memory tests, the

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
EAEE and the performance expectancy scale used in Study 1
(inter-item r =.47, p ⬍.01 and r =.63, p ⬍.01 for the figure and
story recall tasks, respectively) were completed between test
presentation and test completion. During the interview, the
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

participants also completed the STAI-Trait (a =.87).

Results
Factorial Structure
Confirmatory factor analyses (CFA) were conducted on the
participants’responses to the six EAEE items for all but one of the
studies (Study 4), which was omitted because of insufficient
sample size (N = 36). CFAs were conducted using either a
maximum likelihood or a robust parameters estimation method,
depending on data multinormality (according to Mardia’s coeffi-
cient). Participants with missing data were deleted (6% at most).
For the CFAs conducted using the EQS maximum likelihood para-
meters estimation method, we displayed the maximum
likelihood Chi-Square statistic (c2) as an exact-fit index and five
approximate-fit indices. These indices included two incremental
fit indices (the comparative fit index [CFI] and the non-normed
fit index [NNFI]), one absolute fit index (the adjusted goodness
of fit index [AGFI]), and two absolute misfit indices (the root
mean square error of approximation [RMSEA] and the standard-
ized root mean squared residual [SRMR]). CFAs conducted using
the EQS robust parameters estimation method provided robust
fit indices: the Satorra-Bentler Chi-Square statistic (S-B c2), the
robust CFI (CFI*), NNFI (NNFI*) and RMSEA (RMSEA*). We
followed Kline’s (1998) recommendations for interpreting the
robust goodness of fit indices; that is to say, AGFI, CFI and NNFI
values greater than.90 were interpreted as indicating a good fit, as
were SRMR values of less than.10. RMSEA values of less than.08
were taken to indicate a satisfactory fit and values below.05 were
taken to indicate a good fit. For each study, the expected two-

VALIDATION OF A FRENCH TEST ANXIETY SCALE


90
MEP_1_2009 8/04/09 10:25 Page 91

factor model was tested first. Given the factor correlations


observed (.41 ⬍ rs ⬍.97), a single-factor model consisting of a
general state test anxiety factor was then tested. As the two
models were nested within each other, the relative superiority of
the two-factor model was assessed by the means of either the c2-
difference test or the corrected S-B c2-difference test (Byrne,

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
2006), depending on the parameters estimation method used. Fit
indices obtained for each of the two models, factor correlations
and c2-difference test results are presented in Table 3.
Standardized and unstandardized factor loading for the two-
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

factor models are shown in Table 4.


TABLE 3:
Study (N) Models c2 df RMSEA SRMR AGFI NNFI CFI Factor c2-difference
Goodness of Fit
(90% CI) correlation test Summary for the Two-
D c2 (Ddf) factor and
Study 1a 2-factor 4.72 8 .00 .02 .95 1 1 .84 Single-factor Models,
(104) (.00-.08) Two-factor Models’
1-factor 25.47** 9 .13 .05 .76 .90 .94 19.62 (1) ** Inter-factors
Correlations and
(.07-.20)
Models c2-difference
Study 2 2-factor 13.14 8 .08 .06 .90 .92 .96 .41 Test Results.
(101) (.00-.15)
1-factor 42.46** 9 .19 .10 .70 .53 .72 29.32 (1) **
(.14-.25)
Study 3a 2-factor 12.8 8 .07 .05 .88 .96 .98 .97
(112) (.00-.14)
1-factor 13.63 9 .07 .05 .89 .96 .98 0.25 (1)
(.00-.14)
Study 5 2-factor 22.63 ** 8 .09 .04 .92 .93 .97 .79 14.13 (1) **
(225) (.05-.14)
1-factor 36.76** 9 .12 .05 .88 .89 .93
(.08-.16)
Study 6aa 2-factor 70.80** 8 .19 .12 .70 .60 .78 .62 –
(224) (.15-.23)
1-factor 72.12** 9 .18 .10 .75 .64 .78
(.14-.22)
Study 6ba 2-factor 63.59** 8 .18 .11 .68 .79 .89 .72 –
(220) (.14-.22)
1-factor 75.35** 9 .18 .09 .68 .78 .87
(.15-.22)
Note. RMSEA = root mean square error of approximation; SRMR = standardized root mean
square residual; AGFI = adjusted goodness of fit index; NNFI = non-normed fit index; CFI
= comparative fit index. c2 corresponds either to the maximum likelihood c2 or the Satorra-
Bentler c2, depending on the parameter estimation method used. RMSEA, NNFI and CFI
correspond to robust values when the robust estimation method was used. Study 6a and
Study 6b correspond to the figure and the story recall task situations, respectively. Given
the poor fit of both the single-factor and two-factor models in Study 6, the c2-difference test
was not computed.
a Robust estimation method used.

** p ⬍.01.

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


91
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

92
MEP_1_2009
8/04/09

TABLE 4:
Standardized Coefficients of Item-factor Paths for the Two-factor Models.
10:25

Study 1a Study 2 Study 3a Study 5 Study 6aa Study 6ba


Item Factor
b SE b SE b SE b SE b SE b SE
I’m thinking of things EAEE-C. 50* .18 .50* .19 .70* .13 .40* .13 .47* .20 .54* .10
Page 92

I’m preoccupied with


I can’t prevent myself from EAEE-C. 92* .11 .82* .21 .51* .16 .71* .12 .76* .10 .83* .07
thinking the test will go wrong
I’m afraid of losing EAEE-C .85* .13 .64* .20 .68* .16 .67* .14 .72* .10 .79* .07
my head during the test
I feel relaxed EAEE-E .87* .11 .66* .17 .72* .14 .83* .09 .77* .11 .78* .07
I feel my heart beating fast EAEE-E .70* .14 .55* .17 .70* .12 .68* .11 .47* .10 .50* .08

VALIDATION OF A FRENCH TEST ANXIETY SCALE


I feel calm EAEE-E .95* .10 .81* .19 .79* .12 .79* .10 .72* .11 .83* .07

Note. Study 6a and Study 6b correspond to the figure and the story recall task situations, respectively.
aRobust estimation method used.

* p <.05.

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
MEP_1_2009 8/04/09 10:25 Page 93

Table 3 shows that the two-factor model adequately fits the data
in all but two of the studies (Study 5 and Study 6), with each item
significantly contributing to its factor (Table 4). The c2-difference
showed its relative superiority over a single-factor model in all
but one of the studies (Study 3), in which the factor correlation
was unusually high (r =.97) (Table 3). We further examined the

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
reasons why the model did not show an acceptable fit in Study 5
and Study 6. In Study 5, despite the significance of the c2, the
approximate fit indices appeared satisfactory. The case of Study 6
appears more problematic as the two test situations (figure and
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

story recall tasks) in this study gave poor fits for both the two-
factor and single-factor models. We carried out Lagrange
Multiplier and Wald tests to determine whether parameters
should be added or dropped in order to improve their overall fit
(Byrne, 2006). For both test situations the results showed that
allowing the EAEE-E’s “I feel my heart beating fast” item to load
on the Worry factor would significantly increase the overall fit (c2
(1) = 77.123, p ⬍.01 and c2 (1) = 67.06, p ⬍.01 for the figure and
story recall tasks, respectively). Note that this study was
conducted on older adults and it may be the case that some
participants did not complete the scale sufficiently carefully and
that their responses were biased by the fact that this item is the
only EAEE-E item that is similar to the EAEE-C items in that it
describes the presence, rather than the absence, of anxiety.
Because the EAEE-E item may have been problematic, we carried
out CFAs on the data for this elderly population after exclusion of
this item. In the case of the figure recall task, the bifactorial
model appeared well adjusted (S-B c2 (4) = 4.42, p ⬎.05;
RMSEA* =.02; SRMR =.03; AGFI =.97; NNFI* =.99; CFI* = 1). It
gave a better fit than the single-factor model (S-B c2 (5) = 56.70,
p <.01; RMSEA* =.22; SRMR =.11; AGFI =.60; NNFI* =.49; CFI*
=.75; DS-B c2 (Ddf) = 30.67 (1), p <.01) and each item signifi-
cantly contributed to its factor (bs >.46, ps ⬍.05). Consequently,
we decided not to test further for the psychometric properties of
the EAEE-E subscale in this test setting. In the case of the story
recall task, the two-factor model was misadjusted (S-B c2 (4)
= 13.68, p ⬍.01; RMSEA* =.11; SRMR =.04; AGFI =.87; NNFI*
=.94; CFI* =.98), albeit less so than the single-factor model (S-B
c2 (5) = 51.77, p ⬍.01; RMSEA* =.21; SRMR =.09; AGFI =.56;

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


93
MEP_1_2009 8/04/09 10:25 Page 94

NNFI* =.77; CFI* =.89; DS-B c2 (Ddf) = 21.59 (1), p ⬍.01), so we


decided not to test further for the psychometric properties of the
EAEE in this test setting.

Reliability

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Cronbach’s Alpha coefficients and average interitem correlations
for the worry and emotionality subscales of the EAEE in each
study are presented in Table 1. With the exception of the EAEE-C
subscale in study 5, all the coefficients are higher than.60 with
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

average interitem correlations near or above.40, which indicates


adequate internal consistency (Clark & Watson, 1995).

Sensitivity
The inter-individual sensitivity of the questionnaire appears satis-
factory, as the mean scores obtained by the participants for each
EAEE subscale were normally distributed (skew and kurtosis
between –1.5 and 1.5) and ranged from 1 or 1.33 to 6 or 5.67 (on
a scale from 1 to 6) (Table 1).

Convergent Validity
The convergent validity of the EAEE appears good, as each of the
two EAEE scores strongly correlated with the other three anxiety
scales. Indeed, the EAEE-C scores were significantly correlated
with the STAI-State scores in Study 1 (r =.82, p ⬍.05), with the
STAI-Trait scores in Study 6 (r =.25, p ⬍.05), and with the TAI
Worry subscale score in Study 4 (r =.42, p ⬍.05), and marginally
significantly correlated with the STAI-Trait scores in Study 3
(r =.28, p =.05). Similarly, the EAEE-E scores were significantly
correlated with the STAI-State scores in Study 1 (r =.86, p ⬍.05),
with the STAI-Trait scores in Study 3 (r =.43, p ⬍.05), and with
the TAI Emotionality subscale score in Study 4 (r =.48, p ⬍.05).

Criterion-related Validity
We assessed the criterion-related validity of the EAEE in Study 3
by comparing the EAEE scores obtained on two occasions,
namely several months before an end-of-year examination and
just before the start of this examination. Criterion-related validity
is satisfactory, as the results showed that EAEE-C and EAEE-E

VALIDATION OF A FRENCH TEST ANXIETY SCALE


94
MEP_1_2009 8/04/09 10:25 Page 95

scores were significantly higher in the test setting (M [SD] = 4.64


[1.23] and M [SD] = 4.58 [1.23], respectively) than outside the
test setting (M [SD] = 3.07 [1.12], t [50] = 8, p ⬍.01, Cohen’s
d = 1.2, and M [SD] = 2.38 [1.06], t [50] = 10.93, p ⬍.01,
Cohen’s d = 1.53, respectively).

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Construct Validity
We assessed the construct validity of the EAEE by the means of
three indices. First, EAEE-C and EAEE-E scores were regressed on
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

the participants’gender (male and female, coded 1 and 2, respec-


tively) in order to test for the effect of gender on state test worry
and emotionality (with women reporting higher anxiety levels
than men). These relationships were not tested in Study 1, Study
3 or Study 4 because of an insufficient number of cases in each
cell. Second, the expected correlation between performance
expectancies and worry was tested using zero-order correlation
analyses involving EAEE-C and performance expectancy scores.
Third, the participants’actual performances on the tests were
regressed on their EAEE-C scores to test the relationship
between worry and performance. In addition, the relationships
between EAEE-E scores and performance expectancy scores, and
between EAEE-E scores and actual test performance were exam-
ined from an exploratory perspective. Each statistical analysis was
carried out after eliminating outliers (3% at most), which were
diagnosed using distance, leverage and influence statistics
(Howell, 1998).
Gender effect. In line with the data in the literature, we expected
gender to predict EAEE scores. The results showed that women
reported higher EAEE-C and EAEE-E scores than men in Study 2
(b =.32, p ⬍.05, and b =.26, p ⬍.05, respectively) and Study 5
(b =.27, p <.05, and b =.33, p ⬍.05, respectively). However, we
found no gender effect on the EAEE-C scores in Study 6 (b =.03,
p ⬎.05). Thus, the data offer mixed support for the gender effect
found in the test anxiety literature.
Relationship between EAEE scores and performance
expectancy. The results showed negative correlations between
performance expectancy and EAEE-C scores for all the studies
(rs ⬍ –.22, ps ⬍.05). The exploratory analyses carried out on the
EAEE-E scores revealed a significant negative correlation with

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


95
MEP_1_2009 8/04/09 10:25 Page 96

performance expectancy in Study 1, Study 3, and Study 5 (rs


⬍ –.21, ps ⬍.05), and a marginally significant negative correlation
in Study 2 (r = –.19, p =.05). However, the EAEE-E scores were
not correlated with performance expectancy in Study 4 (r = –.14,
p ⬎.05).

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Relationship between EAEE scores and performance. The rela-
tionships between each of the two EAEE scores and the
participants’actual test performance were examined by using
linear regression analyses to predict performance scores from
EAEE-C scores and from EAEE-E scores. EAEE-C scores and EAEE-
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

E scores were entered in the same regression model in order to


avoid possible confounded or suppressor effects. In Study 6,
because there was a problem with one emotionality item in the
factor analysis, the EAEE-E scores were replaced with an emotion-
ality measure consisting of the means of the scores for the two
reliable EAEE-E items. In addition, perceived preparedness was
included as a predictor in the Study 3 and Study 5 analyses in
order to statistically check its influence on performance. This was
done to ensure that the correlation expected between anxiety
and performance was not explained by this common determi-
nant. The results showed that the EAEE-C scores negatively
predicted performance in all of the studies except the study
conducted on the candidates for the written part of the driving
test (Table 5). For the emotionality component of test anxiety, the
results of all but one of the studies revealed a positive correlation
between the EAEE-E scores and actual test performances.
In general, the results support the construct validity of the EAEE-
C subscales as the expected correlations (reported in the
literature) with gender, test performance expectancy and actual
test performance appeared in almost all the studies. Exceptions
were found in Study 6, where the results did not show the
expected gender effect, and in Study 5, where we did not find the
correlation between state test worry and performance that is
commonly reported in the literature. Furthermore, the construct
validity of the EAEE-E is supported by the gender effect we found
in almost all the studies. Although the inconsistencies reported in
the literature indicate that these relationships cannot be used as
evidence of construct validity, we nevertheless found that the
EAEE-Es negatively correlated with performance expectancy

VALIDATION OF A FRENCH TEST ANXIETY SCALE


96
MEP_1_2009 8/04/09 10:25 Page 97

Table 5:
Study (N) Subscale B SE b t
Linear Regressions
Study 2 (93) EAEE-C -0.569 0.292 -.203 -1.945 t
Predicting
EAEE-E 1.122 0.311 .377 3.605 ** Performance by EAEE-
C and EAEE-E Scores.
Study 3 (105) EAEE-C -0.849 0.420 -.219 -2.004 *
EAEE-E 1.722 0.430 .437 4.001 **
Study 4 (35) EAEE-C -2.031 0.726 -.590 -2.799 **

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
EAEE-E 2.165 0.780 .585 2.776 **
Study 5 (220) EAEE-C 0.178 0.175 .086 1.017
EAEE-E -0.124 0.170 -.062 -0.731
Study 6a (212) EAEE-C -0.311 0.128 -.183 -2.435 *
Emotionality 0.052 0.147 .026 0.350
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

Study 6b (212) EAEE-C -5.416 1.347 -.322 -4.020 **


Emotionality 1.895 1.453 .104 1.304
Note. Perceived preparedness was included as a predictor in the Study 3 and Study 5 regres-
sion analyses.
tp ⬍.06, * p ⬍.05, ** p ⬍.01.

scores and positively correlated with actual test performance in


all but one of the studies.

Discussion

The objective of these studies was to draw up, and then evaluate
the validity of a short state test anxiety scale that distinguishes
between the worry and emotionality components of STA and that
can be applied in various test settings. For this purpose, we
created an initial item pool which we then reduced and
submitted to a first confirmatory factor analysis. These construc-
tion steps led to the EAEE, a reliable and factorially valid 6-item
scale comprising both a worry and an emotionality subscale. The
EAEE was administered to several samples in various test settings,
along with other measures that would allow us to confirm its
factorial validity, reliability and sensitivity and to test its conver-
gent, criterion and construct validity. The results of the validation
studies generally supported the psychometric properties of the
EAEE. First, its bifactorial structure was confirmed in various test
settings, as was the reliability and sensitivity of its two subscales.
Second, the EAEE-E and EAEE-C subscales appeared valid as they
converged with valid state, trait and test anxiety measures, they
were sensitive to the presence or absence of a test and, with few

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


97
MEP_1_2009 8/04/09 10:25 Page 98

exceptions, they correlated with variables generally linked to STA,


such as gender, performance expectancy and test performance.
Nevertheless, several points need to be discussed. First, the facto-
rial structure of the EAEE showed a satisfactory fit in all but one
of the studies, namely the one involving elderly participants. It
should be noted that the instructions for the EAEE have to be

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
read very carefully as the test does not have a simple response
scale. It includes items with different directions (absence or pres-
ence of anxiety signs) and asks participants to evaluate to what
extent each statement describes their condition. The examiners
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

who collected the data for this study frequently reported prob-
lems with the administration of the scale. For example, it was
noted that participants often reacted negatively to the scale, with
many stating that they did not care about the result; therefore,
they completed the scale as quickly as possible in order to start
the memory tasks. Thus, it is possible that the older participants
did not complete the EAEE carefully enough. As the bifactorial
structure was confirmed in the other four studies, it is unlikely
that these results call into question the factorial validity of the
EAEE. Nevertheless, the results of this study have to be taken
with caution and further studies are needed in order to assess the
factorial validity of the EAEE in such test situations.
Second, the results revealed the expected negative relation
between EAEE-C scores and actual test performance in all of the
studies except the one involving candidates for the written part
of the driving test. In this case, the absence of a relationship may
be due to the nature of this test. Most test anxiety researchers
postulate that test anxiety has a negative effect on performance
because of the interference it produces during testing (Zeidner,
1998). It is argued that test anxiety affects performance by
increasing the amount of attention given to worrying thoughts,
and this distracts examinees from the task (Eysenck & Calvo,
1992; Keogh, Bond, French, Richards, & Davis, 2004; Kurosawa &
Harackiewicz, 1995). It has also been shown that state anxiety is
linked to a general distractibility, with attention directed towards
worry or various other internal and external cues (Braunstein-
Bercovitz, 2003; Hopko, Ashcraft, Gute, Ruggiero, & Lewis, 1998;
Keogh & French, 2001). Recently, Eysenck, Derakshan, Santos
and Calvo (2007) put forward their attentional control theory,
which postulates that state anxiety is likely to decrease atten-

VALIDATION OF A FRENCH TEST ANXIETY SCALE


98
MEP_1_2009 8/04/09 10:25 Page 99

tional control, thereby decreasing the influence of the controlled,


goal-driven attentional system, which focuses attention on the
task at hand, and increasing the stimulus-driven attentional
system, which automatically diverts attention toward salient and
conspicuous stimuli, such as internal (e.g., worry) and external
distractors. During the written part of the French driving test,

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
candidates are presented with several driving situations in
conjunction with multiple-choice questions about a driver’s
behavior or decisions. These situations are presented by means
of slides, which are shown sequentially and for a limited time.
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

The sequential presentation of the slides controls the speed of


the test and focuses the candidate’s attention. By doing this, the
written part of the driving test is distinct from the tasks
commonly used in test anxiety studies (in which it is usually up
to the examinees to focus their attention on the task). Hence, the
fact that the driving test focuses the participants’attention on the
task to be completed may have precluded the participants’atten-
tion being diverted from the task, thereby leading to
performances that are unaffected by STA. This explanation could
be tested in an experimental design that varies the mode of
presentation of test material. For example, undergraduate
students could undertake a multiple choice questionnaire either
presented on an usual exam paper or with individual items
sequentially presented by the mean of slides.
Finally, it should be noted that the two studies in which the
expected results were not always obtained (memory test for
elderly participants and the written part of the driving test) were
also the only two tests carried out in non-academic settings.
Consequently, the results can be said to provide general support
for the validity of the EAEE in academic test settings, but addi-
tional studies are needed to confirm its applicability in other test
settings.
In conclusion, the EAEE appears to provide a valid measure of
state worry and emotionality test anxiety. In addition to its good
psychometric properties, the EAEE has two practical advantages
over existing scales: it is brief and its items do not contain any
reference to a particular test situation (e.g., academic examina-
tions), such that it can be used across a range of test settings. The
EAEE therefore appears to provide a useful tool for researchers
who have to work within time constraints and who are interested

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


99
MEP_1_2009 8/04/09 10:25 Page 100

in assessing state test anxiety in non-academic test situations.


However, further validation studies are needed in order to
confirm the psychometric validity of the EAEE in non-academic
test settings (e.g., recruitment, neuropsychological assessment,
driving test).

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
References

Ansseau, M. (1997). Inventaire d’anxiété Trait-Etat de Spielberger


- STAI. In J. D. Guelfi (Ed.), L’évaluation clinique standardisée
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

en psychiatrie (pp. 349-354). Castres: Editions Médicales Pierre


Fabre.
Arvey, R. D., Strickland, W., Drauden, G., & Martin, C. (1990).
Motivational components of test taking. Personnel Psychology,
43, 695-716.
Awang-Hashim, R., O’Neil, H. F., & Hocevar, D. (2002). Ethnicity,
effort, self-efficacy, worry, and statistics achievement in Malaysia:
A construct validation of the state-trait motivation model.
Educational Assessment, 8, 341-364.
Bandalos, D. L., Yates, K., & Thorndike-Christ, T. (1995). Effects of
math self-concept, perceived self-efficacy, and attributions for
failure and success on test anxiety. Journal of Educational
Psychology, 87, 611-623.
Bandura, A. (1989). Regulation of cognitive processes through
perceived self-efficacy. Developmental Psychology, 25, 729-735.
Blair, D. V., O’Neil, H. F., Jr., & Price, D. J. (1999). Effects of exper-
tise on state self-efficacy and state worry during a
computer-based certification test. Computers in Human
Behavior, 15, 511-530.
Bors, D. A., Vigneau, F., & Kronlund, A. (2006). L’anxiété face aux
examens: Dimensionalité, similitudes et différences chez les
étudiants universitaires. Canadian Journal of Behavioural
Science, 38, 176-184.
Braunstein-Bercovitz, H. (2003). Does stress enhance or impair
selective attention? The effects of stress and perceptual load on
negative priming. Anxiety, Stress & Coping, 16, 345-357.

VALIDATION OF A FRENCH TEST ANXIETY SCALE


100
MEP_1_2009 8/04/09 10:25 Page 101

Byrne, B. M. (2006). Structural equation modeling with EQS:


Basic concepts, applications, and programming (2nd ed.).
Mahwah, NJ: Lawrence Erlbaum Associates Publishers.

Clark, L. A. & Watson, D. (1995). Constructing validity: Basic


issues in objective scale development. Psychological Assessment,

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
7, 309-319.

Elliot, A. J., & McGregor, H. A. (1999). Test anxiety and the hier-
archical model of approach and avoidance achievement
motivation. Journal of Personality and Social Psychology, 76,
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

628-644.

Endler, N. S., Edwards, J. M., & Vitelli, R. (1991). Endler


Multidimensional Anxiety Scales (EMAS): Manual. Los Angeles,
CA: Western Psychological Services.

Eysenck, M. W., & Calvo, M. G. (1992). Anxiety and performance:


The processing efficiency theory. Cognition & Emotion, 6, 409-
434.

Eysenck, M. W., Derakshan, N., Santos, R., & Calvo, M. G. (2007).


Anxiety and cognitive performance: Attentional control theory.
Emotion, 7, 336-353.

Hong, E. (1998). Differential stability of individual differences in


state and trait test anxiety. Learning & Individual Differences, 10,
51-69.

Hong, E. (1999). Test anxiety, perceived test difficulty, and test


performance: Temporal patterns of their effects. Learning &
Individual Differences, 11, 431-447.

Hong, E., & Karstensson, L. (2002). Antecedents of state test


anxiety. Contemporary Educational Psychology, 27, 348-367.

Hopko, D. R., Ashcraft, M. H., Gute, J., Ruggiero, K. J., & Lewis, C.
(1998). Mathematics anxiety and working memory: Support for
the existence of a deficient inhibition mechanism. Journal of
Anxiety Disorders, 12, 343-355.

Howell, D.C. (1998). Méthodes statistiques en sciences


humaines. Paris: Editions De Boeck Université.

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


101
MEP_1_2009 8/04/09 10:25 Page 102

Keogh, E., & French, C. C. (2001). Test anxiety, evaluative stress,


and susceptibility to distraction from threat. European Journal
of Personality, 15, 123-141.

Keogh, E., Bond, F. W., French, C. C., Richards, A., & Davis, R. E.
(2004). Test anxiety, susceptibility to distraction and examination

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
performance. Anxiety, Stress & Coping: An International
Journal, 17, 241-252.

Kim, S. H., & Rocklin, T. (1994). The temporal patterns of worry


and emotionality and their differential effects on test perfor-
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

mance. Anxiety, Stress & Coping, 7, 117-130.

Kline, R. B. (1998). Principles and practice of structural equa-


tion modeling. New York, NY: Guilford Press.

Kurosawa, K., & Harackiewicz, J. M. (1995). Test anxiety, self-


awareness, and cognitive interference: A process analysis.
Journal of Personality, 63, 931-951.

Malpass, J. R., O’Neil, J. H. F., & Hocevar, D. (1999). Self-


Regulation, goal orientation, self-efficacy, worry, and high-stakes
math achievement for mathematically gifted high school
students. Roeper Review, 21, 281-289.

Morris, L. W., Davis, M. A., & Hutchings, C. H. (1981). Cognitive


and emotional components of anxiety: Literature review and a
revised worry-emotionality scale. Journal of Educational
Psychology, 73, 541-555.

Morris, L. W., & Liebert, R. M. (1970). Relationship of cognitive


and emotional components of test anxiety to physiological
arousal and academic performance. Journal of Consulting and
Clinical Psychology, 35, 332-337.

Mulkey, J. R., & O’Neil, H. F., Jr. (1999). The effects of test item
format on self-efficacy and worry during a high-stakes computer-
based certification examination. Computers in Human Behavior,
15, 495-509.

O’Neil, H. F., & Abedi, J. (1992). Japanese children’s trait and state
worry and emotionality in a high-stakes testing environment.
Anxiety, Stress & Coping: An International Journal, 5, 253-267.

VALIDATION OF A FRENCH TEST ANXIETY SCALE


102
MEP_1_2009 8/04/09 10:25 Page 103

O’Neil, H. F., Baker, E. L., & Matsuura, S. (1992). Reliability and


validity of Japanese children’s trait and state worry and emotion-
ality scales. Anxiety, Stress & Coping: An International Journal,
5, 225-239.
O’Neil, H. F., & Fukumura, T. (1992). Relationship of worry and

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
emotionality to test performance in a Juku environment. Anxiety,
Stress & Coping: An International Journal, 5, 241-251.
Oostdam, R., & Meijer, J. (2003). Influence of test anxiety on
measurement of intelligence. Psychological Reports, 92, 3-20.
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

Pekrun, R., Goetz, T., Perry, R. P., Kramer, K., Hochstadt, M., &
Molfenter, S. (2004). Beyond test anxiety: Development and vali-
dation of the test emotions questionnaire (TEQ). Anxiety, Stress
& Coping: An International Journal, 17, 287-316.
Proost, K., Derous, E., Schreurs, B., Hagtvet, K. A., & De Witte, K.
(2008). Selection test anxiety: Investigating applicants’self- vs
other-referenced anxiety in a real selection setting. International
Journal of Selection and Assessment, 16, 14-26.
Sarason, I. G., Sarason, B. R., Keefe, D. E., Hayes, B. E., & Shearin,
E. N. (1986). Cognitive interference: Situational determinants
and traitlike characteristics. Journal of Personality and Social
Psychology, 51, 215-226.
Signoret, J.L., 1991. Batterie d’efficience mnésique BEM 144.
Paris: Editions Scientifiques Elsevier.
Smith, J. L. (2004). Understanding the process of stereotype
threat: A review of mediational variables and new performance
goal directions. Educational Psychology Review, 16, 177-206.
Smith, R. J., Arnkoff, D. B., & Wright, T. L. (1990). Test anxiety and
academic competence: A comparison of alternative models.
Journal of Counseling Psychology, 37, 313-321.
Smith, T. W., Snyder, C. R., & Handelsman, M. M. (1982). On the
self-serving function of an academic wooden leg: Test anxiety as a
self-handicapping strategy. Journal of Personality and Social
Psychology, 42, 314-321.
Spielberger, C. D. (1980). Test Anxiety Inventory. Palo Alto, CA:
Consulting Psychologists Press.

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


103
MEP_1_2009 8/04/09 10:25 Page 104

Spielberger, C. D. (1983). Manual for the State-Trait Anxiety


Inventory. Palo Alto, CA: Consulting Psychologists Press.
Zeidner, M. (1998). Test anxiety: The state of the art. New York,
NY: Plenum Press.

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

VALIDATION OF A FRENCH TEST ANXIETY SCALE


104
MEP_1_2009 8/04/09 10:25 Page 105

Appendix: French instructions and items of the EAEE

Un certain nombre de phrases décrivant ce que peut ressentir


une personne à un moment donné sont écrites dans le tableau ci-
dessous. Vous devrez bien lire chaque phrase et indiquer, pour
chacune, si vous pensez qu’elle décrit bien ce que vous ressentez

Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble
maintenant.
Pour donner votre réponse, vous devrez cocher en face de
chaque phrase la case associée au chiffre qui correspond le plus
sincèrement à ce que vous pensez.
Document téléchargé depuis www.cairn.info - - - 105.101.146.28 - 21/12/2018 18h29. © Presses universitaires de Grenoble

• Si la phrase ne décrit pas du tout ce que vous ressentez, cochez


1
• Si la phrase décrit mal ce que vous ressentez, cochez 2
• Si la phrase décrit plutôt mal ce que vous ressentez, cochez 3
• Si la phrase décrit plutôt bien ce que vous ressentez, cochez 4
• Si la phrase décrit bien ce que vous ressentez, cochez 5
• Si la phrase décrit tout à fait ce que vous ressentez, cochez 6

1 2 3 4 5 6

pas plutôt plutôt tout


mal bien
du tout mal bien à fait

Maintenant… 1 2 3 4 5 6
1. Je pense à deschoses qui me
préoccupent
2. Je me sens détendu (e)
3. Je ne peux m’empêcher de penser
que le test va peut-être mal se passer
4. Je sens que mon cœur bat vite
5. Je me sens calme
6. J’ai peur de perdre mes moyens
pendant le test

REVUE INTERNATIONALE DE PSYCHOLOGIE SOCIALE 2009 N° 1


105

S-ar putea să vă placă și