This article was downloaded by: [New York University] On: 05 March 2015, At: 11:04 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Personality Assessment Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/hjpa20

Does Test Anxiety Bias Scholastic Aptitude Test Performance by Gender and Sociocultural Group? Moshe Zeidner Published online: 22 Jun 2011.

To cite this article: Moshe Zeidner (1990) Does Test Anxiety Bias Scholastic Aptitude Test Performance by Gender and Sociocultural Group?, Journal of Personality Assessment, 55:1-2, 145-160, DOI: 10.1080/00223891.1990.9674054 To link to this article: http://dx.doi.org/10.1080/00223891.1990.9674054

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

JOURNAL OF PERSONALITY ASSESSMENT, 1990, 55(1&2), 145-160 Copyright o 1990, Lawrence Erlbaum Associates, Inc.

Does Test Anxiety Bias Scholastic Aptitude Test performance by Gender and Sociocultural Group?

Downloaded by [New York University] at 11:04 05 March 2015

Moshe Zeidner School of Education University of Haifa Mount Camel, Israel

This study examines the commonly held contention that test anxiety may serve as a source of bias in the scholastic aptitude test performance of gender and ethnic minority groups. In addition, this study examines sex and sociocultural group differences in the level and pattern of test anxiety among Israeli college students. The sample was composed of 163 male and 198 female students sitting for scholastic aptitude tests routinely administered to all student applicants as part of their college admissions procedures. About 67% were of Western cultural extraction whereas the remainder were of Eastern extraction. Significant differences in text anxiety scores for males and females were observed, with greater sex group differentiation on the Emotionality than on the Worry scale. Test anxiety scores were not discernible by ethnicity or social class. Furthermore, test anxiety was not differentially related to aptitude test scores by sex or sociocukural group membership. Thus, this study lends little evidence to the common contention that test anxiety differentially debilitates the aptitude test scores of females and ethnic minority student candidates.

A vast literature attests t o meaningful sociocultural-group differences in collegeentrance scholastic aptitude test scores i n the U.S. Qensen, 1980) and Israel (Zeidner, 1987a, 1987b), with middle-class majority group candidates meaningfully outscoring their lower class minority counterparts. Furthermore, over t h e past two decades male college students, both in the U.S. (cf. Clark & Grandy, 1984) and in Israel (Safir, 1986; Zeidner, 1986), have been shown to score higher, o n average, than their female counterparts o n both the verbal a n d quantitative sections of Scholastic Aptitude Tests. Although sociocultural and sex group differentials in scholastic aptitude test performance are well documented and a

Downloaded by [New York University] at 11:04 05 March 2015

recurring issue in the field of aptitude testing, the specific reasons underlying the observed group differences (e.g., social, cultural, biological, bias in the test or test situation, etc.) are hotly contested in the literature (cf. Eysenck, 1984; Jensen, 1980; Zeidner, 1986, 1987b). Test anxiety has figured prominently in the literature as one of the many villains in the drama of psychometric test performance and is cited among the diverse factors said to be at play in determining sex and sociocultural group differences in scholastic aptitude and achievement. Thus it is claimed that females and lower class minority group candidates are at a disadvantage in the testing situation as they may be overly anxious about being evaluated and thus react with greater concern and emotional arousal. It has been repeatedly shown that high test anxiety is associated with relatively low cognitive performance at both the school (Zeidner & Safir, 1989) and university levels (Spielberger, Gonzales, Taylor, Algaze, & Anton, 1978). Although there is considerable variability in the specific responses that are labeled as test anxiety, there is a general agreement that both emotional arousal and cognitive concerns are at play (Liebert & Morris, 1967; Morris, Davis, & Hutchings, 1981). The worry (W) and emotionality (E) components have been found to be very strongly related, but distinguishible by virtue of their different patterns of correlations with aptitude test scores (Powers, 1987). Accordingly, the W component has repeatedly been found to be more closely related to academic performance than the E component (Deffenbacher, 1980; Liebert & Morris, 1967; Zeidner, Klingrnan, & Papko, 1988). This supports the notion that in evaluative situations highly test-anxious individuals direct their attention away from the task at hand to self-related cognitions, which serve to debilitate performance. In fact, recent conceptualizations of test anxiety have moved away from emotional reactivity formulations towards cognitive-attentional models and information-processingdeficits (Carver & Scheier, 1989; Tobias, 1985). To the extent that anxiety is inversely associated with aptitude test performance, some examinees would perform less well than their ability would otherwise allow, so that measurement of group differences in ability would be confounded with group differences in anxiety.

SOCIOCULTURAL DIFFERENCES IN TEST ANXIETY It has been claimed (Clawson, Firment, & Trower, 1981; Crocker, Schmitt, & Tang, 1988; Phillips, 1978; Samuda, 1975) that minority groups are at a disadvantage in the testing situation as these examinees may be overly anxious about being evaluated and react with greater concern and hostility. In fact, it is often claimed that minority-group examinees who are motivated to avoid

Downloaded by [New York University] at 11:04 05 March 2015

TEST ANXIETY

Id47

disapproval and to fear negative evaluation may have high test anxiety (Phillips, 1978). Consequently, test anxiety is believed to be systematically disadvantageous to test performance of lower class minority-group students. The etiology of claimed social-class differences has been variously interpreted in the literature. O n the basis of clinical observations, Sieber (1980) suggestled that evaluative and general anxiety have some common antecedents that predispose individuals growing up in lower class environments to feel anxious (e.g., ~ a r e n t a expectations l during early childhood about the level of competence that the child can achieve). Sarason (1980) viewed test anxiety as a personality characteristic that develops during the child's interactions with parents and is stabilized during preschool years. Thus, it is claimed, parents of highly anxious examinees tend to be aversive towards the child, ignore the child's need for security, and fail to offer constructive help in problem solving. Indeed, current research and theory points to test anxiety as resulting from a developmental history of success and failure experiences in evaluative situatior~s. Therefore, differences may have originated in observations of unpleasant consequences of failure or through harsh personal failures. The empirical findings bearing upon ethnic-group differences in test anxiety among adolescent (Zeidner & Safir, 1989) and college populations uensen, 1980) are neither consistent nor sufficiently well established. Furthermore, as suggested by Jensen (1980), the reported ethnic-group differences in test anxiety may be more a reflection of "Type I errors" in the research literature than of true differences among ethnic groups. In addition, much of the research has confounded ethnicity and social class, comparing a middle-class ethnic majority with a lower class ethnic minority. Few studies have examined the separate effectsof social class and ethnicity or their interaction on )testanxiety. Thus the question looms large: Are the occasional sociocultural differences in test anxiety sometimes reported in the literature due mainly to ethnic background or social status variables?

SEX-GROUP DIFFERENCES IN TEST ANXIETY A comprehensive review of the literature (Wine, 1980) reveals that females are more likely to devalue their academic performance and thus show higher levels of text anxiety compared to their male counterparts. Furthermore, Lewis and College (1987) claimed that in testing situations males are more likely to perceive the situation as a personal challenge and exhibit the facilitating responses of low-anxiety individuals (e.g., increased arousal, vigilance, and enthusiasm). Females, by contrast, tend to perceive the test situation as a threat and evidence behaviors characteristic of highly anxious males (fear, worry, anger, lowering of self-esteem, etc.). Thus, in males the evaluative situation serves to increase

Downloaded by [New York University] at 11:04 05 March 2015

efficacy and self-esteem, whereas in females it may lead to increase in arousal, worry, or discomfort. An intuitively appealing explanation for sex differences is given by Sarason, Davidson, Lighthall, and Ruebush (1960). Sarason suggested that boys are more defensive than girls and, consequently, less likely to admit that they are anxious. Maccoby and Jacklin (1974) provided a similar explanation for the tendency for greater female test anxiety in American studies. Deaux (1977) suggested that sex differences in anxiety can best be explained by reference to sex-role socialization: Females may be encouraged to admit to anxiety because anxiety is perceived as a feminine trait. Safir (1986) postulated that test anxiety may be responsible for Israeli females' consistently lower scores on both verbal and math scholastic ability tests. However, the students in the various studies Safir cited were not tested for test anxiety. Mean test anxiety scores for female college students in the U.S. have been reported to be consistently higher than those of males in a number of samples (Spielberger et al., 1980), with major gender differences observed mainly on the E component (Corcoran, Macdougall, & Scarborough, 1985). Furthermore, correlations between SAT scores and total test anxiety scores are reported to be comparable for males and females (Spielberger et al., 1980). For a college sample, Spielberger reported correlations of about .I9 and - .18 between test anxiety scores and aptitude test performance for men and women, respectively. A review of the anxiety literature (Jerusalem, 1985) suggests that sex may be a moderator variable in the anxiety-performance relationship and should be considered when dealing with the effects of W and E in predicting academic achievement. Sharma, Parnian, and Spielberger (1983) reported significant sex-group differences in the test anxiety scores of 160 Iranian and 160 Indian high school and college students. Within each cultural group, females received higher test anxiety scores than males. Hocevar, El-Zahhar, and DeMello (1986) found significant sex- and culturalgroup differences on both total test anxiety scores and the W and E components for 11th- and 12th-grade high school students in Egypt, Brazil, and the U S . Sex-group differences were .22 sigma units for W and .45 for the E component (based on my own calculations-across the three cultural groups). Overall, then, past research has consistently pointed to gender differences in test anxiety, with female students evidencing higher test anxiety levels than their male counterparts, particularly on the emotionality component. Thus, there is substantial evidence that sex, as a correlate of many developmental trends, significantly affects test anxiety.

-

GOALS AND HYPOTHESES Text anxiety may be conceptualized as affecting ability test performance of varying sex and sociocultural groups through one (or both) of the following

Downloaded by [New York University] at 11:04 05 March 2015

TEST ANXIETY

149

processes: (a) The group difference in anxiety process-Due to the inverse relationship between anxiety and aptitude test ~erformancegenerally reported for most subjects, consistently higher levels of test anxiety reported among females alnd minority group members result, in turn, in lower levels of aptitude test performance. (b)The bias process-Due to the possible interaction between test anxiety and sex or sociocultural group in affecting test performance, test anxiety ha!; a more debilitating effect on the test performance of females and minority-group members at each level of anxiety. Unfortunately, the empirical evidence in favor of each of the two foregoing alternative conceptualizations of the influence of test anxiety on aptitude test performance has not yet been systematically scrutinized. Thus, one major goal of the present study is aimed at filling this gap in the literature and shedding needed light on test anxiety as a ~otentialmediator and/or moderator of the group-ability test performance relationship. Furthermore, with all the interest in test anxiety over the past three decades, relatively little information has been directed to understanding the role that test anxiety plays in standardized admission testing. Thus, a further goal of this study is to determine the relationship between self-reported test anxiety and scholastic aptitude test performance. In addition, few data are presently available on sex- or sociocultural-group differences in the level and pattern of test anxiety among college student candidates or the relationship between test anxiety and aptitude test performance among various subgroups in the population in a cultural site outside the U.S. It is quite plausible that cross-cultural variations in cultural or sex-related experiences (e.g., socialization practices, schooling, societal expectations, etc.) may differentially influence the affective development and consequent personality profiles of social and gender groups by culture. Thus, the results of previous studies based mainly on U.S. college populations may not be uncritically applied to other national populations. Thus, a further goal of this study is to shed light on the prevalence and severity of test anxiety among college-student candidates in a true-to-life evaluative situation in a national setting outside the U.S. O n the basis of a number of cross-cultural studies it may be hypothesized that gender is meaningfully related to test anxiety, with higher levels of test anxiety among females than among males. Furthermore, with respect to ethnic-group differences, although it may be assumed that East-West cultural factors may influence reactions to objective exams and consequent test anxiety levels by ethnic group, diametrically opposed predictions may be raised with respect to the direction of Israeli ethnic-group differences in test anxiety. It is noted that Israel's ethnic-minority group consists of Sephardic Jews who immigrated from Northern Africa and the Middle East. Despite the fact that they are found disproportionately in the lower socioeconomic group, they are a majority in Israeli society today, comprising about 60% of the school population. Based on one line of reasoning, evaluative contexts may represent a much greater threat for students of Eastern cultural background, where there is strorlg

150

ZEIDNER

Downloaded by [New York University] at 11:04 05 March 2015

affiliative obedience towards parental authority, strict discipline, and family centeredness-all aspects that have been associated with anxiety (cf. Phillips, 1978). Accordingly, it may be more difficult in Eastern cultures for children to cope effectively with exam situations, because they may be more interested in pleasing adults and more afraid of disappointing parents than their Western counterparts. O n the other hand, it may be hypothesized that Asian children in Israel, coming from a person-oriented culture, may not be severely pressured by parents to achieve. Because in the East competitive achievement and selfsufficiency are not emphasized, a relatively low level of both motivation and test anxiety would result.

METHOD

Sample The present sample of 356 college student candidates was composed of members of the Israeli norm group (n = 587)-on whom the Hebrew version of the Test Anxiety Inventory (TAVHB; Zeidner, Nevo, 6r Lipschitz, 1988) was standardized and for whom complete (or nearly complete) TAI, scholastic aptitude test scores, and demographic data were available. Comparisons of this sample with the original norm group show that the two groups are not significantly different with respect to either mean test anxiety scores or scholastic aptitude test performance. The original norm group was composed of student candidates who sat for the intrauniversity college-entrance scholastic aptitude exams, administered on a national basis, during August, 1987. Approximately half of the examinees in the norm group were given the TAVHB before the scholastic aptitude exam and half immediately following the exam. Time of testing did not interact with sexor sociocultural-groupmembership in affecting TAI, E, or W. Furthermore, time of testing had no meaningful effect on reliability of TAI (before = .91; after = .92), W (before = .80; after = 3 7 ) or E (before = .88; after = .88). The major differences evidenced were a higher level of E found for those responding to the TAVHB before aptitude testing (M = 19.35) than after testing (M = 17.80), with about a .3 1 sigma difference in E scores. These results corroborate previous studies showing that the W component is relatively stable across time and periods of administration and is impervious to environmental influences, immediately before the testing situation and fall whereas E scores reach a ~ e a k rapidly immediately after (Spiegler, Morris, 6r Liebert, 1968). In fact, E scores are expected to be lower following exam periods than prior to it, because E is conceived as a reaction to stress of the specific test situation per se. Examinees in my sample were almost evenly distributed by sex, with 55% of

TEST ANXIETY

151

identifiable sex-group membership female and 45% male. The mean age was 23.45 (SD = S3). Based on father's country of origin, approximately 67% of thie student candidates in the sample were of Western (European/American or Israeli) extraction and about 33% were of Eastern (AsiadAfrican) background. Examinees' social-class background was assessed by a linear composite index, based on two 6-point rating scales gauging father's and mother's formal education. O n the basis of a median split, examinees at the median (3.5) and above were defined as upper class (n = 182,54%)and those below the median (n = 156, 46%) as lower class.

Downloaded by [New York University] at 11:04 05 March 2015

Tests and Procedures

Test Anxiety Inventory. The Hebrew adaptation (Zeidner et al., 1988)of the Test Anxiety Inventory (S~ielbergeret al., 1980) was used to assess student candidates' test anxiety levels. The TAVHB is a 20-item inventory providing a total test anxiety score and separate scores for the W and E components of test anxiety. Students use a 4-point scale, ranging from almost never (1) to almost always (4), to report how frequently they experience specific symptoms of a n x i e ~ ~ in evaluative situations. Each subscale consists of 8 items (score range from 8 to 32). Up to three items were generated for each of the original items in the Test Anxiety Inventory. The Hebrew version was submitted to expert psychometricians for review, final editing, and "face validity" purposes; minor changes in wording and style were made. The original 36-item questionnaire was then administered to 161 student candidates sitting for university-admissions scholastic aptitude testing, and following classical item analysis and factor-analytic procedures, 12 items were omitted-thus narrowing down the questionnaire to 24 items. The 24 items were administered to the original norm group, anal following principal factor analysis and item analysis, the item pool was reducedl to 20 items (including 8 W and 8 E items) on which the norms were based. The high item-remainder (with total score) correlations (from .41 to .72) and. satisfactory alpha internal-consistency coefficientsfor the TAI/HB (.92), and W (.84) and E (.88) subscales compared favorably with those reported for larger U.S. samples. The alpha coefficients for the various Hebrew scales were satisfactory for both male (M) and female (F) groups, as follows: W (M = .84, F = .84), E (M = .88, F = .88) and TAI/HB (M = .92, F = .9 1). Principal factor analysis of the 20 items for the entire sample, followed by varimax rotations, yielded two orthogonal factors-E and W-accounting for 27% and 19% of the total variance, respectively. Similar factor analysis carried out separately by sex, yielded an E and a W factor, accounting for 28% and 18% of the respective variance in the male group and 25% and 19% of the respective variance in the female group.

152

ZEIDNER

Scholastic Aptitude Test. Administered under standardized testing conditions, this battery included five objective multiple-choice-type subtests: (a) General Information, (b) Figural Reasoning, (c) Mathematical Reasoning, (d) Analytic Thinking, and (e) English Comprehension. Subtest scores were standardized to a M = 100 and SD = 20 (relative to norms for college candidates of the National Institute for Testing and Evaluation). Total test score was formed by a linear composite of subtest scores, scaled to a M = 500 and SD = 100.

Downloaded by [New York University] at 11:04 05 March 2015

Statistical Analysis In order to assess the degree to which test anxiety is a primary factor in the differential test performance of sex or sociocultural group, test anxiety, group membership (sex or sociocultural) and test Anxiety x Group was included in a single linear model. Interaction effects between anxiety level and a particular subgroup were assessed by including appropriate product variables in the regression analysis to reveaI any subgroup differences with respect to the relationship of test anxiety to performance.

RESULTS Table 1 presents means and standard deviations for TAVHB scores for the entire sample and grouped separately by sex, ethnicity, and social class. Table 2 presents the sample intercorrelation matrix among the key variables in the TABLE 1 Summary Statistics for TAI Scales and Composite Aptitude Test Scores by Sex, Ethnicity, and Social Class

TAI Subgroups Sex Male Female Ethnicity Western Eastern Social Class Middleclass Lowerclass TotalGroup

worry

Emotion

SAT Scores

M

SD

M

SD

M

SD

M

SD

38.46 42.91

10.46 10.24

13.40 14.14

4.09 4.38

17.30 20.07

5.16 5.00

555.12 482.53

104.46 104.55

40.13 41.66

10.92 9.98

13.46 14.23

4.19 4.35

18.58 18.98

5.48 4.96

546.22 472.32

102.51 101.18

39.89 40.88 40.55

10.85 10.28 10.66

13.35 14.18 13.90

4.08 4.57 4.43

18.53 18.35 18.41

5.21 5.26 5.24

560.29 465.92 514.60

98.18 100.18 109.68

--

Note. Due to missing data for each of the background variables, the subgroup ns do not add up to the sample size of 357: (a) 233 Western and 115 Eastern candidates, (b) 156 lower class and 182

middle class, and (c) 163 males and 198 females.

TABLE 2

Sample Intercorrelation Matrix of Scholastic Aptitude and Test Anxiety Measures (n

=

378) Currelation Matrix Variables

1. Information 2. Figures 3. Logic

4. Math

Downloaded by [New York University] at 11:04 05 March 2015

5. English 6. SAT Total 7. WOIT~ 8. Emotionality 9. TAL'HBa

I

2

3

4

.44 .63 .51 .62 .76 -.I8 -.I5 -.I9

.69 .75 .61 .33 -.24 -.15 -.21

.73 .70 .89 -.28 -.I9 -.27

.62 .86 -.28 -.I9 -.27

5

6

7

8

.53 .83

.90

9 -

-

.85 -.30 -.I5 -.25

-.SO -20 -.28

Note. All correlations were significant at the .05 level of significance. Test Anxiety Inventory scores.

--

study. The TAVHB sample mean was 40.55 (SD = 10.66), with mean scores of 13.90 (SD = 4.43) on the W scale and 18.41 (SD = 5.24) on the E scale. The mean SAT score was 514.60 (SD = 109.68).

Sociocultural Differences Eastern students are observed to fall significantly below their Western counterparts on total scholastic aptitude test scores (472.32 < 546.22), t(346) = -6.3 5, p < ,001. To what extent is this meaningful difference (of about .75 sigma unit) attributable to group differences in test anxiety? A multivariate analysis of variance (MANOVA) for the effects of ethnicgroup membership on the W and E subscales taken together proved to be nonsignificant (F < I), implying no statistically reliable differences between students of Eastern and Western background on the two major components of anxiety. Similarly, no significant differences were observed for Eastern (M -= 41.66) and Western (M = 40.1) student candidates on total TAVHB scores. Correlations between SAT and TAI/HB total scores were nondiscernible among Western (r = - .30) and Eastern (r = - .25) students, with test anxiety explaining about 9% and 6% of the variance in the Western and Eastern groups, respectively. Similar patterns of correlations were found between W and E and total SAT scores among Eastern (W = - .3 1, E = - .14) and Western (W == - .33, E = - .23) candidates. Also, intercorrelations among W and E scores were highly similar in Western (r = .55) and Eastern (r = .47) examinees. Does test anxiety interact with ethnic-group membership in affecting schollastic aptitude test performance?To shed light on this question, we conducted an analysis of covariance (ANCOVA) for the effects of test anxiety, ethnicity,

Downloaded by [New York University] at 11:04 05 March 2015

154

ZEIDNER

and their interaction upon SAT performance. It should be pointed out that testing for heterogeneity of slopes is a natural extension of covariance analysis (Freund & Littell, 1981). The model used was Y = f(X, A, X x A), where Y = SAT composite scores, X = TAI scores (ignoring the grouping), A = grouping (ethnic-group membership in this case), and X x A = interaction term due to different regression coeficients for the groups specified by Factor A. Regression relationships that differ among treatment groups are reflected in an interaction between the treatment groups and the independent variable or covariate. Type I hierarchical sum of squares model is used to test, in the following order, (a) the sum of squares due to a single regression of Y on X, ignoring grouping (= X); (b) the sum of squares due to different intercepts, assuming a single regression relationship (= A); and (c) an additional sum of squares due to different regression coefficients for the groups specified by factor A(= X x A). From the perspective of this study, the crucial test was the test for interaction. The analysis revealed significant effects for T-Anxiety, F(l, 344) = 35.91, p < .001, and ethnicity, F(1, 344) = 38.71, p < .001, but no significant effects for Test Anxiety x Ethnicity interaction (F < 1). Thus, test anxiety has a significant negative effect on aptitude scores; Eastern students fall significantly below their Western counterparts on aptitude test performance at each level of test anxiety; and test anxiety is not differentially related to aptitude test performance among varying ethnic groups. With respect to social class effects, lower class students fall significantly below their upper class counterparts on total aptitude test scores (465.92 < 560.29), t(378) = -9.26, p < .001. Although lower class students evidenced somewhat higher test anxiety levels than their middle-class counterparts (40.88 > 39.89), the mean differences on the total test anxiety inventory as well as its subscales were not statistically significant. Correlations between total SAT and TAI/HB scores were homogeneous for middle-class (r = - .3 17) and lower class (r = - ,233) student candidates. Also similar patterns of correlations between W and E and total SAT scores were found in lower class (W = - .26; E = -17)and middle-class (W = - .33; E = - .22) examinees. Also, much as was abserved for varying ethnic groups, the W and E scales were similarly correlated among lower class (r = .54) and middleclass (r = .51) examinees.

-

Sex Differences Among Israeli Students As shown in Table 1, males outscored their female counterparts on scholastic aptitude test scores (555.12 > 482.53), t(359) = 6.57, p < .001, with gender accounting for about 11% of aptitude test variance. Furthermore, women evidenced higher test anxiety scores than their male counterparts (42.91 > 38.46), t(357) = 6.57, p < ,001. There was about a .40 SD difference among the

Downloaded by [New York University] at 11:04 05 March 2015

TEST ANXIETY

I55

gender groups, with gender accounting for about 5% of the test anxiety variance. A MANOVA for the effects of sex on the W and E components considered together was significant, F(2, 356) = 13.89, p < .001, Wilks's lambda = .93, with sex accounting for about 7% of the vector variable variance. A profile analysis, testing for parallelism in the TAI component profile scores for male and female student candidates, shows significantly divergent sex-group profiles for the two test anxiety scales, F(1, 357) = 16.89, p < .W1. These data imply variations in the magnitude of sex-group differences with the particular components assessed. Separate analyses of variance (ANOVAs) for the effects of sex on the W and the E subscales, in turn, show significant sex-group differentiation on the E scales, F(1,357) = 26.50, p < ,001 (eta square = .07), with no significaint differences on the W scale. Thus, females evidenced meaningfully higher levels than males on the E scale (20.07 > 17.30, d = .55) but were only slightly higher on the W scale (14.14 > 13.40, d = .17). ANOVAs (via least square regression procedures) for the effects of sex, ethnicity, social class, and their interactions, accounted for 5% of the test anxiety scale variance. Sex was the only main effect, F(1, 330) = 13.10, p *< .001, that proved to be significant, with no significant interaction effects observed. The correlations between SAT scores and TAI/HB were nondiscernible for males (r = - .21) and females (r = - .25). Correlations between W and total SAT scores were - .332 among males and - .256 among females, whereas the E component was only negligibly related to aptitude test performance in both gender groups. The relationship between W and E was somewhat stronger among males (r = .57, P < .001) than among females (r = .47, P < .001). Bivariate regression of total SAT on TAVHB scores was y = 635.78 -t (-2.10 x X) for males and y = 591.78 (-2.52 x X) for females. An ANCOVA testing for sex-group differencesin the regression parameters of SAT scores regressed on TAI/HB scores shows significant effects for TAI/HB, F(1, 355) = 33.22, p < .001, and sex, F(l, 355) = 30.49, p < .001, but no significant interaction between test anxiety and sex. Thus, for each level of anxiety males have higher SAT scores, and TAI is a significant predictor of SAT. However, the regression lines are parallel for males and females-indicating no significant "slope bias" by gender. Does test anxiety mediate the relationship between sex and aptitude test scores to any meaningful extent? In order to shed light on this question, sex was entered into the regression equation-with SAT scores as criterion-after controlling for test awiety scores (i.e., using test anxiety scores as a covariate)~. With test anxiety held constant, F(1,356) = 330.30, p < .001, sex effects were still highly significant, F(1, 356) = 30.56, p < .001, incrementing the SAT variance accounted for about 7% over and above that of test anxiety. Thus, the

+

156

ZEIDNER

data imply that sex-group differencesin SAT performance do not appear to be mediated in any meaningful way by test anxiety. Furthermore, an ANCOVA for the effects of sex on test anxiety, controlling for scholastic aptitude (i.e., using SAT scores as the covariate), shows significant effects for SAT scores, F(1,356) = 31.23, p < .01, and for sex, F(1,356) = 6.50, p < .001. However, gender adds very little to test anxiety variance (about 1%) above that of scholastic aptitude. Sex-group differences in test anxiety (squared multiple correlation = .05) may be due, in large part, to gender differences in scholastic ability.

Downloaded by [New York University] at 11:04 05 March 2015

Additional Analyses A series of additional ANCOVA runs were conducted at the SAT subtest level (i.e., General Information, Figural Reasoning, Mathematical Reasoning, Analytical Thinking, English Comprehension) for the interaction between test anxiety scores and each of the background variables (i.e., sex, social class, and ethnic-group membership, in turn) in affecting each of the five SAT subtests. The results at the subtest level were virtually identical with what I found for the SAT as a whole-namely, composite test anxiety scores did not interact with either of the three demographic variables tested (sex, social class, and ethnicity) in affecting any of the SAT subtest scores.

DISCUSSION From a methodological point of view, it should be held in mind that this particular study was based on a sample of Israeli student candidates rather than a truly representative sample from the entire student population in Israel. Therefore, it may be premature to generalize the results to the Israeli student candidate or student population as a whole. There is a possibility that the order of testing interacted in some unknown degree with students' background characteristics so as to affect and perhaps confound criterion score results, although the likelihood that these effects were meaningful in the context of this research appear to be small (particularly because the overall results appear to conform in a lawful way to what we know about group differences in SAT and TAI scores and the relationship among the latter variables). bearing this methodological constraint in mind, these data lend little support to the notion that ethnic- and social-group differences in test performance are meaningfully accounted for by sociocultural-group differences in test anxiety. For one, sociocultural-group differences in test anxiety were minimal. Surprisingly, lower class and Eastern minority examinees were found to be no less anxious, on average, than their middle-class and majority-group counterparts under standardized test situations. Second, when SAT scores were regressed on

Downloaded by [New York University] at 11:04 05 March 2015

TEST ANXIETY

157

composite test anxiety scores, equivalent regression parameters were observied for varying sociocultural groups. Relative to the sizable sociocultural-group differences on test scores, group differences on anxiety scales were truly negligible. Indeed, it is difficult to accept that such a volatile trait as anxiety is distributed differentially across groups so that, on average, members of one group have more anxiety than members of another group, with individual variability within groups seeming to preclude such a generalization. By the same token, these data do not support the commonly held contention that sex-group differences in ability test performance are due, in part, to gender differences in test anxiety. One would indeed be hard pressed to explain gender differences in ability by male-female differences in test anxiety, particularly when sex-group differences are evidenced mainly on the E factor-which is only weakly related to SAT performance. Our data are consistent with a number of previous findings reported in the literature. These findings confirm previous cross-cultural reports (e.g., Manley & Rosemier, 1975; Hocevar et al., 1986) that the anxiety level of females is invariably higher than that of males in various cultural groups. Furthermore, this study points to significant sex-group differences in the profiles of test anxiety components, with greater sex-group differentiation on the E scales than on the W scales (Communian, 1985; Rocklin & Ren-Min, 1989; Spielberger et a\., 1980; van der Ploeg, 1983). Also, consistent with previous research (Powers, 1987), Worry and Emotionality were found to be strongly related, but distinguishable by virtue of their different patterns of correlations with rother variables, including aptitude test scores (cf. Liebert &Morris, 1967; Powers, 1987; Zeidner et al., 1988). W and E were found to be moderately related to test performance on each section of the SAT, with higher test anxiety associated with lower performance. These relationships were essentially equivalent for each ethnic, social, and sex group. Furthermore, the W component of test anxiety was more strongly related to test performance than the E component, and when considered together, only \W remained highly related to test performance. The nonsignificant ethnic-group differences reported in this study are at variance with some previous U.S. findings of significant ethnic differences in test anxiety. How can this be accounted for? First, it should be noted that ethnicgroup differences reported for U.S. minority groups have generally been small and may be attributed in part to Type I errors (Jensen, 1980). If ethnic-group differences in test anxiety reported in the literature are not artifactual, but real, then perhaps we should consider the theoretical and practical implications of the specific minority group in this study, namely, Sephardic Jews. Indeed, relevant processes operative in the U.S., which would lead minoritygroup students to fear negative evaluation in the academic system, may not be at work in the makeup of minority groups in Israel. As hypothesized, Asia11

Downloaded by [New York University] at 11:04 05 March 2015

158

ZEIDNER

children in Israel, coming from a person-oriented culture, are not severely pressured by parents to achieve and compete in school, thus resulting in a relatively low level of both motivation and test anxiety. In contrast to U.S. Blacks and Latinos, who have minority status both in terms of group size and economic and political factors, Sephardic Jews in Israel are not a minority in the numerical sense. They do, however, share economic and political disempowerment with U.S. minorities. The constant security threat in Israel is probably responsible for the greater social integration and consolidation of minority groups around basic social values, as well as greater feelings of cooperation and togetherness, which may carry over into the schools. Furthermore, because students of Sephardic background comprise the majority in the Israeli school system, school may be less threatening, with the numerical majority of Sephardic students serving as a stress buffer in evaluative situations. It should be stressed that no position is taken here with respect to the specific causal direction in the observed relationship between test anxiety and performance. O n the one hand, it is highly plausible that test anxiety, in concert with other situational variables, may work in combination to influence examinees' level of scholastic aptitude test performance. O n the other hand, it is equally ~lausiblethat examinees who do well on tests may possess or develop more efficient test-coping resources and hence lower levels of test anxiety under stressful test situations. In sum, these data lend no empirical evidence in favor of the hypothesis that TAI scores increase while SAT scores decrease differentially for gender and sociocultural groups.

ACKNOWLEDGMENTS I am indebted to Baruch Nevo of the Department of Psychology at the University of Haifa for his professional help in conducting this study and his invaluable comments to an earlier draft of this article. Also, thanks are due to the National Institute of Testing and Evaluation in Jerusalem for institutional support, and to Ms. Ela Lipschitz in particular, for her help in data collection and analysis. I am also indebted to Charles Spielberger of the University of South Florida for his constructive help in the process of norming the Hebrew Version of the Test Anxiety Inventory.

REFERENCES Carver, J. C., & Scheier, M. F. (1989). Expectancies and coping: From test anxiety to pessimism. In R. Schwarzer, H. M. van der Ploeg & C . D. Spielberger (Eds.), Advances in test anxiety research (Vol. 6, pp. 3-1 1). Lisse, Netherlands: Swets & Zeitlinger, B. V. Clark, M. J., & Grandy, J. (1984). Sex differexes in the academic performance of scholastic aptitude t a t takers (Report No. 84-43). Princeton, NJ: Educational Testing Service.

Downloaded by [New York University] at 11:04 05 March 2015

TEST ANMETY

I59

Clawson, T. W., Firment, C. K., &Trower, T . L. (1981). Test anxiety: Another origin for racial bias in standardized testing. Measurement and Evaluation in Guidance, 13, 210-215. Communian, A. L. (1985). The development and validation of the Italian form of the Test Anxiety Inventory. In H. M. van der Ploeg, R. Schwarzer, & C. D. Spielberger (Eds.), Advances in test anxiety research (Vol. 4, pp. 215-220). Lisse, Netherlands: Swets & Zeitlinger, B.V. Corcoran, M. D., Macdougall, M. A., & Scarbrough, W. H. (1985). The interplay of worry and emotionalitywith anxiety and cognitive interference in predicting test performance. In H. M. van der Ploeg, R. Schwarzer, & C. D. Spielberger (Eds.), Advances in test anxiety research (Vol. 4, pp. 103-109). Lisse, Netherlands: Swets & Zeitlinger, B.V. Crocker, L, Schmitt, A., & Tang, L. (1988). Test anxiety and standardized achievement test performance in the middle school years. Measurement and Evaluation in Counseling and Development, 20, 149-157. Deaux, K. (1977). Sex differences. In T . Blass (Ed.), Personality variables in social behavior (pp. 357-372). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Deffenbacher, J. (1980). Worry and emotionality in test anxiety. In I. G. Sarason (Ed.), Test anm'ety: Theory, research, and applications (pp. 111-128). Hillsdale, NJ Lawrence Erlbaum Associates, [nc. Eysenck, H. J. (1984). The effect ofrace on human abilities and menial test scores. In C. R. Reynolds & R. T. Brown (Eds.), Perspectives on bias in mental testing (pp. 249-291). New York: Plenum,. Freund, R. J., & Littell, R. C. (1981). SAS fur linear models. Cary, NC: SAS Institute, Inc. Hocevar, D., El-Zahhar, N., & DeMello, M. (1986). Sex and cultural differences in anxiety and test anxiety: Egypt, Bradl and the U S A . Manuscript submitted far publication. Jensen, A. R. (1980). Bias in mental testing. New York: Free Press. Jerusalem, M. (1985). A longitudinal field study with trait worry and trait emotionality: Methodological problems. In H. M. van der Ploeg, R. Schwarzer, & C. D. Spielbergs (Eds.), Advances in test anxiety research (Vol. 4, pp. 23-34). Lisse, Netherlands: Swets 6r Zeitlinger, B.V. Lewis, E. C., &College, C. (1987). Dierential responses of females and males to evaluative stress: Anxiety, self-esteem, efficacy, and willingness to participate. In R. Schwarzer, H. M. van der Ploeg, & C. D. Spielberger (Eds.), Advances in test anxiety research (Vol. 5, pp. 97-106). Lisse, Netherlands: Swets & Zeirlinger, B.V. Liebert, R. M., & Morris, L. W. (1967). Cognitive and emotional components of test anxiety: A distinction and some initial data. Psychological Reports, 20, 975-978. Maccoby, E. E., & Jacklin, C. N. (1974). The psychology of sex differences. Stanford, CA: Stanfiord University Press. Manley, M. J., & Rosemier, R. A. (1975). Developmental trends in general anxiety among junior and senior high school students. Journal of Genehc Psychology, 20, 219-226. Morris, L. W., Davis, M. A., & Hutchings, C. H. (1981). Cognitive and emotional components of anxiety: Literature review and a revised Worry-Emotionality Scale. lournal of Educational Psychology, 73, 541-555. Phillips, B. N. (1978). School stress and anxiety. New York: Human Sciences. Powers, D. E. (1987). Test anxiety and the GRE general test (Report No. 86-45). Princeton, NJ: Educational Testing Service. Rocklin, T., & Ren-Min, Y. (1989). Development and adaptation of the Chinese Test Anxiety Inventory. In. R. Schwarzer, H. M. van der Ploeg, & C. D. Spielberger (Eds.), Advances in test anxiety research (Vol. 6, pp. 245-252). Lisse, Netherlands: Swets & Zeitlinger, B.V. Safir, M. P. (1986). The effects of nature or of nurture on sex differences in intellectual functioning. Sex Roles, 14, 581-590. Samuda, R. J. (1975). Psychological testing of American minorities. New York: Harper &Row. Sarason, I. G. (Ed.). (1980). Test anxiety: Theory, research, and applications. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Sarason, S. B., Davidson, K. S., Lighthall, F. F., & Ruebush, B. K. (1960). Anxiety in ekment,zv school children. New York: Wiley. Sharma, S., Parnian, S., & S~ielberger,C. D. (1983). A cross-cultural study of test anxiety levels in

Downloaded by [New York University] at 11:04 05 March 2015

Iranian and Indian students. Personality and Individual Differences, 4, 117-120. Sieber, J. E. (1980). Defining test anxiety: Problems and approaches. In I. G. Sarason (Ed.), Test anxiety: Theory, research, and applications (pp. 15-40). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Spiegler, M. O., Morris, L. W., & Liebert, R. M. (1968). Cognitive and emotional components of test anxiety: Temporal factors. Psychologicd Reports, 22, 451456. Spielberger, C. D., Gonzales, H. P., Taylor, C. J., Algaze, B., &Anton, W. D. (1978). Examination stress and test anxiety. In C. D. Spielberger &I. G. Sarason (Eds.), Stress and anxiety (Vol. 5, pp. 167-191). New York: Wiley. Spielberger, C. D., Gonzales, H. P., Taylor, C. J., Anton, W. D., Algaze, B., Ross, G. K., & Westberry, L. G. (1980). Test Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press. Tobias, S. (1985). Test anxiety: Interference, defective skills and cognitive capacity. Educational Psychologist, 20, 135-142. van der Ploeg, H. M. (1983).The validation of the Dutch form of the Test Anxiety Inventory. In H. M. van der Ploeg, R. Schwarzer, & C. D. Spielberger (Eds.), Advances in test anxiety research (Vol. 2, pp. 191-202). Lisse, Netherlands: Swets & Zeitlinger, B.V. Wine, J. D. (1980). Cognitive-attentional theory of test anxiety. In I. G. Sarason (Ed.), Test anxiety: Theory, research, and applicatwns (pp. 349-385). Hilisdale, NJ: Lawrence Erlbaum Associates, Inc. Wine, J. D. (1982). Evaluation anxiety: A cognitive-attentional construct. In H. W. Krohne & L. Laux (Eds.), Achievement, stress and anxiety (pp. 207-219). Washington, DC: Hemisphere. Zeidner, M. (1986).Sex differences in scholastic aptitude: The Israeli scene. P e r s d i t y and Individual Differences, 7, 847-852. Zeidner, M. (1987a). Socio-cultural group differences in the level and pattern of SAT scores: The Israeli scene. International Journal of Psychology, 22, 363-373. Zeidner, M. (1987b). Test of the cultural bias hypothesis: Some Israeli findings. Jouml of Applied Psychology, 72,3848. Zeidner, M., Klingman, A., & Papko, 0. (1988). Enhancing students' test coping skills: Report of a psychological health education program. Journal of Educational Psychology, 19, 114-124. Zeidner, M., Nevo, B., & Lipschitz, H. (1988). The Hebrew Version of the Test Anxiety Inventory. Haifa: University of Haifa. Zeidner, M., & Safir, M. (1989). Sex, ethnic, and social differences in test anxiety among Israeli adolescents. Joumai of Genetic Psychology, 150, 175-185.

Moshe Zeidner School of Education University of Haifa Mount Carme131999 Israel Received July 18, 1989 Revised September 28, 1989

Does test anxiety bias scholastic aptitude test performance by gender and sociocultural group?

This study examines the commonly held contention that test anxiety may serve as a source of bias in the scholastic aptitude test performance of gender...
999KB Sizes 0 Downloads 0 Views