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

net/publication/284501454

Test anxiety

Article · January 2003

CITATIONS READS

8 43,101

2 authors:

Moshe Zeidner Gerald Matthews

University of Haifa University of Central Florida
40 PUBLICATIONS 2,054 CITATIONS 477 PUBLICATIONS 30,306 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Gerald Matthews on 31 July 2017.

The user has requested enhancement of the downloaded file.

 Encyclopedia of Psychological
Assessment
Test Anxiety

Contributors: Moshe Zeidner & Gerald Matthews

Edited by: Rocío Fernández-Ballesteros
Book Title: Encyclopedia of Psychological Assessment
Chapter Title: "Test Anxiety"
Pub. Date: 2003
Access Date: July 31, 2017
Publishing Company: SAGE Publications Ltd
City: London
Print ISBN: 9780761954941
Online ISBN: 9780857025753
DOI: http://dx.doi.org/10.4135/9780857025753.n202
Print pages: 965-969
©2003 SAGE Publications Ltd. All Rights Reserved.
This PDF has been generated from SAGE Knowledge. Please note that the pagination of
the online version will vary from the pagination of the print book.
SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

Introduction

‘Test anxiety’ refers to the set of phenomenological, physiological, and behavioural responses
that accompany concern about possible negative consequences or failure on an examination
or similar evaluative situation (Zeidner, 1998). ‘Test anxious’ students are characterized by a
particularly low response threshold for anxiety in evaluative situations, tending to view test
situations as personally threatening. They tend to react with extensive worry, mental
disorganization, tension, and physiological arousal when exposed to evaluative situations
(Spielberger & Vagg, 1995). Test anxiety is often accompanied by maladaptive cognitions such
as threat perceptions, feelings of reduced self-efficacy, anticipatory failure attributions, and
coping through self-criticism (e.g. Matthews et al., 1999). A widely accepted definition
proposed by Spielberger (e.g. 1980) construes test anxiety as a situation-specific personality
trait. ‘Test anxiety’ may also refer to stressful evaluative stimuli and contexts, and fluctuating
anxiety states experienced in a test situation. In general, trait test anxiety and evaluative
situations may be seen as interacting to provoke states of anxiety (Sarason et al., 1995).

Test anxiety research has prospered, in part, due to the increasing personal salience of test
situations for people in modern society, making tests and their long-term consequences
significant educational, social, and clinical problems for many. Indeed, test anxiety figures
prominently as one of the key villains in the ongoing drama surrounding psycho-educational
testing, as a source of both scholastic underachievement and psychological distress. Many
students have the ability to do well on exams, but perform poorly because of their debilitating
levels of anxiety. Test anxiety may also jeopardize assessment validity in the cognitive domain
and constitute a major source of ‘test bias’, in that anxious examinees may perform less well
than their ability and skills would otherwise allow. Much of the test anxiety research over the
past half century has been conducted to help shed light on the negative effects of test anxiety
on examinee performance and these concerns have stimulated the development of a variety
of assessment methods, to which we now turn.

Self-Report Instruments

Self-report assessments of test anxiety responses are most often elicited via questionnaires.
Self-reports have become the most popular format for assessing test anxiety because they are
considered to provide the most direct access to a person's subjective experiential states in
evaluative situations, they possess good psychometric properties, they are relatively
inexpensive to produce, and they are simple to administer and score. Self-report paper-and-
pencil questionnaire measures of trait measures ask subjects to report symptoms they
typicallyor generally experience in test situations, whereas state anxiety scales ask individuals
to report which of the relevant symptoms of anxiety they are currently experiencing in a
particular test situation. Next, we briefly walk the reader through a number of salient issues in
the development and validation of self-report measures.

What to Measure: Conceptualization and Dimensionality

An initial conceptualization of test anxiety is essential in order to guide the development of the
item pool, and facilitate the initial construct validity research. As a hypothetical construct, test
anxiety may be inferred by measuring cognitive (e.g. self-focused thoughts and worries),
affective (e.g. subjective tension), or behavioural (e.g. escape behaviour) indices. Lack of
precision in defining and observing inner constructs such as test anxiety can lead to serious

Page 2 of 9 Encyclopedia of Psychological Assessment

SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

problems in assessment. Although some early questionnaires were unidi-mensional, most

contemporary researchers accept the distinction made by Liebert and Morris (1967) between
Worry a n d Emotionality as major components of test anxiety. Worry refers to cognitive
concerns about the level of performance, failure, and comparison with others, whereas
Emotionality refers to feelings of tension and self-perceived physiological arousal. Debate
continues on the dimensionality of test anxiety, and so contemporary questionnaires differ
somewhat with respect to their number of scales.

Item Selection and Scale Construction

It is important to begin assessment with well-written items, and awareness of how item format
may influence their measurement properties. Over the years, a wide array of item formats
have been employed; currently, Likert scales are the most popular. Test anxiety self-report
scales are also plagued by a number of conventional threats to validity, including response
biases such as acquiescence and social desirability, defensiveness and repression of test
anxiety, and deliberate faking. Thus, attention to possible biases and careful statistical
analysis of test anxiety scales is essential. Most scales have been constructed using
exploratory factor analytic techniques. Confirmatory factor analysis was used early in the
1980s in test anxiety research to test the adequacy of the indicator-factor relationship in the
measurement model of test anxiety scales and has also recently been employed for purposes
of item analysis and selection. The new latent trait theory methods of scaling have rarely been
used in scale development, but they have considerable promise for scaling of test anxiety
items in the years to come.

In general, test constructors have succeeded in developing measures with fairly high internal
consistencies, typically in the high 0.80s to low 0.90s, and test-retest stabilities typical of
personality traits. The factor structure of subscales is somewhat more uncertain, with
conflicting results from different studies. Also, regrettably, few of the existing self-report
instruments allow standardized comparisons to be made across independent investigations,
although most provide separate norms for males and females.

Validity

Empirical evidence to show that a test purporting to measure test anxiety is indeed valid for
the designated purpose. A main problem for empirical validation lies in finding an acceptable
criterion. Scores on ability tests, grade point average, observer ratings, behaviour in
structured evaluative situations, and the like are good candidates for criterion behaviours.
Evidence on the concurrent predictive validity of test anxiety scales is reviewed by Zeidner
(1998). Test anxiety scores predict academic criteria such as those just listed. Meta-analyses
(e.g. Seipp, 1991) suggest a modest negative association between anxiety and academic
performance of around −0.2. Validity coefficients are higher for worry than for emotionality, as
expected: −0.29 and −0.15 respectively, in Seipp's meta-analysis of 126 studies (156 effect
sizes).

Test anxiety scales also require ‘construct validity’, i.e. the nature of test anxiety is understood
by relating test scores to other psychological constructs within the framework of some
overarching theory. The predominant theories are cognitive-psychological, focusing on the
detrimental effects of worry on attention and retrieval from memory (Spielberger & Vagg,
1995). Sarason (e.g. 1984) has proposed that test anxiety is characterized by self-
preoccupation and worry, such that intrusive self-denigrating thoughts (‘cognitive
interference’) attract attentional resources that might otherwise be allocated to task-related

Page 3 of 9 Encyclopedia of Psychological Assessment

SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

processing. The person's appraisal of their own performance impairment feeds back into
additional worry, generating, in the worst case, a vicious circle of progressively escalating
worry and interference. There is a rich empirical literature showing that test anxiety is
associated with maladaptive self-referent appraisals and ineffective coping strategies
consistent with the Spielberger and Sarason theories (e.g. Matthews et al., 1999).

Key Measures

The Test Anxiety Questionnaire (TAQ), by George Mandler and Seymour Sarason (1952), is
regarded as the first acceptable measure of the trait construct. This 37-item scale inquires
about symptoms of anxiety experienced by examinees under major academic evaluative
situations. Split-half reliability coefficients in the high 0.90s is evidenced for the TAQ, and test-
retest correlations of 0.82 over a 6-week interval have been reported. Correlations of 0.59
between scores on the TAQ and behavioural ratings of overt manifestations of anxiety
(perspiration, restlessness, inappropriate questions, and laughter) provide evidence of
concurrent validity. A ‘downscaled’ version of the TAQ for children, the Test Anxiety Scale for
Children (TASC), by Seymour Sarason and his co-workers (Sarason, Davidson, Lighthall &
Waite, 1958), consists of 30 items dealing with anxiety in evaluative situations at school which
subjects respond to with yes/no (e.g. ‘Do you worry more about school than other children?’).
Test-retest reliability coefficients in the low 0.70s are reported for the elementary school
grades and internal consistency reliability coefficients for the TASC are reported to range from
0.82 to 0.90. TASC scores evidence concurrent validity when correlated with gross intellectual
indicators, such as group IQ and achievement test scores. Factor analytic studies of the TASC
have, with some exceptions, obtained the following four factors: ‘Test Anxiety’, ‘Poor Self-
Evaluation’, ‘Remote School Concern’, and ‘Somatic Signs of Anxiety’.

A second generation of test anxiety scales was spawned by Irwin Sarason, Seymour
Sarason's younger brother. The Test Anxiety Scale (TAS; Sarason, 1978) is a trait measure of
test anxiety, and represents a shift in focus from the situation to the person. The TAS consists
of 37 items (originally consisting of 21 items taken from the TAQ), written in a true-false format
and summed to obtain a total score. Test-retest coefficients in the 0.80s have been obtained
for intervals of several weeks in the current 37-item version. Total TAS scores corrrelate highly
with other test anxiety measures and also have been found to relate to task-debilitating and
reported difficulty working under pressure. A factor analysis of the 37-item TAS yielded the
following two-factor solution: (a) cognitive concern and worry about oneself and one's
performance on tests, and (b) apparent consequences of this intense worry (including
interference with effective cognitive functioning and a variety of physical and emotional
consequences). Though it lacks sufficient normative data and sufficiently replicated
information regarding the psychometric properties of this scale, the scale has been widely
used in research on the nature and treatment of test anxiety. An additional scale developed by
Sarason in 1984, the Reactions to Tests scale (RTT; Sarason, 1984), is a multidimensional
measure of test anxiety, developed to gauge the following four theoretical dimensions of test
anxiety: Worry (e.g. ‘Before taking a test I worry about failure’), Tension (e.g. ‘I feel distressed
and uneasy before tests’), Test-Irrelevant Thinking (e.g. ‘During tests, I wonder how the other
people are doing’), and Bodily Symptoms (e.g. ‘My heart beats faster when the test begins’).
Each scale of the RTT is composed of 10 items, yielding four factorially derived subscale
scores and a total score. Subjects are asked to mark the intensity of their responses on a
scale from 1 (‘not at all typical of me’) to 4 (‘very much typical of me’). Sarason (1984)
reported subscale internal consistency reliabilities ranging from 0.68 to 0.81 for the total scale
and 0.78 for all 40 items.

Page 4 of 9 Encyclopedia of Psychological Assessment

SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

The Worry and Emotionality Questionnaire (WEQ) was developed by Liebert and Morris
(1967) to measure what they believed to be the two major components of test anxiety – Worry
(W) and Emotionality (E). Worry refers to cognitive concerns about such things as level of
performance, failure, and comparison to others (e.g. ‘I do not feel very confident about my
performance on this test’). Emotionality refers to self-perceived physiological arousal and
upset (‘I feel my heart beating fast’). Scoring is along a 5-point Likert scale. In constructing
the WEQ, five W and five E items were chosen from the TAQ on the basis of their general
factor loadings content validity regarding the dimensions involved. Alpha reliabilities for the W
and E scales are in the 0.79–0.88 range. Scale validity has been demonstrated in the inverse
relationship of performance-related indices to both Worry and Emotionality.

The most prevalent contemporary measure is the Test Anxiety Inventory (TAI; Spielberger,
1980). The TAI is a 20-item self-report scale based on two conceptualizations of test anxiety.
The 20-item inventory consists of 8-item subscales for Worry (‘Thoughts of doing poorly
interfere with my concentration on tests’) and Emotionality (‘During tests I feel very tense’).
The instrument has been translated into over a dozen languages including Arabic, Chinese,
Dutch, German, Hindi, Hebrew, Hungarian, Italian, Japanese, Korean, Norwegian, Persian,
Portuguese, and Spanish. The test manual (Spielberger, 1980) reports that alpha coefficients
for the TAI total scores are 0.94 or higher, for both males and females. Although the internal
consistency for the subscale scores is found to be somewhat lower than for the TAI total
scale, the subscale alphas were all reported to be 0.86 or higher, with a median alpha of 0.90.
Furthermore, the manual reports test-retest reliability to be in the range of 0.80–0.81 for 2-
week to 1-month periods in a variety of student groups. The author provides impressive
correlational evidence for the construct validity of the TAI. The TAI demonstrates strong
convergent validity with the TAS (r = 0.82 for men and 0.83 for women), and is negatively
correlated with both study skills (r = −0.48 for men, −0.14 for women) and grade point average
(r = −0.31 for men, −0.18 for women). Furthermore, the TAI is modestly (r = −0.22), though
consistently, related to academic achievement indices. Although a number of exploratory
factor analyses have substantiated the two-factor structure of the TAI, confirmatory factor
analyses tend to yield inconsistent results.

Benson and her co-workers (Benson, Moulin-Julian, Schwarzer, Seipp & El-Zahhar, 1992)
combined the TAI and RTT scales to form the Revised Test Anxiety (RTA) scale, an 18-item
multidimensional scale. The instrument produces four factorially derived subscale scores:
Tension, Worry, Bodily Symptoms, and Test-Irrelevant Thinking. This scale takes advantage of
the strong psychometric foundation of the TAI as well as the promising evidence for the
multiple dimensions of the RTT. A unique feature of the RTA is its construction and validation
through the use of extensive multi-national factor analysis and cross-validation procedures.
Specifically, the scale was developed by subjecting the non-redundant items from the original
RTT and the TAI to extensive multi-national factor analysis and cross-validation procedures
based on a sample of 346 US, German, and Egyptian students. The RTA was then cross-
validated using a second multi-national sample of 353 students. Cronbach's alpha for the 18-
item RTA was 0.88. The Tension and Worry subscales showed reliabilities of 0.82 and 0.79
respectively, whereas the Bodily Symptoms and Test-Irrelevant Thinking subscales were
lower, 0.68 and 0.67, respectively. The chi-square test of model fit for the 18-item four-factor
RTA showed acceptable fit as did the cross-validation of the RTA.

Alternative Assessment Procedures

Think-Aloud Procedures

Page 5 of 9 Encyclopedia of Psychological Assessment

SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

Think-aloud procedures are designed to assess the contents of consciousness in examinees

while they are engaged in test taking, without imposing the researchers' preconceptions on
respondents, as questionnaires may do. Subjects are asked to verbalize anything that comes
into their minds, while working on the cognitive tasks given. The thoughts are then coded on
relevant dimensions, such as their positive valence (‘Problems are simple’) or negative
valence (‘Not enough time left’).

Physiological Measures

Test anxiety researchers have occasionally employed physiological measures of arousal (e.g.
electrodermal activity) as indices of anxiety. Biochemical ‘trace measures’ assayed from blood,
urine or saliva, such as corticosteroids and catecholamines, may also be useful, especially in
studies involving prolonged exposure to evaluative stress (e.g. writing a doctoral dissertation).
These responses should be immune to the problems of response bias endemic to self-report
measures of anxiety. However, physiological indices suffer from a number of formidable
methodological problems, including questionable construct validity, poor reliability, and low
practicality in naturalistic field settings.

Behavioural Observations

Observations of behaviours such as excessive body movement and hand wringing may permit
greater accuracy and more objectivity than self-report. Trained observers may utilize some
standard set of observation categories in documenting test anxious behaviour. However, the
use of observational procedures for measurement of test anxiety is rare and the psychological
processes considered to be relevant to test anxiety are not very amenable to direct
observation. Among the problems that adhere to these measures are: the complexity of
coding schemes, observer bias and reliability, the reactive nature of the observation process
itself, and the new latent trait theory methods of scaling high costs of conducting
observational procedures. Table 1 presents a summary of the key measures of test anxiety.

Future Perspectives: Improving Assessment

We briefly present some suggestions for improving test anxiety scale assessment in light of
current drawbacks, focusing on standardized scale development.

More Complete and Systematic Domain Coverage

Most scales focus mainly on cognitive and affective parameters of anxiety responses. The
restricted content scope can be improved by employing more systematic domain mapping
procedures (e.g. through facet theory) and better representation of additional facets in the
test anxiety inventory. More attention to assessment of the processes supporting dynamic
person-situation interaction across the various phases of a stressful evaluative encounter is
also required.

Making Scales more Relevant for Clinical Purposes

Future scales need to target not just overall test anxiety, but also pathology associated with
various antecedent or latent conditions, and a wider range of symptoms and consequences.
The coverage of items needs to be expanded to reflect the phenomenology of high test

Page 6 of 9 Encyclopedia of Psychological Assessment

SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

anxious examinees, including such manifestations as panic attacks, total blackout, and
anxiety blockage. Conversely, current instruments also tell us rather little about unusually low
anxiety, ranging from lack of concern about evaluation and minimal motivation, to supreme
self-confidence or high self-efficacy.

Differentiating between Adaptive and Maladaptive Manifestations of Anxiety

Current scales do not adequately separate maladaptive anxiety effects from those which may
be adaptive, such as increased motivation. Future measures might distinguish between
facilitating and debilitating arousal, and cognitive processes that are realistic (e.g. worrying
about a genuinely threatening exam) and those that are unrealistic (e.g. those prompted by
an exam covering familiar material).

Conclusions

Self-report test anxiety questionnaires have well-attested predictive and construct validity, but
they may be open to a variety of response sets, including deliberate distortion and
defensiveness. Future work might develop the assessment of physiological and behavioural
test anxiety measures, and ‘triangulate’ assessment by means of converging operations. The
assessment of test anxiety would also benefit from the application of the dynamic process
models of contemporary stress theory. Rather than see test anxiety solely as a fixed property
of the individual, it is important to explore how stable dispositions bias self-appraisal and
coping in the context of the person's active attempts to manage the evaluative situation, and
process cues towards performance adequacy.

Table 1. Classification of test anxiety measures: Self-report versus alternative

assessment procedures
Prevalent self-
Alternative assessment procedures
report scales
Test Anxiety
Think-aloud procedures (e.g. report anything that comes while working
Questionnaire
on the test)
(TAQ)
Test Anxiety Scale Physiological indices (e.g. electrodermal activity, muscle tension, GSR,
for Children (TASC) trace measures, biochemical analysis of blood and urine samples)
Behavioural observations (e.g. dgety behaviour, nail biting, sweating,
Test Anxiety Scale
excessive body movement and tension, hand wringing, speed and
(TAS)
accuracy of performance)
Reactions to Tests
(RTT)
Worry and
Emotionality
Questionnaire
(WEQ)
Test Anxiety
Inventory (TAI)
Revised Test
Anxiety Scale (RTA)

Page 7 of 9 Encyclopedia of Psychological Assessment

SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

Test anxiety scale scores must be understood within the context of a person's life and social
milieu. Assessment requires appreciation of the various influences on anxiety score, including
the person's academic history and learning skills, psychosocial adjustment, and behaviour
during examinations. A simple composite test anxiety score should never be used in
describing, predicting, or explaining an examinee's behaviour. Sound interpretation requires
integration of various sources of data and assimilating them into an exposition that describes
the examinee's functioning, detailing specific strengths and weaknesses, and predicting
specific behavioural manifestations.

test anxiety
test anxiety scale
self-reports
anxiety
testing
scale
anxiety assessment

MosheZeidner, and GeraldMatthews

http://dx.doi.org/10.4135/9780857025753.n202
Related Entries

Applied Fields: Clinical

Theoretical Perspective: Psychometrics
Anxiety Assessment
Anxiety Disorders Assessment

References
Benson, J., Moulin-Julian, M., Schwarzer, C., Seipp, B., & El-Zahhar, N. (1992). Cross-
validation of a revised test anxiety scale using multi-national samples. In Hagtvet, K.A., &
Johnson, B.T. (Eds.), Advances in Test Anxiety Research, Vol. 7 (pp. 62–83). Lisse, The
Netherlands: Swets & Zeitlinger.
Liebert, R.M., and Morris, L.W.Cognitive and emotional components of test anxiety: a
d i s t i n c t i o n a n d s o m e i n i t i a l d a t a. Psychological Reports20975–978. (1967).
http://dx.doi.org/10.2466/pr0.1967.20.3.975
Mandler, G., and Sarason, S.B.A study of anxiety and learning. Journal of Abnormal and
Social Psychology47166–173. (1952).
Matthews, G., Hillyard, E.J., and Campbell, S.E.Metacognition and maladaptive coping as
components of test anxiety. Clinical Psychology and Psychotherapy6111–125. (1999).
http://dx.doi.org/10.1002/%28SICI%291099-0879%28199905%296:2%3C111::AID-
CPP192%3E3.0.CO;2-4
Sarason, I.G. (1978). The Test Anxiety Scale: concept and research. In Spielberger, C.D., &
Sarason, I.G. (Eds.), Stress and Anxiety, Vol. 5 (pp. 193–216). Washington, DC: Hemisphere.
Sarason, I.G.Stress, anxiety and cognitive interference: reactions to tests. Journal of
Personality and Social Psychology46929–938. (1984). http://dx.doi.org/10.1037/0022-
3514.46.4.929
Sarason, I.G., Sarason, B.R., & Pierce, G.R. (1995). Anxiety, cognitive interference, and
performance. In Saklofske, D., & Zeidner, M. (Eds.), International Handbook of Personality
and Intelligence (pp. 285–296). New York: Plenum Press.
Sarason, S.B., Davidson, K., Lighthall, F., and Waite, R.A test anxiety scale for children. Child

Page 8 of 9 Encyclopedia of Psychological Assessment

 SAGE SAGE Reference
© Rocío Fernández-Ballesteros 2003

Development29105–113. (1958).
Seipp, B.Anxiety and academic performance: a meta-analysis of findings. Anxiety
Research427–41. (1991). http://dx.doi.org/10.1080/08917779108248762
Spielberger, C.D. (1980). Test Anxiety Inventory: Preliminary Professional Manual. Palo Alto,
CA: Consulting Psychologists Press.
Spielberger, C.D., & Vagg, P.R. (1995). Test anxiety: a transactional process. In Spielberger,
C.D., & Vagg, P.R. (Eds.), Test Anxiety: Theory, Assessment, and Treatment (pp. 3–14).
Washington, DC: Taylor & Francis.
Zeidner, M. (1998). Test Anxiety: The State of the Art. New York: Plenum Press.

Page 9 of 9 Encyclopedia of Psychological Assessment

View publication stats