the impact of feedback and self-efficacy on performance in training

16
JOURNAL OF ORGANIZATIONAL BEHAVIOR, VOL. 14,379-394 (1993) Research The impact of feedback and self-efficacy on performance in training Note KATHERINE A. KARL Department of Management. Western Michigan University. Kalamazoo. MI 49OO8. US. A. ANNE M. O’LEARY-KELLY Department of Management, Texas A & M University, College Station, TX 77843-4221, U. S.A. AND JOSEPH J. MARTOCCHIO Institute of Labor and Industrial Relations, University of Illinois at Urbana-Champaign. 504 E. Armory venue, Champaign, iL61am, U.S.A. Summary This field experiment examined the main and interactive effects of self-efficacy and feedback (i.e. the use of feedback versus no feedback) on performance in a speed reading class. Results showed that the provision of feedback was beneficial to the performance of all subjects,however as subject self-efficacy increased, the beneficial effects of feedback to the subject’s performance also increased. Subjects who received feedback on their performance experienced signi6cantly greater increases in self-efficacy than subjects who received no feedback. The more positive the performance feedback received, the greater the increase in individual self-efficacy. The implications of these results for training programs are discussed. Introduction Several authors (Baldwin and Ford, 1988; Noe, 1986; Wexley, 1984) have suggested that training effectivenessmay be improved by focusing upon the interaction between training design charac- teristics and individual difference variables. This approach suggests that training methods may be differentially effective for different trainees. In other words, some individuals will be more successful with one training technique whereas other individuals will be more successful with an alternative training technique, or, one training technique may be superior for all individuals, but one group of individuals may benefit considerably more from that technique than other individuals. Performance feedback is one training design characteristic that has received a great deal of research attention. Performance feedback (knowledge of results) provide trainees with infor- mation on whether their efforts toward performing a particular task have been successful. Several researchers have advocated the use of performance feedback to improve learning and perform- The authors wish to thank Daniel R. Ilgen and John R. Hollenbeck for their comments on an earlier version of this article. An earlier version was presented at the 33rd Midwest Academy of Management Meetings, Milwaukee, WI, April 18-21, 1990. Partial support for this research was provided by the Departments of Management and Psy- chology, Michigan State University. 0894-3796/93/040379-16$13.00 0 1993 by John Wiley & Sons, Ltd. Received 24 September 1991 Accepted 8 October I992

Upload: katherine-a-karl

Post on 06-Aug-2016

217 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: The impact of feedback and self-efficacy on performance in training

JOURNAL OF ORGANIZATIONAL BEHAVIOR, VOL. 14,379-394 (1993)

Research The impact of feedback and self-efficacy on performance in training Note

KATHERINE A. KARL Department of Management. Western Michigan University. Kalamazoo. MI 49OO8. US. A .

ANNE M. O’LEARY-KELLY Department of Management, Texas A & M University, College Station, TX 77843-4221, U. S.A.

AND JOSEPH J. MARTOCCHIO Institute of Labor and Industrial Relations, University of Illinois at Urbana-Champaign. 504 E. Armory venue, Champaign, iL61am, U.S.A.

Summary This field experiment examined the main and interactive effects of self-efficacy and feedback (i.e. the use of feedback versus no feedback) on performance in a speed reading class. Results showed that the provision of feedback was beneficial to the performance of all subjects, however as subject self-efficacy increased, the beneficial effects of feedback to the subject’s performance also increased. Subjects who received feedback on their performance experienced signi6cantly greater increases in self-efficacy than subjects who received no feedback. The more positive the performance feedback received, the greater the increase in individual self-efficacy. The implications of these results for training programs are discussed.

Introduction

Several authors (Baldwin and Ford, 1988; Noe, 1986; Wexley, 1984) have suggested that training effectiveness may be improved by focusing upon the interaction between training design charac- teristics and individual difference variables. This approach suggests that training methods may be differentially effective for different trainees. In other words, some individuals will be more successful with one training technique whereas other individuals will be more successful with an alternative training technique, or, one training technique may be superior for all individuals, but one group of individuals may benefit considerably more from that technique than other individuals.

Performance feedback is one training design characteristic that has received a great deal of research attention. Performance feedback (knowledge of results) provide trainees with infor- mation on whether their efforts toward performing a particular task have been successful. Several researchers have advocated the use of performance feedback to improve learning and perform-

The authors wish to thank Daniel R. Ilgen and John R. Hollenbeck for their comments on an earlier version of this article. An earlier version was presented at the 33rd Midwest Academy of Management Meetings, Milwaukee, WI, April 18-21, 1990. Partial support for this research was provided by the Departments of Management and Psy- chology, Michigan State University.

0894-3796/93/040379-16$13.00 0 1993 by John Wiley & Sons, Ltd.

Received 24 September 1991 Accepted 8 October I992

Page 2: The impact of feedback and self-efficacy on performance in training

380 K. A. KARL ET AL.

ance (Goldstein, 1986; Holding, 1965; Ilgen, Fisher and Taylor, 1979; Ilgen and Moore, 1987), and an abundant body of research clearly demonstrates that the provision of specific and timely feedback enhances individual performance (Ilgen et al., 1979).

Self-efficacy (a belief in one’s ability to perform a particular task, Bandura, 1982), has also been highlighted as an individual difference variable worthy of further study in the training context (Noe, 1986; Latham, 1988; Tannenbaum, Mathieu, Salas and Cannon-Bowers, 1991). Because previous research indicates that self-efficacy plays a crucial role in the initiation and maintenance of behaviour change (Bandura, 1986; Gist, Schwoerer and Rosen, 1989; Hill, Smith and Mann, 1987; Locke, Frederick, Lee and Bobko, 1984; Taylor, Locke, Lee and Gist, 1984), this focus on self-efficacy within the training area is warranted.

Given that performance feedback is a design characteristic that is important for maximizing trainee performance, and that four separate calls for research on self-efficacy within the context of training programs have recently appeared (Baldwin and Ford, 1988; Gist, 1987; Noe, 1986; Latham, 1988; Tannenbaum ei al., 1991), research that examines the interactive effect of trainee self-efficacy and performance feedback on trainee learning is timely (Day and Silverman, 1989; Chatman, 1989). The purpose of this research is: (a) to examine the impact of providing objective performance feedback on self-efficacy perceptions, (b) to examine the impact of the sign of feedback on self-efficacy perceptions, (c) to examine the extent to which the provision of objective performance feedback and self-efficacy interactively affect trainee performance, and, (d) to exam- ine the extent to which the sign of performance feedback and self-efficacy interactively affect trainee performance.

The impact of the provision of objective performance feed-back on self-eficacy percept ion According to Bandura (1986), self-efficacy perceptions are dynamic and change based upon the acquisition of new information and experience (enactive mastery). Because performance feedback conveys information regarding trainee knowledge and skills in relation to the task at hand, it follows that feedback will influence self-efficacy perceptions. Past research examining the relationship between feedback and self-efficacy has found that destructive criticism, as opposed to constructive criticism, lowers subsequent self-efficacy (Baron, 1988) and that employees had higher self-efficacy when their performance feedback came directly to them on their computer terminal than when it came from their supervisor (Earley, 1988). This study also found that getting feedback from a computer was associated with higher levels of trust in the feedback than getting it from the supervisor.

The studies reported above have examined the effects of subjective performance feedback on self-efficacy, while the focus of the present study is on the provision of objective performance feedback. One study that has examined the effect of the provision of objective performance feedback information on self-efficacy was conducted by Schunk (1983). This researcher found that elementary school children who received feedback on the number of pages completed had significantly higher self-efficacy and solved more problems correctly than children in the no-feedback condition. The results of this study provide preliminary evidence regarding the effect of provision of feedback on individual self-efficacy. Based on these findings, the following hypothesis is presented.

H 1 : Provision of performance feedback will influence self-efficacy such that individuals who receive performance feedback will show greater change in self-efficacy perceptions than will individuals who receive no performance feedback.

Page 3: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 381

The impact of the sign ofperformance feedback on self-eficacy perceptions The sign of performance feedback (i.e. positive versus negative feedback) is likely to be an important consideration (Ilgen et al., 1979) in predicting the direction of the change in self-efficacy perceptions. According to Bandura (1982, 1986) successful performance accomplishments increase self-efficacy perceptions whereas unsuccessful performance accomplishments decrease self-efficacy. It follows that positive performance feedback is more likely to enhance self-efficacy perceptions than is negative performance feedback. This prediction was supported in a study by Holroyd, Penzien, Hursey, Tobin, Rogers, Holm, Marcille, Hall and Chila (1984). These researchers told persons suffering from tension headaches that their headaches would decrease if they learned to control their facial muscles. Following a series of electromyographic biofeed- back sessions, participants received false feedback that they had achieved either a great deal of control over their facial muscles or very little control. Their results indicated that positive feedback led to higher perceptions of self-efficacy and the higher their perceived self-efficacy, the fewer headaches they reported. The actual amount of change in muscle activity achieved in treatment was unrelated to the report of subsequent headaches.

Research by Stone and Stone (1984, 1985) provides further evidence regarding the influence of sign of feedback on self-efficacy perceptions. In two independent laboratory studies the authors examined the impact of feedback favorability (acceptable versus superior) on self- perceived task competence in an in-basket exercise (study 1) and an assembly task (study 2). Their results showed that feedback favorability had a significant impact on self-perceived task competence such that subjects who received superior performance feedback had higher levels of self-perceived task competence than subjects who received acceptable performance feedback. Based on these findings it is hypothesized that:

H2: The sign of performance feedback will be positively correlated with change in self- efficacy perceptions .

Self-eflcacy, provision of feedback and trainee performance Previous research has not directly examined the possible interaction effect of self-efficacy and provision of feedback on trainee performance, however the relationship between self-esteem and performance feedback has been investigated. According to Gist and Mitchell (1992), self- esteem and self-efficacy differ in that self-esteem is a trait describing an individual’s affective self-evaluation, and self-efficacy is an individual’s perception of personal task capability (which is not inherently an evaluative perception). According to Brockner (1988), however, self-esteem and self-efficacy are constructs that are virtually synonymous and are highly related to one another. Similarly, Eden and Kinnar (1991) suggest that while there are slight conceptual differ- ences between the two, they are unlikely to be different operationally. Therefore, research on the interactive effects of self-esteem and performance feedback is likely to be relevant to the prediction of a self-efficacy-feedback effect on trainee performance.

Brockner’s (1988) plasticity hypothesis suggests that the behavior and attitudes of low self- esteem individuals are more susceptible to influence from their social environment than are those of high self-esteem individuals. As discussed by Eden and Kinnar (1991), the plasticity hypothesis has been supported in relation to several organizational variables (i.e. their effects are stronger for low self-esteem than high self-esteem individuals): role ambiguity and conflict (Mossholder, Bedeian and Armenakis, 1981), layoffs (Brockner, Davy and Carter, 1985), task preview information (Pond and Hay, 1987), and unemployment (Eden and Aviram, 1990; Shamir, 1986).

Page 4: The impact of feedback and self-efficacy on performance in training

382 K. A. KARL ET AL.

The plasticity hypothesis has also been supported in relation to performance feedback infor- mation. Brockner, Derr and Laing (1987) examined the effect of self-esteem and feedback on the exam performance of management students. Their results showed that there was no relation- ship between self-esteem and performance on the initial exam, however, the performance of low self-esteem individuals on the second exam varied considerably as a function of the feedback they received concerning their performance on the initial exam; this was less true for high self-esteem individuals.

Another relevant study examined the relationship between self-esteem and information search (Weiss and Knight, 1980). In this research, individual self-esteem influenced the search for task-related information, with low self-esteem subjects searching for more information than high self-esteem subjects. Consistent with the plasticity hypothesis, this suggest that low self- efficacy subjects seek performance information to a greater extent than do high self-efficacy subjects.

Given that self-esteem and self-efficacy are highly related constructs (Brockner, 1988; Eden and Kinnar, 1991), this research on self-esteem suggests that high self-efficacy trainees may be influenced less by performance feedback information than will low self-esteem trainees. There- fore, the following interaction was hypothesized:

H3: There will be a significant interaction between self-efficacy and the provision of per- formance feedback such that the relationship between the provision of feedback and performance will be greater for low self-efficacy individuals than for high self-efficacy individuals.

Serf-eficacy, sign of feedback and trainee performance Previous research suggests that high and low self-efficacy individuals respond differently to positive and negative feedback information. In regard to positive feedback, Gist and Mitchell (1992) suggested that there is a ceiling effect operating for high self-efficacy individuals. Because these individuals already hold high efficacy regarding the task, there is little room for positive change in their self-efficacy perceptions. For those with low efficacy, however, perceptions have more latitude for positive change and this positive change will be associated with a positive change in performance (Gist and Mitchell, 1992). Based upon this research, the following hypoth- esis is presented:

H4a: In response to positive feedback information, low self-efficacy individuals will show

In regards to negative feedback information, previous research (see Gist and Mitchell, 1992) suggests that different attributions are made by high versus low self-efficacy individuals (although their attributions for positive feedback are similar). This research shows that those high in self-efficacy are likely to attribute negative feedback to low effort or bad luck (factors that are perceived to be changeable), while those with low self-efficacy are more likely to attribute negative feedback to low ability (a factor that is not generally perceived to be changeable in the future). Because low self-efficacy individuals attribute failure to stable factors, they are likely to experience decreased performance as compared to high self-efficacy individuals who perceive that previous failure was due to unstable factors. Based upon this research, the following hypothesis is presented:

H4b In response to negative feedback information, low self-efficacy individuals will show

a greater increase in performance than will high self-efficacy individuals.

a greater decrease in performance than will high self-efficacy individuals.

Page 5: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 383

.26

.24

.22

.2a Abeolute ,18

(T1) .14 -12 .1c

Percent Change in Self-efficacy

.oa

.06

.28

.24

.22

Low Self-ef f icacy

\self-efficacy

I No Feedback Feedback

Feedback Condition

Ab8olute Percent Chanae *

in Self-efficaEy (T3) .14

.12

.10

.08

.06 I

No Feedback Feedback

Feedback Condition Figure 1. The interaction between self-efficacy and provision of feedback on absolute percent change

in self-efficacy

If self-efficacy and performance feedback interact in the manner predicted, this relationship has important implications for the design of training programs. While performance feedback has been shown to be beneficial to the subsequent performance of all individuals (Ilgen et al., 1979), it may be most essential to the performance of low self-efficacy subjects in order to maximize trainee performance. In other words, training sessions which might otherwise not include performance feedback because of time, cost, or manpower considerations, should incor- porate this technique for low self-efficacy trainees in order to maximize their performance.

In order to investigate the aforementioned issues, undergraduate student performance in a professional speed reading training course was examined. The need for understanding how to improve reading skills has never been greater from an original perspective (Goldstein and

Page 6: The impact of feedback and self-efficacy on performance in training

3&4 K. A. KARL ET AL.

Gilliam, 1990). Research has recently suggested that American schools are graduating students who lack the basic skills necessary for survival in organizations, including reading proficiency (Fiske, 1989). Consequently, many organizations have assumed the responsibility for providing employees with such basic skills (Fiske, 1989) and are therefore interested in understanding issues related to the training of reading skills. Thus, the sample employed here allows for examin- ation of the important training-related issues described in the previously stated hypotheses, while also allowing for the examination of a currently vital organizational issue: The training of reading skills.

Subjects Subjects in this study were 122 juniors and seniors enrolled in a general management course at a large midwestern university. All subjects received minimal course credit for their participa- tion. Fifty-seven per cent of the subjects were males (N=69) and 43 per cent were females (N = 53). The mean age was 21.4 (S.D. = 5.1).

Subjects were permitted to sign up for the training time which was most convenient, and this procedure determined the assignment of individuals to one of the two training treatments (feedback, no-feedback). Sixty-five subjects were assigned to the no-feedback condition and 57 subjects were assigned to the feedback condition. Since subjects were not randomly assigned to feedback conditions the two treatment groups were examined for systematic variation. Results indicated that the two treatment groups did not differ significantly on any of the demographic variables including sex, age, or class level, or on self-reported reading ability, comprehension, pre-training reading speed, or pre-training self-efficacy .

Training The training in both treatment groups consisted of a 1%-hour course in speed reading. The training materials used were the same materials used in the professional speed reading courses conducted by the university and the trainer was the director of the University’s Learning Resource Center who is an expert in the area of speed reading.

Measures Six variables were measured in this study: (a) Self-reported reading ability, (b) self-efficacy in relation to the speed reading task, (c) reading speed, (d) comprehension of the materials read, (e) provision of feedback, and (f) sign of feedback. Each of these is described in detail below.

Self-reported reading ability Self-reported reading ability was measured using a six-item scale adapted from the pre-training questionnaire used by the University’s Learning Resource Center. Sample items include ‘I have difficulty deciding what are the main points in reading materials’, ‘I have difficulty concentrating on what I read’, ‘I have difficulty remembering what I have read’ and, ‘I do not understand what I read without rereading’. Responses were measured using a 3-point scale (2 = seldom, 1 = sometime, 0 = often). Cronbach’s alpha for this scale was 0.67. The purpose of this measure was to examine the relationship between ability, self-efficacy, and performance improvement.

Page 7: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 385

It’s possible that high self-efficacy individuals with high ability may have little room for improve- ment (ie. a ceiling effect), or conversely, that low self-efficacy, low ability individuals may not be able to improve.

Self-efkac y Self-efficacy was measured using a five-item scale adapted from Hollenbeck and Brief (1987). Responses were measured using a 5-point Likert scale ranging from strongly disagree to strongly agree. Sample items include ‘I believe that speed reading is a skill that I can learn easily’, ‘I feel confident that I can learn to improve my reading speed’, and ‘It is just not possible for me to learn to read as fast as I would like’ (reverse scored). It should be noted that self-efficacy in relation to the subjects’ ability to learn speed reading, rather than their efficacy in relation to ability to perform speed reading was measured. This focus was appropriate given that subjects were involved in a training program. This measure was administered before training, after trial 1 and after trial 3. Cronbach’s alpha for this scale was 0.85, 0.88 and 0.88, respectively. Percentage change in self-efficacy was computed by subtracting pre-training self-efficacy from self-efficacy after trial 3 and dividing the result by pre-training self-efficacy.

Reading speed Pre-training reading speed was measured by the number of words read per minute by each subject on the pre-test. Post training reading speed was measured by computing the number of words read per minute on the post-test (after trial 4). Percentage change in reading speed was computed by subtracting pre-training reading speed from post training reading speed and dividing the result by pre-training reading speed.

Comprehension Three multiple choice questions were developed for each article used to assess reading speed (i.e. the pre-test, the post-test and the four practice trials). Comprehension was measured as the total number of correct answers on the six pre-test and post-test questions. The purpose of this measure was to control for inflated reading speeds by enabling us to identify subjects who did not understand or remember what they had read. Since all comprehension scores were acceptable, inflated reading speeds due to poor comprehension was not considered a signifi- cant problem in this study.

Provision of feedback Provision of feedback was a dichotomous variable coded 0 = without feedback and 1 = with feedback.

Sign of feedback Sign of feedback was measured as the positive or negative percentage change in words read per minute between the pre-test and trial 3.

Since sign of feedback is purely an objective measure and may not reflect how subjects interpret the feedback they receive, a 3-item manipulation check measure was administered after trial 3 to ensure that subjects were accurately interpreting the objective feedback received. The items included in this measure were: ‘I was very satisfied with my performance on this practice trial’, ‘My reading speed did not increase as much as I expected on this practice trial’ (reverse scored), and ‘My reading speed on this practice trial was a great improvement over my initial reading speed’. Responses were measured using a 5-point Likert scale ranging from strongly disagree to strongly agree. Cronbach’s alpha for this scale was 0.76. The correlation between the manipula-

Page 8: The impact of feedback and self-efficacy on performance in training

386 K. A. KARL ET AL.

tion check measure and sign of feedback was ( r = 0.37, p < 0.01). Thus, objectively positive feedback was associated with positive feedback perceptions.

Procedure The trainer opened the session with a general introduction and administered the pre-test measures in the following order: Demographic information, self-efficacy, and pre-training reading speed. The trainer then gave a 30-minute lecture on speed reading techniques (e.g. pre-reading, organiza- tional patterns, eye and hand movements). The remainder of each session was devoted to four practice trials in which the subjects were instructed to use the techniques discussed in the lecture. The reading material used in the practice trials, as well as that used in the pre-test and post-test, consisted of articles gathered from current business periodicals such as The Wall Street Journal, Business Week, and Fortune. In the feedback condition, subjects were asked to record the time required to read each article (the elapsed time was recorded on a flip chart at the front of the room). In addition, subjects were provided with a conversion chart and were asked to record the number of words per minute they had read after each trial. After recording the number of words read per minute, subjects completed the three-item comprehension measure.

In the no-feedback condition, subjects read each of the four practice articles and completed the four comprehension measures. They did not receive any external feedback regarding the number of words read per minute nor were any timing devices made available to them. In order to monitor changes in self-efficacy across practice trials, subjects in both conditions com- pleted the 5-item self-efficacy measure after the first and third practice trials.

Means and standard deviations by feedback condition and by level of self-efficacy are presented in Table 1. The high and low self-efficacy groups were determined by dividing the total sample at the mean on pre-training self-efficacy (18.24). This split resulted in 55 subjects in the low self-efficacy group and 61 in the high self-efficacy group (six subjects were eliminated due to missing data on the pre-training self-efficacy measure). Means, standard deviations, and intercor- relations among all variables are shown in Table 2.

As can be seen in Table 2, self-reported reading ability was not significantly related to perform- ance improvement (percentage change in reading speed) or change in self-efficacy. Thus, basic reading ability does not appear to play a major role in this study. One should note that the sample consisted of junior and senior level college students, thus, there was little variance in this sample in terms of reading ability.

Hypothesis 1 Hypothesis 1 stated that performance feedback would have a significant impact on absolute change in self-efficacy perceptions such that individuals who received performance feedback would show greater change in self-efficacy perceptions than individuals who did not receive performance feedback. This hypothesis was tested by computing the correlation between feed- back condition and absolute percentage change in self-efficacy from the pre-trial to after practice trial 1 and, between feedback condition and absolute percentage change in self-efficacy from the pre-trial to after practice trial 3. Neither correlation was significant (r = 0.09, p = 0.167, and r = 0.13, p = 0.085). Thus, hypothesis 1 was not supported. However, apost hoc examination

Page 9: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 387

Table 1. Means and standard deviations by training treatment condition and level of self-efficacy

Level of No feedback Feedback self-efficacy Measure M S.D. A4 S.D.

Self-reported ability 12.44 (2.33) 11.85 (2.11) Pre-training selfefficacy 16.11 (1.91) 15.19 (2.45) Self-efficacy (Tl) 16.46 (2.63) 16.07 (2.91) Self-efficacy (T3) 16.75 (2.53) 17.59 (3.07) Percentage change in self-efficacy (Tl) 0.03 (0.15) 0.09 (0.25)

Low Percentage change in self-efficacy (T3) 0.04 (0.15) 0.21 (0.37) Absolute percentage change in self-efficacy (Tl) 0.12 (0.10) 0.20 (0.17) Absolute percentage change in self-efficacy (T3) 0.12 (0.09) 0.26 (0.33) Sign of feedback - - 0.53 (0.31) Pre-training reading speed 241.70 (63.21) 229.26 (42.59) Post training reading speed 241.70 (63.21) 229.26 (42.59) Percentage change in reading speed 0.46 (0.32) 0.56 (0.25)

Training treatment

Self-reported ability 13.67 Pre-training self-efficacy 20.37 Self-efficacy (TI) 19.28 Self-efficacy (T3) 19.38 Percentage change in self-efficacy (Tl) -0.05

High Percentage change in self-efficacy (T3) -0.03 Absolute percentage change in self-efficacy (TI) 0.11 Absolute percentage change in self-efficacy (T3) 0.1 1

Pre-training reading speed 269.73 Post training reading speed 343.70 Percentage change in reading speed 0.33

Sign of feedback -

(2.1 1) (1.47) (3.06) (3.37) (0.14) (0.16) (0.10) (0.12)

(76.59) (71.57) (0.31)

-

13.00 (2.52) 20.77 (1.69) 20.40 (3.18) 20.81 (3.16) -0.02 (0.11)

0.00 (0.12) 0.08 (0.08) 0.08 (0.08) 0.74 (0.38)

249.00 (58.09) 415.61 (98.32)

0.69 (0.30)

of absolute change in self-efficacy by treatment and level of self-efficacy suggested the existence of an interaction. The results of an hierarchical moderated regression analysis (Stone and Hollen- beck, 1984) revealed a significant self-efficacy by feedback interaction such that the provision of feedback had a significant impact on the absolute percentage change in self-efficacy for low self-efficacy subjects but lesser impact on the absolute percentage change in self-efficacy for high self-efficacy subjects (AR2 for the interaction term = 0.084,~ < 0.001, when the absolute percentage change in self-efficacy between trial 1 and pre-training was the dependent variable, and AR2 = 0.112, p < 0.001, when the absolute percentage change in self-efficacy trial 3 and pre-training was the dependent variable) (see Figure 1).

Hypothesis 2 Hypothesis 2 stated that sign of feedback would be positively correlated with change in self- efficacy perceptions. Since within person analyses of self-efficacy increases should be subject to ceiling effects when pre-test self-efficacy is high (Gist and Mitchell, 1992), this hypothesis was tested by computing a partial correlation between sign of objective feedback (i.e. the percent- age change in words per minute between the pre-test and trial 3) and percentage change in self-efficacy, controlling for initial differences in pre-training self-efficacy. The partial correlation was significant (pr (55) = 0.28, p < 0.01). Thus, sign of feedback had a substantial impact on change in self-efficacy perceptions such that subjects who received more positive feedback had greater increases in self-efficacy. It should be noted that sign of feedback was positive

Page 10: The impact of feedback and self-efficacy on performance in training

Tab

le 2.

Means, s

tand

ard

devi

atio

ns a

nd in

terc

orre

latio

ns*

amon

g al

l var

iabl

es

Var

iabl

e M

S.

D.

1 2

3 4

5 6

7 8

9 10

11

12

13

1. 2.

3. 4.

5.

6.

7. 8.

9.

10.

11.

12.

13.

Prov

isio

n of f

eedb

ack?

-

Self-

repo

rted r

eadi

ng ab

ility

12

.94

Pre-

train

ing

self-

effic

acy

18.2

4 Se

lf-ef

ficac

y (T

I)

18.1

8 Se

lf-ef

ficac

y (T3

) 18

.93

Perc

enta

ge ch

ange

in se

lf-

Perc

enta

ge ch

ange

in se

lf-

effic

acy

(T3)

0.

05

Abs

olut

e per

cent

age c

hang

e in

self-

effic

acy (

Tl)

0.

13

Abs

olut

e per

cent

age c

hang

e in

self-

effic

acy (

T3)

0.14

Pre-

train

ing

read

ing s

peed

24

9.81

Po

st-tr

aini

ng re

adin

g sp

eed

366.

43

Perc

enta

ge ch

ange

in re

adin

g sp

eed

0.51

effic

acy (

Tl)

0.

01

Sign

of f

eedb

ack

0.64

-

-

-0.0

6 2.

31

-0.1

4 -

3.12

0.

13

0.17

3.

40

0.18

0.

07

3.43

0.

15

0.12

0.17

0.

13

0.00

0.23

0.

10

0.05

0.12

-0

.16

-0.0

9

0.19

-0

.14

-0.0

9 0.

36

- -

0.28

65

.38

-0.1

5 0.

14

90.7

8 0.3

911

0.05

0.32

0.5

211

-0.1

9

-0.2

1 -0

.07

0.15

0.

14

0.28

$ 0.

28$

0.28

$ 0.

13

0.16

0.

13

0.01

-0

.03

-0.1

2 -0

.04

-

0.421

1 0.

19

0.571

1 -0

.6811

-0.6411

-0.70

11

0.551

1 -

0.701

1 0.4

711

0.11

-0

.15

-0.0

9 0.4

611

0.791

1 -

0.441

1 0.5

211

0.06

0.

28$

0.07

0.8

711

0.671

1 -

0.821

1 0.5

711

0.691

1

-0.0

1 0.5

811

0.881

1 0.7

011

-

0.671

1 0.9

011

-0.0

1 -0

.5311

-0.4011

-0.62

11

-0.41

11

-

0.771

)

0.09

-0

.20

-0.2

2$

-0.2

5$

-0.2

5$

0.551

1 -

0.22

0.44$

0.24

0.

41$

0.16

-0

.29

-0.0

6 0.

20

0.07

-0

.03

-0.0

3 -0

.15

0.00

0.

11

0.27

$ 0.

31$

0.20

0.

20

0.07

-0

.18

-0.0

2

0.14

0.

27$

0.24

$ 0.

22$

0.20

-0

.17

-0.1

0

-

-0.1

2 0.

08

0.18

0.11

0.

14

-0.0

7 -0

.22$

0.561

1 -0

.01

0.06

0.

08

0.40

$ 0.

09

0.10

0.

25*

0.25

-0

.12

0.05

0.

19

0.06

-0

.20

-0.0

8 0.

13

0.15

-0

.12

0.01

0.

14

-0.2

4 -

0.661

1 -0

.21

0.47

8 0.

30$

0.33

1 0.

15

0.621

1 0.

06

0.08

0.

07

-

0.04

0.

548

0.751

)

0.28

0.

355

-

0.571

1

0.691

1 -0.

5211

0.5

711

-

*Cor

rela

tions

abov

e the

dia

gona

l are

for h

igh

self-

effic

acy i

ndiv

idua

ls an

d co

rrel

atio

ns be

low

the d

iago

nal a

re fo

r low

self-

effic

acy i

ndiv

idua

ls.

+Pro

visi

on of

feed

back

is a

dich

otom

ous v

aria

ble c

oded

0 =

with

out f

eedb

ack,

1 =

with

. sp

c 0.

05.

op c 0.

01.

Ilp <

0.00

1.

Page 11: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 389

for all subjects (the minimum percentage change in reading speed was + 2 per cent). The generali- zability of our conclusion, therefore, is limited to the use of positive (versus negative) feedback.

Hypothesis 3 Hypothesis 3 stated that there would be a significant interaction between pre-training self-efficacy and the provision of performance feedback such that the relationship between the provision of feedback and performance would be greater for low self-efficacy subjects than high self-efficacy subjects. An hierarchical moderated regression analysis was used to test this hypothesis. Perform- ance improvement, measured as the percentage change in reading speed from the pre-test to the post test was regressed on (a) feedback condition, (b) pre-training self-efficacy, and (c) the interaction between feedback condition and pre-training self-efficacy. As shown in Table 3, the interaction between feedback and self-efficacy explained a significant amount of variance in performance (R2 change = 0.086, p < 0.0001). Feedback was beneficial to the performance of all subjects, however contrary to what was hypothesized, the relationship between feedback and performance was greater for higher self-efficacy individuals than for lower self-efficacy individuals (see Figure 2).

Table 3. Regression results for the impact of provision of feedback and pre-training self-efficacy on percent- age change in reading speed

Step Variable Beta R R2 AR’

2 Pre-training self-efficacy 0.82 0.369 0.136 O.OO0 1 Feedback condition -1.41 0.369 0.136 0.136*

3 Feedback x self-efficacy -2.00 0.47 1 0.222 0.086*

* p c 0.001

Hypotheses 4a and 4b Hypotheses 4a and 4b concerned the relationship between sign of feedback, self-efficacy and performance improvement, Since sign of feedback in this study was positive for all subjects, the results of hypothesis 4a, which stated that low self-efficacy individuals would show a greater improvement in performance in response to positive feedback than high self-efficacy individuals, are the same as that reported for hypothesis 3. Furthermore, hypothesis 4b which stated that low self-efficacy individuals, in response to negative feedback, would show a greater decrease in performance than high self-efficacy individuals, could not be tested.

Discussion This study has extended previous research by examining the relationship between self-efficacy, feedback, and performance in a training setting. Hypothesis 1 suggested that those subjects receiving feedback on their performance would show a greater change in self-efficacy perceptions than those subjects who received no performance feedback. This hypothesis was not supported by these data. However, a post hoc analysis revealed a significant interaction such that the provision of feedback had a significant impact on change in self-efficacy perceptions for low

Page 12: The impact of feedback and self-efficacy on performance in training

390 K. A. KARL ET AL.

Percent Change in

Reading Speed

-70

.65

.60 5.5 .50 .45 .40 .3 5 .30 .25 .20 .15 .10

High Self-eft icacy

Low Self -ef f icacy

.05 I No Feedback Feedback

Feedback Condition Figure 2. The interaction between self-efficacy and provision of feedback on percent change in reading

speed

self-efficacy individuals but not high self-efficacy individuals. Consistent with the plasticity hypothesis (Brockner, 1988), the self-efficacy of low self-efficacy individuals was more susceptible to influence from external cues than that of high self-efficacy individuals.

Hypothesis 2 suggested that sign of feedback would be positively correlated with change in self-efficacy perceptions. A significant positive correlation was found. Taken together, the results of hypotheses 1 and 2 indicate that, in general, the feedback message rather than the feedback per se has important implications for influencing self-efficacy perceptions. Consistent with social cognitive theory (Bandura, 1986), performance feedback provides information about prior performance and serves as a basis for evaluating one’s capability to perform successfully on subsequent tasks. Negative feedback provides information that an individual’s performance has fallen below an acceptable standard and positive feedback indicates that an individual’s performance has exceeded an acceptable standard. Therefore, to influence self-efficacy percep- tions, the content of the feedback message is critical (Baron, 1988; Gist and Mitchell, 1992). However, with regard to the malleability of self-efficacy (Gist and Mitchell, 1992), the provision of feedback (regardless of sign) is more likely to lead to changes in self-efficacy perceptions for low self-efficacy individuals than high self-efficacy individuals.

Hypotheses 3 and 4 suggested that feedback would have a greater impact on the performance of low self-efficacy individuals than high self-efficacy individuals. A significant self-efficacy- feedback interaction was found, however feedback had a greater impact on the performance of high self-efficacy subjects than low self-efficacy subjects. While this finding is in conflict with previous self-esteem research (Brockner et al., 1987; Weiss and Knight, 1980), it is consistent with recent goal setting research (Locke and Latham, 1990). In their view of studies examining the relationship between goals and feedback, Locke and Latham (1990) concluded that self- efficacy and dissatisfaction with goal-performance discrepancies influences subsequent perform- ance attainment. Specifically, low self-efficacy subjects do not improve in response to goal-

Page 13: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 391

relevant feedback whereas high self-efficacy subjects who are highly dissatisfied with the feedback they receive show the greatest improvement. Although subjects in this study were not assigned goals nor provided with standards for assessing their progress, it is possible that subjects set their own goals. Given the consistent positive relationship between self-efficacy and goal level chosen (see Locke and Latham, 1990), it is also possible that high self-efficacy subjects set higher goals than low self-efficacy subjects. The higher goals set by high self-efficacy subjects may have created higher goal-performance discrepancies and greater motivation to improve subsequent performance. Another explanation is that the studies cited earlier (Brockner et al., 1987; Weiss and Knight, 1980) examined subject self-esteem rather than subject self-efficacy. Perhaps there are greater differences between these two constructs than was origmally assumed. It is also possible that low self-efficacy individuals are less likely to strive to improve their performance because they perceive a tenuous link between effort and performance due lo felt ability deficiencies. High self-efficacy individuals, on the other hand, may be more likely to strive to improve their performance because they perceive a more direct link between effort and performance (Bandura, 1986).

Practical implications The findings of the present study have several practicaI implications for training professionals. In agreement with previous research (Ilgen et al., 1979), performance feedback had a positive impact on the performance of all individuals. However, this study found that feedback had a greater impact on the performance of high self-efficacy individuals, suggesting that trainers should focus on raising the self-efficacy of low self-efficacy individuals prior to training so that these individuals will receive the full benefits provided by performance feedback. Bandura (1 986) had identified four factors that influence self-efficacy: Enactive mastery (repeated perform- ance successes), vicarious experience (observing a successful model), verbal persuasion, and emotional arousal (changes in anxiety associated with performing the task). It is possible that the self-efficacy of low self-efficacy individuals could be raised in a pre-training session that focuses on convincing trainees that they are capable of performing the task (verbal persuasion); that provides models successfully learning to perform the task (vicarious experience); and/or that focuses on decreasing the anxiety associated with the task (emotional arousal). Another strategy would be to initially provide easy tasks to low self-efficacy trainees (i.e. enactive mastery) to ensure success or enhance positive feedback which in this study was shown to influence self-efficacy perceptions.

Limitat ions Limitations of the present research should be discussed. The first limitation involves the external validity of the results obtained here. This research involved the training of college undergraduates in a speed reading task. Although the study of reading skills in an educational setting is appropri- ate and necessary (Fiske, 1989), further research is needed to assess the extent to which findings from this sample, setting, and task are generalizable to the training of employees in organizational settings on a variety of tasks.

A second limitation of this study is that all subjects received positive feedback, and, therefore the generalizability of our conclusions is limited to the use of positive feedback. It would be interesting to examine whether receiving negative feedback is more or less beneficial to receiving no feedback at all in terms of its impact on self-efficacy.

Page 14: The impact of feedback and self-efficacy on performance in training

392 K. A. KARL ET AL.

A third possible limitation relates directly to the complexity of the feedback stimulus (Ilgen et al., 1979). In particular, the source of the feedback needs to be considered with regard to whether it is perceived as credible by recipients. While we did not asess source credibility, we feel confident that our subjects perceived feedback sources as credible because they elected to take this training from a noted instructor in speed reading techniques.

Finally, our subjects were not randomly assigned to the training conditions. Although we showed no significant differences between our subjects on demographic factors or pre-training administrations of our outcome variables, it is possible that other unmeasured differences were operating which could have influenced the results. However, there is no guarantee that the influence of extraneous systematic factors would be neutralized even if subjects had been randomly assigned to training treatments (Cook and Campbell, 1979).

Future research Future research should explore the underlying processes behind the findings reported here. For example, a recent study by Alden (1986) found that individuals tend to interpret feedback in a manner consistent with their self-efficacy perceptions. Specifically, she found that low self- efficacy individuals were more likely to make external (situational) attributions for their perform- ance. Thus, positive feedback may have had less impact on the performance of low self-efficacy individuals in this study, as low self-efficacy individuals may have discounted positive feedback as inaccurate because it was inconsistent with their efficacy expectations. Future research should incorporate attributional measures to test this proposition more directly.

High self-efficacy individuals who are confident in their ability to learn a particular skill may also have a strong desire to achieve that skill. Research on need for achievement shows that high need for achievement individuals have a strong desire for immediate and specific feedback and will perform better in situations where feedback is readily available (McClelland, 1962). Thus, if high self-efficacy subjects also tend to have a high need for achievement this could explain why feedback had a stronger impact on high self-efficacy individuals than low self-efficacy individuals. Future research should examine the relationship between self-efficacy and need for achievement more directly.

A recent review by Locke and Latham (1990) suggests that a number of mediating links between feedback and performance include goal setting, self-efficacy, and dissatisfaction with the perceived goal-performance discrepancy. As noted earlier, high self-efficacy individuals, in this study may have performed better than low self-efficacy individuals in the feedback con- dition because they set higher goals than low self-efficacy individuals. It is also possible that low self-efficacy subjects were not sufficiently dissatisfied with low performance feedback to change their effort because low performance fits with their self-efficacy perceptions. High self- efficacy individuals, on the other hand, may be sufficiently dissatisfied with low performance feedback to change their effort because low performance does not fit with their self-efficacy perceptions. Because subjects were not asked whether or not they set goals for themselves, it is impossible to examine whether individual goal setting may be another important variable in the process underlying the relationship between self-efficacy, feedback and performance. Future research should examine the extent to which feedback influences goal setting in training situations.

Finally, this study accentuates the importance of examining the interaction between individual difference variables and training design characteristics (Baldwin and Ford, 1988; Wexley, 1984 1. Future research should examine how other individual differences in personality or cognitive style (e.g. locus of control, field-dependence, self-esteem, cognitive complexity) interact with

Page 15: The impact of feedback and self-efficacy on performance in training

FEEDBACK AND SELF-EFFICACY 393

training design characteristics (e.g. feedback, modeling, goal setting, practice) to influence train- ing outcomes.

Conclusions This study supports the predictions of Noe (1986), Gist (1987) and Latham (1988) that self- efficacy may pinpoint specific training needs. It was demonstrated that low self-efficacy indivi- duals benefit less from performance feedback than high self-efficacy individuals, and that low self-efficacy trainees may therefore require pre-training sessions to increase their self-efficacy if the maximum benefits of providing performance feedback are to be realized. Future research should focus on clarifying our understanding of the processes underlying this relationship.

References Alden, L. (1986). ‘Self-efficacy and causal attributions for social feedback’, Journal of Research in Persona-

Baldwin, T. T. and Ford, J. K. (1988). ‘Transfer of training: A review and directions for future research’,

Bandura, A. (1982). ‘Self-efficacy mechanism in human agency’, American Psychologist, 37,122-147. Bandura, A. (1986). Social Foundations of Thought and Action: A Social-cognitive View, Prentice-Hall,

Baron, R. A. (1988). ‘Negative effects of destructive criticism: Impact on conflict, self-efficacy, and task

Brockner, J. (1 988). Self-Esteem at Work, Lexington Books, Lexington, MA. Brockner, J., Davy, J. and Carter, C. (1985). ‘Layoffs, self-esteem, and survivor guilt: Motivational, affec-

tive, and attitudinal consequences’, Organizational Behavior and Human Decision Processes, 36,229-244. Brockner, J., Derr, W. R. and Laing, W. N. (1987). ‘Self-esteem and reactions to negative feedback:

Towards greater generalizability’, Journal of Research in Personality, 21,318-333. Chatman, J. A. (1989). ‘Improving interactional organizational research: A model of person-organization

fit’, Academy of Management Review, 14,333-349. Cook, T. D. and Campbell, D. T. (1979). Quasi-Experimentation: Design and Analysis Issues for Field

Settings, Houghton Mifftin Company, Boston. Day, D. V. and Silverman, S. B. (1989). ‘Personality and job performance: Evidence of incremental validity’,

Personnel Psycho/ogy, 42,25-36. Earley, P. C. (1988). ‘Computer-generated performance feedback in the magazine-subscription industry’,

Organizational Behavior and Human Decision Processes, 41,50-64. Eden, D. and Aviram, A. (1990). Self-Eflfcaey Training to Speed Reemployment, Working Paper No.

72/90, Tel Aviv University, Faculty of Management, Tel Aviv, Israel. Eden, D. and Kinnar, J. (1991). ‘Modeling Galatea: Boosting self-efficacy to increase volunteering’, Journal

of Applied Psychology, 76,770-780. Fiske, E. B. (1989, September 25). ‘Impending U.S. Jobs “Disaster”: Work force unqualified to work’,

The New York Times, p. 1. Gist, M. E. (1987). Self-efficacy: Implications for Organizational Behavior and Human Resource Manage-

ment. Academy of Management Review, 12,472-485. Gist, M. E. and Mitchell, T. R. (1992). ‘Self-efficacy: A theoretical analysis of its determinants and malleabi-

lity’, Academy ofManagement Review, 17,183-21 1. Gist, M. E., Schwoerer, C. and Rosen, B. (1989). ‘Effects of alternative training methods on self-efficacy

and performance in computer software training’, Journal of Applied Psychology, 74,884-891. Goldstein, 1. L. (1986). Training in Organizations, 2nd edn, Brooks-Cole, Monterey, CA. Goldstein, I. L. and Gilliam, G. (1990). ‘Training system issues in the year 2000’, American Psychologist,

Hill, T., Smith, N. D. and Mann, M. F. (1987). ‘Role of efficacy expectations in predicting the decision

Holding, D. H. (1965). Principles of Training, Pergamon Press, London.

lity, 20,460-473.

Personnel Psychology, 41,63-105.

Englewood Cliffs, NJ.

performance’, Journal of Applied Psychology, 73,199-207.

45,134-143.

to use advanced technologies: The case of computers’, Journal of Applied Psychology, 72,307-3 13.

Page 16: The impact of feedback and self-efficacy on performance in training

394 K. A. KARL ET AL.

Hollenbeck, J. R. and Brief, A. P. (1987). ‘The effects of individual differences and goal origin on goal setting and performance’, Organizational Behavior and Human Decision Processes, 40,392-4 14.

Holroyd, I(. A., Penzien, D. B., Hursey, K. G., Tobin, D. L., Rogers, L., Holm, J. E., Marcille, P. J., Hall, J. R. and Chila, A. G. (1984). ‘Change mechanisms, in EMG biofeedback training: Cognitive changes underlying improvements in tension headache’, Journal of Consulting and Clinical Psychology, 52, 1039- 1053.

Ilgen, D. R., Fisher, C. D. and Taylor, M. S. (1979). ‘Consequences of individual feedback on behavior - . in organizations’, Journal of Applied Psychology, 64,349-371.

Psychokigy, 72,401-406. Ilgen, D. R. and Moore, C. F. (1987). ‘Types and choices of performance feedback’, JournaL of Applied

Latham, ti. P. (1988). ‘Human resource training and development’, Annual Review of Psychology, 39,

Locke, E. 4., Frederick, E., Lee, C. and Bobko, P. (1984). ‘Effects of self-efficacy, goals and task strategies

Locke, E. A,, and Latham, G. P. (1990). A Theory of Goal Setting and Task Performance. Englewood

McClelland, D. C. (1962). ‘Business drive and actual achievement’, Harvard Business Review, 99-1 12. Mossholder, K. W., Bedeian, A. G . and Armenakis, A. A. (1981). ‘Role perceptions, satisfaction, and

performance: moderating effects of self-esteem and organizational level.’ Organizational Behavior and Human Performance, 28,224-234.

Pond, S . R., 111 and Hay, M. S. (1987, April). ‘The impact of task preview information as a function of recipient self-efficacy’. Paper presented at the Second Annual Conference of the Society for Industrial1 Organizational Psychology, Atlanta, GA.

Noe, R. ‘4. (1986). ‘Trainees’ attributes and attitudes: Neglected influences on training effectiveness’, Academji of Management Review, 11,736-749.

Schunk, D. H. (1983). ‘Progress self-monitoring: Effects of children’s self-efficacy and achievement’, Journal of Experimental Education, 51, 89-93.

Shamir, B. (1986). ‘Self-esteem and the psychological impact of unemployment’, Social Psychology Quar- terl-v, 49,61-72.

Stanley, J. C. (1967). ‘General and special formulas for reliability of differences’, Journal of Educational Measurrmenf, 4,249-252.

Stone, E. F. and Hollenbeck, J. R. (1984). ‘Some issues associated with the use of moderated regression’, Organizational Behavior and Human Performance, 34,195-213.

Stone, E. F. and Stone. D. L. (1984). ‘The effects ofmultiule sources of Derformance feedback and feedback

545-582.

on task performance’, Journal of Applied Psychology, 69,241-25 1.

Cliffs, N.J.: Prentice Hall.

favorability on self-perceived task competence and perceived feedback accuracy’, Journal of Management, 10.371 -378.

Stone, D. L. and Stone, E. F. (1985). ‘The effects of feedback consistency and feedback favorability on self-perceived task competence and perceived feedback accuracy’, Organizational Behavior and Human Decision Processes, 36, 167-185.

Tannenbaum, S. I . , Mathieu, J. E., Salas, E. and Cannon-Bowers, J. A. (1991). ‘Meeting trainees’ expec- tations: The influence of training fulfillment on the development ofcommitment, self-efficacy, and motiva- tion’, Journal of Applied Psychology, 76,759-769.

Taylor, M. S., Locke, E. A., Lee, C. and Gist, M. E. (1984). ‘Type A behavior and faculty research productivity: What are the mechanisms?’ Organizational Behavior and Human Performance, 3 4 , 4 0 2 4 8.

Weiss, H. M. and Knight, P. A. (1980). ‘The utility of humility: Self-esteem, information search, and problem-solving efficacy’, Organizational Behavior and Human Performance, 25,216-223.

Wexley, K. N. (1984). ‘Personnel training’, Annual Review of Psychology, 35, 519-551.