Psychological Bulletin1971, Vol. 76, No. 2, 92- -104

TEST ANXIETY AND DIRECTION OF ATTENTION

JERI WINE a

University of Waterloo

The literature reviewed suggests an attentional interpretation, of the adverseeffects which test anxiety has on task performance. During task performancethe highly test-anxious person divides his attention between self-relevant andtask-relevant variables, in contrast to the low-test-anxious person who focuseshis attention more fully on the task. This interpretation was supported byliterature from diverse areas suggesting that (a) highly anxious persons aregenerally more self-preoccupied than are people low in anxiety; (b) the self-focusing tendencies of highly test-anxious persons are activated in testingsituations; (c) those situational conditions in which the greatest performancedifferences occur are ones which elicit the self-focusing tendencies of highlytest-anxious subjects, and the task-focusing tendencies of low-anxious subjects;(d) research examining the relationship between anxiety and task variablessuggests that anxiety reduces the range of task cues utilized in performance;(e) "worry," an attentionally demanding cognitive activity, is more debilitat-ing of task performance than is autonomic arousal. Treatment and researchimplications of this attentional interpretation of test anxiety are brieflydiscussed.

Highly test-anxious persons typically per-form more poorly on tests than do low-test-anxious persons, particularly when the testsare administered under stressful, evaluativeconditions. The literature reviewed in thisarticle suggests that this performance differ-ence is largely due to a difference in theattentional focuses of high- and low-test-anxious persons during task performance.The low-test-anxious person is focused ontask-relevant variables while performing tasks.The highly test-anxious subject is internallyfocused on self-evaluative, self-deprecatorythinking, and perception of his autonomic re-sponses. Since the difficult tasks on which thetest-anxious person does poorly require fullattention for adequate performance, he cannotperform adequately while dividing his atten-tion between internal cues and task cues.

1 This review is based primarily on a dissertationsubmitted to the University of Waterloo in partialfulfillment of the requirements for a doctoral degree.

2 The author wishes to thank the members of herdissertation committee, Marvin Brown, Chairman,Donald Mcichcnbaum, and Herbert Lefcourt for theirhelpful suggestions and advice. Particular gratitudeis expressed to my husband John J. Wine, forsuggesting the basic theoretical approach.

Requests for reprints should be sent to Jeri Wine,Department of Psychology, University of Waterloo,Waterloo, Ontario, Canada.

Marlett and Watson (1968) have statedthis proposition rather well:

The high-test-anxious person spends a part of histask time doing things which are nol task oriented.He worries about his performance, worries abouthow well others might do, ruminates over choicesopen to him, and is often repetitive in his attemptsto solve the task. Any effort to overcome the self-defeating kind of behavior which the high-anxiousperson exhibits should concentrate on allowing himto perform without the constraints which maximizethe negative effects of the avoidance behavior[p. 203].

TEST ANXIETY THEORY AND MEASUREMENT

The ideas presented above are restatementsof test anxiety theory as advanced by itsoriginators (Handler & Sarason, 1952) andlater contributors. Test anxiety theory wasinitially presented by Handler and Sarasonin their article introducing the Test AnxietyQuestionnaire (TAQ). In contrast to generalanxiety scales such as the Taylor HanifestAnxiety Scale (HAS; Taylor, 19S3), the TAQwas constructed as a measure of anxietyproneness in a specific kind of stressful situa-tion—the testing situation. It was expectedthat items which related specifically to thesubject's reactions to testing situations wouldbe more predictive than general anxiety scalesof his behavior in these situations.

92

ANXIETY AND ATTENTION 93

It was assumed that two kinds of drivesare evoked in the testing situation. The firstof these are learned task drives, which arereduced by "responses or response sequenceswhich lead to completion of the task [Mandler& Sarason, 1952, p. 166]." The second kindis a learned anxiety drive, which can elicittwo classes of responses: those related to taskcompletion, which are anxiety reducing, andthose which interfere with task completion.This latter response class is the one which theTAQ was constructed to measure:

These responses . . . may be manifested as feelingsof inadequacy, helplessness, heightened somatic re-action, anticipations of punishment or loss of statusand esteem, and implicit attempts at leaving the testsituation. It might be said that these responses areself rather than task centered [p. 1661.

The task-relevant responses are ones whichare specific to the task, and must be learnedduring task performance. The self-orientedresponses are a class of generalized responseswhich are readily evoked in a task situation,and interfere with the learning of specifictask-relevant responses. The TAQ, a 37-itemquestionnaire in a rating-scale format, isdesigned to measure these self-oriented re-sponses. The items refer to the kinds ofinternal responses typically experienced bythe subject immediately before and duringexaminations and tests.

I. Sarason (19S8a) constructed a 21-itemmeasure of test anxiety, which he labeled theTest Anxiety Scale (TAS). It is based largelyon items taken from Mandler and S. Sarason'sTAQ rewritten for a true-false format. TheTAQ and the TAS are highly correlated,product-moment correlation = .93 (I. Sara-son, Pederson, & Nyman, 1968). I. Sarasonhas restated the interfering response hypothe-sis, emphasizing that it is a habit interpreta-tion of anxiety:

This interpretation, briefly put, states that 5s scoringhigh and low in anxiety differ in the response tend-encies activated by personally threatening conditions.Whereas low scoring 5s may react to such conditionswith increased effort and attention to the task athand, high scoring Ss respond to threat with self-oriented, personalized responses [1960, p. 405].

As Sarason noted, the attentional focuses ofhigh- and low-anxious subjects differ whenunder threat. The low-anxious subjects turn

their attention to the task, while high-anxioussubjects attend to their internal self-orientedresponses.

Alpert and Haber (1960) constructed a testanxiety questionnaire, the Achievement Anx-iety Test (AAT), which yields two measures.One of these is a 10-item debilitating anxietyscale, a construct similar to that measuredby the TAQ and TAS. Total scores on thescale correlated .64 with total TAQ scores.The other measure is labeled facilitating anx-iety. Mandler and S. Sarason (1952) hadtheorized that test anxiety is debilitating ofperformance only among subjects who hadlearned a habitual class of interfering re-sponses to the test anxiety. Among personswithout this class of interfering responses,test anxiety elicits task-relevant responsesleading to task completion. Persons scoringlow on the TAQ were implicitly assumed byMandler and Sarason to be high in the facili-tating kind of anxiety. Alpert and Haber'sAAT measures separately the two constructsof anxiety and their associated responses. Thenine items on the facilitating anxiety scaleare based on "a prototype of the item—'Anxiety helps me to do better during exami-nations and tests [p. 213].' " The facilitatingand debilitating scales correlated —.48 witheach other (Alpert & Haber, 1960).

Throughout this review, the terms "hightest anxious" and "low test anxious" refer topersons who score at the high and low ex-tremes of measures of debilitating test anx-iety, such as the TAQ, TAS, or debilitatinganxiety scale of the AAT.

SELF-FOCUSING TENDENCIES OF HIGH-ANXIOUS SUBJECTS: EMPIRICAL

EVIDENCE

In a review of the literature on paper-and-pencil anxiety scales, I. Sarason (1960) citeda number of studies that provide evidencethat high-anxious subjects are "more self-deprecatory, more self-preoccupied, and gen-erally less content with themselves than sub-jects lower in the distribution of anxietyscales [p. 404]." This conclusion was pri-marily based on the results of studies whichreported relationships between scores on anx-iety scales and other paper-and-pencil person-ality measures. In general, these investigations

94 JKRl WINE

indicated that persons who score high onanxiety scales describe themselves in negative,self-devaluing terms on other pcncil-and-papermeasures as well.

Two of the studies reviewed in this con-nection by I. Sarason reported relationshipsbetween test anxiety and responses of sub-jects in experimental situations. En one ofthese (Doris & S. Sarason, 19SS) subjectsdiffering in TAQ scores were arbitrarily failedon a number of tasks. Following the tasks,the subjects were required to rank order"blame statements," which included "self-blame" and "other-than-self" blame items.Highly anxious subjects blamed themselvesfor their failures significantly more thandid low-anxious subjects. In the otherstudy, Trapp and Kausler (19S8) comparedWechsler-Bellevue digit-symbol performanceand levels of aspiration of high- and low-test-anxious subjects. Through the actual perform-ance of the high- and low-TAQ groups didnot differ, the levels of aspiration of high-anxious subjects became progressively lowerover the four trials. On the last two trials, thehigh-test-anxious subjects had significantlylower level-of-aspiration scores than the low-anxious subjects. Though they did not ob-jectively perform more poorly than the low-TAQ subjects, the high-anxious subjects pro-gressively became more pessimistic abouttheir performance over the four trials.

A more recent study by Meunier and Rule(1967) reported results leading to similarconclusions regarding the test-anxious person'stendency to devalue his own performance.The effects of positive, negative, or no feed-back on the subjects' confidence in theirjudgments of the length of lines were investi-gated. On no-feedback trials, highly test-anxious subjects rated their confidence levelas low and comparable to their confidence ontrials with negative feedback. In contrast, low-test-anxious subjects expressed high confidencein their judgment on no-feedback trials, andthis level of confidence corresponded to thelevel expressed on positive feedback trials.

Since publishing his review of anxiety scalestudies, I. Sarason has completed severalstudies which further indicate the self-dep-recatory, self-ruminative tendencies of highlytest-anxious subjects (Sarason & Ganzcr,

1962, 1963; Sarason & Koenig, 1965). Ineach of these studies, an unstructured verbalconditioning paradigm was used with collegestudents who scored at the extremes of theTAS. The subjects were required to describethemselves orally for approximately one-halfhour. Nonreinforcement, reinforcement ofnegative self-references, and reinforcement ofpositive self-references have been compared.Some of the conclusions from these experi-ments arc: (a) Regardless of experimentalcondition, highly test-anxious subjects gen-erally describe themselves in more negativeterms than do low-test-anxious subjects, (b)High-tcst-anxious subjects are extremely re-sponsive to reinforcement when the responseclass being reinforced is negative self-refer-ences, (c) However, when the response classbeing reinforced is positive self-references,high-anxious subjects do not condition. Thatis, they do not produce more positive self-references as a result of being verbally rein-forced for them.

The conclusion that high-anxious personsare, indeed, generally self-deprecatory andself-preoccupied seems inescapable in light ofthe research evidence. But whether (hesc self-focusing tendencies are specifically activatedin testing situations is a question which hasnot been answered in the studies reviewed tothis point. Four recent studies (Ganzer, 1968;Mandler & Watson, 1966; Marled & Watson,1968; Neale & Katahn, 1968) provide evi-dence bearing directly on this proposition.

Tn the earliest of these studies, Mandlerand Watson (1966) administered a scries ofdigit-symbol tasks to extreme low- and high-TAQ groups. A posttask questionnaire in-cluded the question, "How often during thetesting did you find yourself thinking howwell, or how badly, you seemed to be doing?|p. 276| ." On a 10-point rating scale, high-test-anxious subjects indicated markedlygreater occurrence of such thoughts than didthe low-test-anxious group. In a partialreplication of this study, Neale and Katahn(1968) reported identical results on thisquestionnaire item.

Marlctt and Watson (1968) reported asimilar result. Ninth-grade males in high- andlow-extreme groups on the children's form ofthe Alpert-Habcr Achievement Anxiety Test

ANXIETY AND ATTENTION

(Stanford, Dember, & Stanford, 1963) werearbitrarily failed for 12 trials on a button-pressing task. A posttask questionnaire in-cluded the question, "How often did you thinkabout how well or badly you were doing?"The ends of the rating scale were labeled"I didn't think about it," and "I thoughtabout it so much I couldn't concentrate [pp.202-203]." The scores of high-test-anxioussubjects indicated that they were significantlymore disturbed by such self-focused thoughts.

Even more direct evidence for the activa-tion of self-oriented tendencies among high-anxious subjects in test situations is providedin the study reported by Ganzer (1968). Theeffects of audience presence and test anxiety(TAS) on the serial verbal learning of femalesubjects were investigated. Tabulations weremade of the frequency and content of all ofthe subjects' task-irrelevant comments whilethey were working on the task. "High TASscorers, especially in the Observed condition,emitted more than any other group. Contentanalysis revealed that the comments weremostly of a self-evaluative or apologeticnature | Ganzer, 1968, p. 194]."

In summary, there is abundant evidencethat the test-anxious person is more self-preoccupied and self-deprecatory than hislow-test-anxious counterpart. It is also clear,in view of the research reviewed, that theseself-focused tendencies are activated by thepressures of the testing situation.

SITUATIONAL CONDITIONS AND THE TASKPERFORMANCE OF HIGH- AND Low-

TKST-ANXIOUS SUBJECTS

The two initial studies published by theauthors who originated test anxiety theory(Handler & S. Sarason, 19S2; S. Sarason,Mandler, & Craighill, 1952) were prototypesof the many subsequent investigations of situ-ational stress variables and test anxiety. Thegeneral predictions of test anxiety theory withregard to situational stress variables are:

When a stimulus situation contains elements whichspecifically arouse test or achievement anxiety, thisincrease in anxiety drive will lead to poorer per-formance in individuals who have task-irrelevantresponses in their response repertory. For individualswithout such response tendencies, these stimuluselements will raise their general drive level and result

in improved performance. (Such elements would heany reference that the individual is being judged, astatement of expected performance, etc.) [Sarasonet al., 1952, p. 5611.

Handler and Sarason (19S2) examined theeffects of feedback and no-feedback on theperformance of subjects scoring at extremeson the TAQ. As predicted, the two feedbackconditions, both success and failure, facili-tated the performance of low-TAQ subjects,and interfered with the performance of high-TAQ subjects. In the second study (Sarasonet al., 1952) similar predictions were madeconcerning an instructional manipulation. Inthe stressful "ego-involving" instructionalcondition, subjects were told that they shouldeasily complete the test within the time limit,when it was, in fact, made impossible forthem to do so. This "expected-to-fmish" con-dition can be described as an advance failuremanipulation. The nonstressful instructions,"not-expectccl-to-finish," informed the sub-ject that no one could expect to finish withinthe time limit. In general, the high-TAQ sub-jects did better following the "not-expected-to-finish" than the "expected-to-finish" in-structions, while the reverse was true for thelow-anxious subjects.

The many subsequent investigations ofstressful situational conditions and test anx-iety have generally provided data consistentwith the predictions of test anxiety theory(Cox, 1966, 1968; Ganzer, 1968; Harleston,1962; Paul & Eriksen, 1964; Russell & I.Sarason, 1965; I. Sarason, 1958b, 1959, 1961;I. Sarason & Harmatz, 1965; I. Sarason &Hinard, 1962; I. Sarason & Palola, 1960).Most of these studies have investigated theeffects of feedback or instructional manipula-tions, though three were concerned with theeffects of audience presence on task perform-ance (Cox, 1966, 1968; Ganzer, 1968). Con-sidering audience presence as a source ofthreatening evaluation, the results of bothstudies support the predictions of test anxietytheory. Audience presence debilitated the taskperformance of high-tcst-anxious subjects,and facilitated the performance of low-test-anxious subjects.

Instructional variations, other than thosediscussed above, have most frequently 'in-volved comparisons of (a) "ego-involving,"

96 JERI WINE

"highly motivating" instructions, which informthe subject that his task performance willreflect his ability, (b) minimal task instruc-tions, and (c) instructions which assure thesubject that his task performance will not beused to evaluate him. Frequently, the subjectis told that his performance will be anonymous.He may be instructed that the experimenteris interested only in properties of the task,or that his performance is important only sothat the experimenter may complete the study.

The results of studies varying instructionalconditions have generally reported an inter-action between level of test anxiety and evalu-ation emphasis: (a) Highly test-anxious sub-jects perform more poorly following highlyevaluative "ego-involving" instructions thannonevaluative "anonymous" instructions whilethe reverse is true for low-test-anxious sub-jects, (b) Following highly evaluative instruc-tions, low-test-anxious subjects perform betterthan high-test-anxious subjects; followingnonevaluative instructions, high-test-anxioussubjects perform better than low. (c) Fol-lowing minimal task instructions, high- andlow-test-anxious-subjects perform at aboutequivalent levels, intermediate between theirperformances in the highly evaluative andnonevaluative conditions.

All of the situational conditions which dif-ferentially affect the performance of high-and low-anxious subjects vary on evaluativedimension—the degree to which the subjectbelieves that he is being or will be evaluatedon the basis of his task performance. Condi-tions at the extremes of the dimension affectthe performance of high- and low-test-anxiouspersons in opposing ways. It might also benoted that situational conditions varying onthis evaluative dimension have opposing ef-fects on the direction of attention of high-and low-test-anxious persons. Highly evalua-tive conditions cause the highly test-anxiousperson to direct his attention internally ratherthan to the task. The same conditions causethe low-test-anxious person to increase his"attention to the task at hand [T. Sarason,1960, p. 405]." Nonevaluative situationalconditions do not elicit the highly test-anxious subject's self-directed interfering re-sponses, thus it is possible for him to directhis full attention to the task. Conversely, non-

evaluative testing conditions presumably donot excite the nonanxious subject's motivationand interest. Though it cannot be said that hetherefore turns his attention inward, it isprobable that his low interest in the taskis accompanied by less than full attentionto it.

I. Sarason (1958b) has investigated anotherinstructional variation, which he labeled a"reassurance" condition. These instructions,like the nonevaluative ones discussed above,were designed to improve the performance ofhigh-test-anxious subjects rather than inter-fere with it. In spite of the "reassurance"label, these instructions can most accuratelybe described as neutral with respect to theevaluative dimension, since there was no ref-erence in them to evaluation of the subject'sperformance. Their nature might more accu-rately be described as information-giving andattention-directing. The subject's task was tolearn a serial list of nonsense syllables. Fol-lowing the standard serial learning taskinstructions, the subject was warned thatmany people become tense when learning thiskind of task and that he should expect hisprogress to be slow. He was further told thathe should concentrate on the list rather thanworrying about how he was doing. When thisset of instructions was compared to standardor minimal task instructions, high-test-anxioussubjects performed better following the reas-surance instructions. The reassurance condi-tion was detrimental to the performance ofthe low-test-anxious subjects, in comparisonto the standard task instructions. Sarason'sstudy suggests that it may not be necessaryto resort to removing the evaluative emphasisfrom the highly anxious subject's perform-ance in order to improve it. The results indi-cate that it may be sufficient to substitutetask-relevant responses for the self-orientingones by means of rather straightforwardattention-directing instructions.

ANXIETY AND RANGE OF CUE UTILIZATION

It was suggested in the preceding sectionthat evaluative testing conditions have op-posite effects on the attentional focuses ofhigh- and low-test-anxious persons. Whenbeing evaluated, the high-test-anxious personturns his attention inward while the low-test-

ANXIETY AND ATTENTION 97

anxious person focuses more fully on the task.The implication is that the high-test-anxiousperson attends to fewer task cues than doesthe low-test-anxious person. This attentionalinterpretation of the effects of test anxiety ontask performance is consistent with an em-pirical generalization advanced by Easter-brook (1959) concerning the relationship be-tween arousal level and task variables. Hereviewed a large body of research which indi-cated that emotional arousal consistently nar-rows the range of cue utilization in taskperformance.

The following are examples of the researchEasterbrook cited to support this generaliza-tion: (a) Studies of the effects of experi-mentally manipulated drive level on responseto a focal task, and simultaneous response toperipheral, only occasionally relevant stimuli.Arousal reduces responsivity to the peripheralstimuli, while maintaining or improving cen-tral performance (Bahrick, Fitts, & Rankin,1952; Bursill, 1958; Davis, 1948; Easter-brook, 1953). (b) Studies of the effects ofemotional arousal on measures of incidentallearning. Again arousal reduces the amountof incidental learning, while having no effecton or improving intentional learning (Aborn,1953; Bahrick, 1954; Bruner, Matter, &Papanek, 1955; Johnson, 1952; Kohn, 1954;Silverman, 1954; Silverman & Blitz, 1956).These are only a few of the studies cited byEasterbrook in support of his formulation.

Easterbrook's definition of emotionalarousal is very broad:

the innate response to a state of biological depriva-tion or noxious stimulation, which underlies oroccurs simultaneously with overt action and affectsits strength and course. This emotional arousal isgreater in neurotic than in normal subjects, greaterthan usual in subjects under stress or threat or infrustration, and in general greater in animals thathave been "motivated" by any of the usual depriva-tions, noxious stimulations, or other incentives thanit is in unmotivated or resting animals of the samespecies [p. 184].

Anxiety is one aspect of this emotionalarousal dimension, and several of the studiesreviewed by Easterbrook indicate that narrow-ing the range of cue utilization accompaniesanxiety arousal.

Since Easterbrook's review, several studieshave reported the effects of experimentally

aroused anxiety on range of cut utilization intask performance (Agnew & Agnew, 1963;Bruning, Capage, Kozuh, Young, & Young,1968; Tecce & Happ, 1964; Wachtel, 1966,1968). The results of these investigationshave been consistent with Easterbrook'sformulation. Stressful situational conditionstended to reduce the range of cue utilization,in comparison to nonstressful conditions. Nomeasures of individual differences were usedin these studies.

Zaffy and Bruning (1966), West, Lee, andAnderson (1969) and Wachtel (1966, 1968)reported relationships between scale-measuredanxiety level and range of cue utilization.Zaffy and Bruning selected subjects fromextreme scorers on the Taylor MAS. Ratherthan varying range of cues, they made theavailable cues either relevant or irrelevant totask performance. The subjects' task was tolearn a correct position for each of 19 presen-tations of 5 zeros, presented by a memorydrum. In the relevant cue condition the zeroswere numbered in order with subscripts from1 to 5; the same subscripts were used in theirrelevant cue condition but in random order.These conditions were compared to a no-cuecontrol condition. The performance of thelow-MAS subjects was affected more by thepresence of the cues, whether relevant orirrelevant, than was the performance of thehigh-MAS subjects. Within the low-MAS anx-iety level, there were differences significantbeyond the .001 level in all three of the com-parisons of the three conditions, with superiorperformance in the relevant cue condition,intermediate performance in the no-cue condi-tion, and poorest performance in the irrelevantcue condition. The performance levels of thehigh MAS subjects followed the same order,but there was a significant difference onlybetween the relevant and irrelevant cueconditions (p < .025).

More pertinent to the purposes of this re-view are the studies reported by West et al.(1969) and Wachtel (1966, 1968), since theyselected subjects on the basis of their scoreson measures of test anxiety. West et al. usedsixth and eighth graders who were high andlow scorers on the Test Anxiety Scale forChildren (TASC; S. Sarason, Davidson,Lighthall, Waite, & Ruehush, I960). The task

98 JERl WINE

was made up of arithmetic problems. Therewere two conditions determined by the natureof the information given in the problems. Inone condition only information essential tosolution of the problems was given; in theother condition the same relevant informationwas provided, but additional irrelevant infor-mation was given in each problem. The inter-action between anxiety level and conditionwas significant. Both high- and low-TASCgroups performed more poorly in the irrele-vant information condition. No tests of sig-nificance for individual pairs of means werereported; the means for the groups indicatethat the performance level of the low-TASCgroup was affected more by the addition ofirrelevant cues than was the performance levelof the high-TASC group: low TASO, relevantinformation only, X = 33.4, relevant plus ir-relevant, X"23.8; high TASC, relevantonly, X — 27.3, relevant plus irrelevant,X = 24.4. The low-tcst-anxious children weremore sensitive to the addition of irrelevantcues than were the high-test-anxious children.

Wachtel's study (1966, 1968) reported rela-tionships between test anxiety level and rangeof cue utilization in task performance on acentral continuous tracking task and reactiontimes to two occasional peripheral lights. Acontrol condition with minimal task instruc-tions was compared with three experimentalconditions: (a) Subjects told that unavoid-able electric shock would be administeredduring task performance; (b) an avoidableshock condition in which subjects were in-formed that they could avoid shock if theircombined performance on the central andperipheral tasks remained high; (c) a per-sonality diagnostic condition in which thetask was presented as a means of evaluatingthe subject's personality; the subject was toldthat the peripheral lights would go on onlyif his performnce fell below an acceptablelevel.

Though Wachtel's study was primarilyfocused on the effects of experimentally ma-nipulated stress on performance, he reportedinteresting relationships between subjects'TAQ scores and task performance. The"avoidable shock" and "personality diagnosis"conditions were the only ones in which testanxiety was appropriately engaged through

threatening the subject with evaluation of hisperformance. In both of these conditions therewas evidence of narrowing attention to taskcues as a result of test anxiety. In the avoid-able shock condition there was a nearly sig-nificant r of —.35 (p < .10) between speedscores (reciprocals of reaction times to theoccasional peripheral lights) and subjects'TAQ scores, and a significant positive cor-relation between TAQ scores and scores onthe central tracking task (r — .58, p < .01).In the personality diagnosis condition thecorrelation between speed scores and TAQscores was —.44 (p < .05), though the cor-relation between TAQ scores and trackingscores in this condition was only .02. Whenall subjects in these two conditions werecombined, the correlation between TAQ scoresand speed scores was —.40 (n — 30, p < .01).Thus in these two conditions test anxietyreduced responsiveness to the peripherallights. In the other two conditions, in whichtest anxiety was not appropriately aroused,there were no significant correlations betweenTAQ scores and task performance.

A proposition offered by Wachtel to explainthe effects of experimentally manipulatedanxiety and test anxiety in this study com-bines Easterbrook's attentional formulationand the interfering response hypothesis of thetest anxiety theorists: "when an individual isanxious, attention is diverted inward to per-ception of his anxiety and therefore lessattention is available for external stimuli[Wachtel, 1966, p. 2149]."

WORKY AND EMOTIONALITY: COMPONENTSOF TEST ANXIETY

Liebert and Morris (1967) have presentedan analysis of test anxiety which is quiteconsistent with the one proposed in this paper.They suggest that test anxiety is composedof two major aspects: worry and emotionality.The worry (W) component is described ascognitive concern over performance, andemotionality (E) is the autonomic arousalaspect of anxiety. Liebert and Morris deviseda brief situational measure of W and E basedon TAQ items. The results of a series ofstudies (Doctor & Altman, 1969; Liebert &Morris, 1967; Morris & Liebert, 1969, 1970;Spiegler, Morris, & Liebert, 1968) suggest

ANXIETY AND ATTENTION 99

that scores on the two components vary pre-dictably with temporal relationships to class-room examinations and with performanceexpectancies. The scale has been administeredseveral days before an examination, immedi-ately before, and immediately after. Scores onW tend to be fairly constant across time;while E scores reach a peak immediately be-fore an examination, falling off rapidly im-mediately after the examination. W scores aresignificantly and negatively correlated withsubjects' preexamination ratings of perform-ance expectancy, while E scores are notrelated to performance expectancy.

Three studies in this scries have yieldedresults on the relationship between worry andemotionality and task performance (Doctor& Altman, 1969; Morris & Liebert, 1969,1970). Morris and Liebert (1969) devisedanother paper-and-pencil measure of the twoanxiety components, this one composed ofitems from the Taylor MAS. The perform-ance measure was a total of five WechslerAdult Intelligence Scale (WAIS) subtests. A2 x 2 factorial design was used: half of thesubjects completed only the easy items ineach subtest, half the difficult items; half ofthe subjects were aware of being timed, halfwere not. There were significant interactionsbetween worry scores and both task difficultyand timing in their effects on total WAISscores. There were no significant results in-volving emotionality and task performance.The authors suggest that "it is worry, not'anxiety,' which affects performance onintellectual-cognitive tasks and which inter-acts with the relevant variables of the testsituation [pp. 243-244]."

Doctor and Altman (1969) examined therelationships between W and E scores onLiebert and Morris' (1967) scale and finalexamination grades in a psychology course.W scores were found to be significantly andnegatively related to exam performance. Escores were negatively related to performanceonly among subjects below the group medianW score. When W was high, the additionof E had little or no effect on performance.

Morris and Liebert (1970) reported similarresults in two studies, one involving collegeundergraduates, the other, high school stu-dents. Partial correlations were computed be-

tween worry and emotionality scores on theLiebert-Morris (1967) scale, which was ad-ministered immediately before a course finalexamination and examination grades. In bothstudies, partial correlations between worryscores and final examination grades were nega-tive and significant. Partial correlations be-tween emotionality scores and grades werenonsignificant.

The worry component seems closely relatedto an attentional interpretation of test anx-iety, which proposes that the adverse effects oftest anxiety are clue to attention being dividedbetween self and the task. Worry has beendescribed as "preoccupation with performance[Doctor & Altman, 1969, p. 564]," "cognitiveconcern about the consequences of failing, theability of others relative to one's own, etc.[Liebert & Morris, 1967, p. 9751," "cognitiveor intellectual concern about one's own per-formance | Spiegler et al., 1968, p. 451]."These descriptions all refer to attentionallydemanding cognitive activity going on duringtask performance. This is not to say that auto-nomic activity is not demanding of attentionas well, but it is less likely to require atten-tion except "at high levels where physiologicalreactivity might be distractive and annoying[Doctor & Altman, .1969, p. 364]."

CONCLUSIONS AND IMPLICATIONS

The literature reviewed supports an atten-tional interpretation of the debilitating ef-fects of test anxiety on task performance. Thehighly test-anxious person responds to evalu-ative testing conditions with ruminative, self-evaluative worry and, thus, cannot direct ade-quate attention to task-relevant variables. Anattentional analysis is closely related to exist-ing theories of test anxiety. Mandler andSarason's (1952) original theoretical positionassumed that anxiety interferes with the per-formance only of subjects who have ahabitual class of self-oriented interfering re-sponses in their response repertoires. For sub-jects without this class of interfering re-sponses, anxiety may facilitate task-relevantresponses. Alpert and Haber's AchievementAnxiety Test was designed to provide inde-pendent measures of these two classes of re-sponse to test anxiety. Liebert and Morris'

100 JER1 WINE

(1967) analysis of test anxiety into the twocomponents of worry and emotionality is alogical extension of test anxiety theory. Emo-tional arousal appears to bear no consistentrelationship to performance on intellectual orcognitive tasks, while worry is consistentlyand negatively related to performance. Anattentional interpretation states simply thatthe reason "worry" debilitates task perform-ance is that it is attentionally demanding anddistracts attention from the task.

Before moving into specific research andtreatment applications, a few implications ofthe attentional analysis of test anxiety arestated below:

1. An attentional approach is explicitlyconcerned with how the subject uses his tasktime—his cognitive activity, what he isthinking about and attending to.

2. This approach implies little interest inautonomic arousal per se. In this context,degree of arousal is irrelevant unless the sub-ject is attending to his arousal. Of course, asnoted earlier, when arousal becomes quiteextreme, it is attentionally demanding.

3. Finally, this analysis implies that thetest-anxious person's performance may beimproved by directing his attention to task-relevant variables, and away from self-evalua-tive rumination.

Experimental Applications

Though test anxiety theorists have notviewed test anxiety as a unitary dimensionof emotional arousal, the experimental studiesin this area have typically been concernedwith manipulations of the evaluative situa-tional conditions which evoke test anxiety.With two notable exceptions (Sarason el al.,1968; Sieber, 1969) studies have not investi-gated conditions designed to alter the inter-fering responses elicited by the anxiety. Re-search manipulating evaluative conditions hasbeen necessary in order to establish test anx-iety as a theoretical construct and to validatemeasures of test anxiety. These goals havebeen realized; it seems reasonable to suggestthat an appropriate next stage for research inthis area is one in which the evaluative dimen-sion is held constant, and in which a searchis begun for variables that will alter thetest-anxious subjects' interfering responses.

An attentional analysis of test anxiety canprovide some direction to such a search.

There is a large body of literature in selec-tive attention, which has recently been re-viewed by Egeth (1967) . The basic premiseof the studies in this research area is thatsubjects' attention can be differentially di-rected to specific stimuli or stimulus attributesand away from others. The nature of themechanisms involved, the completeness of theselectivity, and the level at which the selec-tion takes place arc matters of some dispute(e.g., Broadbent, 1958; Deutsch & Deutsch,J963; Treisman, 1964). It is, however, em-pirically established that persons can be in-structed to be selectively more attentive tospecific stimulus attributes or dimensions andto be less attentive to others. Moreover, withrepeated training under attention-directing in-structions, subjects become more and moreskilled in attending to the "relevant" stimuliand ignoring the "irrelevant" stimuli.

The selective attention literature has notbeen concerned with individual differences.There is no evidence, therefore, on how test-anxious persons might respond to attention-directing instructions. There is a good dealof evidence, however, which indicates thathighly test-anxious persons are very respon-sive to social cues of all sorts, including verbalreinforcement (T. Sarason & Ganzer, 1962,1963) modeling cues (1. Sarason et al.,1968), persuasion (Janis, 1955), and con-formity pressures (Mcunicr & Rule, 1967),as well as evaluative instructional manipula-tions, ft seems a plausible hypothesis thattest-anxious subjects should be susceptible toattentional instructions directing them toselectively attend to task-relevant variablesand to ignore self-relevant variables.

Instructional manipulations are the mostobvious means of directing subject's attentionto task-relevant variables, but they are notthe only available means. For example,modeling has been demonstrated as a veryeffective means of transmitting new behavior(e.g., Bandura, 1969; Bandura & Walters,1963). T. Sarason et al (1968) and 1. Sarason(1968) have reported studies suggesting thatthe verbal learning of highly test-anxious sub-jects is facilitated by prior exposure to busi-nesslike live models learning similar serial

ANXIETY AND ATTENTION 101

lists. The use of models who transmit task-attending, nonworrying cues may be an ef-fective technique for other kinds of learningsituations.

Task cues themselves may provide anothermeans of manipulating the attention of test-anxious subjects. Sieber (1969) has reporteda study in which high- and low-test-anxiouschildren worked on a multistage task. In onecondition, they were provided with visual dis-plays of all of the earlier stages of the task;in another condition, there were no displays.The highly test-anxious children performedbetter in the visual display than the no-display condition, while the displays had nodifferential effect on the performance of thelow-test-anxious children. Seiber describedthe displays as "memory supports." Theymight also be described as means of sustain-ing attention to task-relevant variables. Itshould be possible to develop such "memory-supporting" or "attention-sustaining" tech-niques for use with other tasks and age levels.

To this point, the research possibilities dis-cussed have been concerned only with meansof manipulating the attention of highly test-anxious persons. These suggestions do notexhaust the research possibilities implied byan attentional approach. The approach alsosuggests analysis of other areas, for instance,research designed to examine in detail themanner in which high- and low-test-anxiouspersons deploy their attention during taskperformance. Such research would inevitablylead to concern with the properties of specifictasks and the attentional demands they makeon subjects.

Treatment Applications

The purpose of this section is not to reviewin any detail the treatment literature in testanxiety, but rather to suggest possible treat-ment techniques which might be derived froman attentional approach. The test anxietytreatment literature is described briefly, how-ever, in order to provide some contrast withan attentional treatment approach.

All of the published accounts of treatmentof test-anxious subjects have involved varia-tions of systematic desensitization techniques(Cohen, 1969; Donner & Guerney, 1969;Emery & Krumboltz, 1967; Garlington &

fotler, 1968; 'Ihli & Garlington, 1969; John-son & Sechrest, 1968; Katahn, Strenger, &Cherry, 1966; Kondas, 1967; Laxer, Quarter,Kooman, & Walker, 1969; Snider & Getting,1966; Suinn, 1968). This literature has con-tributed a graet deal to analysis of the rela-tive efficacy of various components of thesystematic desensitization procedures, for in-stance, standardized versus individualizedanxiety hierarchies, partial versus full hier-archies, relaxation only versus full desensi-tization, etc. The research indicates in gen-eral that systematic desensitization has bene-ficial effects on self-reported measures of testanxiety, and often, on scholastic performance.Therefore, it is not with any intent tominimize the value of this treatment literaturethat it is noted that these studies have evolvedfrom interest in specific treatment techniques,rather than from an analysis of the natureand effects of test anxiety. Test anxiety hasbeen considered to be a severe problem, andthus its treatment is considered a stringenttest of systematic desensitization procedures.It has implicitly been assumed that test anx-iety differs only in degree from the specificanxieties and phobias (e.g., rat, spider, snake)dealt with in much of the behavior modifica-tion literature. It should also be noted thatby training test-anxious subjects to relax inthe presence of progressively more stressfulstimuli, a systematic desensitization treatmentapproach assumes that the emotional arousalcomponent of test anxiety is its definingcharacteristic.

The attentional formulation suggests quitea different approach to treatment of test-anxious persons. First of all, the specific con-cern with how highly test-anxious subjectsuse their task time suggests treatment: inwhich subjects are given intensive practice indealing with tests. Second, this task practiceshould be accompanied by instructions tofocus fully on the tasks and to inhibit self-relevant thinking. In a pilot study, the author(Wine, 1970) has compared the relative effi-cacy of attentional training of the kind de-scribed, attentional training plus relaxationtraining, and a "self-explorational" therapyapproach with test-anxious undergraduate stu-dents. Each treatment condition was con-ducted over six hourly sessions. Test-anxious

102 JERI WINE

subjects in the atlentional training conditionchanged positively and significantly on self-report measures of test-related anxiety, aswell as on two standardized performance mea-sures. The performance measures were Forms1 and 2 of the Wondcrlic Personnel Tests(Wonderlic, 1959) and matched forms of adigit-symbol test.3 The attentional trainingplus relaxation also produced positive changes,but added nothing beyond attentional train-ing alone. Subjects in the self-explorationalcondition did not change significantly on anymeasures.

A larger, more adequate treatment study isbeing planned for the near future. Though theresults of this pilot study are no more thansuggestive, one of the suggestions is that at-tentional training not only improves task per-formance but lowers reported test anxiety levelas well. In conclusion, perhaps it is possible todefine test anxiety attentionally. If a personis not attending to his test anxiety, it, ineffect, does not exist. "My experience is whatI agree to attend to [William James, 1890,p. 403]."

:! The digit-symbol tests were selected from a bat-tery of eight, constructed and standardized byMarvin Brown, University of Waterloo, 1968.

REFERENCES

AHOKN, M. The influence of experimentally inducedfailure on the retention of material acquiredthrough set and incidental learning. Journal ofExperimental Psychology, 1953, 45, 225-231.

AONKW, N., & AGNKW, M. Drive level effects on tasksof narrow and broad attention. Quarterly Journalof Experimental Psychology, 1963, 15, 58-62.

AT.PKUT, R., & HABKK, R. N. Anxiety in academicachievement situations. Journal of Abnormal andSocial Psychology, 1960, 61, 207-215.

BAIIUIOK, H. P. Incidental learning under two incen-tive conditions. Journal of Experimental Psychol-ogy, 1954, 47, 170-172.

BAIIKJCK, H. P., FITTS, P. M., & RANKIN, R. E.Effects of incentives upon reactions to peripheralstimuli. Journal of Experimental Psychology, 1952,44, 400-406.

BANIHJKA, A. Principles of behavior modification.New York: Holt, Rinehart & Winston, 1969.

BANDURA, A., & WALTERS, R. H. Social learning andpersonality development. New York: Holt, Rine-hart & Winston, 1963.

BROADBENT, D. E. Perception and communication.New York: Pcrgamon Press, 1958.

BIUJNER, J. S., MATTER, J., & PAPANEK, M. L. Breadthof learning as a function of drive level and mecha-nization. Psychological Review, 1955, 62, 1-10.

BRUNINO, J. L., CAPAGE, J. E., Kozuu, G. F., YOUNG,P. E., & YOUNG, W. E. Socially induced drive andrange of cue utilization. Journal of Personality andSocial Psychology, 1968, 9, 242-244.

BURSILI,, A. E. The restriction of peripheral visionduring exposure to hot and humid conditions.Quarterly Journal of Experimental Psychology,1958, 10, 113-129.

COJIEN, R. The effects of group interaction and pro-gressive hierarchy presentation on desensitizationof test anxiety. Behaviour Research and Therapy,1969, 7, 15-26.

Cox, F. N. Some effects of test anxiety and presenceor absence of other persons on boys' performanceon a repetitive motor task. Journal of ExperimentalChild Psychology, 1966, 3, 100-112..

Cox, F. N. Some relationships between test anxiety,presence or absence of male persons, and boys'performance on a repetitive motor task. Journal ofExperimental Child Psychology, 1968, 6, 1-12.

DAVIS, D. R. Pilot error. (Air publication 3139A)London: Her Majesty's Stationery Office, 1948.

DEUTSCH, J. A., & DKUISCII, D. Attention: Sometheoretical considerations. Psychological Review,1963, 70, 80-90.

DOCTOR, R. M., & ALTMAN, F. Worry and emotional-ity as components of lest anxiety: Replication andfurther data. Psychological Reports, 1969, 24,563-568.

DONNER, L., & GUERNEY, B. G., JR. Automated groupdesensitization for test anxiety. Behaviour Researchand Therapy, 1969, 7, 1-13.

DORIS, J., & SARASON, S. B, Test anxiety and blameassignment in a failure situation. Journal of Ab-normal and Social Psychology, 1955, 50, 335-338.

EASTKRHROOK, J . A. The measurement of persever-ance. Unpublished master's thesis, Queen's Univer-sity Library, Kingston, Ontario, 1953.

EASTERBROOK, J. A. The effect of emotion on cueutilization and the organization of behavior.Psychological Review, 1959, 66, 83-201.

ECIETII, H. Selective attention. Psychological Bulletin,1967, 1, 41-57.

EMERY, J. R., & Kiu MIIOI.TZ, J. D. Standard versusindividualized hierarchies in desensilization to re-duce lest anxiety. Journal of Counseling Psychol-ogy, 1967, 14, 204-209.

GANZER, V. J. Effects of audience presence and testanxiety on learning and retention in a serial learn-ing situation. Journal of Personality and SocialPsychology, 1968, 8, 194-199.

GAKLINGTON, W. K., & COTLER, S. B. Systematic de-scnsitizalion of test anxiety. Behaviour Researchand Therapy, 1968, 6, 247-256.

HAKT.ESTON, B. W. Test anxiety and performance inproblem-solving situations. Journal of Personality,1962, 30, 557-573.

IIILI, K. L., & GARLINGTON, W. K. A comparisonof group versus individual desensitization of testanxiety. Behaviour Research and Therapy, 1969,7, 207-209.

JAMES, W. Principles of psychology. Vol. 1. NewYork: Henry Holt, 1890.

ANXIETY AND ATTENTION 103

JANIS, I. L. Anxiety indices related to susceptibilityto persuasion. Journal oj Abnormal and SocialPsychology, 1955, 51, 663-667.

JOHNSON, E. E, The role of motivational strength inlatent learning. Journal of Comparative andPhysiological Psychology, 1952, 45, 526-530.

JOHNSON, S. M., & SECUREST, L. Comparison of de-sensitization and progressive relaxation in treatingtest anxiety. Journal of Consulting and ClinicalPsychology, 1968, 32, 280-286.

KATAHN, M., STRENCER, S., & CHERRY, N. Groupcounseling and behavior therapy with test-anxiouscollege students. Journal of Consulting Psychology,1966, 30, 544-549.

KOIIN, H. Effects of variations of intensity ofexperimentally induced stress situations upon cer-tain aspects of perception and performance. Jour-nal of Genetic Psychology, 1954, 85, 289-304.

KONDAS, O. Reduction of examination anxiety and'stage-fright' by group dcscnsitization and relaxa-tion. Behaviour Research and Therapy, 1967, 5,275-281.

LAXER, R. M., QUARTER, J., KOOMAN, A., & WALKER,K. Systematic desensitization and relaxation ofhigh-test-anxious secondary school students. Jour-nal of Counseling Psychology, 1969, 5, 446-451.

LIEBERT, R. M., & MORRIS, L. W. Cognitive andemotional components of test anxiety: A distinctionand some initial data. Psychological Reports, 1967,20, 975-978.

MANDLER, G.> & SARASON, S. B. A study of anxietyand learning. Journal of Abnormal and SocialPsychology, 1952, 47, 166-173.

HANDLER, G., & WATSON, D. L. Anxiety and the inter-ruption of behavior. In C. D. Spielbcrger (Ed.),Anxiety and behavior. New York: Academic Press,1966.

MARLETT, N. J., & WATSON, D. Test anxiety andimmediate or delayed feedback in a test-like avoid-ance 'task. Journal oj Personality and Social Psy-chology, 1968, 8, 200-203.

MEUNIER, C., & RULE, B. G. Anxiety, confidence andconformity. Journal of Personality, 1967, 35, 498-504.

MORRIS, L. W., & LIEBERT, R. M. Effects of anxietyon timed and unlimed intelligence tests. Journal ofConsulting and Clinical Psychology, 1969, 33,240-244.

MORRIS, L. W., & LIEBERT, R. M. Relationship ofcognitive and emotional components of test anx-iety to physiological arousal and academic perform-ance. Journal of Consulting and Clinical Psychol-ogy, 1970, 35, 332-337.

NEALE, J. M., & KATAHN, M. Anxiety choice andstimulus uncertainty. Journal oj Personality,1968, 36, 235-245.

PAUL, G. L., & EMKSEN, C. W. Effects of test anx-iety on "real-life" examinations. Journal oj Person-ality, 1964, 32, 480-494.

RUSSELL, D. G., & SARASON, I. G. Test anxiety, sex,and experimental conditions in relation to anagramsolution. Journal of Personality and Social Psy-chology, 1965, 1, 493-496.

SARASON, I. G. Interrelationships among individualdifference variables, behavior in psychotherapy andverbal conditioning. Journal of Abnormal andSocial Psychology, 1958, 56, 339-344. (a)

SARASON, I. G. The effects of anxiety, reassurance,and meaningfulncss of material to be learned inverbal learning. Journal oj Experimental Psychol-ogy, 1958, 56, 472-477. (b)

SARASON, I. G. Relationships of measures of anxietyand experimental instructions to word associationtest performance. Journal oj Abnormal and SocialPsychology, 1959, 59, 37-42.

SARASON, I. G. Empirical findings and theoreticalproblems in the use of anxiety scales. PsychologicalBulletin, 1960, 57, 403-415.

SARASON, I. G. The effects of anxiety and threat onsolution of a difficult task. Journal oj Abnormaland. Social Psychology, 1961, 62, 165-168.

SARASON, I. G. Verbal learning, modeling, and juveniledelinquency. American Psychologist, 1968, 4, 254-266.

SARASON, I. G., & GANZER, V. J. Anxiety, reinforce-ment, and experimental instructions in a freeverbalization situation. Journal oj Abnormal andSocial Psychology, 1962, 65, 300-307.

SARASON, I. G., & GANZKR, V. J. Effects of test anxietyand reinforcement history on verbal behavior.Journal oj Abnormal and Social Psychology, 1963,67, 513-519.

SARASON, I. G., & HARMATZ, M. G. Sex differencesand experimental conditions in serial learning.Journal of Personality and Social Psychology, 1965,1, 521-524.

SARASON, I. G., & KOENIG, K. P. The relationshipof test anxiety and hostility to description of selfand parents. Journal oj Personality and SocialPsychology, 1965, 2, 617-621.

SARASON, I. G., & MINARD, J. Test anxiety, experi-mental instructions and the Wechsler Adult Intel-ligence Scale. Journal of Educational Psychology,1962, 53, 299-302.

SARASON, I. G., & PALOI.A, E. G. The relationship oftest and general anxiety, difficulty of task, andexperimental instructions to performance. Journalof Experimental Psychology, 1960, 59, 185-191.

SARASON, I. G., PEDERSON, A. M., & NVMAN, 13. Testanxiety and the observation of models. Journal ojPersonality, 1968, 36, 493-511.

SARASON, S. B., DAVIDSON, K. S., LIGIITIIALL, F. F.,WAITK, R. R., & RUEBUSH, B. K. Anxiety in ele-mentary school children. New York: Wiley, I960.

SARASON, S. B., HANDLER, G., & CRAIGHILL, P. G.The effect of differential instructions on anxietyand learning. Journal of Abnormal and SocialPsychology, 1952, 47, 561-565.

SIKBER, J. E. A paradigm for experimental modifica-tion of the effects of test anxiety on cognitiveprocesses. American Educational Research Journal,1969, 6, 46-61.

SII/VERMAN, R. E. Anxiety and the mode of re-sponse. Journal of Abnormal and Social Psychol-ogy, 1954, 49, 538-542.

104 JER1 WINK

S I I . V K K M A N , R. I1'/., & BLITZ, B. reaming and twokinds of anxiety. Journal of Abnormal and SocialPsychology, 1956, 52, 301-303.

SNIDER, J. G., & GETTING, K. R. Autogcnic trainingand the treatment of examination anxiety in stu-dents. Journal of Clinical Psychology, 1966, 22,111-114.

SPIKGLER, M. D., MORRIS, L. W., & LIEBEET, R. M.Cognitive and emotional components of test anx-iety: Temporal factors. Psychological Reports,1968, 22, 451-456.

STANFORD, D., DE.MBER, W., & STANI'OHD, F. A chil-dren's form of the Alpert-Haber achievement anx-iety scale. Child Development,, 1963, 34, 1027-1032.

SUINN, R. M. The desensitization of test-anxiety bygroup and individual treatment. Behaviour Re-search and Therapy, 1968, 6, 385-387.

TAYLOR, J. A. A personality scale of manifest anxiety.Journal of Abnormal and Social Psychology, 1953,48, 285-290.

TK.CCE, J. J., & HAPP, S. Effects of shock-arousal ona card-sorting test of color-word interference.Perceptual and Motor Skills, 1964, 19, 905-906.

TRAIT, E. P., & KAUSLER, P. H. Test anxiety leveland goal-setting behavior. Journal of ConsultingPsychology, 1958, 22, 31-34.

TKEISMAN, A. M. Selective attention in man. BritishMedical Bulletin, 1964, 20, 12-16.

WACHTEL, P. L. Anxiety, cue utilization, and patternsof ego defense. Dissertation Abstracts, 1966, 27,2149.

WACHTEL, P. L. Anxiety, cue utilization, and patternsof ego defense. (Doctoral dissertation, Yale Uni-versity) Ann Arbor, Mich.: University Microfilms,1966. No. 66-5420.

WACHTEL, P. L. Anxiety, attention, and coping withthreat. Journal of Abnormal Psychology, 1968, 73,137-143.

WEST, C. K., LEE, J. F., & ANDERSON, T. H. Theinfluence of test anxiety on the selection of rele-vant from irrelevant information. The Journal ofEducational Research, 1969, 63, 51-52.

WINE, J. Investigations of an attenlional interpreta-tion of test anxiety. Unpublished doctoral disserta-tion, University of Waterloo, 1970.

WONDERLIC, E. F. Wonderlic personnel tests. North-field, Illinois: E. F. Wonderlic and Associates,1959.

ZATM?Y, D., & BRUNING, J. L. Drive and range of cueutilization. Journal of Experimental Psychology,1966, 71, 382-384.

(Received February 10, 1970)