self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

12
Journal of Personality and Social Psychology 1983, Vol. 45, No. 5, 1017-1028 Copyright 1983 by the American Psychological Association, Inc. Self-Evaluative and Self-Efficacy Mechanisms Governing the Motivational Effects of Goal Systems Albert Bandura and Daniel Cervone Stanford University The present research tested the hypothesis that self-evaluative and self-efficacy mechanisms mediate the effects of goal systems on performance motivation. These self-reactive influences are activated through cognitive comparison requiring both personal standards and knowledge of performance. Subjects performed a strenuous activity with either goals and performance feedback, goals alone, feedback alone, or without either factor. The condition combining performance information and a standard had a strong motivational impact, whereas neither goals alone nor feedback alone effected changes in motivation level. When both comparative factors were present, the evaluative and efficacy self-reactive influences predicted the mag- nitude of motivation enhancement. The higher the, self-dissatisfaction with a sub- standard performance and the stronger the perceived self-efficacy for goal attainment, the greater was the subsequent intensification of effort. When one of the comparative factors was lacking, the self-reactive influences were differentially related to per- formance motivation, depending on the nature of the partial information and on the type of subjective comparative structure imposed on the activity. The capability for intentional and purposive human action is rooted in cognitive activity. Social learning theory postulates two cogni- tively based mechanisms of motivation that serve such telic purposes. One mechanism op- erates anticipatorily through the exercise of forethought. By representing forseeable out- comes symbolically, future consequences can be converted into current motivators and reg- ulators of behavior. The second major source of cognitive motivation derives from internal standards and self-evaluative reactions to one's performances (Bandura, 1977a). The motivational effects of setting goals, which provides the standard against which performance is gauged, have been amply doc- umented in different lines of research con- ducted under both controlled and naturalistic conditions. The evidence is relatively consis- tent in showing that explicit challenging goals enhance performance motivation (Locke, Shaw, Saari, & Latham, 1981). However, the This research was supported by Public Health Research Grant M-5162-20 from the National Institute of Mental Health. We are indebted to William L. Haskell, Stanford Medical Center, for the loan of the ergometer equipment. Requests for reprints should be sent to Albert Bandura, Department of Psychology, Building 420-Jordan Hall, Stanford University, Stanford, California, 94305. psychological mechanisms through which personal standards create motivational effects and how these mechanisms govern motivation under different patterns of performance in- formation have received less attention. It is to these issues that the present research addresses itself. In the social learning analysis, self-moti- vation through performance standards oper- ates largely through an internal comparison process (Bandura, 1978). When people com- mit themselves to explicit standards or goals, perceived negative discrepancies between what they do and what they seek to achieve creates self-dissatisfactions that serve as motivational inducements for enhanced effort. Both the an- ticipated self-satisfactions for matching ac- complishments and the self-dissatisfactions with substandard performances provide in- centives for heightened effort. Performance motivation is not posited to be a monotonically increasing function of de- gree of perceived discrepancy. Performances that fall markedly short of standards are apt to give rise to discouragement and goal aban- donment. Moderately discrepant perfor- mances, which leave construal of the standard as attainable (Atkinson, 1964; Locke, 1968), are likely to activate self-dissatisfactions that spur efforts to bring performance in line with 1017

Upload: daniel

Post on 14-Dec-2016

223 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

Journal of Personality and Social Psychology1983, Vol. 45, No. 5, 1017-1028

Copyright 1983 by theAmerican Psychological Association, Inc.

Self-Evaluative and Self-Efficacy Mechanisms Governing theMotivational Effects of Goal Systems

Albert Bandura and Daniel CervoneStanford University

The present research tested the hypothesis that self-evaluative and self-efficacymechanisms mediate the effects of goal systems on performance motivation. Theseself-reactive influences are activated through cognitive comparison requiring bothpersonal standards and knowledge of performance. Subjects performed a strenuousactivity with either goals and performance feedback, goals alone, feedback alone,or without either factor. The condition combining performance information anda standard had a strong motivational impact, whereas neither goals alone norfeedback alone effected changes in motivation level. When both comparative factorswere present, the evaluative and efficacy self-reactive influences predicted the mag-nitude of motivation enhancement. The higher the, self-dissatisfaction with a sub-standard performance and the stronger the perceived self-efficacy for goal attainment,the greater was the subsequent intensification of effort. When one of the comparativefactors was lacking, the self-reactive influences were differentially related to per-formance motivation, depending on the nature of the partial information and onthe type of subjective comparative structure imposed on the activity.

The capability for intentional and purposivehuman action is rooted in cognitive activity.Social learning theory postulates two cogni-tively based mechanisms of motivation thatserve such telic purposes. One mechanism op-erates anticipatorily through the exercise offorethought. By representing forseeable out-comes symbolically, future consequences canbe converted into current motivators and reg-ulators of behavior. The second major sourceof cognitive motivation derives from internalstandards and self-evaluative reactions to one'sperformances (Bandura, 1977a).

The motivational effects of setting goals,which provides the standard against whichperformance is gauged, have been amply doc-umented in different lines of research con-ducted under both controlled and naturalisticconditions. The evidence is relatively consis-tent in showing that explicit challenging goalsenhance performance motivation (Locke,Shaw, Saari, & Latham, 1981). However, the

This research was supported by Public Health ResearchGrant M-5162-20 from the National Institute of MentalHealth. We are indebted to William L. Haskell, StanfordMedical Center, for the loan of the ergometer equipment.

Requests for reprints should be sent to Albert Bandura,Department of Psychology, Building 420-Jordan Hall,Stanford University, Stanford, California, 94305.

psychological mechanisms through whichpersonal standards create motivational effectsand how these mechanisms govern motivationunder different patterns of performance in-formation have received less attention. It is tothese issues that the present research addressesitself.

In the social learning analysis, self-moti-vation through performance standards oper-ates largely through an internal comparisonprocess (Bandura, 1978). When people com-mit themselves to explicit standards or goals,perceived negative discrepancies between whatthey do and what they seek to achieve createsself-dissatisfactions that serve as motivationalinducements for enhanced effort. Both the an-ticipated self-satisfactions for matching ac-complishments and the self-dissatisfactionswith substandard performances provide in-centives for heightened effort.

Performance motivation is not posited tobe a monotonically increasing function of de-gree of perceived discrepancy. Performancesthat fall markedly short of standards are aptto give rise to discouragement and goal aban-donment. Moderately discrepant perfor-mances, which leave construal of the standardas attainable (Atkinson, 1964; Locke, 1968),are likely to activate self-dissatisfactions thatspur efforts to bring performance in line with

1017

Page 2: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

1018 ALBERT BANDURA AND DANIEL CERVONE

valued standards. Attainments that match orsurpass personal standards create self-satisfac-tions that serve as positive inducements forfurther pursuits.

Activation of self-evaluative processesthrough internal comparison requires bothpersonal standards and knowledge of the levelof one's performance. It follows from this for-mulation and Locke's goal theory (1968) thatneither knowledge of performance withoutstandards nor standards without knowledge ofperformance provides a basis for self-evaluativereactions and thus has little motivational im-pact. There is some empirical evidence to sug-gest that this is indeed the case (Becker, 1978;Strang, Lawrence, & Fowler, 1978). Simplyadopting goals, whether easy or personallychallenging ones, without knowing how oneis doing seems to have no appreciable moti-vational effects.

Results of studies varying the properties ofgoals are also in accord with the postulatedself-evaluative mechanisms (Bandura &Schunk, 1981; Bandura & Simon, 1977;Locke, 1968; Steers & Porter, 1974). Explic-itness, challengeability, and temporal prox-imity of subgoals and standards are conduciveto enlisting self-reactive influence by specifyingthe amount and type of effort required to fulfillthe goals. Such properties augment the mo-tivational impact of goals. Empirical verifi-cation of the self-reactive causal link wouldprovide a conceptual framework within whichto analyze the features of goal systems thatcarry motivational potential.

The self-efficacy mechanism also plays acentral role in human agency and self-moti-vation (Bandura, 1981, 1982). It is partly onthe basis of self-percepts of efficacy that peoplechoose what to do, how much effort to mo-bilize for given activities, and how long to per-severe at them (Bandura, 1977b; Brown & In-ouye, 1978; Schunk, 1981; Weinberg, Gould,& Jackson, 1979). Whether negative discrep-ancies between standards and performance aremotivating or discouraging is likely to be in-^fluenced by people's perceptions of their ef-ficacy to attain the standards they set forthemselves. Those who have a low sense ofself-efficacy may be easily discouraged by fail-ure, whereas those who are assured of theircapabilities for goal attainment intensify their

efforts when their performances fall short andpersist until they succeed.

The present research was primarily designedto test the notion that self-evaluative and self-efficacy mechanisms operate differentially inperformance motivation, depending on thestructure of comparative performance factors.The guiding conceptual scheme posits thatboth of these self-processes jointly regulate ef-fort under conditions permitting cognitivecomparison between a standard and knowl-edge of performance. To test this notion, con-ditions were created in which both compar-ative factors were present, one of the com-parative factors was lacking, or both wereabsent. Subjects performed a strenuous phys-ical activity on an ergometric device underconditions including either goals with perfor-mance feedback, goals alone, feedback alone,or without the presence of either factor. Toequalize the constituent factors across subjectsand treatment conditions, the goals involveda 40% increase in effortful performance andfeedback of a 24% gain in performance. Thesevalues were selected to create a moderate neg-ative discrepancy sufficient to activate self-dis-satisfied reactions without unduly undermin-ing self-percepts of efficacy for goal attainment.The research of Atkinson (1964) and Locke(1968) indicates that a moderate discrepancyis well suited for this purpose. After the per-formance session in which the requisite goaland feedback conditions were created, subjectsrecorded their level of self-satisfaction withtheir performance and their perceived self-ef-ficacy for goal attainment, whereupon theireffortful performance was again measured.

By systematically varying the comparativefactors and measuring the posited self-pro-cesses antecedently, the integrated design per-mits a dual level of verification of requisiteconditions for motivational enhancement tiedto mediating self-processes. It was predictedthat subjects in the condition combining goalswith performance feedback would display thehighest gains in effortful performance. On thepremise that self-evaluative and self-efficacymechanisms are most consistently activatedin the service of motivation only when goalsand feedback information are both present, itwas hypothesized that subjects receiving eithergoals or feedback alone would lack an essential

Page 3: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

SELF-REACTIVE MECHANISMS IN GOAL MOTIVATORS 1019

comparative element and, hence, would notdiffer from those receiving neither of these fac-tors.

As the primary test of the theory under ex-amination, it was predicted that in the con-dition combining goals with feedback, themagnitude of performance gains would varyas a function of level of self-dissatisfaction andperceived self-efficacy for .goal attainment. Thehigher the subjects' self-dissatisfaction, themore they would increase their performanceto bring it in line with their standard. Thehigher their self-percepts of efficacy, the greatereffort they would mount to attayi their goal.Subjects who were self-dissatisfied with theirsubstandard performance but judged them-selves highly efficacious in attaining their goalwould show the highest performance gains. Incontrast, those who judged themselves inef-ficacious and were not unduly self-dissatisfiedwould mobilize the weakest effort. In condi-tions in which one or both comparative factorsare lacking, the mediating mechanisms arelikely to operate variably depending on whatpartial information is available and on sub-jective provision of the missing comparativefactor.

Method

SubjectsThe subjects were 45 men and 45 women drawn from

an introductory psychology course. Twenty subjects, equallydivided by sex, were randomly assigned to each of fourtreatment conditions. Ten subjects were similarly assignedio a self-judgment control condition designed to determinewhether recording one's self-satisfaction and self-perceptsof efficacy, in itself, had any reactive effects on performance.

General Procedure and ApparatusThe introductory instructions describing the nature of

the study were identical for all subjects. The experimentwas presented as part of a program of research ostensiblydesigned to identify performance tasks that might even-tually prove useful for planning and evaluating postcoro-nary rehabilitation programs. It was further explained thatthe information being gathered would not only aid de-velopment of diagnostic devices but also provide normative,data on physical stamina at different age levels. The relationbetween cardiovascular fitness and performance on aerobictasks was then described to lend further credibility to theactivity.

The performance task apparatus was a Schwinn Air-Dyne ergometer in the form of an exercise device thatuses a wind vane system to provide variable air resistance.

In its modified form, the ergometer was operated by al-ternatively pulling and pushing two arm levers. The exertedforce rotated a wheel with fanlike wind vanes, creatingresistance for the physical effort.

The ergometer task was chosen for a number of reasons.This effortful activity combined with the rationale wasreceived with uniformly high credibility. It yielded a precisemeasure of performance effort with virtually no upperlimit. Because it required considerable effort over extendedperiods, the task provided a stringent test of how the pos-tulated determinants and mechanisms affect the mobili-zation and maintenance of performance motivation. Fi-nally, the task itself provided little implicit feedback re-garding performance level, which allowed for credibleprearranged feedback. Because subjects could not easilydiscern quantitative variations in their physical outputacross sessions, goals and feedback information could besystematically varied without jeopardizing the perceivedveridicality of the feedback.

The ergometer was connected by a cable to a work loadindicator with an odometer in the adjoining room. Theodometer readings were recorded at 1-minute intervalsduring the 5-minute sessions so as to capture any variationsin performance during the session. To measure preciselythe performance effort expended, the odometer readingswere converted to kilopond meter units. Kilopond unitsare indices of work output that consider both the speedat which the ergometer is operated and the exponentialincrease in air resistance with increasing speed. The fivesets of kilopond scores resulting from the five 1-minuteintervals were summed to obtain a total performance scorefor each session.

Before starting the experiment, subjects completed abackground questionnaire that asked about their age, sex,height, weight, and smoking habits. It was included bothto add further credence to the prior instructions and toincrease the naturalness of the assessment, in a later session,of self-reactions imbedded among filler items ostensiblytapping other aspects of physical status. They also filledout a physical-readiness questionnaire designed to excludeany subject for whom extended physical exertion wouldbe medically contraindicated. Only one subject, who re-ported a history of cardiovascular problems, was excludedon this basis.

Subjects removed their watches to control for possiblevariations in the regulation of effort by checking the timeelapsed. They were informed that each performance sessionwould last 5 minutes but were not told how many sessionsthey would complete. The latter procedure was institutedto eliminate the possibility that subjects might intensifytheir performance in the third session if they knew it wastheir final effort.

The experimenter concluded the general instructionsby explaining that he would be in the adjoining roomtending to the recording instruments during each perfor-mance session. The subject would be signaled when tobegin and end each session via an intercom system.

Baseline Performance Session

All subjects performed the ergometer task alone for a5-minute baseline period. Pretesting indicated that a .5-

Page 4: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

1020 ALBERT BANDURA AND DANIEL CERVONE

minute session required substantial performance effortwithout being overly fatiguing. •

Following the baseline assessment, subjects were ran-domly assigned within sex groupings to treatment con-ditions. The random order of assignment was devised forthe entire sample at the outset of the study. After eachsubject completed the baseline session, the experimenterremoved a cover card that revealed the condition to whichthe particular subject was assigned. Thus, the experimenterhad no prior knowledge of the subjects' condition assign-ments during the baseline session.

Goal Setting and Feedback VariationsIn conditions including goal setting, subjects selected a

goal for performance improvement in subsequent sessions.The experimenter explained that in coronary rehabilitationprograms patients have goals for increasing their physicalactivity. These goals vary across cases. Therefore they wouldperform the ergometer task with goals to shed light on theeffects of this goal variability.

Goals were not simply assigned to subjects. Rather, theyostensibly selected their own goal level. An apparent-choiceprocedure was used to increase subjects' sense of self-determination and commitment to the goal (Kiesler, 1971;Langer, 1975). They were told that to study goal levelsrepresentative of the range found in a rehabilitation pro-gram, they would choose one goal from among a widerange of goal levels. It was explained that in rehabilitationprograms, goals are set based on the patient's currentphysical status. However, because this psychological in-formation was not available for participants in the presentstudy, there was no basis for assigning particular goals toparticular subjects. Hence, subjects would simply selectone from among a variety of goals.

Different goal levels representing percentage decrementsand percentage improvements above baseline performancewere printed on cards. After the full range of goal levelswas inspected, the experimenter placed all the goal cardsin a cloth bag attached to a wooden rim and handle, shookthe bag, and presented it to the subject, who selected agoal. Unbeknownst to the subject, the choice was prear-ranged to be a 40% increase in performance above baselineperformance. This was achieved by flicking a hidden switchon the bottom of the handle, automatically switching com-partments of the bag so that subjects were selecting theirgoals from a preloaded set of goal cards, all of whichrepresented a 40% performance increase.

A 40% goal level was chosen for several reasons. It rep-resented an attainable goal, a negative performance dis-crepancy from it would appear credible, and neither thegoal nor the performance discrepancy was so high as toundermine perceived self-efficacy in attaining it.

Subjects in conditions that did not include goals receivedthe identical information as did their goal-setting coun-terparts concerning how coronary rehabilitation patientsstrive for different goals of increased physical activity aspart of their recovery program. They were all given thesame information about goals and striving for performanceimprovement to equate the groups for the suggestion ofincreased effort. The experiment thus provided a clear testof the motivational contribution of actual goal adoption.

To control for any possible experimenter bias, all theinformation for creating the requisite conditions for themain phase of the experiment was presented remotely via

a video system. The experimenter explained that he hadto reset the recording instruments after the second session.Thus, the video system would be used to convey furtherinformation. Subjects then performed the ergometer taskfor 5 minutes alone in the room, whereupon they wereinstructed through the intercom to turn on the video ter-minal.

The performance feedback and goal-setting informationwas printed on subsets of cards that could be combinedto include goals and feedback, goals alone, feedback alone,or neither. A camera in the adjoining room transmittedthe relevant information to the video screen. In the feed-back-alone conditions the sign read "Your performancescore for the last session was % your firstsession." The experimenter wrote "24" and "above," re-spectively, in the blanks. This feedback information, in-dependent of the subject's actual performance, was writtenin the blanks to avoid the impression that the feedbackmay have been prearranged. For subjects in goals-aloneconditions, the sign read, "The goal you were aiming foris ." The experimenter filled in "+40%." The abovetwo subsigns were combined for subjects in the goals andfeedback condition, informing them that they had attaineda 24% increase in performance and were aiming for a 40%increase.

For subjects in all conditions, the next sign that appearedon the screen instructed them to complete a questionnairethat was next to the video terminal.

Measurement of Self-Evaluation andPerceived Self-Efficacy

The questionnaire contained the two measures of centralinterest: subjects' level of self-satisfaction with their per-formance and their perceived self-efficacy at reaching var-ious performance attainments. These scales were embeddedin a set of filler items (cast in the same format) measuringexercise routines and general physical status.

In measuring self-evaluative reactions, subjects ratedtheir self-reactions on a 25-point scale, ranging from "highlyself-satisfied," through "neutral," to "highly self-dissatis-fied." They first rated their level of self-satisfaction or dis-satisfaction with their performance in the second session,which they had just completed. Subjects could be pleasedwith their prior progress but self-dissatisfied were they toachieve the same level of performance on their subsequentattempts. Hence, for the second rating, subjects rated howself-satisfied or self-dissatisfied they would be if they at-tained the same level of perfomance in the next session.

Subjects recorded their perceived self-efficacy for goalattainments on an efficacy scale that described 14 possiblelevels of performance attainments relative to the baselinelevel. The goal attainments varied in 10% intervals froma 50% decrement to an 80% increase above the baselinelevel. For each of the 14 performance levels, subjects ratedthe strength of their perceived efficacy that they couldattain them on a 100-point scale, ranging in 10-unit in-tervals from "high uncertainty" to "intermediate valuesof certainty" to "complete certitude." The strength of sub-jects' perceived efficacy that they could achieve a 40%performance increase was the pertinent efficacy measure.

Performance TestAfter the assessment of self-processes, subjects were in-

structed via the intercom to resume the ergometer task.

Page 5: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

SELF-REACTIVE MECHANISMS IN GOAL MOTIVATORS

They engaged in the effort|ul actiwhich their performance

At the conclusion

ivity for 5 minutes, duringwas recorded,

of t(ie formal experiment, subjectscompleted a questionnaire in which those in the goalditions rated their percept ons of the attainability of a 40%increase in performance,goals were asked if they

iubjects in conditions withoutad spontaneously set any per-

formance goals for themselves and, if they had, to described not differ on the earlier filler

thetheir goals. The groups ditems on wl'hich they ratec their physical stamina and Itype and amount of physical activity they regularly performeach week.

Effects ofSelf-Judgn

suits

tents

To test for possitcording one's self-evspercepts of efficacy,randomly selected fcwith goals andtheir self-evaluativeof efficacy. Thecontained only themance was comparedwho also performedafter recording theirceived self-efficacy,significant differenceperformance change,two groups differ irthe postexperiment <self-judgments thus

Initial Effects of Goal Setting

As a first step tow a:treatment conditionformance half of th«of improving their p;rf<jects increased theiiwithout goals and 8!formance changes wore analyzed by a tanalysis of variancetors. To equalizetransformation wasThe analysis revealedfor goals, F(\, 76) =be shown below, thisbuiin performance mo'the interaction betwisignificant effect on

At the end of thesion, the appropriat;

1021

le reactive effects of re-lative reactions and self-he sample of 10 subjectsr this purpose performed

feed jack but did not recordreactions and self-percepts

questionnaire they completedller items. Their perfor-against that of subjects

with goals and feedbackself-satisfaction and per-

JThe analysis revealed nobetween the groups in

f(28) = .70. Nor did thesehow they responded to

iiuestionnaire. Recordinglad no reactive effects.

jd creating the requisites, after the baseline per-subjects received a goalbrmance by 40%. Sub-

performance level 42%with goals. These per-

;wo-wayivith goals and sex as fac-

vjariances, a square rootperformed on the scores,a significant main effect16.26, p< .001. As willinitial gain carried over,

t goals alone produced no further incrementivation. Neither sex nor

'Sen sex and goals had anychange in performance,second performance ses-groups received perfor-

mance feedback, all groups recorded their self-satisfaction and self-efficacy, and all groupsthen performed the effortful task again. In thesecond session the conditions comprising goalsand feedback and feedback alone exist inname only. It is the performance of subjectsin the final session, when all four treatmentconditions were fully formed and operating,that bears on the major hypotheses, to whichwe turn next.

Effects of Goal Setting and Feedback onEffortful Performance

To test how the structure of comparativefactors effects changes in level of motivation,the percentage increase in effortful perfor-mance above that of the second session wascomputed. In Figure 1 the mean percentageincreases are plotted as a function of treatmentconditions. Subjects who had the benefit ofboth goals and feedback more than doubledtheir performance over and above those sub-jects receiving either the goal alone, feedbackalone, or neither factor.

Baseline performance levels could have in-fluenced later performance change. An analysisof covariance was therefore computed, withtreatment conditions and sex as factors and

60

uiu< 50cc£o:

g30

iii 20

uio:oz 10

,'!'• 1 ' i

I" i

GOALS 1FEEDBACK

FEEDBACK CONTROL

Figure 1. Mean percentage increase in effortful perfor-mance under conditions varying in goals and performancefeedback.

Page 6: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

1022 ALBERT BANDURA AND DANIEL CERVONE

the first session performance as the covariate.A linear contrast showed that subjects in thecondition combining goals with feedback out-performed those in the other conditions, F(l,71) = 18.42, p < .0001, which did not differfrom each other. The same pattern of resultsis obtained if analysis of covariance is per-formed on performance scores in the thirdsession, with performance in the second sessionserving as the covariate. Goals with feedbacksurpassed the other conditions, F(l, 71) = 5.59;p < .025, which did not differ from each other.The analysis failed to yield any significant ef-fects on performance change for sex either in-dependently or in interaction with treatmentconditions.

Mechanisms Governing Motivation Effects

Mean levels of perceived self-efficacy andself-evaluation in the various treatment con-ditions are presented in Table 1. The self-pro-cesses did not differ in mean level, but thecondition combining a goal with feedback ofa substandard gain substantially increased thevariance of how self-satisfied subjects wouldbe with a similar future performance as com-pared both to feedback alone, F(\9, 19) =2.64, p < .05, and to the control condition,F(19, 19) = 2.15, p < .06. However, as hy-pothesized, these self-processes relate to per-formance motivation in strikingly differentways when the requisite comparative factorsare fully present than when they are presentonly partially or not at all.

Product-moment correlations were com-puted between the indices of the self-processesand percentage performance change. Degreeof self-dissatisfaction with the preceding per-formance and self-dissatisfaction if the samelevel of performance was attained in the next

session were each correlated with percentageof subsequent performance change. It is thepredictiveness of the second self-evaluativemeasure that is of greatest interest because itmore closely reflects the future performanceattainments subjects judge they must fulfill tofeel self-satisfied. The role of perceived self-efficacy as a performance motivator was eval-uated by correlating strength of perceived self-efficacy for a 40% goal attainment with percentof performance change. Correlational analyseswere conducted separately for each of the fourtreatment conditions based upon 18 degreesof freedom for each group. The complete setof correlations is presented in Table 2.

Complete Comparative Factors (Goal PlusFeedback)

Social learning theory posits that depend-able activation of self-evaluative mechanismsrequires both goals and performance feedback.Correlational analyses conducted on data fromthe condition combining goals with feedbackinformation indeed confirmed that self-dis-satisfaction is predictive of performancechange (Table 2). The more self-dissatisfiedsubjects were with the substandard perfor-mance they had just completed, the more theyheightened their next performance (r = .37,p = .05). Level of self-dissatisfaction with thesame substandard performance were it to oc-cur on the next session was even more pre-dictive of subsequent performance gains (r =.51,p = .01).

Perceived self-efficacy is also predictive ofthe performance changes exhibited by subjectswho had the benefit of goals and feedback.The more self-efficacious they were that theycould attain the 40% goal, the mpre highlythey boosted their next performance (r = .45,

Table 1Mean Levels of Perceived Self-Efficacy and Self-Evaluation in the Treatment Conditions

Self-process

Goal +feedback Goal Feedback

M SD M SD M SD

Control

M SD

Perceived self-efficacySelf-dissatisfaction

Prior performanceFuture performance

85.0

10.410.1

25.4

3.85.9

81.5

9.89.8

24.1

4.04.5

77.5

11.09.6

20.7

3.03.6

67.8

8.38.6

26.6

3.84.0

Page 7: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

SELF-REACTIVE MECHANISMS IN GOAL MOTIVATORS 1023

Table 2Relation of Self-Evaluation and Perceived Self-Efficacy to Effortful Performance Under Conditions inWhich the Requisite Comparative Factors of a Goal System Are Present or Absent _^

% change in effortful performance(fifth minute relative to first minute of

Session 3)

Self-process

% change in effortful performance (Session3 relative to Session 2)

G+F G F C G + F

Perceived self-efficacySelf-dissatisfaction

Prior performanceFuture performance

.45**

.37*

.51**

.57***

-.44*-.24

-.32

-.18-.59***

-.09

.27

.24

.43**

.36

.68****

.14

.03-.20

-.24

.08-.03

-.19

.22,

.24

Note. G = goal; F = feedback; C = control.* p = .05. ** p < .03. *** p < .01. **** p < .001.

p < .025). Correlations were also computedbetween performance effort and perceived self-eificacy to accomplish each of the 14 perfor-mance levels included in the efficacy scale.Subjects' perceived efficaciousness to attain the40% goal exceeded the correlations at each ofthe 13 remaining levels, a pattern that is highlysignificant (p < .001, as estimated by binomialtest). Self-satisfaction and percepts of self-ef-ficacy were not significantly related in any ofthe conditions.

It is postulated that self-dissatisfaction andperceived self-efficacy jointly determine per-formance changes. To test this, indices of self-dissatisfaction with future substandard per-formance and perceived self-efficacy for goalattainment were converted to standardized Tscores, combined, and correlated with per-formance change. This composite index of themediating self-processes was highly predictiveof subsequent performance change (r = .63,p < .002).

The joint operation of these two motiva-tional self-processes in the condition combin-ing an explicit goal with feedback is mostgraphically revealed by categorizing subjectsin terms of whether they expressed self-dis-satisfaction with future substandard perfor-mance and whether their perceived self-efficacyfor goal attainment exceeded the 50% strengthvalue. The mean percentage changes in per-formance as a function of varying combina-tions of these self-processes are presented inFigure 2. As may be seen in the left-hand panel,subjects who were both self-dissatisfied buthighly self-efficacious displayed huge perfor-mance gains. The self-inefficacious but self-

satisfied subjects manifested little performancechange. If at least one of the two self-processeswas strongly operative, subjects achievedmoderate performance gains. Because the ef-ficacious-dissatisfied and the inefficacious-satisfied subjects performed comparably, thesetwo groups were combined to increase thesubgroup size for statistical analysis. Eventhough the subgroups—formed by dichoto-mizing scores on the two self-processes—weresmall within this total sample of 20 subjects,the variations in effortful performance weresignificant, F(2, 17) = 4.10, p < .04.

.These two self-processes even predictchanges in performance motivation over thecourse of the session. For this temporal analysisthe difference in physical effort exerted be-tween the first minute and the last minute ofthe final session was computed and correlatedwith the indexes of the self-processes (Table2). Self-dissatisfaction with future substandardperformance (r = .68, p < .001) and self-ef-ficacy for goal attainment (r = .43, p < .03),both singly and in combination (r = .72, p <.001), predicted the degree of performancechange over time.

As shown in the right-hand panel of Figure2, the self-dissatisfied but self-efficacious sub-jects greatly accelerated their performance ef-fort, those who were either self-dissatisfied or.self-efficacious sustained their performance ef-fort, and those who judged themselves inef-ficacious to fulfill the goal and were satisfiedwith a 24% future increase slackened their ef-forts and showed a substantial decline in per-formance by the end of the session. Becausethe means for the high-efficacy-low-dissatis-

Page 8: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

1024 ALBERT BANDURA AND DANIEL CERVONE

IUU

S 90z

I80

£ 70UJQ.

D 60

U.

g 50u.u.UJ 40Z

S) 30

LUa. 20uz

-

-

-

-

-

-

-

-

-

>"'V ,)i ' ' .

! l '

'"

'r

i t i

,,i , ,

I'lft

i!1. 1 > pi" ii ii, , ! . , > ' ' ( : ' [,' "i1 ' !' '

1 [ ' [i ' '',! ,

. ' : !;

HIGH S-EFF.HIGH S-DIS.

LOW S-EFF.LOW S-DIS.

HIGH S-EFF.LOW S-DIS.

LOW S-EFF.HIGH S-DIS.

QW

Z

1 50UJ</1 40

O

1 30

D0 20UJo< 10I0 0UUloz -10Zg - 2 0

0:UJ -30

-

-

-

_

-

''•: .•!

', '., i" JL"

' "'^'"i !

!~Jv!i r- "i i r r 11 ' \ I"'-:.?) I

-

-

HIGH S-EFF. LOWS-EFF. HISH S-EFF. LOW S-EFF.HIGH S-DIS. LOW S-DIS. LOW S-DIS. HIGH S-DIS.

Figure 2. Mean percentage change in effortful performance by subjects in the goals and feedback conditionas a function of differential combinations of levels of self-dissatisfaction (S-DIS) and perceived self-efficacy(S-EFF) for goal attainment. (The left-hand panel shows the mean change for the entire session; the right-hand panel shows the mean change between the first and final minute of the session. The number of subjectsin each of the four combinations of self-processes was as follows: High S-EFF-High S-DIS, 4; Low S-EFF-Low S-DIS, 6; High S-EFF-Low S-DIS, 6; Low S-EFF-High S-DIS, 4.)

faction and the low-efficacy-high-dissatisfac-tion subgroups were virtually identical and thens were small, these two subgroups were com-bined for the ANOVA. These differential pat-terns of motivational change are highly sig-nificant, F[2, 17) = 14.73, p < .001.

Of the subjects in the inefficacy-satisfiedsubgroup, 83% perceived the selected goal asvirtually unattainable, whereas only 7% of theremaining subjects considered the goal beyondtheir reach. These differential perceptions ofgoal attainability, which can markedly affectgoal adherence, are highly significant (z = 3.45,p < .001).

Partial Comparative Factors

Subjects who performed with either goalsalone or feedback alone lacked one of the crit-ical elements to regulate their effort effectivelythrough self-reactive influence unless theysupplied the missing element subjectively.Correlational analyses indeed reveal that theconstituent self-reactive influences operate dif-ferentially in performance motivation de-pending on which comparative factor islacking.

Feedback-alone condition. Subjects in thefeedback-alone condition were informed thatthey had improved by 24% but had no explicitgoal to judge whether the gain was exemplaryor insufficient. The extent to which they wereself-satisfied with this performance gain wasunrelated to subsequent performance change,but the more satisfied they were with main-taining a similar future gain, the greater wastheir performance (r = —.59, p < .01, two-tailed). Enhanced effort is thus related to sat-isfaction under feedback of progress alone butto discontent when that same level of progressis viewed in relation to a seemingly difficultstandard of a 40% gain.

When people engage in an ongoing activityand are periodically informed of their perfor-mance attainments, some spontaneously setgoals for themselves (Bandura & Simon, 1977).In the feedback-alone condition 70% of thesubjects indeed reported in the postexperimentquestionnaire that they set performance stan-dards for themselves on their own. Those whoset no goals for themselves achieved no change(.4%), those who aimed to sustain their im-provement realized a modest gain (27%), andthose who set themselves the more demanding

Page 9: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

SELF-REACTIVE MECHANISMS IN GOAL MOTIVATORS 1025

goal of bettering their improvement raisedtheir level of performance substantially (40%).This pattern of differences is significant evenwith the limited sample size, F(2, 16) = 3.51,p = .05. In contrast, self-set goals had no effecton performance effort in the condition inwhich subjects received no performance feed-back.

Whether knowledge of a 24% increase inperformance is indicative of self-efficacy orself-inefficacy depends on what goals, if any,subjects set for themselves as suitable markersof capability. Because subjective goal settingvaried widely, perceived self-efficacy for a 40%goal attainment bore no significant relation toperformance change in the feedback-alonecondition.

Goal-alone condition. In the goal-alonecondition, subjects aimed for a 40% increasein performance but had no objective knowl-edge of how they were doing. The stronger thesubjects' self-efficaciousness that they couldattain such a goal, the greater was their per-formance change (r = .57, p < .005). The cor-relation of perceived self-efficacy to attain the40% goal with performance effort was higherthan the correlations at all but one (60% goal)of the 13 levels of goal attainment. This patternis highly significant (p < .002, as estimated bythe binomial test). In this condition, in whichsubjects had only their subjective performanceimpressions to go on, it was self-satisfactionwith their imagined attainment that was re-lated to subsequent performance change (r =—.44, p < .05, two-tailed), but degree of self-evaluation for a similar future performancewas not.

No Comparative Factors (No Goal,No Feedback)

In the condition in which subjects lackedboth goals and knowledge of how they per-formed, they had little basis for either ap-praising or regulating their ongoing perfor-mance effort. Self-processes were unrelated toperformance change (see Table 2).

Discussion

The findings of the present study supportthe theory that goal systems gain motivatingpower through self-evaluative and self-efficacy

mechanisms activated by cognitive compari-son. Goals enhanced performance effort onlyunder conditions combining a personal stan-dard with performance feedback of progresstoward it. Neither goals alone nor performancefeedback alone, both of which lack an essentialcomparative ingredient, effected change inmotivational level. When first adopted, goalsalone produced a performance gain that car-ried over, but they did not generate any furtherincrements in motivation in the absence ofperformance knowledge. Although goals alonedid not further augment performance moti-vation, persistence of the initial boost enabledthe subjects who performed with goals aloneto surpass the controls. The pattern of resultsat the point at which all conditions were fullyoperative is consistent with Locke's goal theory(Locke etal., 1981).

Analysis of performance effort as a functionof self-set standards under conditions of feed-back alone is also in accord with the view thatboth performance knowledge and a standardof comparison are needed to produce moti-vational effects. For the most part, the self-prescribed goals were of a general sort and notunduly challenging. Nevertheless, participantswho set no goals were outperformed by thosewho set themselves the goal of sustaining theirperformance gain, who, in turn, were outper-formed by those who sought to better theirpast attainment. These goals were of a morequalitative sort of sustaining or surpassingone's prior accomplishments (e.g., "I want todo as well as the time before . . . do betterthan the previous session.") rather than statedin terms of explicit quantitative levels of per-formance change. Although such goals hadmotivating potential in the context of perfor-mance knowledge, the rise in performance ef-fort for feedback with qualitative self-set stan-dards was less than that for the same feedbackwith a challenging (40%) quantified standard.These findings are congruent with those ofprevious studies showing that explicit goalsare more motivating than are general ones(Latham & Yukl, 1975; Locke, 1968). Self-setgoals had no motivating potential withoutperformance information in the control con-dition.

Results of the correlational analyses supportthe proposition that goal systems affect per-

Page 10: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

1026 ALBERT BANDURA AND DANIEL CERVONE

formance motivation in part through self-evaluative and self-efficacy mechanisms.Moreover, the findings shed interesting lighton how these self-mechanisms operate in per-formance motivation when only partial com-parative information is available. When per-formance information is combined with astandard of comparision, the higher the self-dissatisfaction with a substandard performanceand the stronger the perceived self-efficacy forgoal attainment, the greater is the subsequentintensification of effort. In a recently com-pleted study employing a path analysis, Lockeand his colleagues (Locke, Frederick, Lee, &Bobko, Note 1) found that perceived self-ef-ficacy affects the level of self-set goals, strengthof goal commitment, and level of cognitiveperformance.

When one of the requisite comparative fac-tors is lacking, the relation of self-reactive in-fluences to performance motivation dependson the nature of the partial information pro-vided or that performers fashion for them-selves. Thus, in the condition providing onlyfeedback, knowledge of a 24% gain in perfor-mance carried no absolute value. It repre-sented a commendable accomplishment ifjudged against subjective modest aspirationsbut a failure if evaluated against subjectivelyinvoked high standards. Subjects' reports oftheir self-set goals reveal that many of themeither set no goals for themselves (45%) oraimed for the same level of performance gain(25%). For subjects in this condition, a 24%gain constituted positive or success feedback.The more pleased subjects were with sustainingthis level of improvement, the more effortfullythey behaved. In contrast, a 24% gain whenone is aiming for a 40% increment constitutesnegative or failure feedback. Discontent withthe prospect of similar failure in the futurespurred subjects to greater effort. Invertedmeaning of the performance feedback thusproduces inverse relations between self-eval-uation and performance motivation.

The findings that self-evaluative reactionsoperated differently on motivation undervarying comparative structures testify to thecomplexity of the relation between self-satis-faction and motivation. With goals and per-formance feedback, self-dissatisfaction affectseffort (see also Locke, Cartledge, & Knerr,1970), whereas with either goals alone or feed-

back alone, effort seems to be governed bylevel of self-satisfaction.

Variable self-prescribed standards similarlyconfer diverse self-efficacy value on the sameperformance gain. Subjects oriented towardsustaining their level of effortful performanceare likely to raise their self-percepts of efficacyon learning that they surpassed their pastachievements by 24%, whereas those who askmuch of themselves might interpret the sameperformance gain as a sign of physical inef-ficacy. Perceived self-efficacy in attaining a 40%goal therefore bore no consistent relation tosubsequent effort in the feedback-only con-dition. However, in the goal-alone condition,in which all performers aimed for the samechallenging standard but had to guess how theywere doing, the stronger their perceived self-efficaciousness for goal attainment and themore pleased they were with whatever theysurmised their prior performance to be, themore they heightened their effort.

Self-reactive influences are least likely to beactivated in any consistent way in differentindividuals by conditions providing neitherevaluative standards, performance informa-tion, nor even distinct implicit feedback con-cerning the level of performance. Not sur-prisingly, the latter condition, which was de-void of information for monitoring, gauging,and regulating one's effort, yielded no signif-icant correlates.

In the present study, predicted relations weretested under conditions in which performancewas moderately discrepant from personalstandards. The informative next stage for re-search is to clarify further the precise natureof the relation between self-percepts of efficacy,self-evaluation, and performance motivationwhen attainments diverge from personal stan-dards across a wide range of positive and neg-ative magnitudes. Recall that social learningtheory postulates a linear positive function be-tween perceived self-efficacy for goal attain-ment and effort but a nonlinear one betweendegree of goal discrepancy and effort, as me-diated through self-evaluative reactions.

Theorists working within the framework ofachievement motivation have addressed theissue of task difficulty mainly in terms of suc-cess expectancy and valuation of goal attain-ment (Atkinson & Raynor, 1974; Feather,1982). Because these two factors are considered

Page 11: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

SELF-REACTIVE MECHANISMS IN GOAL MOTIVATORS 1027

to be inversely related, performanee is highestfor tasks of moderate difficulty, althoughHeckhausen (1977) posits a somewhat lowersuccess probability than does Atkinson asbeing maximally motivating. In the goal theorydeveloped by Locke (1968), performance islinearly related to goal difficulty as long asperformers continue to adhere to taxing goals.By expanding the self-process probes to includerevision of goals, as well as perceived self-ef-ficacy and self-satisfaction with prior and fu-ture attainments, the paradigm developed forthe present experiment might help shed someadditional empirical light on subprocessesmediating effort when attainments divergefrom standards in degree and direction.

Performances that fall markedly short ofstandards are likely to be demotivating by un-dermining perceived self-efficacy. To the extentthat performers judge the standard as exceed-ing their capabilities, they are apt to lowertheir standard and demand less of themselves.Such adjustments would lower effort and per-formance (Feather, 1982; Locke, 1968). Of in-terest is the threshold strength value belowwhich reduced self-efficaciousness results ingoal abandonment. As already noted, mod-erately discrepant attainments heighten mo-tivation to fulfill standards that appear attain-able through extra effort. When dissatisfactioncombines with self-efficaciousness, effort ismobilized to master the challenge. The thirdpattern of interest concerns attainments thateither fall just short of challenging standardsor exceed them. High accomplishments thatstrengthen perceived self-efficaciousness arelikely to lead performers to raise their stan-dards (Lewin, Dembo, Festinger, & Sears,1944), thus creating new motivating discrep-ancies for themselves. Under the latter cir-cumstance, the pattern of self-subprocess reg-ulation of motivation would include satisfac-tion with prior attainments—but discontentwith similar future ones—high self-effica-ciousness, and raised aspirations.

Depending on their direction and magni-tude, goal discrepancies can raise the moti-vational potential of one of the self-reactivefactors while lowering the motivational po-tential of the other. Thus, large negative dis-crepancies increase self-dissatisfaction butlower perceived self-efficacy for goal attain-ment. However, a decrement in self-efficacy

sufficient to prompt adoption of a lower stan-dard would serve to moderate evaluative self-reactions. For example, when provided withfeedback of a substandard performance, someof the subjects in the present study seemed toabandon their goal as unattainable and wereno longer unduly self-dissatisfied with mod-erate progress. Smaller negative discrepanciesreduce self-dissatisfaction but strengthen self-percepts of efficacy. Further research is neededto determine the relative susceptibility of thesetwo self-reactive factors to failed efforts andhow they may combine and compensate foreach other as motivators of action.

Social learning theory distinguishes betweenthe effects of strength of perceived, self-efficacyon effort during learning and during executionof established skills (Bandura, 1982): In ap-proaching learning tasks, persons who perceivethemselves to be supremely self-efficacious inthe undertaking may see little need to investmuch preparatory effort in it (Salomon, inpress). However, in applying acquired skills,strong belief in one's self-efficaciousness in-tensifies and sustains the effort needed to re-alize challenging goals, which are difficult toattain if one is plagued by self-doubts. In short,self-doubts create an impetus for learning buthinder adept use of established skills.

Reference Note

1. Locke, E. A., Frederick, E., Lee, C., & Bobko, P. Theeffect of self-efficacy, goals and task strategies on taskperformance. Unpublished manuscript, University ofMaryland, 1982.

References

Atkinson, J. W, An introduction to motivation. Princeton,N.J.: Van Nostrand, 1964.

Atkinson, J. W., &-Raynor, J. O. Motivation and achieve-ment. Washington, D.C.: V. H. Winston, 1974.

Bandura, A. Social learning theory. Englewood Cliffs, N.J.:Prentice-Hall, 1977. (a)

Bandura, A. Self-efficacy: Toward a unifying theory ofbehavioral change. Psychological Review, 1977,54, 191-215. (b)

Bandura, A. The self system in reciprocal determinism.American Psychologist, 1978, 33, 344-358.

Bandura, A. The self and mechanisms of agency. In J.Suls (Ed.), Psychological perspectives on the self (Vol.1). Hillsdale, N.J.: Erlbaum, 1981.

Bandura, A. Self-efficacy mechanism in human agency.American Psychologist, 1982, 37, 122-147.

Bandura, A., & Schunk, D. H. Cultivating competence,self-efficacy, and intrinsic interest through proximal self-

Page 12: Self-evaluative and self-efficacy mechanisms governing the motivational effects of goal systems

1028 ALBERT BANDURA AND DANIEL CERVONE

motivation. Journal of Personality and Social Psychology,1981, 41, 586-598.

Bandura, A., & Simon, K. M. The role of proximal in-tentions in self-regulation of refractory behavior. Cog-nitive Therapy and Research, 1977, /, 177-193.

Becker, L. J. Joint effect of feedback and goal setting onperformance: A field study of residential energy con-servation. Journal of Applied Psychology, 1978, 63,428-433.

Brown, I., Jr., & Inouye, D. K. Learned helplessnessthrough modeling: The role of perceived similarity incompetence. Journal of Personality and Social Psy-chology, 1978, 36, 900-908.

Feather, N. T. (Ed.). Expectations and actions: Expectancy-value models in psychology. Hillsdale, N.J.: Erlbaum,1982.

Heckhausen, H. Achievement motivation and its con-structs: A cognitive model. Motivation and Emotion,1977, 1, 283-329.

Kiesler, C. A. The psychology of commitment: ExperimentsUnking behavior to belief. New York: Academic Press,1971.

Langer, E. J. The illusion of control. Journal of Personalityand Social Psychology, 1975, 32, 311-328.

Latham, G. P., & Yukl, G. A. A review of research onthe application of goal setting in organizations. Academyof Management Journal, 1975, 18, 824-845.

Lewin, K., Dembo, X, Festinger, L., & Sears, P. S. Levelof aspiration. In J. McV. Hunt (Ed.), Personality andthe behavior disorders (Vol. 1). New York: Ronald Press,1944.

Locke, E. A. Toward a theory of task motivation and in-centives. Organizational Behavior and Human Perfor-mance, 1968, 3, 157-189.

Locke, E. A., Cartledge, N., & Knerr, C. S. Studies of therelationship between satisfaction, goal setting, and per-formance. Organizational Behavior and Human Per-formance, 1970, 5, 135-158.

Locke, E. A., Shaw, K. N., Saari, L. M., & Latham,G. P. Goal setting and task performance: 1969-1980.Psychological Bulletin, 1981, 90, 125-152.

Salomon, G. Television is "easy" and print is "tough":The differential investment of mental effort in learningas a function of perceptions and attributions. Journalof Educational Psychology, in press.

Schunk, D. H. Modeling and attributional effects of chil-dren's achievement: A self-efficacy analysis. Journal ofEducational Psychology, 1981, 73, 93-105.

Steers, R. M., & Porter, L. W. The role of task-goal at-tributes in employee performance. Psychological Bul-letin, 1974, 81, 434-452.

Strang, H. R., Lawrence, E. C., & Fowler, P. C. Effects ofassigned goal level and knowledge of results on arithmeticcomputation: Laboratory study. Journal of Applied Psy-chology, 1978, 63, 446-450.

Weinberg, R. S., Gould, D., & Jackson, A. Expectationsand performance: An empirical test of Bandura's self-efficacy theory. Journal of Sport Psychology, 1979, 1,320-331.

Received August 27, 1982Revision received April 21, 1983 •