Effects of reward-schedule parameters and attribution ... - Springer Link

Bulletin of the Psychonomic Society

1986, 24(1), 65-68

Effects of reward-schedule parameters and attribution retraining on children's attributions and reading persistence NANCY E. MEYER and DENNIS G. DYCK

University of Manitoba, Winnipeg, Manitoba, Canada

Fourth- and fifth-grade children, identified as having reading difficulties, were given persistence training on a reading task. The effects of various combinations of reward-schedule parameters and attribution retraining (effort feedback) were evaluated. Four groups of children were given either partial reward with nonsuccessive failure (PRF) or continuous reinforcement prior to partial reward (CRF-PRF), either with or without attribution retraining during an 8-day training phase. An additional group received CRF without attribution retraining (effort feedback). The results indicated an increase in reading persistence (an increase from pretest to posttest in the number of difficult sentences attempted) as a function of attribution retraining but not of reinforcement scheduling. Although attributions to lack of ability were related to low pretest reading levels and persistence, the changes in reading persistence were not accompanied by changes on any attribution measure. Taken together, these results support previous studies in documenting the efficacy of effort feedback in promoting persistence, but they do not support the hypothesized mediation of such persistence by altered attributions for failure. fects between the pattern of success and failure outcomes (a reward-scheduling parameter) and attribution retraining on reading persistence; this was done since the two variables had been confounded in Dweck's study, and the resultant effects, therefore, could not be unambiguously ascribed to the attribution-retraining procedure. Chapin and Dyck's results did, in fact, demonstrate an interactive effect between a reward-scheduling parameter, termed N length, and attribution retraining. The specific form of the interaction was that attribution retraining increased persistence only under conditions of nonsuccessive failure. Unfortunately, Chapin and Dyck did not include measures of attributional change in their study; therefore, it was not known which, if any, of the observed changes in persistence were mediated by altered causal attributions. Although subsequent studies have supported the claim that effort-feedback training enhances persistence (Andrews & Debus, 1978; Fowler & Peterson, 1981; Medway & Venino, 1982), the mechanism underlying this effect is by no means clear. Either attributional change has not been directly assessed (Chapin & Dyck, 1976) or it has been inconsistently related to changes in persistence, depending on the measure used (Andrews & Debus, 1978; Dweck, 1975; Medway & Venino, 1982). The purpose of the present study, then, was to further examine the changes in attributions, reading persistence, and their relationship following various effort-feedback and scheduling-parameter manipulations. To that end, separate groups of children, who previously had been identified as having significant reading problems, were given persistence training following the methods of Chapin and Dyck (1976). Within the context of a reading task, four groups of children were provided with success- or

Previous investigators have been interested in studying the effects of effort-feedback (Dweck, 1975) and reward-scheduling parameters (Chapin & Dyck, 1976) on the reversal of the persistence deficits of academically "helpless" children. In the effort-feedback procedure, as initially developed by Dweck (1975), nonpersistent children were exposed to intermittent failure on an arithmetic task, and an adult evaluator verbalized that lack of effort caused the child's failure. This procedure, relative to a success-only procedure, increased the persistence of children previously "helpless" in response to math failure. Based on the premise that persistence is mediated by the perception of personal control (Rotter, 1966; Weiner, 1972), Dweck proposed that her training procedure increased persistence by modifying causal attributions for failure. This notion received partial support from the finding that effort-feedback training produced an increased frequency in attributions to effort on a task-specific, but not on a general, measure of attributions. Chapin and Dyck (1976) followed up Dweck's initial report with an assessment of the possible interactive efThis research was supported by a Manitoba Health Research studentship to N. E. Meyer and by a National Research Council Grant No. 3111665-06 to D. G. Dyck. The research was done in partial fulfillment of the first author's premaster's year in the Department of Psychology. University of Manitoba. We would like to thank Ron Petrinack for his help in all phases of the study: Rhea Brooks. Tom Dahl. and April Machej for their assistance in carrying it out: Linda Wilson for her helpful comments and direction: and the teachers. resource teachers. and principals of Assiniboine South No. 3 School Division for their cooperation and support. Requests for reprints should be sent to D. G. Dyck. Department of Psychology. University of Manitoba. Winnipeg. Manitoba. Canada R3T 2N2.

65

Copyright 1986 Psychonomic Society, Inc.

66

MEYER AND DYCK

failure-feedback patterns, in the form of either partial reinforcement (PRF) or continuous reinforcement (CRF) prior to PRF training (CRF-PRF), which were combined either with or without effort feedback (attribution retraining) . A fifth group received success-only reinforcement (CRF). Both PRF schedules developed for this study contained nonsuccessive failure (N length = 1), because it is under these conditions that the effects of the effort-feedback procedures are most likely to be observed (Chapin & Dyck, 1976). The two-stage schedule (CRF-PRF) was selected for comparison purposes in order to capitalize on the known benefits of the individual CRF and PRF schedules. Specifically, the initial experience with CRF was hypothesized to provide children who had previously experienced much reading failure with increased expectations of success and mastery. The later exposure to PRF , which involved the gradual "fading in" of failure (Terrace, 1966, 1969), was designed to promote reading persistence (Cotler & Nygaard, 1969; Yelen, 1980). The number of difficult sentences that children attempted to read was recorded both prior to and following the persistence training phase; this constituted the measure of reading persistence. Pre- and posttesting was performd by an experimenter who was blind to the children's training condition. Two self-report attribution scales were used to assess the reasons children gave for reading failure and for academic outcomes in general.

METHOD Subjects There were 30 subjects, 5 females and 25 males between the ages of 8 and 11 years, who participated in the treatment conditions. 1 They were selected from 71 Grade 4 and 5 students with reading problems from two schools in a Winnipeg school division. Materials The Informal Reading Assessment (IRA; Bums & Roe, 1980) was administered to all students with reading problems. The IRA was used to identify each student's reading level for the reading task. The inventory contains graded word lists, which each subject read orally. The number of errors in a list determined a subject's grade reading level. The word lists of the IRA were used to construct the sentence inventory used in training and in pre- and posttest. After a subject's reading level was assessed by the IRA, success and failure sentences were assigned from a corresponding level in the sentence inventory. Success and failure sentences for each reading level were drawn from previous and subsequent reading levels, respectively. A JO-item subset of the 34-item Intellectual Achievement Responsibility (IAR) Scale (Crandall, Katkavsky , & Crandall, 1965) was administered in the selection process and again after the posttest in order to assess attributions in academic situations. Each item depicts an achievement situation and an internal and external attribution in which the child must take personal responsibility for outcomes or relegate that responsibility to others. We adopted this subset of items from Diener and Dweck (1978) because we were also concerned with children's attributions to effort in failure situations. An Effort versus Ability Attribution Scale (EAS) for reading, similar to Dweck's (1975) task-specific attribution measure for arithmetic, was administered in pre- and posttest. The scale is designed to assess the extent to which children attribute their reading performance to ability or effort. It depicts 10 negative reading situations (e.g., "You were asked to read a story aloud in class. You had trouble and stopped because: (a) you hadn't practiced reading it the night before, or (b) you weren't a good oral reader").

Procedure Seventy-one students experiencing reading problems were identified by teachers and reading scores from Canadian Achievement Tests. The behavioral measure of persistence, consisting of pretest failure sentences, was administered to all 71 children. The 30 children with the lowest persistence scores were selected and assigned to the treatment conditions. The reading inventory (IRA), which had been used to construct the reading task, was then used to identify subjects' reading levels. Pretest/posttest. Subjects were individually tested on pre- and posttraining persistence and attributions by different persons than those who conducted the training. The pre- and posttest consisted of 20 failure sentences, which the subjects were asked to read aloud. After each sentence, subjects were given feedback about their accuracy and asked whether they would like to continue reading (e.g ., "That's incorrect, would you like to go on to the next sentence or to stop?"). Persistence was defined as the number of sentences attempted. Following the persistence measure, the EAS was administered in pre- and posttest, and the IAR was administered in posttest. Treatment conditions. With the restriction that an equal proportion of females (at least 1) be assigned to each condition, the 30 subjects were also matched on their persistence scores and assigned to one of the five treatment conditions. 2 In three conditions, the reinforcement schedules (CRF, PRF, CRF-PRF) were given without attribution retraining (AR), and in two conditions, the schedules were combined with AR, (PRF-AR, CRF-PRF-AR). During training treatments , the subjects were asked to read sentences that were preassigned at various levels of difficulty to ensure appropriate levels of success or failure. The SUbjects read 10 sentences each day for 8 days . To increase the salience of the feedback , the subjects were asked to keep a tally of their performance (right or wrong) as they read. In the attribution retraining groups, the experimenter provided effort feedback to the subject for success and failure trials: "That's correct, that means you tried hard," or "That's incorrect, that means you didn't try hard enough. " The groups without attribution retraining were told only about the accuracy of their responses . The CRF group received only success sentences . In designing the schedules for the PRF and CRF-PRF groups, an attempt was made to have: (a) a similar percentage of reinforcement in each group, (b) no consecutive failure trials, and (c) a similar number of failure trials on the last day . The schedules of success and failure trials that were used in training are shown in Table I. Following the training and the posttest, all subjects were given an additional set of success sentences to read for debriefing purposes.

RESULTS AND DISCUSSION Persistence Table 2 presents the means and standard deviations of the reading persistence scores on pretest and posttest for each group. A 5 (group) X 2 (trials) mixed-design repeated measures ANOV A revealed only a significant main effect for trials [F(1,24) = 5.89, p < .05]. Although this result implies that all groups were more persistent on the posttest, relative to the pretest, a more detailed inspection of the data indicated that this was not the case. Specifically, correlated t tests indicated that only the two groups that received effort feedback benefited from persistence training [Group CRF-PRF-AR, t(5) = 2.36, p < .05; and Group PRF-AR, t(5) = 2.34, p < .05]. Thus attribution retraining increased reading persistence, but the reward schedules did not. The results are consistent with Chapin and Dyck (1976), showing that PRF schedules with nonsuccessive failure (N length = 1) do not increase reading persistence without the benefit of attribution retraining. The critical scheduling manipulation appears to be N length, the number of successive

READING PERSISTENCE Table 1 Schedules of Success (S) and Failure (F) Trials for Training Sentence 4 Day 2 3 5 7 10 6 8 9 CRF-PRF I S S S S S S S S S S 2 S S S S S S S S S S 3 S S F S S S S S S S 4 S S S F F S S S S S F F F 5 S S S S S S S F F 6 S S S F S S F S 7 S F S F S F F S S S F F 8 S S S S F S F S I 2 3

4 5 6 7 8

S S S S F S S S

S S S S S F S F

F S F F S S F S

PRF S S F S F S S S F S S S S S F S

F S S S S S F F

S F S S S F S S

S S F S S S S F

F S S F S S S S

S S S S F S F S

Table 2 Means and Standard Deviations of the Number of Sentences Attempted for Each Training Condition Posttest Pretest Mean SD Mean SD Condition NoAR 3.4 4 .6 1.1 \.8 CRF 7.5 7.0 3.8 2.3 PRF 4.7 2 .9 3.2 \.5 CRF-PRF AR PRF CRF-PRF

3.7 3.8

2.4 3.0

7.8 8.8

6.5 6.6

Note-NoAR = without attribution retraining; AR = with attribution retraining; CRF = continuous reinforcement; PRF = panial reinforcement; CRF-PRF = two-stage continuous and partial reinforcement. Although the sentence inventory has not yet been formally validated. its sentences prm'ed to be effective in providing the appropriate success and failure trials.

failures prior to success (see Capaldi, 1967; Chapin & Dyck, 1976), rather than the provision of CRF prior to PRF. That the children who received CRF prior to PRF benefited from attribution retraining is noteworthy. This finding suggests that failure can be faded in without any loss in persistence, at least relative to a PRF schedule with nonsuccessive failure. Although not shown here, there may be advantages to the initial use of CRF in remedial programs, due to the fact that children with a history of failure may have difficulty processing success (Diener & Dweck, 1980). For such children, the initial provision of success (CRF) may be important to the modification of attributions for success, whereas the PRF procedure may be more conducive to the modification of attributions for failure.

Attributions Correlation analyses indicated that reading-specific attributions (EAS scores) were modestly related to pretest

67

reading persistence scores [r(29) = - .19, P < .10] and also related to reading achievement level as assessed by the IRA [r(20) = - .25, P < .05] . These results indicate that lower reading performance and persistence levels were associated with the tendency to make attributions for failure in reading to lack of ability as opposed to lack of effort. On the other hand, the more general attributional profile, as revealed by the IAR, was not associated with reading level or persistence. Since a major purpose of this study was to examine the relationship between attributions and persistence under various parameters of persistence training, a number of correlational and analysis of variance procedures were applied to the pretest and posttest scores of the EAS and IAR. No significant correlations or effects were found, suggesting that the observed behavioral changes were not mediated by modified attributions. We are left, then, with an attribution-retraining manipulation that enhances persistence in the manner predicted by theory but, at the same time, has no measurable effect on attributions. In our view, the lack of attributional change could be explained by the comparatively short duration of persistence training, or an EAS measure that assessed attributions to a variety of reading situations, rather than to the sentence task itself. Also, in both the EAS and the IAR, only attributions for failure were assessed. A more composite measure of attributions for success and failure may have yielded a more discriminating attributional profile from which changes could be assessed. Finally, even with an extended training period and improved measures, the relationship between persistence and attributions may be modest. Effort feedback and scheduling procedures may boost persistence through the modification of other mediating processes such as increased "efficacy ~xpectation " (see Bandura, Adams, & Beyer, 1977) and higher levels of accustomed learned effort expenditure (see Eisenberger, Leonard, Carlson, & Park, 1979). In addition to defining the relationshp between causal attributions and performance, further work is needed to identify the contributions of other potential mediators of persistence and achievement behaviors . REFERENCES ANDREWS, G. R. , & DEBUS, R. L. (1978). Persistence and the causal perception of failure: Modifying cognitive attributions. Journal of Educational Psychology, 70, 154-166.

BURNS, P. c., & ROE, B. D. (1980). Informal reading assessment. New York: Houghton Mifflin. BANDURA, A., ADAMS, N. E., & BEYER, J. (1977) . Cognitive processes mediating behavioral change. Journal of Personality & Social Psychology, 35, 125-139.

CAPALDI. E. 1. (1967). A sequential hypothesis of instrumental learning. In K. W. Spence & J. T. Spence (Eds.), The psychology of learning and motivation (Vol. I) . New York: Academic Press. CHAPIN, M. , & DYCK, D. G. (1976). Persistence in children's reading behavior as a function of N length and attribution retraining. Journal of Abnormal Psychology, 85, 511-515.

COTlER. S. B., & NYGAARD, 1. E. (1969). Resistance to extinction following sequences of partial and continuous reinforcement in a human choice task. Journal of Experimental Psychology, 81, 270-274. CRANDALL, V. C., KATKOVSKY, W., & CRANDALL, V.I . (1965) . Chil-

68

MEYER AND DYCK

dren's beliefs in their own control of reinforcement in intellectual academic achievement situations. Child Development, 36, 91-109. DIENER, C. I., & DWECK, C. S. (1978). An analysis of learned helplessness: Continuous changes in performance, strategy and achievement cognitions following failure. Journal of Personality & Social Psychology, 36, 451-462 . DIENER, C. I., & DWECK, C. S. (1980). An analysis of learned helplessness: II. The process of success. Journal of Personality & Social Psychology, 39, 940-952. DWECK, C. S. (1975). The role of expectations and attribution in the alleviation of learned helplessness. Journal of Personality & Social Psychology, 31, 674-685. EISENBERGER, R., LEONARD, J. M., CARLSON, J., & PARK, D. C. (1979). Transfer effects of contingent and noncontingent positive reinforcement: Mechanisms and generality. Journal of American Psychology, 92, 525-535. FOWLER, J. W., & PETERSON, L. (1981). Increasing reading persistence and altering attribution style in learned helpless children. Journal of Educational Psychology, 73, 251-260. MEDWAY, F. J., & VENINO, G. R. (1982). The effects of effort feedback and performance patterns on children's attributions and task persistence. Contemporary Educational Psychology, 7, 26-34. ROTTER, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs, 80(1, Whole No. 609).

TERRACE, H. S. (1966). Stimulus control. In W. K. Honig (Ed.), Operant behavior: Areas of research and application. New York: Academic Press. TERRACE, H. S. (1969) . Extinction of a discriminative operant following discriminative learning with and without errors. Journal of Experimental Analysis of Behavior, 12, 571-582. WEINER, B. (1972). Theories of motivation: From mechanism to cognition . Chicago: Markham. YELEN, D. (1980). Order of partial reinforcement and continuous reinforcement schedules in verbal choice behavior. Psychological Reports, 46, 635-649. NOTES

I. One male subject's results were not used in the data analysis because he did not experience failure in pre- and posttest. He read 3/4 and 11/, 5 of his attempted failure sentences correctly on pre- and posttest, respectively. 2. Fewer girls than boys were initially selected as experiencing reading problems. On the basis of pretest persistence, only 4 girls initially fell within the range of the 30 least persistent children.

(Manuscript received for publication October I, 1985.)