achievement motivation: conceptions of ability, subjective...

Psychological Review 1984, Vol. 91, No. 3, 328-346 Copyright 1984 by the American Psychological Association, Inc. Achievement Motivation: Conceptions of Ability, Subjective Experience, Task Choice, and Performance John G. Nicholls Purdue University Achievement behavior is denned as behavior directed at developing or demonstrating high rather than low ability. It is shown that ability can be conceived in two ways. First, ability can be judged high or low with reference to the individual's own past performance or knowledge. In this context, gains in mastery indicate competence. Second, ability can be judged as capacity relative to that of others. In this context, a gain in mastery alone does not indicate high ability. To demonstrate high capacity, one must achieve more with equal effort or use less effort than do others for an equal performance. The conditions under which these different conceptions of ability function as individuals' goals and the nature of subjective experience in each case are specified. Different predictions of task choice and performance are derived and tested for each case. In this article, predictions of task choice, performance, and subjective experience in ex- perimental achievement settings are derived, and relevant evidence is examined. An inten- tional view of behavior (Dennett, 1978) is adopted. In this, action is construed as a ra- tional attempt to attain goals or incentives. In commonsense terms, individuals' actions serve to achieve purposes efficiently or economically. In the terms of games theory, action maximizes gains and minimizes losses. An action can fail to achieve a purpose but cannot be useless in the sense that it would be if it were not designed to serve a purpose. In other words, the term rational refers to the way goals are pursued. Only in a limited sense (e.g., p. 332 of this article), does it imply anything about the adaptive value of any goal. 1 To use this approach to predict behavior This article is based on a paper presented in a sym- posium, Attributional Approaches to Human Motivation (W-U. Meyer and B. Weiner, chairmen), that was held at the Center for Interdisciplinary Research, University of Bielefeld, West Germany, in July 1980. Preparation of the article was facilitated by National Science Foundation Grant BNS 7914252, University of Illinois subcontract. The author is grateful for discussions with Carol Dweck, Martin Maehr, and Donald Ferris, and comments on earlier versions by Charles Carver, Virginia Crandall, Heinz Heckhausen, Carolyn Jagacinski, Martin Maehr, Ernest McDaniel, Arden Miller, Klaus Schneider, Deborah Stipek, and anonymous reviewers. Requests for reprints should be sent to John G. Nicholls, Educational Psychology, SCC-G, Purdue University, West Lafayette, Indiana 47907. and thought, one must first specify individuals' goals and predict when a given goal will govern behavior. Here, achievement behavior is de- nned as that behavior in which the goal is to develop or demonstrate—to self or to others— high ability, or to avoid demonstrating low ability. This implies that in achievement sit- uations individuals desire success to the extent that it indicates high ability and seek to avoid failure to the extent that it indicates low ability (Kukla, 1978; McFarland& Ross, 1982). This is more precise than McClelland, Atkinson, Clark, and Lowell's (1953) definition of achievement motivation in terms of affect as- sociated with performance that is evaluated in terms of standards of excellence (p. 79). Standards of excellence apply to moral as well as to achievement behavior. V. J. Crandall, Katkovsky, and Preston (1960), Heckhausen (1967), Kukla (1972, 1978), and Maehr and Nicholls (1980) hold that the distinguishing feature of achievement behavior is that its goal is competence or perception of competence. The present theory adds the assumption that adolescents and adults can construe compe- tence or ability in at least two different ways. 1 Consideration of actions in terms of individuals' goals is the stance adopted by naive observers when they em- pathize with others (Hoffman, Mischel, & Mazze, 1981). This gives derivations within the intentional framework a commonsense or naive flavor and might account for the fact that, despite its long history (Bolles, 1974), the ap- proach may be considered heretical (Garcia, 1981, p. 151). 328

Upload: vokhanh

Post on 11-Feb-2019




0 download


Psychological Review1984, Vol. 91, No. 3, 328-346

Copyright 1984 by theAmerican Psychological Association, Inc.

Achievement Motivation: Conceptions of Ability, SubjectiveExperience, Task Choice, and Performance

John G. NichollsPurdue University

Achievement behavior is denned as behavior directed at developing or demonstratinghigh rather than low ability. It is shown that ability can be conceived in two ways.First, ability can be judged high or low with reference to the individual's own pastperformance or knowledge. In this context, gains in mastery indicate competence.Second, ability can be judged as capacity relative to that of others. In this context,a gain in mastery alone does not indicate high ability. To demonstrate high capacity,one must achieve more with equal effort or use less effort than do others for anequal performance. The conditions under which these different conceptions ofability function as individuals' goals and the nature of subjective experience ineach case are specified. Different predictions of task choice and performance arederived and tested for each case.

In this article, predictions of task choice,performance, and subjective experience in ex-perimental achievement settings are derived,and relevant evidence is examined. An inten-tional view of behavior (Dennett, 1978) isadopted. In this, action is construed as a ra-tional attempt to attain goals or incentives. Incommonsense terms, individuals' actions serveto achieve purposes efficiently or economically.In the terms of games theory, action maximizesgains and minimizes losses. An action can failto achieve a purpose but cannot be useless inthe sense that it would be if it were not designedto serve a purpose. In other words, the termrational refers to the way goals are pursued.Only in a limited sense (e.g., p. 332 of thisarticle), does it imply anything about theadaptive value of any goal.1

To use this approach to predict behavior

This article is based on a paper presented in a sym-posium, Attributional Approaches to Human Motivation(W-U. Meyer and B. Weiner, chairmen), that was held atthe Center for Interdisciplinary Research, University ofBielefeld, West Germany, in July 1980. Preparation of thearticle was facilitated by National Science FoundationGrant BNS 7914252, University of Illinois subcontract.

The author is grateful for discussions with Carol Dweck,Martin Maehr, and Donald Ferris, and comments on earlierversions by Charles Carver, Virginia Crandall, HeinzHeckhausen, Carolyn Jagacinski, Martin Maehr, ErnestMcDaniel, Arden Miller, Klaus Schneider, Deborah Stipek,and anonymous reviewers.

Requests for reprints should be sent to John G. Nicholls,Educational Psychology, SCC-G, Purdue University, WestLafayette, Indiana 47907.

and thought, one must first specify individuals'goals and predict when a given goal will governbehavior. Here, achievement behavior is de-nned as that behavior in which the goal is todevelop or demonstrate—to self or to others—high ability, or to avoid demonstrating lowability. This implies that in achievement sit-uations individuals desire success to the extentthat it indicates high ability and seek to avoidfailure to the extent that it indicates low ability(Kukla, 1978; McFarland& Ross, 1982). Thisis more precise than McClelland, Atkinson,Clark, and Lowell's (1953) definition ofachievement motivation in terms of affect as-sociated with performance that is evaluatedin terms of standards of excellence (p. 79).Standards of excellence apply to moral as wellas to achievement behavior. V. J. Crandall,Katkovsky, and Preston (1960), Heckhausen(1967), Kukla (1972, 1978), and Maehr andNicholls (1980) hold that the distinguishingfeature of achievement behavior is that its goalis competence or perception of competence.The present theory adds the assumption thatadolescents and adults can construe compe-tence or ability in at least two different ways.

1 Consideration of actions in terms of individuals' goalsis the stance adopted by naive observers when they em-pathize with others (Hoffman, Mischel, & Mazze, 1981).This gives derivations within the intentional framework acommonsense or naive flavor and might account for thefact that, despite its long history (Bolles, 1974), the ap-proach may be considered heretical (Garcia, 1981, p. 151).



Two forms of achievement goal are therebydistinguished. According to the intentionalframework, our subjective experience and overtbehavior should differ in predictable wayswhen we have different goals. The two con-ceptions of ability are, therefore, keys to thistheory.

Conceptions of AbilityThe two conceptions of ability have in com-

mon the notions that task mastery is improvedby effort or learning and that mastery is notnormally lost. In the first and less differentiatedconception, levels of ability and task difficultyare judged in relation to one's own perceivedmastery, understanding, or knowledge. Themore individuals feel they have learned, themore competent they feel. The more differ-entiated conception is embodied in standardintelligence testing practices (Nicholls, 1978).In this conception, learning is an insufficientbasis for perception of competence. Rather,task difficulty and ability are judged high orlow with reference to the ability of membersof a normative reference group. High abilitymeans above average and low ability meansbelow average. In addition, valid inferences ofability are presumed to require evidence thateffort is equal and optimum across individuals.This implies a conception of ability as capacitythat is not always revealed in performance.Only optimum effort reveals the present limitof one's capacity and this capacity limits theeffect of effort on performance.2

The two conceptions of ability embody dif-ferent criteria for judging one's ability orchances of demonstrating ability. In the lessdifferentiated conception, difficulty and abilityjudgments are self-referenced. Tasks are judgeddifficult if we expect to fail on them, and themore difficult they appear, the more does suc-cess indicate high ability. A greater gain inmastery or mastery of a task that one wasuncertain of being able to master indicatesgreater competence. Furthermore, becausemore effort is seen to lead to more learning(that indicates more ability), the higher theeffort needed for mastery, the higher the per-ceived ability. In the more differentiated con-ception of ability as capacity, task difficulty(normative difficulty) is judged from the per-formance of others, and demonstration of highability demands success on tasks where others

fail. Capacity is inferred by interpersonalcomparison of performance and effort. In thiscase, one could learn through effort or couldmaster a task that was personally very chal-lenging but still fail to demonstrate high ability.Indeed, the more effort or time one needs tolearn something (compared to the effort ortime it takes others) the less capacity is implied.

These conceptions of ability were first notedin research on the development of the conceptof ability (Nicholls, 1978, 1980; Nicholls &Miller, 1983). Young children conceive of abil-ity in a self-referenced manner as learningthrough effort. For them, to have low abilitymeans mere failure to master a task or to im-prove as much as one had hoped. After a num-ber of intermediate levels of differentiation,adolescents conceive of ability as capacity (notmerely performance) relative to that of others.In this case, perception of low ability moreclearly involves perception of inadequacy inthe self—lack of capacity that cannot readilybe altered. The motivational significance ofthese conceptions of ability is indicatedby concomitant developmental changes inachievement affect, task choice, and perfor-mance (Nicholls & Miller, in press). The con-cern here is with causes and consequences ofactivation of the different conceptions in ad-olescents and adults.

To judge our capacity we must compare theeffort and attainment of self and others. Inother words, we must adopt a relatively ex-ternal or social self-evaluative perspective. Forthis reason, the term ego involvement is appliedto states where individuals seek to demonstrateability in the differentiated sense. Use of theless differentiated conception involves a lessexplicitly self-evaluative stance. The concernhere is with improving one's mastery of tasksrather than with one's ability relative to thatof others. Accordingly, the term task involve-

2 Among researchers and, no doubt, among naive adults,there are diverse views on the relative importance of hered-ity and environment in determining the levels of individ-uals' available or present capacities. These diverse viewsare not of concern here. There is, however, agreement thata test score does not reveal one's present capacity if onedoes not apply optimum effort and that it takes muchmore than optimum effort during a test to improve one'scapacity. It is this understanding, which is implicit in in-telligence test administration practices, that is embodiedin the differentiated conception.


ment is applied to states where individuals seekto demonstrate ability in the less differentiatedsense. I do not thereby imply that feelings ofcompetence are absent in task involvement.Young children who lack the differentiatedconception gain strong feelings of competencefrom their accomplishments (Heckhausen, inpress); so do adults when they are task in-volved. However, task involvement requires aless social or external perspective on the self.Studies confirming that both conceptions candefine achievement goals for adults are notedafter the following two sections.

The foundations of the present theory arenow laid. The most general postulate is thatovert action and cognitive activity is rationalin the sense described. The second postulateis that individuals can have development ordemonstration of ability, in either sense, as agoal. We can, therefore, begin derivation ofpredictions—first, with the factors that engagethe more rather than the less differentiatedconception of ability.

Conditions of Usage of Each Conception

It has frequently been observed that humans(and monkeys) attempt to improve their levelof mastery (a) if they are presented skill tasksoffering a moderate challenge and are notplaced under physiological or psychologicalstress and (b) if task-extrinsic incentives arenot salient (Csikszentmihalyi, 1977; Deci,1975; Elkind, 1971; Harlow, 1950; White,1959). That is, these conditions produce taskinvolvement. Under such relatively neutralconditions, the less differentiated conceptionof ability is a goal of action, and perceptionof improved mastery through effort indicatescompetence. The intentional framework in-dicates that individuals do not make complexjudgments if simple ones serve their purposes.In task involvement, the differentiated con-ception is superfluous. We do not need tocompare our own and others' attainment andeffort to establish whether we have gained inmastery. The differentiated conception shouldnot, therefore, be employed. But, as when weseek to assess a person's intelligence, when ourgoal is to establish whether mastery reflectsability rather than task ease or effort, we mustemploy the differentiated conception. Only thedifferentiated conception permits a concep-

tually adequate evaluation of the extent towhich mastery reflects ability as opposed toeffort or task difficulty.

It follows that, when one faces skill tasks,the basis of performance evaluation shifts fromthe less to the more differentiated conceptionas the strength of factors that induce concernsabout evaluation of our competence increases.(I do not consider situations involving explicitbehavioral standards such as a specific levelof achievement or situations where altruisticor other task-extrinsic goals are salient.) An-nouncements that skill tasks are being usedto test subjects should induce concerns aboutpersonal competence, especially if importantor valued skills such as intelligence are at issue.Interpersonal competition on skill tasks alsoposes the question, who is best? Competition—especially on valued tasks—is, therefore, pre-dicted to increase use of the differentiatedconception to evaluate oneself. Finally, themore differentiated conception involves amore public perspective on the self. Thus, inskill situations, manipulations such as use ofan audience (Carver & Scheier, 1981) that in-crease public self-awareness are likely to in-crease self-evaluation in terms of the differ-entiated conception. Tests of these predictionsare considered after the following section.

Subjective ExperienceUse of the differentiated conception involves

a more actively self-evaluative or social viewof the self. It is also predicted that learning ormastery through effort is experienced more asan end in itself as task involvement increasesand more as a means to an end as ego in-volvement increases. This follows from con-sideration of the processes necessary to inferability in each sense.

To assess their chances of demonstratingability, ego-involved individuals must assesswhat they can master and whether this implieshigher capacity than that of others. They mustjudge whether mastery will serve their end.They might, for example, believe they can im-prove through effort but expect such attemptsto imply low capacity more clearly than in-activity would. Mastery, therefore, tends to beexperienced as a means to an end rather thanan end in itself: Action is more exogenouslyattributed (Kruglanski, 1975; see also Deci &Ryan, 1980).


In task involvement, however, improvementis the goal. An increase in mastery is, therefore,an end in itself. In task involvement, percep-tion of an opportunity to develop or exercisean increased level of skill, therefore, reliablyoccasions attempts to do so. Furthermore, aslearning is an end in itself, when individualsfeel they are mastering a task, they feel theyare doing what they want to do. They feelmore intrinsically motivated than when theyare ego involved.3

Conceptions of Ability and SubjectiveStates: Evidence

First, it is claimed that either conception ofability can be employed by adults to evaluatetheir performance in skill situations. Second,it is predicted that the differentiated concep-tion is activated as a goal when, instead ofbeing presented in a neutral fashion, (a) tasksare presented as tests of valued skills, (b) in-terpersonal competition is fostered, and (c) self-awareness is induced. Ego-involving conditionsare also predicted to increase exogenous at-tribution and diminish interest in mastery.

Diener and Srull (1979) used television andvoice recordings—a method that appears toinduce public self-awareness (Carver & Scheier,1981, chap. 16). This manipulation producedself-reinforcement when peer norms were sur-passed. When self-awareness was not induced,however, subjects self-reinforced on the basisof their own previous performance (indicatingself-evaluation in terms of the less differen-tiated conception). When self-awareness wasmanipulated and when individual variation inpublic self-awareness was assessed, Scheier andCarver (1983) also found self-awareness as-sociated with more attention to social com-parison norms. These results support the hy-potheses that increases in public self-awarenessincrease use of the differentiated conceptionto evaluate one's competence.

Jagacinski and Nicholls (in press) asked stu-dents to anticipate feelings of competence onpassing a language test after high versus loweffort. All of the students expected higher effortto lead to greater gains in competence. How-ever, in a competitive (ego-involving) situation,students anticipated feeling less able when ef-fort was high. This indicates use of the dif-ferentiated conception where the "fact" that

effort improves mastery conflicts with the"fact" that higher effort implies lower capacity.No such conflict was apparent when a learningfor learning's sake situation was simulated.These findings (replicated four times) supportthe prediction that competitive rather thanlearning-oriented conditions engage the dif-ferentiated conception. Furthermore, prideand a sense of accomplishment were higherwhen effort was higher in task involvementand lower when effort was higher in ego in-volvement.

Presentation of tasks as tests of intelligenceshould induce ego involvement. Test-anxietyresearchers have often compared this manip-ulation with neutral conditions and found itto increase concerns to meet the performancenorms (integral to the differentiated concep-tion) implied in the manipulation (I. G. Sar-ason, 1975; Wine, 1971). (This research isconsidered in the section on performance.)Patten and White (1977) found that presentinga task as an intelligence test (ego involving)and asking students to make causal attributionsfor their performance had the same effects onperformance. (Performance differed in a neu-tral condition.) When attributing one's out-comes to ability, effort, or difficulty, the dif-ferentiated conception of ability must be ac-tivated if one is to respond meaningfully. Thus,Patten and White's finding is also consistentwith the hypothesis that presentation of tasksas tests of a valued ability is more likely thanis a neutral presentation to activate the dif-ferentiated conception.

Comparisons of attributions and affect incompetitive (ego-involving) and noncompet-itive (task-involving) conditions also indicatemore self-evaluation in terms of the less dif-ferentiated conception (where what is accom-plished through effort is the basis of perceivedcompetence) in noncompetitive conditions.More effort attribution (C. Ames & R. Ames,

3 The exogenous quality of ego-involved action can bedistinguished from the examples of exogenously attributedskill behavior most commonly found in the literature butnot discussed here. For example, surveillance or offers ofprizes can transform endogenous to exogenous involvement(Lepper & Greene, 1975). In ego involvement, the end isto demonstrate superior capacity, whereas in the aboveexamples, demonstration of skill is a means to nona-chievement incentives.


1981) and a stronger positive association be-tween perceived effort and satisfaction (C.Ames, R. Ames, & Felker, 1977) emerged innoncompetitive conditions. Competitive con-ditions produced a positive association be-tween own perceived ability (not effort) andsatisfaction with performance (C. Ames et al.,1977). In the above studies, experimenterselicited performance attributions. When sub-jects could report either attributions or self-instructions (which indicate a focus on mas-tery), a complementary pattern emerged (C.Ames & McKelvie, 1983). Competition pro-duced more ability and difficulty attributionsthan did an individual goal condition, whichproduced more effort attributions and self-in-structions. Thus there appears to be more useof the less differentiated conception in non-competitive conditions and more evaluationof capacity in competitive conditions.

Csikszentmihalyi (1977) has described thesubjective states of individuals when they arehighly involved in tasks as marked by feelingsof competence and reduced public self-aware-ness. As the concept of task involvement im-plies, these states required tasks offering achance to improve or perfect one's mastery.Activities that induced these states were seenas satisfying for endogenous rather than ex-ogenous reasons (p. 17). This was less markedfor activities that produced less task involve-ment. It supports the proposed link betweentask-involvement and endogenous attributionfor action.

Similarly, Deci, Betley, Kahle, Abrams, andPorac (1981) found more interest in puzzlesafter individual performance (task involve-ment) than after competitive success. Ryan(1982) found more interest in tasks after per-formance in a neutral condition than after anintelligence test condition. Diener and Srull's(1979) induction of self-awareness also di-minished students' feelings of freedom aboutdispensing self-reinforcement. Ego involve-ment appears to increase feelings of constraintabout self-evaluation as well as for achievementbehavior itself. Evidence of a different typealso suggests the link between use of socialnorms and exogenous attribution. Teacher re-ports of use of social comparison to controlchildren were more highly correlated with re-ports of use of coercive (exogenous) methodsthan with use of noncoercive methods (Deci,

Schwartz, Sheinman, & Ryan, 1981). Fur-thermore, children whose teachers used coer-cion and social comparison (not separated inanalyses) reported less intrinsic interest inschool.

There is, then, support for the claim thatboth conceptions of ability can be achievementgoals and for the predictions of the circum-stances under which each goal is activated andof subjective experience in each case. Thesephenomena are of interest in their own rightand are necessary for the derivation and testingof predictions of task choice and performance.

Task Choice

It is assumed that, in achievement situa-tions, each individual's purpose is to dem-onstrate high ability and to avoid demonstrat-ing low ability. According to the intentionalframework, individuals should select thosetasks they expect to enable them to maximizetheir chances of demonstrating high ability andavoiding demonstrating low ability.

Individuals can be committed to goals theydo not expect to attain (Klinger, 1975;McFarlin & Blascovich, 1981). But, ultimately,the rational response to an unattainable goalis rejection of that goal and selection of thenext most attractive goal (Klinger, 1975). Thus,as individuals become more certain they can-not demonstrate high ability they should tendto adopt the less attractive goal of avoidingdemonstrating low ability. This goal should,in turn, be relinquished as hope of attainingit dies.

To ascertain the most economical action fordemonstrating ability, individuals must formsubjective probabilities of demonstrating highversus low ability on the available tasks. Theintentional framework indicates that they em-ploy only the relevant data and process theseno more than necessary for this purpose. Be-cause a less differentiated conception of abilityis employed in task involvement, the processof forming expectations of demonstratingability is less complex in this state.

Task Involvement

When individuals are task involved, theysee more effort as leading to more masteryand higher ability. Tasks seen as demandingno effort for success and tasks expected to not


yield to maximum effort offer no chance ofdemonstrating high ability. Tasks that appearlikely to yield to high effort offer the optimumbalance between high chances of demonstrat-ing little ability and low chances of demon-strating high ability (which would involvewaste of much effort). That is, tasks shouldbe most attractive at an intermediate level ofexpectancy of success where one's highestlikely level of competence might be demon-strated. Individuals should differ in the levelof objective difficulty at which they have mod-erate expectancies of success.4 They shoulddiffer correspondingly in the level of objectivedifficulty they prefer; however, all should prefertasks close to their own perceived level of com-petence.5

Ego Involvement

When individuals are ego involved, theirchances of demonstrating ability depend onthe ability of others. Thus, one might dem-onstrate incompetence but not competence onnormatively easy tasks (on which many cansucceed). One might demonstrate ability butnot incompetence on normatively difficulttasks. Moderate normative difficulty levels al-low the possibility of above- or below-averageperformance that would indicate high or lowcapacity.

Individuals with high perceived ability havemoderate expectancies of success on norma-tively moderate to difficult tasks where successindicates high ability. Therefore, they shouldprefer tasks at or above moderate difficultylevels, depending on how able they believe theyare. They do not expect to decline in the levelthey can attain. Therefore, when repeated suc-cess at a given level indicates possession of thecorresponding level of ability, they can gainbut not lose in perceived ability by attemptingmore difficult tasks. Thus they should prefernormatively moderate to difficult tasks wheretheir expectations of success are moderate.

The picture is more complex for those withlow perceived ability. With tasks of moderatenormative difficulty they expect to fail and,thereby, to demonstrate low ability. Theyshould, therefore, avoid such tasks. Choice ofeither very easy or very difficult tasks enablesthem to avoid demonstrating low ability. Theseindividuals' preferences for easy versus difficult

tasks are predicted to depend on how certainthey are that they lack ability and on theirassociated level of commitment to demon-strating high ability or to avoiding demon-stration of low ability. Consider first thosewhose doubts about their competence are notfirm enough to have extinguished commitmentto demonstrating high ability. For these in-dividuals, choice of easy tasks would be ir-rational in that it could not conceivably leadto demonstration of high ability. Only nor-matively difficult tasks offer any chance of at-taining the more attractive goal of demon-strating high ability as well as the certainty ofavoiding the demonstration of low ability.(McFarlin & Blascovich, 1981, show thatcommitment to achievement goals can existdespite very low expectations of attainingthem.) Thus, those who suspect their abilityis low but are committed to demonstratinghigh ability should prefer normatively difficulttasks. Here, their expectation of success is verylow but failure cannot imply low ability andthe possibility that they have high ability can-not be ruled out.

Repeated failure or other experiences may,however, have produced virtual certainty thatone lacks high ability. The intentional frame-work leads to the prediction that, for individ-uals in this second category, commitment todemonstrating high ability is low and the goalof avoiding the demonstration of low ability

4 Objective difficulty refers to the objective propertiesof tasks, such as number of pieces in a jigsaw puzzle ordistance from a target. This is distinguished from normativedifficulty that is inferred from the performance of others(Nicholls & Miller, 1983).

5 The less differentiated conception of ability, being self-referenced, does not allow the possibility of demonstratinglow ability in the decisive fashion that is possible whenability is judged relative to that of others. Nor does lowability in this sense clearly indicate the personal inadequacyassociated with a lack of capacity. Perception of low abilityin the less differentiated sense can occur in environmentsthat offer no opportunity to improve one's level of mastery.Consider, for example, the effect of an extremely unre-sponsive environment on infants, who do not have thedifferentiated conception of ability. They appear to almostnever perceive an opportunity to develop or demonstratecompetence and rarely attempt to do so (Hethcrington &Parke, 1975). In the framework of the less differentiatedconception, this represents something close to absolutezero of perceived ability. This could be termed a state oflearned helplessness—perceived noncontingency of actionand outcome.


is more salient. Withdrawal from the situationwould achieve this. If the situation mandatesa choice, the rational choice is a normativelyeasy task that demands little effort and wheresuccess indicates that one does not lack thelow level of ability that failure would indicate.Thus, those who are more certain that theirability is not high are predicted to prefer nor-matively easy tasks where their expectationsof success are high. Those in the third andmost extreme category—those who are certaintheir ability is low and who are not committedto avoiding demonstration of low ability—should see very easy tasks as offering the mosteconomical way of leaving the situation.

In summary, three types of individuals withlow perceived ability are distinguished. Someare committed to demonstrating high ability.Others, more certain that their ability is nothigh, lack commitment to demonstrating highability. Others are certain their ability is lowand have accepted this. The first type shouldtend to select tasks where they have very lowexpectancies of success. The second and thirdtypes should tend to prefer tasks where theyhave high expectancies of success.6

Atkinson's (1957, 1965) theory predictsgreater preference for moderate probabilitiesof success in those with high resultant achieve-ment motivation and preference for high orlow probabilities in those with low resultantmotivation. These parallel present predictionsfor high and low perceived ability in ego in-volvement. However, Atkinson's theory doesnot distinguish preference for high versus lowprobabilities of success. Nor does it predict ageneral preference for challenging tasks in task-involving situations. Others hold that Atkinsonis wrong: Kukla (1978), and Meyer, Folkes,and Weiner (1976) predict greater preferencefor intermediate probabilities in all subjects.The present position is that Atkinson is correctthough incomplete in ego involvement andthat Kukla and Meyer et al. are correct in taskinvolvement.

Measurement of Perceived AbilityHypothesis testing does not demand esti-

mates of perceived ability in the undifferen-tiated sense beyond expectations of success.An index of perceived ability relevant to egoinvolvement is, however, needed. This shouldindicate individuals' evaluations of their ability

relative to that of others and predict expec-tations of success or perceived ability on ex-perimental tasks. Self-esteem or self-conceptscales serve this purpose effectively. Such scalesrefer primarily to the adequacy of one's com-petence (R. Crandall, 1973), and authors (e.g.,Brockner, 1979; McFarlin & Blascovich, 1981;Shrauger, 1975) commonly employ and de-scribe them as perceived-competence mea-sures. The fact that they predict perceivedability on experimental tasks (McFarlin &Blascovich, 1981) supports their validity forthe present purpose. Jagacinski and Nicholls(in press) also found that college students in-terpreted the statement that someone is ableas meaning that they have high capacity rel-ative to others. Conversely, a statement thatsomeone succeeds through high effort was notinterpreted as implying that they have highability. This indicates that typical self-esteemitems would elicit evaluations of ability in thedifferentiated sense. Many questions aboutpersonal competence would heighten a self-evaluative stance and amplify this trend. Anassociation between a self-concept scale andratings of own skill relative to that of others(Roberts, Kleiber, & Duda, 1981) also supportsthis thesis.

6 Predictions are stated in terms of normative difficultyand expectancies of success for ego involvement and interms of expectancy of success for task involvement. Evenif normative cues are absent, ego-involved individuals ap-pear able to estimate their chances of demonstrating ability.Individuals with either high perceived attainment, generalexpectancies, or perceived ability are more inclined thanare others to attribute success but not failure to abilityon tasks of unspecified normative difficulty (C. Ames, 1978;Feather & Simon, 1971; Nicholls, 1976, 1979a; Simon &Feather, 1973; Valle & Frieze, 1976). It seems that whenthey have moderate expectancies of success, individualswith high perceived ability see tasks as moderately nor-matively difficult and infer high ability from success. Whenindividuals with low perceived ability have moderate ex-pectancies of success they appear to see tasks as easy anddo not infer high ability from success. Predictions can,therefore, be tested with reference to expectancies of suc-cess. This is important because studies of task choice oftendo not employ normative cues.

'According to Atkinson's (1957) theory, subjectiveprobabilities of success and task difficulty are equivalent.Except in young children (Nicholls & Miller, 1983) andin task involvement, where the less differentiated concep-tion of ability is employed, they are not. This means thatAtkinson's predictions (framed in terms of probabilitiesof success) are ambiguous. In this instance I assume thathis predictions refer to subjective probability of success,not to task difficulty.


Test-anxiety scales correlate quite highlywith self-concept or self-esteem scales (R.Crandall, 1973) and thus lack discriminantvalidity. Although one would not guess it fromtest-anxiety-scale titles, the content of thesescales refers directly to perceived ability andexpected adequacy of performance. These factsalone provide a case for considering themmeasures of (low) perceived ability. But thereis more. In the case of the Test Anxiety Scalefor Children, it has been shown that perceived-competence content accounts for the associ-ations of the scale with performance and othervariables that have been presumed to be aconsequence of anxiety (Nicholls, 1976). Lie-bert and Morris (1967) separated test-anxiety-scale content into worry and emotionalitycomponents. A number of studies show thatworry rather than emotionality is related toperformance and that worry but not emo-tionality scores are influenced by performancefeedback in exactly the ways one would expectperceived ability to be affected (Morris, Davis,& Hutchings, 1981; Wine, 1971). Morris etal. characterized the tendency to worry as atendency toward negative self-evaluation andnegative expectation. Indeed, worry items areaccurately described as low perceived com-petence rather than worry items (e.g., "I donot feel confident about my performance onthis test", "I do not feel self-confident"; Morriset al., 1981). Thus, the active ingredient oftest-anxiety scales is perceived competence.The validity of these scales for the present pur-pose is further indicated by evidence that test-anxious students perceive themselves as lessable on experimental tasks (Arkin, Detchon,& Maruyama, 1982) and as encountering moreproblems demonstrating ability on exams (Ar-kin, Kolditz, & Kolditz, 1983).

Resultant achievement-motivation (Atkin-son & Feather, 1966) scores incorporate test-anxiety scores. For this reason and in accordwith the evidence and arguments of Kukla(1972, 1978), individuals high in need forachievement and low in test anxiety (high re-sultant achievement motivation) can be con-sidered high in perceived ability, and those lowin need for achievement and high in test anx-iety (low resultant motivation) can be consid-ered low in perceived ability. Evidence thathigher resultant motivation is associated withhigher expectancies of success on experimental

tasks (Atkinson, 1957, 1969) supports thisview.8 Klinger and McNelly's (1969) claim thatthis motive measure indicates perceived statusis also consistent with this thesis.9

Although they lack discriminant validity,the above measures are obviously not identicaland the unique components of each meritmore study (Nicholls, 1976). Yet there is jus-tification for considering them measures ofperceived ability. Confirmation of present pre-dictions with these different measures supportstheir construct validity as indexes of perceivedability and indicates that a theoretical inte-gration has been achieved.

Task Choice: Evidence

In task involvement, preference for inter-mediate subjective probabilities is predictedfor all individuals. In ego involvement, thosewith high perceived ability are predicted toprefer moderate subjective probabilities andthose with low perceived ability are predictedto prefer more extreme probabilities. I considerevidence on the general contrast between taskand ego involvement and then the extra pre-dictions for low-perceived-ability individualswhen they are ego involved.10

8 Although consistent with the present position, this ev-idence contradicts a fundamental assumption of Atkinson'smathematical model, namely, that task difficulty and sub-jective probability of success can be equated. It also threat-ens the implication of his model that the differences be-tween low- and high-resultant-motive individuals reflectindividual differences in the values associated with successand failure rather than individual differences in expec-tations of success and failure. Alternatively, this evidencecould be seen as indicating that, in terms of Atkinson'stheory, his measure of resultant motivation lacks constructvalidity; it produces results contrary to the assumptionsof the theory.

9 Mehrabian's (1969) resultant achievement-motivationscale has been used extensively. Many of its items elicitpreferences for challenging skill tasks versus easy or nonskilltasks. Evidence that higher scorers on this scale choosemore challenging tasks might, therefore, merely show thatpeople do what they say they do. Such evidence is of ques-tionable relevance to achievement theory (Nicholls, Licht,& Pearl, 1982). Therefore, studies of task choice using thisscale are not considered.

10 In the method now commonly employed to estimatesubjective probabilities of success, individuals who prefertasks on which they have low or high success rates areassumed to prefer low or high probabilities. By anchoringestimates of subjective probabilities of success to each in-dividual's rate of success, this method takes account of


In an ego-involving condition, Raynor andSmith (1966) presented puzzles as an intelli-gence measure and emphasized the test's im-portance. In another condition, the experi-menter acted in a nonevaluative manner andminimized the importance of the task. It ap-peared that subjects experienced the conditionsas less different than was intended (Raynor &Smith, p. 187). Nevertheless, as predicted here,there was a strong tendency for low- more thanhigh-resultant-achievement-motive students toselect extreme probabilities in the ego-involv-ing condition and a greater general preferencefor moderate probabilities in the neutral con-dition.

A similar test arose out of work by Schneider(1973, chap. 4), who, in a series of studiesunder neutral conditions, found no differencebetween individuals high and low in resultantmotivation. However, in one study (chap. 5),low-motive subjects preferred more extremeprobabilities. In this study the experimenterwas a teacher in the subjects' school ratherthan an unfamiliar nonevaluative adult. Thisled to a comparison of a nonevaluative pre-sentation with'presentation of the task as avalid test of ability. Consistent with presentpredictions, more extreme probability pref-erences on the part of low- than high-resultant-motive students occurred only the more ego-involving condition (Jopt, 1974, p. 196).

Studies without direct comparisons of twoconditions provide less adequate tests of thetheory. Nevertheless, such studies could dis-confirm it. Seven studies of choice under (ap-parently) ego-involving conditions were found.DeCharms and Dave (1965), Roberts (1974),and Hamilton (1974) used physical skill taskswith males. The very high value males placeon physical skills (Roberts, in press) and the

individual differences in subjective probabilities. Deviationsof individual preferences from group (rather than indi-vidual) performance means have also been employed (e.g.,McClelland, 1958). This method appears to reflect theassumption (Atkinson, 1957) that "degree of difficulty canbe inferred from the subjective probability of success" (p.362). Because this method fails to deal with individualdifferences in expectation of success (Heckhausen, 1968),studies employing it are avoided in this review. Note thatstudies that use individual rather than group data to testAtkinson's theory (e.g., Hamilton, 1974; Moulton, 1965)embody a rejection of Atkinson's assumption that difficultyand subjective probability are equivalent.

making of a visible public record of perfor-mance outcomes in these studies should haveinduced ego involvement. Use of academicmaterial (DeCharms & Carpenter 1968) andanagrams (Moulton, 1965) presented as im-portant tests of ability would induce ego in-volvement. Brody (1963) presented under-graduates with what was said to be a test oftheir ability, on which they recorded theirnames. Both actions would induce ego in-volvement. Finally, Mahone (1960) studiedoccupational choices that are likely to occurin ego involvement. As predicted here, in theseseven studies where there was evidence of fac-tors that would induce ego involvement, low-resultant-motive individuals preferred moreextreme probability levels than did high-mo-tive subjects.

Trope (1979) employed a nonevaluative sit-uation in which the task was said to be un-related to intelligence and anonymity was as-sured. Thus it is probable that task involve-ment was maintained. Regardless of perceivedability, all of the subjects in this study preferredtasks that discriminated between levels ofability closest to their own perceived level.With similar conditions, Buckert, Meyer, andSchmalt (1979) obtained the same results.Thus, the available results are consistent withpredictions.

The remaining question is whether highversus low subjective probability of successchoices occur as predicted. When ego involve-ment is induced, low-perceived-ability indi-viduals who retain commitment to demon-strating high ability should choose low prob-abilities. Those who are more certain theirability is low should choose high probabilities.Two studies with relevant data were found.

Sears (1940, 1941) gave 9- to 12-year-oldsacademic tasks in a testlike ego-involvingmanner. Children who chose higher and lowerprobabilities had failed consistently in school.Thus, they would have lower perceived ability(Bloom, 1976; Nicholls, 1979a). Within thisgroup, those selecting very low probabilitiesof success showed a stronger wish for highachievement in diverse activities and mademore negative evaluations of their competencein these activities than did others. They acted"as if they never felt they were doing wellenough" (Sears, 1940, p. 523). This indicatesthe predicted commitment to demonstration


of high ability despite perception of low ability.Those selecting high probabilities showedgreater responsiveness to nonachievement in-centives and greater readiness to lower theirgoals after manipulated success (Sears, 1940,1941). Both phenomena indicate the predictedrejection of the goal of demonstrating highability.

Further support comes from Moulton's(1965) study of high school students' choiceson tasks presented as valid tests of their in-tellectual ability. Students low and interme-diate in resultant motivation chose more ex-treme subjective probability levels than didhigh-motive students. Furthermore, the lowestgroup chose high probabilities more than theintermediate group, who favored low proba-bilities. Given that the lowest group was mostcertain that they lacked ability, these resultsalso support the predictions.

In summary, the present theory resolves thecontradiction between the predictions of At-kinson and those of Kukla and Meyer et al.It successfully predicts choices in task versusego involvement. Unlike the above positions,it also predicts high- versus low-probabilitychoices in ego involvement. (See also Heck-hausen, 1977.)

Performance or AttainmentSome theories (e.g., Kukla, 1972) hold that

level of performance increases with intensityof effort. The present position adds the as-sumption that performance can be impairedby self-derogatory ability evaluations evenwhen effort is high (Arkin et al., 1982; I. G.Sarason, 1975; Wine, 1971). Consistent withthe intentional framework, it is assumed thateffort and self-evaluations are a function ofexpectations that effort leads to demonstrationof high rather than low ability. The derivationof performance predictions therefore dependson the previous statements of expectations ofdemonstrating ability as a function of task dif-ficulty. Ego involvement is considered first andthen is contrasted with task involvement.

Performance in Ego InvolvementPredictions. According to the intentional

framework, effort should be high (and produceeffective performance or attainment) when itis perceived that high effort is necessary fordemonstration of high ability in the differ-

entiated sense. For those with high perceivedability, this is on normatively moderate to dif-ficult tasks. For these individuals, effort andthus performance is predicted to be low ontasks that appear to demand little effort andwhere failure appears certain despite maxi-mum effort—tasks perceived as normativelyeasy and extremely difficult, respectively.

In ego involvement, individuals with lowperceived ability expect to demonstrate lowability in the moderate normative difficultyrange. Performance on such tasks should beimpaired, but the mechanism involved de-pends on how certain they are that they lackability (see Carver & Scheier, 1981, for ananalysis in terms of increasing expectancies offailure.) First, consider those whose perceivedability is low, but not low enough to have ex-tinguished commitment to demonstrating highability. This commitment will maintain effort.However, the aversive expectation of demon-strating lack of personal capacity should pro-duce the self-derogation, negative affect, andimpaired performance associated with testanxiety (Arkin et al., 1982; I. G. Sarason, 1975;Wine, 1971).

As predicted in the task-choice section, in-dividuals whose expectations of demonstratinghigh ability are extremely low should have lesscommitment to this goal and should be morecommitted to avoiding demonstration of lowability. The fact that, in ego involvement, fail-ure implies low ability less decisively wheneffort is low is then relevant. Thus, at moderatenormative-difficulty levels, the probability ofa self-protective reduction of effort (Frankel& Snyder, 1978) should be higher for theseindividuals than for others. A more extremeresult is predicted for individuals who are socertain that their ability is low that they haverelinquished commitment to avoiding dem-onstration of low ability. Such individualsshould avoid becoming ego involved with thetasks in question. In this case, effort is em-ployed only to the degree that other incentivesappear contingent on it and even perceptionof the possibility of demonstrating high abilityshould not produce high effort. The termlearned helplessness might be used here, al-though in practice it has been applied to anyperformance impairment consequent on per-ceived noncontingency of action and outcome(Abramson, Seligman, & Teasdale, 1978).


Evidence of the first (Arkin et al., 1982;Brockner, 1979) and last effects (Marecek &Mettee, 1972) exists. Many studies do not dis-tinguish the different levels of low perceivedability or the effects predicted to be associatedwith them. Nevertheless, because performanceimpairment is predicted for all levels of lowperceived ability (on normatively moderatelydifficult tasks), such studies can be used to testthe more general predictions of performance.Furthermore, it is likely that few individualswith the lowest level of perceived ability wouldenter or survive college long enough to findtheir way into subject pools. Thus, it is as-sumed that they play a negligible role in effectsobtained with students simply identified as lowin perceived ability.

When low-perceived-ability individuals facenormatively easy or difficult tasks, they expectto avoid demonstrating low ability, and self-derogatory evaluations do not occur. Effort ishigh on easy tasks where failure—which wouldindicate low ability—appears avoidable byhigh effort. This effort and, thus, performanceis higher for low- than for high-perceived-abil-ity individuals who expect an easier successon easy tasks. If the task is seen as normativelydifficult, high effort will (except at the lowestlevel of perceived ability) be forthcoming be-cause its absence would be inconsistent withcommitment to demonstration of high ability.Accordingly, performance of low-perceived-ability individuals is predicted to be higherwhen tasks are perceived as normatively easyor difficult rather than moderate.

Expectations of demonstrating ability andperformance: Evidence. According to thepreceding predictions, expectations of failureimpair performance to the extent that failureindicates low ability in the differentiated sense.Part of this thesis is supported by Frankel andSnyder (1978). They contrived a series of fail-ures calculated to make subjects doubt theirability and then presented tasks perceived asnormatively difficult or moderate. Perfor-mance was not impaired in the difficult con-dition where expectancies of failure and per-ceived noncontingency would be higher. Aspredicted, performance was only impaired inthe moderately difficult condition where failurewould indicate low ability. Miller (1982) rep-licated these findings (in an evaluative situa-tion) with sixth graders who had mastered the

conception of ability as capacity (Nicholls,1978). On the other hand, sixth graders whohad not yet mastered this conception per-formed similarly in moderate, high-difficulty,and control conditions. The two groups hadsimilar levels of perceived ability, but expec-tations of demonstrating low ability only im-paired the performance of those who had at-tained the differentiated conception. Thesefindings support the proposed role of expec-tations of demonstrating low ability—as op-posed to mere expectancies of failure (Carver& Scheier, 1981) or perceived noncontingency(Abramson et al., 1978)—and, more specifi-cally, of the conception of ability as capacity.

Task difficulty, perceived ability, and per-formance: Evidence. Returning to individualdifferences under ego involvement, it is pre-dicted that high-perceived-ability individualsperform their worst on tasks perceived as nor-matively easy, whereas low-perceived-abilityindividuals perform their worst at moderate-difficulty levels. Furthermore, low-perceived-ability individuals perform better than thosewith high perceived ability on tasks perceivedas normatively easy, whereas this is reversedat intermediate-difficulty levels. These predic-tions resemble those of Revelle & Michaels's(1976) revision of Atkinson's theory."

The following studies are relevant to theabove hypotheses in that normative difficultycues were manipulated and ego-involving ma-nipulations, such as presentation of tasks astests of intelligence, were employed. Kara-benick and Youssef (1968) found that low-resultant-achievement-motive students per-formed most poorly and below high-motivestudents at an intermediate normative diffi-culty level. High-motive students performedworst at an easy level, but this performancewas not, as predicted, below that of low-motivestudents. Comparing conditions of easy andintermediate normative difficulty, Kukla(1974) found all relevant predicted effects.12

'' Like the original theory, this revision holds that taskdifficulty and subjective probability of success can beequated. Because they are not equivalent, the predictionsof the theory are ambiguous. In this case, I assume thatRevelle and Michaels's predictions refer to normative taskdifficulty—the factor manipulated in relevant studies (c.f.Footnotes 7 and 10).

12 Kukla employed the Mehrabian (1969) ResultantAchievement Motivation Scale. High scores on this scale


I. G. Sarason (1961) found test-anxious(low-perceived-ability) students' anagram per-formance lower at moderate to moderately lowdifficulty than at high-difficulty levels. Low-anxious students performed better with mod-erate- than with high-difficulty instructions.These results accord with predictions for theanxious, but not the low-anxious students forwhom similar performance is expected in eachcase. However, the task was presented as anintelligence measure only in the moderate-dif-ficulty condition. This could account for low-anxious students' higher performance in themoderate- than the high-difficulty condition.S. B. Sarason, Mandler, and Craighill (1952)found no difference between high- and low-anxious students in a normatively difficultcondition, whereas low-anxious students outperformed those high in anxiety in an easycondition. If the easy task was seen as mod-erately easy, these results accord with predic-tions. Description of a task as highly difficultproduced higher performance in test-anxiousstudents than did a neutral condition wheredifficulty was presumably seen as moderate(I. G. Sarason, 1958). As predicted, low-anx-ious students' performance was best in the lat-ter condition. Though not perfectly consistent,most findings support the present predictions.

Feedback and performance: Evidence. Nor-mative difficulty cues are important in thatthey affect expectancies of demonstrating abil-ity in the differentiated sense. Normative feed-back during performance indicates the levelof ability the individual is likely to demonstratewith much less ambiguity. It is also likely toincrease or maintain ego involvement. There-fore, studies using normative feedback providebetter controlled tests of predictions of per-formance in ego involvement.

When, in ego involvement, feedback indi-cates that performance is below that of others,low-perceived-ability individuals should expectto appear incompetent and experience anxietyor reduce effort. Performance would thus be

are gained by asserting a preference for challenging skilltasks. Thus, the study can be seen as indicating that thosewho prefer such tasks perform better on them, whereasthose who do not prefer such tasks perform better on easytasks. In this light, the concept resultant achievement mo-tivation is marginally relevant to Kukla's results. With thiscaveat, the results are consistent with present predictions.

lowered. When feedback indicates above-av-erage performance, these individuals should(if not certain their ability is low) believe thatthey are demonstrating high ability and applyhigh effort to maintain this perception. (Ifsubjects are certain they have low ability, suc-cess feedback should not produce high effort.)Feedback indicating below-average perfor-mance would violate high-perceived-abilityindividuals' expectancies and produce high ef-fort and performance. High normative feed-back should confirm their high perceived abil-ity, imply that less effort is necessary for dem-onstration of high ability, and lead to lowerperformance.

Results consistent with all the above pre-dictions (except that in parentheses, which wasnot tested) were obtained by Weiner (1966)who told high- and low-resultant-motive sub-jects that they were doing much better or worsethan most others during performance on atask described as a test of general ability. Thesefindings were replicated by Weiner andSchneider (1971). Perez (1973) and Schalon(1968) examined performance of high- andlow-self-esteem students after they had beentold that their initial intelligence test perfor-mance was below average. High-self-esteemstudents' performance improved. Performanceof those with low self-esteem declined slightly.Thus, the studies of effects of performancefeedback are consistent with predictions.

The further prediction that individuals whoare certain their ability is low would avoiddemonstrating ability was confirmed by Ma-recek and Mettee (1972). When told they haddisplayed above-average ability, subjects withhigh and low self-concepts, but not those withlow self-concepts who were certain of thisevaluation, showed improvement on a retest.The additional finding that the latter individ-uals improved when performance was pre-sented as dependent on luck rather than onability supports the view that, in the skill con-dition, they avoided demonstrating high ability.There was, in this study, no evidence of thediminished performance predicted for high-perceived-ability subjects after success feed-back. However, ego involvement was probablynot aroused on the pretest because the initiallystated purpose of the session was to determinegroup baseline performance levels. Thus, a de-cline in performance would be unlikely.


In this section, predictions for individualswith low perceived ability were confirmed withconsiderable consistency. The expected di-minished performance of high-perceived-abil-ity individuals after success feedback was notfound in one of three cases. Variation in degreeof ego involvement was probably implicatedin this exception. In summary, the better con-trolled tests of this section of the theory sup-port it.

Performance: Task InvolvementVersus Ego Involvement

Predictions. When individuals are task in-volved, effort reflects the extent to which itappears likely to produce improved masteryor mastery of tasks that individuals are un-certain about their ability to master. If indi-viduals believe that high effort is necessary toproduce improvement they apply high effort.Effort, and thus performance, is predicted tobe lower if they believe that little effort isneeded or that high effort has no effect. Whenindividuals differ in the level of difficulty atwhich they expect gains in mastery, they willcorrespondingly differ in the level of difficultyat which they perform most effectively. How-ever, provided difficulty levels (normative orobjective) are not extreme, all or most indi-viduals will expect to be able to gain in mas-tery. As this is an end in itself, they shouldapply high effort to maximize their masteryand perceived ability. Consequently, theyshould perform effectively. These predictionsresemble those of Kukla (1972) more thanthose of Atkinson.

Many studies that compare ego- and task-involving conditions enable testing of thesepredictions and of the differences between taskinvolvement and ego involvement. High-per-ceived-ability ego-involved individuals' con-cern to perform well compared to others, andtheir expectancy of being able to do so, waspredicted to maintain intense effort and ef-fective performance on tasks of moderate nor-mative difficulty. For low-perceived-ability in-dividuals facing moderate normative difficultylevels, impaired performance is predicted inego involvement but not in task involvement.

The preceding predictions apply to exper-imental settings where tasks are of relativelyshort duration and task requirements areclearly specified. In these cases, the fact that

(in ego involvement) learning is a means toan end is unlikely to impair performance. In-deed, a concern with scoring as high as possiblecould lead to item-selection strategies thatwould increase scores at the expense of learn-ing (Harter, 1978). However, task involvement,where learning is more an end in itself, shouldbe superior to ego involvement in sustainingsignificant real-world achievements such as thedevelopment of logical thinking (Piaget, 1960)and original scientific thought (T. S. Kuhn,1968). Here, individuals must themselves de-tect problems or inconsistencies in their ownlogic or in scientific thought. Endogenous at-tribution appears likely to foster sensitivity tosuch problems (Condry & Chambers, 1978)and to maintain the necessary long-term in-volvement (Campbell, 1960; D. Kuhn, 1974).Effort would be less consistent over long pe-riods and sensitivity to contradictions wouldbe reduced by concern with one's score (Con-dry & Chambers, 1978) when, as in ego in-volvement, learning is a means to an end.

Task involvement versus ego involvement:Evidence. Here I review studies where briefexperimental tasks were presented as mod-erately normatively difficult or allowed a widerange of scores and would, therefore, allow thepossibility of demonstrating high or low ca-pacity. (See Footnote 6.) In such cases, egoinvolvement is predicted to impair perfor-mance for those with low perceived ability sothat they perform more poorly than when taskinvolved and more poorly than high-perceived-ability individuals in task or ego involvement.There are strong reasons for not predictingdiminished performance in ego involvementfor high-perceived-ability individuals andweaker reasons for expecting them to performbetter in ego than task involvement. In thefollowing studies, a variety of manipulationsthat should induce ego involvement werecompared with more neutral conditions.

Entin and Raynor (1973) compared high-and low-resultant-motive students in testlikeand in neutral conditions.13 High-motive sub-

13 Entin and Raynor (1973) characterize the ego-in-volving condition as one where opportunity to work onlater problems is contingent on success on the test. However,this, but not the neutral condition, used social comparisonnorms and was testlike. As identical results occur whenthere is no reference to a contingent future task, this aspectof the methodology is of uncertain relevance.


jects performed better under ego than task in-volvement, whereas the reverse occurred forlow-resultant-motive students. Similar findingswere obtained by Raynor and Rubin (1971).Gjeseme (1974) obtained identical results withanagrams presented as a school test versus atype of problem students would be tested onone year later.

Presentations of tasks as tests (usually ofintelligence) have also been compared withneutral conditions with high- and low-test-anxious students. Studies that have reported(a) the presently predicted poorer performancefor anxious students in test than in neutralconditions and (b) no condition effect for low-anxious students are Paul and Eriksen (1964),I. G. Sarason (1959), and S. B. Sarason, Man-dler, and Craighill (1952). I. G. Sarason andMinard (1962) obtained this result for com-prehension but not for three other tests. Witha digit-symbol task, I. G. Sarason and Palola(1960) also found lower performance for anx-ious students and higher performance for low-anxious students in an intelligence test con-dition. (With very easily discriminable sym-bols, anxious students, as expected, scoredhigher than did others in the test condition.)Russell and I. G. Sarason (1965), however,found no effect of testlike instructions on ana-gram performance. Given the number of fac-tors that could lead to failure to demonstratethe expected effects (e.g., tasks could appeareasy), the above studies favor the hypothesisof impaired performance at moderate diffi-culty levels in low- but not in high-perceived-ability individuals in ego-involving conditions.

Also consistent with present predictions isPerez's (1973) finding that an intelligence testcondition led to lower performance for low-self-esteem subjects than did a neutral pre-sentation, whereas high-self-esteem subjectsperformed similarly in each condition. Sim-ilarly, induction of self-awareness producedpoorer performance in low-self-esteem subjectsthan did a task focus (Brockner, 1979; Brock-ner &Hulton, 1978). High-self-esteem subjectswere not affected by conditions, and task focusproduced similar performance in high- and inlow-self-esteem subjects. Also as predicted, thetask condition reduced anxiety in low-self-es-teem subjects (Brockner, 1979). The presenceof observers also lowered performance in low-perceived-ability (Shrauger, 1972) and in test-

anxious subjects (Ganzer, 1968), but not inhigh-perceived-ability and low-anxious sub-jects.

There is, then, good evidence of the pre-dicted effects of ego versus task involvementwhen difficulty is perceived as moderate,problems are clearly defined, and time periodsare relatively short. Compared to task involve-ment, ego involvement produces lower per-formance in low-perceived-ability individualsand equal or higher performance in high-per-ceived-ability individuals.

There appears to be no relevant experi-mental evidence with tasks involving problemfinding or the development of logical thought.However, Amabile (1979) found that an eval-uative condition lowered artistic creativity.Furthermore, evidence that individuals whomake outstanding creative achievements orwho are more successful in school and in sci-ence are distinguished from others by higherlevels of task involvement rather than ego in-volvement (Spence & Helmreich, 1983; Nich-olls, 1979b) provides support for the presentposition.

Other Theories of Performance

A major difference between the present the-ory and those of Atkinson (1965; Atkinson &Raynor, 1974) and Kukla (1972) is that thepresent one distinguishes and makes separatepredictions for task and for ego involvement.The importance of this distinction is sustainedby the evidence reviewed above. (This is notto say that further distinctions are unnecessary;e.g., Maehr, 1983, and Roberts, in press.)

The present theory also distinguishes nor-mative difficulty and expectancies of success.They are imperfectly distinguished by Kukla,whose predictions are framed in terms of ob-jective difficulty that cannot be specified asprecisely as can normative difficulty. This leadsto inconsistent interpretations of data. At onepoint, Kukla (1972) presents as consistent withhis theory, evidence that low-resultant-motiveindividuals perform better when half of theirpeers are predicted to fail than when the oddsof success are clearly higher or lower (Atkinson,1958). At another point, evidence (Kukla,1974) that low-resultant-motive individualsperform better on a normatively easy than ona moderately difficult task is held to support


his theory. Such inconsistency is unlikely whenpredictions are framed in terms of normative-difficulty cues. The differences between thepresent position and Kukla's stem largely fromhis use of the less differentiated conception ofability. Consequently, the predictions of Ku-kla's theory resemble present predictions fortask involvement. There is no place in his for-mulation for the various performance im-pairment and enhancement effects that occurin ego involvement.

The predictions of Atkinson's theory asmodified by Revelle and Michaels (1976) aremore compatible with the present predictionsfor ego involvement. It seems that in Revelleand Michaels's revision, as in the present the-ory, low-resultant-motive (or low-perceived-ability) individuals would be predicted toperform their worst at moderate normative-difficulty levels and high-perceived-ability in-dividuals would be expected to perform theirworst at easy and at extremely difficult tasks.However, as Revelle and Michaels note (p.400), their position does not predict higherperformance in low- than in nigh-resultant-motive subjects after feedback indicatingabove-average performance (Weiner, 1966;Weiner & Schneider, 1971), whereas this the-ory does. More critical, however, is the failureto deal with task involvement where perfor-mance is often, as predicted, as good as orbetter than in ego involvement.14

The concept of learned helplessness has beenused to account for impaired performance.Abramson et al. (1978) have distinguishedpersonal and universal helplessness. This par-allels the present distinction between percep-tion of low capacity and perception of highnormative difficulty. However, the claim thatimpaired performance is a consequence ofperceived noncontingency of action and out-come embodies the less differentiated concep-tion of ability. Bandura's (1977) distinctionbetween lack of a sense of efficacy and beliefin an unresponsive environment also suggeststhe differentiated conception of ability. How-ever, when Bandura claims that increased per-ception of personal efficacy follows improve-ments in mastery (p. 195), the less differen-tiated conception is implied. But, neitherposition explicitly distinguishes the concep-tions of ability or makes different predictionsfor task and for ego involvement. As the theory

and evidence reviewed here show, performanceimpairment is reduced in task-involved states.Carver and associates have demonstrated sim-ilar effects (Carver, Blaney, & Scheier, 1979;Carver, Peterson, Follansbee, & Scheier, 1983;Carver & Scheier, 1981), as has Kuhl (1981).

This very brief comparison of these differentpositions indicates that the present one dealsrelatively effectively with the evidence on per-formance on tasks of relatively short duration,where the problem to be solved is specifiedand major cognitive restructuring is not re-quired.


I have focused on task choice and perfor-mance in experimental settings because thesetopics have long been of interest to researchersand because data enabling comparisons withprevious theories were available. The follow-ing examples illustrate the applicability ofthe approach to other achievement-relatedphenomena.

First, R. Ames (1983) has analyzed students'requests for academic assistance in terms of(a) ego involvement versus task involvementand (b) individual differences in perceivedability. He proposed that in ego-involvingconditions, students with low perceived abilitywould be more likely than those with highperceived ability to see a request for assistanceas a demonstration of lack of capacity. Thus,they would be less likely to seek assistance. Intask-involving conditions, on the other hand,Ames predicted that most students would viewseeking assistance as a way of learning or dem-onstrating ability in the less differentiatedsense. When task involved, students with lowperceived ability should, therefore, not avoidseeking assistance. Ames's (1983) review ofthe available evidence supports his predictions.

Second, there has been considerable interestin the nature of the associations of attributionsof effort, ability, and other factors with different

14 It might be argued that task-involving situations arenot achievement situations. The problem then would beto explain why individuals perform effectively and choosechallenging tasks in these situations. This position wouldalso have the anomalous implication that the motivationthat makes the most distinctive contribution to outstandingcreative achievement is not a form of achievement mo-tivation.


affects (e.g., Weiner, Russell, &Lerman, 1978).In this context, researchers have assumed thatthe meanings of effort and ability are fixed.However, because the meaning of effort andability can change, the links between affectand perceived effort and ability may alsochange. For example, in ego-involving con-texts, Jagacinski and Nicholls (in press) foundperception of lower effort associated with guiltand perception of higher effort (implying lowercapacity) associated with embarrassment. Intask-involving contexts—where effort andability are imperfectly differentiated—percep-tion of lower effort was associated with in-creases in both guilt and embarrassment. Thiswas expected because lower effort generallyimplies lower ability in task-involving contexts.Thus, the present approach provides new in-sights on achievement-related affect.

Finally, the conceptions of ability that playa central role in this theory also play an im-portant role in the development of achieve-ment motivation. As the conception of abilityas capacity develops, children's achievementaffect and overt behavior increasingly approx-imates that of adults when both are in ego-involving situations (Nicholls & Miller, inpress). Whereas young children and adults dif-fer in ego-involving situations, they react sim-ilarly in task-involving situations. There is alsoevidence that, as children progress throughthe grades, school environments generally be-come more ego involving (Eccles, Midgley, &Adler, in press). Thus, there is a close inter-dependence between the present approach toadult achievement motivation and research ondevelopmental change in achievement moti-vation.


Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. D.(1978). Learned helplessness in humans: Critique andreformulation. Journal of Abnormal Psychology, 87,49-74.

Arnabile, T. M. (1979). Effects of external evaluation onartistic creativity. Journal of Personality and Social Psy-chology, 37, 221-233.

Ames, C. (1978). Children's achievement attributions andself-reinforcement: Effects of self-concept and compet-itive reward structure. Journal of Educational Psy-chology, 70, 345-355.

Ames, C., & Ames, R. (1981). Competitive versus indi-vidualistic goal structures: The salience of past perfor-mance information for causal attributions and affect.Journal of Educational Psychology, 73, 411-418.

Ames, C., Ames, R., & Felker, D. (1977). Effects of com-petitive reward structure and valence of outcome onchildren's achievement attributions. Journal of Edu-cational Psychology, 69, 1-8.

Ames, C., & McKelvie, S. (1983, April). Achievement at-tributions and self-instructions under competitive andindividualistic goal structures. Paper presented at themeeting of the American Educational Research Asso-ciation, Montreal, Quebec, Canada.

Ames, R. (1983). Help-seeking and achievement orien-tation: Perspectives from attribution theory. B. M.DePaulo, A. Nadler, & J. D. Fisher (Eds.), New directionsin helping (pp. 165-188). New \brk: Academic Press.

Arkin, R. M., Detchon, C. S., & Maruyama, G. M. (1982).Roles of attribution, affect, and cognitive interferencein test anxiety. Journal of Personality and Social Psy-chology, 43, 1111-1124.

Arkin, R. M., Kolditz, T. A., & Kolditz, K. K. (1983).Attributions of the test anxious student: Self-assessmentsin the classroom. Personality and Social PsychologyBulletin, 9, 271-280.

Atkinson, J. W. (1957). Motivational determinants of risk-taking behavior. Psychological Review, 64, 359-372.

Atkinson, J. W. (1958). Towards experimental analysis ofhuman motives in terms of motives, expectancies, andincentives. In J. W. Atkinson (Ed.), Motives in fantasy,action, and society (pp. 288-305). New York: Van Nos-trand.

Atkinson, J. W. (1965). The mainsprings of achievement-oriented activity. In J. D. Krumboltz (Ed.), Learningand the educational process (pp. 25-66). Chicago: RandMcNally.

Atkinson, J. W. (1969). Comments on papers by Crandalland Veroff. In C. P. Smith (Ed.), Achievement-relatedmotives in children (pp. 200-206). New York: RussellSage.

Atkinson, J. W., & Feather, N. T. (Eds.). (1966). A theoryof achievement motivation. New York: Wiley.

Atkinson, J. W., & Raynor, J. Q (Eds.). (1974). Personality,motivation, and achievement. New \brk: Hemisphere.

Bandura, A. (1977). Self-efficacy: Toward a unifying theoryof behavioral change. Psychological Review, 84, 191-215.

Bloom, B. S. (1976). Human characteristics and schoollearning. New York: McGraw-Hill.

Bolles, R. C. (1974). Cognition and motivation: Some his-torical trends. In B. Weiner (Ed.), Cognitive views ofhuman motivation (pp. 1-20). New York: AcademicPress.

Brockner, J. (1979). Self-esteem, self-consciousness, andtask performance: Replications, extensions, and possibleexplanations. Journal of Personality and Social Psy-chology, 37, 447-461.

Brockner, J., & Hulton, A. J. B. (1978). How to reversethe vicious cycle of low self-esteem: The importance ofattentional focus. Journal of Experimental Social Psy-chology, 14, 564-578.

Brody, N. (1963). «-Achievement, test anxiety and sub-jective probability of success in risk taking behavior.Journal of Abnormal and Social Psychology, 66, 413-418.

Buckert, U., Meyer, W-U., & Schmalt, H-D. (1979). Effectsof difficulty and diagnosticity on choice among tasks inrelation to achievement motivation and perceived ability.


Journal of'Personality and Social Psychology, 37, 1172-1178.

Campbell, D. T. (1960). Blind variation and selective re-tention in creative thought as in other knowledge pro-cesses. Psychological Review, 67, 380-400.

Carver, C. S., Blaney, P. H., & Scheier, M. F. (1979). Reas-sertion and giving up: The interactive role of self-directedattention and outcome expectancy. Journal of Personalityand Social Psychology, 37, 1859-1870.

Carver, C. S., Peterson, L. M., Follansbee, D. J., & Scheier,J. F. (1983). Effects of self-directed attention on per-formance and persistence among persons high and lowin test anxiety. Cognitive Research and Therapy, 7, 333-354.

Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A control-theory approach to human behavior.New York: Springer-Verlag.

Condry, J. D., & Chambers, J. (1978). Intrinsic motivationand the process of learning. In M. R. Lepper & D.Greene (Eds.), The hidden costs of reward: New per-spectives on the psychology of human motivation (pp.61-84). Hillsdale, NJ: Erlbaum.

Crandall, R. (1973). The measurement of self-esteem andrelated constructs. In J. P. Robinson & P. R. Shaver(Eds.), Measures of social psychological attitudes (pp.45-168). Ann Arbor, MI: Institute for Social Research.

Crandall, V. J., Katkovsky, W., & Preston, A. (1960). Aconceptual formulation for some research on children'sachievement development. Child Development, 31, 787-797.

Csikszentmihalyi, M. (1977). Beyond boredom and anxiety.San Francisco: Jossey-Bass.

DeCharms, R., & Carpenter, V. (1968). Measuring mo-tivation in culturally disadvantaged school children. InH. J. Klausmeier and G. T. O'Hearn (Eds.), Researchand development toward the improvement of education(pp. 31 -41). Madison, WI: Dembar Education Services.

DeCharms, R., & Dave, P. N. (1965). Hope of success,fear of failure, subjective probability, and risk-takingbehavior. Journal of Personality and Social Psychology,1, 558-568.

Deci, E. L. (1975). Intrinsic motivation. New York: Plenum.Deci, E. L., Betley, G., Kahle, J., Abrams, L., & Porac,

J. (1981). When trying to win: Competition and intrinsicmotivation. Personality and Social Psychology Bulletin,7, 79-83.

Deci, E. L., & Ryan, R. M. (1980). The empirical explo-ration of intrinsic motivational processes. In L. Ber-kowitz (Ed.), Advances in experimental social psychology(Vol. 13, pp. 39-80). New York: Academic Press.

Deci, E. L., Schwartz, A. J., Sheinman, L., & Ryan,R. M. (1981). An instrument to assess adults' orien-tations toward control versus autonomy with children:Reflections on intrinsic motivation and perceived com-petence. Journal of Educational Psychology, 73, 642-650.

Dennett, D. C. (1978). Brainstorms: Philosophical essayson mind and psychology. Montgomery, VT: Bradford.

Diener, E., & Srull, T. K. (1979). Self-awareness, psycho-logical perspective, and self-reinforcement in relationto personal and social standards. Journal of Personalityand Social Psychology, 37, 413-423.

Eccles, J., Midgley, C., & Adler, T. (in press). Age-relatedchanges in the school environment: Effects on achieve-ment motivation. In J. G. Nicholls (Ed.), The devel-

opment of achievement motivation. Greenwich, CT: JAIPress.

Elkind, D. (1971). Cognitive growth cycles in mental de-velopment. In J. K. Cole (Ed.), Nebraska Symposiumon Motivation (pp. 1-31). Lincoln: University of Ne-braska Press.

Entin, E. E., & Raynor, J. Q (1973). Effects of contingentfuture orientation and achievement motivation on per-formance in two kinds of task. Journal of ExperimentalResearch in Personality, 6, 314-320.

Feather, N. T, & Simon, J. G. (1971). Attribution of re-sponsibility and valence of outcome in relation to initialconfidence and success and failure of self and other.Journal of Personality and Social Psychology, 18, 173-188.

Frankel, A., & Snyder, M. L. (1978). Poor performancefollowing unsolvable problems: Learned helplessness oregotism? Journal of Personality and Social Psychology,36, 628-634.

Ganzer, V. J. (1968). Effects of audience presence and testanxiety on learning and retention in a serial learningsituation. Journal of Personality and Social Psychology,8, 194-199.

Garcia, J. (1981). Tilting at the paper mills of academe.American Psychologist, 36, 149-158.

Gjesme, T. (1974). Goal distance in time and its effectson the relations between achievement motives and per-formance. Journal of Research in Personality, 8, 161-171.

Hamilton, J. O. (1974). Motivation and risk-taking be-havior: A test of Atkinson's theory. Journal of Personalityand Social Psychology, 29, 856-864.

Harlow, H. F. (1950). Learning and satiation of responsein intrinsically motivated complex puzzle performanceby monkeys. Journal of Comparative and PhysiologicalPsychology, 43, 289-294.

Harter, S. (1978). Pleasure derived from challenge and theeffects of receiving grades on children's difficulty levelchoices. Child Development, 49, 788-799.

Heckhausen, H. (1967). The anatomy of achievement mo-tivation. New York: Academic Press.

Heckhausen, H. (1968). Achievement motive research:Current problems and some contributions toward ageneral theory of motivation. In W. J. Arnold (Ed.),Nebraska Symposium on Motivation (pp. 103-174).Lincoln: University of Nebraska Press.

Heckhausen, H. (1977). Achievement motivation and itsconstructs: A cognitive model. Motivation and Emotion,1, 283-329.

Heckhausen, H. (in press). Emergent achievement behavior:Some early developments. In J. G. Nicholls (Ed.), Thedevelopment of achievement motivation. Greenwich, CT:JAI Press.

Hetherington, E. M., & Parke, R. D. (1975). Child psy-chology: A contemporary viewpoint. New Vbrk: McGraw-Hill.

Hoffman, C., Mischel, W., & Mazze, K. (1981). The roleof purpose in the organization of information aboutbehavior: Trait-based versus goal-based categories inperson cognition. Journal of Personality and Social Psy-chology, 40, 211-225.

Jagacinski, C. M., & Nicholls, J. G. (in press). Conceptionsof ability and related affects in task involvement andego involvement. Journal of Educational Psychology.

Jopt, J. U. (1974). Extrinsiche motivation and leistungs-


verhalten. Unpublished doctoral dissertation, RuhrUniversity, Bochum, West Germany.

Karabenick, S. A., & Youssef, Z. I, (1968). Performanceas a function of achievement motive level and perceiveddifficulty. Journal of Personality and Social Psychology,10, 414-419.

Klinger, E. (1975). Consequences of commitment to anddisengagement from incentives. Psychological Review,82, 1-25.

Klinger, E., & McNelly, F. W. (1969'). Fantasy needachievement and performance: A role analysis. Psy-chological Review, 76, 574-591.

Kruglanski, A. W. (1975). The endogenous-exogenouspartition in attribution theory. Psychological Review,82, 387-406.

Kuril, J. (1981). Motivational and functional helplessness:The moderating effect of state versus action orientation.Journal of Personality and Social Psychology, 40, 155-170.

Kuhn, D. (1974). Inducing development experimentally:Comments on a research paradigm. DevelopmentalPsychology, 10, 590-600.

Kuhn, T. S. (1968). A function for thought experiments.Ontario Journal of Educational Research, 10, 211-231.

Kukla, A. (1972). Foundations of an attributional theoryof performance. Psychological Review, 79, 454-470.

Kukla, A. (1974). Performance as a function of resultantachievement motivation (perceived ability) and perceiveddifficulty. Journal of Research in Personality, 7, 374-383.

Kukla, A. (1978). An attributional theory of choice. InL. Berkowitz (Ed.), Advances in experimental socialpsychology (Vol. 11, pp. 113-144). New \brk: AcademicPress.

Lepper, M. R., & Greene, D. (1975). Turning play intowork: Effects of adult surveillance and extrinsic rewardson children's intrinsic motivation, Journal of Personalityand Social Psychology, 31, 479-486.

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

Maehr, M. L. (1983). On doing well in science: Why Johnnyno longer excels; Why Sarah never did. In S. G. Paris,G. M. Olson, & H. W. Stevenson (Eds.), Learning andmotivation in the classroom (pp. 179-210). Hillsdale,NJ: Erlbaum.

Maehr, M. L., & Nicholls, J. G. (1980). Culture andachievement motivation: A second look. In N. Warren(Ed.), Studies in cross-cultural psychology (Vol. 2, pp.221-267. New York: Academic Press.

Mahone, C. H. (1960). Fear of failure and unrealistic vo-cational aspiration. Journal of Abnormal and SocialPsychology, 60, 253-261.

Marecek, J., & Mettee, D. R. (1972). Avoidance of con-tinued success as a function of self-esteem, level of self-esteem certainty, and responsibility for success. Journalof Personality and Social Psychology, 22, 98-107.

McClelland, D. C. (1958). Risk taking in children withhigh and low need for achievement. In J. W. Atkinson(Ed.), Motives in fantasy, action, and society (pp. 306-321). New York: Van Nostrand.

McClelland, D. C., Atkinson, J. W, Clark, R. W., & Lowell,E. L. (1953). The achievement motive. New York: Ap-pleton-Century Crofts.

McFarland, C., & Ross, M. (1982). Impact of causal at-tributions on affective reactions to success and failure.Journal of Personality and Social Psychology, 43, 937-946.

McFarlin, D. B., & Blascovich, J. (1981). Effects of self-esteem and performance feedback on future affectivepreferences and cognitive expectations. Journal of Per-sonality and Social Psychology, 40, 521-531.

Mehrabian, A. (1969). Measures of achieving tendency.Educational and Psychological Measurement, 29, 445-451.

Meyer, W-U., Folkes, V, & Weiner, B. (1976). The perceivedinformational value and affective consequences of choicebehavior and intermediate difficulty task selection.Journal of Research in Personality, 10, 410-423.

Miller, A. T. (1982). Self-recognitory schemes and achieve-ment behavior: A developmental study. Unpublisheddoctoral dissertation, Purdue University.

Morris, L. W., Davis, M. A., & Hutchings, C. H. (1981).Cognitive and emotional components of anxiety: Lit-erature review and a revised worry-emotionality scale.Journal of Educational Psychology, 73, 541-555.

Moulton, R. W. (1965). Effects of success and failure onlevel of aspiration as related to achievement motives.Journal of Personality and Social Psychology, 1, 399-406.

Nicholls, J. G. (1976). When a scale measures more thanits name denotes: The case of the Test Anxiety Scalefor Children. Journal of Consulting and Clinical Psy-chology, 44, 976-985.

Nicholls, J. G. (1978). The development of the conceptsof effort and ability, perception of own attainment, andthe understanding that difficult tasks require more ability.Child Development, 49, 800-814.

Nicholls, J. G. (1979a). Development of perception of ownattainment and causal attributions for success and failurein reading. Journal of Educational Psychology, 71,94-99.

Nicholls, J. G. (1979b). Quality and equality in intellectualdevelopment: The role of motivation in education.American Psychologist, 34, 1071-1084.

Nicholls, J. G. (1980). The development of the conceptof difficulty. Merrill-Palmer Quarterly, 26, 271-281.

Nicholls, J. G., Licht, B., & Pearl, R. (1982). Some dangersof using personality questionnaires to study personality.Psychological Bulletin, 92, 572-580.

Nicholls, J. G., & Miller, A. T. (1983). The differentiationof the concepts of difficulty and ability. Child Devel-opment, 54, 951-959.

Nicholls, J. G., & Miller, A. T. (in press). Developmentand its discontents: The differentiation of the conceptof ability. In J. G. Nicholls (Ed.), The development ofachievement motivation. Greenwich, CT: JAI Press.

Patten, R. L., & White, L. A. (1977). Independent effectsof achievement motivation and overt attribution onachievement behavior. Motivation and Emotion, 1,39-59.

Paul, G. L., & Eriksen, C. W. (1964). Effects of test anxietyon "real-life" examinations. Journal of Personality, 32,480-494.

Perez, R. C. (1973). The effect of experimentally inducedfailure, self-esteem, and sex on cognitive differentiation.Journal of Abnormal Psychology, 81, 74-79.

Piaget, J. (1960). The general problems of the psycho-biological development of the child. In J. M. Tanner &B. Inhelder (Eds.), Discussions on child development(Vol. 4, pp. 3-27). London: Tavistock.


Raynor, J. Q, & Rubin, I. S. (1971). Effects of achievementmotivation and future orientation on level of perfor-mance. Journal of Personality and Social Psychology,17, 36-41.

Raynor, J. Q, & Smith, C. P. (1966). Achievement relatedmotives and risk-taking in games of skill and chance.Journal of Personality, 34, 176-198.

Revelle, W., & Michaels, E. J. (1976). The theory ofachievement motivation revisited: The implication ofinertial tendencies. Psychological Review, 83, 394-404.

Roberts, G. C. (1974). Effect of achievement and socialenvironment on risk-taking. Research Quarterly, 45,42-55.

Roberts, G. C. (in press). Children's achievement moti-vation in sport. In J. G. Nicholls (Ed.), The developmentof achievement motivation. Greenwich, CT: JAI Press.

Roberts, G. C., Kleiber, D. A., & Duda, J. L. (1981). Ananalysis of motivation in children's sport: The role ofperceived competence in participation. Journal of SportPsychology, 3, 206-216.

Russell, D. G., & Sarason, I. G. (1965). Test anxiety, sex,and experimental conditions in relation to anagram so-lution. Journal of Personality and Social Psychology, 1,493-496.

Ryan, R. M. (1982). Control and information in the in-trapersonal sphere: An extension of cognitive evaluationtheory. Journal of Personality and Social Psychology,43, 450-461.

Sarason, I. G. (1958). The effects of anxiety, reassurance,and meaningfulness of material to be learned in verballearning. Journal of Experimental Psychology, 56, 472-477.

Sarason, I. G. (1959). Relationships of measures of anxietyand experimental instructions to word association testperformance. Journal of Abnormal and Social Psy-chology, 59, 37-42.

Sarason, I. G. (1961). The effects of anxiety and threaton solution of a difficult task. Journal of Abnormal andSocial Psychology, 62, 165-168.

Sarason, I. G. (1975). Anxiety and self-preoccupation. InI. G. Sarason & C. D. Spielberger, (Eds.), Stress andanxiety (Vol. 2, pp. 27-44). Washington, DC: Wiley-Hemisphere.

Sarason, I. G., & Minard, J. (1962). Test anxiety, exper-imental instructions and the Wechslcr Adult IntelligenceScale. Journal of Educational Psychology, 53, 299-302.

Sarason, I. G., & Palola, E. G. (I960). The relationshipof test and general anxiety, difficulty of task, and ex-perimental instructions to performance. Journal of Ex-perimental Psychology, 59, 185-191.

Sarason, S. B., Mandler, G., & Craighill, P. G. (1952). Theeffect of differential instructions on anxiety and learning.Journal of Abnormal and Social Psychology, 47, 561-565.

Schalon, C. L. (1968). Effect of self-esteem upon perfor-mance following failure stress. Journal of Consultingand Clinical Psychology, 32, 497.

Scheier, M. F. & Carver, C. S. (1983). Self-directed attentionand the comparison of self with standards. Journal ofExperimental Social Psychology, 19, 205-222.

Schneider, K. (1973). Motivation under Erfolgsriskia.Gottingen, West Germany: Hogrefe.

Sears, P. S. (1940). Levels of aspiration in academicallysuccessful and unsuccessful children. Journal of Ab-normal and Social Psychology, 35, 498-536.

Sears, P. S. (1941). Levels of aspiration in relation to somevariables of personality: Clinical studies. Journal of So-cial Psychology, 14, 311-336.

Shrauger, J. S. (1972). Self-esteem and reactions to beingobserved by others. Journal of Personality and SocialPsychology, 23, 192-200.

Shrauger, J. S. (1975). Responses to evaluation as a functionof initial self-perceptions. Psychological Bulletin, 82,581-596.

Simon, J. G., & Feather, N. T. (1973). Causal attributionsfor success and failure at university examinations. Jour-nal of Educational Psychology, 64, 46-56.

Spence, J. X, & Helmreich, R. L. (1983). Achievement-related motives and behavior. In J. X Spence (Ed.),Achievement and achievement motives: Psychological andsociological approaches (pp. 10-74). San Francisco:Freeman.

Trope, Y. (1979). Uncertainty-reducing properties ofachievement tasks. Journal of Personality and SocialPsychology, 37, 1505-1518.

Valle, V. A., & Frieze, I. H. (1976). Stability of causalattributions as a mediator in changing expectations ofsuccess. Journal of Personality and Social Psychology,33, 579-587.

Weiner, B. (1966). The role of success and failure in thelearning of easy and complex tasks. Journal of Person-ality and Social Psychology, 3, 339-343.

Weiner, B., Russell, D., & Lerman, D. (1978). Affectiveconsequences of causal ascriptions. In J. H. Harvey,W. J. Ickes, & R. F. Kidd (Eds.), New directions inattribution research (Vol. 2, pp. 59-90). Hillsdale, NJ:Erlbaum.

Weiner, B., & Schneider, K. (1971). Drive versus cognitivetheory: A reply to Boor and Harman. Journal of Per-sonality and Social Psychology, 18, 285-263.

White, R. W. (1959). Motivation reconsidered: The conceptof competence. Psychological Review, 66, 297-333.

Wine, J. (1971). Test anxiety and direction of attention.Psychological Bulletin, 76, 92-104.

Received September 19, 1983Revision received December 7, 1983 •