Self-efficacy and behaviour as predictors of subsequent behaviour in an assertiveness training programme

Self-efficacy and behaviour as predictors of subsequent behaviour in an assertiveness training programme

Behar. Res. Thu. Vol. 21,No. 3,pp.225-232.1983 Printed in Great Britain. All rights reserved 00057967/83/03022S-08SO3.0010 Copyright ,C 1983 Pergamon...

877KB Sizes 0 Downloads 28 Views

Behar. Res. Thu. Vol. 21,No. 3,pp.225-232.1983 Printed in Great Britain. All rights reserved

00057967/83/03022S-08SO3.0010 Copyright ,C 1983 Pergamon Press Ltd

SELF-EFFICACY AND BEHAVIOUR AS PREDICTORS OF SUBSEQUENT BEHAVIOUR IN AN ASSERTIVENESS TRAINING PROGRAMME CHRISTINA Department

of Psychology,

University

of Adelaide,

LEE GPO

Box 498, Adelaide,

South

Australia

5001

(Received 22 June 1982) Summary-Bandura and his colleagues have offered evidence that self-efficacy measures are more accurate predictors of maintenance of behaviour change than are behavioural measures. The present article argues that their evidence is flawed by the use of inappropriate statistical techniques, and attempts to replicate these findings using alternative statistical methods. Forty-six female students participated in an Assertiveness Training Programme which ran for 6 weeks, with a follow-up 3 months later. At each session they made efficacy predictions about their performance in specific situations, and then role-played those situations. Efficacy and behavioural measures were compared as predictors of later behaviour, and it was shown that, contrary to Bandura’s findings, efficacy measures were significantly less accurate than behaviours in predicting later behaviour. This result indicates that, although self-efficacy theory is a useful framework for understanding behaviour change, it may be inappropriate in some situations to rely entirely upon efficacy predictions.

INTRODUCTION

Recently, a number of people have described concepts designed to integrate the somewhat fragmented field of behaviour change (e.g. Rachman, 1980; Staats, 1981). One concept which has stimulated a considerable amount of discussion and research is Bandura’s self-efficacy theory (Bandura, 1977a, 1978a, 1980, 1982), which argues for a single process underlying all behaviour change, regardless of the therapeutic method involved. Self-efficacy theory grew out of social learning theory (Bandura, 1977b, 1978b). This approach rejects both the unilateral environmental determinism of radical behaviourism (Ledwidge, 1978; Skinner, 1977) and the human determinism of trait theorists (cf. Mischel, 1977). Instead, it postulates a system of reciprocal determinism, a dynamic equilibrium between environment, behaviour and specific, modifiable person variables, including cognitions and affective states. Central to self-efficacy theory is the premise that people have expectations about their abilities and that these expectations are highly specific to particular behaviours. The concept of “efficacy expectations” thus differs from global concepts such as self-esteem. Bandura argues that, given “appropriate skills and adequate incentives”, (Bandura, 1977a, p. 194) these expectations will determine whether or not a task is attempted, how strongly the effort will continue in the face of failure, and what will be the final outcome. Further, he argues that efficacy expectations are not accompaniments but determinants of behaviour. Changes in behaviour result from changes in efficacy expectations, so that behaviour-change programmes succeed according to the extent to which they alter efficacy expectations. Efficacy expectations are considered better predictors of future behaviour than are previous behaviours (e.g. Bandura and Adams, 1977) which may be influenced by transitory factors such as presence of influential others, situational constraints and ambiguities, ill-defined goals and temporary conditions. Efficacy expectations provide a purer measure, and, given adequate skills and a desire to perform, should predict future behaviour more accurately. Although this theory has met with some criticism (e.g. Rachman, 1978; Biglan, 1981) it does have some advantages over rival explanations of behaviour change, in providing a place for cognitions as determinants of behaviour without losing the unambiguous measurement required in experimental work. Experimental evidence exists to support this theory. These experiments have measured efficacy expectations with a self-report scale listing specific behaviours. Subjects indicate the number of behaviours they will be able to perform (Efficacy Level), and rate their confidence for each item (Efficacy Strength). Enactive tests of identical behaviours are carried out. Since items in the 225

CHKISTINALEE

226

questionnaire are highly specific and matched to behavioural items, a fine-grained, item-by-item analysis (microanalysis) may be performed. This method has been used to show a close match between efficacy and behaviour in experiments 1977; Gauthier and Ladouceur, 1981) involving snake phobia (e.g. Bandura and Adams, agoraphobia (Bandura, Adams, Hardy and Howells, 1980), acrophobia (Biran and Wilson, 1981; Borque and Ladouceur, 1980), fear of lifts and of darkness (Biran and Wilson, 1981), assertiveness development (Kazdin, 1980) vocational choice (Betz and Hackett, 1981) academic performance (Bandura and Schunk, 1981; Keyser and Barling, 1981), smoking cessation (e.g. Condiotte and Lichtenstein, 1981; Owen, Ewins, Bullock and Lee, 1982) recovery from heart attacks (Bandura, 1982) and physical and sporting performance (e.g. Feltz, Landers and Raeder, 1979; Weinberg, Gould and Jackson, 1979; Lee, 1982; Morelli and Martin, 1980). Experiments have been performed to assess the relative predictive accuracy of efficacy and behaviour. Bandura, Adams and Beyer (1977) examining treatments of snake phobia, included a l-month follow-up in their design. They compared behaviour at the end of treatment with efficacy at that time as predictors of behaviour at follow-up. A hierarchy of behaviours involving approaching and handling a snake was used, and all 29 people had reached the top of the hierarchy either during the experiment or later during supplementary treatment. However, not all had maximal efficacy expectations. An examination of the item-by-item match between behaviour at the end of treatment and behaviour at follow-up on the one hand, and between efficacy at the end of treatment and behaviour at follow-up on the other, indicated error rates, for a similar threat, of 21’;; for efficacy and 287: for behaviour. For a dissimilar threat, the error rates were 24% for efficacy and 52”/, for behaviour. Efficacy expectations were significantly more accurate than behaviours for the dissimilar threat and for both threats combined, but not for the similar threat alone. Bandura et al. (1980) also provide evidence, from 11 agoraphobics, that self-efficacy is a better predictor of post-treatment test behaviour than are behavioural measures (10% vs 26’4 error rate). Other researchers (e.g. Keyser and Barling, 1981) have used regression techniques to show that efficacy accounts for a greater percentage of the variance of later behaviour than does behaviour, but have not shown a significant difference in predictive power. If efficacy expectations are indeed more accurate predictors of post-treatment behaviour than are behaviours, this has important implications for assessment. For example, efficacy measures may be used to predict maintenance with more success than behavioural measures, and identify those people most likely to relapse. However, the two experiments which support this contention can be criticized on statistical grounds. The effect found in the Bandura et al. (1977) study depends entirely on the inaccuracy of previous behaviours as predictors of behaviours with a “dissimilar threat” (p. 133). They do not say whether this is the snake used to measure generality at the end of treatment, or, if not, whether it is of equivalent threat value. Also, it is unclear whether the behaviours and efficacy measures used as predictors for behaviour with the ‘dissimilar snake’ involve this snake or some other one. Microanalytical techniques require that both the behavioural and the efficacy predictors involve the same threat, which must be identical to that threat assessed at follow-up, and it is not clear whether this is the case. The comparison of predictors in the Bandura et al. (1980) study of agoraphobics can be criticized on similar grounds. They say that “the behavioral tests were standardized by clearly specifying the community settings to be used for the performance tasks, the sequence in which they were to be administered, and the criteria of successful performance.” (p. 54) But they also add, “to evaluate the generality. . . the community differed from those used in the field mastery

settings chosen for the behavioral treatment” (p. 54)

tests

Presumably the efficacy expectation scales used did not differ between field tests and assessment tests. If behaviours carried out in settings deliberately chosen to be different from test settings arc

Predictors of behaviour change

227

with behaviours in those test settings, it is hardly surprising that there will be some difference. And if efficacy scales are identical, it is equally unsurprising that efficacy measured for one situation will match the other situation quite well. Again, microanalytic techniques are inappropriate. Microanalysis is appropriate only if the items are identical. If they differ, a correlational technique should be used. In these studies, written and enactive items may differ in salient aspects (such as size or liveliness of snake) which are not controlled for and which may affect various people differently. As Kirsch (1980) has pointed out, item-by-item match will be distorted by any differences, while correlational analyses of aggregate scores will not be affected. The question arises of whether the findings of Bandura and his colleagues can be replicated with correlational analysis. This was examined in the present experiment, involving an Assertiveness Training Programme for female university students. Assertiveness was selected as a target behaviour because of evidence (e.g. Alden and Cappe, 1981; Bellack, Hersen and Turner, 1979; Bordewick and Bornstein, 1980; Nietzel and Bernstein, 1979; Schwartz and Gottman, 1976) that non-assertiveness stems from cognitive deficit rather than from inability to perform or recognize the appropriate skills. Therefore, the relationship between efficacy and performance should be strong and direct. Students involved underwent 6 sessions of assertiveness training; during each session they made efficacy predictions of their performance in specific situations and then role-played those situations. With efficacy measures matching role-play situations within sessions, microanalysis of the relationship between efficacy and behaviour within sessions could be performed, and a close match was hypothesized. Efficacy ratings and behavioural scores were averaged within sessions, and correlational methods used with these averaged scores to examine whether efficacy or behaviour in previous sessions explained more of the variance in behaviour in later sessions. It was hypothesized, following Bandura’s findings, that averaged efficacy would be a better predictor of subsequent behaviour than averaged behaviour scores in previous sessions.

compared

METHOD

Participants Unassertive female students were selected as follows. The College Self-Expression Scale (CSES; Galassi, DeLo, Galassi and Bastien, 1974) with minor changes in wording to suit an Australian university population, was administered to 262 students enrolled in a first-year psychology course. Those scoring at least half a standard deviation below the mean, who were female and aged 21 or less, were selected to participate in the study. None refused to take part. Forty-seven students began the study and 46 completed it. Their mean age was 17 years 10 months, and their mean score on the CSES was 103 (Z = - 1.09). Participants were paid $15 each for attending 6 sessions. Apparatus Role-play situations were presented either session began with the following instructions.

by audiotape

or orally

by the experimenter.

Each

-‘In this session you will hear described a number of situations requiring an assertive response. At the end of each description you will hear a click and you will have 60 set in which to respond. Your response will be recorded on the second tape recorder.” There then followed either eight or four descriptions of situations in which an assertive verbal response would be appropriate, each followed by a 60-set pause during which the participant could respond. Scenes used as stimuli were derived from items included in five Assertiveness inventories (Galassi et ul., 1974; Gambrill and Richey, 1975; MacDonald, 1978; Rathus, 1973; Wolpe and Lazarus, 1966). All items involving verbal responses, excepting those with no relevance to female Australian students, were used. Those requiring a male stimulus voice were also deleted, to eliminate possible differential responding to males and females. The remaining items were randomly assigned to five sets of 8 and one of 4. Each item was then rewritten in a standard format, in which the situation

228

CHRISTINALEE

was described briefly, and ended with a “stimulus statement” participant would respond. For example:

in direct speech to which the

“You have four essays due in a 2-week period, and there’s really no way you can get them all done in time. You haven’t been sick or anything, but you really need an extension, and you decide to go and see a tutor about it. When you go into her room, she says, ‘Hello. How’s the essay going?” Ail stimulus tapes were recorded by the experimenter. A second tape recorder was used to record both the stimulus tape and the participants’s response. Procedure

Participants were seen individually by the experimenter for 6 half-hour sessions over a 6-week period. Twenty-six students participated during the first term of the university year, and 20 during the second. Those who participated in the first term were asked to return for a follow-up session 3 months after their training programme finished, and 21 did so. All sessions were tape recorded with the participants’ knowledge and consent. In each session, participants completed a “self-rating scale” (efficacy measure). In these, scenes were described briefly, and the participant was asked to state for each scene whether or not she believed she could deal with it in an appropriately assertive way, and, if so, to rate her confidence on a scale from 0 (quite uncertain) to 100 (certain). “Assertive” was defined as referring to a behaviour which ensured consideration both for her rights and for the rights of others. This provided measures of Efficacy Level and Efficacy Strength. Following completion of the efficacy scale, the participant listened to role-play situations matching the items on the efficacy scale, and responded verbally to each one. Discussion of responses and possible alternatives completed each session. In the sixth session, each participant repeated the efficacy measure and stimulus set she had used in the first session (part a), and also carried out a generalization test by responding to an efficacy scale and corresponding tape which she had not previously heard (part b). The initial and generalization tapes were randomized across participants. In the sixth session participants completed the CSES once more. Follow-up sessions were conducted individually 13 weeks after the final training session. Participants once again completed the efficacy measure they had used in their first session, and responded to the corresponding tape. The CSES was administered for the third time. Audiotape rating

A graduate student, blind to the purposes of the experiment, rated responses to all items on each of three variables. The variables were: Assertiveness, on a scale from 0 to 6 (0 = no response, 1-3 non-assertive, 4.5 assertive, 6 aggressive); Appropriateness (O-3 inappropriate, 4-6 appropriate) and Response Latency, measured in tenths of a second. The experimenter rated a random sample of 16 items on the same measures, and reliabilities appeared satisfactory. Mean differences per item between the raters were: for Assertiveness, 0.6 of a scale point; for Appropriateness, 0.4 of a scale point; for Latency, 1.6 sec. RESULTS Preliminary Analyses

Students participating in the first term did not differ from those in the second term on the CSES (F < l), nor were there differences between terms at the first session (Efficacy Strength: F( 1,42) = 1.31; Efficacy Level, Assertiveness, Appropriateness, Latency: Fs < 1). Half the students heard Tape 1 in Sessions 1 and 6(a), and used Tape 6 as a generalization measure, while this order was reversed for the others. No significant differences were found in Session 1 between these two groups (Efficacy Level: F(l,45) = 3.61, P = 0.06; Efficacy Strength: F( 1,4.5)= 0.09; Latency: F(1,45) = I .43; Assertiveness: F( 1.45) = 1.46; Appropriateness: F( 1.45) = 1.43), nor did they differ on the CSES (F < 1). So it appeared that the two tapes were equal in difficulty, Data from all participants were combined in all analyses.

229

Predictors of behaviour change

Assertiveness Changes Repeated-measures analysis of variance was used to assess changes during the programme. For each dependent variable, averaged scores were computed within each session. For Sessions 1 and 6(a) (see Table I for means), with identical stimulus materials, ail five variables showed improvement significant at or beyond the 0.0005 level (F ratios for CZ”=1,45-Efficacy Strength, 65.0; Efficacy Level, 53.6; Assertiveness, 57.7; Appropriateness, 52.9; Latency, 14.3). Repeatedmeasures analyses of averaged scores from all 6 sessions showed changes significant at the 0.0001 level (F ratios for @= 6,270-Efficacy Strength, 22.5; Efficacy Level, 14.9; Assertiveness, 13.2; Appropriateness, 16.1; Latency, 9.8). The CSES showed an increase [F(1,45) = 57.0, P -CO.OOOl]from an initial mean z-score of - 1.05 to a final mean of -0.22. Thirty-two participants (70%) had final z-scores greater than -0.5, and 41 (89y;) greater than - 1. Generalization and Maintenance Generalization was measured in Session 6 by using a new tape equivalent in difficulty to the test tape. Related-samples t-tests compared generalization data (6b) with data from Sessions 1 and 6(a). Performance and efficacy on the generalization tape were significantly better on ali measures than in Session 1; in comparison with the repeated measures (Session 6a), ‘generalization’ performance was characterized by lower Efficacy Strength and slower response times, but did not differ on the other dependent measures (see Table 1). Twenty-one students participated in follow-up testing 3 months after completing the experiment. Maintenance was assessed by comparing data from this session with data using the identical stimulus tape in Sessions 1 and 6. Performance and efficacy were significantly better on all measures than in Session 1; in comparison with Session 6(a), ‘maintenance’ performance was characterized by signi~~antly higher Efficacy Strength, and no significant change on other dependent measures (see Table 2). Eficacy Expectations as Predictors of Behaviour Accuracy Microanalysis of individual items was used to assess the match between the efficacy prediction made and the ratings of the response to that item. Efficacy Strength, measured on a scale from 0 to 100, and Efficacy Level, measured as a dichotomy, were closely related (Pearson Y = 0.713). Assertiveness and Appropriateness ratings were dichotomized at the scale midpoints, except that an Assertiveness score of 6 (= aggressive) Table 1. Generalization strength. Repeated-measures t-tests (n = 46) comparing Session 6(b) (generalization measure) with initial and final session scores

-_-.

Variable

Mean -_.-._ h(a) -.0.927 59.1 3.95 5.03 1.32

Session

Session ._ 0.750 43.6 3.32 4.48 2.87

Effkacy Level Efhcacy Strength Assertiveness Appropriateness Latency

f Session 6(b) 0.902 54.x 3.78 4.92 1.48

I vs 6(b) 4.79+** 4.29*** 4.50*** 5.44*** - 3.34**

6(a) vs 6(b) _..~_^ - i.to - 2.68*+ -1.91 -1.76 2.45’

*P < 0.05; **p < 0.01; ***fJ < 0.001.

Table 2. Maintenance

strength.

Repeated-measures

f-tests (n = 21) comparing scores

follow-up

scores with initial and final session

t ______.._____

Mean Variable

_- Session

Efficacy Level Efficacy Strength Assertiveness Appropriateness Latency **p < 0.01; ***p < 0.001.

0.750 43.6 3.39 4.39 2.86

1

Session 0.926 59.6 3.94 4.96 I.31

6(a)

Follow-up 0.949 63.9 3.98 4.99 I .32

1 vs F

6(a) vs F

4.9-f*** 9.76*** 4.46*** 4.78*** ---3.38**

5.70*** 0.97 0.43 0.10

1.28

CHRISTINALEE

??(I

was classified as non-assertive. Percentage matches with Efficacy Level were 85.5”?, for Assertiveness, and for Appropriateness 89.7%. The overall xz across all items was 755.4 for Assertiveness (n = 1792) and 652.8 for Appropriateness (n = 1788), both significant beyond the 0.0001 level. Percentage matches and xz were calculated separately for each of the first 5 sessions and both parts of the sixth session. Although there was some increase in accuracy across time (Assertiveness: F(6,246) = 2.10, P = 0.054; Appropriateness: F(6,234) = 3.47, P = 0.003) matches within individual sessions were all extremely high, ranging from 81.5 to 97.0%. All x2 scores were significant at or beyond the 0.001 level. Overall. the correlation between an Efficacy Level rating and the rating of Assertiveness for that item was 0.598, and between Efficacy Level and Appropriateness 0.501. Correlations with Efficacy Strength were 0.476 for Assertiveness and 0.395 for Appropriateness. All these correlations are significant beyond the 0.001 level. Comparison

of’ c$icucy

with preoious hehauiour us predictors

cf hehuaiour

A number of analyses were performed to assess whether efficacy or behaviour was a better predictor of subsequent behaviour. (a) Microanalysis. Since identical items were repeated at Sessions 1, 6(a) and follow-up, a microanalytic analysis across time was possible. Three predictors-Efficacy Level in Session 1, behaviour in Session 1, and Efficacy Level in Session 6(a)-were used as predictors of behaviour in Session 6(a). Also, efficacy and behavior in Session 6(a) and efficacy at follow-up were used as predictors of behaviour at follow-up. Table 3 summarizes the accuracy of the three predictors in each case. This table suggests that, while efficacy and behaviour are both good predictors of subsequent behaviours, behaviour on the whole is slightly better. It should be noted that the x2 analyses take into account chance levels of responding, while percentage-match scores do not (cf. Kirsch, 1980). Measures taken in Session I, while significant predictors of Session 6 performance, are not as accurate as those taken in Session 6 to predict follow-up. This presumably is because of the different degrees of behaviour change occurring in the intervals. (6) Correlation. Average scores within each session for Efficacy Level, Efficacy Strength, Assertiveness and Appropriateness were computed. The accuracy with which these scores predicted scores in later sessions was assessed. For all possible pairs of sessions (counting Session 6 parts a and b as two sessions, and including follow-up, there are 28 such pairs), individuals’ averaged Efficacy Level, Efficacy Strength, Assertiveness and Appropriateness in the earlier session were correlated with Assertiveness and Appropriateness in the later session. Four pairs of the resultant correlations were considered for each pair of sessions [(a) denotes the earlier session, (b) the later]. Each pair of correlations included an efficacy/behaviour and a behaviour/behaviour correlation: (i) Efficacy Strength (a) with Assertiveness (b) compared with Assertiveness (a) with Assertiveness (b). (ii) Efficacy Level (a) with Assertiveness (b) compared with Assertiveness (a) with Assertiveness (b). (iii) Efficacy Strength (a) with Appropriateness (b) compared with Appropriateness (a) with Appropriateness (b). (iv) Efficacy Level (a) with Appropriateness (b) compared with Appropriateness (a) with Appropriateness (b).

Each pair of correlations was compared using a test for differences between related correlations (McNemar, 1969). Table 4 summarizes the results of these 112 tests. All significant differences are in the same direction: all indicate that the behavioural measure at time (a) is a better predictor than the efficacy measure at time (a) of the behaviour at time (b). The sign test also indicates a highly significant tendency for the behaviour/behaviour correlations to be greater than the efficacy/behaviour correlations.

Predictors Table

3. Microanalysis

of behaviour

of behavioural behaviour

change

231

and efficacy measures on identical items

as predictors

Assertiveness Predictor

n

x2

Appropriateness

o/o accurate Predicted:

Session I Assertiveness Appropriateness Efficacy Level Session 6(a) Efficacy Level Session 6(a) Assertiveness Appropriateness Efficacy Level Follow-up Efficacv Level

360 358 360

215.2 121.3 45.4

360

197.6

159 159 167

136.8 82.6 130.7

167

130.7

of later

xz Session

o? accurate

6(a)

I1 1.9 150.1 34.9

75.8 81.3 77.1

161.7 Follow-up

94.7

91.2 91.9 95.9

139.3 84.4 126.8

91.8 93. I 98.3

95.9

126.8

98.3

71.4 80.2 77.5 92.2 Predicted:

All xz are significant beyond the 0.0001 level. The last two rows are identical, as no participants changed their responses to Efficacy Level items between Session 6 and follow-up.

DISCUSSION

These results do not support the contention that efficacy is a better predictor of behaviour than are previous tests of behaviour. Rather, they suggest that, in this case, the reverse is the case. It is not the purpose of this paper to assess the Assertiveness Training Programme; the programme was run to provide a situation in which behaviour and efficacy could be measured. While the participants change in behaviour over time, the lack of a separate control group makes this a version of the recurrent institutional cycle (Campbell and Stanley, 1963) a quasiexperimental design with some flaws. One cannot conclude definitely that the programme changed behaviour, but it is clear that there is a change in behaviour over time, and therefore that the experimental situation is similar to a therapeutic situation. One cannot argue that the main finding results from the use of an inappropriate or vague measure of efficacy, as there is a very close match between behaviour and efficacy, according to both microanalytic and correlational analyses, indicating a high level of specificity in the measures. In fact, the very high item-for-time match would result in a ‘ceiling’ effect which would reduce the differences between the correlations. So it is an indicator of the strength of the effect that significant differences still exist. It is, of course, feasible that the findings reported in this paper are specific to particular types of behaviours. Indeed, it is arguable that assertive behaviour is likely to be determined less strongly by efficacy than are other behaviours. Feedback as to whether one has acted assertively is not instant and unequivocal, as is feedback as to whether one has touched a snake. Although Bandura does not specifically mention accuracy of feedback in his analysis of the development of efficacy, it seems that learning out one’s ability through performance accomplishments and vicarious experience, and thus developing efficacy expectations, requires accurate and consistent feedback (cf. Bandura, 1969). The evidence of this study cannot support complete reliance on efficacy measures as the best predictors of behaviour wherever skills and incentives are appropriate. Other factors must be

Table 4. The results of 112 I-tests comparing efficacy and behavioural predictors of later behaviour

r-value positive I-value negative

measures

P > 0.05

P < 0.05

P < 0.02

P < 0.01

P < 0.001

58

11

5

13

12

13

0

0

0

0

A positive f-value indicates that the behavioural test (ignoring size or significance of f-value):

as

measure was more accurate. Sign N = 112, x = 13, P < 0.0001.

232

CHRISTINALEE

involved, and further research may serve to identify those factors. However, efficacy measures are accurate, and while there is no direct evidence supporting Bandura’s claim that they determine behaviours, the theory must be considered as an interesting and useful framework for considering behaviour change. REFERENCES Alden L. and Cappe R. (1981) Nonassertiveness: skill deficit or selective self-evaluation? Behar. Ther. 12, 107-l 14. Bandura A. (1969) Principles qf Behavior Mod$cution. Holt, Rinehart & Winston, New York. Bandura A. (1977a) Self-efficacy: toward a unifying theory of behavior change. PsychoI. Rev. 84, 191-215. Bandura A. (I 977b) Sociul Learning Theory. Prentice-Hall, New Jersey. Bandura A. (1978a) Reflections on self-efficacy. Adv. Behuv. Rrs. Thrr. 1, 2377269. Bandura A. (1978b) The self system in reciprocal determinism. An?. PsychoI. 33, 344358. Bandura A. (1980) The self and mechanisms of agency. In Social Psychological Perspectives on the Self’(Edited by Suls J.). Erlbaum. Hillsdale, New Jersey. Bandura A. (1982) Self-efficacy mechanism in human agency. Am. P.sychol. 37, 122-147. Bandura A. and Adams N. E. (1977) Analysis of self-efficacy theory of behavioral change. Cog. Thrr. Res. 1, 287-310. Bandura A. and Schunk D. H. (1981) Cultivating competence, self-efficacy, and intrinsic interest through proximal self-motivation. /. Person. sot. PsychoI. 41, 586598. Bandura A.. Adams N. E. and Beyer J. (1977) Cognitive processes mediating behavioral change. J. Person, sot. Psychol. 35, 125- 139. Bandura A.. Adams N. E.. Hardy A. B. and Howells G. N. (1980) Tests of the generality of self-efficacy theory. Cog. Ther. Res. 4, 39-66. Bellack A. S.. Hersen M. and Turner S. M. (1979) Relationship of roleplaying and knowledge of appropriate behavior to assertion in the natural environment. J. consulr. cfin. Psychol. 47, 670-678. Betz N. E. and Hackett G. (1981) The relationship of career-related self-efficacy expectations to perceived career options in college men and women. J. counsel. PsychoI. 28, 399-410. Biglan A. (1981) A critique of Bandura’s Social Learning Theory. Unpublished manuscript, Oregon Research Institute. Biran M. and Wilson G. T. (1981) Treatment of phobic disorders using cognitive and exposure methods: a self-efficacy analysis. J. consult. clin. Psychol. 49, 886899. Bordewick M. C. and Bornstein P. H. (1980) Examination of multiple cognitive response dimensions among differentially assertive individuals. Behuv. Ther. 11, 440448. Borque P. and Ladouceur R. (1980) An investigation of various performance-based treatments with acrophobics. Behau. Res. Ther. 18, 161-170. Campbell D. T. and Stanley J. C. (1963) Experimenlal and Quasi-Experimental Designs for Research. Rand McNally. Chicago. Illinois. Condiotte M. M. and Lichtenstein E. (1981) Self-efficacy and relapse in smoking cessation programs. J. consult. clin. Psj,chol. 49, 648-658. Feltz D. L.. Landers D. M. and Raeder U. (1979) Enhancing self-efficacy in high-avoidance motor tasks: a comparison of modeling techniques. J. Sport Psychol. 1, 112-122. Galassi J. P., DeLo J. S., Galassi M. D. and Bastien S. (1974) The college self-expression scale: a measure of assertiveness. Behac. Ther. 5, 165-171. Gambrill E. D. and Richey C. A. (1975) An assertion inventory for use in assessment and research. Behau. Ther. 6, 550-561, Gauthier J. and Ladouceur R. (1981) The influence of self-efficacy reports on performance. Behau. Ther. 12, 435-439. Kazdin A. E. (1980)Covert and overt rehearsal and elaboration during treatment in the development of assertive behavior. Behar. Res. Ther. 18, 191-201. Keyser V. and Barling J. (1981) Determinants of children’s self-efficacy beliefs in an academic environment. Cog. Ther. Res. 5, 29-40. Kirsch I. (1980) ‘Microanalytic’ analyses of efficacy expectations as predictors of performance. Cog. Ther. Res. 4, 259-262. Ledwidge B. (1978) Cognitive behavior modification: a step in the wrong direction? Psychol. Bull. 85, 353-375. Lee C. (1982) Self-efficacy as a predictor of performance in competitive gymnastics. J. Sporf Psychol. 4, 405409. MacDonald M. L. (1978) Measuring assertion: a model and a method. Behav. Ther. 9, 889-899. McNemar Q. (1969) Psychological Statistics, 4th edn. Wiley, New York. Mischel W. (1977) On the future of personality measurement. Am. Psychol. 32, 246-254. Morelli E. A. M. and Martin J. (1980) Self-efficacy and athletic performance of 800 meter runners. Unpublished manuscript, Simon Fraser University. British Columbia. Nietzel M. T. and Bernstein D. A. (1979) Effects of instructionally mediated demand on the behavioral assessment of assertiveness. J. consult. clin. Psvchol. 44, 500. Owen N.. Ewins A.. Bullock M. and Lee C. (1982) Adherence and relapse in health-related behaviors. In Behavioral _ Medicine (Edited by Sheppard J.). Cumberland College, Sydney. Rachman S. (Ed.) (1978) Perceived self-efficacy: analyses of Bandura’s theory of behavioural change. Adu. Behac. Res. Ther. l(4). Rachman S. (19X0) Emotional processing. Behac. Res. Ther. 18, 51-60. Rathus S. A. (1973) A 30.item schedule for assessing assertive behavior. Behuv. Ther. 4, 3988406. Schwartz R. M. and Gottman J. M. (1976) Towards a task analysis of assertive behavior. J. consult. clin. Psychol. 44, 910-920. Skinner B. F. (1977) Why I am not a cognitive psychologist. Behaviorism 5, I-10. Staats A. W. (1981) Paradigmatic behaviorism. unified theory, unified theory construction methods, and the zeitgeist of separatism. Am. Psychal. 36, 2399256. Weinberg R. S.. Gould D. and Jackson A. (1979) Expectations and performance: an empirical test of Bandura’s self-efficacy theory, J. Spar/ P.sychol. 1, 320-331. Wolpe J. and Lar;aru\ A. A. (1966) Brharwur Thcrap), Techniques: A Guide IO the Treatment of Neuroses. Pergamon Press, Oxford.