alpert & haber 1960 anxiety in academic achievement situations

Upload: hoorie

Post on 14-Apr-2018

506 views

Category:

Documents


32 download

TRANSCRIPT

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    1/9

    Journal of Abnormal an d Social PsychologyI960, Vol.81, No. 2, 207-215

    ANXIETY IN ACADEM IC ACHIEVEMENT SITUATIONS1RICHARD ALPERT

    Harvard University A K DA PART OF the effort to define andmeasure the variables of which testperformance is a function, this studywas designed to evaluate, both experimentallyand theoretically, the paper-and-pencil in-struments currently being used in Americanresearch for the measurement of individualdifferences in anxiety as it affects academicachievement performance. In addition, itincludes a description of a new achievement-anxiety scale which has been devised to indi-cate not only the presence or absence of anx-iety, but also whether the anxiety facilitatesor debilitates test performance.

    Three separate problems are considered: (a)the relationship between scales which aredesigned to measure general anxiety andscales specifically designed to measure testanxiety (specific anxiety scales), and a com-parison of the relative efficacy of the generaland specific scales as predictors of academicachievement performance; (b) the relationshipbetween the construct of anxiety and that ofaptitude and the methodological problemsinvolved in separating these twooperationally;and (c) the direction of the effect of anxietyupon academic achievement performance.

    PROCEDURE AND SUBJECTSProcedurally, this research involved the adminis-tration of a variety of anxiety scales to introductory

    psychology students and to freshmen at StanfordUniversity. The scales used included the TaylorManifest Anxiety Scale (MAS)(Taylor, 1953), theWelsh Anxiety Index (AI) (Welsh, 19S2), the FreemanAnxiety Scale (AS) (Freeman, 1953), the Mandler-Sarason Test Anxiety Scale (TAS) (M andler & Sarason,1952), and the Achievement Anxiety Test (AAT),which has two scales discussed below. Analysis of thedata involved interco rrelating these scales and studyingtheir relationship to a measure of verbal aptitude, the1The support provided by the National ScienceFoundation (G-1787 and G-3045) is gratefully ac-knowledged. We also wish to thank R. R. Sears, Q.McNemar, and Jan Pierce for their help and support,and the Stanford University Office of the Registrarand Computing Center for their cooperation.Part of this report appeared as a portion of adoctoral dissertation at Stanford University by thefirst author.

    RALPH NORMAN HABERYale University

    Scholastic Aptitude Test (SAT) (College Entrance Ex-amination Board, 1956), and to a set of academicperformance indices, including college grade-pointaverage and the final examination, mid-term exam-ination, and course grades in the introductory psy-chology course. Only male 5s were included in thedata analyses. For the students enrolled, the scaleswere either administered during the first class meet-ing of the quarter to the entire class (Fall 1955,Winter 1956, and Spring, 1956, with tfs of 93, 92, and98, respectively), or as part of smaller group experim entswhich were required of all students (Spring, 1956,subsample, N = 40). Therefore, these data are repre-sentative of the population of the students from theintroductory course in psychology, which is taken by75% of the students in the university. For the fresh-men, the scales were group administered as part of thefreshman evaluation program.The scores for the Scholastic Aptitude Test (adminis-tered to the students during their senior year in highschool) and the grade-point averages were obtainedfrom the Registrar of the university. The final exami-nation, course, and midterm grades were obtainedfrom class records. All of these data were collectedwithout the 5s' awareness either that it had beencollected or that data of this nature would be relatedin any way to their responses on the anxiety scales.In view of the finding of Davids (1955) that scoreson anxiety scales are afiected by the 5s' particularmotivation for filling out anxiety questionnaires, itmust be noted that 5s completed the present question-naires with the understanding that the data were beingcollected as part of a large ongoing research programconducted by the Psychology Department and thattheir answers would be used for research purposes only.In some instances, 5s' affiliation motivation might havebeen aroused when, for example, the entire class wasasked to fill out a questionnaire asa favor to the experi-menter; at times when the questionnaire was adminis-tered in experimental session, the fulfillment of courserequirements was the major motivation. In all situ-ations, however, it was emphasized that the datawould no t be made available either to the universityadministration or to the faculty and that no selectionof any kind would be based on the results of the ques-tionnaire. There is no reason to believe that moti-vational differences in any way differentially affectedthe responses.GENERAL VERSUS SPECIFIC ANX IETY SCALESThe first concern in constructing a measuringinstrument which will predict the effect of

    anxiety in academic performance is one ofdefining the population of experiences andbehaviors which the item responses are toreveal. There are current in the literature two

    2 0 7

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    2/9

    208 RICHARD ALPERT AND RALPH NORMAN HABERdivergent positions regarding this matter. Oneposition is exemplified by Taylor (1953) withher Manifest Anxiety Scale and the other byHandler and Sarason (1952) with their TestAnxiety Scale.Taylor's general anxiety scale is made up ofitems drawn from the MMPI and is concernedwith a wide variety of situations other thantest taking. Underlying the construction of theMAS is a theoretical assumption that thereis a relatively constant "level of internal anx-iety or emotionality," and also "that theintensity of this anxiety could be ascertainedby a paper-and-pencil test consisting of itemsdescribing what have been called overt ormanifest symptoms of this state" (Taylor,1953, p. 285). If Taylor is correct in positingsuch a general anxiety state, then it shouldfollow that a single measure of a set of manifestanxiety responses gathered from many situa-tions would be an adequate predictor of thepresence and effects of anxiety responses inany situation, whether it be eyelid conditioning(Spence & Taylor, 1951; Taylor, 1951), thera-peutic sessions (Peck, 1951), serial learning(Taylor & Spence, 1952), or academic per-formance. At present, however, there is someevidence (Child, 1954; Mandler, 1954) ofw/ra-individual differences in anxiety both incontent and intensity from one situation toanother. This evidence warrants further exam-ination of the basic general-anxiety-stateassumption underlying the construction of theTaylor MAS-type scale before applying such ascale as extensively as has been done in thepast. Too often, conclusions have been drawnon the basis of correlations with the MASregarding the presence or absence of anxietyor the effects attributable to anxiety withoutdue consideration of the possible limitationsof a general scale of this type as a sensitiveindicator of anxiety in any limited, recurringtype of situation.

    Mandler and Sarason, who represent thealternative current position, maintain that theitems composing the measuring instrumentshould be concerned with the specific situationsin which it is to be used. These investigators,studying stress in academic achievementsituations, devised the Test Anxiety Scale(TAS), "...a questionnaire which was spe-cifically concerned with the Ss attitudes and

    experiences in a testing situation" (Mandler &Sarason, 1952, p. 166). Thus, the TASprovidesa score indicating the recalled intensity ofcertain experiences and behaviors immediatelyantecedent to or concomitant with the takingof various types of examinations in the past.The implication of this type of scale is thatthe increased situational specificity of its itemcontent will allow for a more sensitive meas-urement of anxiety and its effect in the aca-demic achievement situation. It follows thatfor other situations other specific scales will beneeded; and, only if these specific scales, eachshown to be valid in its own situation, turnout to be highly intercorrelated may one posita general anxiety state and thus justify asingle measure of anxiety for all situations.The relative merits of situational specificityversus generality of item content can beevaluated by two statistical techniques. Thefirst is a comparison of the intercorrelationsamong a variety of general and specific scalesadministered to the same subjects. The TaylorMAS, the Welsh AI, and the Freeman AS areused as measures of general or manifest anx-iety; the Mandler-Sarason TAS and the twoscales of the AAT are used as measures ofspecific anxiety. High intercorrelations amongall six scales would support the theoreticalassumption of a single underlying state, andthus the appropriateness of a single generalmeasure. On the other hand, low intercorrela-tions would support the use of scales whichare specific to the situations in which they aretopredict. Table 1reports the intercorrelationsof the measures of anxiety for an N of 40(the only sample that received all six scales).Because in all instances inspection of thescatter diagrams revealed no departure fromlinearity, only product-moment correlationcoefficients were computed.As Table 1 indicates, the correlations amongthe general anxiety scales range from .32 to .39.The correlations between the general andspecific scales range from .24 to .38. The cor-relations among the specific anxiety scalesrange from .40 to .64, all of which appear tobe higher than any of the correlations involvingthe general scales. This observation is sup-ported by significance tests of the differencesbetween correlations, which indicate that inmost instances the intercorrelations among

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    3/9

    ANXIETY IN ACADEMIC ACHIEVEMENT SITUATIONS 209TABLE 1

    INTESCORRELATION A M O N G Six MEASURES OF ANXIETY(N = 40)Test

    General AnxietyScales:MASAIAS

    Specific AnxietyScales:TASAAT-AAT+

    REL

    .89.84.73

    .82.87.83

    AI

    .39*"

    AS

    .32*.34*

    TAS

    .32*.2 8

    .38*

    AAT-

    .38*

    .37*

    .30

    .64**

    AAT+

    -.33*-.25-.24

    -.40*-.48**

    TABLE 2CORRELATIONS BETWEEN Six ANXIETY SCALES ANDFOUR MEASURES OE ACADEMIC PERFORMANCE(N = 40 to 379)

    Note.Reliabilitiesof the scales appear in Column 1.aBased on the nonoverlapping items. When overlapping items

    are included, r .89.* Significant at .05 level.** Significant at .01 level.

    the specific scales are significantly higher thanthe correlations involving the general scaleseither with each other or with the specificscales. This seems to throw some doubt onthe comparability and, therefore, substituta-bility of a general anxiety scale for a specificanxiety scale.A factor analysis was also performed (Alpert,1957) on this matrix of correlations, but al-though the above conclusions were supportedby this analysis, such support is little morethan suggestive because of the hazards ofdrawing conclusions from a factor analysisbased on only 40 cases.A second technique for determining thesubstitutability of the general for the situa-tionally specific anxiety scales, far morerigorous than merely intercorrelating thescales, is to compare the two types of scalesas predictors of various measures of academicperformance. If the general scales predict aswell as the specific scales, there is justificationfor the use of a single general measure for thedetection of anxiety in all situations. If, onthe other hand, the specific scales are able toaccount for significantly more of the variancein the performance measures than are thegeneral scales, then there is justification forthe use of scales whichare specific to the situa-tion in which they are used.

    Table 2 presents the correlations betweenthe six anxiety scales and four measures ofacademic performance. These are average

    Test

    General AnxietyScales:MASAIAS

    Specific AnxietyScales:TAS

    AAT-AAT+

    Grade-PointAverage

    .01-.04-.06

    -.24*-.35*.37*

    CourseGradePsych 1

    -.08-.05.14

    -.21-.26*.23*

    FinalExamGradePsych 1

    -.02-.03.15

    -.16-.28*.26*

    MidtermExamGrade

    Psych 1

    -.19-.22*

    -.32*-.25*.21

    * Significant at .05 level.correlations, based on combined samples withtotal Ns ranging from 40 to 379. Thus, theserelationships are based on all samples forwhich both the appropriate anxiety andper-formance scores were available. Before averag-ing was attempted, a check was made to besure that neither the means and the standarddeviations of all the samples nor the correla-tions to be averaged were significantlydif-ferent. No differences were found in any ofthese comparisons.The data summarized in Table 2 indicatethat the specific anxiety scales are more oftensignificantly correlated with academic per-formance measures than are the generalanx-iety scales. Accepting the 5% level of signifi-cance (two-tailed) as a criterion, only one ofthe 11 available correlations for the generalscales reached significance, while 9 of the12 correlations involving the specific scalesreached significance.

    The implications of the findings are reason-ably clear. Specific anxiety scales and generalanxiety scales measure, to a significant extent,something different. Furthermore, it appearsthat the variable which the specific scalesmeasure, and which the general scales donot,is involved in academic performance to suchan extent that the specific scales are betterpredictors of academic performance than arethe general anxiety scales.One may tentatively conclude, therefore,that further use of a general anxiety scale as

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    4/9

    210 R I C H A R D ALPERT A N D RALPH N O R M A N HABERTABLE 3

    C O R R E L A T I O N S B E T W E E N Six ANXIETY S C A L E S A N D AMEASURE or V E R B A L APTITUDE (SAT)(N = 40 to 379)Teat Verbal AptitudeN

    General AnxietyScales:MASAIASSpecific AnxietyScales:TASAAT-AAT+

    19815340

    154379379

    + .10+.13-.24

    -.18-.29+.21

    nsnsns

    .05.001.001

    an appropriate operation for the measurementof academic achievement anxiety is unwar-ranted. However, the next section, on therelationship of these various anxiety scales toaptitude must be considered before this canbecome a final conclusion.THE R E L A T I O N S H I P B E T W E E N A N X I E T Y

    A N D A P T I T U D EIn evaluating the efficacy of the variousscales in the academic situation, one musttake cognizance of the relationship betweenthe anxiety measures and intellectual abilitybecause interest is centered in a scale whichpredicts performance variance attributable tosomething other than aptitude. On this basis,it seems desirable for an anxiety scale to cor-relate highly with perform ance and onlyslightly with aptitude.The equivocal results reported in the litera-ture regarding the relationship between anxietyand aptitude become more clear if one takesinto consideration the type of anxiety scaleused, the heterogeneity of the populationmeasured, and the nature of the aptitudemeasure (most importantly, whether it wastimed and, therefore, under pressure or un-timed). (See Alpert, 1957, for a mo re extensivereview of these variables.) Since the presentstudy is concerned with a highly intellectuallyhomogeneous college population and with anaptitude test given under pressure conditions,the literature involving comparable popula-tions and test conditions is of most directconcern. It indicates that the general anxiety

    scales are not related to timed aptitude testsin homogeneous college populations (Davids& Eriksen, 1955; Schulz & Calvin, 1955),whereas specific anxiety scales are related tothis kind of aptitude test (Sarason & Mandler,1952). The data presented in Table 3 supportthese findings. Here the general anxiety scalesare not significantly related to aptitude, butall the specific scales are significant beyond atleast the 5% level.There are four possible explanations forthis difference:1. The specific scales are in part a measure ofintellectual ability. That is, for some reason,intelligence, independent of actual anxiety level,affects the individual's response to a specificanxiety measure. General scales are free of anintellectual ability component according tothis argument.2. A specific anxiety scale is an appropriatevehicle for rationalization or justification ofpoor academic performance. Because peopleof lower intelligence are more apt to haveexperienced the effects of the results of pooracademic performance in the past, more ofthem would be inclined to use such rationaliza-tion than would people of higher intelligence,and the specific scales are more sensitive torationalization attempts.3. There are stress cues connected with thetaking of timed college aptitude tests. Thesecues elicit anxiety responses which affect theaptitude test performance. The specific scalesare sensitive to the presence of anxiety of thetype reflected in the aptitude tests, but thegeneral scales are not similarity sensitive.

    4. The more intelligent the individual, theless anxiety he manifests in test situationsbecause he has less objective reason to fearthe experience. The specific scales, which areable to measure this interaction, are, there-fore, a sensitive index of anxiety in a collegepopulation, whereas the general scales are not.The last of these arguments is the onlyone which maintains that the obtained correla-tion is a measure of the relationship betweenthe two constructs of anxiety and aptitude.The first three suggest that the correlationbetween the measures is an artifact whichappears because of poor test construction orthe conditions of measurement. While thepresent studies were not designed to evaluate

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    5/9

    ANXIETY IN A C A D E M I C ACHIEVEMENT SITUATIONS 211these various explanations, they do providesuggestive evidence of their relative validity.The first explanation holds that if an anxietyscale which is correlated with an aptitudemeasure is, because of some characteristic suchas its format, simply a measure of aptitude it-self, then one would expect that the aptitudeand anxiety measures would account for thesame variance in academic performance . If thiswere, in fact, the case, then a multiple correla-tion based on the best weighted com bination ofaptitude and anxiety would not be expected toaccount for significantly more of the variancein academic performance than would aptitudealone. Tables 4 and 5 present multiple correla-tions of aptitude and anxiety, predicting grade-point average and final examination grade.Also, they indicate the significance of the addi-tion of the anxiety measure in the predictions.The general anxiety scales are not included inthe tables because they are not correlated w ithachievement performance and hence, couldmake no contribution to the prediction ofsuch perform ance.The debilitating scale of the AAT addssignificantly, over and above aptitude, to theprediction of the grade-point average (GPA)in two out of three instances, and in one outof three for the final examination grade. Thefacilitating scale of the AAT adds signifi-cantly, over and above aptitude, to the predic-tion of GPA in the three instances reported,and in two out of the three instances for thefinal examination grade. The Mandler-SarasonTA S adds significantly to prediction of finalexamination grade, but not for GPA, in theonly samples which data are available. These

    TABLE 4MULTIPLE CORRELATIONS PREDICTING GRADE-POINTAVERAGE (GPA) FROM A COMBINATION OFV E R B A L A P T I T U D E A N D A N X I E T YTest N R* rb r " pd

    TABLE 5MULTIPLE CORRELATIONS PREDICTING FINALEXAMINATION GRADE FROM A COMBINATIONOF APTITUDE AND ANXIETY

    Test N JZ" rb r" t

    AAT- 93 .46AAT- 92 .29AAT- 98 .53TA S 40 .46 -AAT+ 93 .42AAT+ 92 .43AAT+ 98 .58

    GPA Aptitude & Anxiety.b GPA Anxiety.0 GPA Aptitude.d Foradding anxiety toaptitude.

    .45.08.40.2 7.36

    .32.50

    .27.2 9.43.41.27

    .2 9.43

    .001ns.001ns.01

    .01.001

    AAT-AAT-AAT-TA SAAT+AAT+AAT+

    93929840939298

    .49.42.35.40.36.47.32

    -.48-.02-.34-.35.25.2 3.28

    .30.42.16.25.30.42.16

    .001nsns.05.05.05nsaFinal examination grade Anxiety &Aptitude.b Final examination grade Anxiety.0 Final examination grade Aptitude.

    findings refute the argument that the specificscales which are correlated with aptitude aresimply measures of aptitude, because theresults indicate that specific anxiety scales areable to account for variance in academicperformance other than that accounted for bya measure of aptitude. This does not mean tosay that these anxiety instruments are not,in part, measures of intellectual ability; itdoes say, however, that they are somethingmore than that and that therefore they areable to account for added variance in academicperformance, and, as such, are valuable supple-ments to a measure of aptitude.The second explanation holds that thesignificant correlation of specific anxiety scaleswith aptitude is really the result of a m ediatingprocess of rationalization and justification forpoor performance. In order to evaluate thisargument, the entering freshman class wasgiven a specific anxiety scale. Freshmen enter-ing Stanford have, in the majority, a pasthistory of similar academic success experiencesin high school, and they have had no pastexperience with the college academic per-formance criteria which anxiety scales wereto predict. The AAT was administered to theentire freshman class at Stanford during theirorientation week in the Fall of 1955. After theend of their first year, a random sample of96 was drawn and the AAT was correlatedwith their grade-point averages. Correlationsof .42 with the facilitating scale and .2 9with the deb ilitating scale were obtained. Bo thcorrelations are significant beyond the 1%level. Comparing these correlations with thoseobtained for a sample of 70 upperclassmen

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    6/9

    212 RICHARD ALBERT AND RALPH N O R M A N HABERfor whom exactly comparable data were onhand, no significant differences were found.This is a weak refutation of the secondexplanation, at least at the college level. But,of course, college students have had anxietyexperiences regarding high school achievement,and thus, it is still possible that the freshmencould be rationalizing via the AAT for theirhigh school grades. But the two scales werecorrelated w ith high school grade-point average.29 and .21, respectively. These correlationsare both significant at the 5 % level, and theyare not significantly different from .42 and .29. Thus, it seems the da ta cannot provide adefinite answer to the issue involved nor,probably, can any correlational study of thisnature.Third, if a timed aptitud e test arouses stresscues, which elicit anxiety responses whichaffect the aptitude test performance, then acorrelation between an anxiety measure andan aptitude measure could be interpreted tomean that the anxiety scale measures theanxiety which is a component in the aptitudetest situation and which is reflected in theaptitude test scores. If this were true, then itwould follow that an aptitude measure wouldcorrelate m ore highly w ith an anxiety-provok-ing college examination than with one thataroused little anxiety. Because the final exam-inations require more involvement on thepart of the student and count more towardhis total grade, these tests should be moreanxiety-provoking than the midterm examina-tion. From this reasoning, these tw o examina-tions were compared to provide the anxietydifference against which to test the aptitudemeasure. In no sample was there a significantlygreater correlation between aptitude and finalexamination than between aptitude and mid-term examination. These findings constitutea weak refutation of the third explanation.A more definitive technique for evaluationof this explanation would be to administer theaptitude test to a group of students before an dafter they had been subjected to a course ofpsychotherapeutic treatment directed at thereduction of examination anxiety. If the in-crease in the mean aptitude score for thegroup receiving psychotherapy were signifi-cantly larger than any increase in the controlgroup which had not received the psycho-therapy, then we would have strong support

    for the third explanation of the relationshipbetween aptitude and anxiety. This experi-ment would be similar to that by Pomeroy(1950), who demonstrated increases in finger-maze performance after only 10 min. ofcathartic-type therapy. Until such time as ap-propriate data are available, however, onemust hold in abeyance any final conclusionsregarding the validity of this explanation.In summ ary, regarding these three explana-tions, the first is refuted, doubt cast on thesecond, and the third neither refuted norsupported. Present evidence allows one to sayonly that either the third or the fourth explana-tion (that is, that the obtained correlation is ameasure of the relationship between the twoconstructs anxiety and aptitude) may becorrect, or that both of them may be true.Had the first and second explanations beensupported, they would support a preferencefor the use of a general anxiety scale becauseit is free of the contaminating influence ofaptitude. The last two explanations support apreference for a specific anxiety scale becauseof its greater sensitivity. Since the evidencedoes indicate that these explanations are moreprobably correct, one has further supportbeyond that derived from the analysis of theinterrelationships of the anxiety scales for theuse of the specific rather than the generalanxiety scales for prediction in academicperformance situations.

    DIRECTION O F EFEECTA study of the anxiety literature leads tothe question of how anxiety, when aroused in

    an examination, will affect performance; thatis, whether it will facilitate it, debilitate it, orperhaps have no effect on it at all. On thebasis of their earlier theoretical considerations,Sarason, Handler, and Craighill (1952) statethe following hypothesis:When a stimulus situation contains elements whichspecifically arouse test or achievement anxiety, thisincrease in anxiety drive will lead to poorer performance

    in individuals who have test-irrelevant [incompatibleor interfering] anxiety responses in their responserepertory. For individuals without such response t enden-cies, [italics added] these stimulus elements will raisetheir general drive level and result in improved per-formance (p. 561).

    The implication is that measurable anxietyresponses, when present, are debilitating toperformance, an implication which is reflected

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    7/9

    A N X I E T Y IN A C A D E M I C A C H IE V E M E N T S IT U A T I O N S 213in the Mandler-Sarason TAS. All the items ofthe TAS are unidimensional, i.e., anxietyresponses are either debilitating or not (e.g.,"While taking a course examination, to whatextent do you worry? Worry a lot... [to]...Worry not at all." [Item No. 35]). From theabsence of negative responses for an S, thatis, a low score on the test, they infer thatwhen anxiety-provoking cues are present inthe environment, "these stimulus elementswill raise [his] general drive level and result inimproved performance for that S." (Sarason,Handler, & Craighill, 1952, p. 561). Thisresults in their confounding the two alter-natives of a facilitating effect of anxiety andno effect of anxiety on academic performanceand leads to their failure to allow for a thirdalternative, the possible existence of an indi-vidual whose anxiety responses in an anxiety-provoking situation do not affect test per-formance either by improving or by de-pressing the score. Such an individual wouldhave to be described in their formulation asone whose negative or debilitating responseeffect is just counteracted by the increasedtask-performance drive.There is another approach which does notrequire this unwarranted inference and accord-ing to which an anxiety scale would measurethe presence and intensity of both kinds ofanxiety responses, those which facilitate per-formance and those which interfere with it.If an individual scored low on both of thesescale components, he would be consideredto be unaffected (insofar as test performancewas concerned) by anxiety-provoking cues. Inother words, the facilitating effects of anxietywould then be measured independently andnot inferred from the absence of negativeresponses. This independent measure wouldallow for the possibility of the absence ofboth types of response as well as for thepresence of either one.

    Theoretically, Handler and Sarason wouldpredict, on the basis of their ideas, that twosuch independent measures of anxiety wouldbe highly negatively correlated. That is, if anindividual had a great deal of facilitatinganxiety, he would have little debilitatinganxiety and vice versa. This is the assumptionunderlying their scale because it allows themto use only the debilitating index and to inferfacilitating anxiety from the absence of debili-

    tation. This assumption may be unnecessary.In fact, these two constructs of debilitatingand facilitating anxiety may be uncorrelated.Thus, an individual may possess a largeamount of both anxieties, or of one but notthe other, or of none of either. The nature ofthis correlation can be determined empiricallyfollowing the construction of two such inde-pendent measures of anxiety, a scale measuringthe facilitating effects of anxiety on achieve-ment performance, and a separate scale meas-uring the debilitating effects of anxiety onachievement performance.

    The Achievement Anxiety Test (AAT) wasconstructed to make this determination. Itconsists of two independent scales: a facili-tating scale of nine items based on a prototypeof the item"Anxiety helps me to do betterduring examinations and tests"; and a debili-tating scale of 10 items based on a prototypeof the item"Anxiety interferes with myperformance during examinations and tests."Both scales have gone through numerousrevisions based upon item analyses, correla-tions with various criteria, and theoreticalreformulations. The test-retest reliabilities fora 10-week interval are .83 and .87, respectively.The test-retest reliability over an 8-monthperiod is .75 for the facilitating scale and .76for the debilitating scale. The two scales areadministered in one questionnaire, the itemsrandomly mixed (as presented below). The5s answer each item on a five-point scale,indicating the degree to which the item appliesto them. The AAT includes neutral bufferitems in addition to the items listed. Thenumbers in the parentheses to the right are theactual item numbers on the AAT.Facilitating Anxiety Scale

    1. I work most effectively under pressure, as whenthe task is very important. AlwaysNever. (2)2. While I may (or may not) be nervous before tak-inganexam, once I start, I seem to forget to benervous.I always forget I am always nervous during an exam.(9)3. Nervousness while taking a test helps me dobetter. It never helpsIt often helps. (11)4. When I start a test, nothing isable to distract me.This is always true of meThis is not true of me. (12)5. In courses in which the total grade is basedmainly on one exam, I seem to do better than otherpeople. NeverAlmost always. (14)6. I look forward to exams. NeverAlways. (16)7. Although "cramming" under pre-examinationtension is not effective for most people, I find that ifthe need arises, I can learn material immediately before

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    8/9

    214 RICHARD ALPERT A N D RALPH N O R M A N HABERan exam, even under considerablepressure, and success-fully retain it to use on the exam. I am always able touse the "crammed" material successfully I am neverable to use the "crammed" material successfully. (19)8. I enjoy taking a difficult exam more than aneasy one. Always Never. (21)9. The more important the exam or test, the betterI seem to do. This is true of me This is not true ofme. (24)Debilitating Anxiety Scale

    1. Nervousness while taking an exam or test hindersme from doing well. Always Never. (1)2. In a course where I have been doing poorly, myfear of a bad grade cuts down my efficiency. NeverAlways. (3)3. When I am poorly prepared for an exam or test,I get upset, and do less well than even my restrictedknowledge should allow. This never happens to meThis practically always happens to me. (5)4. The mo re imp ortant the e xam ination, the lesswell I seem to do. Always Never. (6)5. During exams or tests, I block on questions towhich I know the answers, even though I might re-member them as soon as the exam is over. This alwayshappens to m e I never block on questions to which Iknow the answers. (10)6. I find that my mind goes blank at the beginningo f an exam, and it takes me a few minutes before I canfunction. I almost always blank out at first I neverblank out at first. (IS)7. I am so tired from worrying about an exam, thatI find I almost don't care howwell I do by the time Istart the test. I never feel this w ay I almost alwaysfeel this way. (17)8. Time pressure on an exam causes me to do worsethan the rest of the group under similar conditions.Time pressure always seems to make me do worse onan exam than others Time pressure never seems tomake me do worse on an exam than others. (18)9. I find myself reading exam questions withoutunderstanding them, and I must go back over them sothat they will make sense. Never Almost always. (23)

    10. When I don't do well on a difficult item at thebeginning of an exam, it tends to upset me so that Iblock on even easy questions later on. This neverhappens to me This almost always happens to me. (26)Part of the refinement of the scales wasaccomplished through means of empiricaltechniques. After a large number of items hadbeen constructed for each scale, the scaleswere used to predict several performancecriteria, such as grade-point averages andfinal examination grades. These datawere item-analyzed to give the correlation of each itemwith the criterion. Those items were retainedwhich were highly correlated with the criteriabut which were not correlated w ith each other.In this way, it was hoped to minimize the

    intercorrelation of the scales without affectingtheir validity coefficients. The final correla-tions between the facilitating and debilitatingscales were . 37 , .34, .43, and .48,drawn from four different samples. Fo r thesefour samples, the average means an d standarddeviations for the facilitating an d debilitatingscales of the AAT were as follows: for theA A T +, A f = 27.28, S Z > = 4.27; for the AAT-,M=26.33, SD=5.33. The average correlationfor the combined N (379) was . 37 . All thesecorrelations are significant beyond the 1%level. Hence, in spite of efforts to separatethe two scales empirica lly, a low but significantcorrelation remains. The point seems made,however, that the correlation is neither perfec tnor high as Handler and Sarason imply andrequire as an assumption for the use of theirscale.In general, the two scales of the AAT cor-related about equally, though opposite insign, with the other variables discussed in thispaper (see Tables 1, 2, and 3). The importantquestion is whether, when predicting anaca-demic performance score such as college grade-point average, there is any advantage in usingboth scales rather than using merely a con-ventional debilitating anxiety scale such asthe AAT alone or the Handler and SarasonTAS. These two measures of debilitatinganxiety (theAAT and the TAS) correlate.64 with an N of 40, indicating considerableequivalence. Table 6 answers the question ofwhether a scale which measures the facilitatingeffect of anxiety justifies itself. In the lasttwo columns to Table 6 it can be seen thatfor three separate samples, the multiple cor-relations, using both the plus and minus scalesto predict grade-point average, are signifi-

    TABLE 6MULTIPLE C O R R E L A T I O N S P R E D I C T I N G G R A D E - P O I N TA V E R A G E (GPA) F R O M THE AATN

    939296

    -R".50.32.54

    r*.36.32.50

    rc

    -.45-.08-.40

    r <

    -.37-.34-.43

    #".001ns.05

    t1.05.01.001

    * GPA AAT+&AAT-.bGPA AAT+.0GPA AAT-.d AAT- AAT+.8 Significance of adding AAT1 Significance of adding AAT+.

  • 7/30/2019 Alpert & Haber 1960 Anxiety in Academic Achievement Situations.

    9/9

    ANXIETY IN A C A D E M I C ACHIEVEMENT SITUATIONS 215cantly better predictors than the minus or theplus scales alone.It seems clear, therefore, that the incorpora-tion of items designed to measure facilitatinganxiety into a scale which already effectivelymeasures debilitating anxiety can significantlyincrease the prediction of academic per-formance scores.

    S U M M A R YThe results support the following con-

    clusions:1. Specific anxiety scales (scales having

    items specific to the academic test situation)and general anxiety scales are measuring, to asignificant extent, something different fromone another.2. The specific anxiety scales are betterpredictors of academic performance than arethe general anxiety scales.3. The general anxiety scales are not signifi-cantly related to verbal aptitude, while thespecific anxiety scales are all related toaptitude.

    4. The specific anxiety scales, althoughmore highly correlated with aptitude than thegeneral anxiety scales, are, nevertheless, moreoften than the general anxiety scales, able toaccount fo r variance in academic performanceother than that accounted for by a measure ofaptitude.5. An explanation of the relationship be-tween anxiety scales and measures of aptitude,based on the use of an anxiety scale as avehicle fo r rationalization or justification bystudents with poor past performance in aca-demic situations, is not supported by thesedata.6. On the basis of these data, no conclusionmay be reached about explanations of therelationship between anxiety scales and meas-ures of aptitude on the assumption that apti-tude scores are, themselves, affected byanxiety. A possible experiment which mayclarify this issue is suggested.A scale for the measurement of "facilitating"anxiety was constructed, which added signifi-cantly to the prediction of grade-point averagewhen it was combined with a measure of"debilitating" anxiety. The theoretical andpractical significance of these findings wasdiscussed.

    REFERENCESAIPEKT, R. Anxiety in academic achievement situ-ations: Its measurement and relation toaptitude.Unpublished doctoral dissertation, Stanford

    Univer., 1957.CHILD, I. L. Personality. Anna. Rev. Psychol., 1954, 6,149-170.C O L L E G E E N T R A N C E E X A M I N A T I O N B O A R D . A descrip-tion of the C o l l e g e Board Scholastic Aptitude Test .Princeton: Educ. Testing Serv., 1956.D A V I D S , A. Relationsamong severalobjective measuresof anxietyunder different conditions of motivation./.consult. Psychol., 1955,19, 275-279.D A V I D S , A., & E R T K S E N , C. W. The relation of manifestanxiety to association p roduc tivity and intellectualattainment. /.consult. PsycM., 1955,19, 219-222.FREEMAN, M. J. The development of a test for themeasurement of anxiety: A study of its reliabilityand validity.Psychol. Monogr., 1953, 67(3, WholeNo. 353).HANDLER, G. Anxiety and performance: Empirical andtheoretical aspects of anxiety scales. Unpublishedmanuscript, Lab. Soc. Relat., Harvard Univer.,1 9 5 4 .

    M A N D L E R , G., & S A R A S O N , S. B. A study of anxiety andlearning. /. abnorm. soc.Psychol., 1952, 47, 166-173.P E C K , RUTH. The influence of anxiety factors upon theeffectiveness of an experimental "counsellingsession." Unpublished doctoral dissertation, State

    Univer. Iowa, 1951.P O M E R O Y , D. S. Ameliorative effects of "counselling"upon maze performance following experimentallyinduced stress. Amer. Psychologist, 1950, 6, 327.(Abstract)S A R A S O N , S. B., & HANDLER, G. Some correlates oftest anxiety. /.abnorm. soc. Psychol., 1952, 47,810-817.S A R A S O N , S. B., H A N D L E R , G., & C R A I G H I L L , P. G.The effect of differential instructions on anxietyand learning. /.abnorm. soc. Psychol., 1952, 47,561-565.S C H U L Z , R. E., & C A L V I N , A. D. A failure to replicate

    the finding of a negative correlation betweenmanifest anxiety and ACE scores. /. consult.Psychol., 1955,19, 223-224.S P E N C E , K . W., & T A Y L O R , J A N E T A. Anxiety andstrength of the USC as determiners of the amountof eyelid conditioning. /.exp. Psychol., 1951, 42,183-188.T A Y L O B , J A N E T A. The relationship of anxiety to theconditioned eyelid response. /. exp.Psychol., 1951,41, 81-92.T A Y L O R , J A N E T A. A personality scale of manifestanxiety. /.abnorm. soc.Psychol., 1953, 48, 285-

    2 9 0 .T A Y L O R , J A N E T A., & S P E N C E , K . W. The relationshipof anxiety level to performance in serial learning.J. exp.Psychol., 1952, 44, 61-64.WELSH, G. S. An anxiety index and an internalizationratio for the HHPI. /. consult. Psychol., 1952,16,65-72.

    (Received Hay 21, 1959)