Effective Interventions on Test Anxiety Reduction A Meta-Analysis

This meta-analysis synthesized the results from test anxiety reduction programs. Analyses were based on 56 studies (n = 2,482); the overall mean effect size (ES) for test anxiety reduction programs was E++ = 0.65 (95 percent confidence intervals [CI] + 0.58 to 0.73). On measures of anxiety reduction, the average individual completing treatment is seen as better off than 74 percent of those individuals who did not receive treatment. The treatment of test anxiety has been quite successful in reducing the test anxiety level of clients. The most effective treatments appear to be those that combine skill-focussed approaches with behaviour or cognitive approaches. Individually conducted programs, along with programs that combined individual and group counselling formats, produced the greatest changes. There is a serious lack of research on test anxiety reduction programs for primary, secondary and high school students. Test anxiety is a serious problem for many student populations. Test anxiety among students has been negatively associated with test performance, achievement of degrees and the selection of occupations (Topp, 1989). Students at all levels who suffer from test anxiety choose and pursue careers which involve infrequent evaluations and which, consequently, may not fully challenge their cognitive abilities (Krohne and Laux, 1982). The term ‘test anxiety’ as a scientific construct, refers to the set of phenomenological, physiological and behavioural responses that accompany concern about possible negative consequences or failure of an exam or a similar evaluative situation (Sieber et al., 1977). A particularly Please address correspondence to: Dr Tuncay Ergene, Hacettepe University, Faculty of Education, Department of Counseling and Guidance, 06590, Turkey. Email: [email protected] School Psychology International Copyright © 2003 SAGE Publications (London, Thousand Oaks, CA and New Delhi), Vol. 24(3): 313–328. [0143–0343 (200211) 24:3; 332–328; 034944] at California State University, Stanislaus on March 7, 2016 spi.sagepub.com Downloaded from 314 School Psychology International (2003), Vol. 24(3) low response threshold for anxiety in evaluative situations characterizes test anxious students. As a result, they tend to react with threat perceptions, reduced feelings of self-efficacy, self-derogatory conditions; anticipatory failure attributions and more intense emotional reactions and arousal at the very first sign of failure. Test-anxious behaviour is typically evoked when a person believes that her or his intellectual, motivational and social capabilities are affected by the test situation (Sarason and Sarason, 1990). In test anxiety literature Spielberger’s (1972) test anxiety definition is widely accepted (test anxiety is an ‘unpleasant state characterized by feelings of tension and apprehension, worrisome thoughts and the activation of the autonomic nervous system when an individual faces evaluative achievement-demanding situations’). Test anxiety is a significant issue for counselling as evidenced by its rapidly growing literature. Based on a number of estimates of the prevalence rates of test anxiety in school and college age populations the phenomenon appears to be widespread. Hill and Wigfield (1984) projected that two or three students in a typical classroom are ‘highly anxious’, with as many as a total of ten million elementary and secondary school students under-performing due to their high test anxiety. Furthermore, a study of 1,648 Canadian elementary and high school students indicated that 22 percent of the students were significantly worried about schoolwork and this was the most prevalent stressor by grade 12 (McGuire et al., 1987). Researchers have estimated test anxiety prevalence rates of anywhere between 15– 20 percent for college students (Hill and Wigfield, 1984). Reduction of test anxiety levels has been one of the important concerns in test anxiety research. Just as there have been different ways of conceptualizing the problem of test anxiety, there have also been different ways of approaching its treatment. With the early understanding of test anxiety in terms of a physiological/emotional phenomenon, treatment efforts were directed toward reducing the physiological arousal through behavioural methods. Later, treatments moved more in the direction of cognitive and combined approaches. Various treatments have been developed or applied to test anxiety: (a) behavioural approaches including systematic desensitization, relaxation training, biofeedback, modelling, anxiety induction, anxiety management training and other behavioural techniques; (b) cognitive approaches including rational emotive therapy, cognitive restructuring and other cognitive techniques; (c) cognitive-behavioural approaches including cognitive-behavioural modification, stress-inoculation training and other cognitive behavioural techniques and (d) skill-deficit approaches including study skills training, test-taking skills training, other skill deficit approaches and cognitive behavioural and skill-focussed approaches combined (Beck et al., 1996; Jones and Petruzzi, 1995; Kondo and Gifu, 1997; Onwuegbuzie and at California State University, Stanislaus on March 7, 2016 spi.sagepub.com Downloaded from 315 Ergene: Effective Interventions on Text Anxiety Reduction Daley, 1996; Sanghvi, 1995; Sud and Prabha, 1996). Test anxiety reduction programs have gained widespread acceptance and are used as an intervention for students and persons in different stages of education. An abundance of literature exists that describe the process and outcome of test anxiety reduction programs with students at different levels. The majority of this information is presented as descriptive or anecdotal reports. Controlled research does exist that provides evidence of the effectiveness of test anxiety reduction interventions. These interventions have been shown to be effective in modifying test anxiety levels of clients. The generalizibility of this research has often been hampered by small samples, different research settings and conditions and conflicting results. A methodologically sound synthesis of the available controlled research is needed to answer broad questions about the effectiveness of test anxiety reduction programs and the impact of the moderating variables on their overall effectiveness. Meta-analysis can address these shortcomings by synthesizing the available research into one comprehensive study. The purpose of this study was to determine the overall effectiveness of test anxiety reduction programs and to find out if effectiveness is related to particular interventions, studies and client characteristics. These potential moderators included characteristics of the intervention and the client, as well as design features of the studies themselves. The results of this meta-analysis should provide the information needed for test anxiety theorists and for practitioners regarding effective test anxiety reduction programs and conditions under which the benefits of intervention may be maximized. Methods This meta-analysis included published and unpublished studies on the effectiveness of test anxiety reduction programs. The following methods were used to locate studies for inclusion: (a) computerized search of PsycLIT (1974–1998), Educational Resources Information Center (ERIC; 1966–1998), Social Science Citation Index (SSCI; 1990–1998), PsycINFO (1967–1998), Medline (1985–1998) and Digital Dissertations Abstracts (1977–1998) databases using the keywords test anxiety, examination anxiety, test fear, reduction programs, treatment, intervention, psychotherapy, counseling and outcome; (b) manual search of Journal of Counseling Psychology, Journal of Counseling and Development, Behavior Research and Therapy, Journal of Experimental Education, Journal of Consulting and Clinical Psychology, Behavior Therapy, Psychological Reports and Cognitive Therapy and Research; (c) manual search of numerous review articles, chapters and books as well as the reference lists of all located studies and (d) non-published studies were at California State University, Stanislaus on March 7, 2016 spi.sagepub.com Downloaded from 316 School Psychology International (2003), Vol. 24(3) sought by contacting anxiety and test anxiety associations, training and treatment centres and posting messages to electronic newsgroups. As a result of these search procedures, 286 studies were identified as possibly meeting the inclusion criteria. A number of the studies failed to meet at least one inclusion criteria such as a lack of a control group, no random assignment of clients or data that were non-susceptible to analysis. After reviewing these studies, 39 published (included 76 experiment groups) and 17 unpublished (included 38 experiment groups) were included in the meta-analysis. In the current meta-analysis 56 studies (114 effect sizes) were analysed. Coding process. Coding procedures for translating critical study information into coded form amenable to computer processing were established after all studies were retrieved for meta-analysis. Three graduate students, trained in coding the variables included on the coding sheet, coded the articles. These trained coders used a coding manual, which was developed by the researcher. Coders attended a two-hour training session about the coding process. (Coding sheets and coding manual can be obtained from the author on request.) Three coders independently coded all studies for eight potential moderator variables that were thought to be theoratically relevant to the outcome of test anxiety reduction programs. These moderators included characteristics of intervention, client and study; intervention approach (behavioural, cognitive, cognitive-behavioural, skill-focussed, behavioural and skill focussed combined, cognitive and skill focussed combined, cognitive and behavioural and skill focussed combined and other techniques); type of intervention technique (systematic desensitization, relaxation training, anxiety management training, hypnotherapy, other behavioural techniques, rational-emotive therapy, cognitive restructuring, other cognitive techniques, cognitive-behavioural modification, stress inoculation training, other cognitive-behavioural techniques, study-skills training, other skill focussed interventions, behavioural and skill-focussed approaches combined, cognitive and skill-focussed approaches combined, cognitive behavioural and skillfocussed combined, other interventions); intervention modality (group, individual, both); time in therapy in minutes (0–60, 61–200, 201–350, 351–500, 501–650, 651 or above); education level of clients (primary school, middle school, high school, college and university); publication status (published versus unpublished). Inter-rater reliability among the coders was calculated by using intraclass correlation and was calculated as 0.84. The coding process was considered reliable. Computation of effect sizes and statistical analysis. Effect sizes and their variances were calculated using formulas derived by Hunter and Schmidt (1990) for post-test comparisons of control and treatment groups. The data were entered into MetaWin Statistical Program at California State University, Stanislaus on March 7, 2016 spi.sagepub.com Downloaded from 317 Ergene: Effective Interventions on Text Anxiety Reduction (Rosenberg et al., 1997) after effect size calculations. The combined effect sizes and mean effect size differences were analysed using this program. The meta-analysis was based on fixed-effect models (Rosenberg et al., 1997); in other words, it is assumed that a class of studies with similar characteristics shares a common true effect size. Differences among studies in the actual effect size measured are assumed to be due to sampling error. In this analysis, the null hypothesis that all effect sizes are equal, versus the alternative hypothesis that at least one of the true effect sizes in a series of comparisons differs from the rest, is tested with the homogeneity statistic, Q, which has approximately a X2 distribution with degrees of freedom equal to one less than the total number of studies. The greater the value of Q, the greater the heterogeneity in effect sizes among comparisons. The total homogeneity, QT, much as one can partition variance in an ANOVA. In the formule QB is homogeneity between groups and QW, homogeneity within groups. QB + QW = QT The statistical analysis results were presented on ‘Q Tables’ as Halvorsen (1994) and Light et al. (1994) suggested.