the effects of contingent feedback on perceived control and performance

12
European Journal ofPs of Education 1995, Vol. X, n" I, 13-24 © 1995, I.S.P.A. The Effects of Contingent Feedback on Perceived Control and Performance Frans Siero Jan Pieter van Oudenhoven University of Groningen, The NetherLands The purpose of this study was to examine the impact on students' performance of three contingent feedback strategies used by teachers. Contingency means that the feedback strongly corresponds with task behaviour that can be controlled by the students. Elementary school students (N=296) received individuali zed feedback about their performance during a series of five lessons. Within this contingent feedback structure. three strategies were applied which were assumed [0 enhance the perceived controllability of the task situation: (1) enhancing the perceptibility of the contingency between feedback and task behavior; (2) explicit reference to effort as part of the feedback. and (3) setting of goals. As predicted, the three contingent feedback strategies had a significant positive effect on perceived controllability, and led to better task performance. It has long been recognized that the feedback teachers give to students influences students' achievement. Depending on whether the feedback is positive or negative, a student may be encouraged or discouraged, sometimes to the point of becoming an underachiever. An important aspect of feedback is the relationship between the judgment and those elements in the task behaviour which are under the student's control, such as effort. When this correspondence is strong, i.e, judgments covary with controllable aspects of the task behaviour, judgments are called contingent. Is it possible to improve the performance of elementary school children by enhancing this contingency of judgments? The purpose of this study was to answer that question. The criterion used by the teacher to judge performance is essential for the degree of contingency. In the case of a group-based evaluation, performance is assessed in relation to other students' performance ; a performance above a specified level (e.g., average) is viewed more positively than a performance below that level. If feedback is individualised , achievement is judged in relation to the student's past performance: improvement results in a positive judgment, deterioration results in a negative or less positive judgment. Intrapersonal comparison of performances instead of interpersonal comparison diminishes differences in abilities and enhances the importance of effort. Consequently, if feedback is individualised, contingency between judgment and task behaviour is stronger. Contingent judgments given to a student are subject to fluctuations depending on the

Upload: rug

Post on 10-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

European Journal ofPs)'ch(}I(}~y of Education1995, Vol. X, n" I, 13-24© 1995, I.S.P.A.

The Effects of Contingent Feedback on PerceivedControl and Performance

Frans SieroJan Pieter van OudenhovenUniversity ofGroningen, The NetherLands

The purpose of this study was to examine the impact on students'performance of three contingent feedback strategies used by teachers.Contingency means that the feedback strongly corresponds with taskbehaviour that can be controlled by the students . Elementary schoolstudents (N=296) received individualized feedback about theirperformance during a series of five lessons . Within this contingentfeedback structure. three strategies were applied which were assumed[0 enhance the perceived controllability of the task situation: (1)enhancing the perceptibility of the contingency between feedback andtask behavior; (2) explicit reference to effort as part of the feedback.and (3) setting of goals. As predicted, the three contingent feedbackstrategies had a significant positive effect on perceived controllability,and led to better task performance.

It has long been recognized that the feedback teachers give to students influencesstudents' achievement. Depending on whether the feedback is positive or negative, a studentmay be encouraged or discouraged, sometimes to the point of becoming an underachiever. Animportant aspect of feedback is the relationship between the judgment and those elements inthe task behaviour which are under the student's control, such as effort. When thiscorrespondence is strong, i.e, judgments covary with controllable aspects of the taskbehaviour, judgments are called contingent. Is it possible to improve the performance ofelementary school children by enhancing this contingency of judgments? The purpose of thisstudy was to answer that question .

The criterion used by the teacher to judge performance is essential for the degree ofcontingency. In the case of a group-based evaluation, performance is assessed in relation toother students' performance ; a performance above a specified level (e.g., average) is viewedmore positively than a performance below that level. If feedback is individualised ,achievement is judged in relation to the student's past performance: improvement results in apositive judgment, deterioration results in a negative or less positive judgment. Intrapersonalcomparison of performances instead of interpersonal compari son diminishes differences inabilities and enhances the importance of effort. Consequently, if feedback is individualised,contingency between judgment and task behaviour is stronger.

Contingent judgments given to a student are subject to fluctuations depending on the

14 F. SIERO & 1. P. VAN OUDENHOVEN

effort that the student put into the task. In addition to these intrapersonal alterations inoutcomes, there is hardly any correspondence between the results achieved by those of fellowstudents. Hence, contingency between judgments and task behaviour results in outcomeinformation with a low temporal consistency (alternate pattern) and neutral consensus (lowcorrespondence with outcome information of other students). The combination of these twoprinciples increases the probability that the judgment is ascribed to a unstable cause, such aseffort and luck (e.g., Forsyth & McMillan, 1981a; Hewstone & Jaspers, 1983, 1987; Weiner,1982, 1985). Owing to the covariation between judgment and effort, particularly effort (beingunstable, internal, and controllable) will be perceived as the main cause.

Attributions to controllable causes contribute to the willingness to exert oneself nexttime. According to Forsyth and McMillan (1981 b), people estimate the probability of apositive outcome higher when they attribute their outcomes to controllable causes. Forsythand McMillan also indicate a positive relationship between perceived controllability ofoutcomes and affective reactions after success and failure: the more people perceive anoutcome as controllable, the more they appraise the outcome in a positive way. Therefore,attributions of outcomes to controllable causes go together with high outcome expectationsand high outcome evaluations. Both elements, outcome expecations and outcome evaluations,should motivate a person to show greater effort and achievement improvement.

As a starting-point for all experimental conditions in this study, an evaluation method wasdeveloped in which the judging of performances is based on the achievement improvementand achievement deterioration of the individual student. Other elements were added to thisindividualised feedback. It was expected that these elements would also emphasize therelationship between judgment and task behaviour. On the basis of practical considerations,three strategies were applied: (1) enhancing the perceptibility of the contingency betweenfeedback and task behavior; (2) explicit reference to effort as part of the feedback, and (3)setting of goals. Effective use of these strategies should result in more perceivedcontrollability and higher achievement. The central variable is perceived controllability,specified as the subjective contingency, derived by students from their experiences with thetask and judgment. High perceived controllability means that a person attributes judgmentsabout performance mainly to factors which he can influence such as effort and to a lesserdegree to factors which are not under his control, such as ability, luck, and task difficulty.

Perceptibility ofcontingency

To discern a relationship between task behaviour and judgment, the contingency has tobecome sufficiently visible. Several studies (e.g., Alloy & Tabachnik, 1984; Crocker, 198I;Kelley, 1973; Nisbett & Ross, 1980) have shown that people are often not able to perceive theactual covariation between behaviour and outcomes. The gap between behaviour and outcomecan be too large. Similarly, a cognitive barrier may prevent them from noticing thecontingency. For example, a person may be used to applying fixed causal schemata (such as "Ifail exams because I'm too stupid") as a result of which he/she expects no covariation betweeneffort and achievement. Therefore, the effort made by the student has to be a manifestlyrecognizable component of the judgment. Hence, the first strategy implies that the visibility ofthe contingency between task behaviour and judgment is increased. As a result of receivingfeedback of which the contingency is more visible, students will feel more in control and willachieve better than students who receive the 'standard' individualised feedback.

Reference to effort

In addition to enhancing the visibility of the contingency, it is also possible to alertsomeone to more controllability in a more direct way, namely, by pointing to the causalsignificance of effort for the performance and the judgment. For that purpose, a remark aboutthe sufficiency or insufficiency of the effort shown during the task is added to the positive and

CONTINGENT FEEDBACK 15

negative feedback respectively. Direct information about the role of effort should encouragestudents to perceive more controllability and to try harder.

Goal setting

The two strategies mentioned above are interventions with respect to the presentation ofthe judgment. Goal setting is an intervention at the task level. Directing a person to a task goalis a strategy by which the control perceptions about the task stituation can be enhanced. Anumber of researchers have shown that goal setting can result in better performance (for areview, see: Locke & Latham, 1990; Mento, Steel, & Karren, 1987). However, goal setting isonly effective when a number of conditions are fulfilled. Firstly, goal setting has to beaccompanied by feedback about task behaviour so that a person is able to adapt his behaviourwhen he does not reach his task goal (Becker, 1978). Secondly, goals may be high, but theyhave to be feasible for a person, by virtue of his capabilities (Locke & Latham, 1990; Garland,1982). Thirdly, a person has to direct his effort to a specific goal. The more specific the goal,the more goal setting should result in better performances (e.g., Jackson & Zedeck, 1982).

The setting of goals was made part of the method of individualised feedback. By doingso, the specified requisites about feedback, feasability, and specificity are reasonably satisfied.The imposed goal is nothing more than the presentation of a specific amount of progress thatthe student should make. By comparing the judgment with the goal, the student can find outwhether the goal has been attained. The task goals are adapted to the individual level of thestudent. Therefore, each student is able to realise the imposed goal. It was expected that thepresence of a attainable goal amounts to the controllability of the task situation andconsequently encourages students to improve their performance (Bandura & Wood, 1989). Inthe literature on goal setting, the role of perceived controllability is not explicitly dealt with.This study provides an opportunity to examine this relationship directly.

Direction of the judgment

Generally, people are inclined to attribute success to internal causes (ability, effort) andfailure to external causes such as task difficulty or chance factors ("attribution bias": Weiner,1985; Zuckerman, 1979). In this study, success and failure refer to achievement improvementand achievement deterioration respectively. The dependence of the judgment on individualchanges in task behavior will induce the student to take more responsibility for negative,controllable outcomes and will also lead to a more solid causal connection between failure andeffort. Consequently, the more the judgment in the three experimental conditions is associatedwith changes in task behaviour, the more failure will be attributed to lack of effort.

Hypotheses

To test the effects of contingency and control, four judgment conditions were realised. Inthese conditions the relationship between judgment and controllable components of the taskbehaviour was made cumulatively more explicit. In the first condition, judgments referred toachievement improvement and achievement deterioration. Students received a judgmentwhich was, because of the comparison with previous performance, equally contingent for allstudents. In the second condition, judgments were based on improvement or deterioration ofachievement as well, but the visibility of the contingency was enhanced. The third conditionwas identical to the second condition except that remarks about effort were added to thefeedback. In the fourth condition, goal setting was combined with the elements of the thirdcondition. The following hypotheses were tested:

I) Enhancing the visibility of the contingency of judgments leads to higher perceivedcontrollability (Hypothesis 1a) and higher achievement (Hypothesis 1b): Condition IIversus Condition I.

16 F. SIERO & 1. P. VAN OUDENHOVEN

2) Making references to effort in addition to the enhanced visibility of contingency leadsto even higher perceived controllability (Hypothesis 2a) and even higher achievement(Hypothesis 2b): Condition III versus Condition II.

3) The combination of goal setting with enhanced visibility of contingency and referencesto effort will lead to the highest achievement: Condition IV versus Condition III.

Method

Subjects and task

Thirteen randomly selected elementary schools (296 fifth grade students, age 10-11)participated in the study. All schools made use of a group-based system of judgingperformances. The 13 schools were randomly assigned to the four conditions. The number ofstudents in conditions I, II, III" and IV were 76, 79, 63, and 78 respectively.

During five lessons, the students worked on a relatively new topic that had been used in aprevious study (Siero, Van Oudenhoven & Veen, 1983), i.e., a unit on the social history ofNorthern America. The lessons were given by a teacher who was appointed especially for thatpurpose. All the lessons had the same set-up and structure and were, with the exception of thefifth lesson, completed with a multiple choice test. The lessons, which lasted about fiftyminutes, were given within a period of two weeks.

Judgment conditions

As indicated above, the feedback in the four experimental conditions consisted ofjudgments about individual achievement improvement or achievement deterioration. Thestudents received points which were based on a comparison of their score on the test with theirscore on the preceding test or, when possible, on the two preceding tests. The calculation ofthe differences between the scores on the equally difficult tests always yielded points which,in Condition J, varied between --15 (deterioration) and + I5 (improvement). This range wasbased on the test results in Siero et al. (1983).

The feedback was presented as a "mini-report" in which the improvement or deteriorationafter each test was indicated. The meaning of the number of points received was specifiedwith labels, such as "strongly backwards" ... "strongly forward". An example of a feedbackform is presented in Figure I.

Stronglybackwards

-30 -20

Backwards

-10

A little bitbackwards

o

A little bitforward

X8 points

+10

Forward

+20

Stronglyforward

+30

Figure J. Example of a feedback form (comparison of the performance during lesson 2 withthe performance during lesson I)

In Condition II the visibility of the contingency was enhanced by assigning three times asmany points for a particular improvement or deterioration. The student who received animprovement score of eight points ("a little forward") in Condition I (see Figure 1), receivedan improvement score of 24 points ("strongly forward") for the same improvement inCondition fl. Consequently, the number of points in this condition could vary between -45and +45. In the case of an extreme change (more than +30 points or less than -30 points), thepoints were set outside the row with labels.

CONTINGENT FEEDBACK 17

The enhanced visibility of the second condition was also applied in Condition Il/.Furthermore, in this condition the causal relationship between the effort expended and thefeedback received was emphasized by including a written remark in the report (e.g., "try to dobetter" in the case of an achievement deterioration). In Condition IV, students received, inaddition to the elements of the third condition, a task goal. The teacher informed the studentthat he/she should be able to realise a particular improvement at the end of the lessonconcerned. The teacher mentioned that the task goal was based on the student's testperformance: in the previous lesson. Actually, the goals were randomly chosen; they variedbetween 10 points and 20 points improvement, registered in our previous study (Siero et aI.,1983). In addition, the goal was made clear to the student with the help of a figure. The taskgoal information was given in an extra page, which was added to the report.

Dependent variables

Perceived controllability. This variable was assessed by asking students to attribute theiroutcome (i mprovement or deterioration) to ability, effort, luck, and task difficulty.Attributions to effort and luck were considered as indicators of perceived controllability. Thefour causal factors were offered in six pairs, the order of which was varied. The two causalfactors of each pair were the anchors of a four-point scale. The subject had to indicate on thisscale to what extend the judgment was caused by the one factor relative to the other factor.Depending on the judgment, each pair of causal factors began with the text "I was goingbackwards because ...." or "I was going forward because ...." On the basis of the responses onthe six four-point scales, a percentage score was calculated for each of the four causal factors.

Achievement. To assess achievement, the students were tested at the end of each of thefirst four lessons. Each test contained 20 forced choice items concerning the content of thelesson. Only the third and fourth test were used as effect variables because the feedbackstarted at the beginning of the third lesson.

Procedure

The subject of the series of five lessons was introduced in the first lesson. The teacherannounced that the students had to answer questions about the lessons. Assessment ofachievement took place at the end of this lesson. At the beginning of this test (and thebeginning of the three other tests) the teacher mentioned explicitly that the four tests wereequally difficult.

At the beginning of the second lesson, the most difficult items from the first test werediscussed. The teacher announced that each student would receive a combined evaluation ofthe test results of the first and the second lesson at the start of the third lesson. To accustomthe students to the unusual judgments about achievement improvement and deterioration, theteacher clarified the feedback form and the system of backwards points and forward points. Atthe end of this lesson, measurement of the second test took place.

The third and fourth lesson were structured identically. After the discussion of the resultsof the preceding test, the students were told how to fill out the attribution items. Then theteacher once again clarified the feedback form briefly, and distributed the attributionquestionnaire and the report, i.e., the feedback about the improvement or deterioration fromlesson I to lesson 2 (feedback third lesson) or from lesson 2 to lesson 3 (feedback fourth lesson).Next, the students had to fill out their outcome, the number of points backwards or forward, onthe first page of the attribution questionnaire. Moreover, the students in Condition III and IVhad to fill in the remark about their effort on the same page. The teacher asked all students tomemorize the judgment as they filled in the questionnaire. The students in Condition IV, whoreceived a report with a extra goal setting page, were told that the task goal for each student wasbased on the test results of the preceding lesson. Actually, the height of the goal was chosenrandomly. After the content of the lesson was discussed, students had to do the achievement test.

18 F. SIERO & 1. P. VAN OUDENHOVEN

At the beginning of the fifth, last lesson, the students received feedback about theirimprovement or deterioration from lesson 3 to lesson 4. Afterwards, the teacher rounded offthe lecture about the social history of Northern America. The five lessons, which each tookabout 75 minutes, were given within two weeks at intervals of two/three days. A secondperson assisted the teacher and, in addition, watched to see whether the teacher followed thescript specified beforehand during the implementing of the judgment conditions and lessons.

Analysis

The model for the testing of the effects contains two between-group factors: the judgment(2 levels, forward and backwards) and contingency (4 levels). To test the cumulative effect ofeach judgment strategy, the difference between two subsequent conditions was analysed:Condition II versus Condition I for the effect of visibility of the contingent judgments,Condition III versus Condition II for the effect of "Reference to effort", and Condition IYversus Condition III for the effect of setting of goals. The data after the first and after thesecond judgment were analysed separately.

The responses on the attribution items were analysed by means of MANDYA'.ANCDYA was used to analyse the effects on the achievement scores. The scores on the firstand the second achievement test acted as covariates.

Results

Perceived controllibility

Visibility (comparison of Condition I and Il). In Hypothesis I a, it was predicted thatenhancing the visibility of the contingency of judgments should lead subjects to perceive morecontrol over their outcomes (more effort attributions, fewer luck attributions). The resultssupport this hypothesis. Mean attributions are given in Table I (after the first judgment) andTable 2 (after the second judgment). InTable 3, F-values for the test of this hypothesis and allthe subsequent hypotheses are summarized.

Table IMean attributions ofcausality within the four judgment conditions and within the two feedbackconditions, after the first judgment (lesson 3)

Attribution of outcome to

Judgment conditionsLow visibilityHigh visibility

Directing of attributionsSetting of goals

Direction of judgmentBackwardsForward

Noles. Range of attribution scores: 10-40.

Effort23.723.9

25.326.7

19.830.0

Luck

26.424.4

22.821.4

25.821.6

Ability

21.623.9

25.524.9

24.623.4

Task difficulty

28.127.826.427.0

29.825.0

After the first judgment, students in the strong visibility condition attributed their outcometo luck to a lesser degree than students in the weak visibility condition (P<.001). Students inthe two conditions did not differ in their attributions of this first judgment to effort.

After the second judgment, the attributions to effort were also in the predicted direction:Students who received a judgment in the strong visibility condition attributed their outcome toeffort to a higher degree and to luck to a lesser degree than students with a judgment in the weakvisibility condition (for both differences: p<.Ol). The mean attribution scores are shown in Table 2.

CONTINGENT FEEDBACK 19

The effects of the two contingency conditions on effort and luck were also supported bythe multivariate test results: p<.OOI after the first judgment, and p<.OI after the secondjudgment (see Table 3). For the rest, students attributed their outcome in the strong visibilitycondition to a higher degree to ability (M=23.9) than students in the weak visibility condition(M=2 1.6; p<.Ol). This unpredicted difference was absent after the second judgment.

Directing of attribution (comparison Condition II and 1II). Making references to effort inaddition to the enhanced visibility of contingency should lead to an even higher perceivedcontrollability (Hypothesis 2a). The means in Table I and Table 2 support this hypothesis.After both the first judgment (p<.OI) and the second judgment (p<.05), the addition ofreference to effort enhanced the causal attribution of the outcome to effort.

Table 2Mean attributions ofcausality within the four judgment conditions and within the two feedbackconditions, after the second judgment (lesson 4)

Attribution of outcome to

Judgment conditions

Low visibilityHigh visibilityDirecting of attributionsSetting of goals

Effort

23.925.926.327.8

Luck Ability

26.7 22.124.5 22.925.2 22.822.9 23.1

Task difficulty

27.326.725.726.2

Direction of judgmentBackwardsForward

Nilles. Range of attribution scores: 10-40.

21.430.6

26.723.0

22.822.6

29.123.8

Table 3Effects ofdirection ofjudgment, visibility, directing ofattributions, and setting ofgoals uponcausal attributions (F-ratios)

Source MANOVA3 Effort Luck Ability Task difficulty

Afterfirst judgmentDirection of judgment 63.34*** 183.93*** 26.91 *** 1.85 45.88**Judgment conditions

Visibility 5.91 *** 2.45 11.29*** 10.83***Directing of attributions 7.50*** 7.34** 14.10*** 8.65** 2.92Setting of goals 5.37*** 6.89** 12.63*** 3.37 <1.0

After second JudgmentDirection of judgment 46.42*** 128.37*** 17.90*** <1.0 60.85***Judgment conditions

Visibility 4.30** 7.82** 9.37** 1.57 1.07Directing of attributions 3.13* 6.63* 5.95* <1.0 1.01Setting of goals 4.13** 5.51* 11.00*** 1.35 <1.0

Nilles. There are, no significant interactions between direction of judgment and the separate judgments conditions; aMultivariate test upon three of the four attribution variables; degrees of freedom: 3 and 2; b Degrees of freedom foreach ANOVA: I and 288; *** p<.OO I; ** p<.OI; * p<.05.

The attributions to luck were made only after the first judgment in accordance withhypothesis 2a: less attribution of causality to luck in Condition III (p<.OOI). After the secondjudgment, the causal reactions were the reverse of the prediction (p<.05).

These univariate effects on effort and luck were supported by multivariate tests: p<.OOI

20 F. SIERO& J. P. VAN OUDENHOVEN

after the first judgment and [1<.05 afte r the second j udgme nt. As rep orted above for thecomparison of Condition I and II , the addition of references to effort in Co ndition III led to astronger attribution of ca usa lity to ability as well (p<.0 I ). After the second judgment, studentsdid not differ in the ir attributions to ability.

Settin g of goals (compa rison of Condition III and IV). In the introduction , the behavioureffects of the setting of goals were discussed. Yet , forcing students to strive after a parti cularimprovement seeme d to be of co nse que nce for the percei ved co ntrolla bi lity. Th e meanattribution scor es are di spl ayed in Table 1 and Table 2. In co mparison with stude nts inCondition III (s trong visibility of continge ncy plus reference to effor t, bu t no task goa ls).stud ents in Condition IV (addition of task goa ls) attributed their outcome after the first and thesecond judgment to a higher degree to effort and to a lesser degree to luck (for e ffort: p<.OIfor the first judgment and [1<.05 for the second judgment ; for luck : p<.OO I for both the firstand the second judgment). The test result s in Table 3 show that the effects of goal setting onthe attributions to effort and luck were also multivariate sign ificant (p<.OOI and p<.O I afterthe first and after the second judgment respectively).

Direction of the judgment. We expected that the more the j udgment from Condition I toCondition IV was associated with changes in task behaviour, the more the students wouldattribute failure to effort. Indeed, trend analysis on the cumulative effect of the contingencyrising in the four conditions on the effo rt attributions showed a linear trend from Condition Ito Condition IV in the backwards condition which was different from the linear trend in thefor ward condition. The int era ct ion effec t was the mo st obvio us aft er the fir st judgment(F(I .288)=2.36. [1<.01 ). After the second judgment, the interaction effect was in the predicteddirection but . two-sidedly tested , not significant (F( I .288)=2.38, [1<.12).

Table 4Mean attributions ofcausa lity to effo rt fo r combinations ofjudgment conditions and dir ectionofjudgment

Judgment conditions

Directionof judgment Low visibility

AJierfi rst j udgmen tBackwards 17.3Forward 30.2

After second judgmentBackwards 18.0Forward 29.7

Notes. Rangeofattributionscores: 10-40.

Highvisibility Directing of attributions Setting of goals

18.7 20.3 23.029.1 30.2 30.4

21.9 21.4 24.129.8 3 1.1 31.5

The mean attributions for effort after the first judgment in Table 4 show that the more thecontingency was expressed from Condition I to Condition IV, the more the students attributedthe ir failure (i.e., achievement deterioration) to this cau sal fac to r (linear trend within thebackwards condition: F(I.288)=:]4.09, p<.OO I). On the other hand , stude nts in the forwardcondition attributed their success to the four conditions equally (no line ar trend within thefo rward condition: F( 1,288)< I). Concerning the weaker inte raction effec t after the secondj udg ment. a separate test of the linear trend for the back wards and a lin ear trend for theforward condition sho wed an identical result. The attributions of failure to effort were stronglydep endent on the degr ee of contingency in the four co ndi tions (line ar trend within thebackward s condition: F(1.288)=9.90, p<.O I). Within the forward condition, the linear trendwas not significant (F( 1.288)= 2.35).

In addition to the interaction effect between the four contingency conditions and directionof the judgment on effort att ributions, direction of judgm ent also had a main effect onattribution of causality. In spite of the indi viduali zed feedb ack co ntext. students were subject

CONTINGENT FEEDBACK 21

to the so-called attribution bias. After the first judgment as well as after the second judgment,students who showed improvement attributed their success to effort (internal factor) to ahigher degree and to luck and task difficulty (external factors) to a lesser degree. Meanattributions are displayed in Table I (first judgment) and Table 2 (second judgment).Strikingly, the pattern of causal reactions after the first and the second judgment is almost thesame: An identical contrast between the forward and backwards condition for attributions toeffort, luck, and task difficulty and no differential attributions to ability. The univariate andmultivariate F-values are depicted (for all effects: p<.OI) in Table 3.

Achievement

Each of the three judgment strategies, enhancing the visibility of contingency, directingof attributions, and setting of goals, should encourage students to achievement improvement(Hypothesis Ib, 2b, and 3). Indeed, the performances after the first judgment support thesehypotheses. Table 5 presents the means for achievement test 3 (administered in lesson 3) andachievement test 4 (administered in lesson 4), by means of analysis of covariance adjusted forinitial differences on achievement test I and achievement test 2.

Table 5Mean achievement scores within the four judgment conditions

Judgment conditions

Achievement testa

Test 3Test 4

Low visibility

17.330.2

High visibility

18.729.1

Directing of attributions

20.330.2

Setting of goals

23.030.4

Notes. Range of attributionscores: 10-40.

After the first judgment, students had higher achievement scores in the high visibility ofcontingency condition than in the low visibility condition (F(I ,286)=5.80, p<.05). Theaddition of references to effort to the highly visible contingent judgments also resulted inhigher achievement scores (F(l ,286)=4.92, p<.05). Finally, imposing a task goal in addition tothe two other strategies led to the highest achievement scores (F(I ,286)=7.58, p<.O I).However, this strong cumulative effect of the three strategies from Condition I to ConditionIV on achievement was absent after the second judgment.

Discussion

Does enhancing control over outcomes within the classroom encourage students toperceive more control and to try harder? The results of this field-experiment appear to answerthis question affirmatively. Enhancing the visibility of the contingent judgment, directingattributions, and setting of goals evoke, even when used repeatedly, a strong sense of control.Under the influence of these three control strategies, students appear to conceive effort as animportant determinant of the judgments. Moreover, under the influence of the three controlenhancing strategies, the students exert themselves more and show higher achievement.

Initially, enhancing the visibility of the contingency and directing of attributions led to anunpredicted, stronger attribution of causality to ability, a stable internal factor. This is notsurprising. Students are accustomed to comparing results and to linking judgments aboutperformances to their abilities. Therefore, in the first instance, receiving feedback aboutachievement improvement and achievement deterioration will evoke feelings of competenceor incompetence: "I have improved more than you have, thus I am better than you".

In this study, students attributed their success (i.e., improvement) mainly to effort and

22 F. SIERO & J. P. YAN OUDENHOYEN

explained away their failure (deterioration) to luck and task difficulty. This pattern isanalogous to the contrast in causal reactions between high and low achievers in a former study(Siero et aI., 1983) in which we applied a classical system of judgments, thus assessingachievement in relation to other students' achievement, on the basis of a group-basedcriterion. Strikingly, students also take less responsibility for negative outcomes within thecontext of this study in which feedback was individualised, in order to enhance theresponsibility for negative results. Apparently, the negative or positive nature of the judgmentis such a dominant element that, in spite of the individualized character of the feedback,students are inclined to external attribution of their deterioration.

On the other hand, students can be orientated towards effort as the explanation for theirnegative outcomes by emphasizing the controllability of the judgment and the task situation.In this study, "failure" was connected to effort most strongly when the three strategies,enhancing the visibility of the contingent judgment, directing attributions, and setting goals,were combined. Likewise, these strategies may be applied successfully in other educationalsettings provided that the judgments are based on improvement and deterioration of individualperformances.

The cumulative effect of the three strategies on perceived control and achievement ismainly present in the beginning. In particular, the directing of attributions which was added tothe highly visible contingency has less effect on perceived control after the second judgment.

After that judgment, the stronger perception of controllability finds expression only inthe attributions to effort. Yet, the other two strategies, enhancing the visibility of thecontingent judgment and setting goals, prompted students, after both the first judgment andthe second judgment, to perceive more control over their outcomes.

The three strategies resulted in systematically better achievement only after the firstjudgment. Perhaps the unusual and strong modifications in the nature of the judgments werethe cause of the short-lived achievement effects. However, this does not mean that controlenhancing strategies could not be effective for a longer period. In a four month field­

experiment (Van Oudenhoven, Siero, Veen, & Withag, 1983), third grade studentsdemonstrated such long-term effects. Three conditions were applied: the usual classicalsystem of judgments, a condition with judgments about improvement of achievement, and acondition in which the improvement feedback was combined with individualized instruction.In the conditions with improvement feedback, the students showed greater achievement, butthese effects only consolidated in the fourth and final month.

That these effects did not consolidate until later, is probably due to the fact thatcontingency was just one principle of the existing spelling instruction programme. The pupilshad to deduce contingency from tiny (subjective) steps of improvement. From a more generalapplied point of view, Van Oudenhoven et al. (1983) show how principles such ascontingency, individualised feedback, and goal setting, which were artificially distinguished inthis study, can be successfully integrated in existing spelling instruction methods.

Notes

Because of the use of percentage scores, the fourth attribution variable is a perfect linear combination of the threeother variables. Therefore, MANOY A was applied on three of the four variables.

References

Alloy, L., & Tabachnik, N. (1984). Assessment of covariation by humans and animals: The joint intluence of priorexpectations and current situationa. information. Psychological Review, 91, 112-149.

Bandura, A., & Wood. R. (1989). Effect of perceived controllability and performance standards on self-regulation ofcomplex decision making. Journal ojPersonaiity andSocial Psvcholog», 56,805-814.

CONTINGENT FEEDBACK 23

Becker, L. J. ( 1978). Joint effect of feed back and goal sett ing on performan ce: A fie ld study of residential energyco nservation. Journal of Applied Psyc/wIOKY, 63, 428-433.

Crocker, J. (198 1). Judgment of covariatio n by social perceivers. Psych ological Bulletin, 90,272-292.

Forsyth , D. R., & McMill an, J .H. (1981). Attributions, affect, and expecta tions: a test of Weiner' s three-d imensionalmodel. J ournal ofEducat ional Psycho logv, 73,393-403.

Garland, H. (1983). Influence of ability, assig ned goals, and normative information on perso nal goals and performance:A challenge to the goal attainability assumption. Journal ofApplied Psych ology, 68, 20-30.

Hewstone, M., & Jaspers, J. (1983). A re-examination of the roles of consensus, consiste ncy and distinct iveness:Kelley 's cube revisited. British.Journal ofSocia l Psvchotogy, 22, 41-50.

Hewstone, M., & Jaspers , J. ( 1987). Covariation and causal attribution: A logical model of the intuitive analysis ofvariance. Jou rnal ofPersonality and Socia l Psycholog», 53, 663-672.

Jack son , S. E., & Zedeck , S. ( 1982) . Expl ain ing per formance va riabi li ty : Contri but ion s of go al se tti ng , taskchuracteristics, and evaluative contexts. Journal ofApplied Psych ology, 67, 759-768.

Kelley, H. H. ( 1973). The processes of causal attribution. American Psychologist, February, 107-128.

Locke, E. A., & Latham , G. P. (1990). A theory of goal setting and task performance . Englewood Cliffs, NJ: PrenticeHall.

Mento, A., Steel, R., & Karren , R. (1987). A meta-analytic study of the effects of goal setting on task performanc e:1966-1984. OrKallizatirJllalBehaviur alld Humall DecisiolJ Processe s, 39, 52-83.

Nisbett , R. E., & Ross, L. (1980). Human inference: Strategies and short com ings ojsocial judgment. Englewood Clif fs,NJ: Prentice-Hall.

Sier o, F. W.. Oudenhoven , J . P. Van . & Veen , P. (1983). Caus ule attributie s nu oo rdelen over prestat ies: ee nveld experiment in het onde rwijs [Causal attributions afte r performance ju dg ments: An experiment in aneducational selling]. Gedrax - tydschrift voor psychologie, 11, 190- 204.

Van Oudenhoven, J. P., Siero, F. W., Veen, P., & Withag, J. (1983). Effects ofindividualised feedbuck and ins tructionOil effo rt attributions ahilit), attribution s Will "pell illKachievement. Educational Studies, 9, 105-113.

Weiner, B. (1982). An attributionul based theory of motivation and emotion: Focus, range, and issues. In N. T. Feather(Ed.), Expec tation , and Actions: Expectancy-Value Models III P"ychoiOKY (pp. 163-203). Hillsdale , NJ: LawrenceErlbaurn Associates, 163-203.

Weiner, B. (1935). An attributi onal theory of achievement motivation. Psych ologicul Review. 92, 548-573.

Zuckerman , M. (1979). Attribu tions of success and failu re revisited , or: The motivational bias is alive and well inattribution theory. Journal ofPersona lity, 47, 245-287.

Key words: Contingency, Elementary schools, Feedback, Performance.

Received: November 1993

Revision received: March 1993

FransSiero, University of Cron ingen , Social and Organizational Psychology. Grote Kruisstraat 2/1, 9712TS, Groningen, The Netherlands.

Current theme ofresearch:

Persuasion.

Most relevant publications ill thefield ofEduca tional P.\) ·cho/OKY:

Van Oudenho ven, J. P., Siero, E, Veen, P., & Withag, J. (1983). Effect, of individualized feedback and instruction oneffort attributions, abil ity attributions and spelling achievement. Educat ional Series, 9, 105-113.

Van Oude~~ven , J. P: ,.Siero, ~. , & Withag ~ J. (1984). La lang~e offic ielle et l'evaluation de redactions cornposees pardes eleves de milieux socio-cultures differents, Psych ologic Fruncuise, 29, 204-208.

24 F. SIERO & 1. P. VAN OUDENHOVEN

Van Berkum, G., Buisman, W. Kok, G. J., & J. P., Siero, F. (1991). Mass communication and drug education.Evaluation of the Dutch national Company booklet "What everybody ought to know about drugs". Journal (~{

Alcohol and Drug Education, 36(3),63-74.

Jan Pieter van Oudenhoven. University of Gronin~en, Social and Organizational Psychology, GroteKruisstraat 2/[,9712 TS, Groningen, The Netherlands.

Current theme ofresearch:

Cooperation and ethnic intergroup relations Social Psychological factors of international mergers,

Most relevant publications ill thefield ofEducational Psychology:

Van Oudenhoven, J. P., Berkum, G. Van, & Swen-Koopmans, T. (1987). Effect of cooperation and shared feedback onspelling achievement. Journal ofEducutional Psychology, 79,92-94.

Van Oudenhoven, J. P., Wiersema, B., & Yperen, N. van (1987). Effects of cooperation and feedback by fellow-pupilson spelling achievement. European Journal ojPsychology ofEducation, 2, 83-91.

Van Oudenhoven, J. P., & Willemsen, T. M. (Eds.) (1989). Ethnic Minorities. Social Psychological Perspectives.Amsterdam: Swetz en Zeitlinger.

Wiersema, B., & Van Oudenhoven, I. P. (1992). Effects of cooperation on spelling achievement at three age levels(grades 2, 4 AND 6). European Journal ofPsychology ofEducation, 7,95-108.