gender differences in supervisors’ use of performance feedback

18
Gender Differences in Supervisors’ Use of Performance Feedback’ NEIL BREWER,2 LYNNE SOCHA, AND ROB POTTER Flinders University of South Australia This study examined whether male and female supervisors differed in their delivery of performance feedback to subordinates. Male and female subjects supervised 2 confederate subordinates whose performance was stable and either above or below average. Supervisors regularly checked each subordinate’s performance and could deliver one of several feedback messages. A MANOVA was performed on frequency of specific negative, general negative, specific positive, general positive, and neutral feedback messages, followed by stepdown analyses to isolate which dependent vari- ables contributed uniquely. Male and female supervisors were distinguished by their use of specific negative feedback. Males were more likely to provide such messages to poorly performing subordinates, a result consistent with suggestions that males are characterized by a more directive leadership style. One of the important functions of a supervisor is to provide subordinates with feedback that informs them about how they are performing on the job. Feedback represents a means by which supervisors can inform subordinates how they are progressing in relation to set goals, recognize or acknowledge improving or good performance, and correct any errors or deficiencies in subordinate performance. The principal dimensions along which feedback messages can vary include sign (positive, negative, neutral), specificity (gen- eral, specific), frequency, timing, and affect (cf. Larson, 1986). Not surprisingly, a considerable amount of research has been concerned with the effects of feedback on performance and factors influencing feedback delivery (Larson, 1984; Nadler, 1979). In this study, we focused on whether or not there are gender-related differences in feedback-giving behavior. The existence of such differences is likely to have implications for supervisory effectiveness, the training and evaluation of supervisors, and even for the training and evaluation of subordinates. ‘This research was supported by grants from the Flinders University of South Australia Research Budget and the Australian Research Council. The authors gratefully acknowledge the assistance of Alli, Tony, and David, who willingly acted as confederates. 2Correspondence concerning this article should be addressed to Neil Brewer, School of Psychology, The Flinders University of South Australia, GPO Box 2100, Adelaide, South Aus- tralia 5001, Australia. 786 Journal of Applied Social Psychology, 1996, 26, 9, pp. 786-803. Copyright 0 1996 by V. H. Winston 8 Son, Inc. All rights reserved.

Upload: neil-brewer

Post on 20-Jul-2016

230 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Gender Differences in Supervisors’ Use of Performance Feedback

Gender Differences in Supervisors’ Use of Performance Feedback’

NEIL BREWER,2 LYNNE SOCHA, AND ROB POTTER Flinders University of South Australia

This study examined whether male and female supervisors differed in their delivery of performance feedback to subordinates. Male and female subjects supervised 2 confederate subordinates whose performance was stable and either above or below average. Supervisors regularly checked each subordinate’s performance and could deliver one of several feedback messages. A MANOVA was performed on frequency of specific negative, general negative, specific positive, general positive, and neutral feedback messages, followed by stepdown analyses to isolate which dependent vari- ables contributed uniquely. Male and female supervisors were distinguished by their use of specific negative feedback. Males were more likely to provide such messages to poorly performing subordinates, a result consistent with suggestions that males are characterized by a more directive leadership style.

One of the important functions of a supervisor is to provide subordinates with feedback that informs them about how they are performing on the job. Feedback represents a means by which supervisors can inform subordinates how they are progressing in relation to set goals, recognize or acknowledge improving or good performance, and correct any errors or deficiencies in subordinate performance. The principal dimensions along which feedback messages can vary include sign (positive, negative, neutral), specificity (gen- eral, specific), frequency, timing, and affect (cf. Larson, 1986).

Not surprisingly, a considerable amount of research has been concerned with the effects of feedback on performance and factors influencing feedback delivery (Larson, 1984; Nadler, 1979). In this study, we focused on whether or not there are gender-related differences in feedback-giving behavior. The existence of such differences is likely to have implications for supervisory effectiveness, the training and evaluation of supervisors, and even for the training and evaluation of subordinates.

‘This research was supported by grants from the Flinders University of South Australia Research Budget and the Australian Research Council. The authors gratefully acknowledge the assistance of Alli, Tony, and David, who willingly acted as confederates.

2Correspondence concerning this article should be addressed to Neil Brewer, School of Psychology, The Flinders University of South Australia, GPO Box 2100, Adelaide, South Aus- tralia 5001, Australia.

786

Journal of Applied Social Psychology, 1996, 26, 9, pp. 786-803. Copyright 0 1996 by V. H. Winston 8 Son, Inc. All rights reserved.

Page 2: Gender Differences in Supervisors’ Use of Performance Feedback

GENDER AND PERFORMANCE FEEDBACK 787

Models of supervision and leadership typically include behaviors associ- ated with the provision of performance feedback or consequences as key elements of the supervisory process (e.g., Fleishman, 1953; Komaki, Zlotnick, & Jensen, 1986; Likert, 1967; Luthans & Lockwood, 1984; Mintzberg, 1973; Yukl, 1989). The central role assigned to feedback in such models reflects its acknowledged importance in shaping and refining new skills and behaviors in subordinates, and in maintaining and strengthening established behaviors (cf. Erez, 1977; Larson & Callahan, 1990; Nadler, 1979; Podsakoff, Todor, & Skov, 1982).

Despite the acknowledged importance of feedback in the supervisory proc- ess, we know that supervisors’ use of performance feedback is characterized by considerable variability both within and between supervisors (Brewer, Wilson, & Beck, 1994; Dansereau, Graen, & Haga, 1975; Katerberg & Hom, 198 1 ; Komaki, Desselles, & Bowman, 1989; Larson, 1984). Yet systematic examination of the determinants of feedback delivery has been limited (Larson, 1984). Several studies have explored the relationship between level of subordinate perform- ance and the delivery of different types of feedback message (e.g., Fisher, 1979; Hobson, 1986; Larson, 1986). A number of studies have also examined the relationships between supervisors’ attributions regarding the causes of subordinate performance and their administration of organizational rewards and punishments or feedback (e.g., Dobbins, Pence, Orban, & Sgro, 1983; Green & Mitchell, 1979). Despite the likelihood that individual differences between supervisors will exert a significant influence on the nature and frequency of feedback delivery (cf. Larson, 1984), exploration of the contribution of indi- vidual difference variables has attracted little empirical investigation.

One individual difference variable which has attracted considerable atten- tion in the broader literature on leadership and supervision is gender. There has been considerable discussion in both the popular and scientific literatures regarding the existence of stable gender differences in leadership style. The psychological literature generally has been less supportive of the popular notion that males and females have distinctive leadership styles. While reviews (e.g., Bass, 1990; Nieva & Gutek, 1981) have noted occasional findings of gender differences in style (e.g., Bass, 1985), they have also pointed to a broad array of evidence in which such differences have not emerged. Nevertheless, recent evidence from a meta-analysis of leadership style studies (Eagly & Johnson, 1990) suggests a different conclusion from that of some previous general reviews. Eagly and Johnson found that, across all studies, women’s leadership styles were characterized by a slightly greater concern for both interpersonal relations and task accomplishment. These differences were much less evident in organizational settings than in the laboratory, a finding which was attributed to organizational selection and socialization processes leveling

Page 3: Gender Differences in Supervisors’ Use of Performance Feedback

788 BREWER, SOCHA, AND POTTER

out any tendencies to exhibit gender stereotypic styles. They also found that, regardless of the setting, women were characterized by a more democratic or participative style, whereas men were more autocratic or directive. Eagly and Johnson speculated that this difference might reflect gender differences in social skills among those selected into leadership roles, with women perhaps possessing a repertoire of skills that allowed them to behave in a more demo- cratic or participative manner. They also suggested that women may need to exhibit a more democratic style than men because of attitudinal biases against women exhibiting directive behaviors.

As Eagly and Johnson (1990) acknowledged, predicting how such differ- ences in leadership style might impact on leader effectiveness is highly prob- lematic because of the likely interactions between style and various task and organizational variables. There are also difficulties associated with spelling out precisely how such differences in leadership style might be translated into leader behaviors. The leadership style dimensions tapped in those studies sampled in the Eagly and Johnson meta-analysis rely principally on self-reports or the reports of others on a diverse array of behaviors that, ultimately, are clustered together by the measurement instrument in order to characterize leaders on a few broad dimensions or categories. How accurately self-reports or the reports of others predict actual behaviors, and how differences in quite broad style dimensions translate into specific behaviors such as giving per- formance feedback are questions that can only be answered by careful investi- gation. Nevertheless, the research on leadership style does provide a basis for some hypotheses which can be used to guide our investigations.

The most marked gender difference in leadership style to emerge from Eagly and Johnson’s (1990) meta-analysis was the tendency for males to be more directive or autocratic. Thus, when subordinates are performing below expectations, we might expect that male supervisors would be more likely to step in quickly to tell them that their performance is unsatisfactory, to indicate exactly what they should be doing, or both. When confronted with ongoing poor performance from subordinates, males should, therefore, use general negative feedback (i.e., indications that performance is clearly unsatisfactory) more frequently than females, and they also should use corrective or specific negative feedback more liberally. Further support for this prediction is pro- vided by Bass’ (1985) finding that male managers are more likely to be characterized as “managing by exception” (Le., by taking corrective action).

Eagly and Johnson (1990) also found that consideration or concern for interpersonal relations and emphasis on task accomplishment was slightly greater in female than in male leaders. While measures of consideration do not suggest direct predictions about feedback delivery, inspection of the instru- ments for measuring initiating structure or task focus (e.g., LBDQ, LOQ)

Page 4: Gender Differences in Supervisors’ Use of Performance Feedback

GENDER AND PERFORMANCE FEEDBACK 789

clearly indicates that a greater emphasis on task accomplishment should be associated with more extensive use of encouraging feedback (e.g., typical initiating structure items focus on behaviors such as getting subordinates to work to capacity, encouraging them to work harder, and encouraging after- hours work). This should be reflected in females providing more specific positive feedback for good performers and more general encouragement (i.e., general positive or neutral feedback) for both good and poor performers. Thus, the present study incorporated the independent variables of gender and per- formance (goodpoor) with dependent measures designed to tap general nega- tive feedback, corrective or specific negative feedback, specific positive feedback, and measures of encouragement (general positive or neutral feed- back) for subordinates. These dependent measures allowed us to examine several major dimensions of feedback delivery: sign (positive, negative, neu- tral), specificity (general, specific), and, of course, frequency and timing. The other major dimension, affect, was held constant.

There are also grounds for suggesting that the gender of the supervisor may interact with that of the subordinate to influence feedback delivery. Carli (1989), for example, found that when men and women discussed topics on which they disagreed, the nature of the dyadic interactions (i.e., percentage of task behaviors, agreements, disagreements, positive social behaviors) varied depending on whether the partner was of the same or the opposite gender, with gender stereotypic interactions more prominent in same-gender than in mixed- gender pairs. Also, the nature of their attempts to influence their partner varied with the partner’s gender.

Other indications of the importance of subordinate gender have emerged from studies exploring the contributions of attributional theories to the expla- nation of supervisory behavior. In these studies, the focus has been on the extent to which supervisors differentially attribute responsibility for the per- formance outcomes of male and female subordinates to internal and external factors, and the impact of these attributions on supervisor-subordinate interac- tions. Specifically, Dobbins and his colleagues (Dobbins, 1986; Dobbins et al., 1983) found that, when provided with a description of a performance lapse by a subordinate, males and females playing the role of supervisor rated the likelihood of their implementing, or the appropriateness of, various corrective actions differently depending on whether the subordinate was male or female. In one study, harsher corrective measures were more likely to be implemented against female subordinates (Dobbins et al., 1983). In another (Dobbins, 1986), poorly performing female subordinates were treated less harshly than were males by female supervisors, whereas males and females were treated similarly by male supervisors. Since these studies lead to conflicting predictions about likely gender differences in feedback delivery, and none of them has actually

Page 5: Gender Differences in Supervisors’ Use of Performance Feedback

790 BREWER, SOCHA, AND POTTER

examined the ongoing use of performance feedback by individuals in a super- visory role, it was decided to control for subordinate gender by including it as an independent variable in the design.

Although some studies (e.g., Baker, DiMarco, & Scott, 1975; Szilagyi, 1980) have looked at the influence of gender on the delivery and withdrawal of rewards (with no differences being reported), most of the previous investiga- tions of the actual use of performance feedback by supervisors have used male supervisors only and have not examined gender differences in feedback deliv- ery (e.g., Fisher, 1979; Hobson, 1986). One study that did examine gender differences in feedback delivery (Larson, 1986) reported no significant effect. In Larson’s study, all supervisor-subordinate pairs were of the same gender and the dependent measures did not, for example, distinguish between positive, general negative, and corrective components, focusing instead on overall feed- back frequency and specificity. These factors may have contributed to gender differences being obscured. The focus of the present study was, therefore, to reexamine the issue of gender differences in feedback delivery using a design that was more sensitive to any differences that may exist. Supervisor gender, subordinate gender, and subordinate performance level were included as inde- pendent variables, and the dependent variables were the frequency of use of a number of different feedback messages (specific negative, general negative, specific positive, general positive, and neutral feedback or encouragement).

It was hypothesized that male supervisors, when confronted by poor per- formance, would use general negative and specific negative or corrective feedback sooner and more frequently. It was also hypothesized that the use of specific positive feedback for good performance and encouraging feedback (i.e., general positive and neutral) for both good and poor performance would be more prevalent among female supervisors. Earlier research into the relation- ships between the level of subordinates’ performance and feedback delivery (cf. Hobson, 1986; Larson, 1986) also suggested the following minor hypothe- ses. Overall, poor performers were expected to receive more feedback than good performers. Although it was expected that poor performers would seldom receive specific positive feedback, they were expected to receive encouraging (i.e., general positive and neutral) feedback in addition to any specific and general negative feedback; good performers were not, however, expected to receive either kind of negative feedback.

The study involved a laboratory experiment, using a paradigm loosely based on that employed by Larson (1986). Male and female subjects each supervised two confederate subordinates (one of each gender) while they worked on a set of puzzles for 10 short work periods. Subordinates maintained stable performance, with both members of each confederate pair performing either well or poorly. At the end of each work period, supervisors scored each

Page 6: Gender Differences in Supervisors’ Use of Performance Feedback

GENDERANDPERFORMANCEFEEDBACK 791

subordinate’s puzzle and had the option of providing one of the five general types of feedback message to the subordinate.

Method

Subjects, Setting, and Design

Thirty male and 30 female undergraduate students (two of each gender were Asian born, the rest were Australian born) participated as supervisors in the study. Subjects were selected from a pool of students who had volunteered to participate in psychology experiments. Four subjects from the original sample had to be replaced after postexperimental interviews revealed that they suspected that their behavior, rather than the subordinates’ performance, was the focus of the study. The study was conducted in a suite of cubicles, with the subject and the two confederate subordinates each located in separate cubicles.

All subjects supervised a male and a female subordinate. For half of the supervisors, both subordinates maintained stable, high performance; for the other half, both subordinates maintained stable, low performance.

Procedure

Each subject was met by the experimenter and introduced to the two confederate subordinates. Subjects were advised that they would be supervis- ing the performance of the subordinates while they worked on a series of word puzzles, and also that the task had just been explained to the subordinates. Subjects were led to believe that supervisor and subordinate roles were deter- mined by order of arrival, with the experimenter “saving time” by assigning the first two subjects to arrive to the role of subordinates and explaining the task to them while waiting for the third subject (the supervisor). The experimenter then took the subject to the supervisor’s cubicle and, using a standardized instruction sheet, explained both the subordinates’ and the supervisor’s tasks.

Subordinates ’ task. Supervisors were led to believe that the subordinates were working on a series of embedded word puzzles. Each puzzle consisted of a matrix of 15 x 11 letters, embedded in which were 33 whole words. The embedded words were listed immediately below the matrix. Subordinates were required to circle each word located in the puzzle and to cross it out on the list provided. Each subordinate had a set of 10 puzzles to work through, and was required to tackle one puzzle in each of ten 2-min work periods, with their goal being to find as many embedded words as they could in each period.

The puzzle sheets had been previously completed by the experimenter so

Page 7: Gender Differences in Supervisors’ Use of Performance Feedback

792 BREWER, SOCHA, AND POTTER

that subordinates would appear to be maintaining either a consistently high or low level of performance. To ensure that these levels were realistic, the puzzles had been previously administered to 8 pilot subjects in order to identify average levels of performance. (Although no systematic testing was conducted to determine whether or not the task was gender-neutral, no obvious gender differences were apparent in this pilot testing.) Across the 10 work periods, subordinates located 14-16 items and 6-8 items in the high and low perform- ance conditions, respectively.

Supervisors ’ task. The experimenter explained the nature of the subordi- nate’s task to the supervisor, and indicated that the subordinates’ goal was to find as many embedded words as they could in each work period. The supervi- sor’s goal was to supervise the performance of each subordinate. At the end of each work period, the experimenter collected the completed puzzle sheets from the two subordinates and gave them to the supervisor for scoring. This involved checking that the circled words corresponded with those marked off on the list, and counting and recording the total number of words found by each subordi- nate. After completing the scoring, they were able to provide subordinates with feedback about their performance, if they so chose. They were permitted to give one feedback message only after each work period, and it was emphasized that whether or not they gave a feedback message was entirely up to them. In order for them to assess their subordinates’ performance, they were given a sheet which summarized typical levels of performance achieved in a 2-min work period by high (14-16 words), average (10-12 words), and low (6-8 words) performers who had previously completed the task.

If supervisors chose to give feedback, they selected a message from a fixed menu, entered it onto a feedback sheet (coded by supervisor, subordinate, and work session), and advised the experimenter when the scored puzzles were being collected. The experimenter then brought the subordinate(s) into the supervisor’s cubicle and the supervisor handed the message to the subordinate. Supervisors and subordinates were instructed not to initiate conversation or to ask questions. The feedback menu contained five different types of message: general negative feedback (“You are doing poorly. Try harder,” or “Your performance is extremely poor. Concentrate and try harder.”), specific negative or corrective feedback (“You missed the goal on the last one by - . Try harder.”), specific positive feedback (“You beat the goal on the last one by - . Keep it up,” or “You met the goal exactly on the last one. Keep it up.”), general positive feedback (“You are doing well. Keep it up,” or “Your performance is excellent.”), and neutral encouragement or feedback (“Keep working as hard as you can.”).

Dependent measures. The dependent measures were the frequency with which each subject delivered general negative, specific negative, specific

Page 8: Gender Differences in Supervisors’ Use of Performance Feedback

GENDER AND PERFORMANCE FEEDBACK 793

positive, general positive, and neutral feedback messages to each subordinate across the 10 work periods.

Manipulation Check

After the last work period, each subordinate was escorted to the supervi- sor’s room to receive feedback on their overall performance. The supervisor rated the performance of each subordinate on a 7-point scale, ranging from 1 (extremelypoor) to 7 (Outstanding) and gave the rating form to the subordinate. When the subordinate left the room, the supervisor was asked to provide another rating of each subordinate’s performance that would only be seen by the experimenter. Subjects were then debriefed and dismissed.

Results

Manipulation Check

The overall ratings of each subordinate’s performance which were com- pleted after the last work period were analyzed in a 2 x 2 x 2 x 2 (Supervisor Gender x Performance x Subordinate Gender x Rating Recipient) ANOVA, with repeated measures on the last two factors. High performing subordinates were perceived as performing above average ( M = 6.15) and significantly better than low performing subordinates whose performance was rated below average (M = 3.17), F(1, 55)3 = 317.81, p < .001. Ratings of overall performance were not affected by supervisor gender, subordinate gender, or by their being transmitted to the subordinate or the experimenter. There was, however, a significant Performance x Subordinate Gender interaction, F( 1, 55) = 7.86, p < .01, which was due to males being rated a little higher than females (6.23 vs. 6.07) in the high performance condition only (Tukey, p < .05).

Feedback Measures

Average frequencies for each of the different types of feedback message and the total feedback messages provided in the different conditions are shown in Table 1. General positive and neutral feedback messages were used most frequently, with these messages administered following both good and poor performance. Specific positive and specific negative feedback messages, which were generally constrained to instances of good and poor performance,

3Degrees of freedom are one less than expected because of missing data from one subject.

Page 9: Gender Differences in Supervisors’ Use of Performance Feedback

Tabl

e 1

4

(D

P

Aver

age

Freq

uenc

ies f

or D

iffer

ent T

ypes

of F

eedb

ack

Mes

sage

s and

Tot

al F

eedb

ack

Mes

sage

s by

Con

ditio

n

Perf

orm

ance

Hig

h Lo

w

Mal

e Fe

mal

e su

bord

inat

e su

bord

inat

e O

vera

ll M

ale

Fem

ale

subo

rdin

ate

subo

rdin

ate

Ove

rall

Mal

e su

perv

isor

Sp

ecifi

c ne

gativ

e G

ener

al n

egat

ive

Spec

ific

posi

tive

Gen

eral

pos

itive

N

eutra

l To

tal

Fem

ale

supe

rvis

or

Spec

ific

nega

tive

Gen

eral

neg

ativ

e Sp

ecifi

c po

sitiv

e G

ener

al p

ositi

ve

Neu

tral

Tota

l

0.07

0.

00

1 .oo

2.53

0.

47

4.07

0.00

0.

07

1.07

2.

07

0.47

3.

68

0.27

0.

00

1.07

2.

07

0.60

4.

01

0.00

0.

00

0.93

2.

27

0.40

3.

60

0.17

0.

00

1.04

2.

30

0.54

4.

05

0.00

0.

04

1 .oo

2.17

0.

44

3.65

1.87

0.

60

0.20

0.

87

1.60

5.

14

0.86

0.

43

0.2 1

0.

43

1.79

3.

72

1.73

0.

80

0.13

0.

87

1.67

5.

20

0.7 1

0.

64

0.29

0.

64

1.36

3.

64

1 .so

0.70

0.

17

0.87

1.

64

5.18

0.79

0.

54

0.25

0.

54

1.58

3.

70

Page 10: Gender Differences in Supervisors’ Use of Performance Feedback

GENDERANDPERFORMANCEFEEDBACK 795

respectively, occurred less frequently. General negative messages were used most sparingly.

A 2 x 2 x 2 (Supervisor Gender x Performance x Subordinate Gender) multivariate analysis of variance (MANOVA), with repeated measures on subordinate gender, was performed on the five dependent variables: frequency of general negative, specific negative, specific positive, general positive, and neutral feedback messages.

Using Wilks’s criterion, the main effect for supervisor gender was not significant, F(5, 5 1) = 1.49, p > .05. There was a significant main effect on the combined dependent variables for performance, F(5, 51) = 1 6 . 6 5 , ~ < .001. As hypothesized, poor performers received more feedback messages overall than good performers ( M = 4.86 vs. 3.88). A significant effect for subordinate gender also emerged, F(5, 51) = 11.10, p < .001, with male subordinates receiving slightly more feedback messages than females ( M = 4.16 vs. 4.12). Significant multivariate interaction effects between supervisor gender and subordinate gender, F(5, 51) = 2.44, p < .05, and between performance and subordinate gender, F(5, 51) = 1 0 . 3 1 , ~ < .001, were also found.

Since pooled within-cell correlations, adjusted for independent variables, for many of the dependent variables exceeded .30 (Table 2), a stepdown analysis was conducted in an attempt to isolate which of the dependent variables contrib- uted uniquely to the effects of the independent variables on the combined depend- ent variables (Tabachnick & Fidell, 1989). To prioritize dependent variables for the stepdown analysis, the different types of feedback message were entered into a regression equation via a forced entry procedure, with total feedback frequency as the dependent variable. The feedback types which independently accounted for the most variance in the total feedback score were, in decreasing order, specific negative, specific positive, general negative, neutral, and general positive feedback. In the stepdown analysis, each of these dependent variables was analyzed in turn, with the highest priority dependent variable tested in a univariate ANOVA and each subsequent dependent variable analyzed with the higher priority dependent variables as covariates. To achieve an overall alpha equal to .05, tests for each of the dependent variables were evaluated against an alpha equal to .01. Table 3 summarizes the outcome of these analyses.

Although a significant effect for supervisor gender did not emerge in the multivariate analysis, the stepdown analysis revealed that the highest priority dependent variable, specific negative feedback, uniquely predicted differences between male and female supervisors, stepdown F( 1, 55) = 7.42, p < .01. Male supervisors gave more than twice as much specific negative or corrective feedback as did female supervisors (0.99 vs. 0.38 messages). We also expected that male supervisors would respond to poor performers more rapidly. Indeed, male supervisors (93%) were more likely than were female supervisors (57%)

Page 11: Gender Differences in Supervisors’ Use of Performance Feedback

796 BREWER, SOCHA, AND POTTER

Table 2

Pooled Within-Cell Correlations for the Five Dependent Variables

General Specific General negative positive positive Neutral

Specific negative .40 -.20 -.22 .42 General negative - .18 -.12 .36 Specific positive S O -.02 General positive -.14

to provide their first feedback message to low performers after the first or second work period than at some later stage, x2 ( 1 , N = 5 8 ) = 8 . 4 8 , ~ < .01. This pattern was not found with high performing subordinates. Although, as hy- pothesized, male supervisors used specific negative feedback more frequently than did females, this difference did not extend to the use of general negative feedback messages, as had been predicted. Nor was there any evidence in support of the hypothesis that female supervisors would use specific positive, general positive, or neutral feedback more frequently than males.

Stepdown analyses exploring the significant mutivariate effects of Subor- dinate Gender and Performance x Subordinate Gender showed that none of the dependent variables made unique contributions to these effects. Examination of the Supervisor Gender x Subordinate Gender interaction revealed a signifi- cant stepdown, F(1, 5 5 ) = 1 0 . 7 9 , ~ < .01, q2 = .01, for neutral feedback, with the higher priority dependent variables, specific negative, specific positive, and general negative feedback, as covariates. The absence of a significant univari- ate effect, F( 1 , 5 5 ) 2 . 9 7 , ~ = .09, makes interpretation of this stepdown effect particularly difficult and, given the very low proportion of shared variance, no further explanation of this effect was attempted.

Exploration of the hypotheses regarding performance level and feedback revealed that unique contributions to the overall differences in feedback pro- vided to high and low performing subordinates were made by three dependent variables: specific negative, specific positive, and general positive feedback. The largest contribution was made by the highest priority dependent variable, specific negative feedback, stepdown F( 1 , 5 5 ) = 3 1.13, p < .OO 1 , with high and low performing subordinates receiving 0.08 and 1.3 1 corrective messages, respec- tively. With specific negative feedback as a covariate, specific positive feed- back also made a significant contribution, stepdown F( 1 , 54) = 9.61, p < .01,

Page 12: Gender Differences in Supervisors’ Use of Performance Feedback

GENDER AND PERFORMANCE FEEDBACK 797

Table 3

Tests of Supervisor Gender, Subordinate Gender, Performance, and Their Interactions

~~ ~

Independent Dependent Univariate variable variable F df Step F df P

Supervisor gender

Subordinate gender

Performance

Supervisor Gender

Subordinate Gender

Performance

Supervisor Gender

X

X

Spec neg Spec pos Gen neg Neutral Gen pos

Spec neg Spec pos Gen neg Neutral Gen pos

Spec neg Spec pos Gen neg Neutral Gen pos

Spec neg Spec pos Gen neg Neutral Gen pos

Spec neg Spec pos Gen neg Neutral Gen pos

7.42* 0.01 0.18 0.07 0.47

0.12 0.03 2.09 0.53 0.03

31.13* 12.05* 15.53* 12.96* 20.3 1 *

0.90 0.03 0.05 2.97 8.50*

3.82 0.06 0.42 0.00 0.08

1,55 1, 55 1, 55 1,55 1,55

1,55 1,55 1,55 1,55 1, 55

1,55 1,55 1,55 1,55 1,55

1 , 5 5 1, 55 1, 55 1,55 1,55

1,55 1,55 1,55 1,55 1 ,55

7.42** 0.09 0.00 0.12 0.35

0.12 0.03 1.91 0.10 0.40

31.13** 9.61** 5.27 3.25 7.97**

0.90 0.03 0.32

10.79** 0.17

3.82 0.15 0.11 0.27 0.06

1, 55 1,54 1,53 1, 52 1, 51

1,55 1, 54 1,53 1, 52 1, 51

1,55 1, 54 1,53 1, 52 1, 51

1, 55 1, 54 1,53

1,51

1 , 5 5 1,54 1,53 1, 52 1,51

1, 52

,009 .768 .994 .73 1 ,554

.73 1 ,868 .173 .748 .532

.ooo

.003

.026

.077

.007

.346

.863

.573

.002 ,681

.056

.70 1

.739

.604

.800

(table con tin ues)

Page 13: Gender Differences in Supervisors’ Use of Performance Feedback

798 BREWER, SOCHA, AND POTTER

Table 3 (continued)

variable variable Independent Dependent Univariate

~

Performance

Subordinate Gender

X

Performance

Supervisor Gender

Subordinate Gender

X

X

Spec neg Spec pos Gen neg Neutral Gen pos

Spec neg Spec pos Gen neg Neutral Gen pos

F

4.67 0.04 4.01 1.13 2.53

0.75 0.8 1 0.1 1 0.53 2.24

df Step F df P

1, 55 4.67 1, 55 ,035 1, 55 0.02 1, 54 .875 1, 55 2.03 1, 53 ,160 1, 55 1.04 1, 52 .312 1, 55 1.25 1, 51 ,268

1, 55 0.75 1, 55 .391

1, 55 0.70 1, 53 .406 1, 55 0.24 1, 52 .626 1,55 1.18 1, 51 ,283

1, 55 0.80 1, 54 ,374

Note. Spec neg = specific negative; Gen neg = general negative; Spec pos = specific positive; Gen pos = general positive. *Significance level would reachp < .01 in univariate context. **p < .01.

with high and low performing subordinates receiving 1.02 and 0.21 specific positive messages, respectively. Finally, general positive feedback, with the other four dependent variables as covariates, contributed significantly, step- down F( 1 , s 1) = 7 . 9 7 , ~ < .01, with high and low performers receiving 2.23 and 0.7 1 general positive feedback messages, respectively. Although univariate analyses also revealed significant effects of performance on general negative, F(1, 55) = 1 5 . 5 3 , ~ < .001, and neutral messages, F(1, 5 5 ) = 12.96, p = .001, these differences were already accounted for by higher priority dependent variables. In summary, specific negative and specific positive feedback were seldom given inappropriately to good and poor performers, respectively. But, while good performers were more likely than poor performers to receive general positive feedback, the latter still occasionally received general positive messages and, thereby, more feedback overall.

Discussion

The outcomes of previous research on gender differences in leadership style

Page 14: Gender Differences in Supervisors’ Use of Performance Feedback

GENDERANDPERFORMANCEFEEDBACK 799

(Eagly & Johnson, 1990) provided grounds for predicting gender differences in feedback delivery. Contrary to specific hypotheses about the relationships between supervisor gender and the use of particular types of performance feedback, the frequency with which general negative, specific positive, general positive, and neutral encouraging messages were used was unaffected by supervisor gender. The prediction that males, who tend to lead in a more directive style, would react more quickly to poor performance and use specific negative feedback more frequently was, however, supported by the stepdown analysis examining the unique contributions of each dependent variable. Thus, this finding provides one illustration of how gender differences in leadership style may be translated into specific supervisory behaviors.

We can only speculate as to why male supervisors might use specific negative feedback more frequently than female supervisors. Perhaps, as Eagly and Johnson (1 990) suggested, there are stronger attitudinal biases against females behaving in a directive manner than there are against males. Or, again following Eagly and Johnson, perhaps males more readily resort to correcting subordinate performance because they are characterized by deficiencies in social skills which mitigate against them dealing with poorly performing subordinates in other ways. Another possibility is that male and female super- visors may differ in the nature of the attributions they make regarding the causes of poor performance, with these attributions underlying the differences in performance feedback. For example, if males are more likely than females to attribute poor subordinate performance to unstable, external, and/or control- lable causes rather than to stable, internal, and/or uncontrollable causes (cf. Weiner, 1985), they might be more likely to perceive that they are able to influence their subordinates’ performance using appropriate specific negative feedback. All of these possibilities are amenable to systematic investigation in future research.

Previous research also suggested that subordinate gender may interact with supervisor gender to influence feedback delivery. Although significant effects on the composite dependent variable were found for subordinate gender, and for Supervisor Gender x Subordinate Gender, the stepdown analyses revealed that none of the dependent variables made a unique contribution to these effects. In those previous studies which had pointed to the possibility that subordinate gender may be an important variable, actual behavioral interac- tions between individuals in supervisory and subordinate roles had not been examined. Perhaps when individuals are confronted with the actual supervision of subordinates, any influence of subordinate gender is far outweighed by the performance characteristics of subordinates.

While our dependent measures allowed us to tap frequency, timing, sign, and specificity of feedback delivery, the affective component of the message

Page 15: Gender Differences in Supervisors’ Use of Performance Feedback

800 BREWER, SOCHA, AND POTTER

was fixed. When feedback delivery is completely under the supervisor’s con- trol, this dimension will also vary across supervisors and may prove to be as strong a source of gender differences as any of the other dimensions.

It is, of course, possible that the influence of both supervisor and subordi- nate gender may have been constrained by the particular performance charac- teristics exhibited by subordinates in this experiment. Performance levels were manipulated so that subordinates were clearly perceived as performing either above or below average. It is possible that this highlighted the most appropriate feedback message, thereby working against the emergence of any gender differences. The stability of subordinates’ performance and their lack of re- sponsiveness to the supervisors’ feedback messages may also have constrained gender-related differences in feedback delivery. If, for example, supervisors had been confronted with subordinates whose performance improved or dete- riorated across work periods, rather than remaining stable, perhaps gender effects may have been reflected in some of the dependent measures other than specific negative feedback. Similarly, if performance actually appeared to be responsive to any feedback messages delivered, providing feedback may have seemed more worthwhile. Likewise, had the content of the feedback messages focused more on the process by which subordinates could improve their per- formance, rather than simply on the overall level of performance, supervisors might have felt more inclined to react to subordinate performance. Finally, allowing subjects to generate their own feedback messages (and subsequently having judges categorize them) might have given rise to greater inter- and intra-individual variability than prescribing a set menu of feedback messages. Despite these weaknesses, it is worth noting that the ratio of feedback messages to feedback opportunities was relatively high, suggesting that supervisors were, in fact, not overly constrained. Nevertheless, these issues warrant inves- tigation in future work.

The results of this experiment also reinforced some of the findings from previous studies into the relationships between the level of subordinates’ performance and the nature of the performance feedback given. For example, as previously found by Hobson (1986), poor performers received more feed- back messages (overall) than did good performers. This reflected the fact that supervisors frequently provided encouraging messages following poor per- formance, whereas negative messages seldom followed good performance (cf. Larson, 1986).

To what extent the findings of this study will generalize beyond the laboratory setting is a question which can only be resolved by appropriate field studies. We must acknowledge that the supervisors in this study were not actual managers, had no prior or anticipated future interactions with their subordinates, and selected the feedback messages from a predetermined menu.

Page 16: Gender Differences in Supervisors’ Use of Performance Feedback

GENDERANDPERFORMANCEFEEDBACK 801

Nevertheless, it is interesting to note that Eagly and Johnson’s (1 990) meta- analysis of gender differences in leadership style found that the tendency for males to lead in a more directive manner-the finding which led to the hypothesis that males would use specific negative feedback more frequently than females-prevailed in both laboratory and organizational settings. Given the broad dimensions of behavior tapped by leadership style measures, gender differences in leadership style could, of course, translate into a number of other specific differences in behavior. Clarification of the precise nature of any such behavioral differences, and of the conditions under which they result in im- proved performance and satisfaction from their subordinates, will provide abundant opportunities for further research in this area.

References

Baker, L. D., DiMarco, N., & Scott, W. E., Jr. (1975). Effects of supervisor’s sex and level of authoritarianism on evaluation and reinforcement of blind and sighted workers. Journal of Applied Psychology, 60,28-32.

Bass, B. M. (1 985). Leadership and performance beyond expectations. New York, NY: Free Press.

Bass, B. M. (1990). Bass and Stogdill’s handbook of leadership: Theory, research, and managerial applications (3rd ed.). New York, NY: Free Press.

Brewer, N., Wilson, C., & Beck, K. (1994). Supervisory behaviour and team performance amongst police patrol sergeants. Journal of Occupational and Organizational Psychology, 67,69-78.

Cadi, L. L. (1989). Gender differences in interaction style and influence. Journal of Personality and Social Psychology, 56,565-576.

Dansereau, F., Jr., Graen, G., & Haga, W. J. (1975). A vertical dyad linkage approach to leadership within formal organizations: A longitudinal inves- tigation of the role making process. Organizational Behavior and Human Performance, 13,46-78.

Dobbins, G. H. (1986). Equity vs. equality: Sex differences in leadership. Sex Roles, 15, 513-525.

Dobbins, G. H., Pence, E. C., Orban, J. A., & Sgro, J. A. (1 983). The effects of sex of the leader and sex of the subordinate on the use of organizational control policy. Organizational Behavior and Human Performance, 32,

Eagly, A. H., & Johnson, B. T. (1990). Gender and leadership style: A meta-

Erez, M. (1977). Feedback: A necessary condition for the goal setting-

325-343.

analysis. Psychological Bulletin, 108, 233-256.

performance relationship. Journal of Applied Psychology, 62, 624-627.

Page 17: Gender Differences in Supervisors’ Use of Performance Feedback

802 BREWER, SOCHA, AND POTTER

Fisher, C. D. (1979). Transmission of positive and negative feedback to subor- dinates: A laboratory investigation. Journal of Applied Psychology, 64,

Fleishman, E. A. (1953).The description of supervisory behavior. Journal of Applied Psychology, 37, 1-6.

Green, S. G., & Mitchell, T. R. (1979). Attributional processes of leaders in leader-member interactions. Organizational Behavior and Human Per- formance, 23,429-458.

Hobson, C. J. (1986). Factors affecting the frequency, timing, and sign of informal supervisory feedback to subordinates in a simulated work setting. Multivariate Behavioral Research, 21, 187-200.

Katerberg, R., & Hom, P. (1981). Effects of within-group and between-group variation in leadership. Journal of Applied Psychology, 66,2 18-223.

Komaki, J. L., Desselles, M. L., & Bowman, E. D. (1989). Definitely not a breeze: Extending an operant model of effective supervision to teams. Journal of Applied Psychology, 74,522-529.

Komaki, J. L., Zlotnick, S., & Jensen, M. (1986). Development of an operant based taxonomy and observational index of supervisory behavior. Journal of Applied Psychology, 71, 260-269.

Larson, J. R., Jr. (1984). The performance feedback process: A preliminary model. Organizational Behavior and Human Performance, 33,42-76.

Larson, J. R., Jr. (1986). Supervisors’ performance feedback to subordinates: The impact of subordinate performance valence and outcome depend- ence. Organizational Behavior and Human Decision Processes, 37,

Larson, J. R., Jr., & Callahan, C. (1990). Performance monitoring: How it affects work productivity. Journal of Applied Psychology, 75, 530-538.

Likert, R. (1967). The human organizution. New York, NY: McGraw-Hill. Luthans, F., & Lockwood, D. L. (1984). Toward an observation system for

measuring leader behavior in natural settings. In J. G. Hunt, D. Hosking, C. Schriesheim, & R. Stewart (Eds.), Leaders and managers: International perspectives on managerial behavior and leadership (pp. 1 17-141). New York, NY: Pergamon.

Mintzberg, H. (1973). The nature of managerial work. New York, NY: Harper & Row.

Nadler, D. A. (1979). The effects of feedback on task group behavior: A review of the experimental research. Organizational Behavior and Human Performance, 23, 309-338.

Nieva, V. F., & Gutek, B. A. (1981). Women and work: A psychological perspective. New York, NY: Praeger.

Podsakoff, P. M., Todor, W. D., & Skov, R. (1982). Effects of leader contingent

533-540.

3 9 1 -408.

Page 18: Gender Differences in Supervisors’ Use of Performance Feedback

GENDERANDPERFORMANCEFEEDBACK 803

and noncontingent reward and punishment behaviors on subordinate per- formance and satisfaction. Academy of Management Journal, 25,8 10-82 1.

Szilagyi, A. D. (1980). Reward behavior of male and female leaders: A causal inference analysis. Journal of Vocational Behavior, 16, 59-12.

Tabachnick, B. G., & Fidell, L. S. (1989). Using multivariate statistics (2nd ed.). New York, NY: Harper Collins.

Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological Review, 92, 548-513.

Yukl, G. A. (1989). Leadership in organizations (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall.