315 m LIBRARIES

llllllll llll“l“"l\ lZ‘l ll ”ill“

3 12930

l

 

 

 

 

 

This is to certify that the

thesis entitled
EFFECTS OF GENDER, PERFORMANCE TRENDS AND CAUSAL
INFORMATION ON THE FORMATION OF A'ITRIBUTIONS FOR HIGH
PERFORMERS: IMPLICATIONS FOR PERSONNEL DECISIONS
presented by

Stanley Morris Gully

has been accepted towards fulfillment
of the requirements for

M. A. Psychology

degree in

 

 

    

 

Major profes/so/

Date November 17, 1993

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove thh chockom from your record.
TO AVOID F INES return on or before data duo.

DATE DUE DATE DUE DATE DUE

 

 

 

J3; 1 1 2000

 

 

 

 

 

 

 

 

 

 

 

 

___I_I
:l‘ir—T—‘i

 

 

MSU I. An Affirmative Adlai/Emu Opportunity lmtltulon
Wm:

 

 

 

EFFECTS OF GENDER, PERFORMANCE TRENDS AND CAUSAL
INFORMATION ON THE FORMATION OF ATTRIBUTIONS FOR HIGH
PERFORMERS: IMPLICATIONS FOR PERSONNEL DECISIONS

BY
Stanley Morris Gully

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology

1993

 

ABSTRACT
EFFECTS OF GENDER, PERFORMANCE TRENDS AND CAUSAL

INFORMATION ON THE FORMATION OF ATTRIBUTIONS FOR HIGH
PERFORMERS: IMPLICATIONS FOR PERSONNEL DECISIONS

BY
Stanley Morris Gully

The current study was designed to investigate how
gender and performance trends interact with causal
attributions of locus, controllability, and stability to
affect personnel decisions. The study was a 2 (Employee
Gender) X 3 (Performance Trend) X 3 (Informational Cues)
experiment, which included undergraduate students as
participants. The results indicate informational cues of
effort, ability, and ease of job vary in perceived locus of
causality, controllability, and stability, as predicted by
attribution theory. These attributions, in turn, interact
with performance trends to affect personnel decisions.
Results also indicate employee gender interacts with causal
information and performance trends to affect attributions
and personnel decision outcomes, though not as hypothesized.

Implications for organizational contexts and future research

are discussed.

 

This is dedicated to my grandfather, James L. Gully.

iii

 

 

en:
he

5 A]
5 UK

N,
Vi

'0

.I
-

. 7‘

ACKNOWLEDGMENTS

I would like to thank my family and friends who have
encouraged and helped me to pursue my education. Without
their love and support, I would not be writing this page
today.

I would also like to thank the members of my thesis
committee, Steve Kozlowski, Dan Ilgen, and Galen
Bodenhausen. I appreciate their thoughtful insights,
suggestions, and guidance. I am especially grateful for

their willingness to take the time to read my thesis and

help develop my skills as a researcher.

I would like to thank Steve Kozlowski, in particular,
for his incredible ability to return quickly my written
products with astute comments scribbled all over them. I
also want to thank him for elucidating the research process;
I learned much about conducting quality research by working
with him. Steve also provided words of encouragement just

when I needed them most. Without his help, I would never

have finished this thesis.

I would like to express gratitude to John Hollenbeck
and Neal Schmitt for their assistance with various aspects
of the data analyses. I learned much about repeated

measures regression through my conversations with them.

iv

 

87.:

an!
UV
0 a

on

ass

§
.he

I would also like to thank Jean Phillips for her
encouragement, assistance, and support. I will always
appreciate her willingness to proofread and provide feedback
on my thesis during those times when I was too brain—weary
or bleary-eyed to understand my own writing.

In closing, I would like to thank countless others,
including Fernanda Marchi and Garrett Gamet, for their
assistance in completing this work. This thesis is truly

the product of the efforts of many.

 

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . . . . . . . . .
INTRODUCTION . . . . . . . . . . . . . . . . . . . .
REVIEW OF THE LITERATURE . . . . . . . . . . . . . .

Attribution Theory and Leader Decisions . . . .
Attribution theory . . . . . . . . . .
Effects on leader decisions and behaviors .

Controllability as an attributional
dimension . . . . . . . . . . . . . . . .

Gender Effects on Attributions and
Personnel Decisions . . . . . . . . . . . . . .

Gender stereotypes and attributions . . . .
Gender stereotypes and female managers . .

Gender stereotypes and personnel decisions

Effects of gender group membership . . . .
Effects of information availability . . . .
Limitations of past research . . . . . . .
Summary of the literature review . . . . .

Integration and Conceptual Model . . . . . . .
Performance trends . . . . .

Effect of attributions and
performance trends . . . . . . . . . . . .

vi

ll

15

25

25

31

33

39

46

51

56

59

61

66

Attributions of informational cues . . . . . 74

Employee gender effects on attributional
perceptions . . . . . . . . . . . . . . . .

Employee gender effects on
personnel decisions . . . . . . . . . . . . . 86

Gender group membership effects on
attributional perceptions . . . . . . . . . . 90

Gender group membership effects on
personnel decisions . . . . . . . . . . . . . 99

Conclusion . . . . . . . . . . . . . . . . 102

METHOD . . . . . . . . . . . . . . . . . . . . . . . 105
Participants . . . . . . . . . . . . . . . . . . 105
Design . . . . . . . . . . . . . . . . . . . . . 106
Pilot Studies . . . . . . . . . . . . . . . . . 108
First pilot study . . . . . . . . . . . . . 108

Second pilot study . . . . . . . . . . . . 111
Manipulations . . . . . . . . . . . . . . . . . 113
Measures . . . . . . . . . . . . . . . . . . . . 116
First questionnaire . . . . . . . . . . . . 116

Second questionnaire . . . . . . . . . . . 119
Procedure . . . . . . . . . . . . . . . . . . . 121
RESULTS . . . . . . . . . . . . . . . . . . . . . . . 128
Overview . . . . . . . . . . . . . . . . . . . . 128
Establishment of the Measurement Model . . . . . 134
WAMS . . . . . . . . . . . . . . . . . . . 135

Perceived causes of performance
scales 0 g g g g g o o o o o o o o o o o o 142

Perceived attributions scales . . . . . . . 148

vii

Personnel decisions scales . . . . . . . . 156

Manipulation Checks . . . . . . . . . . . . 177
Effects of Performance Trend, Informational

Cues, and Employee Gender on

Personnel Decisions . . . . . . . . . . . . . . 183

Effects of Performance Trend, Informational
Cues, and Employee Gender on
Attributions . . . . . . . . . . . . . . . . . . 212

Effects of Performance Trend, Informational

Cues, and Gender Group Membership on

Personnel Decisions . . . . . . . . . . . . . . 233
Effects of Performance Trend, Informational

Cues, and Gender Group Membership on

Attributions . . . . . . . . . . . . . . . . . . 235

Effects of Perceived Attributions and
Performance Trend on Personnel Decisions . . . . 237

Effects of WAMS and Decision Maker Gender on
Personnel Decisions . . . . . . . . . . . . . . 259

Summary of Results . . . . . . . . . . . . . . . 272

DISCUSSION 0 O O I O O O I O O O O O O O O O O O O O 277

Attribution Theory . . . . . . . . . . . . . . . 280
Performance Trend . . . . . . . . . . . . . . . 283
Gender Stereotype Research . . . . . . . . . . . 285
Group Membership Research . . . . . . . . . . . 291
Generalizability Issues . . . . . . . . . . . . 292
Implications of Findings . . . . . . . . . . . . 295

Limitations and Future Directions . . 301

APPENDICES
APPENDIX A: Complete Listing of Hypotheses . . . . . 304
APPENDIX B: Power Analysis . . . . . . . . . . . . . 310

APPENDIX C: Results of Pilot Study . . . . . . . . . 312

viii

APPENDIX D: Examples of Experimental Manipulations . 315

APPENDIX E: Measurement Scales . . . . . . . . . . . 338
APPENDIX F: Protocol . . . . . . . . . . . . . . . . 348
APPENDIX G: Definition of Terms . . . . . . . . . . 357
IJST OF REFERENCES . . . . . . . . . . . . . . . . . 359

ix

 

hi

I
t

a

la

alt

Table
Table

Table

Table

Table
Table
Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

10:

11:

12:

13:

14:

15:

16:

LIST OF TABLES

Manipulations of Informational Cues . . .
Summary of Questionnaires . . . . . . . .

Independent and Dependent Variables of
Interest . . . . . . . . . . . . . . . . .

Hypotheses, Relevant Constructs,
and Analyses . . . . . . . . . . . . . . .

Factor Analysis on the WAMS . . . . . . .
Intercorrelations of WAMS Sub-scales . .
Reliability Analysis for the WAMS . . . .

Factor Analysis on the Perceived Causes
of Performance . . . . . . . . . . . . .

Three Factor Solution for Perceived Causes
of Performance . . . . . . . . . . . . . .

Intercorrelations of the Perceived Causes
of Performance . . . . . . . . . . . . . .

Factor Analysis on the Attributions of
Causes of Performance . . . . . . . . . .

Reliability Analyses for the Attributions
Scales . . . . . . . . . . . . . . . . . .

Intercorrelations of the Attributions
Scales . . . . . . . . . . . . . . . . . .

Factor Analysis on the Personnel Decisions
scales 0 O O O O O O O O O O O O O O O 0

Reliability Analyses for the Personnel
Decisions Scales . . . . . . . . . . . .

Intercorrelations of Personnel Decisions
Scales . . . . . . . . . . . . . . . . .

115

122

124

130

136

139

140

144

146

147

150

153

155

158

161

165

 

lab

la]

la

Te

Fr?

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

17:

18:

19:

20:

22:

23:

24:

25:

26:

27:

28:

29:

Means, Standard Deviations and

Reliabilities of All Scales . . . . . . 167
Descriptions of Measured Variables Used

in MANOVA and ANOVA . . . . . . . . . . . 168
Means, Standard Deviations and Intercorrelations
of All Measured Variables, Configured by
Informational Cues Provided . . . . . . . 171
Means, Standard Deviations and Intercorrelations

of All Measured Variables Based on Entire
Set of Observations . . . . . . . . . . . 178
MANOVA for the Perceived Causes of
Observed Performance . . . . . . . . . . . 182
Cell Means of Perceived Ability, Effort, or
Ease of Job Split by Performance Trend and
Informational Cues . . . . . . . . . . . . 184
Repeated Measures MANOVA for the Overall
Effect of Performance Trend, Informational
Cues, and Employee Gender on Personnel
DeCiSions O O O O O O O O O O O O O O O O 186
Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Promotion
Decisions . . . . . . . . . . . . . . . . 187
Cell Means of Promotion Decisions, Split by
Performance Trend, Informational Cues,
and Employee Gender . . . . . . . . . . . 188
Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Pay Deci51ons . 193
Cell Means of Pay Decisions, Split by
Performance Trend, Informational Cues,

and Employee Gender . . . . 194

Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Expectations

for Future Performance . . . . . . . . . . 200
Cell Means of Expectations, Split by
Performance Trend, Informational Cues,

and Employee Gender . . . . . . . . . . . 201

xi

 

Tab

Tat

Tat

Ia

Dv—l

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

Table

31:

32:

33:

34:

35:

36:

37:

38:

39:

40:

Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Affect Toward

the Employee . . . . . . . . . . . . . . .

Cell Means of Affect, Split by
Performance Trend, Informational Cues,
and Employee Gender . . . . . . . . . . .

Repeated Measures MANOVA for the Overall
Effect of Performance Trend, Informational
Cues, and Employee Gender on

Attributions . . . . . . . . . . . . . . .

Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Perceived
Controllability . . . . . . . . . . . .

Cell Means of Controllability, Split by
Performance Trend, Informational Cues,
and Employee Gender . . . . . . . . .

Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Perceived

Locus of Causation . . . . . . . . . . . .

Cell Means of External Locus of Causality,
Split by Performance Trend, Informational
Cues, and Employee Gender . . . . . . . .

Repeated Measures ANOVA for the Effects
of Performance Trend, Informational Cues,
and Employee Gender on Perceived
Stability . . . . . . . . . . . . . . . .

Cell Means of Stability, Split by
Performance Trend, Informational Cues,
and Employee Gender . . . . . . . . .

Repeated Measures MANOVA for the Overall
Effect of Performance Trend, Informational
Cues, and Gender Group Membership on
Personnel Decisions . . . . . . . . . . .

Repeated Measures MANOVA for the Overall
Effect of Performance Trend, Informational
Cues, and Gender Group Membership on
Attributions . . . . . . . . . . . . . . .

xii

207

208

213

215

216

221

222

226

227

234

236

Tab

Tat

Tat

Tat

Ta

Table

Table

Table

Table

Table

Table

Table

Table

Table

41:

44:

45:

46:

47:

48:

49:

Results of Repeated Measures Regressions for
Pay Decisions Using Employee Gender,
Performance Trend, and Attributions . . . 243

Regression Equations for Prediction of Pay
Decisions Using Employee Gender, Performance
Trend, and Attributions . . . . . . . . . 244

Results of Repeated Measures Regressions for
Promotion Decisions Using Employee Gender,
Performance Trend, and Attributions . . . 251

Regression Equations for Prediction of
Promotion Decisions Using Employee Gender,
Performance Trend, and Attributions . . . 252

Results of Repeated Measures Regressions for
Pay Decisions Using WAMS Scores, Decision
Maker Gender, Employee Gender, Performance
Trend and Attributions . . . . . . . . . . 262

Regression Equations for Prediction of Pay
Decisions Using WAMS Scores, Decision Maker
Gender, Employee Gender, Performance Trend

and Attributions . . . . . . . . . . . . . 264

Results of Repeated Measures Regressions for
Promotion Decisions Using WAMS Scores,

Decision Maker Gender, Employee Gender,
Performance Trend and Attributions . . . . 266

Regression Equations for Prediction of Promotion
Decisions Using WAMS Scores, Decision Maker
Gender, Employee Gender, Performance Trend

and Attributions . . . . . . . . . . . . . 268

Outcomes for Tests of Hypotheses . . . . . 275

xiii

Fig
Fig

. 1w.
“It

LIST OF FIGURES

Figure 1: Eight-Cell Taxonomy of Causal Attributions
Figure 2: Six-Cell Taxonomy of Causal Attributions .

Figure 3: Integrative Conceptual Model of Gender
Stereotyping Effects on Attributions and
Personnel Decision Making . . . . . . . . .

Figure 4: Hypothesized Effects of Locus of
Causality and Performance Trends on
Personnel Decisions . . . . . . . . . . . .

Figure 5: Hypothesized Effects of Controllability
and Performance Trends on Personnel

Decisions . . . . . . . . . . . . . . . . .
Figure 6: Hypothesized Effects of Performance Trends
on Promotion Decisions . . . . . . . . . .

Figure 7: Hypothesized Effect of Employee Gender and
Performance Trend on Perceived Stability .

Figure 8: Hypothesized Effect of Employee Gender and
Performance Trend on Perceived
Locus of Causality . . . . . . . . . . .

Figure 9: Hypothesized Effect of Employee Gender and
Performance Trend on Perceived
Controllability . . . . . . . . . . . .

Figure 10: Hypothesized Effect of Employee Gender and
Performance Trend on Personnel Dec151ons .

Figure 11: Hypothesized Effect of Gender Group
Membership and Performance Trend on
Perceived Stability . . . . . . . . . . . .

Figure 12: Hypothesized Effect of Gender Group

Membership and Performance Trend on
Perceived Locus of Causality . . . . . . .

xiv

Fig

Fig

Fi:

Fi

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

13:

14:

15:

16:

17:

18:

19:

20:

21:

22:

23:

24:

25:

26:

Hypothesized Effect of Gender Group
Membership and Performance Trend on
Perceived Controllability . . . . . . . .

Hypothesized Effect of Gender Group
Membership and Performance Trend on
Personnel Decisions . . . . . . . . . . .

Performance Trend X Informational Cues
Interaction Effect for Promotion

Decisions . . . .

O O O O O O I O O O O 0

Employee Gender X Informational Cues
Interaction Effect for Pay Decisions . .

Performance Trend X Informational Cues

Interaction Effect for Pay
Decisions . . . . . . . . . . . . . .
Performance Trend X Informational Cues

Interaction Effect for Expectations .
Employee Gender X Informational Cues
Interaction Effect for Expectations . . .

Trend X Informational Cues

Performance
Effect for Affect . . . . .

Interaction

Trend X Informational Cues

Performance
Interaction Effect for Perceived
Controllability . . . . . . . . . . . . .

Performance Trend X Informational Cues
Interaction Effect for Perceived
External Locus . . . . . . . . . .

Performance Trend X Informational Cues
Interaction Effect for Perceived

Stability . . . . . . . . . . . . . . . .

Interaction Effect of External Locus of
Causality and Performance Trend for

Pay Decisions . . . . . . . . . . . .

Interaction Effect of Perceived
Controllability and Performance Trend for

Pay Decisions . . . . . . . . . . . . . .

Interaction Effect of External Locus of
Causality and Performance Trend for
Promotion Decisions . . . . . . . . . . .

95

101

191

196

198

204

205

210

219

224

230

248

249

256

Fig

Fig

F“
.t

Figure

Figure

Figure

Figure

Figure

27:

28:

29:

30:

31:

Interaction Effect of Perceived
Controllability and Performance Trend

Promotion Decisions . . . . . . . . .

Sample of an Improving Trend for a
Female Managerial Employee . . . . .

Sample of a Maintaining Trend for a
Male Managerial Employee . . . . . .

Sample of a Declining Trend for a
Female Managerial Employee . . . . .

Sample Graph for Explanation to
Participants . . . . . . . . . . . .

for

257

329

332

335

352

 

INTRODUCTION

Organizations have paid an increasing amount of
attention to diversity issues in work environments. This
trend is expected to continue as a result of changing
demographics within the work force and increasing amounts of
legislation designed to promote equal opportunity and
prevent discriminatory practices (e.g., Civil Rights Act of
1991). Between now and the year 2000, the number of 16—24
year olds available for employment is expected to decrease,

while the number of women and minorities available for

employment is expected to increase. Between 1989 and the

year 2000, 92% of work force growth is projected to consist
of women, minorities, and immigrants (London, 1989).

With regard to work opportunities, management and
executive positions are projected to be among the fastest
growing occupations between 1984 and 1995 (U.S. Department
of Labor, 1987). Typically, management and executive
positions are perceived to be of high status, highly valued,
and expected to require skills and leadership on the part of
the performer. However, women and minorities have
traditionally been underrepresented in these occupations,
especially in higher levels of management (Morrison & Von

Glinow, 1990). For example, women fill nearly a third of

1

 

all

dirt

Gli
Res
hel

of

2
all management positions, but only 3.6% of board
directorships and 1.7% of corporate officerships in Fortune
500 companies were held by women in 1988 (Morrison & Von
Glinow, 1990: Von Glinow & Krzyczkowska-Mercer, 1988).
Researchers have suggested that stereotyping processes may
help to explain why women who work as managers earn only 61%
of what men earn as managers and only .6% of managers who
earned $100,000 a year or more were women (Brenner,
Tomkiewicz, & Schein 1989).

Lord and Maher (1991), and Fiske, Bersoff, Borgida,
Deaux, and Heilman (1991) documented an interesting case of
how gender stereotypes affected decisions made with regard
to the promotion of a female executive manager. Ann B.
Hopkins joined Price Waterhouse in 1978 and quickly became a
successful manager by developing over 40 million dollars in
new business. She was the only woman of 88 nominees for
partnership in 1982, and she had brought more business to
the firm than any of the other nominees (Greenhouse 1989).
Despite her achievements, she was not made a partner, while
47 of the remaining 87 male nominees were made partners.
During discussion of her qualifications, many comments
centered on the sex of Ann B. Hopkins and her fit with
gender stereotypes. Supporters described her as outspoken,
independent, self-confident, assertive, and courageous while
her detractors interpreted the same behavior as overbearing,
arrogant, self-centered, and abrasive (American

Psychological Association, 1988: Fiske et a1., 1991). One

eva
whi

lad

ger.

in:

3
evaluator suggested that she take a "course at charm school"
while another said complaints came up only because "she is a
lady using foul language" (Bales, 1988).

The example above is consistent with the idea that
gender stereotypes can have effects on the encoding,
interpretation, and retrieval of information relevant to a
person and that person's behavior (Deaux, 1976: Dipboye,
1985; Heilman, 1983; Ilgen & Youtz, 1986; Feldman, 1990;
Wyer & Srull, 1984). Researchers have found that women are
less likely than men to be rated positively, rewarded,
promoted, or hired in many personnel situations (Cohen &
Bunker, 1975: Eagly, Makhijani, & Klonsky, 1992: Fidell,
1970: Landy & Farr, 1980: Olian, Schwab, & Haberfeld, 1988:
Rosen & Jerdee, 1973, 1974: Terborg & Ilgen, 1975). This
effect is especially strong when the job is a traditionally
masculine one (Heilman, Martell, & Simon, 1988: Plake et
al., 1987, Rosenwasser & Dean, 1989; Terborg & Ilgen, 1975),
a physical one (Mendel & Shoenfelt, 1992), or if the female
is particularly attractive (Heilman & Stopeck, 1985a; 1985b:
Spencer & Taylor, 1988). Some conflicting findings have
also been found (Kinicki & Griffeth, 1985; Lefkowitz &
Battista, 1992), and estimates of effect sizes have been
small in some studies (Swim, Borgida, & Maruyama, 1989;
Eagly et al., 1992).

Some investigators have hypothesized that the biases

described above are the result of the influence of gender

stereotypes on inferred causes of performance (Deaux, 1976;

St'

at

pc

de

4

Heilman, 1983). A stereotype is a set of inferred relations

that associate a social category with personal attributes
(Ashmore & Del Boca, 1979). Because people often possess
knowledge structures about a variety of social categories
(e.g., race, sex, and age), these categories are used as
sources of information about the underlying dispositional
qualities of individuals (Stangor, Lynch, Duan, & Glass,
1992). Hamilton, Sherman, and Ruvolo (1990) noted that a
stereotype is an important source of expectancies for
attributes that individual group members are likely to
Thus, researchers have converged on the following

possess.

definition of a stereotype: A set of attributes ascribed to

a group and imputed to its individual members simply because

they belong to that group (Taylor, Fiske, Etcoff, &
Ruderman, 1978: Heilman, 1983). Schein's (1978) definition
of a sex role stereotype is consistent with this definition.
She defines a sex role stereotype as the belief that a set
of traits and/or abilities is more likely to be found among
one sex than the other. Because stereotypes are sources of
expectancies for categories of people, they may serve as
Simplifying heuristics which allow people to quickly and
economically organize and retrieve information (Deaux, 1984;

Fiske & Taylor, 1984).

On the basis of the ideas presented above, it is clear

that stereotypes are expectations which structure social

perception (Deaux, 1976: Hamilton, 1981). That is, they are

likely to affect attention to information and the way in

thi
int
i l
Hal
is
the

bi;

5
which information is processed, encoded, retrieved, or
interpreted (Bierhoff, 1989; Bodenhausen & Wyer, 1985; Fiske
8 Neuberg, 1990; Hamilton, Sherman, & Ruvolo, 1990: Lord &
Maher, 1991). The way that performance-related information
is treated as a result of stereotypes may explain some of
the findings cited earlier with regard to gender-related
bias in personnel ratings and decisions.

When leaders, managers, or supervisors make ratings or
personnel decisions, they look not only at the occurrence of
the performance itself, but they also look for information
with regard to the cause of the observed performance (Dugan,
1989: Feldman, 1981: Gioia & Sims, 1986; Green & Mitchell,
1979: Ilgen & Youtz, 1986; Mitchell & O'Reilly, 1990). A
large body of research has indicated that gender stereotypes
play an important role in influencing the perceived causes
of good or poor performance of males and females in a wide
variety of contexts (Deaux, 1976; Deaux & Emswiller, 1974;
Feldman-Summers & Kiesler, 1974; Etaugh & Brown, 1975;
Garland & Price, 1977; Heilman & Stopeck, 1985a;
L'Heureux-Barrett & Barnes-Farrell, 1991: Nieva & Gutek,
1980: Pazy, 1986; Pence, Pendleton, Dobbins, & Sgro, 1982).
Research also suggests that this effect can be attenuated by
providing information which is clearly related to
performance outcomes or causes of performance outcomes

(Deaux & Emswiller, 1974: Heilman & Guzzo, 1978: Terborg &

Ilgen, 1975).

35C

wit

IE

of

of

de

1.:

6
Most studies of how stereotypes influence causal
ascriptions to observed performances have been conducted
within the framework of attribution theory. The causal
ascriptions, or attributions, which a person makes to an

observed event are hypothesized to impact on a person's

affect, attitude, future expectancies, behavior, motivation,

judgments, and decisions (Green & Mitchell, 1979: Heider,

1958; Weiner, 1974, 1985, 1986; Weiner, Frieze, Kukla, Reed,

Rest, & Rosenbaum 1971). These attributions can be affected

by gender stereotypes, resulting in gender-related bias with

regard to personnel decisions and evaluations (Feldman,

1981: Heilman, 1983; Ilgen & Youtz, 1986).

The current research effort attempts to integrate
recent developments in attribution theory with the effects

of gender stereotypes on personnel decision making.

Previous research using attribution theory as an explanation

of personnel decision outcomes has not considered the effect

of causes perceived to be due to controllable factors on

decision making processes. The perceived controllability or

responsibility of a performance outcome has been identified

by attribution researchers as an important dimension of

causal factors (Folkes & Marcoux, 1984; Weiner, 1985; 1986;
Weiner, Perry, & Magnusson, 1988). Also, research has

infrequently investigated the presence of an

ingroup/outgroup gender bias on attributions, though a

growing body of research indicates that this might be an

important component of the attributional process (Hewstone,

md
rig
the

er;

in

u.
Ll

7
1990: Hewstone & Ward, 1985; Jackson, Sullivan, & Hodge,
1993: Taylor & Jaggi, 1974).

This study also looks more carefully at the conditions
under which personnel decisions are often made and how these
might actually affect attributional processes. For example,
when considering promotions, decision makers rarely consider
employees who are below average; instead, decision makers
typically only consider above average performers. An
interesting question is thus posed: Do gender-related
attributional biases occur for employees who are all above
average in performance? Conflicting hypotheses can be
generated on the basis of the previously cited research.

Finally, this research effort also attempts to develop
and test a conceptual model of how gender stereotypes
influence attributional processes and how the attributions
made as a result of those processes can affect different

types of personnel decisions. In order to develop and

explain the conceptual model, it will be helpful to review

the literature on gender stereotypes, attribution theory,
and personnel decision making.

The literature review will occur in three parts.
First, the conceptual foundations of attribution theory and
effects of attributions on leader and managerial decisions
and behavior will be examined. Recent advances in
attribution theory will be mentioned as an important but
neglected area of personnel decision making research.

Second, the effects of gender stereotypes on personnel

dec:

att:

res
per
prc

vil

8

decision making will be reviewed, and the role of
attributions in making these decisions will be explored.
Finally, a conceptual model will be presented to integrate

research on gender stereotypes, attribution theory, and

personnel decision making. This model will deal with

problems not considered in prior research efforts, and it

will yield several testable hypotheses.

ca:
lit
9“:
is

be

LL

REVIEW OF THE LITERATURE

Attribution Theory and Leader Decisions
Attribution theory. Attribution theory is about the

causal explanations of events which occur as a part of daily
life (Heider, 1958). The theory mainly developed as an
explanation of social perception. For example, if someone
is successful at a task, different causal explanations can
be made about why that person is successful. Kelly (1971,
1973), Jones (1979), and Jones and Davis (1965) describe the
process as one whereby people function as "naive
psychologists" and automatically try to make sense of the
world around them. Part of this process includes inferring
the reason or cause of a particular observed event.
Research has supported the idea that people spontaneously
generate a variety of causes for various types of observed
events and outcomes (Anderson, 1983: Frieze, 1976: Kelley,
1973: Meyer, 1980; Weiner et al., 1971: Weiner, 1986).
According to attribution researchers, the cause which
is attributed to the observed behavior of a person will have
different effects on how that person is perceived and
treated. Weiner et a1. (1971), identified locus of

causation and stability as two important causal dimensions

bel:

tOl

Crc
cle

Cal

’Q’

10
believed to account for most of the attributions people make

to explain success and failure.

Locus of Causality (Internal or External): The degree

to which a cause is located within a person,
rather than outside the person (Weiner, 1979).
Stability: The degree to which a cause can be expected
to be present at the same level every time the
same situation arises (Weiner, 1979).
Crossing these two dimensions yields a two by two
classification system with four possible conditions of
causality: internal and stable (e.g., ability); internal
and unstable (e.g., effort): external and stable (e.g., task
difficulty); and external and unstable (e.g., luck). As
explanations of observed behaviors, these four causes have
been linked to a wide variety of observer reactions,
including helping behavior, affect, and expectancies for
future behaviors (Valle & Frieze, 1976: Harvey & Weary,
1984; Weiner, 1972, 1974, 1985, 1986).

According to Weiner (1972), affective or evaluative
reactions to task outcomes are held to be a function of the
perceived reasons for these outcomes. For example, Weiner
and Kukla (1970), Leventhal and Michaels (1971), and Rest,
Nierenberg, Weiner, and Heckhausen (1973) found that
attributions of effort led to more intense reactions to good
performers than did attributions of ability.

It should be noted that more recent work by

attributional researchers (e.g Weiner, 1985, 1986) has

ma:

res

has

1'8!

at‘

pe

la

Gr

3C

E;

I‘.

11
suggested the presence of a third attributional dimension,
termed controllability. This third dimension of causal
attributions will be discussed in a later section.

0 s eh ' 5. Performance
attributions to ability, effort, task difficulty, and luck
have been linked to a wide variety of supervisory,
managerial, and leader behaviors and decisions made in
response to observed performance outcomes. Research which
has examined the effects of attributions on supervisory
responses to subordinate performance has found that the
attributions which are made about a subordinate's
performance will affect feedback and communication processes
(Dugan, 1989: Gioia & Sims, 1986: Ilgen & Knowlton, 1980:
Martinko & Gardner, 1987; Mitchell & Wood, 1980: Heneman,
Greenberger, & Anonyuo, 1989) and supervisory decisions and
actions (Goodstadt & Kipnis, 1970; Kipnis & Cosentino, 1969:
Green & Mitchell, 1979; Knowlton & Mitchell, 1980; Mitchell
& O'Reilly, 1990).

Dugan (1989) studied leader communication patterns
using 52 MBA students who played the part of a manager in a
feedback session. She found a consistent relational
communication pattern, or script, occurred during the
feedback process and this script was influenced by the
manager's initial attributions of effort or ability as the
reason for the subordinate's poor performance. The
attributions made also affected salary decisions such that

when poor performance was attributed to lack of effort,

12
decisions were more punitive. Similarly, in a study using
40 people who acted as supervisors of three person work
groups, Ilgen and Knowlton (1980) found the nature of
feedback given by supervisors was strongly affected by
performance level when the attribution of performance was to
effort, but no difference was found in the nature of
feedback when the attribution was to ability. These studies
indicate attributions of effort and ability interact with
observed performance to affect feedback behaviors of
supervisors and managers such that effort attributions
result in more extreme responses than do ability
attributions.

In addition to affecting attitudes and communication
behaviors, attributions have been linked to a variety of
leader/supervisory decisions and reactions to performance.
Green and Liden (1980) found a supervisor's belief about the
cause of a subordinate's poor performance affected the
supervisor's reaction to that performance and the extent to
which he or she used control policies (verbal warnings vs.
docked pay). When the subordinate was seen as the causal
factor of performance, control responses were directed more
at the subordinate, more punitive, and included more change
to his or her job than when external factors were seen as
the cause of performance. Again, causal attributions were
found to affect supervisory behaviors.

Knowlton and Mitchell (1980) examined the effects of a

supervisor's attributions about the causes of a

sul

sul

as:

til:

we

1e

ef

te

13
subordinate's poor performance on the evaluation of that
subordinate. Forty undergraduate students were randomly
assigned to one of four conditions in which they supervised
three other students, actually confederates, working on a
questionnaire coding task. One subordinate performed either
well or poorly relative to the other two, and the leader was
led to believe that this performance was due to either
effort (behavioral cues) or ability (scores on an ability
test). The supervisor then evaluated the "key" subordinate
on his/her performance. Results indicated that attributions
to effort led to more extreme evaluations than attributions
to ability: this is consistent with the literature cited
above. An attribution of effort as a cause of performance
led to higher evaluations when performance was high and to
lower evaluations when performance was low than when ability
was seen as a cause of performance.

In an effort to explain how attributions can affect
leader behaviors and decisions, Green and Mitchell (1979)
proposed a two-stage process model in which the leader
functions as an information processor. In the first stage,
leaders engage in diagnosis or explanation of observed
behaviors or performance (attributions). In the second
stage, leaders respond to the behaviors or performance in a
manner consistent with the attributions made. Green and
Mitchell state that the attributions which the supervisor
makes can affect leader behaviors such as 1) rewarding and

punishing member performance: 2) closeness of supervision;

3)
4)
t0]

Si]

50
C3.

Se

i:

14
3) expectancies about the member's future performance: and
4) aspirations the leader might hold for a member. Support
for the model has been consistent (Dugan, 1989: Gioia &
Sims, 1986; Mitchell & Kalb, 1982: Mitchell & Wood, 1980:
Wood & Mitchell: 1981) .

Taken together, the results of the studies cited above
suggest three things. First, supervisors and managers make
causal attributions for observed performance outcomes.
Second, attributions affect supervisory reactions to
performance such that perceived internal causes are more
likely to result in responses directed at the subordinate
than are perceived external causes. Third, effort
attributions result in more extreme evaluations and
responses to performance than do ability attributions; this
is particularly true when supervisors are attempting to
change subordinate behaviors. In other circumstances, like
recruitment, ability may lead to stronger supervisory
responses. For example, Nicholls (1976) found that people
preferred working with people of high ability over working
with people who put forth high effort, although they
considered high effort to be a very positive characteristic.
While many researchers have viewed the differential effects
of ability and effort as a function of the stability
dimension of attributions, recent work suggests that the
strong effect of effort attributions as compared to ability
attributions can be explained by the presence of a third

dimension in the attributional framework.

add

IEE

in
un

cl:-

:1

15

Controllability as an attributional dimension. In
addition to locus of causation and stability, attribution
researchers have hypothesized the existence of a third
dimension of causation, termed controllability (Weiner,
1979, 1985). Controllability is not the same as locus of
causation because a cause may be internal to a person but
the cause still may not be considered intentional or
controllable.

For example, attributions of beauty or ability, while
internal and stable, are generally viewed as unmodifiable or
uncontrollable because one cannot change one's innate
characteristics. On the other hand, consistent effort may
also be considered to be an internal and stable cause, but
it will be considered to be more controllable than ability
or beauty (Weiner, 1985, 1986).

Researchers have frequently interpreted locus of
causality in a fashion similar to controllability. Mitchell
and Wood (1980) stated: "When an internal attribution is
made, we would expect the leader to direct the response
toward the subordinate in an attempt to change the
subordinate's behavior through feedback, punishment, or
training" (p. 125, emphasis added). Ilgen and Knowlton
(1980) used the following citation as an explanation for how
attributions may affect the feedback process: "With regard
to performance feedback, the primary attributional concerns

are those related to internal factors. Since feedback is

information directed at the subordinate in order to change

OI

fa

llC-

16
or maintain his or her behavior, we can assume that the
conveyor of the feedback believes that the individual to
whom the feedback is directed gan_haye_§gm§_effegt on
performance." (Knowlton & Mitchell, Note 3, cited in Ilgen &
Knowlton, 1980, emphasis added).

The notion that a subordinate can change or have an
effect on his or her behavior or performance level suggests
the subordinate can control the observed performance level
or behavior to some degree. However, as noted above, the
fact that a cause has an internal locus of causation does
not mean the cause is within a person's control.

Weiner (1974, 1979, 1985) modified the original
classification system of locus of causation and stability to
include a third dimension, labeled intentionality, or
controllability. Intentionality had been originally
identified by Heider (1958) as an important causal
attribution that people make, but more recently, Weiner
(1979, 1985) has argued that a more appropriate name for it
is controllability, although he acknowledges that the two
concepts may not be identical.

Support for the existence of this third dimension has
been provided by a number of studies (e.g., Anderson, 1983;
Meyer, 1980; Weiner, 1986). Controllability has been
variously labeled as the degree to which a cause is a factor
that the person has control over (Weiner, 1979, 1986); the
degree to which a cause reflects an intention (Weiner,

1974): or the degree to which the cause implies that the

per

no:
no:
ge:
ca'
21
10
co

C6

17
person can change the factors that caused the outcome
(Anderson, 1983).

Anderson (1983) investigated causal structures for each
of four types of situations: interpersonal failure,
noninterpersonal failure, interpersonal success, and
noninterpersonal success. One group of subjects (n = 24)
generated causes, another group (n = 22) clustered these
causes into a variety of categories, and a third group (n =
21) rated these causes on the dimensions of changeability,
locus, stability, intentionality, globality, and
controllability. Anderson found that different types of
causes were generated for different situations and these
causes differed in dimensional location. He also found that
changeability, controllability, and intentionality were
highly intercorrelated (rs ranged from .81 to .93). The
results suggest that causes which are spontaneously
generated by subjects for diverse classes of events vary
along the dimensions of locus of causality, stability, and
controllability. Results also support the unidimensionality
of intentionality, controllability, and changeability
factors.

Meyer (1980) had 200 subjects judge how strongly each
of nine different causes may have influenced the performance
of hypothetical students on an exam. Results using a
PARAFAC factor analysis procedure (Harshman, 1976, cited in
Meyer, 1980) revealed three factors that corresponded to the

stability, locus, and control dimensions proposed by Weiner.

Age
cor
cat
ex:

tci

Gi

le

18

Again, based on the findings by Meyer, it appears that
controllability should be included as a third dimension of
causal attributions. Weiner (1979, 1986) recognized the
existence of the controllability dimension and gave it the
following definition:

QQDELQIIQ2111£¥= The degree to which a cause is a

factor that the person has control over.
Unfortunately, until recently (Dugan, 1989; Ford, 1985;

Gioia & Sims, 1986), most of the research on supervisory and

leader decisions has only looked at the effects of locus of
causation and stability. Little work has been done to
integrate developments in attribution theory by using the
controllability dimension in research on organizational
behavior (Ilgen & Klein, 1988). While most of the findings
cited earlier have been interpreted as supporting the idea
that the internal/external dimension of causation is a
primary determinant of supervisory reactions, the alternate
explanation of controllability has been ignored in the
interpretation of results. For example, Green and Mitchell
(1979) only considered locus of causation and stability as
causal dimensions, but they state that a leader's perception
of responsibility is a crucial judgment which accompanies
attributions which might be made. Ford (1985), building on
Weiner's revised attributional framework, noted that
perceived controllability of a performance downturn could

affect a decision makers' responses to that downturn.

rel
ott

fut

(l!
at‘
at

1'1

19

In other domains of research, controllability has been
related to the likelihood to compliment, date, or help
others; interpersonal affect: and predictions made for
future behaviors (Folkes & Marcoux, 1983; Weiner, 1985;
Weiner, Perry, & Magnusson, 1988). Folkes and Marcoux
(1983) manipulated the controllability and stability
attributions of physical attractiveness, and examined the
effects of those attributions on perceived attractiveness,
likelihood to date, and likelihood to compliment.

The controllability of attractiveness was found to
affect the likelihood to compliment, and the stability of
attractiveness was found to affect the likelihood to date.
It was hypothesized that controllability affected
complimenting behaviors because it was related to the amount
of credit or blame an individual received (Heider, 1958).

In other words, if a person exercised volitional control
over a change which occurred, then that person received
corresponding amounts of credit or blame based on that
control.

Stability was believed to affect the likelihood to date
because it predicted and increased expectancies for future
outcomes (Valle & Frieze, 1976). The prediction of future
outcomes was important for affecting the likelihood to date
a person, because dating a person implied a longer term
relationship than complementing a person. Instability in
behaviors was found to affect likelihood to compliment

because complimenting unstable behaviors was more

at

10

20

instrumental for changing future behaviors than
complementing stable ones (Weiner, 1979). The findings of
Folkes and Marcoux (1983) and Weiner and colleagues (Weiner,
1985, 1986; Weiner, Perry, & Magnusson, 1988) indicate that
controllability has effects on behaviors, attitudes, and
affect, separate from stability and locus of causation.

Integrating the controllability dimension into the
attributional taxonomy by crossing controllability with
locus of causation and stability yields eight cells (2 x 2 X
2). These eight cells are depicted in Figure 1, based on
modifications of Weiner's (1979) original taxonomy by Gioia
and Sims (1985) for the purposes of organizational research.

Support for the existence of the causal attributions
defined by the eight cell taxonomy was found previously by
Frieze (1976), who had subjects generate and classify causes
for a variety of performance outcomes. Causes which were
generated by one group of subjects were categorized by
another group of raters into clusters of categories which
were subsequently identified as ability, effort, stable
effort, task, luck, mood, other person, and miscellaneous
causes. A comparison of these categories with the 2 X 2 X 2
taxonomy depicted in Figure 1 reveals consistencies between
the taxonomy and the categories. For example, both contain
two types of effort, typical and specific, though they are
not given the same labels. Also, both contain ability,
luck, task difficulty, mood, and other person causes of

observed outcomes. These categories also closely parallel

21

Stable Unstable

 

Typical Effort Specific Effort
Controllable

 

 

 

Internal
Ability Mood
Uncontrollable
Controllable Supervision Coworkers
External
Uncontrollable Task Difficulty Luck

 

 

 

 

Adapted from Gioia and Sims (1986)

1199111.;- Eight—Cell Taxonomy of Causal Attributions

the
res

5pm

ext
lllll
CO

th

tl

22
the causes spontaneously generated by the subjects in the
research done by Anderson (1983). Thus, empirical work on
spontaneously generated and classified causal attributions
supports the integration of the controllability dimension
into the attributional taxonomy.

However, when looking at how attributions might affect
the evaluation of another's behavior, a conceptual
difficulty occurs when including the controllability
dimension in the attributional framework. When a cause is
external to a person, it is, by definition, likely to be
uncontrollable by that person. While that same cause may be
controllable by people other than the actor, as implied in
the above matrix, controllability of a cause by others often
will have little or no bearing on how the actor is treated
or evaluated. This is because controllability connotes the
concept of moral responsibility. In many situations, and
especially in leader/subordinate relationships, people will
think it unreasonable to think of holding a person
responsible for the outcomes of actions of others. Because
of this problem, controllability is reconceptualized as a
subdimension of locus of causation when one is evaluating a
target actor. This taxonomy provides for three dimensions
of responsibility: internal—controllable, internal-
uncontrollable, external-uncontrollable.

Crossing stability with internal-controllable,
internal-uncontrollable, and external-uncontrollable yields

the six cells presented in Figure 2. Using this framework,

Internal
Controllable

Internal
Uncontrollable

External

, Uncontrollable

23

Stable

Unstable

 

Typical Effort

Specific Effort

 

 

 

Ability Mood
Luck/
Task Difficulty Influence of
Others

 

 

Figure-1. Six-Cell Taxonomy of Causal Attributions

 

one
int:

sub<

to '
con
att
a s

su;

24
one can begin to integrate the controllability dimension
into research on leader or managerial responses to
subordinate performance.

Since locus of causality and stability have been found
to be related to supervisory responses, and since
controllability seems to be an important dimension of causal
attributions, it follows that controllability will also have
a strong effect on personnel decisions made by leaders or
supervisors. Mitchell and O'Reilly (1990) note:

"One crucial judgment which accompanies attributions is

that of responsibility, referring principally to an

evaluation of moral accountability. Judgement of
responsibility is seen as moderating the leader's
response to an attribution. The more an outcome is
seen as caused by some aspect of the subordinate, the
more likely the leader is to take action toward the

subordinate" (p. 169).

In summary, the literature review on attribution theory
has highlighted several important points. First, people
seem to make spontaneous inferences of causes of performance
outcomes. Second, these inferred causes vary on dimensions
of locus of causality, stability, and controllability.

Third, these dimensions of causation affect leader behaviors
such that the more internal the cause of a performance
outcome, the more the leader's response is directed at the

subordinate; the more controllable the cause, the more

ext]

C305

25

extreme the response to the behavior; the more stable the
cause, the more it affects expectancies for future outcomes.

Since attributions seem to play an important part in
leader or supervisory behaviors and decisions, it is
important to identify processes by which attributions are
formed. As noted in the introduction, stereotypes can
affect the way in which information is processed, encoded,
retrieved, or interpreted. Consequently, one would expect
that stereotypes will affect attributional formation.
Gende; Effects on Attributions and Personnel Decisipns

Gender stereotypes and attributions. Researchers have
integrated stereotype and attribution theory by arguing that
stereotypes, when activated, can bias subsequent information
processing of causal attributions (Ilgen & Youtz, 1986;
Feldman, 1990; Lord & Maher, 1991: Pazy, 1986), since
knowledge structures serve as reference systems according to
which new information is categorized (Wyer & Srull, 1986),
and stereotypes are knowledge structures (Bierhoff, 1989).

Stereotypes serve to provide expectancies about traits
and behavior (Bierhoff, 1989: Deaux, 1976) and stereotype
consistent behavior is considered to be more stable,
internal, and innate than stereotype inconsistent behavior
(Bodenhausen & Wyer, 1985: Deaux & Emswiller, 1974; Deaux,
1976: Gordon, 1990: Hewstone, 1990). Predictions about the
effects of stereotypes on controllability are nebulous
because no research was identified which investigated the

effects of stereotypes on the controllability dimension of

atti
per:
{Elli
due

per

att

rer
cu:
th
$11

to

26
attributions. However, as previously noted, good
performance which is attributed to internal causes is
rewarded more than good performance which is believed to be
due to external causes: thus, gender stereotypes may affect
personnel decisions through attributional processes.

In a widely cited study of the effects of gender on
attributional processes, Deaux and Emswiller (1974) had 130
undergraduate students listen to confederates attempt to
recognize common objects (lug wrenches, screwdrivers, irons,
curlers, etc.) as part of their task. The performance of
the confederates was manipulated (good or bad) and the
subjects then made ratings of attributions to luck or skill
for the performance of the confederate. When the task was
masculine and performance high, male performance was
attributed to ability, while an equivalent performance by a
female on the same task was attributed to luck. These
attributional biases occurred in spite of the fact that the
subjects viewed the males and females as having performed
equally well, as measured by a manipulation check.

The findings were explained by Deaux (1976), who argued
that a woman's work success is ordinarily attributed to
factors other than her capability because of sex-role
expectations. Success for women is more likely to be
attributed to luck (Deaux & Emswiller, 1974), ease of task

(Feldman-Summers & Kiesler, 1974), or to hard work

(Feldman-Summers & Kiesler, 1974) than to skill or ability.

expl
Fel:
dev-
The
of
Sut
ex;
prc
fa:
be
st

at

27

Another example of how gender stereotypes affected
expectations and attributions was provided when
Feldman-Summers and Kiesler (1974) were attempting to
develop research materials for a gender stereotype study.
They showed 84 male and female undergraduates descriptions
of several people in different professional occupations.
Subjects were asked to indicate how successful he or she
expected a presented professional to be: half of the
professionals were presented as male, half as female. There
was not a single instance in which a woman was expected to
be a more successful professional than a man. In the same
study they found that male subjects provided more
attributions to ability for a male physician, and effort and
easy task attributions for female physicians. Female
subjects did not attribute more ability to the male
physician than the female physician, but they did see
greater effort as a cause of the female doctor's success.
Again, findings support the concept that gender stereotypes
affect expectancies and attributions for performance:
success for men was perceived to be due to ability while
success for women was perceived to be due to effort, though
this effect was diminished for female subjects.

In a study which attempted to directly measure gender
stereotypes and effects of those stereotypes on
attributions, Garland and Price (1977) had 123 male
undergraduates read a description of a successful or

unsuccessful female manager. They later made causal

attr

VEIE

dew

Gen

the

5’01

att

pos

at‘

an

80

28
attributions for her success or failure; gender stereotypes
were measured by the Women as Managers Scale (WAMS),
developed by Terborg, Peters, Ilgen, and Smith (1977).
Generalized attitudes toward women managers as measured by
the WAMS were unaffected by success/failure descriptions of
women managers, but WAMS scores were strongly related to
attributions of female success in management. WAMS
positively correlated with internal (ability and effort)
attributions and negatively correlated with external (luck
and easy job) attributions for subjects reading the female
success descriptions. WAMS scores did not correlate with
attributions for subjects who read female failure
descriptions.

These results suggest that gender stereotypes affect
attributions for successful performance of women managers
but not for unsuccessful performance. When successful women
managers were presented, positive attitudes toward women
managers led to internal attributions of performance, and
reduced external attributions of performance. Findings also
suggest that gender stereotypes are not easily affected by
presentations of successful or unsuccessful women managers.

Stevens and DeNisi (1980) replicated the Garland and
Price findings when they investigated the effects of gender
stereotypes on attributions made with regard to performance
and subsequent personnel decisions. In study one, they used
the responses of 102 undergraduate subjects to test the

hypothesis that WAMS scores would not be affected by the

pres
tha‘
suc
and
eas

per

SCl

pe:
st
te

SC

ta

fl

In

29
presentation of a good or poor female employee. They found
that scores on the WAMS scale were not affected by the
success/failure manipulation. This supported the Garland
and Price (1977) findings; gender stereotypes were not
easily affected by presentation of a female employee's
performance.

Study two was designed to test the hypothesis that WAMS
scores would be related to the causal attributions of
performance for a good or poor female employee. For this
study, 124 subjects (80 males and 44 females) were used to
test their hypotheses. For male subjects, they found WAMS
scores were positively related to attributions of ability (;
= .49), and effort (; = .33) for high performing female
target employees and negatively related to ability (; =
-.18) and effort (3 = -.41) attributions for low performing
female target employees. For female subjects, WAMS scores
were not significantly related to any attributions in either
condition, though all of the correlations were in directions
consistent with the male subjects' findings. The authors
suggest the lack of significant effects for the female WAMS
scores on attributions may have been due to restriction in
range on WAMS scores.

Findings from the Stevens and DeNisi (1980) study
support the Garland and Price (1977) findings. They suggest
the content of stereotypes were not easily affected by
presentations of poor or good performing female managers and

that stereotypes of women managers affected attributions of

fen;
man
suc
at:
was
sul
neg

mi;

St
ca

st

30
female performance. Positive attitudes toward women
managers were positively related to internal attributions of
successful performance and negatively related to internal
attributions of unsuccessful performance. This relationship
was found to be true for male subjects but not for female
subjects; possibly because women were less likely to have
negative stereotypes regarding female managers and this
might produce restriction in range.

In conclusion, it appears that gender and gender
stereotypes can have a significant impact on perceptions of
causes of employee performance. People who have negative
stereotypes of women are more likely to attribute successful
performance of women to external causes than to internal
causes, and more likely to attribute unsuccessful
performance to internal causes than to external ones. The
attributional biases which have been detected as a result of
gender stereotypes may result in large effects on personnel
decision making, because attributions affect leader
responses to performance outcomes. Several researchers have
hypothesized and found attributional biases to be especially
strong for sex-typed tasks or jobs (Deaux, 1976; Deaux &
Emswiller, 1974; Heilman, 1983). In accord with the logic
stated above, if gender stereotypes are expected to affect
managerial personnel decisions through attributional
processes, it must be demonstrated that managers actually
have gender stereotypes which may be biased against

perceptions of females as managers.

 

197
Blc
we:
we:
lil
sue
Us

me

31

Gender stereotypes and female managers. Schein (1973,
1975), Brenner, Tomkiewicz, and Schein (1989), and Heilman,
Block, Martell, and Simon (1989) found that men in general
were perceived to be more similar to middle managers than
were women in general. Schein (1973) had 300 male middle
line managers rate women in general, men in general, and
successful middle managers on a 92-item descriptive index.
Using intraclass correlation coefficients, she found that
men were perceived to be more similar to middle mangers (p;
= .62) than were women (n; = .06). In a follow-up study,
Schein (1975) had 300 male and 167 female managers complete
the same ratings, and she found men and successful managers
were perceived to have leadership, competitiveness, self-
confidence, objectivity, and aggressiveness characteristics,
but women were not. Again, managers perceived men in
general to be more similar to managers in general than women
in general; this was true for both male and female managers,
but less so for female managers.

Two recent studies by Heilman et al. (1989) and Brenner
et al. (1989) suggest attitudes toward women, men, and
managers have changed very little since Schein (1973, 1975)
conducted her work. Heilman et al. (1989) had 268 managers
rate one of seven target groups using the 92-item inventory
developed by Schein. These managers rated men (in general,
managers, or successful managers), women (in general,
managers, or successful managers), and successful managers.

The results of the study closely replicated Schein's

IGS‘

50C

00:

SEE

me:

In

32

results. Men in general were described as more similar to
successful managers (p; = .54) than were women in general
(3; = -.24). The correspondence between descriptions of
women and successful managers increased dramatically when
women were depicted as managers, but they continued to be
seen as more different from successful managers than were
men.

Furthermore, even when women were depicted as managers,
perceived differences in many characteristics central to
managerial performance still existed, and these differences
dissipated only when women were clearly labeled as
successful managers. These findings suggest people who are
actually employed as managers still perceive women to be
less similar to successful managers than men and these
perceptions appear to be resistant to change.

Findings by Brenner et al. (1989) indicate male
managers still have stereotypes about the fit between
females in general and managers in general, but female
managers seem to have changed in their perceptions about the
similarity between women in general and managers in general.
Brenner et al. (1989) conducted a study which involved 420
men and 173 women who were managers. They found gender of
the subject affected perceptions of how similar women in
general were to successful middle-managers. For male
managers, the intraclass correlation coefficient for
descriptions of men in general and managers was higher

(r; = .72) than it was for women in general and managers

bet
fez
ma]
be
su

di

33
(p; = -.01). For female managers, there was little
difference in how similar men and women in general were to
successful managers (p; = .59 and .52, respectively).

Thus, it appears that attitudes regarding the fit
between women and management positions have changed for
female managers, but attitudes have changed little for male
managers. Female managers seem to see little difference
between how similar men and women in general are to
successful managers, while male managers seem to see a large
difference between how similar men and women in general are
to successful managers.

The literature reviewed thus far has suggested several
ideas. First, people make causal attributions about
performance outcomes and these attributions affect leader
and managerial decisions and behavior. Second, stereotypes
affect attributions made about successful female
performance, particularly for sex—typed occupations. Third,
managers who are actually employed in businesses seem to
have gender stereotypes which lead to differences in
expected fit between women and men in general and successful
managers. Since stereotypes affect expectancies, and
expectancies affect attitudes and behavior, one would
surmise that gender might affect personnel decision making,
especially for strongly sex-typed occupations. Research
findings support this logic.

d ste o s a d ersonn decisions. Cohen and

Bunker (1975) conducted a study using 150 job recruiters who

 

made
des

The

e:;
til
rec
Thc
co
jo

st

34
made a series of recommendations and responses to job
descriptions, application blanks, and interview transcripts.
They found a high degree of sex discrimination among job
recruiters at two university placement centers.
Significantly more females than males were recommended for
employment in an editorial assistant position (female-
typed), while significantly more males than females were
recommended for a personnel technician job (male-typed).
These decisions appeared to be made on the basis of
congruence of sex of the applicant to the sex-type of the
job. Similarly, Terborg and Ilgen (1975) found that male
students who were asked to allocate starting and second year
salaries for engineering jobs to males or females with
identical qualifications and performance records recommended
higher levels of financial compensation for males than for
females.

Heilman and Stopek (1985a, 1985b) hypothesized that
attractiveness might exaggerate gender-related perceptions.
An attractive male was expected to be perceived as more
prototypically masculine, while an attractive female was
expected to be perceived to be more prototypically feminine.
If attractiveness affects perceptions of masculinity or
femininity, then it should also affect gender-related bias
in attributions and personnel decisions.

Their findings partially support their hypotheses.
Attractiveness affected the perceived femininity of females,

but attractiveness did not affect the perceived masculinity

 

of 1
rel.
per
ml

mi

1101
ur.
a1

fi

35

of males. For males, there were no differences in work-
related evaluations nor in perceived appropriateness of
personnel decisions for either attractive or unattractive
males in either occupation. For females, promotions, pay
raises, and merit pay in managerial jobs were deemed more
appropriate for unattractive females than for attractive
females; unattractive females in managerial jobs also
received higher work-related evaluations. In contrast,
promotions and pay raises for clerical jobs were perceived
more appropriate for attractive females than for
unattractive females; attractive females in clerical jobs
also received higher work-related evaluations. These
findings are quite consistent with studies cited earlier:
personnel decisions can be affected by gender stereotypes.

If gender stereotypes affect personnel decisions and
evaluations, they may do so through effects on causal
attributions. In a study which examined the effects of male
and female attractiveness on causal attributions and
perceived masculinity and femininity, Heilman and Stopek
(1985a) found evidence that perceived masculinity or
femininity affected attributions made about the successful
performance of male and female managers.

They presented 113 working men and women (52% men and
48% women) with information on a hypothetical managerial
employee. They found that attractiveness of a manager was
related to the causal attributions of successful performance

for that manager. Success of unattractive female managers

was

man

att

Att

att

for

llllE

an

tel

00

'le

36
was attributed more to ability than for attractive female
managers, and luck was attributed more as a cause for
attractive female success than unattractive female success.
Attributions were the opposite for males; ability was
attributed more as a cause for attractive male success than
for unattractive males, and luck was attributed more to
unattractive males than to attractive males. There was also
an effect of attractiveness on ratings of masculinity and
femininity such that attractive males were perceived to be
more masculine than other employees and attractive females
were perceived to be more feminine. Heilman and Stopek
(1985a) conclude that being attractive can have negative
consequences for women managers, even when they clearly have
been successful and have reached the executive level. For
attractive females, success was attributed less to ability,
and they were consistently judged to be less capable than
were unattractive managers.

The results of the work by Heilman and Stopek (1985a,
1985b) were consistent with other findings cited earlier:
gender stereotypes (i.e. perceived masculinity or
femininity) affected personnel decisions and attributions
for successful performance of male and female managers.
Successful managerial performance of individuals perceived
to be more masculine was attributed to internal and stable
causes (e.g., ability) and it was rewarded more highly than
successful managerial performance of people perceived to be

less masculine. Successful performance by less masculine

 

man

hi
the
fa:
su].
un
uh
pe
as

ta

37
managers was attributed to external and unstable causes
(e.g., luck) and it was not rewarded as highly.

Pence, Pendleton, Dobbins, and Sgro (1982) did a study
which examined the effects of sex of the evaluator, sex of
the employee, and different causal explanations for employee
failure upon ratings of the appropriateness of eleven
supervisory actions. One hundred male and 100 female
undergraduate subjects were presented with employee files
which consisted of four males or four females whose
performance in a management-trainee position was described
as unsatisfactory. Causal attributions of ability, effort,
task difficulty, or luck were each assigned to one of the
four target employees. Subjects were then asked to indicate
why they thought the employee failed and to rate the
appropriateness of 11 corrective actions for each of the
four employees.

Results indicated that gender and causal attributions
influenced subject ratings of the appropriateness of various
corrective actions such that coercive actions were
considered most appropriate for failure due to lack of
effort, and that in general, subjects considered it more
appropriate to work with and encourage employees of the same
sex. With regard to causal attributions, the authors state
"that in addition to the internality of the cause for
failure, subjects may concomitantly consider stability of
the cause for failure". The authors go on to say, "coercive

actions are considered most appropriate when failure is due

 

to I
com!
a c
act
sch
the
efl
de<
an:
hit
in
di

in

38
to lack of effort which is internal and unstable". Their
conclusion is that researchers must examine the stability of
a causal factor and the intentionality or choice of the
actions of the employee as dimensions underlying the causal
schema used by subjects. It seems appropriate to interpret
the conclusions drawn by these authors as support for the
effects of the dimension of controllability on personnel
decisions, though they only investigated locus of causality
and stability of attributions. Thus, the earlier
hypothesized distinction between internal-controllable,
internal-uncontrollable, and external-uncontrollable
dimensions has received partial support in the
interpretation of findings by the authors of this study.

The findings regarding the effects of gender on
personnel decisions and causal attributions have been
explained by the "lack of fit" model proposed by Heilman
(1983). In this model, she hypothesizes that perceived job
requirements are "fitted" against the perceived attributes
of an individual. This fitting process affects performance
expectations such that poor fits lead to expectations of
failure and good fits lead to expectations of success.
Expectations are believed to affect the evaluation process,
which in turn, affects selection decisions, performance
appraisals, and reward allocation.

Interestingly, Heilman's (1983) model is consistent in
several ways with Green and Mitchell's (1979) two-stage

model of leader behavior. In Green and Mitchell's model,

lea
aft

ide

EVE

at:
an:
be
ch

te

39
leaders make diagnoses or attributions, and these in turn
affect leader behaviors. Heilman's model simply adds the
idea that the person/job fit (i.e. gender stereotypes) can
affect performance expectations, which will affect the
evaluation or attributional process.

It is clear that stereotype-based processing is often
affected by the perceived salient characteristics of others
and they tend to result in some identification of group
membership. For example, gender is an easily identified
characteristic, and people often identify themselves in
terms of whether they are a member of the male or female
category, especially for gender salient environments. Some
researchers postulate that we not only have stereotypes
about other groups, but that we have stereotypes about the
group to which we perceive ourselves as belonging, and that
this may affect attributional processes (Taylor & Jaggi,
1979; Taylor & Moghaddam, 1987; Hewstone, 1990; Hewstone &
Ward, 1985).

Effects of_gender gropp membership. The theory that
group membership affects attributional processes is built
upon Heider's (1958) 'fundamental attribution error', which
is the tendency to underestimate situational factors and
overestimate personal factors as causes of an actor's
behavior. Zuckerman (1979) found that perceivers often give
relatively more internal attributions for their own success
and relatively more external attributions for their own

failure. Actor/observer differences and self-serving biases

haw

(Br

and
em
S€I
auc
cor
ac
be
ha

DC

at

40
have also been found in leader-subordinate relationships
(Brown, 1984; Gioia & Sims, 1985; Green & Mitchell, 1979).
Actors often attribute their failure to external causes

and their success to internal causes, while observers

exhibit the opposite pattern. Actor biases can be self—

serving and can also be based on the fact that actors have

much more information than observers about the particular

context in which a behavior takes place. These

actor/observer and self-serving attributional biases have
been extended to the group context, though most of the work
has been done on ingroup/outgroup differences in ethnicity,

not gender.
Pettigrew (1979) extended self-serving attributional

biases to the group context. He referred to an ultimate

attribution error, which is a "systematic patterning of
intergroup misattributions shaped in part by prejudice" (p.
464). He crossed controllability with locus of causation to

yield four main attributional categories for ultimate

attribution error:

1) the exceptional case (low control/internal)
2) luck or special advantage (low control/external)
3) high motivation and effort (high control/internal)

4) manipulable situational context (high

control/external)

Pettigrew makes five predictions about the attributions one

group member will make about another outgroup member's

behavior:

l)

2)

3)

4)

5)

41
Negative acts (antisocial or undesirable) by an

outgroup member will be attributed to personal,

dispositional causes. Often internal causes will

be perceived as innate characteristics of that

outgroup member.

Positive acts (prosocial or desirable) by an
outgroup member will be attributed to an
exceptional case, luck, special advantage (seen as

unfair by the observer), situational context, or

high motivation and effort.
Ultimate attributional error is likely to
characterize the attributions of most people

toward outgroup members, but this bias is likely

to be stronger for prejudiced people.
Ultimate attributional error is likely to occur

when perceivers are conscious of their own and the

actor's group memberships.
The intensity of the ultimate attributional error

will be greatest when the groups involved have

histories of intense conflict, ethnic differences

covary with national and socioeconomic

differences, and when groups have negative

stereotypes of each other.

Taylor and Jaggi (1974) did a study in India in which

Hindu adults made attributions for desirable and undesirable

acts made by another Hindu (ingroup) or another Muslim

(outgroup).

They found that for socially desirable

beh

tha

beh

out

per

ou‘

bi

an

ir.

42
behaviors, internal attributions were greater for ingroup
than outgroup actors, and for socially undesirable
behaviors, internal attributions were lower for ingroup than
outgroup actors.

Hewstone (1990) reviewed 19 studies which examined
perceptions of causal attributions for acts by ingroup and
outgroup members, and reported that attributions made for
the behavior of ingroup and outgroup members are often
biased. For example, in a study involving Malay (majority)
and Chinese (minority) groups in Malaysia, Malays made more
internal attributions for a positive act by a Malay than a
negative act. It was also found that Malays attributed a
positive act by a Malay actor to more internal factors than
the same act by a Chinese actor. There was clear evidence
of attributional bias as a result of ethnic group
membership, presumably as a result of the stereotypes each
group had about the other. Positive acts by ingroup members
were attributed more to internal factors than positive acts
by outgroup members, while negative acts showed the opposite
pattern.

Jackson, Sullivan, and Hodge (1993) also found support
for ingroup/outgroup attributional biases. Subjects were
presented with college applications which manipulated the
ethnic orientation of the applicant and the strength of
credentials (high vs. low). Pretesting indicated that
strong credentials were considered less stereotype

consistent for blacks than whites, whereas weak credentials

wer
bla

ill}:

the

fa:

me

me

43
were considered less stereotype consistent for whites than
blacks. It was found that attributions and affect played an
important role in the evaluation of black outgroup targets
(e.g., probability of acceptance to the university), such
that strong credentials were "explained away" by external
factors.

Ingroup and outgroup effects can also occur in leader-
member relations. In a study of ingroup/outgroup leader-
member linkages and effects on attributional processes,
Heneman, Greenberger, and Anonyuo (1989) hypothesized that
attributions may be influenced by the nature of the leader-
member relationship (Green & Mitchell, 1979; Mitchell,
Green, & Wood, 1981). They had 188 supervisors assess the
exchange relationship of a subordinate with whom they had
the worst working relationship and with whom they had the
best working relationship. The supervisors generated a
critical incident for effective and poor performance for
both the ingroup and outgroup members. They were then asked
to indicate the extent to which ability, effort, luck, and
task difficulty caused each of the critical incidents.

Significant differences were found between ingroups and
outgroups for attributions to ability. Supervisors were
found to attribute effective performance to ability more for
ingroup members than for outgroup members. Supervisors also
were more likely to attribute low effort and ability as
causes of ineffective performance of outgroup members than

for ingroup members.

 

ingz
gen:
In l
Sig:
was
all?
sex

lee

0t?
at

no

44

If gender affects leader-member relations such that an
ingroup bias exists in the leader-member relationship, then
gender may result in ingroup/outgroup attributional biases.
In the Pence et al. (1982) study cited earlier, a
significant sex of subject by sex of employee interaction
was found. In general, subjects considered it more
appropriate to work with and encourage employees of the same
sex; thus, gender may affect ingroup/outgroup biases through
leader/member linkages.

Medway and Lowe (1976) found that the success of liked
others resulted in stronger internal attributions than
attributions for disliked others. Liked others were seen as
more responsible for success and less for failure.

Similarly, Fontaine (1972, cited in Frieze, 1976, p. 100)
found that outcomes shared with highly similar other people
often led to ability attributions.

The studies cited above suggest that gender is a
salient characteristic in masculine—typed jobs and people
may consider themselves according to gender group category
(Hewstone, 1990). If this is true, then it stands to reason
that the interaction of the gender of the decision maker and
the gender of the employee may affect attributional biases
through group membership. Since outgroup members tend to
receive external or effort attributions for success while
ingroup members tend to receive internal or ability
attributions and the reverse is true for failure conditions,

ingroup/outgroup biases may affect reward and promotion

deci
resi

for

bia
out
the
big
lflll
get
be

Cd

45
decisions. Nicholls (1976) found although people admire and
respect effort, they tend to prefer high ability as a trait
for themselves.

In sum, evidence exists to suggest that attributional
bias might occur as a function of gender ingroup and
outgroup biases of the decision maker. Extrapolating from
the research on ethnic groups and attributional biases, one
might expect attributional processes to vary according to
whether managers and employees are of the same or different
gender. High performance by employees of same gender would
be more likely to be attributed to internal and stable
causes than high performance by employees of different
gender. Conversely, low performance by employees of the
same gender would be more likely to be attributed to
external and unstable causes than low performance by
employees of different gender. These attributions would
then affect personnel decisions made by managers or
supervisors.

While gender stereotypes have been fairly consistently
linked with attributional biases and personnel decisions and
group membership stereotypes have been linked with
attributions, some research has found conflicting findings.
Often, conflicting findings have been due to a lack of
consideration for other important variables such as
differences in the nature of the job or task or in the
amount of information which has been provided to the

decision maker.

effl
'1an
l L
lie
Eir
whi
in
He.
ge
be

th

46

c o ' o 'o vailabil't . Stereotyping
effects may be ameliorated by the presentation of
information relevant to the job or to job performance (Deaux
& Lewis, 1984; Heilman & Guzzo, 1978; Heilman, et al., 1988:
Nieva & Gutek, 1980; Ruble, Cohen, & Ruble, 1984; Tosi &
Einbender, 1985), or by the presentation of information
which would reduce stereotype-based processing about women
in general (Heilman, 1983; Heilman & Martell, 1986).
Heilman (1983) stated that stereotypes about women in
general are more likely to be used when ambiguity in the fit
between the employee and the job exists. She further stated
that gender is unlikely to be viewed as the ultimate source
of information about an individual when more job-relevant
information is available (Renwick & Tosi, 1978), when
information is unambiguous and not subject to
reinterpretation (Pheterson, Kiesler, & Goldberg, 1971), or
when the causal basis of performance is known (Heilman &
Guzzo, 1978).

In a meta-analysis of 21 studies, Tosi and Einbender
(1985) found amount and type of information available was
related to the amount of gender-related discrimination which
occurred. According to the authors, there was a lower
likelihood of discrimination when information that was not
gender-salient was presented. On the other hand, Eagly,
Makhijani, and Klonsky (1992) in a recent meta-analysis of
61 studies found no support for the hypothesis that bias

against women weakens with increases in the amount of

ind
the
the

3V6

co:
in:
ge
lle

llE

47
individuating information available, though they state that
their conclusions should be interpreted carefully because
there was little variance in the amount of information
available across most studies.

Heilman and Guzzo (1978) conducted a study which is
commonly cited as support for the idea that causal
information regarding performance attenuates the effects of
gender on attributions and personnel decision making.
Heilman and Guzzo (1978) had 29 men and women enrolled in a
MBA program rate the appropriateness of several personnel
actions for successful employees. Different causal
attributions (ability, effort, luck, task difficulty) were
provided as explanations for the work success of four male
or female target employees. No significant effects
involving the sex of the stimulus person were found.

Results suggested that the causal attributions of luck,
effort, and task difficulty resulted in fewer and less
desirable organizational rewards when contrasted with
attributions to ability. This was true whether the
successful employee being considered was male or female.
Many researchers have interpreted these findings as
supportive of the idea that sex discrimination will not
occur when enough performance relevant or causal information
is presented, but this conclusion may be premature.

A closer inspection of the obtained cell means

indicates the means for promotion and pay raises may have

been affected by gender. With a sample size of 29, it is

not
tho
tes
stu
Hei

l'EE

an
st

in

48
not surprising that the gender effect was not significant,
though it appeared to be present, because the power of the
test was extremely low. In fact, the Pence et al. (1982)
study cited earlier was a conceptual replication of the
Heilman and Guzzo study, and they found effects of gender on
responses to poor employee performance in spite of the
causal information provided. The interpretation is that
gender-related bias may be diminished with increasing
amounts of performance-related information, but they may
still exist; gender bias may occur even when causal
information is provided.

A study by Pazy (1986) which looked at the influence of
causal information and gender on performance ratings found
that a pro-male bias existed, even when identical
information regarding performance was provided. Forty-eight
middle-level managers (mean age 39 years) from a variety of
organizations in Israel reviewed performance appraisal
material of fictitious male and female employees whose work
success was attributed to identical causes. For each
fictitious employee, information on ability, effort, and
task difficulty as causes of performance was presented.
Although causal information and performance-related
information influenced decisions, pro-male bias did not
disappear. Males were consistently ranked higher than
females, in spite of the fact that they both had identical

performance levels and causal information.

inp
int
pos
inc
ove
dis
va:
at
ma

bi

49

Pazy (1986) hypothesized that people can engage in
input assimilation such that new information is likely to be
integrated into an existing conceptual structure whenever
possible. Pazy states that information input which is
inconsistent with the conceptual structure one has must be
overwhelming for it not to be disregarded, discounted, or
distorted. Since sex-linked perceptions are related to a
variety of contextual and perceptual factors, controlling
attributional biases simply by presenting causal information
may not be enough to overcome treatment and evaluation
biases. Her findings support her hypotheses.

Martell (1991) examined the impact of attentional and
memory demands on work performance ratings of male and
female employees presented in traditionally male jobs. Two-
hundred and two college students read a vignette depicting
the work behavior of a male or female police officer and
then rated the individual's work performance. Men were
rated higher than women when attentional demands were
imposed on subjects while reading the vignette and when time
pressures were made salient. It was only when subjects were
able to carefully allocate all of their attentional
resources to the rating task that sex bias in work
performance ratings subsided. Another finding was that
subjects seemed to rely more heavily on positive stereotypes
of men under heightened attentional demand conditions, since
attentional demands inflated ratings of males, but no

significant differences were found in female ratings between

low
fin
occ

rel

muc
prt
Opt
ta:
by

Fi

50
low and high attentional demand conditions. The research
findings cited above suggest gender-related bias can still
occur in spite of unambiguous and specific performance-
related information.

It appears that there is some disagreement as to how
much impact information can have on stereotype-based
processing. Under some conditions, sex stereotypes can
operate in spite of large amounts of information regarding a
target employee because encoding of information is affected
by expectancies (Bierhoff, 1989; Bodenhausen & Wyer, 1985;
Fiske & Neuberg, 1990; Hamilton, Sherman, & Ruvolo, 1990;
Lord & Maher, 1991: Tversky & Kahneman, 1973, 1974), and
stereotypes are a form of expectancies. The effect of
stereotypes on information processing is further
strengthened by the fact that people often engage in
hypothesis confirmation in cognitive processing, especially
when decisions are complex or when processing loads are
placed on the decision maker (Bierhoff, 1989; Bodenhausen &
Lichtenstein, 1987).

The hypothesis confirmation effect occurs because
stereotypes can serve as simplifying heuristics which help
us to quickly and economically organize and retrieve
information (Bodenhausen & Wyer, 1985; Deaux, 1984; Martell,
1991: Taylor, 1981). Since stereotypes are a source of
expectancies, and the sex-type of the job or task can elicit
stereotype-based processing, it stands to reason that gender

bias can occur in spite of clear performance information if

the

pro

ass
pl:
llal
n0‘
ti‘
at

la

51
the job is strongly sex-linked, and when information
processing demands are placed upon the decision maker.

Some researchers have stated that it is reasonable to
assume that such information processing demands may often be
placed upon individuals functioning in organizations (Lord &
Maher, 1991; Feldman, 1981). For example, Feldman (1981)
notes that raters often have many competing demands on their
time. Thus, it appears that stereotypes can affect
attributional processes and personnel decisions even when
large amounts of job or performance relevant information is
provided. This is likely to occur when the decision maker
is under cognitive load, the job or task is strongly sex-
typed, and when the decision maker has strong gender
relevant stereotypes.

Limitations oflpast research. A problem with previous
research on stereotype based biases is few studies have
examined the impact of gender stereotypes on attributional
processes with only high performing or above average
employees. It is likely personnel decision makers in
organizations only consider above average performers for
promotions and pay reward; overall performance level might
have a large effect on personnel decisions, thus "washing"
out the effect of gender.

For example, in a meta-analysis on the effects of
applicant gender and qualifications on hiring
recommendations, Olian, Schwab, and Haberfeld (1988) found

that gender accounted for 4% of the variance in hiring

dec
bl

ger
cor
til
en;
c0
pr

st

52
decisions as compared with 35% of the variance accounted for
by qualifications of the applicant. Investigating whether
gender bias exists when only above average performers are
considered for promotions or reward may result in different
findings regarding stereotype-based biases than when
employees at all performance levels are considered;
competing hypotheses can be generated on the basis of
previous research. Models of the effects of gender
stereotypes on attributions and personnel decisions need to
deal with processes which occur in the above average range
of performance.

A second problem with previous work on gender
stereotype based biases is researchers have often overlooked
the controllability dimension of the attributional
framework. Often, internal locus of causation has been
interpreted in a fashion consistent with the controllability
dimension, though Weiner and others provide strong evidence
that locus of causation is not the same as controllability
(e.g., Weiner, 1986). Green and Mitchell (1979) and
Mitchell and O'Reilly (1990) have noted the importance of
leader perceptions of subordinate responsibility (i.e.
controllability) for performance, but most organizational
research has failed to integrate this important dimension
into the attributional framework.

A third limitation of past research is gender group
membership effects on attributions of performance have

rarely been directly investigated. A growing body of

res
sup
res
lbs
in‘
in
at

fc

53
research (e.g., Hewstone, 1990; Jackson et al., 1993)
suggests ingroup/outgroup stereotypes and perceptions may
result in attributional biases. While some studies have
looked at gender interactions, none of them have explicitly
investigated gender group based attributional biases. If
ingroup/outgroup perceptions affect attributions, and
attributions affect personnel decisions, then it would
follow that personnel decisions would be affected by
ingroup/outgroup distinctions. Results found by Kraiger and
Ford (1985) and Landy and Farr (1980) are consistent with
this view. In quantitative and qualitative reviews of
performance appraisal research, they found that blacks
tended to rate black ratees higher than whites, and whites
tended to rate white ratees higher than blacks. Future
research should investigate whether these findings
generalize to gender group membership perceptions.

Another problem is attributions have either been
manipulated by providing causal information and not
measured, or they have been measured but no contextual
performance information has been provided. It is
problematic to provide causal information and infer a
particular attribution was made by the decision maker
without further investigation because one cannot be sure
individuals have the same perceptions of controllability,
stability, and locus of a given cause. The formation of
attributions is the outcome of an individual perceptual

process, by definition. For example, for some decision

pale
per
lea
the
per
ob:

0C

at

54
makers, presentation of low ability as a cause of low
performance may result in low perceived effort as a cause
leading to low ability. Thus, decision makers may vary in
the perceived controllability of the cause of the
performance, depending on their interpretation of the
observed cause. Tests of most attributional models have to
occur within person, not between people.

On the other hand, doing research by measuring
attributions while providing little or no performance
relevant information may be a problem because people in
organizations are likely to have access to larger quantities
and more varied types of information about an employee than
a one or two line description of whether someone is doing
well or poorly.

A fifth problem with previous research on the impact of
gender stereotypes is many studies have failed to measure
whether or not a stereotype actually existed in the decision
maker's frame of reference. Findings by Garland and Price
(1977) and Stevens and DeNisi (1980) using the WAMS suggest
the degree to which people have negative stereotypes
regarding the fit between women and management positions
will affect attributions and personnel decisions.

Research on gender stereotypes, attributions, and
personnel decision making has also had methodological
problems. For example, different personnel decisions and
ratings have been summed and treated as a single dependent

variable without any corresponding theoretical or empirical

sup

his

th

in

55
support for such analyses (e.g. Kinicki & Griffeth 1985).
Also, attributions are often measured using single items
(e.g. Deaux & Emswiller, 1974) or by using ipsative scales
(e.g., Russell, 1982). The items themselves are often
ambiguous and allow subjects to interpret the questions in a
variety of ways.

A final problem with earlier research is that it is in
the past. While gender-related bias has been found to occur
in a wide variety of contexts, some researchers suggest that
gender-related bias may be on the decline (e.g., Kinicki &
Griffeth, 1985: Nieva & Gutek, 1980), though much work
suggests this is not true (e.g., Heilman, et al. 1989: Pazy,
1986). Kinicki and Griffeth (1985) found no significant
sex-related interaction effects on attributions and
performance ratings, and they interpreted their findings as
suggesting that sex-role stereotypes no longer affect
perceptions of female performance. The authors state that
their results are consistent with more recent studies which
have found a lack of sex-related bias on causal attributions
of performance, evaluations of performance effectiveness,
and selection decisions. They suggest this recent pattern
of results "might be interpreted optimistically as a trend
toward increasing equity in evaluation" (Nieva & Gutek,
1980, p. 274, cited in Kinicki & Griffeth). Thus,
researchers are drawing conflicting conclusions regarding
the prevalence and effects of gender stereotypes for

personnel decisions made in recent times.

56

Summary of the literature review. Women are
underrepresented in many occupations which are
stereotypically male and many of these occupations are of
high status and pay (e.g., higher level managers). While
women may be underrepresented partially as a function of
choice, research has indicated that a major reason for the
underrepresentation of females may be due to gender
stereotyping processes and the resulting influences on
causal attributions which might be made for observed
performances (e.g., Cochran, 1993; Stevens & DeNisi, 1980).

Research in attribution theory suggests that leaders
infer causes of observed performance outcomes and these
causes vary on the dimensions of locus of causality,
stability, and controllability. Research also suggests that
attributions are linked to variety of leader cognitions and
behaviors (e.g., Mitchell & O'Reilly, 1990). The more
internal and controllable the attribution, the more extreme
the leader's response and the more the response is directed
toward the subordinate; the more stable the attribution, the
more it affects the leader's expectancies for the future
performance. Thus, causal attributions are expected to
affect promotion and pay decisions.

Gender stereotypes can influence the attributions which
are made regarding performance, and these attributions can
affect personnel decisions made by leaders, managers, or
supervisors (e.g., Garland & Price, 1977; Pence et al.,

1982). For male-typed jobs, high performance by men tends

to
pet

ati

st
an
Su

be

57
to elicit internal and stable attributions while high
performance by women tends to elicit external and unstable
attributions (e.g., Deaux, 1976; Stevens & DeNisi, 1980).

Research has indicated that ingroup/outgroup gender
stereotyping processes may also contribute to attributional
and personnel decision making bias (e.g., Hewstone, 1990).
Successful performance by same-sex members is more likely to
be attributed to internal and stable causes than successful
performance by other-sex members. Studies have shown that
findings regarding stereotype bias may be generalizable
across sample types and across lab and field settings (Eagly
et al., 1992; Murphy, Herr, Lockhart, & Maguire, 1986; Olian
et al., 1988).

Some researchers suggest that performance-related
information or reductions in ambiguity will ameliorate
gender-related bias (e.g., Nieva & Gutek, 1980), but recent
research has found that gender stereotypes can still affect
personnel decision making under unambiguous performance
information conditions (e.g., Martell, 1991; Pazy, 1986).
This may occur when the job or task is highly sex-typed,
gender is made salient, the decision maker has strong
gender-related stereotypes, or when the decision maker is
under cognitive load. Research suggests that many of these
conditions are replicated in actual organizational settings.
For example, Heilman et al. (1989) found that many managers
in organizations had the perception that men in general were

more like successful managers than women in general. This

til

th.

58
finding suggests that managerial positions are sex-typed and
that many managers have stereotypes about women and their
fit with managerial positions. Also, since women are still
generally underrepresented in the managerial work force
(especially at higher levels), gender is likely to be a
salient characteristic of managerial employees and
ingroup/outgroup bias can have important effects on
perceptions of managerial performance.

Some important limitations in previous research were
also identified. First, no studies were identified that
expressly integrated the controllability dimension of
attribution theory into studies of leader perceptions,
gender stereotyping, and personnel decision making. Second,
few studies examined gender bias and attributional processes
in the context of only high performing or above average
employees. Third, gender group membership effects on
attributions of performance have rarely been directly
addressed. Fourth, the effect of attributions on different
leader responses has received inadequate treatment (e.g.,
pay versus promotion decisions). Fifth, most of the
research cited has either manipulated attributions with
performance-related information and failed to measure them,
or it has measured attributions without providing
performance-related information beyond overall performance
level. Sixth, many studies on attribution have failed to
measure the degree to which leaders or decision makers

actually have or use gender-related stereotypes: they assume

tha

male

ips

ti:
6?

re

59

that such stereotypes exist in the minds of the decision
makers. Also, most of the studies used single-item,
ipsative, or ambiguous scales to measure attributions.

Finally, research conducted in the present may produce
findings different from research conducted in the past. It
appears that social values may be changing and some
researchers suggest that gender stereotypes may be on the
wane (Kinicki & Griffeth, 1985; Nieva & Gutek, 1980). On
the other hand, the meta-analysis by Eagly et al. (1992)
found the tendency for male leaders to receive more
favorable evaluations than female leaders is larger in more
recently published studies. This finding contrasts with the
assumption that equality of opportunity for women has
increased and prejudice and stereotyping have declined.

Given the conflicting conclusions, problems with
previous methodologies, and lack of integration with current
attribution theory, it seems that an integrative conceptual
model would be helpful for future research endeavors.
Integration and Conceptual Model

The model presented in Figure 3 was developed to
integrate various research efforts and address some of the
problems noted in previous studies. First, it deals with
gender stereotypes and attributional processes within a
performance level by explaining how performance trends or
histories might affect leader perceptions and decisions,
even when performance level is constant. This will be

explained below. Second, it integrates the controllability

60

mcwxc: coamwoua Honcomuom can ncoflusnfiuuu< so muouuum

unwguouumum Hoop—mu mo Homo: Hmsumoocou m>wunuoousH .Illllmi: .H

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

35.329.
.6 36:00
3.3.2-5332.
393.5 All .0 .3230
358.2qu m
.0 >~:_na~w Dem
5:326:50 v 89:20
.283 . f ooceeuotod .0
oozootoa A0 3250 2 3:0
muoEOun— m. 3:23.525
1_
2 3595...
N :86 .E: as:
_ nose;
mwmacom P 3553an
._O wwwmthr:

 

 

 

 

3.2mm 222 1|.
2 Sconces

 

 

dip
ace
of
Th
me

CO

Sl

61

dimension of attribution theory into its framework, and it
models the effects of locus, controllability, and stability
of perceived causes of performance on personnel decisions.
Third, it relates the effect of gender and gender group
membership to perceptions of locus of causality,
controllability, stability, and personnel decisions.
Specifically, personnel decisions refer to promotion and pay
decisions.

Epomotion decisions: The willingness to promote an
employee to a higher managerial position (may
imply a commensurate pay increase).

Pay decisions: The willingness to allocate merit pay,
bonus, or a general salary increase to an
employee.

Psrformance trends. The following two limitations of
previous research were mentioned earlier: 1) leaders in
organizations typically consider only above average
performers for promotions and pay: and 2) leaders generally
have more performance-related information than has typically
been supplied in stereotype research studies. The
conceptual model deals with these two problems by modeling
how performance trends affect personnel decision making. A
decision maker may only consider above average performers

for promotions or pay raises, but he or she can still be
affected by performance history or trends. A decision maker
in an organization receives more performance-related

information than just the overall performance level of an

enp

obs

pee

st

01'

pe

62
employee; information about past performance or causes of
observed performance is likely to be available when making a
personnel decision.

As mentioned earlier, most of the work done with gender
stereotypes and causal attributions has dealt with the good
or poor credentials of a job applicant, or the good or poor
performance of a particular employee. This is problematic
because decision makers may only consider above average
employees for pay rewards or promotions. In other words,
the decision making context is such that only a specific
subset of employees are considered for a particular reward
(promotion or pay); not all employees are eligible or likely
to be considered for such rewards. In such a context only
above average employees are likely to be considered for pay
or promotion rewards.

However, even when one only considers performers within

a particular performance level, performance patterns within
that level can still vary (Hofmann, Jacobs, & Baratta,
1993). An individual employee can randomly deviate around a
mean, he/she can improve, or he/she can decline and still
exhibit the same mean performance. There is evidence to
support the idea that knowledge about prior performance or
performance trends will affect later performance evaluation
and reactions, even when the current performance level is
held constant across conditions (DeNisi & Stevens, 1981;
Gully & Rodin, 1990; Jones, Rock, Shaver, Goethals, & Ward,

1968: Scott & Hamner, 1975).

63
Esrformance Trend: Given an overall performance level,
performance can vary over time and yield the same
overall average. Performance trend indicates the way
in which the performance can vary over time.

Improving: Performance begins below the overall
performance level mean and ends above the
overall performance level mean.

a' a‘ 'n : Performance shows minor random
shifts around the overall performance level
mean.

Declining: Performance begins above the overall
performance level mean and ends below the
overall performance level mean.

Jones et al. (1968) found a primacy effect such that a
worker whose initial performance was high and declined over
time was rated as more able and more likely to do well in
the future than a worker whose performance was initially low
and improved over time, even though the overall performance
was the same for the two workers. Although overall
performance ratings were not collected in their study, the
results suggested workers who start high and decline receive
more positive ratings than workers who start low and
improve.

A study by Scott and Hamner (1975) found results which
were opposite to the Jones et al. study. They presented

subjects with employees whose performance was either stable

or variable around improving, declining, or random patterns

of

per
the
val

in}

110

de

64
of performance, but they all showed equal overall
performance levels. Findings indicated that subjects
thought the stable worker was more motivated than the
variable worker, but not as highly motivated as the
improving worker. Thus, the results of Scott and Hammer
(1975) suggest employees who start low and improve receive
more positive ratings than employees who start high and
decline.

DeNisi and Stevens (1981) conducted a study to try to
resolve the conflicting findings of Jones et al. (1968) and
Scott and Hamner (1975). They used performance profiles as
stimuli to assess decision processes associated with the use
of appraisal information. One hundred and forty-seven
students (67 female, 80 male) were given data representing
hypothetical subordinates whose performance was stable or
variable and at high, medium, or low average levels. They
also looked at two other conditions: an ascending or
descending trend in performance.

DeNisi and Stevens (1981) reported the overall average
level of performance was the most important determinant of
evaluations and other personnel decisions (e.g., salary
decisions), while stability of the performance was less
important. Improving performance received higher ratings on
personnel decisions than declining performance, and stronger
attributions to effort and lower attributions to ability
were made. The authors measured a variety of attributions

but were disappointed not to find stronger effects on

ral
at‘
se

p9

Ell

lllE

 

65

attributions. They suggested future research should provide
raters with the contextual cues they need to make meaningful
attributions. An important point the authors make is raters
seemed to equate improving performance with high
performance, and declining performance with low performance.

Gully and Rodin (1990) provided 56 subjects with
employee performance profiles which depicted improving,
maintaining, and declining performance trends for above
average employees. They also presented subjects with
information which would allow them to infer the cause of the
observed performance trends, and then measured the perceived
controllability and stability of those trends. When
compared with the other trends, it was found the maintaining
performance trend was perceived to be the most stable trend,
while the improving and maintaining performance trends were
rated as more controllable than the declining trend. These
findings occurred in spite of the fact that the average
performance level presented was identical across conditions.

Based on the above findings, it is hypothesized that
performance trends will affect personnel decision making.
Since improving or ascending performance is equated with
high performance and declining or descending performance is
equated with low performance (DeNisi & Stevens, 1981) the
following hypotheses can be made. They are represented in

link number one of the conceptual model.

66

H1: Employees who show improving performance trends
will be more likely to be promoted or to receive
monetary increases than those showing declining
trends.

H2: Employees who show maintaining performance trends
will be more likely to be promoted or to receive
monetary increases than those showing declining
trends.

Because it is not clear how improving and maintaining trends
might differ in their effects on personnel decisions,
specific hypotheses regarding this issue are not stated.

Effect of attributions and_performangs_trsngs. For
ease of comprehension, promotion and pay decisions will
simply be referred to as personnel decisions until specific
hypotheses regarding each of them are made. Based on
previous research, the main effect of performance trend on
personnel decisions will be qualified by the attribution and
performance trend interaction. This is represented by link
number two in the conceptual model.

For example, although good performance is rewarded more
than poor performance in general, researchers have
consistently found that good-internally caused performance
is rewarded more than good-externally caused performance,
while poor-internally caused performance is rewarded less
than poor-externally caused performance (e.g., Green &
Mitchell, 1979; Knowlton & Mitchell, 1980). This is because

an employee is more likely to be considered responsible for

67
the observed performance if the cause is internal (Mitchell
& O'Reilly, 1990).

If an employee is responsible for improving or
maintaining performance, then he or she will be considered
to be more deserving of positive personnel decisions. If an
employee is responsible for declining performance, then he
or she will be considered to be less deserving of positive
personnel decisions. Since the maintaining trend is likely
to be considered stable, by definition, and the improving
and declining trends are considered to be unstable, by
definition, any variance within a trend in their effects on
personnel decisions can be attributed to the effects of
perceived controllability and locus of causality of the
observed performances. This suggests that trend will
interact with perceived controllability and locus of
causation in their effects of personnel decisions.
Hypothesis three is represented by link two in the
conceptual model and it is depicted graphically in Figure 4.

H3: The effect of performance trends on personnel
decisions will be moderated by the decision
maker's perceptions of locus of causality.

H3a: Those employees who show improving or maintaining
performance trends which are perceived by the
decision maker as due to causes which are internal
will receive more positive personnel decisions
than employees who show improving or maintaining

performance trends which are due to causes

68

Personnel Decisions

 

 

 

 

 

 

 

 

 

 

 

 

71— -_ -

l l l

l V l
Improving Maintaining Declining

Performance Trend
*lnternal —+—External

Eigpfs_fi. Hypothesized Effects of Locus of Causality and

Performance Trends on Personnel Decisions

69
perceived to be external, when mean performance
among employees is held constant.

H3b: Those employees who show declining performance
trends which are perceived by the decision maker
as due to causes which are internal will be isss
likely to receive positive personnel decisions
than employees who show declining performance
trends which are due to causes perceived to be
external, when mean performance among employees is
held constant.

Hypotheses made with regard to the dimension of
controllability will closely parallel the hypotheses made
with regard to the locus of causality dimension because the
controllability dimension is a subdimension of the locus of
causality dimension. Work by Mitchell and O'Reilly (1990)
and others has consistently noted the parallel between locus
of control and controllability. As mentioned above, if a
decision maker perceives an employee to be responsible for
improving performance, then he or she is more likely to
reward that behavior than if he or she were not perceived to
be responsible for the improving performance. Conversely,
if a decision maker perceives an employee to be responsible
for declining performance, then he or she will be less
likely to reward that behavior than if he or she were not
perceived to be responsible for the declining performance.
These hypotheses are represented by link number two in the

conceptual model and they are depicted in Figure 5.

7O

 

Personnel Decisions

‘ -

 

/

 

 

 

 

 

 

 

L L J_
1 l 1

Improving Maintaining Declining

Performance Trend

 

“*- Controllable —l— Uncontrollable

Eigp:s_§. Hypothesized Effects of Controllability and

Performance Trends on Personnel Decisions

71

H4: The effect of performance trends on personnel
decisions will be moderated by the decision
maker's perceptions of controllability.

H4a: Those employees who show improving or maintaining
performance trends which are perceived by the
decision maker as due to causes which are
controllable will receive more positive personnel
decisions than employees who show improving or
maintaining performance trends which are due to
causes perceived to be uncontrollable, when mean
performance among employees is held constant.

H4b: Those employees who show declining performance
trends which are perceived by the decision maker
as due to causes which are controllable will be
isss likely to receive positive personnel
decisions than employees who show declining
performance trends which are due to causes
perceived to be uncontrollable, when mean
performance among employees is held constant.

No interactions between perceptions of stability and
performance trend are expected because perceptions of
stability of causation cannot be separated from performance
trends. By definition, maintaining trends are more stable
and less variable than improving or declining trends.
Therefore, stability perceptions are actually perceptions of
performance trend. Improving and declining performance will

be considered less stable than maintaining trends and

pe

wi

72
personnel decisions which are strongly affected by stability
will also be strongly affected by performance trend.

Gully and Rodin (1990) found that maintaining trends
were considered more stable than improving or declining
trends. Since maintaining performance trends show less
variability than improving or declining performance trends,
by definition, it is reasonable to expect perceptions of
causes of maintaining trends to be more stable.

H5: Causes of maintaining performance trends for
employees will be perceived by the decision maker
to be more stable than improving or declining
performance trends, when mean performance among
employees is held constant.

This is represented by link three in the conceptual
model. It is possible that performance trends will strongly
affect promotion decisions because of differences in
perceived stability hypothesized above. Folkes and Marcoux
(1984) found stability was related to the likelihood to date
a person of the other sex. This was explained because
stability could affect future expectancies, and the decision
to date someone depends in some sense on expected future
outcomes. If one considers promotions to be related to
expectancies for future performance, then one might expect
promotion decisions to be related to performance trend and
attributions of stability. This hypothesis is presented in

Figure 6.

73

Promotions

 

 

'////”

 

.__..l_—__»7 v -...r-...~-7 .

 

 

 

 

 

 

 

l l l
r T 7—

lmproving Maintaining Declining

Performance Trend

Eignrs_§. Hypothesized Effects of Performance Trends on

Promotion Decisions

74
H6: Employees who show maintaining performance trends
will be more likely to receive positive promotion
decisions than employees who show improving or
declining performance trends, when perceived
controllability or locus of causality is held
constant.

In sum, performance trends are expected to affect
personnel decisions in a manner consistent with performance
level. Improving performance is expected to be perceived in
a fashion similar to high performance levels, while
declining performance is expected to be perceived in a
fashion similar to low performance levels, while hypotheses
regarding maintaining performance are more equivocal. Also,
performance trend is expected to interact with perceptions
of controllability and locus such that the more internal or
controllable the improving performance trend, the more
positive the personnel decisions. The more internal or
controllable the declining performance trend, the less
positive the personnel decisions. No interactions between
stability and performance trend are expected because
performance trend is, by definition, the representation of
stability.

Attributions of informational cues. One of the
limitations of previous research was most decision makers
have access to more information with regard to observed
performance outcomes of an employee and the causes of that

performance than is typically provided in research designs.

llu

th

01'.

75

Much of the research conducted on stereotypes, attribution
theory and personnel decision making has simply provided a
one or two line description of a good or poor employee and
then asked subjects to infer the cause of that employee's
performance with no additional information. This is not
likely to occur in the actual decision making context. The
conceptual model being presented takes these factors into
account. Martell (1991) and Pazy (1986) found attributions
can affect performance ratings and personnel decisions even
under high information circumstances, given that some degree
of cognitive work load is place upon the decision maker.

Attribution theory states people will spontaneously
generate or infer causes of observed events and these causes
can vary along the dimensions of internal/external locus of
causality, controllability, and stability. Green and
Mitchell (1979) state leaders attempt to diagnose the cause
of a performance outcome in the first stage of their model.
In order to do this, leaders have to act as information
processors and they must integrate informational or
contextual cues into their cognitive framework.
Informational cues are forms of information which would
allow a decision maker to more effectively infer the cause
of an employee's observed performance. If informational
cues allow a decision maker to infer the cause of an
observed performance, then they should also influence
attributional processes and the attributions should vary

along the three dimensions identified earlier.

76
H7: Informational cues regarding causes of observed
performance trends will result in variability of
perceptions along the controllability and locus of
control dimensions of attributions.

As noted earlier, stability perceptions are
inextricably linked to performance trends. For example, one
would have a difficult time finding an unstable cause of a
maintaining performance trend. Therefore, informational
cues are not expected to affect stability perceptions
separate from the effect of performance trend; informational
cues regarding stability are tied to performance trends
while perceptions of locus of causation and controllability
are not. Reconceptualizing locus and controllability
dimensions as part of the same dimension yields three
factors (internal-controllable, internal-uncontrollable, and
external-uncontrollable) which can then be crossed with
stability of performance.

Since the current work is exploratory and relatively
complex, hypotheses will only be stated about a subset of
the 2 X 3 taxonomy depicted earlier (see Figure 2). The
specific subset of causes includes effort, ability, and task
difficulty attributions, as depicted in the stable column of
the 2 X 3 taxonomy. Ability refers to cognitive traits or
skills which allow a person to accomplish a particular job
or task. Effort is the exertion of energy or work level to
accomplish a job or a task. Task difficulty refers to how

hard or easy a particular task or job is for people in

99'
iii
in
ex
st

ea

77

general. For the purposes of this research, task difficulty

will be termed "ease of job". Using the taxonomy presented

in Figure 2, the following hypotheses can be made regarding
expected variations along the dimensions of controllability,

stability, and locus of causation for effort, ability, and

ease of job cues.

H7a: Informational cues of effort will be perceived to

be more controllable than will ability or ease of

job cues.

H7b: Informational cues of ability will be perceived
to be less controllable than effort cues, but more
controllable than ease of job cues.

H7c: Informational cues of ability and effort will be
perceived to be more internal than ease of job
cues.

H7d:

Informational cues of effort, ability, and ease
of job will be perceived to be more stable when
associated with a maintaining trend than when
associated with an improving or declining trend.
Hypotheses H7a through H7d are derived from crossing the
first column of the 2 X 3 taxonomy presented in Figure 2
with the three types of performance trend (improving,
maintaining, declining).

Employee gender effectsnon attributional perceptions.
The gender of the managerial employee is expected to affect
perceptions and attributions of informational cues made by

the personnel decision maker through stereotype based

pre
th
re

19

78
processing. Gender stereotypes lead to expectancies and
these expectancies can bias processing of information with
regard to causes of observed performance (e.g., Heilman,
1983; Heilman & Stopek, 1985a; Martell, 1991).

H8: Performance trends will interact with gender of
the managerial employee to affect attributions of
stability and locus of causality made by the
decision maker.

Gender stereotypes result in high expectations of good
male performance for management jobs and low expectations of
good female performance for the same jobs (e.g. Deaux, 1976;
Heilman, 1983; Heilman et al., 1989). Improving performance
trends by male managerial employees will be considered by
the decision maker to be more internal and stable than will
improving performance trends by female managerial employees.
Conversely, declining performance trends by male managerial
employees will be considered by the decision maker to be
more unstable and external than will declining performance
trends by female managerial employees. Hypotheses regarding
the interaction effect of employee gender and performance
trend on perceived stability are presented below and in
Figure 7.

H8a: Improving or maintaining performance trends for
male managerial employees will be perceived by the
decision maker to be more stable than will the
same performance trends for female managerial

employees.

79

Perceived Stability

 

 

 

 

 

 

 

 

 

 

l L— l
T j E
Improving Maintaining Declining

Performance Trend
"" Male + Female

Eiguzs_1. Hypothesized Effect of Employee Gender and

Performance Trend on Perceived Stability

 

 

80

H8b: Declining performance trends for male managerial
employees will be perceived by the decision maker
to be more unstable than will the same performance
trends for female managerial employees.

Hypotheses regarding the interaction effect of employee
gender and performance trend on perceived locus of causation
are presented below and in Figure 8.

H8c: Improving or maintaining performance trends for
male managerial employees will be perceived by the
decision maker to be more internal than will the
same performance trends for female managerial
employees.

H8d: Declining performance trends for male managerial
employees will be perceived by the decision maker
to be more external than will the same performance
trends for female managerial employees.

Because controllability is a sub-dimension of locus of
causality, predictions similar to those for the interactive
effect of employee gender and performance trend on
perceptions of locus of causality are made for perceived
controllability. These hypotheses are stated below and
graphically represented in Figure 9.

H9: Performance trends will interact with the gender
of the managerial employee to affect attributions
of controllability made by the decision maker.

H9a: Improving or maintaining performance trends for

male managerial employees will be perceived by the

 

 

81

 

Perceived External Locus

 

 

 

 

 

 

 

 

 

 

 

 

l L g
F N j
Improving Maintaining Declining

Performance Trend

A” Male —l— Female

Eignrs_§. Hypothesized Effect of Employee Gender and

Performance Trend on Perceived Locus of Causality

 

82

Perceived Controllability

 

 

 

 

 

 

 

 

 

l _L l

 

 

 

l l l

Improving Maintaining Declining

Performance Trend

—*—' Male —l— Female

Eign:s_2. Hypothesized Effect of Employee Gender and

Performance Trend on Perceived Controllability

 

 

 

 

pe

C0

83
decision maker to be more controllable than will
the same performance trends for female managerial
employees.

H9b: Declining performance trends for male managerial
employees will be perceived by the decision maker
to be more uncontrollable than will the same
performance trends for female managerial
employees.

The interaction effects of employee gender and
performance trend on perceived locus of causality,
controllability, and stability are expected to be qualified
by the effects of specific informational cues. Gender
stereotypes usually result in perceptions that good male
performance is due to high ability, while good female
performance is due to high effort or ease of job. As stated
earlier, many researchers have found that people engage in
hypothesis confirmation processing when evaluating
information relevant to a stereotype.

H10: Informational cues will interact with gender of
the employee and performance trend to affect
attributions of controllability and locus of
causality made by the decision maker.

Since ability is an uncontrollable-internal trait, high
ability cues which are associated with improving or
maintaining performance trends for male managerial employees
Will be perceived by the decision maker to be more internal

and uncontrollable than will high ability informational cues

ass
for
fee

ab:

iii
an
as

ma

84
associated with improving or maintaining performance trends
for female managerial employees. On the other hand, since
female managerial employees are expected to be low in
ability, low ability cues which are associated with
declining performance trends for female managerial employees
will be perceived by the decision maker to be more internal
and uncontrollable than will low ability informational cues
associated with declining performance trends for male
managerial employees. These hypotheses are represented
below:

H10a: Informational cues which indicate high ability
attributions for improving or maintaining
performance trends for male managerial employees
will be perceived by the decision maker to be more
internal and uncontrollable than will improving or
maintaining performance trends and high ability
informational cues for female managerial
employees.

H10b: Informational cues which indicate low ability
attributions for declining performance trends for
female managerial employees will be perceived by
the decision maker to be more internal and
uncontrollable than will declining performance
trends and low ability informational cues for male
managerial employees.

Since effort is a controllable and internal trait, high

effort cues associated with improving or maintaining

 

 

85
performance trends for female managerial employees will be
perceived by the decision maker to be more internal and
controllable than will high effort informational cues
associated with improving or maintaining performance trends
for male managerial employees. Conversely, since male
managerial employees are expected to exert less effort (due
to their high ability), low effort cues associated with
declining performance trends for male managerial employees
will be perceived by the decision maker to be more internal
and controllable than will low effort informational cues
associated with declining performance trends for female
managerial employees. These hypotheses are represented
below:

H100: Informational cues which indicate high effort
attributions for improving or maintaining
performance trends for female managerial employees
will be perceived by the decision maker to be more
internal and controllable than will improving or
maintaining performance trends and high effort
informational cues for male managerial employees.

H10d: Informational cues which indicate low effort
attributions for declining performance trends for
male managerial employees will be perceived by the
decision maker to be more internal and
controllable than will declining performance

trends and low effort informational cues for

female managerial employees.

 

 

CE

86
Since female managerial employee performances (good or
poor) are more likely to be attributed to ease of the job
(or to its difficulty) than will the same performances by
male managerial employees, and since ease of job
attributions are uncontrollable and external in locus of
causality, the following hypothesis can be stated:

H10e: Informational cues which indicate high or low
ease of job attributions for improving,
maintaining, or declining performance trends for
female managerial employees will be perceived by
the decision maker to be more external and
uncontrollable than will the same performance
trends and informational cues for male managerial

employees.

Employee gender effects on personnel decisions. Since
employee gender is expected to affect causal attributions of
performance, and since causal attributions of performance
are expected to affect personnel decisions, employee gender
is expected to interact with performance trends and
informational cues to affect personnel decision making.

As noted in Hypotheses 3-3b and 4-4b, controllable or
internal maintaining and improving performance trends are
expected to be rewarded more than uncontrollable or external
maintaining and improving performance trends. In contrast,

controllable or internal declining performance trends are

expected to be rewarded less than uncontrollable or external

declining performance trends.

 

 

87

As stated earlier, improving and maintaining
performance trends for male managerial employees are
hypothesized to be perceived by the decision maker to be
more internal, controllable, and stable than the same
performance trends for female managerial employees.
Conversely, declining performance trends by female
managerial employees are hypothesized to be perceived by the
decision maker to be more internal, controllable, and stable
than the same performance trends for male managerial
employees.

If both sets of hypotheses stated above hold true, then
employee gender should interact with observed performance
trends to affect perceived attributions and personnel
decisions. Since improving and maintaining performance
trends are hypothesized to be perceived by the decision
maker to be more internal and controllable for male
managerial employees than for female managerial employees,
male employees should receive more reward than female
employees for improving and maintaining performance trends.
Conversely, since declining performance trends are
hypothesized to be perceived by the decision maker to be
more internal and controllable for female managerial
employees than for male managerial employees, female
employees should receive less reward than male employees for
declining performance trends. These hypotheses are

presented below and depicted graphically in Figure 10.

 

88

Personnel Decisions

 

///s’\

 

/

 

 

 

 

 

 

 

 

 

L l L
I j l

 

Improving Maintaining Declining

Performance Trend

“'— Male —i— Female

Eignfs_1_. Hypothesized Effect of Employee Gender and

Performance Trend on Personnel Decisions

89

H11: Performance trends will interact with gender of
the managerial employee to affect personnel
decisions made by the decision maker.

H11a: Improving or maintaining performance trends for
male managerial employees will be rewarded more by
the decision maker than will the same performance
trends for female managerial employees.

Hllb: Declining performance trends for female
managerial employees will be rewarded less by the
decision maker than will the same performance
trends for male managerial employees.

The interaction between performance trend and gender of
the managerial employee is expected to be qualified by the
interaction effect between informational cues, performance
trend and employee gender. This interaction effect will
result in variability in perceptions of stability, locus of

causation, and controllability, depending on the employee
gender, performance trend, and the specific informational
cues provided. These attributional perceptions, in turn,
are hypothesized to affect pay and promotion decisions.

Specific hypotheses regarding the effects of interactions

involving each of the informational cues on personnel

decisions are not presented since conflicting predictions
can be made depending on whether the primary determinant of
personnel decisions are perceptions of locus of causation,

controllability, or stabilitY-

9O

gendef group membership effects on attributional

psfceptions. It is expected that improving performance of

managerial employees who are of the same gender as the
decision maker will be attributed more to internal and
stable factors than improving performance of managerial

employees who are of the opposite gender (Hewstone,

1990;
Pettigrew, 1979). However, hypotheses regarding declining

performance for people who are from the same gender group
are more equivocal.

If the person being evaluated is from
an ingroup and doing poorly then one might "penalize" or

downgrade that person in order to maintain one's self-image

On the other hand, if stereotypes bias information encoding,

one may still perceive the ingroup performer to be more

competent, relatively speaking, than the outgroup performer
(Jackson et al.,

1993; Hewstone, 1990).

Based on the "hypothesis confirmation" theory

(Bodenhausen & Wyer, 1985), it is hypothesized that gender
group membership will affect attributional processes (e.g.,
Hewstone, 1990) such that successful performance by someone
of the same gender will be perceived to be due to internal

and stable factors while successful performance by someone

of the opposite gender will be perceived to be due to

external and unstable factors.

The converse should be true
for declining performance.

These hypotheses are depicted
graphically in Figures 11 and 12.

H12: Performance trends will interact with gender

ingroup/outgroup membership of the managerial

91

Perceived Stability

 

 

 

 

 

 

 

 

 

 

s a e
Improving Maintaining Declining

Performance Trend

—"" Same Gender —+— Other Gender

Eignrs_11. Hypothesized Effect of Gender Group Membership

and Performance Trend on Perceived Stability

92

Perceived External Locus

 

 

 

 

 

 

 

 

 

 

 

: a +
Improving Maintaining Declining

Performance Trend

"" Same Gender + Other Gender

Eiguzs_1z. Hypothesized Effect of Gender Group Membership

and Performance Trend on Perceived Locus of Causality

 

 

93
employee to affect attributions of stability and
locus of causality made by the decision maker.

H12a: Improving or maintaining performance trends for
an employee of the same gender as the decision
maker will be perceived to be more stable than
will the same performance trends for an employee
of the opposite gender.

H12b: Declining performance trends for employees of
the same gender as the decision maker will be
perceived to be more unstable than will the same
performance trends for employees of the opposite
gender.

H12c: Improving or maintaining performance trends for
employees of the same gender as the decision maker
will be perceived to be more internal than will
the same performance trends for employees of the
opposite gender.

H12d: Declining performance trends for employees of
the same gender as the decision maker will be
perceived to be more external than will the same
performance trends for employees of the opposite
gender.

Because controllability is a sub-dimension of locus of

causality, predictions similar to those for the interactive
effect of employee gender ingroup/outgroup membership and

performance trend on perceptions of locus of causality are

 

 

ma

st

94
made for perceived controllability. These hypotheses are
stated below and graphically represented in Figure 13.

H13: Performance trends will interact with gender
ingroup/outgroup membership of the managerial
employee to affect attributions of controllability
made by the decision maker.

H13a: Improving or maintaining performance trends for
employees of the same gender as the decision maker
will be perceived to be more controllable than
will the same performance trends for employees of
the opposite gender.

H13b: Declining performance trends for employees of
the same gender as the decision maker will be
perceived to be more uncontrollable than will the
same performance trends for employees of the
opposite gender.

The interactions of employee gender ingroup/outgroup
membership and performance trend on perceived locus of
causality, controllability, and stability are expected to be
qualified by the effects of specific informational cues.
Ingroup/outgroup stereotypes result in perceptions that good
ingroup performance is due to high ability, while good
outgroup performance is due to high effort or ease of job.
As stated earlier, many researchers have found that people

engage in hypothesis confirmation processing when evaluating

information relevant to a stereotype.

 

95

Perceived Controllability

 

 

 

 

 

 

 

 

 

 

l l l

 

 

 

7 T l

Improving Maintaining Declining

Performance Trend

**— Same Gender + Other Gender

E1gurs_i;. Hypothesized Effect of Gender Group Membership

and Performance Trend on Perceived Controllability

 

96

H14: Informational cues will interact with performance
trend and gender ingroup/outgroup membership of
the employee to affect attributions of
controllability and locus of causality made by the
decision maker.

Since ability is an uncontrollable-internal trait, high
ability cues which are associated with improving or
maintaining performance trends for gender ingroup managerial
employees will be perceived by the decision maker to be more
internal and uncontrollable than will high ability
informational cues associated with improving or maintaining
performance trends for gender outgroup managerial employees.
On the other hand, since outgroup managerial employees are
expected to be low in ability, low ability cues associated
with declining performance trends for gender outgroup
managerial employees will be more likely to be perceived by
the decision maker to be due to internal and uncontrollable
causes than will low ability informational cues associated
with declining performance trends for gender ingroup
managerial employees. These hypotheses are represented
below:

Hl4a: Informational cues which indicate high ability
attributions for improving or maintaining
performance trends for gender ingroup managerial
employees will be perceived by the decision maker
to be more internal and uncontrollable than will

improving or maintaining performance trends and

 

97
high ability informational cues for gender
outgroup managerial employees.

H14b: Informational cues which indicate low ability
attributions for declining performance trends for
gender outgroup managerial employees will be
perceived by the decision maker to be more
internal and uncontrollable than will declining
performance trends and low ability informational
cues for gender ingroup managerial employees.

Since effort is a controllable and internal trait, and

since effort is attributed to the successful performance of
outgroup employees, high effort cues associated with
improving or maintaining performance trends for gender
outgroup managerial employees will be perceived by the
decision maker to be more internal and controllable than
will high effort informational cues associated with
improving or maintaining performance trends for ingroup
managerial employees. Conversely, since ingroup managerial
employees are expected to exert less effort (due to their
high ability), low effort cues which are associated with
declining performance trends for gender ingroup managerial
employees will be more likely to be perceived by the
decision maker to be due to internal and controllable causes
than will low effort informational cues associated with

declining performance trends for gender outgroup managerial

employees. These hypotheses are represented below:

 

 

98

H14c: Informational cues which indicate high effort
attributions for improving or maintaining
performance trends for gender outgroup managerial
employees will be perceived by the decision maker
to be more internal and controllable than will
improving or maintaining performance trends and
high effort informational cues for gender ingroup
managerial employees.

H14d: Informational cues which indicate low effort
attributions for declining performance trends for
gender ingroup managerial employees will be
perceived by the decision maker to be more
internal and controllable than will declining
performance trends and low effort informational
cues for gender outgroup managerial employees.

Since gender outgroup managerial employee performances

(good or poor) are more likely to be attributed to ease of
the job (or to its difficulty) than will the same
performances by gender ingroup managerial employees, and
since ease of job attributions are uncontrollable and
external in locus of causality, the following hypothesis can
be stated:

H14e: Informational cues which indicate high or low
ease of job attributions for improving,
maintaining, or declining performance trends for
gender outgroup managerial employees will be

perceived by the decision maker to be more

 

 

99
external and uncontrollable than will the same

performance trends and informational cues for

gender ingroup managerial employees.

Gsndsp gfoup membership effects on personnel decisigns.

Since employee gender ingroup/outgroup membership is

 

expected to affect causal attributions of performance, and
since causal attributions of performance are expected to
affect personnel decisions, employee gender group membership
is expected to interact with performance trends and
informational cues to affect personnel decision making.

As noted earlier, controllable or internal maintaining
and improving performance trends are expected to be rewarded

more than uncontrollable or external maintaining and

improving performance trends. In contrast, controllable or

internal declining performance trends are expected to be
rewarded less than uncontrollable or external declining
performance trends.

Additionally, improving and maintaining performance
trends for gender ingroup managerial employees are
hypothesized to be perceived by the decision maker to be
more internal, controllable, and stable than the same
performance trends for gender outgroup managerial employees.
Conversely, declining performance trends by gender outgroup
managerial employees are hypothesized to be perceived by the
decision maker to be more internal, controllable, and stable

than the same performance trends for gender ingroup

managerial employees.

 

 

 

Ell

ol

100

If both sets of hypotheses stated above hold true, then
employee gender group membership should interact with
observed performance trends to affect perceived attributions
and personnel decisions. Since improving and maintaining
performance trends are hypothesized to be perceived by the
decision maker to be more internal and controllable for
gender ingroup managerial employees than for gender outgroup
managerial employees, ingroup employees should receive more

reward than outgroup employees for improving and maintaining

performance trends. Conversely, since declining performance

trends are hypothesized to be perceived by the decision
maker to be more internal and controllable for gender
outgroup managerial employees than for gender ingroup
managerial employees, outgroup employees should receive less
reward than ingroup employees for declining performance
trends. These hypotheses are presented below and they are
depicted graphically in Figure 14.

H15: Performance trends will interact with gender
group membership of the managerial employee to
affect personnel decisions made by the decision
maker.

H15a: Improving or maintaining performance trends for
gender ingroup managerial employees will be
rewarded more by the decision maker than will the

same performance trends for gender outgroup

managerial employees.

 

 

101

Personnel Decisions

 

 

 

 

 

 

 

 

 

 

 

L

P

l

 

j

Improving

fl

Maintaining

fi

Declining

 

Performance Trend

"“— Same Gender "l— Other Gender

Em.

and Performance Trend on Personnel Decisions

Hypothesized Effect of Gender Group Membership

 

102
H15b: Declining performance trends for gender outgroup
managerial employees will be rewarded less by the
decision maker than will the same performance
trends for gender ingroup managerial employees.
The interaction between performance trend and gender
group membership of the managerial employee is expected to

be qualified by the interaction effect between informational

cues, performance trend and employee group membership. This
interaction effect will result in variability in perceptions
of stability, locus of causation, and controllability,
depending on the employee gender group membership,
performance trend, and the specific informational cues
provided. These attributional perceptions, in turn, are
hypothesized to affect pay and promotion decisions.

Specific hypotheses regarding the effects of interactions
involving each of the informational cues on personnel
decisions are not presented since conflicting predictions

can be made depending on whether the primary determinant of

personnel decisions are perceptions of locus of causation,
controllability, or stability.

Cpnclusion

If a decision maker does not have stereotyped attitudes
toward women as managers, then stereotype based biases are

not likely to occur because gender is not a salient

characteristic to the decision maker. Therefore, all of the
hypotheses regarding gender and gender group membership are

only expected to occur for people who have negative

 

 

103

stereotypes toward women managers. The hypotheses which

were presented were derived from the literature review and

conclusions which are briefly reviewed below. A complete

listing of the hypotheses is in Appendix A.
Leaders are expected to engage in attributional search

whenever making personnel decisions. Attributions can vary

on the dimensions of controllability, locus of causation,

and stability and they are expected to interact with

observed performance trends to affect personnel decisions.

Improving performance which is due to internal or

controllable causes is expected to be more highly rewarded

than declining performance which is due to internal or

controllable causes. Perceptions of stability are expected

to be directly linked to performance trends and they are

expected to affect promotion decisions in a manner which is

different from pay or salary decisions. These effects are

expected to occur in spite of constant overall performance

levels.
Gender and gender group membership is expected to

affect the perceived locus of causation, stability, and

controllability of observed performance trends. Improving

or maintaining trends for male or same gender employees are

expected to be perceived by the decision maker to be more

internal, controllable, and stable than improving or

maintaining trends for female or opposite gender employees.

These effects will be qualified by interactions involving

specific informational cues. Furthermore, the attributions

104
which decision makers make as a function of gender and

gender group membership are hypothesized to affect

subsequent decision making, such that male or same gender

employees will tend to receive higher reward than female or

opposite gender employees.

The conceptual model presented thus explains how gender

and group membership stereotypes might affect personnel
decisions in spite of the presence of specific information

regarding performance causation and even when all performers

are considered at the same performance level. It also

provides an integration and extension of previous work on

the effects of stereotypes on attributional processes.
Since males are dominant in upper level management

positions, the model also helps to explain why women have a

difficult time moving into higher levels of management.

Males are frequently in high status positions and are

responsible for making many personnel decisions, and since

personnel decision making is hypothesized to be influenced

by the gender of the decision maker and evaluatee, it is

easy to see how these processes can operate to maintain a

"glass ceiling" effect for women. The purpose of the

current research effort is to test the conceptual model

presented above.

METHOD

an s

Two hundred and eighty-nine participants were recruited
from the psychology department participant pool at Michigan
State University for involvement in this research project.
Each participant spent approximately 1 and 1/2 hours reading
fictional employee files and providing responses to various
questions (these will be described in a later section). All
participants were treated in accord with the ethical
standards of the American Psychological Association (APA,
1992).

Of the 289 participants, 256 provided responses which
were usable; the other 33 participants did not follow
instructions properly, which rendered their responses
unsuitable. Of the 256 participants who provided usable
responses, 115 (45%) were male and 141 (55%) were female.
Participants ranged in ages from 18 to 29 years old.

The choice of sample size was based on a power analysis
of the effects of gender on personnel decisions. The power
analysis is presented in Appendix B. This sample of
participants was selected for ease of accessibility, large

sample size, and because testing of the conceptual model

105

106
presented earlier needed to occur within a laboratory
setting.
Design

The study was conducted in a lab setting for several
reasons. First, the purpose of the study was simply to
discover if gender related biases could affect evaluations
and responses to variations in performance trends and
performance-related information. The model presented was a
new and untested model, so it was reasonable to conduct this
test in a lab setting prior to attempting its validation in
a field setting (Ilgen, 1986).

Second, the nature of the hypotheses were such that
perfect similarity in performance level, performance trend
variability, and informational cues had to occur between
different employee performance profiles. This was not
likely to occur in a field context. Generalizability issues
regarding findings involving this sample and research design
will be reviewed in the Discussion section.

The completely crossed factorial design consisted of
three conditions of informational cues (effort, ability,
ease of job), three conditions of performance trend
(improving, maintaining, declining), two conditions of
gender of target employee (male or female), and two
conditions of gender of decision maker (male or female).
Thirty-six cells (3 X 3 X 2 X 2) were included in the

complete design.

107

Performance trend (improving, maintaining, declining),
gender of target employee, and gender of decision maker were
between-subjects variables, whereas type of informational
cue was a within-subjects variable. Scores on attitudes
toward women as managers (WAMS) were collected as a separate
independent variable, since they were expected to influence
the effect of target employee gender on attributions and
personnel decisions.

There were three sets of dependent variables measured
as part of this study. The first set of dependent variables
measured included some of the manipulation checks used in
the study. Specifically, participants provided responses to
whether they perceived the target employee as exhibiting low
or high effort and low or high ability. They also provided
responses to whether they perceived the job in question to
be low or high in ease of accomplishment.

The second set of dependent variables measured included
perceived locus of causality, controllability, and stability
of the causes of the observed performance trends. These
were the attributions hypothesized to vary in response to
informational cues, and they were expected to affect
personnel decision making.

The third and final set of dependent variables measured
included promotion, salary, and merit pay bonus decisions.
This third set of dependent variables also included
likability and expected future performance of the target

employee. Research has indicated that these two variables

ple
hh
ea
re

di

ll

d.

108
play a role in making personnel decisions, though specific
hypotheses regarding these last variables were not presented
earlier. The specific analyses used to test hypotheses
regarding attributions and personnel decisions will be
discussed later.
u 'es

Two separate pilot studies were conducted to ensure:
I) manipulations using fictional employee files would be
effective; 2) participant attitudes, perceptions and
decisions could be reliably measured; 3) the instructions
for the study were clear and unbiased; and 4) demand effects
regarding gender were minimized. The exact nature of the
manipulations and questionnaires used in the complete study
which were derived from the results of the pilot studies
will be described later.

Eifst pilot study. The first of the two pilot studies
investigated whether research instructions were clear and
unbiased and whether fictional employee files created for
use in the complete study would effectively manipulate the
perceived causes of managerial employee performance. The
perceived causes of observed performance (ability, specific
and typical effort, mood, ease of job, and other person
causes) were manipulated in combination with the different
performance trends and target employee gender. The
selection of manipulated causes was based on the six-cell

matrix presented in Figure 2.

109

Factor analyses and reliability analyses were conducted
on the responses given by the 250 participants in the pilot
study to questions about perceived causes of performance,
attributions, and personnel decisions. Results of factor
analyses and reliability analyses of the pilot study are
presented in Appendix C. A repeated measures MANOVA
indicated that the fictional employee files did manipulate
perceived causes of performance, but factor analyses
indicated that participant perceptions involving certain
causes of performance tended to covary. For example,
perceptions of typical or long-term effort and ability
formed one factor, perceptions of specific or short-term
effort and mood formed another factor, and perceptions of
ease of job, other person, and luck formed a third factor.

These results suggested that causes within each of the
cells in the six-cell matrix presented in Figure 2 could not
be reliably measured separate from perceptions involving
other cells, and that people look at patterns of causes when
making attributional inferences, not single causes. The
results also suggested that these patterns of causes tended
to fall into three main areas: consistent or trait-like
qualities (ability or consistent effort), short-term or
motivated qualities (mood, attitude, or specific effort),
and external causes (others, luck, ease of job). These
findings are consistent with previous work which has
identified ability, effort, and task difficulty as the three

primary attributions which affect behavior. Thus, the

110
results of the first pilot study suggested that several
areas of the complete study needed further improvement: 1)
the manipulations in the personnel files needed to be
rewritten to suggest patterns of causation; 2) these
patterns had to correspond to stable or trait-like causes,
temporary or motivational causes, and external causes of
performance; and 3) items in the scales needed to be
rewritten to maximally discriminate between locus of
causation and perceived controllability.

In response to these findings, the fictional employee
files were revised to include only those causes of
performance which tended to create separate factors in the
first pilot study. The three causes of performance
manipulated in the new set of fictional employee files
included ability, effort, and ease of job informational
cues. Also, since perceptions regarding the different
causes of performance could influence perceptions about
other causes, an effort was made to provide informational
cues which would eliminate other causes as reasons for the
observed performance trends. Specific characteristics of
these manipulations will be discussed in the following
section on manipulations.

Participants in the first pilot study also provided
responses to a questionnaire about the clarity of the
research instructions. Participants indicated that more
time was needed to complete the questionnaires and that

certain instructions were unclear. As a result, the

111
research instructions were clarified in response to
participant feedback, and time to complete the
questionnaires was increased from one hour to one and one-
half hours.

Participants also answered open-ended questions
regarding their beliefs about the topic of interest in the
research being conducted. Fifty-two participants were
unable to finish these questions due to time constraints.

Of the remaining 198 participants who did answer the
questions, 100 (51 %) indicated that the purpose of the
research was to investigate personnel decision making or the
process of evaluating the work of others. Thirty-six (18%)
of the 198 participants indicated that the purpose of the
research was to evaluate their ability to answer questions
consistently or accurately, while 44 (22%) indicated that
they had no idea what the research was about. Only 18 (9%)
of the 198 participants indicated that gender was a topic of
consideration for this research. These results suggest that
experimenter demand effects regarding gender were minimized
using the current experimental procedure.

second pilot study. As noted above, the fictional
employee files and measurement scales used in the first
pilot study were revised and a second pilot study was
conducted. There were 60 participants in this second pilot
study. The purpose of this study was to ensure that
perceptions regarding causes of performance, attributions,

and personnel decisions were distinguishable and that they

112
could be reliably measured. This pilot study also tested
revisions in the research instructions and ensured that
experimenter demand effects were minimized.

The results of the factor analyses and reliability
analyses were similar to those obtained in the complete
study. For this reason, only the results of the complete
study will be presented. The obtained results of the
complete study are presented in the Results section.
Manipulation checks indicated that the fictional employee
files affected perceived causes of performance, and that
these could be reliably measured.

Twenty participants were randomly selected for a five-
minute interview immediately following the completion of
their questionnaires. All 20 stated that the research
instructions were clear and easy to follow. Nineteen of the
20 participants indicated that the purpose of the research
was to investigate personnel evaluation or decision making.
Only one participant suggested that gender of the employee
was an important variable in the research.

The results of the pilot studies yielded fictional
employee files which were effective in manipulating
performance trends, perceived causes of observed
performance, and employee gender. Furthermore, the research
instructions were clear and easy to follow and participants
were rarely aware that gender of the target employee was
manipulated or important to the study. The nature of the

specific manipulations are described below.

 

 

It;

113
We

A set of fictional employee files were developed for
presentation to each of the participants who took part in
the study. Attributions and personnel decisions were made
in response to the fictional employee files presented as
part of the study. The fictional employee files consisted
of biographical data, current and past performance history
for the year, and a written statement by the target
employee's hypothetical manager reviewing the cause of that
employee's performance for the year.

The variables which were manipulated in the employee
files were: 1) the gender of the target employee: 2) the
observed performance trend; and 3) informational cues of the
causes of the observed performance. The 2 (gender of target
employee) X 3 (performance trend) X 3 (informational cues)
matrix yielded 18 different employee files needed for the
purposes of the study. Each file was identical in content
with several exceptions.

First, names and pronouns were changed to reflect the
gender of the target employee. The names of the male and
female target employees were matched on perceived
attractiveness and competence to remove effects which might
be due to stereotypes which were associated with specific
names (Kasof, 1993).

Second, performance trend information was changed
according to whether the manipulated performance trend was

improving, maintaining, or declining. The performance trend

114
information was presented in graphical and tabular form with
the overall average performance of each employee for the
year clearly marked.

Third, informational cues were changed to reflect the
presentation of different possible causes of the observed
performance trends. These informational cues were
manipulated as displayed in Table 1. Some examples of the
18 fictional employee files are presented in Appendix D,
along with an outline for creating the employee files.

Each of the participants were presented with three of
the 18 possible employee files. These three files were
identical in the gender of the target employee and the
observed performance trend. The ability, effort, and ease
of job informational cues were the only manipulations which
varied among the three files presented to each participant.
Thus, informational cues was a within-subjects variable
while performance trend and gender of the target employee
were between-subjects variables.

Participants were asked to answer a variety of
questions regarding the causes of the observed performance
of the employees presented in the fictional files, perceived
attributions, and personnel decisions. They were also asked
to provide data regarding their gender, age, other
demographics, and their attitudes toward women as managers.
The following section discusses the exact nature of the two
questionnaires used in the complete study to measure

independent and dependent variables of interest.

lie-x1

115

Table 1

IMPROVING MAINTAINING
33293: High effort High effort
Observed Hard job Hard job
performance Low ability Low ability
in spite of:
ABILITX High ability High ability
Observed Hard job Hard job
performance Low effort Low effort
in spite of:
EASE OF JOB Easy job Easy job
Observed Low effort Low effort
performance Low ability Low ability

in spite of:

QEQLININQ

Low effort

Easy job

High ability

Low ability

Easy job

High effort

Hard job

High effort
High ability

116

Measures
Eipst gnesfionnaire. The perceived causes,

attributions and personnel decisions questionnaire measured

several sets of constructs: 1) perceived causes of observed

performance trends; 2) perceived controllability, locus, and

stability of the causes: and 3) promotion decisions, pay

decisions, expectations for future performance, and

likability. It also contained three questions which ‘

functioned as manipulation checks. A total of 55 block

 

randomized five-point Likert scale items were presented in l
the first questionnaire. The 55 items will be discussed in
four blocks which correspond with the above three sets of
constructs, plus manipulation checks.

The first block of questions measured perceptions of
ability, effort, and ease of job as possible causes of
performance. Three items were written for each of the seven
identified causes and this yielded a total of 9 questions
for the first block.

The items were developed by modifying and adapting
instruments used by previous researchers for the current
study. For example, Deaux and Emswiller (1974) used a
question to ask whether participants thought observed
performance was due to ability or luck. Unfortunately,
their procedure resulted in an ipsative scale, and it was
ambiguous in interpretation. The Deaux and Emswiller
question was modified and adapted for use in the current

study; it became two questions in the first questionnaire.

117
An example of one of these questions is provided below (see
Appendix E for the full set of scales):
'lit
I believe the observed performance of the store was due
to the Store Manager's high ability.
1=Strongly Disagree
2=Disagree
3=Neither Disagree Nor Agree l

4=Agree

 

5=Strongly Agree l

The 9 questions, which were written through the process
described above, were block randomized and presented in the
early part of the perceived causes, attributions, and
personnel decisions questionnaire. These items were used in
a manipulation check of the effectiveness of the
informational cues contained in the employee files read by
the participants.

The next block of the first questionnaire dealt with
perceptions of controllability, locus of causation, and
stability of the causes of observed performance. These
items were derived from the Causal Dimension Scale (CBS)
developed by Russell (1982). The items in the CDS are
ipsative; thus, they were rewritten and adapted for use in
the current research effort.

A total of 18 items were generated for the three
dimensions of causation identified by attribution theory

(eight controllability, five locus of causation, and five

118

stability items). These eighteen items were in five-point

Likert scale format and they comprised the second block of

the first questionnaire. A sample of one of the questions

is provided below (see Appendix E for the full set of

scales):

0 t o ab' i

The CAUSE of the observed performance of this Store

Manager was something which the Store Manager could

directly control.
1=Strongly Disagree
2=Disagree
3=Neither Disagree Nor Agree
4=Agree
5=Strongly Agree
The 18 questions which were generated by adapting and
modifying the CDS to the current study were block randomized

and they were presented in the second part of the first

questionnaire. These items were used as dependent and

independent variables in the analyses to be discussed later.

The third block of the first questionnaire consisted of

items designed to measure expectations for future
performance and general positive or negative affect toward

the employee in question. Items also measured the

likelihood of making a decision to promote the employee, to
increase the employee's wage or salary, or to provide a
merit pay bonus for that employee's performance. Some of

the items were derived from previous research efforts (e.g.

119

Pazy, 1986) though most were specifically written for the

purposes of the current study.

Five items were generated for promotion and salary

decisions, five more were generated for expectations for

future performance, four items were generated for merit pay

decisions, and six items were generated for affect toward

the employee, yielding a total of 25 items. These items

were randomized and placed at the end of the third block of

the first questionnaire. A sample of one of the questions

is provided below (see Appendix E for the full set of

scales):
Promotion
I would recommend this person for permanent promotion
to Division Manager.
1=Strongly Disagree
2=Disagree
3=Neither Disagree Nor Agree
4=Agree
5=Strongly Agree
The three manipulation check items which were in the
first questionnaire asked questions regarding the average
profit level of the target employee, the average profit
level of all employees in the region, and whether the target

employee was average, above average, or below average in

profit level.

c s ' n 're. After participants had completed

the perceived causes, attributions and personnel decisions

 

 

120
questionnaire for all three employee files, they were asked
to complete the WAMS (second) questionnaire. The WAMS
questionnaire consisted of 21 questions which comprise the
WAMS (Terborg, et al., 1977), several demographic variables,
and manipulation checks. There were a total of 32 questions
in the second questionnaire.

The first 21 items of the second questionnaire came
from the WAMS (Terborg, et al., 1977). The WAMS is a 21—
item scale which is designed to measure the extent to which
people have negative stereotypes of women as managers. It
has a split-half reliability of .92, and views toward
women's rights have been positively correlated with WAMS
scores for both men and women (Terborg et al., 1977). WAMS
scores appear to be resistant to experimental manipulations
and it has been successfully used in research relating
gender stereotypes with personnel decisions and attributions
of performance (Garland & Price, 1977; Stevens & DeNisi,
1980).

The last eleven questions of the WAMS questionnaire
consisted of manipulation checks and demographic variables.
The first six of the last eleven questions asked whether
participants based promotion, pay raise, and merit pay bonus
decisions on expected future performance or perceptions of
deservedness or reward.

The last five questions of the WAMS questionnaire were
included at the end of the study to avoid biasing the

responses of participants. If these questions had been

121
included earlier in the study, participants might have
become sensitized to the nature of the manipulations. The
first of these last five remaining questions was designed to
determine whether or not participants noticed the gender of
the target employee. The next two questions provided data
about the gender of the participant and age of the
participant. The final two questions were about the
supervisory work experience of the participant.

Table 2 is a summary of variables which were measured
by each of the two questionnaires and the number of items
used for each variable. Table 3 summarizes the major
independent and dependent variables of interest, along with
manipulation checks.

Prosegure

Each participant read a scenario which described a
specific personnel decision making context. Participants
were asked to imagine themselves as managers of a rapidly
growing computer products company. The descriptive scenario
provided allowed participants to better understand the exact
characteristics of the presented computer company and the
context within which the personnel decisions were to be
made. The specific scenarios participants read are
presented in Appendix D.

After reading the scenario which described the decision
making context, participants were then asked to read one of
three fictional employee files. These files presented

information which described the performance of various

122
Table 2

a o uestionnaires

Espcsivsd Causes, Attributionsl and Personnel Decisions
Qusstionnaire (First Questionnaire)

 

 

First Block: Perceived Causes of Performance (Man. Checks)

Effort: 3 items
Ability: 3 items
Ease of job: 3 items

Second Block: Perceived Locus, Controllability, Stability
Locus of Causation: 5 items
Controllability: 8 items
Stability: 5 items

Third Block: Expectations, Affect, and Personnel Decisions

Expectations future performance: 5 items
Affect toward employee: 6 items
Promotion decisions: 5 items
Salary decisions: 5 items
Merit pay bonus decisions: 4 items

Fourth Block: Manipulation checks of perceptions
Profit level for target employee: 1 item
Profit level for all employees: 1 item

Above average target employees: 1 item

Total number of items in first questionnaire = 55

 

 

IS‘.

123

Table 2 (cont'd)

e ic ' ula ion hecks uestionna're

(Second Questionnaire)

First block: WAMS

Attitudes toward women as managers: 21 items

Second block: Demographics, Manipulation Checks, and Other
Promotion based on future performance: 1 item
Promotion based on reward: 1 item
Pay raise based on future performance: 1 item
Pay raise based on reward: 1 item

Merit pay based on future performance: 1 item

Merit pay based on reward: 1 item
Perceived gender of target employees: 1 item
Gender of participant: 1 item
Age of participant: 1 item
Supervisory work experience: 1 item
Length of supervisory experience: 1 item
= 32

Total number of items in second questionnaire

124

Table 3

Ingspsndenp sng Qspendent Variables of Interest
Ingspengent Variables

1) Gender of target person: male or female

2) Ingroup/outgroup gender: same or different

3) Performance trend: improving, maintaining, or declining
(all performances above average)

4) Informational cues:
a. controllable-internal: effort
b. uncontrollable-internal: ability

c. uncontrollable-external: ease of job

5) Women as Managers Scale (WAMS)

Dependent variables:

1) Attributions formed for controllability, stability, and

locus of causation.
2) Promotion decisions
3) Monetary reward decisions
4) Affect toward employee
5) Expectations for future performance
Manipulation checks:
1) Target employee gender recognition
2) Dependent variables of perceived causes (effort,

ability, ease of job)

3) Performance level recognition for target employee

125
managers over the course of the past year. The files
described managers who were exactly equal in overall
performance level, but who exhibited different performance
trends. Participants also received information which

allowed them to infer causes of the observed performance

trends of the managers for the past year. This information

was expected to affect perceptions of locus of causality
(internal/external) and controllability of the observed
performance trends.

Participants answered a variety of questions after each
presentation of a fictional employee file. These questions
included items which measured perceived causality,
attributions, and decisions regarding promotions, salary
increases, and merit pay bonuses for lower level managers.
Data was collected in groups of 20 to 30 people at a
time. Participants received three fictional target employee
personnel files when they entered the experiment, plus they
received a sample employee file for instructional and
familiarization purposes.

Participants first opened the sample file and they read
a description of the fictional company and the context
within which they were to make their personnel decisions.
The researcher then entertained questions regarding the
description of the company and the nature of the personnel

decision making task.

Next, participants were asked to review a sample of an

employee's performance history. The purpose of the sample

126
file was to ensure that participants understood how to read
and use graphs, rating forms, and managerial comments which
were related to an employee's work history. Again, the
researcher entertained questions regarding the use of the
employee files for making personnel decisions.

Participants then read one of the three employee files
which had been preselected to counterbalance order of
presentation of the files. As noted above, these files
differed on information regarding the cause of the presented
performance trend variable, but they did not differ on
gender of the target employee nor on the observed
performance trend itself.

After reading through the first of the three employee

files, participants answered the first questionnaire in

response to that file. The questionnaire consisted of 55

five-point Likert scale questions regarding perceptions of
possible causes of the observed performance trend,
controllability of causes, stability of causes, locus of

causation, expectations for future performance, affect, and

personnel decisions. After the questionnaire was completed

for the first employee, participants read the next of the

remaining employee files which had been preselected for

counterbalancing, and answered the same 55 questions for the

second employee. This process was repeated for the final

remaining employee file.

Answers to questionnaires were recorded on optical scan

forms for scoring convenience, and these answers provided

127
data on independent and dependent variables of interest.
Certain answers to the questionnaires also functioned as
manipulation checks for the experiment.

Once all of the information on the three employee files
was collected, participants were asked to answer a second
questionnaire. This questionnaire was only answered once
since it was not specifically related to the employee files.

The second questionnaire contained 32 questions.
Twenty-one of the questions were from the WAMS developed by
Terborg et al. (1977), and the last eleven questions
consisted of manipulation checks and demographic
information.

Participants were asked to answer a total of 165
questions (3 * 55) about the three employee files and
another 32 on the second questionnaire. Once the
participants completed the questionnaires, they were
appropriately debriefed. The debriefing form and the
experimental protocol for the instruction and administration

of the experimental procedure are presented in Appendix F.

RESULTS

e 'ew
Results from the study were analyzed in several ways.
First, the measurement model for the various scales was
established. An exploratory factor analysis was performed

on each of the three blocks in the first questionnaire and

on the entire second questionnaire. Next reliability

analyses were conducted on each of the scales within the
questionnaires. Various items were dropped from their
respective scales as a result of the factor analyses and
reliability analyses. This process was repeated in an
iterative fashion until maximally reliable scales were

developed which effectively measured the constructs of

interest. Once the measurement model was established,

analyses of manipulation checks were conducted.
The manipulation checks included participant
perceptions regarding the overall level of target employee

performance, the perceived gender of the target employee,

and perceived causes of performance. Most of the

manipulation checks were descriptive, though the perceived
causes of performance manipulation check was tested using a
doubly multivariate repeated measures MANOVA. The dependent

variables for this MANOVA were perceived effort, ability,

128

129
and ease of job as causes of the observed performance
trends. The independent variables were type of performance
trend and informational cues presented.

Once the measurement model was established and the
manipulation checks proved effective, a variety of
analytical techniques were used to test the hypotheses
presented earlier. These analyses were done in six steps.
First, the effects of performance trend, informational cues,
and gender of target employee on personnel decisions were
analyzed. Second, the effects of performance trend,
informational cues, and gender of target employee on
perceived external locus of causation, controllability, and
stability were analyzed. Third, the effects of performance
trend, informational cues, and gender group membership
(same/different) on personnel decisions were analyzed.
Fourth, the effects of performance trend, informational
cues, and gender group membership (same/different) on
perceived external locus of causation, controllability, and
stability were analyzed. Fifth, the effects of performance
trend and perceived controllability, stability, and external
locus on personnel decisions were established using repeated
measures regression analyses. Finally, a set of analyses
were conducted to investigate the effect of WAMS on
attributions and personnel decisions, though specific
hypotheses for these analyses were not presented.
Hypotheses, relevant constructs, and methods of analysis are

presented in Table 4.

 

In:

Table 4

130

 

Hypptheses, Relevant Constructsi,and Analyses

Experiences—is

1 8 2

8a-9b

10-10e

11-1lb

12-13b

l4-14e

15-15b

IVs

Performance Trnd

Performance Trnd
ExtLocus Causality

Performance Trnd
Controllability

Performance Trnd

Performance Trnd

Informational Cues

Performance Trnd
Gender of Employee

Performance Trnd
Gender of Employee
Informational Cues

Performance Trnd
Gender of Employee

Performance Trnd
Gender Grp Mmbrship

Performance Trnd
Gender Grp Mmbrship
Informational Cues

Performance Trnd

DVS
Promotion
Pay

Promotion
Pay

Promotion
Pay

Stability

Promotion

Cntrllablty
Exths Cslty
Stability

Cntrllablty
Exths Cslty
Stability

Cntrllablty
Exths Cslty
Stability

Promotion
Pay

Cntrllablty
Exths Cslty
Stability

Cntrllablty
Exths Cslty
Stability

Promotion

Gender Grp Mmbrship Pay

amnesia
MANOVA ,

ANOVAs,
Regression

Regression

Regression

ANOVA,
Regression

ANOVA,
Regression

MANOVA,
ANOVAs

MANOVA,
ANOVAs

MANOVA,
ANOVAs

MANOVA,
ANOVAs
MANOVA,
ANOVAs

MANOVA,
ANOVAs

MANOVA,
ANOVAs

131

Three types of analyses were used to test hypotheses.
These included MANOVAs, ANOVAs, and repeated measures
regressions. Each of these will be briefly described below,
and a more detailed discussion of the repeated measures
regression analysis will follow later.

A doubly-multivariate repeated measures MANOVA was
performed, using each of the last two sets of dependent
variables measured in the first questionnaire. The two sets
of dependent variables included 1) attributions (perceived
controllability, external locus of causality, and
stability); and 2) personnel decisions (expectations for
future performance, affect toward target employee, and
promotion and pay decisions). The independent variables
used in various MANOVAs included participant gender, target
employee gender, gender ingroup/outgroup status, performance
trend, and informational cues presented. The MANOVA
procedure was used to minimize Type I error associated with
conducting repeated tests on multiple dependent variables
and to take into consideration interrelationships between
the dependent variables of interest.

It should be noted, however, that findings involving
the specific effects of the independent variables on the
dependent variables of interest have to be interpreted with
caution when using the MANOVA procedure (Hunter, 1987). In
the first step of the MANOVA procedure, independent
variables are regressed on the dependent variables; that is

to say, the causal direction of the analysis is reversed.

132
Linear combinations of the dependent variables which predict
the independent variables become the dependent variables
which are analyzed in subsequent procedures.

This procedure often leads to uninterpretable results,
particularly when it is used with intercorrelated dependent
variables (Hunter, 1987). For example, f tests and effect
sizes for each of the dependent variables are influenced by
suppressor effects, error of measurement, and problems of
collinearity in the dependent variables, much like problems
which can occur in discriminant analyses (Hunter, 1987;
SPSS, 1988, p. 214, 223). A more appropriate method of
analysis is to conduct specific univariate ANOVAs or
regressions for each of the dependent variables of interest,
particularly when hypotheses suggest specific relations
should exist (Hunter, 1987). This was the primary method of
analysis for testing most of the hypotheses.

Once the MANOVAs and ANOVAs were conducted, repeated
measures regression analyses were employed to support the
findings of the MANOVAs and ANOVAs and to test other
hypotheses which were untestable using those procedures
(Cohen & Cohen, 1983; Hollenbeck, Ilgen, & Sego, in press).

In a repeated measures regression analysis, the
variance in the dependent variable is partitioned into
between and within-subjects effects. The resulting E and p
tests for the regressions and beta weights are then adjusted
to reflect the increment.iJiI8 as a function of whether the

effect is a between or within—subjects variable. The

133
results of these analyses also provide specific tests for
differences between cell means.

Some advantages of the repeated measures regression
approach are that it effectively uses the degrees of freedom
available in a within-subjects design, the power of the F
test is increased by the method of calculating the 3 ratio,
and it provides opportunities to use continuous independent
variables without dichotomizing measures. Especially
important for this study, repeated measures regression
analysis allows one to regress personnel decisions on the
actual perceptions of controllability, stability, and
external locus of causality provided by the participants in
the study. This analysis is similar to a policy capturing
study, but in this case, the attributions are the cues which
are weighted by the participants in the decision making
process. The strength of this approach will become more
apparent as the specific analyses are described.

The combination of MANOVAs, ANOVAs, and repeated
measures regressions provided reasonable methods for testing
the hypotheses presented earlier using the data collected in
this study. Each step of the analyses and the results
obtained are discussed in the following sections.

In summary, the results of the full study were analyzed
in several steps. First, measurements of the constructs of
interest were evaluated using exploratory factor analyses

and reliability analyses. Second, manipulations were

checked using various items and by performing a MANOVA using

134
the nine employee files (collapsed across gender of
employee) as independent variables and by using the three
dependent variables of perceived causes of performance
(effort, ability, ease of job) as manipulation checks.
Third, MANOVAs were conducted to test for overall effects in
attributions and personnel decisions while controlling for
Type I error and intercorrelations in dependent variables,
and univariate ANOVAs were conducted to test for specific
hypothesized effects. Finally, repeated measures regression
analyses were conducted to test all remaining linkages in
the model and to provide further tests of various
hypotheses.
Estsblishment of ths_Measurement Modsi

The evaluation of the measurement model was conducted
using factor analyses, calculation of reliabilities,
examination of scale intercorrelations, and examination of
parallelism of items within scales with constructs external
to the scales.

Analyses of the measurement model included one
independent variable scale, and three blocks of dependent
variable scales. The independent variable scale was the
Women As Managers Scale (WAMS). All other independent
variables were either manipulated (e.g. performance trend)
or were demographic variables (e.g. gender of participant),
so they were not evaluated using factor analyses or
reliabilities. The three blocks of dependent variable

scales analyzed were: 1) manipulation checks of perceived

135

causes of performance (effort, ability, ease of job), 2)
perceived attributions (controllability, stability, and
external locus of causation), and 3) personnel decisions
(promotion, salary/pay increases, merit pay,
positive/negative affect, expectations for future
performance). Analyses for the WAMS and each of the
dependent variable blocks will be described separately.

WAM_. An examination of the means and standard
deviations of the 21 items comprising the WAMS scale
indicated that they were within normal ranges for a
questionnaire in five point Likert scale format, though some
ceiling and floor effects were noted. Participant responses
(n = 252) to WAMS items were factor analyzed using a
principal components analysis with a varimax rotation. The
results of this factor analysis are presented in Table 5.

The factor analysis yielded a three factor solution and
the first factor accounted for 52.4% of the variance with an
eigenvalue of 10.99. The next two factors had eigenvalues
less than 1.5 and each of the factors accounted for less
than 7% of the variance. The cross-loadings of each of the
items across factors were positive and most were fairly
high. Use of eigenvalues greater than one as the extraction
criteria for factors has been shown to be too liberal and to
result in overfactoring. For this reason, a scree test was
employed as the criteria for determining the number of

factors in the analysis. The large break in eigenvalues and

136

Table 5

Esgpp; Anaiysis 9n phe WAMS

PRINCIPAL-COMPONENTS ANALYSIS (PC)
FINAL STATISTICS:

VAR COMMUNALITY* FACTOR EIGENVALUE PCT OF VAR CUM PCT
52.4 52.4

 

V1 .67631 * 1 10.99924

V2 .56782 * 2 1.31895 6.3 58.7

V3 .48392 * 3 1.08466 5.2 63.8

V4 .76846 *

V5 .72339 *

V6 .71610 *

V7 .49307 *

V8 .81475 *

V9 .78947 *

V10 .69687 *

V11 .66066 *

V12 .60298 *

V13 .62267 *

V14 .44065 *

V15 .27170 *

V16 .63098 *

V17 .63359 *

V18 .71826 *

V19 .62057 *

V20 .72081 *

V21 .74983 *

ROTATED FACTOR MATRIX:

FACTOR 1 FACTOR 2 FACTOR 3

V16 .76215 .18913 .11973
V6 .74174 .33311 .23443
V17 .71859 .33904 .04760
V18 .71221 .38468 .25108
V21 .70766 .44528 .22530
V1 .69177 .22700 .38241
V20 .67228 .42203 .30124
V3 .65869 .13311 .17982
V2 .51005 .44806 .32696
V15 .42187 .20077 .23111
V9 .37229 .79429 .14136
V4 .32587 .79415 .17776
V8 .38656 .77613 .25089
V5 .37247 .73684 .20427
V10 .06131 .69572 .45725
V19 .46732 .58362 .24813
V14 .42676 .42752 .27524

137

Table 5 (cont'd)

 

 

FACTOR 1 FACTOR g FACTOR 3
V11 .10484 .27359 .75816
V12 .31988 .10047 .70040
V7 .23100 .21903 .62589
V13 .46513 .40760 .49009
FACTOR TRANSFORMATION MATRIX:

FACTOR 1 FACTOR 2 FACTOR 3
FACTOR 1 .67873 .60728 .41297
FACTOR 2 -.73395 .54149 .41001
FACTOR 3 .02537 -.58139 .81323

138
the high cross-loadings suggested that the WAMS was
comprised of a single underlying factor.

An examination of correlations of the items within the
WAMS with various external scales and constructs (e.g.
gender, attributions, etc.), indicated that they had a high
degree of parallelism, also supporting the presence of a
single factor solution.

Reliabilities and intercorrelations for each of the
three sub-factors identified in the factor analysis were
computed. The intercorrelations and reliabilities of these
subscales are presented in Table 6. After correction for
attenuation, the sub-scale intercorrelations ranged from
.819 to .869, again supporting the use of the WAMS as a
unidimensional scale.

Item-total correlations, and coefficient alphas were
calculated and the results are presented in Table 7. The
overall reliability for the scale was high (.95) and only
three items would slightly increase the obtained reliability
if they were removed, again suggesting a single factor
solution. Given that the WAMS was designed to measure a
unidimensional construct, and based on the results of the
above analyses, it was determined that the WAMS scale
measured a single construct with high reliability.

These findings support previous research using the WAMS
as a measure of stereotypical attitudes toward women as

managers (e.g. Peters, et al., 1973). All items were

139
Table 6
e r el ti ns f WAMS Sub-scales
WAMSl = Items V1-V3, V6, V15-V18, V20-V21
(alpha=.917, 10 items)
WAMSZ = Items V4-V5, V8-V10, V14, V19
(alpha=.905, 7 items)
WAMS3 = Items V7, V11-V13

(alpha=.752, 4 items)

Sub-scale Intercorrelations

WAMSl WAMS2 WAMS3
WAMSl 1.0000 .7919** .6980**
WAMSZ .86938 1.0000 .6753**
WAMS3 .84058 .81863 1.0000
* - Signif. LE .05 ** - Signif. LE .01 (2-tailed)

Note: 3 indicates the correlations reported below the
diagonal are corrected for attenuation due unreliability in

measurement.

 

 

140

Table 7
Eslispiiity Analysis for the WAMS
ITEM MEAN STD DEV CASES
1. V1 4.3611 .9983 252.0
2. V2 4.1786 .9168 252.0
3. V3 4.1746 1.0301 252.0
4. V4 4.5595 .7524 252.0
5. V5 4.5040 .7548 252.0
6. V6 4.4008 .8666 252.0
7. V7 4.1349 1.1594 252.0
8. V8 4.4841 .7386 252.0
9. V9 4.5437 .7211 252.0
10. V10 4.4484 .8042 252.0
11. V11 3.6349 1.2408 252.0
12. V12 3.6429 1.1704 252.0
13. V13 4.2262 .9862 252.0
14. V14 4.1706 1.0705 252.0
15. V15 3.8492 1.1745 252.0
16. V16 4.2857 .9685 252.0
17. V17 4.4683 .8626 252.0
18. V18 4.4127 .8950 252.0
19. V19 4.2897 .9273 252.0
20. V20 4.3929 .8794 252.0
21. V21 4.4127 .8304 252.0
# OF CASES = 252.0
# OF
STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
WAMS 89.5754 193.6955 13.9175 21
SCALE SCALE CORRECTED
MEAN VARIANCE ITEM- SQUARED ALPHA
IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
E ELETED DELETED CORRLTN CORRLTN DELETED
V1 85.2143 173.2447 .7403 .6352 .9429
V2 85.3968 175.4515 .7166 .6174 .9434
V3 85.4008 177.2531 .5608 .4177 .9458
V4 85.0159 178.4778 .7288 .7178 .9436
V5 85.0714 178.2499 .7381 .7215 .9435
V6 85.1746 175.3001 .7689 .6983 .9427
V7 85.4405 176.3431 .5199 .3887 .9469*
V8 85.0913 177.4458 .7980 .7967 .9428
V9 85.0317 178.8675 .7418 .7212 .9436
V10 85.1270 179.7766 .6154 .5676 .9448
V11 85.9405 174.9168 .5253 .4299 .9472*
V12 85.9325 175.5532 .5408 .4289 .9466
V13 85.3492 173.4314 .7427 .6208 .9429
V14 85.4048 174.9271 .6223 .4626 .9449

 

 

 

Table 7 (cont'd)

V15
V16
V17
V18
V19
V20
V21

85.7262
85.2897
85.1071
85.1627
85.2857
85.1825
85.1627

177.6658
176.2305
177.6259
174.2882
174.9220
174.1418
175.1567

141

.4679 .3041 .9479*
.6427 .5544 .9444
.6666 .6127 .9441
.7873 .7401 .9424
.7303 .6505 .9431
.8092 .7483 .9421
.8121 .7919 .9423

* These are items in which alpha would get higher if they
were removed from the scale.

RELIABILITY COEFFICIENT
ALPHA = .9467

21 ITEMS
STANDARDIZED ITEM ALPHA =

.9526

142
retained and composite scores based on the sum of
participant responses across all 21 items were computed.
The means, standard deviations, and reliabilities of all
final scales used in subsequent analyses will be presented
later.

W. The perceived
causes of performance scales were designed to determine if
the information in the fictional employee files effectively
manipulated the perceived causes of performance. These
scales included measurements of whether the participants
perceived a target employee's performance to be due to the
high or low ability of the employee, high or low effort of
the employee, or high or low ease of the job for the
employee.

Since the study was a repeated measures design, each
participant rated three different target employees on the
different perceived causes of performance. Thus, three
separate factor analyses could be conducted, one for each
set of ratings provided by participants. The selection of
participant responses for factor analyses were based upon
the order in which they provided their answers. The
responses of the participants to the first target employee
presented were the first set of items which were factor
analyzed.

Similarly, participant responses to the second and
third target employees presented were the second and third

sets of items which were factor analyzed. Some participants

143
failed to properly note the order of their responses, so the
usable sample size varied by analysis (n = 208, 195, and 192
for the first, second, and third sets of ratings,
respectively). The item means and variances across all
three sets of ratings were within acceptable ranges for
scales presented in 5-point Likert format. The results of
all three factor analyses were highly similar. For
presentational purposes, only the results of the first
factor analyses will be presented.

When using eigenvalues greater than one as the factor
extraction criteria, the factor analysis for the first set
of responses yielded a two-factor solution, with most of the
ability and ease of job items loading on one factor and
mostly effort items loading on the second factor. These
results are presented in Table 8. In the second and third
factor analyses, similar results were obtained, though the
patterns of loadings for the ease of job items differed
slightly. Because each of the perceived causes of
performance were simultaneously manipulated in the study, it
is not particularly surprising that items within each of the
scales tended to covary with one another.

When a factor analysis with a three factor solution was
computed, every item loaded on the factor which corresponded
to the scale for which that item was written. This was true
for all three sets of factor analyses computed with a three

factor solution. The results of the three factor solution

144

Table 8

Essppr Analysis on the Perceived Causes of Performance

PRINCIPAL-COMPONENTS ANALYSIS (PC)
FINAL STATISTICS:

VAR COMMUNALITY* FACTOR EIGENVALUE PCT OF VAR CUM PCT

 

V1 .74432 * 1 4.08919 45.4 45.4
V2 .69245 * 2 2.51976 28.0 73.4
V3 .77987 *
V4 .59919 *
V5 .74661 *
V6 .84405 *
V7 .75980 *
V8 .70116 *
V9 .74150 *

ROTATED FACTOR MATRIX:

E = Perceived high or low effort as a cause of observed
performance

A = Perceived high or low ability as a cause of observed
performance

JE = Perceived high or low ease of job as a cause of
observed performance

 

FACTOR 1 FACTOR 2
V6(E) .91717 -.05346
V3(E) .88197 -.O4458
V7(E) .87164 -.00665
V9(JE) -.68445 -.52252
V5(A) -.05619 .86224
V1(A) -.22466 .83298
V8(A) .11955 .82878
V2(JE) -.55862 -.61677
V4(JE) -.49616 -.59416

FACTOR TRANSFORMATION MATRIX:

FACTOR 1 FACTOR 2

FACTOR 1 .77952 .62637
FACTOR 2 -.62637 .77952

145
with the first set of responses are presented in Table 9.
The three factors obtained corresponded to effort, ability,
and ease of job scales.

Computation of reliabilities for the three sets of
participant responses were conducted in a fashion similar to
the way the factor analyses were done. Participant
responses to the first, second, or third target employee
were selected separately and reliabilities were then
calculated. Alphas for each of the scales (effort, ability,
ease of job) ranged from .86 to .96 across all three sets of
analyses. There were only two cases in which reliabilities
might be slightly improved if an item was deleted, but this
pattern was inconsistent across the three sets of
participant responses analyzed. The content of the items in
combination with the high reliabilities suggest that three
different perceived causes of performance were being
measured and manipulated. Further support was found by
examining the parallelism of the items in their correlations
with other scales.

Scale scores were computed on the basis of the three
factor solution and intercorrelations were calculated. This
was done three times, once for each set of participant
responses. Similar findings were obtained across all three
sets of participant responses and only the results of the
first set will be presented in Table 10. After correction
for unreliability, the correlations between ability, effort,

and ease of job scales ranged from -.044 to -.590 for the

 

146
Table 9

Inpss Esgpgf Solution for Perceived Causes of Performance

PRINCIPAL-COMPONENTS ANALYSIS (PC)
FINAL STATISTICS:

VAR, COMMUNALITY* FACTOR EIGENVALUE PCT OF VAR CUM PCT

V1 .74820 * 1 4.08919 45.4 45.4
V2 .72678 * 2 2.51976 28.0 73.4
V3 .81733 * 3 .70744 7.9 81.3
V4 .84869 *
V5 .85533 *
V6 .90387 *
V7 .79176 *
V8 .79449 *
V9 .82993 *

ROTATED FACTOR MATRIX:

E = Perceived high or low effort as a cause of observed
performance

A = Perceived high or low ability as a cause of observed
performance

JE = Perceived high or low ease of job as a cause of
observed performance

 

EFFORT ABILITY EASEJOB
V6(E) .93030 -.03313 -.19314
V3(E) .87555 -.04638 -.22045
V7(E) .85504 -.01965 -.24554
V5(A) -.Ol628 .91293 -.14704
V8(A) .13388 .85167 -.22634
V1(A) -.27941 .77611 -.26034
V4(JE) -.14751 -.21035 .88470
V9(JE) -.41263 -.22567 .78022
V2(JE) -.33751 -.37670 .68627

FACTOR TRANSFORMATION MATRIX:

 

FACTOR 1 FACTOR 2 FACTOR 3
FACTOR 1 .61392 .45012 -.64846
FACTOR 2 -.65743 .74625 -.10442

FACTOR 3 .43691 .49042 .75406

 

147
Table 10
Intezcorfelations of the Perceived Causes of Performance

EASEJOB = Items V2, V4, V9 (alpha=.853, 3 items)

ABILITY = Items V1, V5, V8 (alpha=.855, 3 items)
EFFORT = Items V3, V6, V7 (alpha=.914, 3 items)
EFFORT ABILITY EASEJOB
EFFORT 1.0000 —.O388 -.5210**
ABILITY -.04398 1.0000 -.4850**
EASEJOB -.5900a -.56798 1.0000
* - Signif. LE .05 ** — Signif. LE .01 (2-tailed)

Note: a indicates the correlations reported below the
diagonal are corrected for attenuation due unreliability in

measurement.

 

 

148
first set of responses. The moderate correlations suggest
that it is appropriate to use the different scales to
measure different perceived causes of performance.

These scales function primarily as manipulation checks
to ensure the employee files effectively manipulated
perceived causation in the desired manner. Three sets of
scores were computed for each of the perceived causes by
summing across items comprising each of the scales. Thus,
each participant had nine total composite scores for
perceived cause of performance: effort (first, second,
third target employee); ability (first, second, third target
employee); and job ease (first, second, third target
employee). The means, standard deviations, and
reliabilities of these scales will be presented later.

Perceived attributions scales. The perceived
attributions scales were designed to measure whether the
causes of observed performances were perceived to be high or
low in controllability, high or low in stability, and
internally or externally caused.

As noted earlier, the study was a repeated measures
design, so each participant rated three different target
employees on the different attributions for performance.
Thus, three separate factor analyses were computed, one for
each set of attribution ratings provided by participants.
The selection of participant responses for factor analyses
were based upon the order in which they provided their

answers. Also noted above, some participants failed to

 

149
properly note the order of their responses, so the usable
sample size varied by analysis (n = 208, 195, and 192 for
the first, second, and third sets of ratings, respectively).
The item means and variances across all three sets of
ratings were within acceptable ranges for scales presented
in 5-point Likert format. The results of all three factor
analyses were highly similar. As before, only the results
of the first factor analyses will be presented.

The factor analyses for the first set of participant
responses yielded a four—factor solution, with most of the
items loading on the factor for which they were written to
measure. These results are presented in Table 11. In the
second and third factor analyses, similar results were
obtained, though the patterns of loadings for some of the
items differed slightly. In all three analyses, items V16,
V10, and V19 (reverse worded controllability items) tended
to create a factor separate from the main controllability
dimension for which they were written to measure. The four
factors obtained corresponded to the external locus,
controllability, stability attributions, with the
problematic controllability items creating the fourth
factor.

Computation of reliabilities were done in a fashion
similar to the method used to conduct factor analyses.
Participant responses to the first, second, or third target
employee were selected separately and reliabilities were

then calculated. Alphas for each of the scales (perceived

 

F-

‘u‘ '1‘ V‘ ‘1 ‘ ‘ I ‘

150
Table 11

Egsfpp Anaiysis on the Attributions pf Causes of PepfOfnance

PRINCIPAL-COMPONENTS ANALYSIS (PC)
FINAL STATISTICS:

VAR QQMMUNALII!* FACTOR EIGENVALUE PCT OF VAR CUM PQT
V10 .32050 * 1 4.98546 27.7 27.7
V11 .47552 * 2 2.09665 11.6 39.3
V12 .57327 * 3 1.82477 10.1 49.5
V13 .53280 * 4 1.21787 6.8 56.2
V14 .56166 *
V15 .42985 *
V16 .62548 *
V17 .60198 *
V18 .51839 *
V19 .72723 *
V20 .58492 *
V21 .51277 *
V22 .62613 *
V23 .53124 *
V24 .67761 *
V25 .56061 *
V26 .65693 *
V27 .60786 *

ROTATED FACTOR MATRIX:

EL = Perceived external or internal locus of causation for
observed performance

C = Perceived high or low controllability of causation for
observed performance
S = Perceived high or low stability of causation for

observed performance

 

EXTLOCUS CONTROL STABILITY CONTROL;
V12(EL) .74579 -.05404 .11658 .02362
V20(EL) .74526 -.16755 -.03052 .02244
V22(EL) .74367 -.26565 -.05002 -.00166
V14(EL) .71677 -.15058 .03026 .15590
V25(EL) .67563 -.27258 -.15674 .07260

V17(C) -.58256 .49314 -.07093 .11988

VV“

151

Table 11 (cont'd)

 

EXTLQCUS CONTROL STABILITY CONTROLZ
V13(C) -.os946 .72117 -.05147 .04535
V24(C) -.39485 .72052 -.03352 .03783
V27(C) -.32579 .68705 -.15016 -.os447
V21(C) -.45282 .54586 -.05439 .08242
V26(S) -.20054 -.31381 .65385 .30119
V23(S) .21599 -.13622 .65321 .19834
V18(S) -.06401 .14741 .63928 -.28962
V15(S) -.20227 -.10949 .60906 -.o7745
V11(S) .36884 .04644 .57871 -.04908
V19(C) .05519 -.09922 -.06508 .84268
V16(C) .07504 .07874 .02842 .78284
V10(C) -.01337 .35680 .01462 .43909

FACTOR TRANSFORMATION MATRIX:

FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4
FACT 1 .80446 -.58073 .11704 .04359
FACT 2 -.27801 -.20529 .93223 -.10735
FACT 3 .05950 .18923 .17056 .96517

FACT 4 .52155 .76472 .29693 -.23455

 

 

152
controllability, perceived stability, and perceived external
locus of causation) ranged from .64 to .93 across all three
sets of analyses.

Table 12 presents the results of the reliability
analyses conducted on the first set of participant responses
to items designed to measure attributions. Results across
the three sets of participant responses indicated that
removal of items V10, V16, and V19 consistently tended to
improve the reliabilities of the perceived controllability
scale. These were the same items which created their own
factor in the factor analyses. Almost all the other items
improved the reliabilities of the scale for which they were
written.

Based on the results of the reliability and factor
analyses, Items V10, V16, and V19 were removed from all
further analyses. Scale scores were computed for the three
attributions (controllability, stability, and external locus
of causation) and intercorrelations were calculated. This
was done three times, once for each set of participant
responses. Similar findings were obtained across all three
sets of participant responses and only the results of the
first set will be presented in Table 13.

After correction for unreliability, the correlations
between perceived controllability, external locus of
causation, and stability ranged from .073 to -.751 for the

first set of responses. While the corrected correlation

 

Table 12

Beiiability Analyses for the Attribution Scales

153

Eszpsiysd_Controllabilitv of Cause of Performance Scale

 

 

 

ITEM MEAN STD DEV CASES
1. V10 3.1731 .9524 208.0
2. V13 2.7885 .9395 208.0
3. v16 3.4135 .8525 208.0
4. V17 3.0192 1.0855 208.0
5. V19 3.2837 .9122 208.0
6. V21 2.9760 .9900 208.0
7. V24 2.9952 1.0745 208.0
8. V27 3.1971 1.0743 208.0
# OF CASES = 208.0
# OF
STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
CONTROL 24.8462 21.0004 4.5826 8
SCALE SCALE CORRECTED
MEAN VARIANCE ITEM- SQUARED ALPHA
IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
ITEM DELETED DELETED CORRLTN CORRLTN DELETED
V10 21.6731 17.8250 .2821 .1360 .7145
V13 22.0577 16.7020 .4448 .2488 .6824
v16 21.4327 18.9037 .1848 .2860 .7292*
V17 21.8269 14.9651 .5783 .4490 .6492
V19 21.5625 19.2618 .1132 .3189 .7437*
V21 21.8702 15.9106 .5204 .3687 .6654
V24 21.8510 14.7071 .6235 .5532 .6382
V27 21.6490 15.5429 .5080 .4740 .6666

* These are items in which al

pha would get higher if they

were removed from the scale.

8 ITEMS
STANDARDIZED ITEM ALPHA =

RELIABILITY COEFFICIENTS

ALPHA = .7175 .7048

£§£§§12§0,External Locus of Causation of Performance Scale

1.
2.
3.
4.
5
#

 

ITEM MEAN STD DEV CASES
V12 2.7692 .9950 208.0
V14 3.0288 1.0211 208.0
V20 2.8606 1.1483 208.0
V22 3.0000 1.1207 208.0
V25 2.9135 1.1886 208.0

OF CASES =

208.0

 

Table 12 (cont'd)

154

 

# OF
STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
EXTLOCUS 14.5721 17.8788 4.2283 5
SCALE SCALE CORRECTED
MEAN VARIANCE ITEM- SQUARED ALPHA
IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
ITEM DELETED DELETED CORRLTN CORRLTN DELETED
V12 11.8029 12.7387 .5843 .3950 .8068
V14 11.5433 12.5972 .5848 .3874 .8065
V20 11.7115 11.4913 .6511 .4498 .7878
V22 11.5721 11.4151 .6877 .5112 .7769
V25 11.6587 11.4143 .6290 .4355 .7951

* These are items in which alpha would get higher if they

were removed from the scale.

 

 

 

RELIABILITY COEFFICIENTS 5 ITEMS
ALPHA = .8291 STANDARDIZED ITEM ALPHA = .8294
ESISSiEed_gtébilitx_2£_gggsg:pf Performanqe_8calg

ITEM MEAN STD DEV CASES
1. V11 2.8606 1.0332 208.0
2. v15 3.4087 .9384 208.0
3. v18 2.8894 .9285 208.0
4. V23 3.2933 .9758 208.0
5. V26 3.2163 .9663 208.0
# OF CASES = 208.0

# 0F

SCALE SCALE CORRECTED

MEAN VARIANCE ITEM- SQUARED ALPHA

IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
ITEM DELETED DELETED CORRLTN CORRLTN DELETED
V11 12.8077 6.7648 .3226 .1209 .6182
V15 12.2596 6.8791 .3669 .1819 .5938
V18 12.7788 6.9364 .3614 .1625 .5964
V23 12.3750 6.4094 .4461 .2512 .5545
V26 12.4519 6.4035 .4557 .2512 .5500

* These are items in which alpha would get higher if they

were removed from the scale.

RELIABILITY COEFFICIENTS
ALPHA =

.6364

5 ITEMS

STANDARDIZED ITEM ALPHA =

.6376

155
Table 13
Intercorrelations Q: the Attributions Scales
CONTROL = Items V13, V17, V21, V24, V27
(alpha=.816, 5 items)
EXTLOCUS= Items v12, V14, V20, V22, V25
(alpha=.829, 5 items)
STABLE = Items V11, V15, V18, V23, V26

(alpha=.636, 5 items)

CONTROL EXTLOCUS STABLE

CONTROL 1.0000 -.6176** -.2013**

EXTLOCUS -.75098 1.0000 .0532

STABLE -.27948 .07338 1.0000
* - Signif. LE .05 ** - Signif. LE .01 (2-tailed)

Note: 3 indicates the correlations reported below the
diagonal are corrected for attenuation due unreliability in

measurement.

156
between external locus and controllability was fairly high,
this would be expected since by definition, external causes
of performance are also uncontrollable. Also, the
correlation between controllability and external locus of
causation was smaller in the second and third sets of
participant responses (-.598, —.643, corrected for
unreliability).

The factor analyses, reliabilities, and
intercorrelations between the scales support the notion that
the different scales measure conceptually distinct
attributions. Thus, all three attributional dimensions were
retained for further analyses. Three sets of participant
scores were computed for each of the three perceived
attributions by summing across items comprising each of the
attributional scales. Each participant had nine total
composite scores for perceived attributions for causes of
performance: perceived controllability (first, second,
third target employee); perceived stability (first, second,
third target employee); and perceived external locus of
causation (first, second, third target employee). The

means, standard deviations, and reliabilities of these

scales will be presented later.

Personnel decisions scales. The personnel decisions

scales were designed to measure whether the participants
would be willing to award promotions, pay increases, or
merit pay bonuses to a target employee. They were also

designed to measure positive or negative affect toward the

157
target employees, and expectations for the future
performance of those employees.

As before, each participant rated three different
target employees on the different personnel decisions they
might make based on the observed performance and the
observed causes of performance. Three separate factor
analyses were conducted, one for each set of personnel
decisions provided by participants. As before, the
selection of participant responses for factor analyses were
based upon the order in which they provided their answers,
and the usable sample size varied by analysis (Q = 208, 195,
and 192). The results of all three factor analyses were
highly similar. For presentational purposes, only the
results of the first factor analyses will be presented.

The factor analyses on the first set of participant
decisions yielded a four-factor solution. Most of the items
loaded on the factor for which they were written to measure,
with the exception of the pay raise and merit pay items,
which tended to load onto the same factor. These results
are presented in Table 14. The obtained factors
corresponded to affect, promotion, pay, and expected future
performance personnel decisions.

Computation of reliabilities was done by selecting
participant responses to the first, second, or third target
employee. Alphas for each of the scales (promotion, pay
raise, merit pay, affect, expectation) ranged from .82 to

.94 across all three sets of analyses.

158

Table 14

Eagtor Analysis on the Personnel Decisions Scales

PRINCIPAL-COMPONENTS ANALYSIS (PC)
FINAL STATISTICS:

 

2A3 QQMMUNALII!* FACTOR EIGENVALUE PCT OF VAR CUM BOT

V28 .72622 * 1 11.53886 46.2 46.2

V29 .52665 * 2 2.12779 8.5 54.7

V30 .48206 * 3 1.64898 6.6 61.3

V31 .73667 * 4 1.08887 4.4 65.6

V32 .71579 *

V33 .76906 *

V34 .71369 *

V35 .50533 *

V36 .75448 *

V37 .61678 *

V38 .64511 *

V39 .55653 *

V40 .70320 *

V41 .70455 *

V42 .74265 *

V43 .56970 *

V44 .62049 *

V45 .62489 *

V46 .77090 *

V47 .63602 *

V48 .70873 *

V49 .71897 *

V50 .6537? *

V51 .64396 *

V52 .55830 *

ROTATED FACTOR MATRIX:

A = Positive affect toward the target employee

E = Positive expectations of future performance of the

target employee

PR = Positive pay raise decisions for the target employee

PM = Positive promotion decisions for the target employee

MP = Positive merit pay decisions for the target employee
AFFECT PROMOTION PAY EXPECT

V51(A) .71869 .14651 .18556 .26746

V40(A) .69844 .30340 .29264 .19414

V47(A) .69415 .35327 .17010 .02107

V37(A) .68122 .15178 .32177 .16169

V35(A) .66742 .17724 -.01790 .16778

V45(A) .61537 .21040 .33811 .29600

Table 14 (cont'd)

V48(P)
V52(E)
V50(MP)
V43(MP)

V33(PM)
V28(PM)
V46(PM)
V49(PM)
V42(PM)

V31(PR)
V32(PR)
V30(MP)
V38(PR)
V36(MP)
V29(PR)

V34(E)
V41(E)
V44(E)
V39(E)

FACT 1
FACT 2
FACT 3

159

 

AFF C PROMOTION PAY XPECT
.58669 .42352 .39700 .16597
.55324 .22422 .31598 .31955
.53474 .25442 .51978 .18144
.53005 .20019 .48630 .11039
.13852 .85850 .10058 .05228
.22585 .78782 .13251 .19233
.30004 .76972 .08591 .28465
.25136 .74514 .19126 .25293
.29345 .74164 .10729 .30820
.12962 .01140 .72754 .43638
.22317 .04312 .71267 .39525
.13155 .16498 .65960 -.04966
.32159 .07994 .62680 .37739
.55022 .27857 .60971 .04887
.35756 .43046 .45595 -.07496
.05941 .21203 .07992 .81168
.26269 .26349 .12610 .74177
.23813 .22902 .20620 .68470
.42297 .12863 .26283 .54038
FACTOR TRANSFORMATION MATRIX:
FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4
.62001 .48250 .47633 .39483
-.14849 .81160 -.55865 -.08468
-.41878 -.07032 -.12673 .89645
-.64666 .32180 .66705 -.18254

FACT 4

160

Table 15 presents the results of the reliability
analyses conducted on the first set of participant responses
to items designed to personnel decisions, affect, and
expectations. Results indicated that removal of various
items would improve the reliabilities of various scales, but
these results were inconsistent across the different sets of
participant responses.

Scale scores were computed for the five different
scales (promotion, pay raise, merit pay, affect, and
expectations) and intercorrelations were calculated. This
was done three times, once for each set of participant
responses. Similar findings were obtained across all three
sets of participant responses and only the results of the
first set will be presented in Table 16. The correlation
between merit pay and pay raise decisions was .932 after
correction for unreliability, which was consistent with the
results of the factor analyses. The correlations between
affect and pay raise and merit pay decisions were high, but
the results of the factor analyses suggested they were
measuring different constructs. Furthermore, while affect
might be expected to be strongly related to various
personnel decisions, this does not imply that affect and pay
decisions are the same construct.

The factor analyses, reliabilities, and
intercorrelations between the scales support the notion that
the different scales measure conceptually distinct personnel

decisions and responses, with the exception of merit pay and

161
Table 15

Beiiebiliey Analysee for the Personnel Decisions Scales
Eesitiye Promotion Decisions Scale

 

ITEM MEAN STD DEV CASES
1. V28 2.6827 1.1273 208.0
2. V33 2.4375 1.0839 208.0
3. V42 2.5962 1.0404 208.0
4. V46 2.7212 1.0628 208.0
5. V49 2.8125 1.2427 208.0
# 0F CASES = 208.0

# OF
STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
PROMOTE 13.2500 23.0193 4.7978 5

SCALE SCALE CORRECTED

MEAN VARIANCE ITEM- SQUARED ALPHA

IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
ITEM DELETED DELETED CORRLTN CORRLTN DELETED
V28 10.5673 14.8650 .7919 .6404 .8910
V33 10.8125 15.4768 .7467 .5864 .9002
V42 10.6538 15.5318 .7810 .6175 .8938
V46 10.5288 15.1586 .8133 .6768 .8872
V49 10.4375 14.2280 .7730 .6238 .8968

* These are items in which alpha would get higher if they
were removed from the scale.

5 ITEMS
STANDARDIZED ITEM ALPHA = .9146

RELIABILITY COEFFICIENTS
ALPHA = .9132

BQ§1£1¥§1Pav Raise Decisions Scale

 

ITEM MEAN STD DEV CASES
1. V29 3.3125 1.2291 208.0
2. V31 4.2019 .8091 208.0
3. V32 4.2356 .8149 208.0
4. V38 4.2019 .7340 208.0
5. V48 3.2740 1.2106 208.0

# 0F CASES = 208.0

Table 15 (cont'd)

162

 

SCALE SCALE CORRECTED

MEAN VARIANCE ITEM- SQUARED ALPHA

IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
ITEM DELETED DELETED CORRLTN CORRLTN DELETED
V29 15.9135 8.6591 .5271 .4319 .8216*
V31 15.0240 10.0043 .6493 .7336 .7764
V32 14.9904 9.8550 .6770 .7328 .7692
V38 15.0240 10.4390 .6335 .4956 .7848
V48 15.9519 7.8238 .6930 .5652 .7594

* These are items in which alpha would get higher if they
were removed from the scale.

 

 

RELIABILITY COEFFICIENTS 5 ITEMS
ALPHA = .8179 STANDARDIZED ITEM ALPHA = .8415
Positive Merit Pay Decisions Scale

ITEM MEAN STD DEV CASES
1. V30 3.7019 .9568 208.0
2. V36 3.5048 1.1378 208.0
3. V43 3.6731 1.0672 208.0
4. V50 3.7788 1.0809 208.0
# OF CASES = 208.0

# OF

STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
MERIT PAY 14.6587 11.9747 3.4604 4

SCALE SCALE CORRECTED

MEAN VARIANCE ITEM- SQUARED ALPHA

IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
T M DELETED DELETED CORRLTN CORRLTN DELETED
V30 10.9567 8.4087 .4776 .2511 .8575*
V36 11.1538 6.2274 .7841 .6314 .7237
V43 10.9856 6.9998 .6793 .5314 .7760
V50 10.8798 6.8309 .7037 .5112 .7645

* These are items in which alpha would get higher if they
were removed from the scale.

RELIABILITY COEFFICIENTS

ALPHA =

.8303

4 ITEMS

STANDARDIZED ITEM ALPHA =

.8265

163
Table 15 (cont'd)

Pesiiive Expectations for Future Performance Scale

 

ITEM MEAN STD DEV CASES
1. V34 3.4615 1.0302 208.0
2. V39 3.8029 .9297 208.0
3. V41 3.4615 1.0488 208.0
4. V44 3.4087 .9836 208.0
5. V52 3.5577 .9959 208.0
# OF CASES = 208.0
# 0F
STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
EXPECT 17.6923 15.0740 3.8825 5
SCALE SCALE CORRECTED
MEAN VARIANCE ITEM- SQUARED ALPHA
IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM
EM DELETED DELETED CORRLTN CORRLTN DELETED
V34 14.2308 9.9948 .6168 .4961 .8099
V39 13.8894 10.4177 .6318 .4123 .8060
V41 14.2308 9.4151 .7083 .5338 .7832
V44 14.2837 9.8080 .6977 .4899 .7872
vs2 14.1346 10.5712 .5421 .3770 .8297

* These are items in which alpha would get higher if they
were removed from the scale.

 

 

RELIABILITY COEFFICIENTS 5 ITEMS
ALPHA = .8366 STANDARDIZED ITEM ALPHA = .8369
EgéitiyeeAffect Toward for Future Performance Scale

ITEM MEAN STD DEV CASES
1. V35 3.6731 .8840 208.0
2. V37 3.5625 1.1060 208.0
3. V40 3.3942 1.0625 208.0
4. V45 3.7981 1.0061 208.0
5. V47 2.8990 1.1482 208.0
6. V51 3.5962 1.0074 208.0
# 0F CASES = 208.0

# OF
STATISTICS FOR MEAN VARIANCE STD DEV VARIABLES
AFFECT 20.9231 23.9457 4.8934 6

164

Table 15 (cont'd)

SCALE SCALE CORRECTED

MEAN VARIANCE ITEM- SQUARED ALPHA

IF ITEM IF ITEM TOTAL MULTIPLE IF ITEM

DELETED DELETED CORRLTN CORRLTN DELETED
V35 17.2500 18.7585 .5754 .3666 .8704
V37 17.3606 16.4636 .6974 .5161 .8510
V40 17.5288 16.3180 .7571 .5863 .8402
V45 17.1250 17.3370 .6680 .4679 .8559
V47 18.0240 16.2458 .6895 .5002 .8529
V51 17.3269 17.1390 .6944 .4895 .8515

* These are items in which alpha would get higher if they
were removed from the scale.

RELIABILITY COEFFICIENTS 6 ITEMS
ALPHA = .8753 STANDARDIZED ITEM ALPHA = .8753

165

Table 16
Intezeoggeietiens of Personnel Decisions Seaies
PROMOTE = Items V28, V33, V42, V46, V49

(alpha=.913, 5 items)

PAY RAISE= Items V29, V31, V32, V38, V48
(alpha=.818, 5 items)

MERIT PAY= Items V30, V36, V43, V50
(alpha=.830, 5 items)

EXPECT = Items V34, V39, V41, V44, V52
(alpha=.837, 5 items)

AFFECT = Items V35, V37, V40, V45, V47, V51

(alpha=.875, 6 items)

PROMOTE PAY MERIT EXPECT AFFECT
PROMOTE 1.0000 .5605** .5173** .5668** .6107**
PAY .64868 1.0000 .7676** .6463** .7367**
MERIT .59428 .9316a 1.0000 .5715** .7369**
EXPECT .64848 .7811a .68578 1.0000 .6395**
AFFECT .68338 .8708a .86478 .7473a 1.0000
* - Signif. LE .05 ** - Signif. LE .01 (2-tailed)

Note: 5 indicates the correlations reported below the
diagonal are corrected for attenuation due unreliability in

measurement.

 

166
pay raise decisions. The merit pay and pay raise decisions
appear to be tapping into the same construct, as evidenced
by the results of the factor analyses and scale
intercorrelations. Thus, four major personnel decision
scales were retained for further analyses, and these scales
corresponded with the factors obtained in the factor
analyses. Three sets of participant scores were computed
for each of the four personnel decision scales by summing
across items comprising those scales. Each participant had
twelve total composite scores for personnel decisions:
positive promotion decisions (first, second, third target
employee); positive pay decisions (first, second, third
target employee); positive affect (first, second, third
target employee) and positive expectations for future
performance (first, second, third target employee).

The means, standard deviations, and reliabilities of
all scales are presented in Table 17. The reliabilities of
the scales are all very high, with the exception of the
perceived stability of the cause of performance scale, which
has alphas which ranged from .64 to .73.

Since the original data set was configured in the order
Of participant responses, participant scores had to be
reordered to reflect the manipulation of informational cues
to run the MANOVA and ANOVA analyses. Table 18 contains the
descriptions of the variables used in the MANOVA and ANOVA
analyses. These are the variables which were obtained by

reconfiguring the data set based upon informational cues.

167

 

 

 

 

Table 17

Standard Deviations and Reliabilities of All Sc leS

No. of

SCALE MEAN STDDEV ALPHA ITEMS
WAMS 89.58 13.92 .947 21
Se; 1 - First Presented Tarqet Employee
Effort 9.29 3.77 .914 3
Ability 8.92 3.07 .855 3
Easejob 7.47 3.24 .853 3
Control 14.98 3.92 .816 5
ExtLocus 14.57 4.23 .829 5
Stability 15.67 3.09 .636 5
Promotion 13.25 4.80 .913 5
Pay 33.88 6.77 .895 9
Expect 17.69 3.88 .837 5
Affect 20.92 4.89 .875 6
Set 2 - Second Presented Tarqet Employee
Effort 9.31 3.96 .951 3
Ability 8.93 3.34 .875 3
Easejob 7.72 3.46 .918 3
Control 15.50 4.00 .811 5
ExtLocus 14.05 4.61 .882 5
Stability 16.57 3.03 .659 5
Promotion 14.37 5.54 .941 5
Pay 33.43 7.22 .919 9
EXpect 17.98 4.04 .863 5
Affect 21.16 5.11 .898 6
§§t_3 - Third Presented Tarqet Employee
Effort 8.20 3.97 .963 3
Ability 8.33 3.59 .923 3
Easejob 8.70 3.94 .920 3
Control 14.24 4.65 .874 5
ExtLocus 14.82 5.46 .934 5
Stability 16.66 3.40 .731 5
Promotion 13.75 5.73 .940 5
Pay 31.97 9.05 .951 9
EXPect 17.07 4.61 .899 5
Affect 20.31 6.20 .939 6

168

Table 18

Qeeeziptions o; Measuged Variables Used in MANOVA end ANOVA

1. Tgender = Gender of fictional target employee
(0=male, 1=female)

2. Gendprt = Gender of participant providing responses
(0=male, 1=female)

3. Samediff = Ingroup/outgroup gender
(0=different gender, 1=same gender)

4. WAMS = Attitudes toward women as managers
(higher number indicates more positive attitudes)

5. Abilityl = Perceived ability in response to ability
information provided

6. Effortl = Perceived effort in response to ability
information provided

7. Easejobl = Perceived ease of job in response to ability
information provided

8. ExtLocsl = Perceived external locus of causality in
response to ability information provided
9. Controll = Perceived controllability in response to

ability information provided

10. Stablel = Perceived stability in response to ability
information provided

11. Affectl = Positive affect in response to ability
information provided

12. Expectl = Positive expectations for future performance
in response to ability information provided

13. Promote1 = Positive promotion decisions made in
response to ability information provided

14. Payl = Positive pay decisions made in response to
ability information provided

15. Ability2 = Perceived ability in response to effort
information provided

16. Effort2 = Perceived effort in response to effort
information provided

17. Easejob2 = Perceived ease of job in response to effort
information provided

18. ExtLocsZ = Perceived external locus of causality in
response to effort information provided

19. Controlz = Perceived controllability in response to
effort information provided

20. Stablez = Perceived stability in response to effort
information provided

21. Affect2 = Positive affect in response to effort
information provided

22. Expect2 = Positive expectations for future performance
in response to effort information provided

23. Promotez = Positive promotion decisions made in
response to effort information provided

169

Table 18 (cont'd)

24.
25.
26.
27.
28.
29.
30.
31.
32.
33.

34.

Pay2 = Positive pay decisions made in response to
effort information provided
Ability3 = Perceived ability in response to ease of job

information provided
Effort3 = Perceived effort in response to ease of job

information provided

Easejob3 = Perceived ease of job in response to ease of
job information provided

ExtLocs3 = Perceived external locus of causality in
response to ease of job information provided

Control3 = Perceived controllability in response to
ease of job information provided

Stable3 = Perceived stability in response to ease of
job information provided

Affect3 = Positive affect in response to ease of job
information provided

Expect3 = Positive expectations for future performance
in response to ease of job information provided
Promote3 = Positive promotion decisions made in
response to ease of job information provided

Pay3 = Positive pay decisions made in response to ease

of job information provided

170
Table 19 contains the means, standard deviations, and
intercorrelations of all measured variables used in those
analyses. The sample size for each correlation ranged from
252 to 256.

Many of the correlations were significant, but they
carry little interpretive substance without further analyses
because many are the result of the manipulations provided in
the study. For example, the fact that Effortl is negatively
correlated with Effort2 (—.68) and positively correlated
with Effort3 (.68) is a product of the informational cues
provided in the employee files (see Table 1).

Also, the correlations between the personnel decisions
and attributions, and between perceived causes and
attributions are uninterpretable without further analyses
because they too are the product of manipulations.
Furthermore, the effects of hypothesized interactions are
masked when viewing the zero-order correlations.

For example, the correlation between Effortl and
Controll was -.17. This correlation suggests the more
effort the employee was perceived as exerting, the less
controllable the cause of observed performance. This
finding would be puzzling, but when one realizes that both,
high AND low effort can be related to perceived
controllability, simple explanations become possible. For
example, a group of participants might view low effort which
leads to declining performance as highly controllable. In

this case, the correlation between effort and perceived

171
Table 19
neene, §tandazd Peviations and integcorrelations of Ail
fleeegged VariableeL Configured by Informational Cuee

Previded

 

Scale Mean Sthev i g g A
1. Tgender .504 .501 -
2. Gendprt .551 .498 .05 —
3. Samediff .523 .500 .10 .00 -
4. WAMS 89.575 13.918 .05 .56“ .00 .-
5. Abilityl 10.160 3.410 .04 .04 -.04 .13“
6. Effortl 8.180 3.215 —.02 -.02 .00 -.03
7. Easejobl 7.379 2.867 .01 -.09 .03 -.23“
8. ExtLocsl 12.867 3.886 -.03 -.08 .01 -.19“
9. Controll 14.867 3.738 -.04 .03 -.09 .04
10. Stablel 16.807 3.226 .03 .09 .01 .01
11. Affectl 21.451 4.421 -.07 .00 -.01 .21“
12. Expectl 17.808 3.992 .04 .05 -.05 .31“
13. Promote1 14.475 5.031 -.03 -.05 -.08 .06
14. Pay1 34.180 6.510 -.02 .06 -.00 .34“
15. Ability2 8.119 2.643 -.07 .01 .02 -.10
16. Effort2 10.555 3.981 -.02 .02 -.05 .14“
17. Easejobz 7.098 3.163 .01 -.07 .05 - 23“
18. ExtLocs2 12.718 3.981 -.14“ .05 .00 -.15“
19. Controlz 17.320 3.605 .13“ .12 -.06 .19“
20. Stable2 15.824 3.090 .08 .09 .01 .21“
21. Affectz 21.110 5.305 -.04 -.01 .05 .12a
22. Expectz 17.922 4.086 -.06 .02 -.05 .17“
23. Promote2 14.086 5.128 -.08 -.09 .02 .03
24. Pay2 34.008 7.841 -.04 -.00 .00 .20“
25. Ability3 8.028 3.576 .05 —.01 -.03 -.12
26. Effort3 8.000 3.919 .02 .01 .03 -.08
27. Easejob3 9.632 4.095 -.02 -.04 -.00 .02
28. ExtLocs3 18.020 4.388 .09 .05 -.03 .13“
29. Control3 12.439 3.953 -.06 -.00 -.04 -.07
30. Stable3 16.094 3.403 .03 -.08 -.02 -.03
31. Affect3 19.606 6.185 .01 -.03 -.01 .01
32. Expect3 16.742 4.443 .09 .05 -.12 .08
33. Promote3 12.723 5.809 .02 -.07 -.04 -.09
34. Pay3 30.757 8.370 .13“ -.01 .04 .04
Note: a indicates p S .05
Note: b indicates p S .01
Note: n = 252 to 256

Table 19 (cont'd)

' indicates p
b indicates p s .01

Scale
1. Tgender
2. Gendprt
3. Samediff
4. WAMS
5. Abilityl
6. Effortl
7. Easejobl
8. ExtLocsl
9. Controll
10. Stablel
11. Affectl
12. Expectl
13. Promote1
14. Pay1
15. Ability2
16. Effort2
17. Easejobz
18. ExtLocsZ
19. Controlz
20. Stable2
21. Affect2
22. Expeth
23. Promote2
24. Pay2
25. Ability3
26. Effort3
27. Easejob3
28. ExtLocs3
29. Contr013
30. Stable3
31. Affect3
32. Expect3
33. Promote3
34. Pay3
Note:
Note:
Note: 3 =

5

-.57“

la

.63“
.53“

S .05

252 to 256

172

N

§ 2
-.22“ .—
—.35“ -.20“
-.22 .07
-.33“ .25“
-.32“ .22“
-.25 .15“

.23“ -.11
-.34“ .29“
.39“ -.24“
.50“ -.01
-.19“ -.13“
-.05 .01
-.24“ .24“
-.29“ .25“
-—.15“ .16“
- 24“ .23“
.35“ -.30“
.31 - 24“
- 32“ .33“
- 38“ .09

.27“ .08
-.04 .07

.25“ -.24“

.17“ -.16“

.37“ -.30“

.27“ -.2o“

-.19“
-.06
-.07
.02
-.06
-.03

-.19“
-.11
-.02

.18“
.09

.15a

-.12a
.03
.02
.02

-.16a
.04

Table 19 (cont'd)

WGQO‘UIPUNH

NIOAJNrJh‘HPHP‘HFJF“HfH
h3NF4C>©(D\Jm(fl£waOF‘O
O O O O O O O O O O O O O O

25.
26.
27.
28.
29.
30.
31.
32.
33.
34.

Note:
Note:
Note:

Scele
Tgender

Gendprt
Samediff
WAMS
Abilityl
Effortl
Easejobl
ExtLocsl
Controll
Stablel
Affectl
Expectl
Promote1
Pay1
Ability2
Effort2
Easejob2
ExtLocsZ
Controlz
Stable2
Affect2
Expectz
Promote2
Pay2
Ability3
Effort3
Easejob3
ExtLocs3
Control3
Stable3
Affect3
Expect3
Promote3
Pay3

a
b
g—

12

-.04
-.13“
-.36“
-.16“
-.36“
-.28“

indicates 9 s .05
indicates p S .01
- 252 to 256

173

.37“
-.O6
-.15“
-.65“
-.60“
-.51“
-.67“

.74“

.72“
-.72“
-.37“

.14“

.04

.62“

.41“

.68“

.55“

-.49“
-.05
-.25“
-.26“
-.17“
-.24“
.30“
.32“
-.27“
-.40“
.31“
-.16“
.16“
.11
.25“
.14“

Table 19 (cont'd)

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.

Note:
Note:
Note:

Scale
Tgender
Gendprt
Samediff
WAMS
Abilityl
Effortl
Easejobl
ExtLocsl
Controll
Stablel
Affectl
Expectl
Promote1
Pay1
Ability2
Effort2
Easejob2
ExtLocsz
Contro12
Stable2
Affect2
Expect2
Promote2
Pay2
Ability3
Effort3
Easejob3
ExtLocs3
Control3
Stable3
Affect3
Expect3
Promote3
Pay3

12

.04

eg

' indicates p s .05
b indicates p s .01

fl = 252 to 256

174

I
ox
Ch

0'

.64“
.76“
-.59“
-.59“
.61“
.19“
-.07
.04
-.50“
-.25“
-.58“
-.49“

.61“
-.47“
-.49“

.53“

.11

.01
-.06
-.45“
-.36“
-.43“
-.41“

24 e5
-.67“ .-
-.67“ .91“

.69“ -.89“

.25“ -.36“
-.04 .17“

.02 .09
-.52“ .77“
-.34“ .59“
-.63“ .87“
-.46“ .72“

Table 19 (cont'd)

‘OQQGUIhUNH

Note:
Note:
Note:

Seals
Tgender
Gendprt
Samediff
WAMS
Abilityl
Effortl
Easejobl
ExtLocsl
Controll
Stablel
Affectl
Expectl
Promote1
Pay1
Ability2
Effort2
Easejob2
ExtLocs2
ControlZ
Stable2
Affect2
Expectz
Promote2
Pay2
Ability3
Effort3
Easejob3
ExtLocs3
Control3
Stable3
Affect3
Expect3
Promote3
Pay3

25 21

-.89“ .-

-.39“ .35“
.17“ -.13“
.00 -.04
.78“ -.78“
.59“ -.57“
.84“ -.83“
.73“ -.72“

° indicates p s .05
indicates p s .01
n = 252 to 256

175

28 22
-.63“ .-

05 -.07
- 25“ .02
—.20“ .04
-.37“ 15“
- 27“ 07

22 31
.08 .-

.19“ .72“
.09 .77“
.10 .86“

176
Table 19 (cont'd)

$221.9 22323.4
1. Tgender
2. Gendprt
3. Samediff
4. WAMS
5. Abilityl
6. Effortl
7. Easejobl
8. ExtLocsl
9. Controll
10. Stablel
11. Affectl
12. Expectl
13. Promote1
14. Pay1
15. Ability2
16. Effort2
17. Easejob2
18. ExtLocs2
19. Controlz
20. Stable2
21. Affect2
22. Expect2
23. Promote2
24. Pay2
25. Ability3
26. Effort3
27. Easejob3
28. ExtLocs3
29. Contr013
30. Stable3
31. Affect3
32. Expect3 .-
33. Promote3 .58b .-
34. Pay3 .69“ .73“ .-

Note: 3 indicates p S .05
Note: b indicates p s .01
Note: 3 = 252 to 256

177
controllability would be negative. Another group of
participants might view high effort which leads to improving
performance as highly controllable. In this case, the
correlation would be positive. Thus, the obtained
correlations are not interpretable without consideration of
the independent variables, interactions, and specific
hypotheses. They are merely presented here for
completeness.

Intercorrelations between variables were also computed
across the entire set of observations, without regard to the
specific informational condition provided. This correlation
matrix was based on a sample size of 768 (256 * 3), the
total number of observations available in the study. This
is the matrix used to conduct the repeated measures
regression analyses. This intercorrelation matrix is
presented in Table 20.

Manipulation Checks

A variety of items and methods were used to ensure the
manipulations were effective in achieving their desired
outcomes. Manipulation checks included order of
presentation, perceived performance levels, and perceived
employee gender. A MANOVA using the perceived causes of
observed performance as dependent variables was also
conducted.

As noted earlier, the order of the presentation of
employee files was counterbalanced. Manipulation checks

indicated that this procedure was properly conducted. For

Table 20

e

omqmmbuwp

HHH
MPG
00.

13.

p.-
b
O

mmflathNH
O.

10.
11.
12.
13.
14.

Note:
Note:
Note:

nda

178

'a i ns and nter orrelations of l

neeeuzeg Veziaeies Pesed on Entire Set of Observetiegs

 

Mean Sthev

Seaie
Tgender .504
Gendprt .551
Samediff .523
WAMS 89.575
Ability 8.773
Effort 8.914
Easejob 8.030
ExtLocs 14.539
Control 14.869
Stable 16.241
Affect 20.720
Expect 17.490
Promote 13.764
Pay 32.983
Scale 5
Tgender
Gendprt
Samediff
WAMS
Ability .-
Effort -.02
Easejob -.49b
ExtLocs -.23b
Control .15b
Stable .11b
Affect .33“
Expect .35b
Promote .51b
Pay .32b
° indicates p S
b indicates p S
g = 756 to 768

.500
.498
.500
13.899
3.379
3.894
3.591
4.774
4.258
3.265
5.407
4.206
5.378
7.764

6

Ix!

;

.05
.10“
.05
.01
-.00
.00
-.02
.01
.05
-.03
.03
-.03
.03

[CD

-.56“
-.11“
-.28“
-.29“
-.32“
-.32“

2 3 4
.00 .-
.56“ .00 .-
.01 -.02 -.02
.01 -.01 .01
-.06 .02 -.12“
.01 -.01 -.05
.04 -.06 .04
.03 -.00 .09“
-.01 .01 .10“
.04 -.07“ .18“
-.07 -.03 -.01
.01 .01 .18“
2 12
-.11“ -
.07 .15“
.15“ .21“
.17“ .17“
.14“ .16“

Table 20 (cont'd)

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.

Note:
Note:
Note:

Scale
Tgender

Gendprt
Samediff

WAMS

Ability

Effort

Easejob
ExtLocs
Control

Stable

Affect .-
Expect .71”
Promote .66”
Pay .82“

11

a
b
a =

indicates p 5
indicates p s .01
756 to 768

179

180
example, ability informational cues were presented first 33%
of the time, second 35% of the time, and third 32% of the
time. There were some minor discrepancies from the ideal 33
1/3% for each condition, but this was mostly due to missing
data on the order of presentation, not because the order of
presentation was not properly counterbalanced.

Participants were randomly assigned to performance trend
and employee gender conditions. For example, each type of
performance trend was presented so that approximately equal
numbers of participants were presented with each type of
performance trend (34.3%, 32.4%, 33.2% for improving,
maintaining, and declining trends, respectively).

A manipulation Check was used to ensure that
participants correctly identified the gender of target
employees. Every participant (100%) correctly identified
the gender of the target employees presented in the
fictional files. This question was asked very near the end
of all questions to avoid biasing participant responses.

The gender of male and female target employees were
equally distributed in presentation across participants (127
or 49.6% received male target employees, and 129 or 50.4%
received female target employees). The gender of the target
employee was the same as the gender of the participant 48%
of the time, while gender was different between the target
employee and participant 52% of the time. AS noted earlier,
the gender of participants included 45% males and 55%

females.

181

Several items were used to determine if participants
correctly perceived that all employees in the files
presented were above average in performance. The first
question asked participants about the average profit level
of the employee presented in the file. Ninety-two percent
of participants correctly answered this question. The next
question asked participants what the average profit level of
all stores in the region was. Ninety-four percent of
participants correctly answered this question. The third
question asked participants whether the employee presented
in the file was above or below the regional average in
profit. Ninety-one percent of the participants correctly
answered this third question. ANOVAs and MANOVAs were
conducted on the entire sample and then again only for those
who answered the third question correctly. The results of
these analyses were then compared. Few differences existed
between both sets of analyses. Due to space considerations,
only the results of analyses involving the entire sample
will be reported.

A doubly-multivariate MANOVA was conducted on the
perceived causes of observed performance, with performance
trend and informational cues as the independent variables.
The results are presented in Table 21. The MANOVA indicated
that performance trend, informational cues, and the trend X
information interaction all significantly affected perceived
causes of performance. This is exactly what would be

expected based on the research design. The means of

Table 21

182

MAHQXA for the Pegceived Causes of Obeerved Performance

Effect g;
Constant 3
Main Effects

Performance Trend (T) 6
Informational Cues (I) 6
Interactions

T X I 12
Note:

 

All P tests are reported using Wilk's Lambda

Error
Q: E
241 15361.36
482 26.09
238 31.87
476 57.50

Q-Vfilug
.000

.000

.000

.000

183

perceived high or low ability, effort, and ease of job for
the observed performances are presented in Table 22. The
means correspond to the manipulations provided in the
fictional employee files and the significant MANOVA for main
effects and interactions indicate the manipulations were
successful. Further analyses of manipulation checks using
ANOVAs or specific cell comparisons were deemed unnecessary.

In summary, the manipulation checks indicate that the
order of employee file presentation did not represent a
threat to the research design. They also Show that most of
the participants correctly perceived the target employees as
performing above average in performance. When participants
who misperceived the level of target employee performance
were removed from analyses, only minor differences in the
results were obtained. The manipulation checks show that
approximately even numbers of male and female target
employees were presented, and about half of those presented
were of the same gender as the participant. All of the
participants correctly perceived the gender of the target
employee. Finally, manipulation checks indicate that the
fictional employee files effectively manipulated the
perceived causes of observed performance.

Effects of Performance Trend. Informational Cues. and
Empioyee Gender on Personnel Decisions

 

A doubly multivariate repeated measures MANOVA was
performed with performance trend, informational cues, and

gender of the target employee as independent variables, and

Table 22

geli Meegs of Perceived Ability, Effort, or Ease 0; Job

Selig py Performance Trend and Informational Cues

INFO TYPE

Abilifv Info C229
Perceived ability
Perceived effort

Perceived easejob

Effort Info Cues
Perceived ability

Perceived effort

Perceived easejob

EaseJob Info Cues
Perceived ability

Perceived effort
Perceived easejob
AVGS ACROSS

INFO CELLS
Perceived effort

Perceived ability

Perceived easejob

Note: Numbers in parentheses indicate number of

PERFORMANCE TRENDS

Imp rov

11.898

(88)

6.580
(88)

6.136
(88)

7.200
(85)

13.080

(88)
5.375
(88)
5.837
(86)

5.448
(87)

12.129

(85)

8.380
(263)

8.344
(259)

7.831
(261)

Mntain

12.072

(83)

6.639
(83)

6.349
(83)

7.145
(83)

12.651

(83)
5.482
(83)
5.940
(83)

5.759
(83)

12.169

(83)

8.349
(249)

8.386
(249)

8.000
(249)

Declin

6.494
(85)

11.341
(85)

9.671
(85)
9.988
(85)

5.894
(85)

10.459
(85)
12.282
(85)

12.857
(84)

4.659
(85)

10.020
(254)

9.588
(255)

8.263
(255)

AVGS ACROSS
TEEEQ_Q§LL§

10.160

(256)

8.180
(256)

7.379
(256)

8.119
(253)

10.555

(256)
7.098
(256)
8.028
(254)

8.000
(254)

9.632
(253)

OVERALL
GRAND MEANS

8.914
(766)

8.773
(763)

8.030
(765)

observations

185
with promotion, pay, affect, and expectations as dependent
variables. This analysis was performed to test the
hypothesis that performance trends, informational cues, and
employee gender would affect overall personnel decisions.
The results of this MANOVA are presented in Table 23.

The results indicate that performance trends and
informational cues significantly affected personnel
decisions. These main effects were qualified by significant
effects for the performance trend X informational cues and
the informational cues X employee gender interactions.

As noted earlier, interpretational problems with MANOVA
can occur when evaluating univariate effects using
multivariate procedures. For this reason, separate repeated
measures univariate ANOVAs were conducted for each of the
four dependent variables (promotion, pay, affect, and
expectations). These analyses provided a more specific
explanation of the manner in which different dependent
variables were affected by the independent variables in the
MANOVA.

The first ANOVA evaluated the effects of performance
trend, informational cues, and target employee gender on
promotion decisions. The results of this analysis are
presented in Table 24, while the cell means are presented in
Table 25.

The results of the ANOVA indicate that informational
cues had an effect on promotion decisions, but this was

qualified by the significant informational cues X

186

Table 23

Bepeeted Measures MANOVA for the Overall Effect of

0..

en Personnel Decisions

Effect df
Constant 4

Between-Subjects Main Effects
Performance Trend (T) 8

Employee Gender (EG) 4

Between-Subjects Interactions

EG X T 8

Within-Subjects Main Effects

Informational Cues (I) 8

Within-Subjects Interactions

T X I 16
EC X I 8
EG X T X I 16

Note: All P tests are reported using Wilk's Lambda

0182! ‘ - ‘11 O_ .tional Cues

Error
e;

239

478

239

478

235

470
235

470

and Em. .

I“?!

5685.20

.46

.000

.120

.884

.000

.000

O 046

.430

187

Table 24

Bepeeted Measures ANOVA for the Effects of Performance

Irene, Informational Cues, and Employee Gender on Promotien

9.2212989.

Egfiect e:
Between-Subjects Effects
Employee Gender (EG) 1
Performance Trend (T) 2
EG X T 2
Error 246

Within-Subjects Effects

Informational Cues (I) 2
T X I 4
EG X I 2
EG X T X I 4
Error 492

Note: All P tests for within-subjects effects are

22.

19.

18.

194

2157.

29.

7.

16

using averaged tests of significance

03

68

.84

55

.76

94

59

26

.12

I'll

12.08

133.84

.45

p-value

.277
.348

.813

.000
O 000
.161

.772

reported

188
Table 25

gel; Means of Promotion Decisions, Split by Perfopnance
Ipeng. Informational Cues, and Employee Gender

PERFORMANCE TRENDS
AVGS ACROSS

INFO TYPE Improv nnpnin Declin TREND CEL 8
apiiipy Info gues 15.76 16.43 11.19 14.47
(88) (83) (84) (255)
5.20 4.63 3.39 5.03
Male Employees 15.50 16.83 11.67 14.62
(44) (40) (43) (127)
4.84 4.66 3.06 4.76
Female Employees 16.02 16.07 10.68 14.32
(44) (43) (41) (128)
5.57 4.62 3.68 5.30
ngort Info Cues 15.78 16.43 10.04 14.08
(88) (83) (85) (256)
4.31 4.83 3.58 5.13
Male Employees 16.18 16.83 10.65 14.51
(44) (40) (43) (127)
4.72 4.74 3.43 5.12
Female Employees 15.39 16.07 9.40 13.67
(44) (43) (42) (129)
3.86 4.93 3.66 5.12
Paseaob Info Cues 9.62 9.36 19.34 12.72
(87) (83) (83) (253)
3.25 3.36 3.93 5.81
Male Employees 9.43 9.45 18.86 12.58
(44) (40) (42) (126)
2.80 3.26 4.39 5.67
Female Employees 9.81 9.28 19.83 12.87
(43) (43) (41) (127)
3.68 3.49 3.37 5.96

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

Table 25 (cont'd)

AVGS ACROSS OVERALL
INZQ QPLLS Improv Mntain Peclin QPPPQ_MPAP§
AVGS ACROSS 13.74 14.08 13.48 13.76
GENDER CELLS (263) (249) (252) (764)
5.20 5.45 5.50 5.38
Male Employees 13.70 14.37 13.69 13.91
(132) (120) (128) (380)
5.18 5.49 5.15 5.27
Female Employees 13.77 13.81 13.27 13.62
(131) (129) (124) (384)
5.23 5.41 5.85 5.49

189

PERFORMANCE TRENDS

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

190
performance trend interaction. No other effects were
significant. The findings indicate performance trend did
not have a significant main effect on promotion decisions,
though means were in the hypothesized directions; thus, they
do not provide support for hypotheses H1, H2, and H6. The
results of the ANOVA suggest there were no effects on
promotion decisions for employee gender nor for the employee
gender X performance trend interaction, and they do not
provide support for hypotheses Hll-Hllb.

The significant performance trend X informational cues
interaction provides indirect support for hypotheses H3 to
H4b. Employees who showed improving or maintaining
performance trends which were due to high ability or high
effort were rewarded significantly more than employees who
showed improving or maintaining performance trends which
were due to ease of job. Employees who showed declining
performance trends which were due to low effort were
rewarded significantly less than employees who showed
declining performance trends which were due to low ability,
who were in turn rewarded significantly less than employees
who showed declining performance trends which were due to
non-ease of job. If effort and ability are viewed as more
controllable and internal than ease of job, then hypotheses
H3 to H4b will have been supported. The performance trend X
informational cues interaction effect for promotion

decisions is graphically presented in Figure 15.

191

Promotion Decisions

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20
[/5
16 ;;, .
12
*_

8 ____ -

4 1.. -1 M.“ .1 __I

0 t + 4

Improving Maintaining Declining

Performance Trend

"'— Effort —+— Ability + Ease of Job

Eigu:e_1§. Performance Trend X Informational Cues

Interaction Effect for Promotion Decisions

192

The next repeated measures ANOVA evaluated the effects
of performance trend, informational cues, and target
employee gender on pay decisions. The results of this
analysis are presented in Table 26, while the cell means are
presented in Table 27.

The results of the ANOVA for pay indicate that
informational cues had a significant main effect on pay
decisions. This main effect was qualified by two
significant interaction effects, which included the trend X
informational cues and employee gender X informational cues
interactions. While the significant gender X informational
cues interaction does not provide support for hypotheses
Hll-Hllb, it does suggest that employee gender affects how
decision makers use performance related information.

The employee gender X informational cues interaction
appears to be primarily due to the effect of the ease of job
informational cues and gender on pay decisions. There were
no significant differences in pay reward for male and female
employees in the ability and effort informational cue
conditions, though males tended to be rewarded slightly more
than females in these conditions. There was a significant
difference in pay reward in the ease of job informational
cues condition such that females were rewarded significantly
more than males when above average performance was due to
the ease of job. Figure 16 graphically depicts the employee
gender X informational cues interaction effect on pay

decisions.

193

Table 26

{-9-. ‘9 4e._u -s 4.0V) 0 t - _ e ts 0 re 01H: -
Irene, Infiormational Cues, and Employee Gender on Pay
12921512119

Effect Qf MP P p-value
Between-Subjects Effects

Employee Gender (EG) 1 34.04 .68 .410
Performance Trend (T) 2 12.46 .25 .779
EC X T 2 16.37 .33 .721
Error 249 49.90

Within-Subjects Effects

Informational Cues (I) 2 953.52 30.26 .000
T X I 4 3922.72 124.50 .000
EC X I 2 165.79 5.26 .005
EC X T X I 4 8.85 .28 .890
Error 498 31.51

Note: All P tests for within-subjects effects are reported

using averaged tests of significance

194

Table 27

gel; Means of Pay Pecisions, Split by Performance Trend,
Internatienal Cues, and Employee Gender

PERFORMANCE TRENDS
AVGS ACROSS

INFO TYPE Improv Mntain Declin TREND EL
npiiipy Inge Cues 34.83 34.80 32.91 34.18
(88) (83) (85) (256)
6.67 6.78 5.93 6.51
Male Employees 35.18 34.53 33.26 34.32
(44) (40) (43) (127)
5.55 5.77 5.39 5.58
Female Employees 34.48 35.05 32.55 34.04
(44) (43) (42) (129)
7.68 7.66 6.50 7.33
Effort Info Cues 37.75 37.77 26.51 34.01
(87) (83) (85) (255)
4.66 5.65 6.86 7.84
Male Employees 37.59 38.18 27.35 34.31
(44) (40) (43) (127)
4.66 4.95 6.27 7.28
Female Employees 37.91 37.40 25.64 33.71
(43) (43) (42) (128)
4.71 6.27 7.39 8.37
Ensegop Info Cues 26.90 26.55 38.81 30.76
(87) (83) (85) (255)
7.35 6.28 4.40 8.37
Male Employees 25.66 25.33 37.86 29.69
(44) (40) (43) (127)
6.64 6.27 4.51 8.28
Female Employees 28.16 27.70 39.79 31.82
(43) (43) (42) (128)
7.89 6.15 4.11 8.36

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

195
Table 27 (cont'd)

PERFORMANCE TRENDS
AVGS ACROSS

OVERALL
IMPO CELLS Improv Mntain Declin GRAND MEANS
AVGS ACROSS 33.16 33.04 32.74 32.98
GENDER CELLS (262) (249) (255) (766)
7.80 7.84 7.68 7.76
Male Employees 32.81 32.68 32.82 32.77
(132) (120) (129) (381)
7.65 7.83 6.91 7.45
Female Employees 33.52 33.38 32.66 33.19
(130) (129) (126) (385)
7.97 7.86 8.42 8.07

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

196

Pay Decisions
40

 

36 .-- Isl—"HM-..“ .--_---..

 

 

32 1..-”-.. ,2

 

28 ,m_21,...._.~_1- _

 

24 ~- ~~~~~

 

 

 

 

20 i ‘r 4:
Effort Ability Ease of Job

Employee Gender

 

—*— Male + Female

£1gn:e_1§. Employee Gender X Informational Cues Interaction

Effect for Pay Decisions

197

Again, the findings do not provide support for
hypotheses H1 and H2, because performance trend did not have
a significant main effect on pay decisions, but again, means
were all in the hypothesized directions. It is useful to
note that the overall MANOVA indicated that performance
trend did have a significant main effect on personnel
decisions.

As before, the significant performance trend X
informational cues interaction provides indirect support for
hypotheses H3 to H4b. Employees who showed improving or
maintaining performance trends which were due to high effort
were rewarded significantly more than employees who showed
improving or maintaining performance trends which were due
to high ability, who in turn were rewarded significantly
more than employees who showed improving or maintaining
performance trends which were due to ease of job. Employees
who showed declining performance which was due to low effort
were rewarded significantly less than employees who showed
declining trends which were due to low ability, who in turn
were rewarded less than employees who showed declining
performance trends which were due to non-ease of job. The
performance trend X informational cues interaction effect
for pay decisions is graphically presented in Figure 17.
These results exactly reflect hypothesized effects of locus
and controllability on personnel decisions.

A repeated measures ANOVA was performed to analyze the

effects of performance trends, informational cues, and

198

Pay Decisions

 

 

 

 

 

 

 

 

 

 

 

 

4O
36
32 “Nimble,“ _-- - .- - ., . ...__
28 ___I_____. -1”-.- .-.__I
HF

24 ~ - - 5 ~ 5
20 i t i

Improving Maintaining Declining

Performance Trend

“"‘— Effort *‘i— Ability + Ease of Job

Eiggze_11. Performance Trend X Informational Cues

Interaction Effect for Pay Decisions

199
employee gender on expectations for future performance. The
results of this analysis are presented in Table 28, while
the cell means are presented in Table 29.

The results of the ANOVA for expectations indicate that
performance trend and informational cues each had a
significant effect on expectations for future performance.
These main effects were qualified by the trend X
informational cue interaction, and to a certain degree, by
the informational cue X employee gender interaction (p S
.07) .

The results show expectations for future performance
were significantly higher for employees who showed improving
and maintaining trends than they were for employees who
showed declining trends, but the difference between
expectations for employees who showed improving and
maintaining performance trends was not significant. These
results provide indirect support for H1 and H2. It should
be noted however, that differences in expectations for
future performance did not translate into a significant main
effect of performance trend on pay and promotion decisions.

The significant performance trend X informational cues
interaction provides indirect support for hypotheses H3 to
H4b. Employees who showed improving performance which was
due to effort or ability were expected to do better than
employees who showed improving performance which was due to
ease of job. Employees who showed maintaining performance

which was due to effort were expected to do better than

200
Table 28
;-,-. ‘9 ‘as ‘ ;.0V; 0, t - _, ‘ ts O_ " fog": '
Itgng, Informational Cues, and Employee Gender on

Expectations for Euture Performance

Effect g: fig E p-value
Between-Subjects Effects

Employee Gender (EG) 1 5.50 .33 .567
Performance Trend (T) 2 133.79 7.98 .000
EC X T 2 13.83 .83 .439
Error 249 16.76

Within-Subjects Effects

Informational Cues (I) 2 98.19 9.18 .000
T X I 4 857.93 80.23 .000
EG X I 2 29.19 2.73 .066
EC X T X I 4 7.65 .72 .582

Error 498 10.69

Note: All E tests for within-subjects effects are reported

using averaged tests of significance

201
Table 29

gg11_ngap§_of Expectationsl Split by Performance Trendl
Ipfprmational Cues, and Employee Gender

PERFORMANCE TRENDS
AVGS ACROSS

INFO TYPE IEEIQX E23213 222112 IB§§Q_Q§LL§
hpility Info Cues 18.86 19.08 15.44 17.81
(88) (83) (84) (255)
4.01 3.35 3.51 3.99
Male Employees 18.39 18.90 15.62 17.62
(44) (40) (42) (126)
4.00 3.22 3.35 3.81
Female Employees 19.34 19.26 15.26 17.98
(44) (43) (42) (129)
4.01 3.49 3.70 4.17
Effort Info Cues 19.33 19.90 14.53 17.92
(88) (83) (85) (256)
3.02 3.24 3.66 4.09
Male Employees 19.43 20.30 14.91 18.17
(44) (40) (43) (127)
3.35 2.99 3.36 4.00
Female Employees 19.23 19.53 14.14 17.67
(44) (43) (42) (129)
2.68 3.45 3.95 4.17
EaseJob Info Cues 14.66 15.49 20.12 16.74
(88) (83) (85) (256)
4.29 3.78 3.05 4.44
Male Employees 14.02 15.65 19.40 16.35
(44) <40) (43) (127)
4.08 3.79 3.20 4.33
Female Employees 15.30 15.35 20.86 17.12
(44) (43) (42) (129)
4.44 3.80 2.74 4.53

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

Table 29 (cont'd)
AVGS ACROSS

1329 CELLS

AVGS ACROSS
GENDER CELLS

Male Employees

Female Employees

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

PERFORMANCE TRENDS

Improv

17.62
(264)
4.34

17.28
(132)
4.47

17.95
(132)
4.20

Mntain

18.16
(249)
3.94

18.28
(120)
3.86

18.05
(129)
4.04

Declin

16.70
(254)
4.20

16.65
(128)
3.83

16.75
(126)
4.56

OVERALL
GRAND fiEAES

17.49
(767)
4.21

17.38
(380)
4.11

17.59
(387)
4.30

203
employees who showed maintaining performance which was due
to ability, who were expected to do better than employees
who showed maintaining performance which was due to the ease
of job. Employees who showed declining performance which
was due to effort or ability were expected to do worse than
employees who showed declining performance which was due to
ease of job. The performance trend X informational cues
interaction effect for expectations of future performance is
graphically presented in Figure 18.

The findings did not support hypotheses Hll-Hllb, but
they do provide support for the idea that employee gender
affects future expectations for performance, based on the
performance related information provided. The employee
gender X informational cues interaction appears to be
primarily due to the effect of the ease of job informational
cues and gender on expectations for future performance.
There were no significant differences in expectations for
male and female employees in the ability and effort
informational cue conditions, but there was a significant
difference of expectations in the ease of job informational
cues condition such that females were expected to do better
in the future than were males, particularly in the declining
performance condition. Figure 19 graphically depicts the
employee gender X informational cues interaction effect on
expectations for future performance.

A final repeated measures ANOVA was conducted for the

effects of performance trend, employee gender and

204

Performance Expectations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

20
15
10 ~—
5 --_111_flm-m111-111
O l l l
Improving Maintaining Declining
Performance Trend
—'— Effort —+— Ability —*— Ease of Job
figure_1§. Performance Trend x Informational Cues

Interaction Effect for Expectations

 

205

Performance Expectations

 

20

 

 

4.\

 

16

 

14

 

 

 

 

10 f .L l
Effort Ability Ease of Job

Employee Gender

—-'— Male + Female

Eigptg_12. Employee Gender X Informational Cues Interaction

Effect for Expectations

 

206

informational cues on positive affect toward the employee.
The findings replicate some of the findings of the three
previous ANOVAs. The results are presented in Table 30,
while the cell means are presented in Table 31.

Informational cues had a main effect on affect toward
the employee, but this was qualified by the performance
trend X informational cues interaction. Decision makers had
significantly more positive affect for employees who showed
improving or maintaining performance which was due to effort
than for employees who showed improving or maintaining
performance which was due to ability. Decision makers also
had significantly more positive affect for employees who
showed improving or maintaining performance which was due to
ability than for employees who showed improving or
maintaining performance which was due to ease of job.
Decision makers had significantly less positive affect for
employees who showed declining performance which was due to
lack of effort than for employees who showed declining
performance which was due to lack of ability. Decision
makers also had significantly less positive affect for
employees who showed declining performance which was due to
lack of ability than for employee who showed declining
performance which was due to the non-ease of the job. These
results are graphically depicted in Figure 20.

The results indirectly support hypotheses H3-H4b, but
they do not provide support for hypotheses Hl-Hz since no

main effect for performance trend was found. The results

207

Table 30

Bepgatgg Measures ANQVA tor the Effects of Performance
ngpg, Informational Cues, and Employee Gender on Affect
onapg the Employee

EiIQQE Q: MS 2 -va e
Between-Subjects Effects

Employee Gender (EG) 1 18.55 .93 .336
Performance Trend (T) 2 36.65 1.83 .162
E6 X T 2 7.93 .40 .673
Error 247 19.98

Within-Subjects Effects

Informational Cues (I) 2 221.46 14.65 .000
T X I 4 2286.09 151.19 .000
EC X I 2 12.94 .86 .425
E6 X T X I 4 20.62 1.36 .245

Error 494 15.12

Note: All E tests for within-subjects effects are reported

using averaged tests of significance

208

Table 31

Cell Heaps of Affect, Split by Performance Trendr

Infpzmatippal CuesI and Employee Gender
PERFORMANCE TRENDS

 

AVGS ACROSS

INFO TYPE Improv Mntain Declin TREND CELLS
ppitity Info Cues 21.63 21.83 20.89 21.45
(88) (83) (84) (255)
5.10 4.10 3.93 4.42
Male Employees 21.98 21.80 21.47 21.75
(44) (40) (42) (126)
4.67 4.10 4.23 4.32
Female Employees 21.27 21.86 20.31 21.16
(44) (43) (42) (129)
5.53 4.14 3.56 4.52
Ettort Info Cues 23.32 24.17 15.80 21.11
(87) (83) (84) (254)
3.37 3.23 4.55 5.31
Male Employees 23.41 24.38 16.30 21.31
(44) (40) (43) (127)
3.46 2.90 4.95 5.28
Female Employees 23.23 23.98 15.27 20.91
(43) (43) (41) (127)
3.31 3.54 4.09 5.34
EspeJob Info Cues 15.93 16.70 26.25 19.61
(88) (83) (85) (256)
4.60 4.21 3.12 6.18
Male Employees 15.39 17.38 25.84 19.55
(44) (40) (43) (127)
5.10 4.68 2.78 6.27
Female Employees 16.48 16.07 26.67 19.66
(44) (43) (42) (129)
4.03 3.67 3.43 6.13

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

209

Table 31 (cont'd)

PERFORMANCE TRENDS

AVGS ACROSS OVERALL
W _p__Im rov M___ntain ___Declin W
AVGS ACROSS 20.28 20.90 21.00 20.72
GENDER CELLS (263) (249) (253) (765)
5.42 4.96 5.79 5.41
Male Employees 20.26 21.18 21.20 20.87
(132) (120) (128) (380)
5.65 4.88 5.64 5.42
Female Employees 20.31 20.64 20.79 20.57
(131) (129) (125) (385)
5.21 5.04 5.94 5.39

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

210

0 Positive Affect

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

24 =__
18
.19,
12 1---... - 1
6 1.
o l l l
Improving Maintaining Declining

Performance Trend

—'— Effort —+— Ability + Ease of Job

Eisure_zg. Performance Trend X Informational Cues

Interaction Effect for Affect

 

211
also do not provide support for hypotheses H11-H11b since no
main effect for employee gender was found, nor was an
employee gender X performance trend interaction found.

In summary, the results of the above analyses for the
effect of performance trend, informational cues, and gender
of the target employee support the following statements.
Performance trends affect expectations for future
performance, but this does not necessarily translate into
differences in promotion and pay decisions. These findings
only partially support hypotheses H1 and H2.

Informational cues interact with performance trends to
yield a strong effect on pay and promotion decisions,
positive affect, and future expectations for employees.
These findings support the hypothesized effects of locus and
controllability on personnel decisions (H3-H4b), though
indirectly. It is yet to be determined that effort is
perceived to be more controllable than ability, and that
ability is more controllable than ease of job. It also
still remains to be shown that ease of job is considered
more external in locus than ability or effort.

Employee gender seems to affect personnel decision
making such that above average female employees who are
improving, maintaining or declining in performance as a
result of job characteristics are rewarded more than above
average males employees who are improving, maintaining, or
declining in performance as a result of job characteristics.

These findings do not support Hll-Hllb, but they do suggest

 

212
gender of the employee can affect personnel decision making,
even when large amounts of performance related information
are available. Given that performance trend, informational
cues, and employee gender were found to affect personnel
decision making, the next series of analyses were designed
to determine if these same factors could affect attributions
of causes of performance.
Effects of Performance Trend! Informational Cuegy and
Employee Gender on Attributions

A doubly multivariate repeated measures MANOVA with
gender of the target employee, performance trend and
informational cues as independent variables and perceived
controllability, stability, and locus of causation as
dependent variables was performed. The results of this
analysis are presented in Table 32. The results suggest
that performance trends and informational cues each
significantly affected attributions, but these main effects
were qualified by the performance trend X informational cues
interaction.

Separate repeated measures univariate ANOVAs were
conducted for each of the three dependent variables
(perceived controllability, external locus of causation, and
stability of causes of observed performance). These
analyses provided a more Specific explanation of the manner
in which different attributions were affected by the

independent variables in the MANOVA. The first ANOVA

213
Table 32

BepeateQ Measures MANOVA for the Overall Effect of

Eeptppmance Trend, Informational Cues, and Employee Gender

en Atttibutions
Error

Effect g; g; E p-value
Constant 3 240 23513.07 .000
Between-Subjects Main Effects

Performance Trend (T) 6 480 2.41 .027
Employee Gender (EG) 3 240 .89 .449
Between-Subjects Interactions

EG X T 6 480 1.43 .202
Within-Subjects Main Effects

Informational Cues (I) 6 237 35.78 .000
Within-Subjects Interactions

T X I 12 474 7.95 .000
EC X I 6 237 1.37 .233
EG X T X I 12 474 .96 .484

Note: All E tests are reported using Wilk's Lambda

214
evaluated the effects of performance trend, informational
cues, and target employee gender on perceived
controllability. The results of this analysis are presented
in Table 33, while the cell means are presented in Table 34.

A significant main effect of informational cues on
perceived controllability was obtained, but this was
qualified by the informational cues X performance trend and
the informational cues X employee gender interactions. The
main effect for informational cues indicates that decision
makers considered effort to be significantly more
controllable than ability, which was considered to be
significantly more controllable than ease of job. These
results provide strong support for hypotheses H7-H7b.

The test for the performance trend X informational cues
interaction shows that improving performance which was due
to effort or ability was considered to be significantly more
controllable than improving performance which was due to
ease of job, but effort and ability were not significantly
different from one another. Maintaining performance which
was due to effort was considered to be significantly more
controllable than maintaining performance which was due to
ability, which was considered to be significantly more
controllable than maintaining performance which was due to
ease of job. Declining performance which was due to effort
was considered to be significantly more controllable than
declining performance which was due to either ability or

non-ease of the job, but ability and non-ease of job were

Table

 

215

33

Repeeted Measures ANOVA for the.Effects of Performance

 

Irene, Informational Cues, and Employee Gender on Perceived

Centrellsbilitx

rifles; d.f. ME
Between-Subjects Effects

Employee Gender (EG) 1 .26
Performance Trend (T) 2 20.99
EG X T 2 9.35
Error 246 11.84

Within-Subjects Effects

Informational Cues (I) 2 1505.14
T X I 4 101.33
EG X I 2 45.51
EG X T X I 4 12.39
Error 492 14.35
Note:

using

All E tests for within-subjects effects are

averaged tests of significance

Im

.02

.79

104.90

7.06

3.17

.86

p-velue

.882
.172

.455

.000
O 000
.043

.486

reported

216

 

Table 34
Qe;1 Means of ContrelIabilityll Split by Performance Trend,
'ona ues nd Em lo ee Ge der
PERFORMANCE TRENDS
AVGS ACROSS
INFO TYPE Improv Mntain Declin zeppp_eppte
Ability Info Cues 15.63 15.80 13.18 14.87
(88) (83) (85) (256)
3.75 3.59 3.31 3.74
Male Employees 16.30 15.63 13.16 15.02
(44) (40) (43) (127)
3.44 3.38 3.29 3.61
Female Employees 14.95 15.95 13.19 14.71
(44) (43) (42) (129)
3.95 3.81 3.37 3.87
Effort Info Cueg 16.56 17.67 17.79 17.32
(88) (81) (84) (253)
3.82 3.27 3.58 3.61
Male Employees 16.34 17.02 17.17 16.83
(44) (38) (42) (124)
3.85 3.64 3.55 3.67
Female Employees 16.77 18.23 18.40 17.79
(44) (43) (42) (129)
3.83 2.84 3.55 3.49
EaseJob Info Cues 12.74 11.80 12.74 12.44
(88) (82) (85) (255)
3.84 3.95 4.05 3.95
Male Employees 12.84 12.40 12.77 12.68
(44) (40) (43) (127)
4.09 4.34 3.66 4.00
Female Employees 12.64 11.24 12.71 12.20
(44) (42) (42) (128)
3.62 3.49 4.46 3.90

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

217
Table 34 (cont'd)

PERFORMANCE TRENDS

AVGS ACROSS OVERALL
INEQ_£§LL§ Improv Mntain Declin GRAND MEANS
AVGS ACROSS 14.97 15.08 14.55 14.87
GENDER CELLS (264) (246) (254) (764)
4.12 4.35 4.30 4.26
Male Employees 15.16 14.98 14.34 14.83
(132) (118) (128) (378)
4.12 4.25 4.00 4.12
Female Employees 14.79 14.77 14.77 14.91
(132) (128) (126) (386)
4.14 4.46 4.59 4.39

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

218

not significantly different from each other. These results

are graphically depicted in Figure 21.

The employee gender X performance informational cues
interaction effect shows that effort informational cues were

considered to be significantly more controllable for female

employees than for male employees. None of the other

differences between males and females were found to be

significant. These results provide partial support for

hypotheses H10 and H10c. The results do not support

hypotheses H10a-H10b, nor do they support Hlod-Hloe. The

results do suggest however, that employee gender exerts some

effect on attributional formation, which in turn, appears to

affect personnel decision making.

In sum, perceptions of controllability are strongly
affected by the nature of informational cues provided, as

hypothesized in Weiner's three-dimensional attributional

taxonomy. Effort is considered to be more controllable than

is ability, which is considered to be more controllable than

is ease of job. These statements are qualified by the

performance trend X informational cues and employee gender X

informational cues interactions. Ability is considered to

be less controllable than effort, but this is particularly
true for employees exhibiting declining performance trends.
Employee gender also affects attributional formation such

that effort cues are considered to be more controllable for

female employees than for male employees.

219

Perceived Controllability

 

20

 

 

 

16, .

 

 

12

 

 

.1. ”M1,--.g—‘_._r1ev..w ”11.. 1. ..v w-

 

L

 

 

 

0 1 IL m

Improving Maintaining Declining
Performance Trend

7"“ Effort —1— Ability + Ease of Job

Eigute_zl. Performance Trend X Informational Cues

Interaction Effect for Perceived Controllability

220

The next analysis was a repeated measures ANOVA which

evaluated the effects of performance trend, informational

cues, and target employee gender on perceived external locus

of causation. The results of this analysis are presented in

Table 35, while the cell means are presented in Table 36.

Significant main effects were obtained for performance

trend and informational cues. The main effect for

performance trend indicates that improving and declining

performance trends were considered significantly more

external than maintaining performance trends. The main

effect for informational cues indicates that ability and
effort cues were considered significantly more internal than

ease of job cues, but ability and effort were not

significantly different from each other. These effects were

qualified by the performance trend X informational cues and

employee gender X informational cues interactions.

The test for the performance trend X informational cues

interaction shows that though effort and ability were

considered to be more internal than ease of job, the size of

mean differences varied by performance trend. The overall

pattern, however, remained the same. Figure 22 graphically

depicts this interaction effect. The results provide strong

support for hypotheses H7 and H70.
The significant employee gender X informational cues
interaction indicates that effort cues were considered to be

more internal for female employees than for male employees.

No other gender differences were significant. These results

221

Table 35

eas es V the E fe s

ero

Irene, Informational Cues, and Employee Gender on Perceiyeg

Leeus of Causation
Ettect Qi Mg

Between-Subjects Effects

Employee Gender (EG) 1 5.43
Performance Trend (T) 2 75.05
EG X T 2 6.18
Error 248 12.55
Within—Subjects Effects

Informational Cues (I) 2 2339.19
T X I 4 220.74
EG X I 2 56.77
EG X T X I 4 13.11
Error 496 16.84

Note:

using averaged tests of significance

I'TI

.43

.49

138.87

13.10

p-value

.511
.003

.612

.000
.000
.035

.540

All E tests for within-subjects effects are reported

222

Table 36

Qell Means of External Locus of Causality, Split by
Bettermance Trend, Informational Cues, and Employee Gender

PERFORMANCE TRENDS
AVGS ACROSS

 

INFO TYPE IEREQX EQLAiE 222112 ZB§!2_Q§LL§
521112! Info Cues 12.78 11.41 14.38 12.87
(87) (83) (85) (255)
3.76 3.12 4.16 3.88
Male Employees 12.89 11.78 14.19 12.98
(44) (40) (43) (127)
3.43 2.66 3.81 3.46
Female Employees 12.67 11.07 14.57 12.76
(43) (40) (42) (128)
4.11 2.66 4.53 4.28
Effort Info Cues 12.72 11.24 14.15 12.72
(87) (83) (85) (255)
3.26 3.23 4.75 3.98
Male Employees 13.16 11.78 14.77 13.27
(44) (40) (43) (127)
3.33 3.36 4.31 3.86
Female Employees 12.28 10.74 13.52 12.17
(43) (43) (42) (128)
3.17 3.07 5.14 4.04
EaseJob Info Cues 18.45 19.13 16.48 18.02
(88) (83) (85) (256)
4.12 4.20 4.46 4.39
Male Employees 18.61 18.13 16.14 17.62
(44) (40) (43) (127)
4.01 4.52 4.67 4.50
Female Employees 18.30 20.07 16.83 18.41
(44) (43) (42) (129)
4.26 3.70 4.25 4.26

Note: Numbers in parentheses indicate number of observations

Note:

Bottom numbers in cells indicate standard deviations

223

Table 36 (cont'd)
PERFORMANCE TRENDS

AVGS ACROSS OVERALL
Inzg_g§pp§ Improv Mntain Declin GRAND MEANS
AVGS ACROSS 14.67 13.93 15.00 14.54
GENDER CELLS (262) (249) (255) (766)
4.59 5.11 4.57 4.77
Male Employees 14.89 13.89 15.03 14.62
(132) (120) (129) (381)
4.45 4.66 4.32 4.49
Female Employees 14.45 13.96 14.98 14.46
(130) (129) (126) (385)
4.74 5.51 4.82 5.04

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

224

Perceived External Locus

 

20

“.._-.._..- .._. ..fi -7 VN'.‘ -._4

 

16

 

12

 

 

 

 

 

 

O 4r I
Improving Maintaining Declining

Performance Trend

—"— Effort —+‘ Ability + Ease of Job

Eiguze_zz. Performance Trend X Informational Cues

Interaction Effect for Perceived External Locus

225

provide partial support for hypotheses H10 and Hloc,
suggesting that employee gender affects the formation of
attributions about causes of employee performance.

In sum, informational cues strongly affect perceptions
of locus of causality in a manner consistent with Weiner's
attributional taxonomy. Effort and ability are both
considered to be more internal in locus of causality than is
Effort and ability do not seem to differ in

ease of job.
perceived locus of causality. Performance trends affect
perceptions of locus such that maintaining performance

trends are considered to be more internal than are improving

or declining performance trends. The results also show that

the main effects of performance trend and informational cues
are qualified by performance trend X informational cues and
employee gender X informational cues interactions.

The next analysis on attributions involved a repeated
measures ANOVA which evaluated the effects of performance
trend, informational cues and employee gender on perceived

stability. The results are presented in Table 37, and the
cell means are presented in Table 38.

A main effect for informational cues was obtained.
Ability was considered to be significantly more stable than

effort or ease of job, but ease of job and effort were not

significantly different from one another. This main effect

was qualified by the significant effect of the performance

trend X informational cue interaction, and to a certain

226

Table 37
Measures AN VA the Effec s f Per 0 nc

Irene, Informational Cues, and Employee Gender on Perceived

Stability

E13199; d1f1 14.5. E 223111;:

Between-Subjects Effects

Employee Gender (EG) 1 18.79 1.75 .187

Performance Trend (T) 2 7.92 .74 .479

EG X T 2 28.32 2.64 .073

Error 248 10.73

Within-Subjects Effects

Informational Cues (I) 2 64.89 6.41 .002

T X I 4 24.49 2.42 .048

EC X I 2 .88 .09 .917

ES X T X I 4 19.69 1.95 .102

Error 496 10.12

Note: All E tests for within-subjects effects are reported

using averaged tests of significance

227

 

Table 38
gell Menns of Stabi1ity, Split by Performance Treng,
ues lo Ge de
PERFORMANCE TRENDS
AVGS ACROSS
INFO TYPE Improv Mntain Declin TREND CELLS
Ability Info Cuee 16.76 17.27 16.41 16.81
(87) (82) (85) (254)
3.09 2.80 3.69 3.23
Male Employees 16.12 16.79 17.19 16.70
(43) (39) (43) (125)
3.00 2.66 3.65 3.15
Female Employees 17.39 17.70 15.62 16.91
(44) (43) (42) (129)
3.08 2.89 3.60 3.30
Ettort Info Cues 16.06 16.23 15.19 15.82
(88) (82) (85) (255)
3.02 2.97 3.21 3.09
Male Employees 15.98 15.72 15.05 15.58
(44) (39) (43) (126)
2.89 2.69 3.27 2.97
Female Employees 16.14 16.69 15.33 16.06
(44) (43) (42) (129)
3.17 3.15 3.18 3.19
EaseJob Info Cues 15.93 15.82 16.53 16.09
(88) (83) (85) (256)
3.65 3.46 3.05 3.40
Male Employees 16.18 15.33 16.42 15.99
(44) (40) (43) (127)
3.62 3.64 2.80 3.37
Female Employees 15.68 16.28 16.64 16.19
(44) (43) (42) (129)
3.72 3.27 3.33 3.44

Note: Numbers in parentheses indicate number of observations

Note:

Bottom numbers in cells indicate standard deviations

Table 38 (cont'd)
AVGS ACROSS

I320 CELLS

AVGS ACROSS
GENDER CELLS

Male Employees

Female Employees

Note: Numbers in parentheses indicate number of observations

Note: Bottom numbers in cells indicate standard deviations

228

PERFORMANCE TRENDS

Improv

16.25
(263)
3.28

16.09
(131)
3.16

16.40
(132)
3.39

Mntain

16.44
(247)
3.14

15.94
(118)
3.08

16.89
(129)
3.14

Declin

16.04
(255)
3.37

16.22
(129)
3.35

15.87
(126)
3.39

OVERALL
GRAND MEANS

16.24
(765)
3.26

16.09
(378)
3.20

16.39
(387)
3.33

229
degree, by the employee gender X performance trend
interaction (p S .07).

The significant performance trend X informational cues
interaction indicates that perceptions of stability varied
depending on the performance trend and informational cues
provided. In the improving performance trend condition, no
significant differences were obtained in perceived stability
as a function of informational cues provided. In the
maintaining performance trend condition, ability was
perceived to be significantly more stable than was effort or
ease of job, but effort and ease of job were not
significantly different. In the declining performance trend
condition, effort was viewed as more unstable than was
ability or ease of job, but ability and ease of job were not
significantly different from each other. These results are
graphically depicted in Figure 23.

The significant employee gender X performance trend
interaction effect shows that causes of above average
maintaining performance for female employees were considered
to be significantly more stable than causes of above average
maintaining performance for male employees. There were no
other significant differences in perceived stability for
males and females.

The results of the analysis for perceived stability do
not support hypotheses H5 and 7d, since no main effect for

trend was obtained. The results do provide support for

230

Perceived Stability

 

 

 

 

 

 

 

 

 

 

20

18

16

1 4 .._-.1 ._..1 _---1._... .-11---..-..-.----.._..1_ -1.-- ._-.---. -11-.111-.- _.-- - .--

12 .L__.__1_-.._..--------- 1-.- -1 _

10 f 7‘ 4:

Improving Maintaining Declining
Performance Trend
—*‘— Effort —+— Ability + Ease of Job

Eignte_21. Performance Trend X Informational Cues

Interaction Effect for Perceived Stability

231
hypothesis H8, but they directly contradict hypothesis flea
and they do not support H8b.

As a whole, the results of the analyses on attributions
indicate that effort, ability, and ease of job vary along
the dimensions of perceived locus of causation,
controllability, and to some degree, stability. The
findings are very consistent with the attributional
dimensions proposed by Weiner which were modified and
presented in Figure 2.

Findings regarding the effects of gender are more
equivocal. While gender of employee interacts with
informational cues to affect perceived locus and
controllability of performance, obtained findings are not
consistent with hypotheses. Employee gender also interacts
with performance trend to affect perceived stability, but
again, these findings are not consistent with hypothesized
effects. The findings do indicate, however, that employee
gender affects the formation of attributions of causation.

When analyses for personnel decisions and attributions
are considered together, several conclusions can be drawn.
First, informational cues affect perceived locus of
causation and controllability as hypothesized. Effort is
considered to be more controllable than ability, which is
considered to be more controllable than ease of job. Also,
effort and ability are considered to be more internal in

locus of causation than is ease of job.

232

Second, results of both analyses suggest that
perceptions of locus of causation and controllability
interact with observed performance trends to strongly affect
personnel decisions, even when all employees show the same
above average performance level. Employees who show
improving or maintaining above average performance trends
which are due to controllable or internal causes are
rewarded more than employees who show the same performance
trends which are due to uncontrollable or external causes.
Conversely, employees who show declining above average
performance trends which are due to internal and
controllable causes are rewarded much less than employees
who show declining above average performance trends which
are due to external and uncontrollable causes.

Third, gender of the employee interacts with
informational cues to affect personnel decisions, but it is
not clear whether these effects occur through attributional
processes. For example, gender of the employee interacts
with informational cues to affect perceived locus and
controllability of effort, and also with performance trend
to affect perceived stability. However, pay and promotion
decisions are affected by the interaction of employee gender
with ease of job cues, not effort and performance trend as
noted above. Also, female employees receive more reward
than male employees, but only when their performance is due

to ease (or non-ease) of the job, particularly when the

233
employee exhibits a declining performance trend (in above
average performance).

The results of the above analyses have supported
hypotheses regarding the important role that causal
attributions play in the process of making personnel
decisions. It was hypothesized earlier, based on group
membership studies, that gender group status would affect
attributional processes and that it would affect personnel
decision making. The next set of analyses were designed to
investigate these hypothesized effects.

Effects of Performance Trend. Informational Cue§y and Gender

Group Membership on Personnel Decisions

A doubly multivariate repeated measures MANOVA was

 

conducted using gender group membership status, performance
trend, and information cues as independent variables and
with promotion, pay, affect, and expectations as dependent
variables. Participant responses to target employees were
coded as belonging to either the same or different gender
group. This analysis was performed to test the hypothesis
that performance trends, informational cues, and employee
gender group membership would affect overall personnel
decisions. The results of this MANOVA are presented in
Table 39.

As in previous analyses, the results indicate that
performance trends and informational cues significantly
affected personnel decisions. These main effects were

qualified by the significant performance trend X

234
Table 39
Repenten Measures MANOVA for the Overall Effect of
Eepfotnance Trend, Infopnational Cuee. and Gender Group
Mempetenip on Personnel Decisions

 

 

Error

Effect n; n; E p-value
Constant 4 239 5667.38 .000
Between-Subjects Main Effects

Performance Trend (T) 6 478 4.22 .000
Gender Group (CG) 4 239 1.82 .125
Between-Subjects Interactions

GS X T 8 478 1.52 .149
Within-Subjects Main Effects

Informational Cues (I) 8 235 6.87 .000
Within-Subjects Interactions

T X I 16 470 33.45 .000
GG X I 8 235 1.60 .126
66 X T X I 16 470 1.11 .345

Note: All 2 tests are reported using Wilk's Lambda

235
informational cues interaction. The main effect for gender
group membership was not significant, nor were any of the
interactions involving gender group membership. No further
analyses involving gender group membership were conducted.

The results do not provide support for hypotheses H15
to Hle. Gender group membership did not interact with cues
to affect personnel decision making. Even though gender
group membership did not affect personnel decision making,
it is possible that it could have an impact on attributional
processes. The following section describes analyses used to
investigate this possibility.

Effects of Performance Trendy Informational Cues, and gender
Gpoup Membership on Attributions

A doubly multivariate repeated measures MANOVA was
conducted to investigate the effects of performance trend,
informational cues, and gender group membership on causal
attributions. The results are presented in Table 40.

As previous analyses have indicated, informational cues
and performance trend significantly affected perceived
attributions. The performance trend x informational cues
interaction qualified the main effects of informational cues
and performance trend on attributions. No significant
effects involving gender group membership were detected, and
no further analyses were conducted. These results do not
support hypotheses H12-H14e.

The results suggest that gender group membership does

not have an affect on attributional processes and personnel

236

Table 40
Bepeeted Measuree MANOVA for the-Overall Effect of

Eetfotnance Trend, Informational Cues, and Gender Group
Membership on Attributions

Error
EI£§§L Q2 Qfi E szelus
Constant 3 240 23088.53 .000
Between-Subjects Main Effects
Performance Trend (T) 6 480 2.36 .030
Gender Group (GG) 3 240 1.18 .320
Between-Subjects Interactions
66 X T 6 480 .63 .705
Within-Subjects Main Effects
Informational Cues (I) 6 237 35.53 .000
Within-Subjects Interactions
T X I 12 474 7.86 .000
66 X I 6 237 .18 .983
GG X T X I 12 474 .79 .666

Note: All E tests are reported using Wilk's Lambda

237

decision making for above average employees. These findings
suggest one of two possibilities. Either group related
attributional biases may be ameliorated by providing enough
performance related information, or such attributional
biases do not generalize to gender group membership.

Earlier it was demonstrated that ability, effort, and
ease of job varied in hypothesized directions on perceived
controllability, locus and stability. Further, it was shown
that these informational cues interacted with performance
trend to affect personnel decision making. It is still
necessary, however, to demonstrate that the actual
perceptions of controllability, stability, and locus of
causation interact with performance trend to affect

personnel decision outcomes.

Effects of Perceived Attributions and Performance Trend on
Bensonnel Decisions

Though previous analyses provided strong support for
hypotheses H3-H4b, the evidence provided was indirect. It
was demonstrated that effort, ability, and ease of job cues
varied along attributions as hypothesized. It was further
demonstrated that these informational cues interacted with
performance trend to influence personnel decisions, also as
hypothesized. It was inferred, based on these results, that
attributions interacted with performance trend to affect
personnel decision making.

Based on the model presented in Figure 3, however, it

is the actual perceptions of the decision maker regarding

238
the three attributional dimensions that need to be linked to
personnel decisions. In order to accomplish the analyses
necessary to provide support for such linkages, a repeated
measures regression approach was used (Cohen & Cohen, 1983:
Hollenbeck, Ilgen, & Sego, in press).

A repeated measures regression analysis begins by
partitioning the variance in the dependent variable into
between and within subjects effects. For the purposes of
discussion, assume the dependent variable is pay decisions
made by participants. Using normal regression methods, one
would have to dummy code all of the participants in the
study with n - 1 dummy codes in order to determine the
variance in pay decisions which is due to within subjects
effects. An alternative to this unwieldy method is provided
in Cohen and Cohen (1983, Chapter 11).

To partition the variance in the dependent variable of
pay decisions, one begins by computing averages across the
responses provided by each participant. In this case,
average scores in pay decisions for each participant across
the three target responses are computed. This provides a
total of 256 average scores for pay decisions, since the
number of participants was 256. The variance of these
nyetage scores gives the variance due to between subjects
effects. Assume this is 20. Next, the variance across elf
768 scores provided by participants is computed (768=256 x
3): this is the tpte; variance in the dependent variable.

Assume this is 80. The proportion of variance accounted for

239
by between subjects effects is the ratio of variance of the
average scores (n=256) to the variance across all scores
(n=768). In this case, the proportion of variance due to
between subjects effects is .25 (20/80). The corresponding
degrees of freedom are 255 (256-1).

Since between and within subjects effects are assumed
to be orthogonal, the proportion of variance due to within
subjects effects is simply one minus the proportion due to
between subjects effects. In this case, the proportion due
to within subjects effects is .75 (1-.25). The
corresponding degrees of freedom for the within subjects
effects are 512 (768-256).

Once the variance in the data is partitioned, one must
enter the independent variables of interest in several
hierarchical steps. All within subjects variables are
entered first, including interactions of those variables, if
they are relevant to the design of the experiment. The E
tests and increments in R2 for these variables are based
upon the within subjects proportion of the total variance
(.75) and the 512 degrees of freedom.

Next, all between subjects variables and their
interactions are entered into the regression equation. All
of the 2 tests and increments in R‘2 for these variables are
based upon the between subjects proportion of the total
variance (.25) and the 255 degrees of freedom.

Finally, all interaction terms which include within eng

between subjects variables are entered. The E tests and

240
increments in R? for these interaction terms are based upon
the within subjects proportion of the total variance (.75)
and the 512 degrees of freedom.

The normal p and t statistics provided by standard
computer printouts must be adjusted for the proportion of
variance and the degrees of freedom in the numerator and
denominator. The resulting analysis yields the same outcome
as if one had dummy coded all 256 participants in the study
to partition the variance into between and within subjects
effects. In a factorial design with equal sample sizes in
each cell, the repeated measures regression approach will
yield 3 tests which are identical, within rounding error, to
the mixed model analysis of variance (Winer, 1971). See
Cohen and Cohen (1983, Chapter 11) and Hollenbeck, et al.
(in press) for a more extensive treatment of this procedure.

The primary advantage of the repeated measures
regression approach in this study is that one can regress
personnel decisions on the actual perceptions of
controllability, stability, and locus of causation.
Furthermore, interaction terms between perceived
attributions and performance trend or employee gender can be
included in the analyses. This can be done without having
to dichotomize participants' perceptions into low and high
categories, and all available degrees of freedom are used.

The resulting analysis is similar in a sense, to a
policy capturing analysis.. The "weight" of various

attributions and their interactions can be determined using

241
standard regression techniques, adjusted for the variance
partitioning procedure. A final advantage is that
partitioning of the variance into within and between
subjects effects reduces error terms and increases the power
of all analyses conducted.

The first repeated measures regression analysis
regressed pay decisions on employee gender, performance
trend, perceived locus of causation, perceived
controllability, and perceived stability, along with their
interaction terms. The first step entered perceived locus,
controllability and stability (within subjects effects).
The next step entered the interactions between the three
perceptions (within).

Performance trend was entered in the third step
(between), while gender of the target employee was entered
in the fourth step (between). Two dummy codes represented
performance trend, one for improving (Trend 1) and one for
declining (Trend 3), and maintaining performance trend was
the comparison group. A single dummy code represented
gender, with male employees as the comparison group. In the
fifth step, the interaction terms for performance trend and
gender were entered (between).

In the sixth step, the interactions between employee
gender and attributional perceptions (perceived
controllability, locus, and stability) were entered
(within). The seventh, eighth, and ninth steps entered the

interaction terms between performance trend and attributions

242
in a hierarchical fashion (within). The interaction terms
for perceived controllability and performance trend were
entered last, to control for the effects of interactions
involving locus and stability. In the final three steps,
triple interaction terms between performance trend, employee
gender, and attributions were entered, also in a
hierarchical fashion (within). Again, interaction terms
involving controllability were entered in the last step.
Table 41 presents the results of the significance tests, and
Table 42 presents the regression equations associated with
the entry of each block of variables.

The variance of pay decisions across all possible cases
was 60.202. The variance of average pay decisions across
the participants was 16.307. The proportion of variance due
to between subjects effects was 27.1% (16.307/60.202), while
the variance due to within subjects effects was 72.9% (1 -
27.1%). These are the figures used to adjust the E and t
tests in the regression analysis described above. Missing
data reduced the total number of observations to 757.

The effects of attributions were the only statistically
significant main effects in the results. Above average
performance which was internal or stable was rewarded more
than above average performance which was external or
unstable. Employee gender and performance trend did not
significantly affect pay decisions, and none of the

interactions involving employee gender were significant.

243

Table 41
Beenfte pf Repeated Measures Regressions for Pay Decisions
a- an! 9 -e _éno‘ '*._Oguance ._end .no A t 'o- To 3

Indpndnt 18 for R2 Incrmnt Incrmnt

gtep Vars Egpation Incrmnt Btwn Within

1 Control (C) .119 .119 .163**
Locus (L)
Stability (S)

2 L X S .121 .002 .003
C X S
L X C

3 Trend 1 (T1) .122 .001 .003
Trend 3 (T3)

4 Employee .123 .001 .002

Gender (EG)

5 EG X T1 .123 .000 .001
EC X T3

6 EG X S .124 .001 .001
EC X L
EG X C

7 T1 X S .127 .003 .004
T3 X S

8 T1 X L .240 .113 .155**
T3 X L

9 T1 X C .271 .031 .043**
T3 X C

10 T1 X S X EG .274 .003 .004
T3 X S X ES

11 T1 X L X EG .275 .001 .001
T3 X L X EG

12 T1 X C X EC .278 .003 .004
T3 X C X EG

Note: * indicates p S .05

Note: ** indicates p s .01

244
Table 42

E a ' Pred' ' Pa Dec' '
Using Employee Gender. Performance Trend. and Attributione

Step 1 - Enter Attributions
Y' = -.523(L)** + .302(S)** + -.056(C) + 36.53

Step 2 - Enter interactions in Attributions
Y' = -.826(L)** + .006(S) + -.506(C) + .014(LXC) +
.015(CXS) + .007(LXS) + 44.57

Step 3 - Enter Performance Trend
Y' = -.814(L)* + -.004(S) + -.495(C) + .013(LXC) + .015(CXS)
+ .OO7(LXS) + .471(Tl) + .321(T3) + 44.14

Step 4 - Enter Employee Gender
Y' = -.843(L)* + -.O30(S) + -.516(C) + .014(LXC) + .Ol6(CXS)
+ .008(LXS) + .477(Tl) + .324(T3) + .374(EG) + 44.56

Step 5 - Enter Employee Gender by Trend interactions

Y' = -.854(L)* + -.044(S) + -.519(C) + .014(LXC) + .016(CXS)
+ .008(LXS) + .621(T1) + .709(T3) + .726(EG) +
-.271(EGXT1) + -.764(EGXT3) + 44.65

Step 6 - Enter Gender by Attributions interactions

Y' = -.838(L)* + -.129(S) + -.489(C) + .013(LXC) + .017(CXS)
+ .009(LXS) + .598(T1) + .728(T3) + 1.090(EG) +
-.228(EGXT1) + -.723(EGXT3) + -.047(EGXL) +
-.091(EGXC) + .102(EGXS) + 44.92

Step 7 - Enter Trend by Stability interaction

Y' = -.807(L)* + -.l98(S) + -.429(C) + .013(LXC) + .015(CXS)
+ .008(LXS) + .496(T1) + -4.139(T3) + 1.430(EG) +
-.278(EGXT1) + -.756(EGXT3) + -.062(EGXL) +
-.097(EGXC) + .105(EGXS) + .302(T3XS) + .007(T1XS) +

45.54

Step 8 - Enter Trend by External Locus interaction

Y' = -l.270(L)** + -.253(S) + —.266(C) + .006(LXC) +
.008(CXS) + .014(LXS) + -l.165(T1) + -24.115(T3)** +
3.875(EG) + -.580(EGXT1) + -l.001(EGXT3) +
-.076(EGXL) + -.110(EGXC) + -.010(EGXS) + .398(T3XS)*
+ .030(T1XS) + .119(T1XL) + 1.264(T3XL)** + 52.09

Note: * indicates p s .05
Note: ** indicates p s

245

Table 42 (cont'd)

Step
Y' =

Step
Y' =

Step
Y' =

Step

Note:
Note:

9 - Enter Trend by Controllability interaction

-1.067(L)** + -.1l7(S) + .062(C) + .006(LXC) +
.004(CXS) + .011(LXS) + -1.348(T1) + -2.395(T3) +
1.989(EG) + -.286(EGXT1) + -.679(EGXT3) + -.047(EGXL)
+ -.056(EGXC) + .022(EGXS) + .157(T3XS) + .010(T1XS)
+ .109(T1XL) + .857(T3XL)** + -.819(T3XC)** +

.029(T1XC) + 43.662

10 - Enter Trend x Stability x Employee Gender

-1.064(L)** + -.275(S) + .067(C) + .008(LXC) +
.003(CXS) + .011(LXS) + -6.140(T1) + -4.574(T3) +
-3.989(EG) + 10.703(EGXT1) + 4.788(EGXT3) +
-.045(EGXL) + -.048(EGXC) + .376(EGXS) + .329(T3XS) +
.358(T1XS) + .075(T1XL) + .835(T3XL)** +
-.819(T3XC)** + .009(T1XC) + -.331(T3XSXEG)

-.672(T1XSXEG) + 46.084

+

11 - Enter Trend X Locus x Employee Gender

-1.109(L)** + -.290(S) + .053(C) + .008(LXC) +
.004(CXS) + .010(LXS) + -7.490(T1) + -5.551(T3) +
-5.722(EG) + 13.724(EGXT1) + 6.769(EGXT3) +
.047(EGXL) + -.042(EGXC) + .397(EGXS) + .334(T3XS) +
.368(T1XS) + .163(T1XL) + .901(T3XL)** +
-.841(T3XC)** + .004(T1XC) + -.350(T3XSXEG) +
-.712(T1XSXEG) + -.166(T1XLXEG) + -.117(T3XLXEG) +

47.034

12 - Enter Trend x Controllability x Employee Gender

-1.022(L)** + -.254(S) + .144(C) + .008(LXC) +
.004(CXS) + .008(LXS) + -1.256(T1) + —5.537(T3) +
-.386(EG) + -.048(EGXT1) + 5.938(EGXT3) + -.070(EGXL)

+ -.247(EGXC) + .361(EGXS) + .358(T3XS) + .344(T1XS)

+ .032(T1XL) + .882(T3XL)** + -.849(T3XC)** +
-.258(T1XC) + -.370(T3XSXEG) + -.633(T1XSXEG) +
.125(T1XLXEG) + -.075(T3XLXEG) + .035(T3XCXEG) +

.552(T1XCXEG) + 44.210

* indicates p S .05
** indicates p S .01

246
The only significant interaction effects found were for the
external locus of causality X performance trend and
controllability X performance trend variables.

It is noteworthy that controllability interacted with

performance trends to exert a significant effect on pay

decisions, even when locus and the locus x trend interaction

variables were previously entered into the regression

equation. This suggests that though external locus and

controllability are related constructs (p = -.56),

controllability adds prediction in pay decisions beyond the

influence of external locus of causality.

The n weights for equation 8 indicate that the slope of
the regression line for the effect of external locus on pay

decisions was not significantly different between improving

trends and maintaining trends (p = .119 for T1 X Locus,

n.s.), but the slope for the effect of locus on pay was

significantly different between declining trends and
maintaining trends (p = 1.264 for T3 X Locus, p S .01).
Similarly, the p weights for equation 9 indicate that
the slope for the effect of controllability on pay decisions
was not significantly different between improving trends and
maintaining trends (p = .029 for T1 X Control) but the
difference in the slope for the effect of controllability on

pay decisions was significantly different between declining

trends and maintaining trends (p = -.819 T3 X Control, p S

.01).

247
These results strongly support hypotheses H3 to H4b.
Declining trends which are external or uncontrollable are

rewarded more than declining trends which are internal or
controllable, and improving or maintaining trends which are

external or uncontrollable are rewarded less than improving

trends which are internal or controllable.
Figures 24 and 25 graphically represent the

interactions between controllability or external locus of

causality and performance trends. They were created by

substituting appropriate group values and means into

equations 8 and 9, and then plotting the equations for

perceptions of external locus of causality and

controllability one SD above and below their respective

means.
A second repeated measures regression analysis was

conducted by regressing promotion decisions on employee
gender, performance trend, perceived locus of causation,

perceived controllability, and perceived stability, along

with their interaction terms. The same hierarchical order

of entry employed for pay decisions was used for promotion

decisions. First, perceived locus, controllability and

stability were entered. Next, the interactions between the

three attributional perceptions were entered. Performance

trend was entered in the third step, and gender of the

target employee was entered in the fourth step. As before,

performance trend was represented by two dummy codes, one

for improving trend (Trend 1) and one for declining trend

248

Pay Decisions

 

 

 

 

 

 

 

 

4o ,
36 \ -
32 ——~ «—
28 -——~ -
24 - ,1 -
20 4 L 4 I 4
8 1o 12 14 16 18 20

Perceived External Locus

*lmproving +Maintaining +Declining

W.

Interaction Effect of External Locus of

Causality and Performance Trend for Pay Decisions

249

Pay Decisions

 

 

 

 

 

 

 

 

4o
36 1 - 11-
32 W ——-~
28
24 -
20 1 1 s L A
8 1o 12 14 16 18 20

Perceived Controllability

“—‘Improving "—i—Maintaining +Declining

Eignze_2§. Interaction Effect of Perceived Controllability

and Performance Trend for Pay Decisions

250
(Trend 3), with maintaining performance trend as the
comparison group. Employee gender was represented by a
single dummy code. The interaction terms for performance
trend and gender were entered in the fifth step. In the
sixth step, interactions between employee gender and
attributional perceptions (perceived controllability,
locus,and stability) were entered. In the seventh, eighth,
and ninth steps, interaction terms between performance trend
and attributions were entered. In the final three steps,
the triple interactions between performance trend, employee
gender, and attributions were entered. Table 43 presents
the results of the significance tests, and Table 44 presents
the regression equations associated with the entry of each
block of variables. The findings for promotion decisions
were very similar to the findings for the pay decisions.

The variance of promotion decisions across all possible
cases was 28.887. The variance of average promotion
decisions across the participants was 6.637. The proportion
of variance due to between subjects effects was, therefore,
23.0% (6.637/28.887), while the variance due to within
subjects effects was 77.0% (1 - 23.0%). These are the
figures used to adjust the E and t tests in the regression
analysis described above. Missing data reduced the total
number of observations to 755.

As was found in the results for pay decisions, the only
main effects obtained in the results were for the effects of

attributions. Above average performance which was internal

Table 43

251

Beenlts of Repeated Measures Regressfons for Prometion

s' s Usin Em lo ee Gender Performance Trend and

 

Attribstiens
Indpndnt 18 for R2 Incrmnt Incrmnt

§tep Vars Eqnation Incrmnt Btwn Within

1 Control (C) .119 .119 .154**
Locus (L)
Stability (S)

2 L X S .133 .014 .019*
C X S
L X C

3 Trend 1 (T1) .133 .000 .000
Trend 3 (T3)

4 Employee .134 .001 .005

Gender (EG)

5 EG X T1 .135 .001 .002
EC X T3

6 EG X S .137 .003 .004
EC X L
EG X C

7 T1 X S .138 .001 .001
T3 X S

3 T1 X L .238 .100 .130**
T3 X L

9 T1 X C .256 .018 .023**
T3 X C

10 T1 x s x EC .258 .001 .002
T3 X S X EC

11 T1 x L x E6 .259 .001 .001
T3 X L X EG

12 T1 X C X EC .259 .000 ~001
T3 X C X EG

Note: * indicates p S .05

Note: ** indicates p S .01

252

Table 44

Regression Egnations for Prediction of Promotion Decisions

gsing Employee Gender, Performance Trend, and Attributions

Step 1 - Enter Attributions
Y' = -.320(L)** + .245(S)** + .026(C) + 14.03

Step 2 - Enter interactions in Attributions
Y' = ~.638(L)** + -.382(S) + -.831(C)** + .015(LXC) +
.038(CXS)** + .007(LXS) + 27.77

Step 3 - Enter Performance Trend
Y' = -.639(L)* + -.383(S) + -.834(C)** + .015(LXC) +
.038(CXS)* + .007(LXS) + -.084(T1) + .000(T3) + 27.83

Step 4 - Enter Employee Gender

Y‘ = -.610(L)* + -.359(S) + -.814(C)** + .014(LXC) +
.038(CXS)* + .006(LXS) + -.090(T1) + -.003(T3) +
-.374(EG) + 27.43

Step 5 - Enter Employee Gender by Trend interactions

Y' = -.603(L)* + -.349(S) + -.807(C)** + .014(LXC) +
.038(CXS)* + .005(LXS) + -.338(T1) + -.192(T3) +
.664(EG) + -.365(EGXT3) + .482(EGXT1) + 27.36

Step 6 - Enter Gender by Attributions interactions

Y' = -.615(L)* + -.417(S) + -.790(C)** + .014(LXC) +
.038(CXS)* + .005(LXS) + -.306(T1) + -.141(T3) +
~3.577(EG) + .370(EGXT3) + .479(EGXT1) + .053(EGXL) +
-.020(EGXC) + .151(EGXS) + 28.54

Step 7 - Enter Trend by Stability interaction

Y' = -.595(L)* + -.470(3) + —.765(C)** + .013(LXC) +
.037(CXS)* + .004(LXS) + -1.477(T1) + -2.386(T3) +
-3.387(EG) + .330(EGXT3) + .418(EGXT1) + .047(EGXL) +
-.025(EGXC) + .152(EGXS) + .140(T3XS) + .073(T1XS) +
29.18

Ste - Enter Trend by External Locus interaction

Y' .884(L)** + -.503(S) + -.659(C)* + .009(LXC) +
.032(CXS)* + .008(LXS) + -2.168(Tl) + -15.168(T3)** +
-1.864(EG) + .158(EGXT3) + .221(EGXT1) + .040(EGXL) +
-.030(EGXC) + .075(EGXS) + .196(T3XS) + .084(T1XS) +
.055(T1XL) + .814(T3XL)** + 33.208

"'6
I6

Note: * indicates p S .05
Note: ** indicates p S

253
Table 44 (cont'd)

Step 9 - Enter Trend by Controllability interaction

Y' = -.797(L)** + -.430(S) + -.528(C) + .009(LXC) +
.030(CXS)* + .006(LXS) + -4.762(T1) + -5.484(T3) +
-2.822(EG) + .300(EGXT3) + .366(EGXT1) + .055(EGXL) +
-.001(EGXC) + .092(EGXS) + .084(T3XS) + .090(T1XS) +
.104(T1XL) + .636(T3XL)** + -.364(T3XC)** +
.113(T1XC) + 30.01

Step 10 - Enter Trend x Stability x Employee Gender

Y' = -.795(L)** + -.399(S) + -.535(C) + .009(LXC) +
.030(CXS) + .005(LXS) + -5.558(Tl) + -3.791(T3) +
-2.491(EG) + -2.795(EGXT3) + 2.078(EGXT3) +
.055(EGXL) + .010(EGXC) + .062(EGXS) + -.016(T3XS) +
.145(T1XS) + .098(T1XL) + .633(T3XL)** +
-.365(T3XC)** + .114(T1XC) + .193(T3XSXEG) +
-.106(T1XSXEG) + 29.695

Step 11 - Enter Trend X Locus X Employee Gender

Y' = -.842(L)** + -.410(5) + .547(C) + .009(LXC) +
.030(CXS)** .005(LXS) + -6.851(T1) + -5.027(T3) +
-4.394(EG) + -.308(EGXT3) + 4.907(EGXT3) + .155(EGXL)
+ .018(EGXC) + .085(EGXS) + -.010(T3XS) + .154(T1XS)
+ .183(T1XL) + .717(T3XL)** + -.369(T3XC)** +
.109(T1XC) + -.173(T3XSXEG) + -.l43(T1XSXEG) +
-.157(T1XLXEG) + -.150(T3XLXEG) + 30.655

Step 12 - Enter Trend x Controllability x Employee Gender

Y' = -.850(L)** + -.388(S) + .567(C)* + .009(LXC) +
.029(CXS)* + .004(LXS) + -7.642(T1) + -7.105(T3) +
-6.564(EG) + 3.928(EGXT3) + 6.671(EGXT3) + .203(EGXL)
+ .100(EGXC) + .101(EGXS) + .009(T3XS) + .157(T1XS) +
.201(T1XL) + .757(T3XL)** + -.288(T3XC)** +
.142(T1XC) + .131(T3XSXEG) + -.157(T1XSXEG) +
-.l95(TlXLXEG) + —.233(T3XLXEG) + -.159(T3XCXEG) +
-.066(T1XCXEG) + 31.150

Note: * indicates p S .05
Note: ** indicates p S .01

254
or stable was rewarded more than performance which was
external or unstable. There was also a significant
interaction between controllability and stability for
promotion decisions. Above average performance which was
considered controllable and stable was rewarded more than
performance which was either uncontrollable or unstable (or
both).

Employee gender and performance trend did not
significantly affect promotion decisions, and none of the
interactions involving gender were significant. The
interactions for external locus of causality X performance
trend and controllability X performance trend were
significant. As before, controllability interacted with
performance trend to exert a significant effect on promotion
decisions, even when the effects of locus and locus x trend
variables were partialled.

The p weights for equation 8 indicate that the slope of
the regression line for the effect of external locus on
promotion decisions was not significantly different between
improving trends and maintaining trends (p = .055 for T1 X
Locus, n.s.), but the slope for the effect of locus on
promotion was significantly different between declining
trends and maintaining trends (n = .814 for T3 X Locus, p S
.01).

The p weights for equation 9 indicate that the slope
for the effect of controllability on promotion decisions was

not significantly different between improving trends and

255
maintaining trends (p = .114 for T1 X Control, n.s.) but the
difference in the slope for the effect on controllability on
promotion decisions was significantly different between
declining trends and maintaining trends (p = -.365 for T3 X
Control, p S .01).

As before, these results strongly support hypotheses H3
to H4b. The relationship between perceived controllability
or external locus of causality and personnel decisions
varied, depending on observed performance trends. Figures
26 and 27 graphically represent the interactions between
controllability or external locus of causality and
performance trends. They were created by substituting
appropriate group values and means into equations 8 and 9,
and then plotting them for perceptions of external locus of
causality and controllability one SD above and below the
mean of locus or controllability perceptions.

The results of the two repeated measures regression
analyses indicate that when employees are all considered
above average in performance and when information about
causes of performance are provided, then perceptions of the
three attributional dimensions of locus of causality,
controllability, and stability have important effects on
personnel decision making. Above average performance which
is considered to be internal in locus and which is stable is
rewarded more than external or unstable performance,
regardless of the performance trend exhibited. Furthermore,

above average performance which is controllable and stable

256

Promotion Decisions

 

20

 

 

 

 

 

 

o I l _J l L

8 10 12 14 16 18 20
Perceived External Locus

*lmproving +Maintaining +Declining

Eigu:e_z§. Interaction Effect of External Locus of

Causality and Performance Trend for Promotion Decisions

257

Promotion Decisions

 

 

 

 

 

 

 

 

 

 

20
15
10 W
5 1 1 . 1 4
e 10 12 14 16 18 20

Perceived Controllability

"‘“Improving +Maintaining +Declining

Eignte_21. Interaction Effect of Perceived Controllability

and Performance Trend for Promotion Decisions

258
is rewarded with more positive promotion decisions than

performance which is either uncontrollable or unstable.

The results also indicate that even when all employees
are above average in performance, observed performance
trends interact with perceptions of locus of causality or

controllability to strongly affect personnel decision

making. Improving or maintaining performance which is

controllable or internal is rewarded more than declining

performance which is controllable or internal. Conversely,

improving or maintaining performance which is uncontrollable

or external is rewarded less than declining performance

which is uncontrollable or external.

The results of previous analyses involving the MANOVA
and ANOVA procedures indicate that employee gender has an

effect on perceived attributions and personnel decision

making. The results of the repeated measures regression

analyses indicate that employee gender has no effect, either

alone or in interactions. While the results cannot prove

that the effect of employee gender on personnel decisions
occurs primarily through attributions, they are suggestive

of that conclusion. Employee gender interacts with

informational cues and performance trend to affect

attributions and personnel decisions. Attributions, in

turn, affect personnel decisions. When the actual

attributions are used to predict personnel decisions,

employee gender no longer has a significant effect.

259
The results of the analyses also suggest that
stereotype-based biases against female employees are
diminished (or even reversed) when decision makers are
provided with extensive performance related information, and

when the employees are all clearly above average in

performance.
It is possible that failure to include actual measures

of stereotypical attitudes toward women in analyses might

change findings involving employee gender. For example, if

a decision maker does not have gender related stereotypes of

women as managers, then employee gender may not have an

effect on decision making processes. It is also possible

that the gender of the decision maker may interact with

 

employee gender to affect personnel decisions. The next
series of analyses were designed to test these
possibilities.

Effects of WAMS and Decision Maker Gender on Personnel
Qecisions

A variety of analyses were conducted to investigate the
possibility that scores on the WAMS or decision maker gender

might interact with the gender of the target employee to

affect personnel decisions and attributions. The basic

findings obtained are presented in the following two
repeated measures regression analyses.

In the first repeated measures regression, pay
decisions were regressed on WAMS scores, decision maker

gender, employee gender, performance trends, attributions,

260

and various interactions. The regressions were exploratory

in nature, so an unusually large number of variables were

used in the analyses.
Perceived stability, locus, and controllability were

entered in the first step, followed by their interactions in

the second step. Next, the gender of the target employee

was entered, and performance trend was entered in the fourth

step. As before, performance trend was represented by two

dummy variables (Trend 1 - improving, Trend 3 = declining).

Gender of the decision maker was entered in the fifth step.

In the sixth step, WAMS scores were entered. Gender of

the decision maker was entered previously to partial its

effects prior to the entry of WAMS scores because the two

were strongly related (f = .56). Next, the variables

representing employee gender X trend, decision maker gender

X trend, and WAMS scores X trend interactions were entered.

In the eighth step, variables representing the employee
gender X decision maker gender, and the employee gender X

WAMS interactions were entered. The triple interaction

terms for employee gender X decision maker gender X
performance trend effects were entered in the ninth step.

In the tenth step, the triple interaction for employee

gender X WAMS X performance trend was entered. Next, the

terms for the interaction between performance trend and

stability were entered, followed by the interaction between

performance trend and locus. In the thirteenth step, the

interaction between performance trend and controllability

261

was entered. The next step entered the employee gender X

attributions terms, the decision maker gender X attributions

terms, and the WAMS scores X attributions terms. The

fifteenth step entered the employee gender X trend X

attributions terms, followed by the decision maker gender X

trend X attributions terms. The seventeenth step contained

the WAMS X trend X attributions terms, and the final step
contained the decision maker gender X employee gender X

attributions and the WAMS X employee gender X attributions

terms.
The results of this analysis are presented in Table 45.

For conservation of space, the regression equations

presented in Table 46 correspond only to those steps which

significantly incremented R?. .Any non-significant blocks of

variables do not have equations completely written out for

them.
As before, main effects for perceived external locus of

causation and stability were obtained. The more internal

and stable the above average performance, the more positive
the pay decisions. Interestingly, a main effect for WAMS

scores was also obtained. The higher the WAMS score, the

more positive the pay decisions made by the decision maker.

This finding suggests that people who have stereotypical
attitudes toward women as managers tend to differ in the

personnel decisions they make from people who do not have

such attitudes toward women as managers.

262

Table 45

Besnlts of Repeated Meaeures Regressions for Pay

' ' ns Usin WAMS Scores Decision Maker Gender Em lo ee

genget, Eerformance Trend. and Attributione

 

Indpndnt I? for R2 Incrmnt Incrmnt
§tep Vars Eguation Incrmnt Btwn Within
1 Control (C) .120 .120 .165**

Locus (L)

Stability (S)

2 L X S .124 .004 .006

C X S

L X C
3 Employee .125 .000 .001

Gender (EG)
4 Trend 1 (T1) .125 .000 .001

Trend 3 (T3)

5 Decision Maker .125 .000 .000
Gender (DG)

6 WAMS .160 .034 .127**

7 EG X T1 .166 .007 .024

EC X T3

DG X T1

DG X T3

WAMS X T1

WAMS X T3
8 EG X DC .169 .002 .008

EC X WAMS
9 EG X DG X T1 .170 .001 .005

EC X DG X T3
10 EG X WAMS X T1 .173 .003 .010

EC X WAMS X T3
11 T1 X S .178 .005 .007

T3 X S
Note: * indicates p s .05

S .01

Note:

** indicates p

Table 45 (cont'd)

12

13

14

15

16

17

18

Note:
Note:

T1
T3

T1
T3

EG
EG
EG
DG
DG
DG
WAMS
WAMS
WAMS

><><><><><>< XX XX
xxxormnrwz GO t"t"

T1
T3
T1
T3
T1
T3

T1
T3
T1
T3
T1
T3

T1
T3
T1
T3
T1
T3

EG
EG
EG
EG
E6
E6

><><><><><>< ><><><><><>< ><><><><><>< ><><><><><><

* indicates p S .05
** indicates p 5 ~01

OOt'HfiUJUI (301-thaw:
><><><><><>< ><><><><><>< ><><><><><>< 01-103

nnbbmm

EG
ES
EG
EG
EG
EG

DG
DG
DG
DG
DG
DG

WAMS
WAMS
WAMS
WAMS
WAMS
WAMS

XXXUIOL"
IDOL"

.287

.316

.318

.325

.330

.336

.340

.109

.030

.002

.007

.005

.006

.004

.149**

.041**

.002

.009

.007

.008

.005

264

Table 46

Regression Eguations for Prediction of Pay Decisions
Qefng WAMS Scores, Decision Maker Gender- Employee
eengep, Berformance Trend, and Attributions

Step 1 - Enter Attributions
Y' = -.539(L)** + .291(S)** + -.079(C) + 37.317

 

 

Step 6 - Enter WAMS scores

Y' = -1.324(L)** + -.665(S) + ~1.107(C)* + .026(LXC)* +
.026(LXS) + .038(CXS) + .201(EG) + .279(T1) +
.144(T3) + -1.921(DG)** + .137(WAMS)** + 47.506

Step 12 - Enter Trend by External Locus interaction

Y' = -1.790(L)** + -l.032(S)* + -.985(C)* + .021(LXC) +
.031(LXS) + .033(CXS) + 2.380(EG) + 8.025(T1) +
-14.229(T3) + -3.278(DG)* + .180(WAMS)** +
-l7.089(EGXT3) + -ll.706(EGXT1) + 2.833(DGXT1) +
4.129(DGXT3) + -.156(WAMSXT1) + -.160(WAMSXT3) +
-.749(EGXDG) + -.011(EGXWAMS) + .686(EGXDGXT1) +
-1.670(EGXDGXT3) + .117(EGXWAMSXT1) +
.487(EGXWAMSXT3) + .216(T1XS) + .542(T3XS)* +
.124(T1XL) + 1.245(T3XL)** + 55.641

Step 13 - Enter Trend by Controllability interaction

Y' = -1.493(L)** + -.751(S) + -.548(C) + .020(LXC) +
.025(LXS) + .026(CXS) + 4.016(EG) + 11.795(T1) +
9.030(T3) + -3.6l3(DG)* + .200(WAMS)** +
-15.929(EGXT1) + -l7.529(EGXT3) + 2.918(DGXT1) +
4.612(DGXT3)* + -.200(WAMSXT1)* + -.181(WAMSXT3)* +
-.515(EGXDG) + -.O32(EGXWAMS) + .493(EGXDGXT1) +
~2.474(EGXDGXT3) + .169(EGXWAMSXT1) +
.298(EGXWAMSXT3) + .184(T1XS) + .298(T3XS)* +
.111(T1XL) + .838(T3XL)** + .044(T1XC) +

-.806(T3XC)** + 42.188

Note: * indicates p S .05
Note: ** indicates p S .01

265
Significant interactions which were consistent with

earlier findings were obtained for the effects of trend X

locus and trend X controllability on pay decisions. There

was little or no evidence to suggest that either the gender

of the decision maker or WAMS scores interacted with

employee gender to affect pay decisions.
A final repeated measures regression analysis was

performed for promotion decisions. In this analysis,

promotion decisions were regressed on WAMS scores, decision

maker gender, employee gender, performance trends,

attributions, and various interactions. As before, the

regressions were exploratory in nature, so an unusually

large number of variables were used in the analyses. The

same variables used in the analysis for pay decisions

described above were used for promotion decisions. The

order in which variables were entered was identical to the

previous analysis. The results of this analysis are

presented in Table 47. The corresponding regression

equations are presented in Table 48. In the interest of

conservation of space, only equations which correspond to

blocks of variables which significantly increased R2 are

presented.
Main effects were obtained for the effects of stability

and locus of causality on promotion decisions. Above

average performance which was considered to be stable and
internal was rewarded more than when it was considered to be

unstable or external. There was also a significant

266

Table 47
3g§p1;§_of Repeated Measures Regressions for Promotion

Qggisions Using WAMS Scores, Decision Maker Gender, Employee

fipnd r1 Performance TrendJ and Attributions

 

Indpndnt I? for R2 Incrmnt Incrmnt
Spgp Vars Eguation Incrmnt Btwn Within
1 Control (C) .118 .118 .153**
Locus (L)
Stability (S)
2 L X S .133 .015 .019**
C X S
L X C
3 Employee .134 .001 .006
Gender (EG)
4 Trend 1 (T1) .134 .000 .000
Trend 3 (T3)
5 Decision Maker .139 .005 .020*
Gender (DG)
6 WAMS .140 .001 .005
7 EG X T1 .143 .004 .016
E6 X T3
06 X T1
DG X T3
WAMS X T1
WAMS X T3
3 EG X DC .145 .002 .008
EC X WAMS
9 EG X DG X T1 .152 .007 .032*
EG X DG X T3
10 EC X WAMS X T1 .155 .002 .011
EC X WAMS X T3
11 T1 X S .157 .002 .002
T3 X S
.05

Note: * indicates p 5
Note: ** indicates p s .01

267

Table 47 (cont'd)

12 T1 X L .258 .101 .132**
T3 X L
13 T1 X C .276 .018 .023**
T3 X C
14 EG X S .280 .004 .005
E6 X L
EG X C
DG X 8
06 X L
DC X C
WAMS X S
WAMS X L
WAMS X C
15 T1 X S X EG .284 .003 .004
T3 X S X EG
T1 X L X EG
T3 X L X EG
T1 X C X EG
T3 X C X ES
16 T1 X S X DG .303 .019 .025**
T3 X S X DG
T1 X L X DG
T3 X L X DG
Tl X C X DG
T3 X C X DG
17 T1 X S X WAMS .309 .007 .008
T3 X S X WAMS
T1 X L X WAMS
T3 X L X WAMS
T1 X C X WAMS
T3 X C X WAMS
18 EC X DG X L .316 .007 .009
EG X DG X C
EG X US X 8
EG X WAMS X L
EG X WAMS X C
EG X WAMS X S

Note: * indicates p s .05
Note: ** indicates p s .01

268
Table 48

Eggrggsipn Egpations for Prediction of Promotion Decisions
ggipg WAMS Scores, Decision Maker Gender, Employee

figpger, Performance TrendJ and Attributions

Step 1 - Enter Attributions
Y' = -.316(L)** + .240(S)** + .03l(C) + 13.969

Step 2 - Enter interactions in Attributions
Y' = -.684(L)** + -.462(S) + -.887(C)** + .016(LXC)* +
.041(CXS)** + .009(LXS) + 29.319

Step 5 - Enter Gender of Decision Maker

Y' = -.621(L)** + -.382(S) + -.849(C)** + .015(LXC) +
.005(LXS) + .039(CXS)** + -.370(EG) + -.026(Tl) +
-.025(T3) + -.745(DG)* + 28.364

Step 9 - Enter interactions for Gender of Decision Maker by

Gender of Employee by Performance Trend
Y' = -.651(L)* + -.438(S) + -.926(C)** + .016(LXC)* +
.006(LXS) + .042(CXS)** + -2.950(EG) + 1.153(Tl) +

-.397(T3) + -2.338(DG)* + .021(WAMS) + 3.048(TGXT1)*

+ 1.423(TGXT3) + 3.925(DGXT1)** + 1.874(DGXT3) +
-.O40(WAMSXT1) + -.009(WAMSXT3) + 1.030(EGXDG) +

.019(EGXWAMS) + -4.577(EGXDGXT1)* + -2.004(EGXDGXT3)

+ 28.867

Step 12 - Enter Trend by External Locus interaction

Y' = -.936(L)** + -.639(S) + -.832(C)* + .011(LXC) +
.OlO(LXS) + .039(CXS)* + .591(EG) + 4.545(Tl) +
-13.551(T3)* + -2.600(DG)* + .041(WAMS) +
-6.877(EGXT1) + -2.351(EGXT3) + 5.355(DGXT1)* +
2.363(DGXT3) + -.ll7(WAMSXTl) + -.038(WAMSXT3) +

1.688(EGXDG) + -.022(EGXWAMS) + -6.829(EGXDGXT1)* +

-2.991(EGXDGXT3) + .122(EGXWAMSXT1) +
.046(EGXWAMSXT3) + .148(TlXS) + .229(T3XS)* +
.038(T1XL) + .810(T3XL)** + 55.641

.05
.01

Note: * indicates p
Note: ** indicates p

IA IA

269

Table 48 (cont'd)

Step 13 - Enter Trend by Controllability interaction

Y' = -.802(L)** + -.500(S) + -.641(C)* + .0ll(LXC) +
.007(LXS) + .035(CXS)* + 1.150(EG) + 4.260(Tl) +
-2.757(T3) + -2.710(DG)* + .047(WAMS) + —8.947(EGXT1) +
-2.353(EGXT3) + 5.3ll(DGXTl)* + 2.538(DGXT3) +
-.139(WAMSXT1)* + -.046(WAMSXT3) + 1.762(EGXDG) +
-.029(EGXWAMS) + -6.848(EGXDGXT1)* + —3.336(EGXDGXT3) +
.147(EGXWAMSXT1)* + .050(EGXWAMSXT3) + .l47(TlXS) +
.111(T3XS)* + .077(T1XL) + .621(T3XL)** + .lO7(TlXC)

+ -.380(T3XC)** + 28.278

Step 16 - Enter Trend by Attributions by Gender of Decision

Maker interactions

Y' = -l.l38(L)* + -.362(S) + -.633(C) + .015(LXC) +
.003(LXS) + .031(CXS)* + -3.802(EG) + -1.927(Tl) +
-3.590(T3) + -4.638(DG) + .035(WAMS) + -3.6l9(EGXTl) +
2.399(EGXT3) + 12.194(DGXT1) + 3.042(DGXT3) +
-.126(WAMSXT1) + -.043(WAMSXT3) + 1.840(EGXDG) +
-.044(EGXWAMS) + -6.751(EGXDGXTl)* + -3.739(EGXDGXT3) +
.l49(EGXWAMSXTl) + .065(EGXWAMSXT3) + .413(T1XS) +
.359(T3XS) + .256(T1XL) + .483(T3XL)* + .018(T1XC) +
-.471(T3XC)* + .215(EGXL) + -.055(DGXL) + .115(EGXC)
+ -.l72(DGXC) + .094(EGXS) + .325(DGXS) + .001(WAMSXC)
+ -.002(WAMSXS) + .003(WAMSXL) + -l.39(EGXTlXL) +
-.329(EGXT3XL) + -.067(EGXT1XC) + -2.69(EGXT3XC) +
-.l70(EGXTlXS) + .178(EGXT3XS) + -.201(DGXT1XL) +
.442(DGXT3XL) + .189(DGXT1XC) + .304(DGXT3XC) +
-.427(DGXT1XS) + -.717(DGXT3XS)* + 31.536

Note: * indicates p S .05
Note: ** indicates p s .01

270
interaction between locus and controllability and
controllability and stability. Above average performance
which was considered to be due to internal and controllable
causes was rewarded more than when it was considered to be
external, uncontrollable, or both. Also, above average
performance which was due to stable and controllable causes
received more positive promotion decisions than when it was
either unstable, uncontrollable, or both, unstable and
uncontrollable.

Gender of the decision maker also affected promotion
decisions. Female decision makers tended to make lower
promotion decisions than male decision makers (p = -.745, p
S .05). This effect was qualified by the significant triple
interaction for the employee gender X decision maker gender
X performance trend effect. Female employees received
significantly less positive promotion decisions from female
decision makers than when either the employee or the
decision maker was male, and this was particularly true for
the improving performance trend condition (p = -4.566, p S
.05).

Consistent with all previous findings, results show
that performance trends interacted with perceptions of
external locus of causality and controllability to affect
promotion decisions. There was also a significant
interaction between the gender of the decision maker,
stability, and performance trend. The slope of the

relationship between stability and promotion decisions in

271
the declining performance trend condition was significantly
different for male and female decision makers (p = -.717, p
S .05). Females tended to give lower promotion decisions
for declining performance which was due to stable causes.

In sum, the two regressions which explored for the
effects of WAMS and decision maker gender on personnel
decisions support the following statements. Attributions
have a strong effect on personnel decisions, and
controllability interacts with locus and stability to affect
promotion decisions. Also, external locus and
controllability interact with performance trends to affect
personnel decision making.

The gender of the decision maker affects promotion
decisions such that female decision makers tend to give less
positive decisions than male decision makers. This is
particularly true when the employee is female and improving,
or when the performance is declining and due to stable
causes.

There was no evidence to support the notion that
attitudes toward women as managers would interact with the
gender of the employee to affect personnel decision making.
The results did suggest however, that people who have
stereotypical attitudes toward women as managers tend to
make personnel decisions differently than people who do not

have such attitudes.

272

summary pf Besplts

The results of the MANOVAs and ANOVAs for the effects
of gender of the employee, informational cues, and
performance trend on attributions indicate that effort,
ability, and ease of job informational cues vary in locus of
causality, controllability, and stability. These findings
are consistent with attribution theory. Effort and ability
cues are both considered to be more internal than ease of
job cues. Also, effort is considered to be more
controllable than ability, which is considered to be more
controllable than ease of job. Finally, effort as a cause
of performance, is considered to be more unstable than
either ability or ease of job as a cause of performance.

Gender of the employee interacts with informational
cues to affect perceived locus and controllability. Effort
cues for female employees are considered to be more internal
and controllable than are the same cues for male employees.
Gender of the employee also interacts with performance trend
to affect perceived stability. Maintaining performance
trends of female employees are considered to be more stable
than are the same performance trends for male employees.

The results of the MANOVAs and ANOVAs also indicate
that informational cues interact with the gender of the
employee to affect personnel decisions. Female employees
who exhibit above average performance trends which are due

to task or job characteristics are rewarded more than male

273
employees who exhibit above average performance trends which
are due to the same causes.

The results also suggest that informational cues
interact with performance trends to affect personnel
decisions through the three attributional dimensions
proposed by Weiner (1985, 1986). Improving or maintaining
performance which is due to uncontrollable or external
causes is rewarded less than the same performance which is
due to controllable or internal causes. Conversely,
declining performance which is due to uncontrollable or
external causes is rewarded more than declining performance
which is due to controllable or internal causes.
Perceptions of stability do not interact with performance
trends to affect personnel decisions, but they do have a
main effect on those decisions. Employees who show above
average performance which is considered to be due to stable
causes receive more reward than when the causes are
considered to be unstable. These findings suggest that the
differences in personnel decisions found in past studies
using effort and ability cues were not only due to perceived
stability, but that they were also due to the effects of
controllability.

The results also suggest that the effects of employee
gender on personnel decisions occur through attributional
processes. When the actual perceptions of controllability,
stability, and locus are used to predict personnel

decisions, employee gender effects disappear.

274

Finally, WAMS scores do not interact with the gender of
the employee to affect personnel decision making. However,
gender of the decision maker does interact with the gender
of the employee to affect personnel decisions. When both
the decision maker and the employee are female, the employee
will receive less positive personnel decisions, especially
if the employee exhibits an improving or declining
performance trend. The main effect for WAMS scores suggests
that people who have stereotypical views of women as
managers differ in the types of personnel decisions they
make from those who do not have such attitudes.

A summary of the results for the tests of the various
hypotheses is provided in Table 49. Most hypotheses
involving gender of the employee or gender group membership
are not supported, while nearly all hypotheses involving the
predicted effects of informational cues and attributions on
personnel decisions are supported. The implications of

these findings will be discussed in the following section.

Table 49

275

Qppppmpg for Tests of Hypotheses

o esis

8a-8b

8c-8d

9a-9b

10

lOa-lOe

11-11b

Test

MANOVA,
ANOVA

MANOVA,
ANOVA,
Regression
MANOVA,
ANOVA,
Regression
ANOVA
ANOVA

MANOVA,
ANOVA

ANOVA

MANOVA,
ANOVA

ANOVA

ANOVA
ANOVA
ANOVA
MANOVA,
ANOVA

MANOVA,
ANOVA

MANOVA,
ANOVA,
Regression

Result

Limited support

Strong support

Strong support

No support
No support

Strong support

Limited support
Limited support
Findings conflict
with hypotheses
No support
Limited support

Findings conflict
with hypotheses

Limited support
Findings conflict
with hypotheses

No support

276
Table 49 (cont'd)

12-12d MANOVA No support
13-13b MANOVA No support
l4-l4e MANOVA No support

15-15b MANOVA No support

DISCUSSION

The analyses of this study resulted in several major
findings. First, the impact of performance trends on
personnel decisions is negligible in the presence of
specific information related to causes of observed
performance trends. Second, informational cues of effort,
ability, and ease of job vary in perceived locus of
causality, controllability, and stability as hypothesized by
attribution theory. Third, perceived locus of causality,
controllability, and stability are linked to personnel
decisions. Specifically, causes of above average
performance which are considered internal, stable, or
controllable and stable, lead to more positive personnel
decisions than when causes are considered external,
unstable, or uncontrollable and unstable.

Fourth, the dimensions of locus of causation and
controllability interact with observed performance trends to
strongly affect personnel decision making. Decision makers
give more reward to employees who show improving or
maintaining performance trends which are perceived to be
controllable or internal than to employees who show
improving or maintaining performance trends which are

perceived to be uncontrollable or external. Conversely,

277

278
decision makers give less reward to employees who show
declining performance trends which are perceived to be
controllable or internal than to employees who show
declining performance trends which are perceived to be
uncontrollable or external.

Fifth, employee gender influences personnel decision
making and attributional formation, but these effects are
not consistent with hypotheses presented earlier. Female
employees receive more positive pay decisions than male
employees when their observed performance trend is
attributable to job characteristics. Decision makers
consider informational cues for effort to be more
controllable for female employees than for male employees,
and they perceive causes of maintaining performance trends
to be more stable for female employees than for male
employees. Findings also suggest employee gender effects on
personnel decisions occur through attributional formation,
because such effects disappear when attributional
perceptions are used to predict personnel decisions.

Sixth, gender group membership does not appear to
affect attributions nor does it seem to affect personnel
decision making. These findings suggest that either group
membership attributional biases do not extend to conditions
of gender group membership, or that gender group membership
biases can be ameliorated by providing large amounts of

performance related information.

 

279

Seventh, the results indicate that attitudes toward
women as managers do not interact with the gender of above
average employees to affect personnel decisions, though such
attitudes do predict pay decisions. Decision makers who
have more positive attitudes toward women as managers make
more positive pay decisions for all employees than decision
makers who have less positive attitudes toward women as
managers.

Finally, the gender of the decision maker does not
affect pay decisions, but it does affect promotion
decisions. Female decision makers give less positive
promotion decisions to all employees than do male decision
makers. Also, the gender of the decision maker interacts
with the gender of the employee to affect pay decisions.
Employees receive lower pay decisions when both the decision
maker and employee are female, and when employees show
improving performance trends.

While the basic findings of this study are summarized
above, elaboration and explanations of these findings will
be discussed in the following sections. The implications of
obtained findings for attribution theory, gender related
research, and group membership studies will be discussed,
followed by a section on the generalizability of the
findings. In the section following generalizability issues,
the implications of findings for "real world" situations

will be reviewed. Finally, the discussion will close with a

280
review of the limitations of this study and suggestions for
future research directions.
appripppipn Theopy

The findings provide strong support for the existence
of a three dimensional classification system for causes of
performance. Causes of outcomes vary along the dimensions
of locus of causality, stability, and controllability as
hypothesized in Figure 2. Calls for the integration of
controllability as an attributional dimension in
organizational research have been made (e.g. Ford, 1985;
Ilgen & Klein, 1988), but few researchers have integrated
this dimension into their work.

Previous research using locus of causality and
stability as attributional dimensions has overlooked the
influence of controllability. As a result, the influence of
stability attributions often has been posited to account for
observed differences in the effects of ability and effort on
leader responses to performance. For example, Dugan (1989,
p. 108) stated:

"...altering attributions from stable to unstable

causes raises expectations that goals can be achieved

(Anderson, 1983; Zoeller, Mahoney, & Weiner, 1983).

Managers who attribute subordinate poor performance to

lack of effort, an unstable cause, may be more open in

a feedback process because they believe they can change

a subordinate's level of effort more readily than they

can change ability."

281
In contrast, the results of this study indicate that
interactions between observed performance and effort or
ability cues are primarily due to the influences of
controllability on personnel decision making, not stability.

The fact that controllability has effects on personnel
decisions which are separate from the effects of locus of
causality is also important. As Green and Mitchell (1979)
have pointed out, the perception of responsibility (i.e.
controllability) is a crucial judgment which leaders make.
They state, "when attributions are external, obviously this
issue is moot, but when attributions are internal,
especially in the case of negative outcomes, it is of utmost
importance." For example, when a person fails for a lack of
motivation, but has high ability, punishment is most severe,
but when failure is seen as due to a lack of ability, but
the person has really tried, the punishment is ameliorated
(Weiner & Kukla, 1970). The findings of this study
generalize this effect to positive personnel decisions made
in response to observed performance trends for above average
employees.

The results of the study also support the view of
leaders as information processors which has been suggested
by Green and Mitchell (1979). In their two step model,
leaders first make an attribution regarding the cause of an
observed performance, then they respond with behaviors
directed toward the cause of that performance. Results

clearly show that decision makers use causal information

282
regarding the observed above average performance trends. In
fact, contrary to expectations, the results of this study
suggest that performance trends have little or no effect on
personnel decisions separate from interactions with the
causal information available to the decision maker.

The results of the pilot studies conducted as a part of
this research also provide evidence for the existence of the
discounting effect. The discounting effect occurs because
leaders look at multiple causes for the same event. The
role of any given cause in creating an observed outcome is
discounted if other plausible causes are also present
(Kelley, 1973; Ajzen & Fishbein, 1978). In this study,
patperns of attributions had to be manipulated in order to
measure each of the causes reliably and separately from each
other.

In sum, the existence and importance of controllability
as an attributional dimension is demonstrated by the results
of this study. Furthermore, previous findings involving
causal attributions are generalizable to situations in which
above average employees exhibit different patterns of
performance. These findings support the view of leaders or
decision makers as information processors. Since leaders or
decision makers are viewed as information processors,
information about observed performance trends for above
average performers should affect personnel decision making.

The next section discusses the implications of findings

283
involving the effect of observed performance trends on
personnel decision making processes.
Engprmappe Trend

The results of the study indicate that in the presence
of specific causal information, observed performance trends
for above average employees have no effect on personnel
decisions which are separate from those involving causal
information. These findings were unexpected and they stand
in contrast to findings obtained by DeNisi and Stevens
(1981), who found that ascending performance was viewed as
more favorable than descending performance.

The differences in the findings of this study with
those of the previous study can be synthesized in the
following manner. The view promulgated by Green and
Mitchell (1979) and Lord and Maher (1991) is that leaders or
decision makers function as information processors. In the
absence of performance information, or when it is ambiguous,
decision makers use schemas to guide their decision making
processes. The two step model by Green and Mitchell (1979)
proposes that first, leaders make causal attributions about
observed performances, and second, they make responses on
the basis of those attributions.

In the DeNisi and Stevens (1981) study, they did not
provide any information regarding the possible causes of the
observed performance. In this situation, decision makers
could only attend to the information available and use

schemas to guide their subsequent decisions. DeNisi and

 

284
Stevens noted that raters seemed to equate ascending
performance with high performance and descending performance
with low performance. Raters apparently assumed that the
improving performer would continue to improve and the
declining performer would continue to decline.

In the current study, decision makers were provided
with detailed and specific information regarding causes of
the observed employee performance, the overall level of
employee performance, and the trend of the observed
performance. In this circumstance, all employees were
clearly viewed as above average in performance. DeNisi and
Stevens (1981) found that level of performance was found to
be the most important determinant of performance
evaluations. In addition to this, information about the
causes of observed performances strongly affected the
attributions made by the decision makers, the first step of
the two step model proposed by Green and Mitchell (1979).
The impact of the provision of specific performance related
information was to reduce the use of schemas (good performer
v. bad performer) by decision makers and to increase the use
of specific causal information. Thus, performance trends
had no effects separate from the causal information
provided. These findings are not necessarily at odds with
those obtained in previous studies. In fact, this study
indirectly supports previous findings by showing that

improving and maintaining performance trends which are

285

internal and controllable are rewarded more than declining
performance trends which are internal and controllable.

While the findings of the study involving performance
trend effects are not incongruent with previous studies,
findings involving the effects of employee gender,
participant gender, and attitudes toward women as managers
are more difficult to interpret. The next section reviews
the implications of the findings involving the effects ‘

mentioned above for gender stereotype research.

 

Gender Stereotype Research |

The current study was designed to determine if previous
findings involving gender related biases can occur when
decision makers make decisions about above average employees
who exhibit different performance trends, and when specific
causal information is provided. The findings indicate that
employee gender exerts some effect on the formation of
attributions and personnel decision making, but the effect
is not always in a manner consistent with earlier
predictions. For example, effort cues are perceived to be
more controllable for above average female employees than
for male employees. While this was not predicted, previous
research has obtained results which can explain these
findings. Feldman-Summers and Kiesler (1974) found that
subjects attributed greater motivation to the female
stimulus person than to her male counterpart, regardless of
other factors. 0n the other hand, subjects attributed

greater ability to the male stimulus person than to the

286

female counterpart. Since motivation, or effort, is
considered to be more controllable than is ability, it is
likely that effort perceptions will increase the perceived
controllability for the performance of above average female
employees.

Informational cues also interact with employee gender
to affect personnel decisions. Above average female
employees whose performance is attributable to the effects ‘

of job or task characteristics receive more positive pay

 

decisions than above average male employees whose l
performance is attributable to the same factors. This
statement is particularly true when the employees are above
average overall, but are declining in performance trend.

Part of the reason for this effect may be explained by
research by Hogan (1987). Hogan has found that prior
expectations interact with actual performance to affect
ratings, implying that supervisors respond negatively to
disconfirmation of expectations. Hogan's findings are also
consistent with the expectancy-violation theory (Jussim,
Coleman, & Lerch, 1987) which states that an employee will
be evaluated more extremely in the direction of the
violation of a stereotyped based expectancy.

If male employees are expected to do better than
females, and their good performance is expected to be due to
ability or effort, then they are likely to receive less
positive personnel decisions when their performance is

attributed to characteristics of the job. This will be

287
particularly true if they are declining in performance trend
because of job characteristics. Thus, it may not be that
decision makers inflate positive personnel decisions for
female employees in such conditions, but rather, decision
makers depress personnel decisions for male employees in
such situations.

It is interesting that WAMS scores did not interact
with employee gender to affect personnel decision making.
Part of the reason this interaction did not occur may be
that, in general, gender stereotypes are not available or
used by the decision maker when large amounts of unambiguous
performance related information is available (Deaux & Lewis,
1984: Fiske & Neuberg, 1990; Heilman, 1983: Nieva & Gutek,
1980). As Heilman (1983) has noted, gender is unlikely to
be viewed as the ultimate source of information about an
individual employee when more job-relevant information is
available (Renwick & Tosi, 1978), when that information is
clear~cut and not subject to reinterpretation (Pheterson et
al., 1971), or when the causal basis of performance is known
(Heilman & Guzzo, 1978).

If a gender stereotype is not activated or used, then
the nature of employee gender becomes an irrelevant fact,
and it would not be expected to interact with other factors
or variables. However, employee gender effects were found
to exist in the context of this study. Thus, additional
explanations have to be sought to explain why WAMS scores

were not found to interact with employee gender to affect

288
personnel decisions. Research by Cohen and Leavengood
(1978) provides some possible answers.

Cohen and Leavengood (1978) found a significant gender
bias against female employees in their study, but the WAMS
was not useful in predicting such bias. They suggested that
part of the reason the WAMS did not relate to bias in
personnel decisions was because internal attitudes and
public behaviors may be under two different kinds of control
processes. Internal attitudes are private and they may or
may not be related to external or public behaviors, thus,
there may be little relation between the two (Calder & Ross,
1973, cited in Cohen & Leavengood, 1978). According to
Cohen and Leavengood (1979), the WAMS is a measure of sex-
role stereotypic attitudes only - it can't measure actual
behavior or discrimination.

It is also interesting to note that the WAMS does
predict personnel decisions, even after controlling for the
effects of decision maker gender. This finding suggests
that people who have stereotypical views of women as
managers tend to make decisions differently than do people
who do not have such views of women. While such attitudes
may or may not predict actual discriminatory behaviors, if
such views are related to other personality traits like
authoritarianism and dogmatism, then there is research to
suggest that these other personality traits will affect the
formation of attributions and personnel decision outcomes

(Carroll, Perkowitz, Lurigio, & Weaver, 1987; Dovidio,

289
Campbell, & Kahn, 1982; Feather, 1985). Currently, however,
it is not entirely clear why WAMS scores would predict
general decision making processes but not interact with
gender of the employee, unless employee gender was truly
unimportant to most decision makers in the given context of
the study.

This last statement may be true, to a large degree,
given the types of performance information provided in the
current study. Research cited by Heilman (1983) has found
that evidence of a woman's success in a field ordinarily
populated only by men is sufficient to totally eliminate
biased evaluation of her. Similarly, Hamner, Kim, Baird,
and Bigoness (1974) and Bigoness (1976) have found that
women are rated as superior to men when they are shown to
perform equivalently doing the heavy physical chores of a
grocery store clerk. In the current study, women were
clearly shown to perform comparably to men as managers in a
computer related field.

Gender of the decision maker also has an effect on
personnel decision making. Female decision makers tend to
make less positive personnel decisions overall than do male
decision makers. This effect, however, is qualified by its
interaction effect with the gender of the target employee
and the observed performance trend.

For example, it was found that female decision makers
give less positive promotion decisions to female employees,

especially when they are exhibiting an improving performance

 

 

 

290
trend. It is not clear why this effect would occur, but
when the results are viewed a different way, social identity
theory (Tajfel & Turner, 1986) provides a possible
explanation. If employee gender had been dummy coded in a
reverse fashion, the obtained results would have indicated
that male decision makers give more positive promotion
decisions to male employees, particularly when they are
improving in performance. It is possible that it is not
that female decision makers give less positive promotion
decisions to female employees, but rather, male decision
makers give more positive promotion decisions to male
employees, particularly when they are improving in
performance.

Social identity theory predicts that evaluations of
favorable and unfavorably described in—group targets should
be more extreme than evaluations of similarly described out-
group targets. This is predicted because social identity
needs are best served by extreme evaluations. The "black
sheep hypothesis" states that more favorable evaluations
will be given to favorably described ingroup targets than to
favorably described outgroup targets. If male decision
makers tend to think in terms of ingroup and outgroup gender
relationships, and they believe that improving performance
is a favorable outcome, then social identity theory might
partially explain observed findings. It should be noted,
however, that there was little evidence to suggest such

ingroup/outgroup effects existed for both male and female

 

291
decision makers in the given context of the study. It is
possible that ingroup/outgroup biases only occur for one
gender or the other.

Researchers have found that members of groups who are
in the majority or who are in power tend to use stereotypes
more than members of groups who are in the minority or who
are not in power (Fiske, 1993; Hewstone, 1990). Men are
typically considered to be in the majority or in power for
masculine-typed jobs, so they may be more likely to use
ingroup/outgroup stereotypes than are women. This logic is
consistent with the findings described above.

In sum, there is evidence to suggest that employee and
decision maker gender affects the formation of attributions
and personnel decision outcomes. The obtained findings,
however, are not congruent with predictions, and a
systematic pattern of gender related effects has not emerged
from the findings. In appears that employee gender may not
have been an important attribute to decision makers when
more relevant performance related information was available.
The next section will review the findings obtained for
effects which involve group membership.

embershi Research

It was hypothesized that group membership would affect
attribution formation and personnel decision making
outcomes. No support was found for such hypothesized
effects. Overall, gender group membership effects seem to

be small or nonexistent when all employees are considered

 

292

above average in performance, and when they exhibit
different performance trends due to clearly identifiable
causes. As is the case with use of gender stereotypes,
group membership effects are not likely to occur when there
is performance related information available which reduces
the use of schema based processing (Fiske & Neuberg, 1990).
It is also possible that theories regarding the effect of
ethnic group membership do not generalize to gender group
membership effects. Based on the findings of this study, it
is not possible to distinguish which of these statements are
true, and specific research designed to tease these findings
apart needs to be conducted.
Generalizability Issues

The findings cited above were found in a laboratory
environment using a sample of undergraduate college
students. Furthermore, findings were obtained by having
participants make ratings in response to fictional employee
files. Given these statements, generalizability issues
become important, even though the study was designed to test
a new integrative model. These issues will be dealt with in
four steps. First, the issue of the "paper person" research
paradigm will be discussed. Next, comparisons of findings
involving attributional research to those found using
professional samples will be reviewed, followed by a
discussion of the generalizability of findings involving
gender related biases using college samples. Finally,

characteristics of the sample which participated in this

 

293
study will be used to support the generalizability of
findings.

First, though the "paper-person" research paradigm has
received strong criticism, a meta-analysis by Murphy, Herr,
Lockhart, and Maguire (1986) suggests that findings using
this paradigm may be generalizable. They conducted a
meta-analysis of performance appraisal research that
contrasted the effect sizes of "paper people" with "behavior
observation" studies. No significant differences in the
effects of ratee sex on performance ratings between the two
experimental paradigms was found.

Also, "paper people" are commonly used in business and
organizational environments to make reward, selection, and
placement decisions. Military promotion review boards often
make personnel decisions on the basis of paper
recommendations given by superiors. College admissions,
particularly for graduate students, are also often initially
made on the basis of resumes, transcripts, letters of
recommendation, etc. Many large companies have a
centralized human resource department which makes hiring,
pay, and promotion decisions on the basis of applications or
written recommendations provided by an employee's superior.

As Knowlton and Mitchell (1981) noted, "In many
organizations, performance evaluations serve as the basis
for pay increases, promotions or demotions, and even
continued employment. If leaders or supervisors make

attributions for subordinate behavior that are subject to

 

294
motivational or informational biases and these attributions
affect performance evaluations, then their behavior may be
inappropriate and may have a substantial influence on the
satisfaction and motivation of subordinates and the
effectiveness of the organization" (p. 465). Used in this
context, performance evaluations are paper people, and
attributional processes can play an important role not only
in how those people are evaluated, but also in how that
evaluation results in personnel decisions.

Second, findings involving informational cues and
attributions have been found to occur in a wide variety of
"real world" contexts, using professional samples. For
example, attributional processes were found to occur in
samples which included 24 experienced managers (Gioia &
Sims, 1986), 188 supervisors (Heneman et al., 1989), 25
nursing supervisors (Mitchell & Wood, 1980) and 48 Israeli
mid-level managers (Pazy, 1986). A meta-analysis by DeVader
and Lord (1984, cited in Kinicki & Griffeth, 1985) indicates
that laboratory research on attributional processes are
consistent with findings in the field.

Third, qualitative reviews of sex-bias research have
indicated that professional decision makers render personnel
decisions no less sensitive to sex bias than ratings made by
students (Nieva & Gutek, 1980). Several quantitative
reviews have had similar findings. For example, the Eagly
et a1. meta-analysis (1992) found no effect size for MBA (d

= -.05)’ manager (g = .01), or college student samples (d

 

295
= -.Ol). The Olian et al. (1988) meta-analysis found that
mean responses of professional and student samples were not
significantly different, although students provided more
homogeneous evaluations in studies of gender and
qualifications.

Fourth, generalizability of the findings using the
present sample to supervisory situations is bolstered by
actual characteristics of the participants. For example,
46% of the sample has previously had some form of work
experience in which they supervised other employees. Of the
46% who have had some supervisory work experience, 18% have
had more than a year's worth of experience.

Based on the discussion above, it can be concluded that
the findings obtained and discussed in this study are likely
to be generalizable to a variety of work contexts. Thus,
these findings are likely to have implications for actual
managerial or supervisory decision making outcomes.
Implications of Findings

The findings have shown that decision makers use causal
information, whenever it is available, to diagnose and
determine appropriate responses to observed performances.
This occurs even in situations in which employees are all
above average in performance and when they exhibit different
patterns of performance trends. Previous research has shown
that lack of information related to causes of performance
outcomes may increase the use of schema based decision

making (Fiske & Neuberg, 1990). Schema based processing, in

 

296
turn, will increase the probability that gender stereotype
based biases will exist, since stereotypes are a form of
person schema (Bierhoff, 1989).

The likelihood that decision makers will engage in the
use of stereotypes when making decisions in organizational
contexts is particularly strong since, as Feldman (1981,
1990) noted, raters and decision makers are likely to be
cognitively busy and have multiple demands on their time.

The high cognitive work loads for decision makers increases

 

the probability that schema based processing will occur.
Also, direct information about subordinates' job behavior is
often fragmentary. Direct personal contact with
subordinates may be minimal and restricted to a particular
set of situations, depending on the nature of the job.
Often jobs themselves are incompletely understood, and
specific duties may be inadequately described or entirely
unspecified, especially at higher managerial levels
(Campbell, Dunnette, Lawler, & Weick, 1970; cited in
Feldman, 1981). Furthermore, research suggests that those
who are in power are more likely to use stereotypes than
those who are not (Fiske, 1993).

A large body of research has shown gender related
biases to exist, particularly for leadership or managerial
positions (Eagly, et al., 1992; Morrison & Von Glinow,
1990). For example, using a sample of 2,812 managers,
Cochran (1993) found that differences in performance

ratings, job qualifications, self-confidence, educational

297
level, could not explain discrepancies in salaries between
male and female managers. Much research has also shown that
many managers view the role of management as a masculine-
typed occupation (e.g. Brenner, et al., 1989), and this is
even true in other countries (Shein & Meuller, 1992). An
equally large body of research has shown that gender related
attributional biases exist, particularly when decision
makers do not have enough clear or unambiguous performance
related information to make effective decisions (e.g. Deaux,
1976, 1984; Heilman, 1983).

The findings in this study suggest that even when
decision makers are provided with ample information related
to the cause of an observed performance, and even when the
performance level of all employees are clearly seen as above
average in performance level, gender related biases can
still occur. It is likely, however, that the effect of
employee gender is substantially reduced in such situations.

Thus, the findings suggest that in order to be fair and
reasonably unbiased when making personnel decisions,
decision makers must have information related to the cause
of the observed performance, even if all employees are seen
as above average in performance. Furthermore, information
which allows leaders or decision makers to diagnose the
amount of controllability or locus of the observed
performance is particularly critical. Decision makers will

provide completely different responses to the same observed

298
performance, depending on the inferred locus and
controllability of that performance.

The implication for organizations is clear. In order
to reduce stereotyped based biases and schema based
processing, several things must happen. First,
organizations must develop methods which allow personnel
decision makers to have access to quality information which
contains not just the overall level of some attained
performance, but perhaps more importantly, the cause of that
performance. Second, organizations must present decision
makers with information about performance trends, not just
the overall level of performance. That is, decision makers
need to have access to histories of performance, not just
the current overall performance level. This too, will help
decision makers to diagnose and respond to some given
performance level. Third, decision makers must be given
enough time to review or deliberate information carefully
prior to making personnel decisions. These procedures will
help to reduce not only the influence of gender stereotypes,
but also reduce the influence of other types of schema based
processing, like ingroup/outgroup attitudes toward
employees.

Other issues which have implications for organizational
contexts are those of self-serving attributional biases
(Zuckerman, 1979), and actor-observer differences (Brown,
1984: Jones & Davis, 1965). Self-serving attributional

biases, actor-observer differences, and managerial decision

 

299
making style can interact to influence an employee's
perceptions of distributive or procedural justice (Cohen,
1987; Folger & Greenberg, 1985; Leventhal, 1980). The
effect of self-serving attributional bias is to enhance an
employee's perception of the internality and controllability
of his or her own successful performance and to decrease the
perception of internality and controllability of his or her
own unsuccessful performance. Actor-observer differences,
however, have shown that observers (managers) are more
likely than actors (employees), to attribute successful
performance of an actor to external and uncontrollable
factors. Conversely, observers (managers) are more likely
than are actors (employees) to attribute unsuccessful
performance of an actor to internal and controllable
factors.

Earlier it was shown that perceptions of locus and
controllability have a strong effect on decision making
regarding employee performance, even given the same
performance trend and overall performance level. If
managers view successful employee performances as due to
external or uncontrollable factors, they are less likely to
provide positive promotions and pay decisions to that
employee. On the other hand, the employee will be likely to
attribute his or her successful performance to internal and
controllable factors, and as a result, will feel that he or
she deserves more positive pay and promotion decisions.

Conversely, if managers view unsuccessful or poor employee

300
performances as due to internal or controllable factors,
they are more likely to engage in punitive behaviors and
less likely to engage in positive pay and promotion
decisions. On the other hand, the employee will be likely
to attribute his or her unsuccessful performance to external
or uncontrollable factors, and as a result, will feel that
he or she does not deserve punitive responses and instead,
deserves positive pay and promotion decisions. The
implications for perceived procedural and distributive
justice, employee satisfaction, and employee motivation are
strong.

Procedural justice is concerned with the perceived
fairness of procedures used in making decisions (Folger &
Greenberg, 1985), while distributive justice is concerned
with the perceived fairness of decision outcomes (Cohen,
1987). Obviously, if managers are making attributions for
performance outcomes differently from the way that employees
are making attributions for those outcomes, and if they are
strongly influenced in their decision making by those
attributions, then the fairness of the decision making
process may be questioned by the employee. Perceived
fairness of decision making procedures and the outcomes of
those decisions have been linked to the satisfaction and
motivation of those affected by the decisions (Greenberg,
1987; Greenberg & Tyler, 1987). Thus, it is not only
important that decision makers receive and use quality

information in the decision making process, but it is

301
probably important to tell those affected by decisions how
that information was used in arriving at some decision
outcome. Specifically, issues of controllability and locus
should be directly addressed and conveyed to others during
the decision making process.

.Limitatippg and Future Directions

This study has demonstrated that performance trends do

 

not affect personnel decisions separate from the effects of
causal information. It has also shown that gender related
effects can still occur, even when extensive performance
related information is provided. What is not clear,
however, is whether gender related biases will change in
form or size when employees are all above average in
performance, they exhibit different performance trends, and
when causal information is pp; provided. In other words, it
is not clear whether removal of causal information regarding
observed performance trends will change the pattern of
gender effects or whether decision makers will respond
differently to observed performance trends. Future research
could provide decision makers with male and female employees
who show different performance trends at different
performance levels, but without providing causal
information, and then investigate how decision makers
respond to that information.

Another limitation of the current research is that,
though evidence exists to support the generalizability of

findings to various work contexts, researchers have not

302
investigated the combined effects of performance trends and
causal attributions on personnel decisions. It is possible
that various environmental factors, like organizational
policy, will overshadow any gender or attributional effects
on personnel decisions. It is also possible that the
experience of the decision maker with performance evaluation
and personnel decision making could change the nature of the
effects obtained in the current study. Future research ‘

should also be conducted to determine the effectiveness of

 

various training procedures to ameliorate or remove gender l
stereotype related biases.

Future research should also investigate how different
types of causal cues might change the effects obtained in
the current study. The current study only used effort,
ability, and ease of job as possible causes of performance.
Weiner (1986, 1987), Anderson (1983), and others have
identified many other potential causes of performance which
decision makers might use in the decision making process.
For example, future research could look at how mood,
attitude, health, and other person cues might affect
personnel decision making processes.

In sum, the current study has proposed and shown that
decision makers function as information processors. It has
also shown that decision makers are affected by perceived
causes of performance, performance trends, and to a certain
degree, gender of the employee. An important contribution

of this study to research on personnel decision making and

303
attribution theory is that perceptions of controllability
have effects separate from those of stability and locus of
causation on personnel decision making processes. Another
important finding is that observed performance trends have
effects on decision making processes, even when the average
performance levels are identical. Results of this study
should stimulate future research in the areas of performance
evaluation and attribution theory which could diminish the
effects of gender related bias, improve organizational
reward practices, and improve the perceived justice of work

environments.

 

APPENDICES

 

APPENDIX A

 

304

Appendix A

Complete Listing of Hypotheses

H1:

H2:

orma r nds on rsonnel Dec's' n

Employees who show improving performance trends will be
more likely to be promoted or to receive monetary
increases than those showing declining trends.
Employees who show maintaining performance trends will
be more likely to be promoted or to receive monetary
increases than those showing declining trends.

Effegg pf Performance Trends and Locus of Causality on

H3:

H3a:

H3b:

Personnel Decisions

The effect of performance trends on personnel decisions
will be moderated by the decision maker's perceptions
of locus of causality.

Those employees who show improving or maintaining
performance trends which are perceived by the decision
maker as due to causes which are internal will receive
more positive personnel decisions than employees who
show improving or maintaining performance trends which
are due to causes perceived to be external, when mean
performance among employees is held constant.

Those employees who show declining performance trends
which are perceived by the decision maker as due to
causes which are internal will be less likely to
receive positive personnel decisions than employees who
show declining performance trends which are due to
causes perceived to be external, when mean performance

among employees is held constant.

 

Effiggt of Perfprmance Trends and Controllability on

H4:

H4a:

H4b:

Personnel Decisions

The effect of performance trends on personnel decisions
will be moderated by the decision maker's perceptions
of controllability.

Those employees who show improving or maintaining
performance trends which are perceived by the decision
maker as due to causes which are controllable will
receive more positive personnel decisions than
employees who show improving or maintaining performance
trends which are due to causes perceived to be
uncontrollable, when mean performance among employees
is held constant.

Those employees who show declining performance trends
which are perceived by the decision maker as due to
causes which are controllable will be less likely to

305

receive positive personnel decisions than employees who
show declining performance trends which are due to
causes perceived to be uncontrollable, when mean
performance among employees is held constant.

Effiggt pf Pgrfopmance Trends on Perceived Stability

H5: Causes of maintaining performance trends for employees
will be perceived by the decision maker to be more
stable than improving or declining performance trends,
when mean performance among employees is held constant.

Effiggt pf Pegformance Trends on Promotion Decisiops

H6: Employees who show maintaining performance trends will
be more likely to receive positive promotion decisions
than employees who show improving or declining
performance trends, when perceived controllability or
locus of causality is held constant.

Effect of Informational Cues on Perceived Controllability
and Locus of Causation

H7: Informational cues regarding causes of observed
performance trends will result in variability of
perceptions along the controllability and locus of
control dimensions of attributions.

H7a: Informational cues of effort will be perceived to be
more controllable than will ability or ease of job
cues.

H7b: Informational cues of ability will be perceived to be
less controllable than effort cues, but more
controllable than ease of job cues.

H7c: Informational cues of ability and effort will be
perceived to be more internal than ease of job cues.

Effigg; pf Performance Trend and Informational Cues on
Perceptions of Stability

H7d: Informational cues of effort, ability, and ease of job
will be perceived to be more stable when associated
with a maintaining trend than when associated with an

improving or declining trend.

Effigcp pf Performance Trends and Employee Gender on
ngcgived Stability and Locus of Causation

H8: Performance trends will interact with gender of the
managerial employee to affect attributions of stability
and locus of causality made by the decision maker.

H8a: Improving or maintaining performance trends for male
managerial employees will be perceived by the decision

 

306

maker to be more stable than will the same performance
trends for female managerial employees.

H8b: Declining performance trends for male managerial
employees will be perceived by the decision maker to be
more unstable than will the same performance trends for
female managerial employees.

H8c: Improving or maintaining performance trends for male
managerial employees will be perceived by the decision
maker to be more internal than will the same
performance trends for female managerial employees.

H8d: Declining performance trends for male managerial
employees will be perceived by the decision maker to be
more external than will the same performance trends for

female managerial employees.

Effigg; pf Performance Trends and Employee Gender op
Perceived Controllability

H9: Performance trends will interact with the gender of the
managerial employee to affect attributions of
controllability made by the decision maker.

H9a: Improving or maintaining performance trends for male
managerial employees will be perceived by the decision
maker to be more controllable than will the same
performance trends for female managerial employees.

H9b: Declining performance trends for male managerial
employees will be perceived by the decision maker to be
more uncontrollable than will the same performance
trends for female managerial employees.

ct of Informational Cues Performance Trend and

Employee Gender on Perceived Controllability and Locus

of Causation

H10: Informational cues will interact with gender of the
employee and performance trend to affect attributions
of controllability and locus of causality made by the
decision maker.

H10a: Informational cues which indicate high ability
attributions for improving or maintaining performance
trends for male managerial employees will be perceived
by the decision maker to be more internal and
uncontrollable than will improving or maintaining
performance trends and high ability informational cues
for female managerial employees.

H10b: Informational cues which indicate low ability
attributions for declining performance trends for
female managerial employees will be perceived by the
decision maker to be more internal and uncontrollable
than will declining performance trends and low ability
informational cues for male managerial employees.

H10c: Informational cues which indicate high effort
attributions for improving or maintaining performance

307

trends for female managerial employees will be
perceived by the decision maker to be more internal and
controllable than will improving or maintaining
performance trends and high effort informational cues
for male managerial employees.
H10d: Informational cues which indicate low effort
attributions for declining performance trends for male
managerial employees will be perceived by the decision
maker to be more internal and controllable than will
declining performance trends and low effort
informational cues for female managerial employees.
Informational cues which indicate high or low ease of
job attributions for improving, maintaining, or
declining performance trends for female managerial
employees will be perceived by the decision maker to be
more external and uncontrollable than will the same
performance trends and informational cues for male

managerial employees.

H108:

Effect of Performance Trends and Employee Gender on
Personnel Decisions

Performance trends will interact with gender of the
managerial employee to affect personnel decisions made
by the decision maker.

H11a: Improving or maintaining performance trends for male
managerial employees will be rewarded more by the
decision maker than will the same performance trends
for female managerial employees.

Hllb: Declining performance trends for female managerial

employees will be rewarded less by the decision maker

than will the same performance trends for male

managerial employees.

Effect of Performance Trends and Gender Group Membership on

Perceived Stability and Locus of Causation

H11:

H12: Performance trends will interact with gender
ingroup/outgroup membership of the managerial employee
to affect attributions of stability and locus of
causality made by the decision maker.

H12a: Improving or maintaining performance trends for an
employee of the same gender as the decision maker will
be perceived to be more stable than will the same
performance trends for an employee of the opposite
gender.

H12b: Declining performance trends for employees of the
same gender as the decision maker will be perceived to
be more unstable than will the same performance trends
for employees of the opposite gender.

H12c: Improving or maintaining performance trends for
employees of the same gender as the decision maker will
be perceived to be more internal than will the same

 

308

performance trends for employees of the opposite
gender.

H12d: Declining performance trends for employees of the‘
same gender as the decision maker will be perceived to
be more external than will the same performance trends
for employees of the opposite gender.

fo ance ends and Gender Grou Memb s ' o

Perceived Controllability

H13: Performance trends will interact with gender
ingroup/outgroup membership of the managerial employee
to affect attributions of controllability made by the
decision maker.

H13a: Improving or maintaining performance trends for
employees of the same gender as the decision maker will
be perceived to be more controllable than will the same
performance trends for employees of the opposite
gender.

H13b: Declining performance trends for employees of the
same gender as the decision maker will be perceived to
be more uncontrollable than will the same performance
trends for employees of the opposite gender.

Effect of Informational Cues, Performance Trend, and Gender

Group Membership on Perceived Controllability and Locus

of Causation

H14: Informational cues will interact with performance
trend and gender ingroup/outgroup membership of the
employee to affect attributions of controllability and
locus of causality made by the decision maker.

H14a: Informational cues which indicate high ability
attributions for improving or maintaining performance
trends for gender ingroup managerial employees will be
perceived by the decision maker to be more internal and
uncontrollable than will improving or maintaining
performance trends and high ability informational cues
for gender outgroup managerial employees.

H14b: Informational cues which indicate low ability
attributions for declining performance trends for .
gender outgroup managerial employees will be perceived
by the decision maker to be more internal and
uncontrollable than will declining performance trends
and low ability informational cues for gender ingroup
managerial employees.

H14c: Informational cues which indicate high effort
attributions for improving or maintaining performance
trends for gender outgroup managerial employees will be
perceived by the decision maker to be more internal and
controllable than will improving or maintaining
performance trends and high effort informational cues
for gender ingroup managerial employees.

 

309

Informational cues which indicate low effort
attributions for declining performance trends for
gender ingroup managerial employees will be perceived
by the decision maker to be more internal and
controllable than will declining performance trends and
low effort informational cues for gender outgroup

managerial employees.
Informational cues which indicate high or low ease of

job attributions for improving, maintaining, or
declining performance trends for gender outgroup
managerial employees will be perceived by the decision
maker to be more external and uncontrollable than will
the same performance trends and informational cues for

gender ingroup managerial employees.

Effggp of Performance Trends and Gender Group Membership on

Personnel Decisions

H14d:

H146:

Performance trends will interact with gender group
membership of the managerial employee to affect
personnel decisions made by the decision maker.

Improving or maintaining performance trends for
gender ingroup managerial employees will be rewarded
more by the decision maker than will the same
performance trends for gender outgroup managerial
employees.

Declining performance trends for gender outgroup
managerial employees will be rewarded less by the
decision maker than will the same performance trends

for gender ingroup managerial employees.

H15:

H153:

H15b:

 

APPENDIX B

 

310

Appendix B

Power Analysis

The following format will be used to indicate the results of
the power analysis. First, a variable and relevant studies
are listed, followed by the effect size (T) obtained in the
results of that study. The p is then listed, which is one
less than the number of expected cells in the new study for
the variable being considered. Next, the a is indicated,
which is the probability of type I error or alpha. Next,

presented and—it was determined from the an examination of
the power tables presented in Cohen (1988), based on the
statistics identified above.

Effect of Performance Trend on Personnel Decisions

DeNisi and Stevens (1981): T = .30 (T = 1/2 p)
p 2, p = .05, power .80: required p = 36 per cell
3 2, a = .05, power required p = 48 per cell

II ll
\0
0

Effect of Target Gender on Personnel Decisions

Eagley et al., meta-analysis (1992), T = .05
Olian et al., meta-analysis (1988), T = .235
Heilman and Stopeck (1985b), T = .229 to .274
Cohen and Bunker (1975), f = .314
f ranges from .05 to .314

T = .10, p = 1, g = .05, power = .80: required p = 375
f = .15, p = 1, g = .05, power = .80: required p = 170
I = .20, p = 1, g = .05, power = .80: required 3 = 96

Attriputions and Leader Decisions

Mitchell and Wood (1980), f .40 .
Q = 10, g = .05, power .80: required p
Q = 10, g = .05, power .90: required p

QQEEQrng Differences in r Between Trends and Personnel
922m

The g statistic is required for comparing differences in
correlations across conditions. Correlations between

0 O 0 be
erformance trend and personnel dec151ons will
Ealculated. A small difference in correlations between the

improving and maintaining performance trends and personnel
decisions is expected, while a large difference in l' .
correlations between improving or maintaining and dec 1n1ng

trends is expected.

10
13

 

 

 

311

From Cohen (1988):

Small g = .10, g = .05, power = .80: required p = 1573

Med g = .30, g = .05, power = .80: required p = 177

Large g = .50, p = .05, power = .80: required p = 66
gpnplpsions

At least 300 to 400 subjects will be needed in each cell in
order to ensure adequate power to test for the effects of
gender on leader decisions. Because power is lower when
testing for interaction effects, power problems are expected
when testing sex of subject by sex of performer by
performance trend effects. Over 1000 subjects may be needed
to test some of the three-way interactions in the proposed
study. The number of subjects needed to test three-way
interactions can be reduced by using a within subjects
design.

APPENDIX C

312

Appendix 0

Results of Pilot Study

Each subject provided one set of responses to the WAMS scale
and four sets of responses to the other scales (perceived
causes, attributional dimensions, personnel decisions).

WAME and demographics

Factor analysis (FA) on the WAMS and demographic items
yielded 6 factors, which were not easily interpretable. FA
using 2, 3, and 4 factor solutions yielded one overall
factor which accounted for most of the variance, and other
factors which consisted of the demographic items. For
example, using a two factor solution, all of the WAMS items
loaded onto the first factor and it accounted for 34.2% of
the variance. All of the demographic items loaded onto
factor two, and they accounted for 6.3% of the variance.
The results suggest the WAMS scale taps into one overall
factor.

Coefficient alpha was computed on the WAMS (.914). Only one
item showed an increase in alpha (.918) after removal.

After content analysis, it was kept for the future study,
since the result could be due to sampling error.

Pepceived Qauses of Performance

Each subject provided four responses to the scales. One
factor analysis was conducted for each set of responses, for
a total of four factor analyses. The initial FA yielded
four, five, and six factor solutions, depending on the set
of responses being analyzed. The factors appeared to be
uninterpretable. After careful examination, it was found
that certain reverse coded items were loading onto their own
factor, in spite of being recoded appropriately. When these
items were removed, the factor structures began to make
sense. Reliability analyses confirmed the results. The
items which were creating odd and uninterpretable factors
were also items which were substantially reducing the
reliabilities of the scales from which they came. Upon
removal, the following factor structures emerged for all
four sets of data.

Factor 1: Consisted primarily of mood/attitude/recent
effort items. Taps into an unstable or changeable affective
component of performance causation. Accounted for 27.9,
30.4, 29.6, 25.9 percent of the variance, respectively.

313

Factor 2: Consisted primarily of ability/consistent effort
items. Taps into a stable or traitlike component of
performance causation. Accounted for 17.3, 10.1, 18.0, 15.9
percent of the variance, respectively.

Factor 3: Consisted primarily of luck/ease of job items.
Taps into an external source of performance causation. When
a three factor solution was derived (two of four sets),
items measuring the influence of superiors on performance
fell into factor 3. Accounted for 11.4, 18.2, 10.1, 12.2
percent of the variance, respectively.

Factor 4: Consisted entirely of the influence of superiors
on job performance when a four factor solution was derived
(two of four sets). Accounted for 7.7, null, null, 9.0
percent of the variance, respectively.

Reliabilities were calculated for each of the individual
perceived causes and also for each of the factors identified
in the FA. Poor items were removed and reliabilities were
recalculated.

Reliabilities (2 items):

Luck: .5185 .5849 .5398 .6623
Constant Effort: .7898 .7033 .7767 .7474
Division Mnger: .5380 .5127 .6309 .5583
Mood: .7407 .8012 .8147 .7096
Ease of Job: .5670 .6066 .5078 .4427
Ability: .7891 .8138 .8543 .7969
SpeCifiC effort: .7122 .8147 .7953 .7375
Reliabilities by Cause Factors:
External (6 items): .6522 .7290 .7103 .6654
Attitude (4 items): .8305 .8729 .8679 .8401
Trait (4 items): .8158 .7869 .7926 .7639

Peppeived Controllability, Locus of Causation, and Stability

Each subject provided four responses to the scales. One
factor analysis was conducted for each set of responses, for
a total of four factor analyses. The initial FA yielded two
and three factor solutions, depending on the set of
responses being analyzed. The factors appeared to be
uninterpretable. After careful examination, it was found
that certain reverse coded items were loading onto their own
factor, in spite of being recoded appropriately. When these
items were removed, the factor structures began to make
sense. Reliability analyses confirmed the results. The
items which were creating odd and uninterpretable factors
were also items which were substantially reducing the
reliabilities of the scales from which they came. Upon

314

removal, the following factor structures emerged for all
four sets of data.

Factor 1: Consisted of both controllability and locus of
causation items. Approximately 35-40% of the variance.

Factor 2: Consisted of stability items. Approximately 15-
20% of the variance.

Factor 3: Consisted of reverse coded items and the
occasional locus of causation items. Approximately 5-10% of
the variance when it appeared (two of four).

It appears that only two factors consistently emerged. They
were locus of causation/controllability and stability
factors.

Reliabilities were calculated only on the original scales,
though the overlap between controllability and locus
perceptions was readily apparent.

Controllability: .8086 .8307 .8321 .7856
Locus of Cause: .7866 .8113 .8067 .7892
Stability: .6936 .7712 .7673 .7540

szspnnel Decisions

Factor analyses indicated one main factor which consisted of
affect, expectancy, promotion, merit pay, and pay raise
decisions. Another factor consisted of affect and
expectancy items or of reverse coded items. The last factor
usually appeared to be garbage. Attempting to extract more
factors did not eliminate the problem.

It appears that two or three main factors exist, but
conclusions are equivocal.
Factor 1: Pay, promotion, reward decisions.
Factor 2: Affect and expectancy.
Factor 3: Reverse coded items.

Reliabilities for the original scales are as follows:

Expectancy: .7697 .7600 .7956 .7703
Raise: .7795 .8032 .7885 .7972
Merit Pay: .8411 .8635 .8314 .8750
Affect: .8638 .8474 .8277 .8268
Promotion: .8696 .9125 .8846 .8558

These scales are all highly intercorrelated and show a high
degree of parallelism in the intercorrelation matrix.

APPENDIX D

 

Appendix D

315

Examples of Experimental Manipulations

Male Manipulations

Ability

Effort

Ease Job

Tmproang

Brian Jones
Store 038

Tom Miller
Store 029

Gary Smith
Store 002

Female Manipulations

Ability
Effort

Ease Job

Improving

Karen Jones
Store 039

Patricia Miller
Store 030

Lisa Smith
Store 004

Maintaining

Gary Miller
Store 040

Brian Smith
Store 005

Tom Jones
Store 017

Maintaining

Lisa Miller
Store 035

Karen Smith
Store 006

Patricia Jones
Store 024

Declining

Tom Smith
Store 021

Gary Jones
Store 018

Brian Miller
Store 001

Declining

Patricia Smith
Store 023

Lisa Jones
Store 019

Karen Miller
Store 003

316
DESCRIPTION OF COMPANY

Organizational Structure of CompuPro. Inc.

Board of Directors

:
v

I --------------- President --------------

I E I

v v v

VP Manfctrng/Prsnnl VP Oprtns/Sales VP Engnrng/Planning
I
l
v
5 Regional Managers under the VP Operations/Sales
l
l
v
8 Division Managers under each Regional Manager
I
I
v
5 Store Managers under each Division Manager
(Total of 40 Stores)
I
v
2-5 Sales Managers under each Store Manager

The President reports to the Board of Directors. There are
three Vice-Presidents under the President of the company.
Five Regional Managers (Pacific Northwest, Pacific
Southwest, Midwest, Atlantic Northeast, Atlantic Southeast)
report to the Vice-President of Operations and Sales. You
are one of the five Regional Managers. There are eight
Division Managers who work under you. Five Store Managers
report to each of the eight Division Managers. There are a
total of 40 stores for which you are responSible.

The name of your company is CompuProt CompuPro .
manufactures, repairs, and sells a wide variety of high
technology computer products. For example, CompuPro has
engineered new computers based on the 80486DX and 80486SX
families of microprocessor chips using EISA, ISA,
Microchannel computer architectures. They are also
developing new printers, networks, and Video monitors to .
remain competitive in the industry. Working as a manager in
CompuPro requires a high degree of familiarity with .
computers, electronics, and other advanCing technologies.

 

317

OUTLINE
THE CURRENT SITUATION:

You are the Regional Manager in a computer/electronics
manufacturing and retailing firm with stores all over the
United States. You were promoted about two years ago and
you are now in charge of the Pacific Northwest region.

Since your promotion, things have been going well.

In fact, the company has been expanding so quickly in
your area that the Vice-President of Operations/Sales has
asked you to recommend one of your Store Managers for
promotion to Division Manager. The person whom you promote
will oversee the management and performance of five stores
in a division, and you will have to work with this person
frequently. Keep in mind that the recommendation you give
to the Vice-President reflects your knowledge of those with
whom you work. Additionally, the performance of this person
will directly affect your performance, so be sure that you
recommend someone who will do well in the future.

Also, it is about the time of the year that you will
want to recognize your most productive Store Managers and
reward them accordingly by giving them salary increases and
bonuses. Raises and bonuses should only be given to those
who deserve them the most. It may be useful to recognize
that decisions regarding pay raises and bonuses might not be
based on the same types information which are used to make
promotion decisions. Promotion decisions will probably
strongly affect your personal performance and the
performance of the company as a whole. Pay raise and bonus
decisions, on the other hand, will reward a person, but they
might have less impact on your personal performance and on
the performance of the company as a whole.

You have eight Division Managers and forty Store
Managers who work for you. For your region, an average
store will generate a profit of $540,000 in a year, which
works out to an average profit rate of $45,000 each month
per store. Out of the forty Store Managers there are three
who have been recommended for promotion by their Division
Managers. These three Store Managers have each been able to
generate about $720,000 of profit during the year. This
works out to an average profit level of $60,000 a month,
which is $15,000 a month higher than the regional average.

You have decided to review the performances of these
three Store Managers so that you can make a better decision
regarding promotions, raises, and bonuses. NAMEl, NAMEZ,
and NAME3 are the three managers who have stores which have
performed above the regional average. There are many
reasons which can account for the observed performance of
these employees. Regions and stores may vary in management
competence and in customer demand for particular computer
brands.

The different causes of performance may be related to
how likely a person is to do well in other future situations

 

 

318

and they may also be related to how you feel about that
person. For example, some possible causes of performance
might be natural intelligence level, effort, and job
difficulty. Some of these causes might be considered more
stable than other causes. They may vary as to whether they
reflect something about the person or situation. They may
also vary on the degree of control or responsibility the
person has for HIS/HER performance, regardless of whether it
is internal or not. These factors may affect whether you
promote a person, reward a person with pay raises or
bonuses, or do nothing for the person. Thus, you have
decided to examine the factors involved in the performance
of each of these employees prior to making your personnel
decisions. Consequently you have asked for and have
obtained the personnel files and performance evaluations
supplied by the Division Managers for each of these

employees.

 

319
- s 5 er ormance o m o ees

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out (poorly, well) (but, and)
HE/SHE has (improved, maintained, declined) in performance

as the year has progressed.

It (also) seems that NAMEl's (recent, overall) (good, poor)
performance has been (due to, in spite of) the (high, low)
natural ability HE/SHE possesses. It takes (much, little)
time for NAMEl to understand new and constantly changing
computer technologies, and HE/SHE (seems, does not seem) to
relate easily with HIS/HER employees. Compared to other
Store Managers, NAMEl seems to be (above, below) average in

intelligence.

It (also) seems that NAMEl's (recent, overall) (good, poor)
performance has been (due to, in spite of) the (ease,
difficulty) of managing this particular store. The store
NAMEl manages is one of the (most difficult, easiest) stores
to manage. There are (many, few) other competitors and
(few, many) customers in the area, so sales are (hard, easy)
to make and maintain. Most other managers who have worked
in this store in the past have performed (well, poorly).

It (also) seems that NAMEl's (recent, overall) (good, poor)
performance has been (due to, in spite of) the (low, high)
effort that HE/SHE has exerted. NAMEl has worked very
(hard, little) to keep the store functioning at a high
level. HE/SHE has been putting in (many, few) overtime
hours to institute or maintain new sales and employment
development programs. Most other managers exert (more,

less) effort than NAMEl.

320

OUTLINE gapsgs pf Perfprmpnpe: Abilipy-Imppovipg

Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out poorly but HE/SHE has
improved in performance as the year has progressed.

It seems that NAMEl's recent good performance has been due
to the high natural ability HE/SHE possesses. It takes
little time for NAMEl to understand new and constantly
changing computer technologies, and HE/SHE seems to relate

easily with HIS/HER employees. Compared to other Store
Managers, NAMEl seems to be above average in intelligence.

It seems that NAMEl's recent good performance has been in
spite of the difficulty of managing this particular store.
The store NAMEl manages is one of the most difficult stores
to oversee. There are many other competitors and few
customers in the area, so sales are hard to make and
maintain. Most other managers who have worked in this store

in the past have performed poorly.

It also seems that NAMEl's recent good performance has been
in spite of the low effort that HE/SHE has exerted. NAMEl
has worked very little to keep the store functioning at a
high level. HE/SHE has been putting in few overtime hours
to institute or maintain new sales and employment
development programs. Most other managers exert more effort

than NAMEl.

321
0 J . s-s . ‘e_fo .nce: 13' T - :1 3:11 .
Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out well and HE/SHE has
maintained in performance as the year has progressed.

It seems that NAMEl's overall good performance has been due
to the high natural ability HE/SHE possesses. It takes
little time for NAMEl to understand new and constantly
changing computer technologies, and HE/SHE seems to relate

easily with HIS/HER employees. Compared to other Store
Managers, NAMEl seems to be above average in intelligence.

It seems that NAMEl's overall good performance has been in
spite of the difficulty of managing this particular store.
The store NAMEl manages is one of the most difficult stores
to oversee. There are many other competitors and few
customers in the area, so sales are hard to make and
maintain. Most other managers who have worked in this store

in the past have performed poorly.

It also seems that NAMEl's overall good performance has been
in spite of the low effort that HE/SHE has exerted. NAMEl
has worked very little to keep the store functioning at a
high level. HE/SHE has been putting in few overtime hours
to institute or maintain new sales and employment
development programs. Most other managers exert more effort

than NAMEl.

322

QETLTEE Qauses of Performance: Ability-Declining

Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out well but HE/SHE has
declined in performance as the year has progressed.

It seems that NAMEl's recent poor performance has been due
to the low natural ability HE/SHE possesses. It takes much
time for NAMEl to understand new and constantly changing
computer technologies, and HE/SHE does not seem to relate
easily with HIS/HER employees. Compared to other Store
Managers, NAMEl seems to be below average in intelligence.

It seems that NAMEl's recent poor performance has been in
spite of the ease of managing this particular store. The
store NAMEl manages is one of the easiest stores to oversee.
There are few other competitors and many customers in the
area, so sales are easy to make and maintain. Most other
managers who have worked in this store in the past have

performed well.

It also seems that NAMEl's recent poor performance has been
in spite of the high effort that HE/SHE has exerted. NAMEl
has worked very hard to keep the store functioning at a high
level. HE/SHE has been putting in many overtime hours to
institute or maintain new sales and employment development
programs. Most other managers exert less effort than NAMEl.

323

s 0 ce: 0 -I v
Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out poorly but HE/SHE has
improved in performance as the year has progressed.

It seems that NAMEl's recent good performance has been due
to the high effort that HE/SHE has exerted. NAMEl has
worked very hard to keep the store functioning at a high
level. HE/SHE has been putting in many overtime hours to
institute or maintain new sales and employment development
programs. Most other managers exert less effort than NAMEl.

It seems that NAMEl's recent good performance has been in
spite of the low natural ability HE/SHE possesses. It takes
much time for NAMEl to understand new and constantly
changing computer technologies, and HE/SHE does not seem to
relate easily with HIS/HER employees. Compared to other
Store Managers, NAMEl seems to be below average in

intelligence.

It also seems that NAMEl's recent good performance has been
in spite of the difficulty of managing this particular
store. The store NAMEl manages is one of the most difficult
stores to oversee. There are many other competitors and few
customers in the area, so sales are hard to make and
maintain. Most other managers who have worked in this store

in the past have performed poorly.

324

u s e formance: Ef or -Maintai 'n
Review of NAMEI by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out well and HE/SHE has
maintained in performance as the year has progressed.

It seems that NAMEl's overall good performance has been due
to the high effort that HE/SHE has exerted. NAMEl has
worked very hard to keep the store functioning at a high
level. HE/SHE has been putting in many overtime hours to
institute or maintain new sales and employment development
programs. Most other managers exert less effort than NAMEl.

It seems that NAMEl's overall good performance has been in
spite of the low natural ability HE/SHE possesses. It takes
much time for NAMEl to understand new and constantly
changing computer technologies, and HE/SHE does not seem to
relate easily with HIS/HER employees. Compared to other
Store Managers, NAMEl seems to be below average in

intelligence.

It also seems that NAMEl's overall good performance has been
in spite of the difficulty of managing this particular
store. The store NAMEl manages is one of the most difficult
stores to oversee. There are many other competitors and few
customers in the area, so sales are hard to make and
maintain. Most other managers who have worked in this store

in the past have performed poorly.

 

 

325

QQTLIHE Causes of Performance: Effort-Declining

Review of NAMEl by the Division Manager

 

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out well and HE/SHE has
declined in performance as the year has progressed.

It seems that NAMEl's recent poor performance has been due
to the low effort that HE/SHE has exerted. NAMEl has worked
very little to keep the store functioning at a high level.
HE/SHE has been putting in few overtime hours to institute
or maintain new sales and employment development programs.
Most other managers exert more effort than NAMEl.

It seems that NAMEl's recent poor performance has been in
spite of the high natural ability HE/SHE possesses. It
takes little time for NAMEl to understand new and constantly

changing computer technologies, and HE/SHE seems to relate

easily with HIS/HER employees. Compared to other Store
Managers, NAMEl seems to be above average in intelligence.

It also seems that NAMEl's recent poor performance has been
in spite of the ease of managing this particular store. The
store NAMEl manages is one of the easiest stores to oversee.
There are few other competitors and many customers in the
area, so sales are easy to make and maintain. Most other
managers who have worked in this store in the past have

performed well.

326

QUILIHE Qauses pf Performance: Ease of Jop-Tmproyipg

Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out poorly but HE/SHE has
improved in performance as the year has progressed.

It seems that NAMEl's recent good performance has been due
to the ease of managing this particular store. The store
NAMEI manages is one of the easiest stores to oversee.
There are few other competitors and many customers in the
area, so sales are easy to make and maintain. Most other
managers who have worked in this store in the past have

performed well.

It seems that NAMEl's recent good performance has been in
spite of the low effort that HE/SHE has exerted. NAMEl has
worked very little to keep the store functioning at a high
level. HE/SHE has been putting in few overtime hours to
institute or maintain new sales and employment development
programs. Most other managers exert more effort than NAMEl.

It also seems that NAMEl's recent good performance has been
in spite of the low natural ability HE/SHE possesses. It
takes much time for NAMEl to understand new and constantly
changing computer technologies, and HE/SHE does not seem to
relate easily with HIS/HER employees. Compared to other
Store Managers, NAMEl seems to be below average in

intelligence.

327

OUTLINE Qeuses of Performance: Ease of Job-Eeinfaiping

Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out well and HE/SHE has
maintained in performance as the year has progressed.

It seems that NAMEl's overall good performance has been due
to the ease of managing this particular store. The store
NAMEl manages is one of the easiest stores to oversee.
There are few other competitors and many customers in the
area, so sales are easy to make and maintain. Most other
managers who have worked in this store in the past have

performed well.

It seems that NAMEl's overall good performance has been in
spite of the low effort that HE/SHE has exerted. NAMEl has
worked very little to keep the store functioning at a high
level. HE/SHE has been putting in few overtime hours to
institute or maintain new sales and employment development
programs. Most other managers exert more effort than NAMEl.

It also seems that NAMEl's overall good performance has been
in spite of the low natural ability HE/SHE possesses. It
takes much time for NAMEI to understand new and constantly
changing computer technologies, and HE/SHE does not seem to
relate easily with HIS/HER employees. Compared to other
Store Managers, NAMEl seems to be below average in

intelligence.

328

QHILIHE Causes of Performance: Ease of Job-Declipipg

Review of NAMEl by the Division Manager

NAMEl's store has performed well above average overall for
the past year. HE/SHE started out well but HE/SHE has
declined in performance as the year has progressed.

It seems that NAMEl's recent poor performance has been due
to the difficulty of managing this particular store. The
store NAMEI manages is one of the most difficult stores to
oversee. There are many other competitors and few customers
in the area, so sales are hard to make and maintain. Most
other managers who have worked in this store in the past

have performed poorly.

It seems that NAMEl's recent poor performance has been in
spite of the high effort that HE/SHE has exerted. NAMEl has
worked very hard to keep the store functioning at a high
level. HE/SHE has been putting in many overtime hours to
institute or maintain new sales and employment development
programs. Most other managers exert less effort than NAMEI.

It also seems that NAMEl's recent poor performance has been
in spite of the high natural ability HE/SHE possesses. It
takes little time for NAMEl to understand new and constantly
changing computer technologies, and HE/SHE seems to relate
easily with HIS/HER employees. Compared to other Store
Managers, NAMEI seems to be above average in intelligence.

Karen Jones
Profit

oooooooooooooooooo

 

   

    

\

330

JONES, Karen Social Security Number

712 Woodland Road 913-45-7243
Middleborough, CA 92083 Date of Birth:
(515) 793-0563 5-15-60

WORK HISTORY:

Date of Hire: 10-29-90

Current Title: Store Manager

Area of Responsibility: Store #39

prfip for the Year 1991:

$720,000

Ayefage Profit per Month for the Year 1991:
$60,000

Ayefage Profit by Month and Revenue Type:
Computer Printer Monitors Accssrs Repair/

21te

Month Amount Sales Sales Sales Sales Labor
January 30000 9900 6900 6300 5400 1500
February 40000 12400 8800 9600 6400 2800
March 38000 14060 9500 7980 4560 1900
April 52000 13520 11960 12480 8840 5200
May 47000 15510 10810 9870 9400 1410
June 58000 18560 12760 13920 8700 4060
July 62000 22940 15500 13020 7440 3100
August 75000 21750 18750 17250 12750 4500
September 70000 23100 16100 14700 14000 2100
October 82000 26240 18040 19680 12300 5740
November 78000 28860 19500 16380 9360 3900
December 88000 25520 22000 20240 14960 5280

170620 161420 114110 41490

Totals 720000 232360

331

Review of KAREN JONES by the Division Manager

Karen's store has performed well above average overall for
the past year. She started out poorly but she has improved
in performance as the year has progressed.

It seems that Karen's recent good performance has been due
to the high natural ability she possesses. It takes little
time for Karen to understand new and constantly changing
computer technologies, and she seems to relate easily with
her employees. Compared to other Store Managers, Karen
seems to be above average in intelligence.

It seems that Karen's recent good performance has been in
spite of the difficulty of managing this particular store.
The store Karen manages is one of the most difficult stores
to oversee. There are many other competitors and few
customers in the area, so sales are hard to make and
maintain. Most other managers who have worked in this store
in the past have performed poorly.

It also seems that Karen's recent good performance has been
in spite of the low effort that she has exerted. Karen has
worked very little to keep the store functioning at a high
level. She has been putting in few overtime hours to
institute or maintain new sales and employee development
programs. Most other managers exert more effort than Karen.

EEBEONNEL EVALUATION FORM
Te Be Completed by the Division Manager

Rated on a scale of 1 to 10
10=BEST, l=WORST

 

 

 

 

 

 

 

 

Rating for Rating for

Last 6 Months Previous 6

of Year Months of Year
General Ability as a Manager 10 6

o the Job 5 5
W1 5 5
19:21; 20 16
6

 

W

Gary Miller
Profit

oooooooooooooooooo

 

 

 

 

 

   

      
          

N \

      

 

\\\ [\r \
.\. .\
6 0c 0 Dec

I ,\ ( k
“NN‘VN NN§6‘(
\§ §§§§§§§§§

N
N

V
f a i i a a e

119912.22. Sam
a

Man gerial Employee

333

MILLER, Gary

4125 Gemstone Way
Harrison, CA 94548
(652) 232-6794

Social Security Number
654-12-4912

Date of Birth:

9-29-60

WORK HISTORY:

Date of Hire: 12-13-90

Current Title: Store Manager
Area of Responsibility: Store #40

Prefif for phe Xear 1991:
$720,000

Ayerage Profit per Month for the Year 1991:
$60,000

Average Profit by Month and Revenue Type: .
Total Computer Printer Monitors Accssrs Repair/

Month Amount Sales Sales Sales Sales Labor
January 59000 19470 13570 12390 10620 2950
February 60500 18755 13310 14520 9680 4235
March 59000 21830 14750 12390 7080 2950
April 59500 15470 13685 14280 10115 5950
May 62000 20460 14260 13020 12400 1860
June 58000 18560 12760 13920 8700 4060
July 61000 22570 15250 12810 7320 3050
August 58500 16965 14625 13455 9945 3510
September 59000 19470 13570 12390 11800 1770
October 61500 19680 13530 14760 9225 4305
November 60000 22200 15000 12600 7200 3000
December 62000 17980 15500 14260 10540 3720
Totals 720000 232360 170620 161420 114110 41490

334

Review of GARY MILLER by the Division Manager

Gary's store has performed well above average overall for
the past year. He started out well and he has maintained in
performance as the year has progressed.

It seems that Gary's overall good performance has been due
to the high natural ability he possesses. It takes little
time for Gary to understand new and constantly changing
computer technologies, and he seems to relate easily with
his employees. Compared to other Store Managers, Gary seems
to be above average in intelligence.

It seems that Gary's overall good performance has been in
spite of the difficulty of managing this particular store.
The store Gary manages is one of the most difficult stores
to oversee. There are many other competitors and few
customers in the area, so sales are hard to make and
maintain. Most other managers who have worked in this store
in the past have performed poorly.

It also seems that Gary's overall good performance has been
in spite of the low effort that he has exerted. Gary has
worked very little to keep the store functioning at a high
level. He has been putting in few overtime hours to
institute or maintain new sales and employee development
programs. Most other managers exert more effort than Gary.

PERSONNEL EVALUATION FORM .
To Be Completed by the Division Manager

 

Rated on a scale of 1 to 10
10=BEST, l=WORST

 

 

 

 

 

 

 

 

 

Rating for Rating for
Last 6 Months Previous 6
of Year Months of Year
General Ability as a Manager 8 8
Effort on the Job 5 5
FoTlows Company Policy 5 5
123811 18 18
6

 

gyezeTl Average Score:

Patricia Smith
Profit

oooooooooooooooooo

     
     

 

60' \ Average 01‘ Store ,
Mmmm \N

\MMMMM

336

SMITH, Patricia

732 Handlebar Road
Rancho Verde, CA 91222
(613) 445-5245

Social Security Number
882-91-3766

Date of Birth:

11-9-60

WORK HISTORY:

Date of Hire: 8-17-90

Current Title: Store Manager
Area of Responsibility: Store #23

Pzpfip for the Year 1991:
$720,000

Average Profit per Month for the Year 1991:
$60,000

Average Profit by Month and Revenue Type:
Total Computer Printer Monitors Accssrs Repair/

 

Month Amount Sales Sales Sales Sales Labor
January 88000 29040 20240 18480 15840 4400
February 78000 24180 17160 18720 12480 5460
March 82000 30340 20500 17220 9840 4100
April 70000 18200 16100 16800 11900 7000
May 75000 24750 17250 15750 15000 2250
June 62000 19840 13640 14880 9300 4340
July 58000 21460 14500 12180 6960 2900
August 47000 13630 11750 10810 7990 2820
September 52000 17160 11960 10920 10400 1560
October 38000 12160 8360 9120 5700 2660
November 40000 14800 10000 8400 4800 2000
December 30000 8700 7500 6900 5100 1800
Totals 720000 232360 170620 161420 114110 41490

337

Review of PATRICIA SMITH by the Division Manager
Patricia's store has performed well above average overall
for the past year. She started out well but she has
declined in performance as the year has progressed.

It seems that Patricia's recent poor performance has been
It takes much

due to the low natural ability she possesses.
time for Patricia to understand new and constantly changing

computer technologies, and she does not seem to relate
easily with her employees. Compared to other Store
Managers, Patricia seems to be below average in

intelligence.
It seems that Patricia's recent poor performance has been in
The

spite of the ease of managing this particular store.

store Patricia manages is one of the easiest stores to
There are few other competitors and many customers
Most

oversee.
in the area, so sales are easy to make and maintain.
other managers who have worked in this store in the past

have performed well.
It also seems that Patricia's recent poor performance has

been in spite of the high effort that she has exerted.

Patricia has worked very hard to keep the store functioning
She has been putting in many overtime

at a high level.
hours to institute or maintain new sales and employee
Most other managers exert less effort

development programs.
than Patricia.

PERSONNEL EVALUATION FORM
To Be Completed by the Division Manager

Rated on a scale of 1 to 10
10=BEST, l=WORST

 

Rating for Rating for
Last 6 Months Previous 6
of Year Months of Year

 

 

 

 

 

 

genepal Ability as a Manager 2 6

Effort on the Job 7 7

F0 3 Com an Polic 7 7
16 20

.Ieteli. ______ .______
6

V8178 9 core 3

APPENDIX E

338

Appendix E

Measurement Scales
The

These are the items which comprised the questionnaires.
items were block randomized during the construction of the

questionnaires.
Participant Instructions:

Please fill out a single Opscan or Scantron sheet for each
of the three managers whose personnel files you have read.
Please do not write on the personnel files nor on the
questionnaires themselves. Thanks!

Spore Number

Please make sure that the store number on the Scantron sheet

(found under SECTION) matches the store number of the Store
These numbers will

Manager you are currently evaluating.
They are used to track

NOT be linked with your identity.
which personnel files you received.

Thank you.

On a Scale of 1 to 5
1=Strongly Disagree

2=Disagree

3=Neither Disagree Nor Agree

4=Agree

=Strongly Agree

Causation of Observed Performance

't
I believe the observed performance of this store was due
(ability)

1.
to the Store Manager's high ability.

I believe this Store Manager is naturally intelligent.

2.
(ability)

I believe the observed performance of the store was due
(ability)

3.
to the Store Manager's high natural ability.

Effert
4. I believe the observed performance of this store was due
to the Store Manager's high level of effort. (effort)

5. I believe this Store Manager has had to work hard to

perform well. (effort)
I believe the observed performance of the store was due
(effort)

6.
to high effort on the part of the Store Manager.

339

Eeee pf Job
7. I believe the observed performance of this store was
because the Store Manager's job would be easy to accomplish

by anyone. (easejob)

8. I believe this Store Manager's performance was because
this store was easy to run. (easejob)
9. I believe this Store Manager's job was easy. (easejob)

340

Attributions of the Inferred Causes of Performance

abilit
The CAUSE of the observed performance of this Store

10.
Manager was something the Store Manager could directly

control. (control)

The CAUSE of the observed performance of this Store

11.
Manager was due to something which this person could affect.

(control)
The CAUSE of the observed performance of this Store

12.

Manager was due to something which the Store Manager could
easily influence. (control)

The CAUSE of the observed performance of this Store

13.
Manager was due to something which was from within the Store

Manager and which the Store Manager was in direct control

over. (control)
The CAUSE of this store's observed performance was due

14.
to something which was internal to the Store Manager and
which the Store Manager was directly able to change.

(control)
The CAUSE of the observed performance of this Store

15.
Manager was due to something which was from within the Store
Manager but which was not easily controlled by the Store

.Manager. (R-control)
16. The CAUSE of this store's observed performance was due
to something which was internal to the Store Manager but
which the Store Manager was not easily able to affect. (R-

control)
The CAUSE of this Store Manager's performance was due

17.
to something which reflects some aspect of the Store Manager

but the CAUSE is not something that the Store Manager can
easily influence. (R-control)

Ipfernality
18. The CAUSE of this store's observed performance was due
to something which was external to the Store Manager.

(extlocus)
The observed performance of this Store Manager was a

19.
result of something which was from outside that person.

(extlocus)
The CAUSE of this store's performance was because of

20.
(extlocus)

the situation.

341

21. The CAUSE of this store's performance was something
that reflects some aspect of the situation and not some
aspect of the Store Manager. (extlocus)

22. The CAUSE of this Store Manager's observed performance
was due to something which was in the environment.
(extlocus)

bi
23. The CAUSE of the store's observed performance was due
to something which will be likely to change in the future.
(R-stable)

24. The CAUSE of the store's observed performance was due
to something which will be likely to stay the same in the
future. (stable)

25. The observed performance of the Store Manager was a
result of something which will always be there. (stable)

26. The CAUSE of the store's observed performance was
something that is permanent. (stable)

27. The CAUSE of the store's observed performance was
something that is temporary. (R-stable)

342
Expectations and Affect
erformance

28. I expect the Store Manager to perform above average in
the future. (expect)

29. I expect the Store Manager to perform below average in
the future. (R-expect)

30. I expect the Store Manager to perform well in the
future. (expect)

31. I think the Store Manager is likely to continue in
above average performance. (expect)

32. I don't think this Store Manager will do well in the
future. (R-expect)

Affect

33. I feel uncomfortable about the intentions of this Store
Manager. (R-affect)

34. I have concerns about this Store Manager. (R-affect)

35. I think this Store Manager is a questionable employee.
(R-affect)

36. I don't feel good about this Store Manager. (R-affect)

37. I think this Store Manager is a poor employee. (R-
affect)

38. I dislike this Store Manager. (R-affect)

343

Personnel Decisions

222111211193
39. I would recommend this person for permanent promotion
to Division Manager. (promote)

40. I would like to interview this person as a candidate
for promotion to Division Manager. (promote)

41. I believe this person deserves a promotion. (promote)

42. I would promote this person to a position higher than
the current position of Store Manager. (promote)

43. I would promote this Store Manager. (promote)

Paise
44. I would not recommend this Sales Manager for a 10%
salary increase. (R-pay)

45. I would not recommend this Sales Manager for a 20%
salary increase. (R-pay)

46. I think this Sales Manager deserves reduced pay. (R—
PaY)

47. I would make a decision to lower this Sales Manager's
pay by 10%. (R-pay)

48. This Sales Manager should receive a lower salary. (R-
PaY)

Mepit Pay Bonus

49. I would not give this person a one-time bonus. (R—
merit)

50. I would not give this person a merit pay bonus for a
job well done. (R-merit)

51. I would not recommend this person for a merit pay
bonus. (R-merit)

52. This person deserves NO bonus pay. (R-merit)

53.

54.

344

 

Pegfpfmance Level

The average profit level for this Store Manager was (see

graph):

1 = $70,000
2 = $60,000
3 = $50,000
4 = $40,000
5 = $30,000

The average profit level for all stores in the region was
(see graph):

1 = $75,000
2 = $65,000
3 = $55,000
4 = $45,000
5 = $35,000

55. Compared to the regional average (see graph and use
answers to #53 and #54), the profit level of the Store
Manager in this scenario was depicted as: (please
indicate your answer using 1, 2, or 3 from below).

1=Below average for the region
2=Average for the region
3=Above average for the region

345
Completed After All Other Questionnaires Are Done

On a Scale of 1 to 5
1=Strongly Disagree
2=Disagree

3=Neither Disagree Nor Agree
4=Agree

=Strongly Agree

Women as Managers Scale (WAMS)

1. It is less desirable for women than men to have a job
that requires responsibility. (R)

2. Women have the objectivity required to evaluate business
situations properly.

3. Challenging work is more important to men than it is to
women. (R)

4. Men and women should he give equal opportunity for
participation in management training programs.

5. Women have the capability to acquire the necessary
skills to be successful managers.

6. On the average, women managers are less capable of
contribution to an organization's overall goals than are
men. (R)

7. It is not acceptable for women to assume leadership
roles as often as men. (R)

8. The business community should someday accept women in
key managerial positions.

9. Society should regard work by female managers as
valuable as work by male managers.

10. It is acceptable for women to compete with men for top
executive positions.

11. The possibility of pregnancy does not make women less
desirable employees than men.

12. Women would no more allow their emotions to influence
their managerial behavior than would men.

13. Problems associated with menstruation should not make
women less desirable than men as employees.

14. To be a successful executive, a women does not have to
sacrifice some of her femininity.

346

15. On the average, a woman who stays at home all the time
with her children is a better mother than a woman who works
outside the home at least half time. (R)

16. Women are less capable of learning mathematical and
mechanical skills than are men. (R)

17. Women are not ambitious enough to be successful in the
business world. (R)

18. Women cannot be assertive in business situations that
demand it. (R)

19. Women possess the self-confidence required of a good
leader.

20. Women are not competitive enough to be successful in
the business world. (R)

21. Women cannot be aggressive in business situations that
demand it. (R)

Promotion Decisions

22. I made my promotion decisions on the basis of expected
future performance.

23. I made my promotion decisions on the basis of whether
or not the Store Manager deserved a reward.

Pay Raise Decisions

24. I made my pay raise decisions on the basis of expected
future performance.

25. I made my pay raise decisions on the basis of whether
or not the Store Manager deserved a reward.

Mefit Pay Decisions

26. I made my merit pay bonus decisions on the basis of
expected future performance.

27. I made my merit pay bonus decisions on the basis of
whether or not the Store Manager deserved a reward.

347
Final Questions
gepger of the Target
28. What was the gender of the Store Managers in the
scenarios?

l=Male 2=Female

gepder of the Subject

29. What is your gender?

1=Male 2=Female

Age of the Subject
30. What is your age?

1) 19 or less 2) 20-22 3) 23-25 4) 26-28 5) more than 29

Sppervisory Work Experience
31. Have you ever had any previous supervisory work
experience?

1) Yes 2) No

32. How much supervisory experience have you had (in
months)?

1) 12 or less 2) 13-18 3) 19—24 4) 25-30 5) more than 31

APPENDIX E

348

Appendix F

Protocol

HAIEEIAL§

Each subject should get one sample file, three personnel
files, four Scantrons, the first of two questionnaires
(1-manager questionnaire; 2=WAMS and final questions),
consent form (to be handed out), a subject number, and a
pencil (second questionnaire and Feedback to Participants
comes later).

Females are even, Males are odd in subject number
PROTOCOL

1. Explain consent forms, have them signed, and collect
them.

2. Pass out materials and pencils. Explain the importance
of the store manager number. Used to group the materials.
Very important to match (or fill in) appropriately.

Explain how the store numbers are used. Store Number of the
Store Manager is located in the area marked Section.

3. Introduction (please don't write on materials - to be
reused)

4. Description of graphs and sample survey questions to
test graph understanding and to provide survey examples.
Don't actually need to answer the questions.

5. Initial description of company and situation

6. Read one of the three personnel files then answer
questions. Pick the next one and repeat. Read them in
order, one at a time.

7. Answer questions on the Scantron (fill in subject number
and store number), one for each file as you read them.
Treat each file independently.

8. When done, fill out last questionnaire (WAMS scale)

9. Collect all data

10. Feedback to participants

11. Track subject hours and participation

1. ersent forms

Before we begin, please read over the consent forms. If you
agree to participate, please sign form and turn it over.
I'll come by later to collect them. I am collecting the
consent forms separately from the rest of the study so that
your identity will remain separate from your responses.

Participation is strictly voluntary. You are free to
discontinue the experiment at any time for any reason.

349

2, Check manager store numbers.
Explain that they are used to match surveys, not to identify

the participants in the study. Explain how the numbers are
recorded on the Scantron sheets. The manager store number
is NOT linked to your identity.

3, Ipffoduction

Good evening. Thanks for participating in today's research
effort. Please do not begin until I've completed the
instructions. Today, you are about to participate in
research which is conducted with the use of personnel files
and surveys. There will be various scenarios presented to
you in a written form, and they will be about various types
of managers. After you have read the scenarios, you will be
asked to answer some questions about the people presented in
the scenarios. None of the questions are of a personal
nature and your identity will remain anonymous. However,
please remember that the response each of you gives is
important and may affect the results of the experiment. It
is important that each of you answer the questions as
thoughtfully and as carefully as possible. Also, please try
to answer each question, even if you aren't sure of your
answer. You may ask me about a question if you need to.
Also, PLEASE DON'T WRITE ON THESE MATERIALS. I have to
reuse them to collect data.

 

4. Description
Included with the personnel files will be a graph depicting

the performance of stores of various Store Managers. Within
the file marked Sample, is a sample graph which you can
view. You may open it now. (Pause). At the bottom of the
graph are the months of the year, starting with January,
ending in December. (Show the graphs). It is important to
read the graphs carefully in order to understand what the
manager's performance trend has been like. You will notice
that there are two lines which indicate the average for the
stores of the region by year and the average that this
person has for the year (point to averages). This is done
so that you can compare the performance of this manager to
the average performance of all the managers of the region.

Please turn to the questionnaire in the Sample file. Does
everyone have one? Please read the first question, "What is
the regional average in profit?" Next read the second
question, "What is this person's average in profit?"
Finally, read the third question, "What were the profits for
this person in the month of July?" (Wait two minutes). The
answer to the question about the regional average was around
$45,000. The answer to the question about this person's
average was around $55,000. The answer to the question
about the profit in July was around $70,000. Does anyone
have any questions about how to read the graphs? (Pause).
You will now notice that underneath the questions you just

 

350

answered are some other sample questions. You do not have
to answer these. They are just example questions to give
you an idea of what type of questions you will be asked
later.

When you are asked to fill out the surveys, you will be
asked to answer many questions which use a scale similar to
the ones you are presently viewing. You will be providing
your answers on a Scantron, so you will need a Number 2
pencil. Does everyone have a pencil? (Pause). When you
are answering the questions, you may find that the way in
which the question was asked will change the way you answer.
When filling out the surveys, the scales may vary in the
same manner as well, so read each question and each answer
carefully before writing your answer. Also, I know that you
will see questions which are similar to each other. The
purpose of multiple questions on the same topic is to
analyze the quality of the questions themselves. Again, do
you have any questions? (Pause).

(You may now replace the Sample File into the folder)

§,,§., Q 7. Administration-Initial Description,
Instruction, and Surveys

We are just about to begin, but before we do, I have some
other instructions. Find the page which contains an initial
general description of the company and the situation within
which you are to evaluate each manager. Please read it now.
(Pause and wait). You may refer back to the initial
description provided on the company and situation at any
time. After reading the initial description, you may choose
the first of the three personnel files in front of you and
begin reading. PLEASE READ THE FILES IN ORDER. After you
have read the first file, open the questionnaire marked #1.
Next, find the Scantron form with the store number (under
Session) which matches the file you just read. Mark in one.
undef Form. When you do the second managerial file, mark in
two, under Form. Do the same for three. You have one
Scantron form for each of the three managerial files.
Answer questionnaire one for each of the three managers.
Please answer the questions for each of the files
independently of your answers for the other files. Also,
please read through the file carefully before answering any
of the questions about that manager. When you are through
with this part, you should have completed three of the four
Scantron sheets.

§,,2,, & 10. Second and LasteQuestionnaire. Collect Data.
Peedback

When you have completed these three sheets, please bring
them to the front. Also, please bring your employee
folders. Please do NOT place Scantrons back into the files.
Pick up questionnaire number two. This is your last

351

questionnaire. This will be marked "111" in Section.
Please answer this questionnaire honestly. Remember, your
answers are completely anonymous. After you have answered
this questionnaire, please turn it in at the front.

You will then receive a Feedback to Participants sheet which
explains the purpose of the experiment. If you have any
questions, feel free to ask them as soon as the time is up
and the data is collected. Thank you for your help and
participation. You may now begin.

11. Make Sure Participants get Credit for Participation

Sample X
Profit

oooooooooooooooooo

 

soL N\ I N

Average of Sa\\\I\§\§ NiNNNrN\\\\\\ N V \ ‘ x \\ .
— \\\\\\‘\E egional AV §\\>\\§!\\\\\\\\jl(\\\\\le\\\\\\t-N\\\\\IN\\\\\\IN\\\\\ N\\\\\§
.. N N N N NN \

N \N\

NNNNNNN\NN
NNNNNNNNNN

N N

 

 

 

 

D

\
,NNNN
8 MONTHS

Mar Ap Ma
P

so
o na n Participants

r
Kiwi. am Gra h XP

 

353

SAMPLE QUESTIONS

Please do not write on these questions. These materials
will be reused later. Thanks.

FIRST SET

1. What is the regional average in profit for the
year?

 

2. What is this person's average in profit for the
year?

 

3. What were the profits for Manager X in the month of
July?

 

SECOND SET

On a Scale of 1 to 5
1=Strongly Disagree
2=Disagree

3=Neither Disagree Nor Agree
4=Agree

5=Strongly Agree

1. I think the manager is a great person.
2. I don't like the manager at all.

NOTE THE CHANGE BETWEEN QUESTION 1 AND 2

354
FEEDBACK TO PARTICIPANTS

You have just participated in a study which is designed to
examine the influence of gender stereotypes on personnel
decision-making. I am a graduate student in the
Industrial/Organizational Psychology program and I am
interested in the effect of gender stereotypes on perceived
causes of performance and personnel decisions.

Specifically, I expect information about causes of
performance to affect perceptions of the controllability and
stability of those causes. Furthermore, I expect this
information to affect whether the cause of the observed
performance is perceived to be something which is internal
or external to the performer. The factors of
controllability, stability, and internality are predicted to
affect decisions about promotion, merit pay, and salary
decisions in different ways for different observed
performances. For example, controllable improving
performance should be rewarded more than uncontrollable
improving performance, while controllable declining
performance should be rewarded less than uncontrollable
declining performance. Similar predictions are made for
internality and stability.

Furthermore, I expect gender to affect the perceived causes
of high performing individuals. I believe that males will
tend to view male employee improving performance as more
controllable, internal, and stable than female employee
improving performance. I also believe that females will
tend to view female employee improving performance as more
controllable, internal, and stable than male employee
improving performance. I expect the opposite to be true for
declining performance. Since personnel decision—making is
expected to be linked to controllability, stability, and
internality, I eXpect to find gender bias in the decisions
made with regard to promotions, raises, and merit pay.

The implication is that gender can affect perceptions of
decision makers and that these perceptions will bias
personnel decision- making. This is expected to be true in
spite of large amounts of specific information about the
observed performance and its causes. These findings could
have important theoretical and applied implications for
business and organizational settings.

Your responses to the questionnaires will allow me to
analyze the whether gender of employees or decision-makers
will affect the personnel decisions which are made. They
will also help to determine whether locus of cause,
controllability, and stability play a role in the formation
of personnel decisions. If you are interested in the
results of the study, you can contact me through the

 

355
psychology department. I will send you an abstract after my
thesis is completed. If you have further interests, I can
discuss my findings with you in person.

Thank you very much for your time.

356
CONSENT FORM

The questionnaires which you are about to answer are
designed to examine the effects of different types of
performance information on personnel decision-making. You
will be asked to read about different types of fictional
managers and then you will be asked to answer questions
about what you believed caused the performance of that
manager to occur. Also, you will be asked to rate different
types of personnel decisions in response to the presented
manager's performance. These personnel decisions will be
made with regard to promotions, salaries, and merit pay
bonuses. Also, you will be asked how you feel about the
quality of the performance and how well you expect the
manager to do in the future. Finally, you will be asked to
answer questions about how you feel about different types of
managers and their qualifications for different types of
jobs. Your involvement in this study is expected to take
about one and one-half hours. At the end of your
involvement, I will provide you with a one page sheet
explaining the purpose of this research effort.

Participation is strictly voluntary. You are free to
discontinue the experiment at any time for any reason.

Also, your responses will be anonymous and treated with the
strictest confidence. After your involvement with this
study you are free to ask any questions you might have. You
may contact Stanley Gully through the Department of
Psychology by mail or you may call directly at 353-9166.

By signing this sheet, you agree that the researcher has

explained the study to you and that you understand the
procedures of the study.

By signing this form, I agree to participate in this study.

 

Signature

After you have signed this form, set it to the side. The
research assistant will come by and collect it separately
from the questionnaires. Your responses to the

questionnaires will not be associated with your identity.

APPENDIX G

357

Appendix G

Definition of Terms

Afnnipnfiene: The perceived cause (attribution) of a
particular event. Managers act as "naive
psychologists" when judging the cause and meaning of
organizational events (Heider, 1958; Gioia & Sims,
1986).

Eiee: The likelihood that information will be processed
differently when a stereotype or category is activated
than when it is not (Bierhoff, 1989, p. 27).

Eontrollability: The degree to which a cause is a factor
that the person has control over (Weiner, 1979; 1986).

Declining Performance Trend: Performance begins above the
overall performance level mean and ends below the
overall performance level mean.

Impppving Performance Trend: Performance begins below the
overall performance level mean and ends above the
overall performance level mean.

Ingzenngutgroup: Refers to whether a person is perceived
as part of the group one identifies oneself with on the
basis of some salient characteristic (skin color,
religion, gender, etc.). Operationalized in this study
as whether or not the target person is of the same
gender as the decision maker.

3 a xter : The degree to which

the cause is located within the person (internal),

358
rather than outside the person (external) (Weiner,
1979).

ng1nneining_Pe;fe;nenee_Tfeng: Performance shows minor
random shifts around the overall performance level
mean.

Pey Decisions: The willingness to allocate merit pay,
bonus, or a general salary increase to an employee.

Pegfeznence Levei: Whether overall performance level is
considered above average, average, or below average.

Penformance Trend: Given an overall performance level,
performance can vary over time and yield the same
overall average. Performance trend indicates the way
in which the performance can vary over time.

Personnel Decisions: Promotion, merit pay bonus, and salary
decisions.

Premption Decisions: The willingness to promote an employee
to a higher managerial position (may imply a
commensurate pay increase).

Eex or Gender: Gender of stimulus person or decision maker.

Sfebiiity: The degree to which the cause can be expected to
be present at the same level every time the same
situation arises (Weiner, 1979).

Sgefeetype: A set of attributes ascribed to a group and
imputed to its individual members simply because they
belong to that group (Taylor, Fiske, Etcoff, &

Ruderman, 1978: Heilman, 1983)

LIST OF REFERENCES

LIST OF REFERENCES

American Psychological Association. (1992). Ethical
principles and code of conduct. American Psychologist,
11, 1597-1611.

American Psychological Association. (1988). Brief of
amicus curiae in support of respondent. Labor Lay
Series (1988—89 term of court), PE, 233-274.

Ajzen, I., & Fishbein, M. (1978). Use and misuse of Bayes'
theorem in causal attribution: Don't attribute it to
Ajzen and Fishbein either. Ps cholo ical Bulletin, pp,
244-246.

Anderson, C. A. (1983). The causal structure of
situations: The generation of plausible causal
attributions as a function of the type of event
situation. Journal of Experimental Social Psychology,
12. 185-203.

Ashmore, R. D., & DelBoca, F. K. (1979). Sex stereotypes
and implicit personality theory: Toward a cognitive-
social psychological conceptualization. Sen Roies, E,
219-248.

Bales, J. (1988, August 23). Sex stereotyping data valid,
brief says. APA Monitor.

Bierhoff, H. W. (1989). Person perception and attfibufipn.

Berlin: Springer-Verlag.

Bigoness, W. J. (1976). Effect of applicant's sex, race,
and performance on employers' performance ratings:
Some additional findings. Journal of Applied

Psyehoiogy, pi, 80-84.

Bodenhausen, G. V., & Lichtenstein, M. (1987). Social
stereotypes and information-processing strategies: The

impact of task complexity. Journal of Personality and
Eeeiei Psychology, E;(5), 871-880.

359

 

360

Bodenhausen, G. V., & Wyer, R. 8., Jr. (1985). Effects of
stereotypes on decision making and
information-processing strategies. Toupnal of

Pepsonality and Social Psychology, eE(2),267-282.

Brenner, O. C., Tomkiewicz, J., & Schein, V. E. (1989).
The relationship between sex role stereotypes and
requisite management characteristics revisited.

Aeademy of Management Journal, 3;, 662-669.

Brown, K. A. (1984). Explaining group poor performance:
An attributional analysis. Academy of Management

Reyiew, 2(1),54-63.

Calder, B. J., & Ross, M. (1973). Attitudes and nehavi0f.
Morristown, NJ: General Learning Press. (Cited in
Cohen & Leavengood, 1978).

 

Campbell, J. P., Dunnette, M. D., Lawler, E. E., III, &

Weick, K. (1970). Managerial behavior, performance,
and effectiveness. New York: McGraw-Hill.

Carroll, J. S., Perkowitz, W. T., Lurigio, A. J., & Weaver,
F. M. (1987). Sentencing goals, causal attributions,
ideology, and personality. Journal of Personality and
Social Psychology, pg, 107-118.

Cochran, C. C. (1993, April). The relationshipbetween

gender differences in salary and ratings of competence,
pptential. and performance. Symposium paper presented

at the Eighth Annual Conference of the Society for
Industrial and Organizational Psychology, in San
Francisco, CA.

Cohen, R. L. (1987). Distributive justice: Theory and
research. Social Justice Research, T, 19-40.

Cohen, J. (1988). Statistical power analysis for the
pehavioral sciences (2nd ed.). Hillsdale, NJ:

Lawrence Erlbaum Assoc.

Cohen, J., & Cohen, P. (1983). Applied multiple

negression1correlation analysis for the behavioral
sciences (2nd edition). Hillsdale, NJ: Lawrence
Erlbaum Assoc.

Cohen, S. L., & Bunker, K. A. (1975). Subtle effects of
sex role stereotypes on recruiters' hiring decisions.

Journal of Applied Psychology, pp, 566-572.

Cohen, S. L., & Leavengood, S. (1978). The utility of the
WAMS: Shouldn't it relate to discriminatory behavior?

Academy of Management Journal, ET, 742-748.

361

Deaux, K. (1976). Sex: A perspective on the attribution
process. In J. H. Harvey, W. J. Ickes, and R. F. Kidd
(Eds.), new directions in attribution research (pp.

335-352). New Jersey: Lawrence Erlbaum Associates.

Deaux, K. (1984). From individual differences to social
categories: Analysis of a decade's research on gender.

Anepican Psychologist, EP, 105-116.

Deaux, K., & Emswiller, T. (1974). Explanations of
successful performance on sex-linked tasks: What is
skill for the male is luck for the female. goufnal of

Peneonality and Social Psychology ;P(1), 80-85.

Deaux, K., & Lewis, L. (1984). Structure of gender
stereotypes: Interrelationships among components and
gender label. Journal of Personality and Social

Psychology, ep, 991-1004.

DeNisi, A. S., & Stevens, G. E. (1981). Profiles of
performance, performance evaluations, and personnel
decisions. Academy,of Management Journal, EA, 592-602.

Dipboye, R. L. (1985). Some neglected variables in
research on discrimination in appraisals. Academy of
Management qurnal, T2, 116-127.

Dovidio, J. F., Campbell, J. B., & Kahn, S. (1982)
Authoritarianism and sex-related attributional biases.

Journal of Social Psychology, 118, 199-205.

Dugan, K. W. (1989). Ability and effort attributions: Do
they affect how managers communicate performance
feedback information? Academy of Management Journal,
EE, 87-114.

Eagly, A. H., Makhijani, M. G., & Klonsky, B. G. (1992).
Gender and the evaluation of leaders: A meta-analysis.

Psychological Bulletin, 111(1), 3-22

Etaugh, C., & Brown, B. (1975). Perceiving the causes of
success and failure of male and female performances.
Develgpmental Psychology, ii, 103.

Feather, N. T. (1985). Attitudes, values, and
attributions: Explanations of unemployment. Journal

pf Pepsonalipy and Social Psychology, Ag, 876-889.

Feldman, J. M. (1981). Beyond attribution theory:
Cognitive processes in performance appraisal. Journal

pf Applied Psychology, pp, 127-148.

362

Feldman, J. M. (1990). Instrumentation and training for
performance appraisal: A perceptual-cognitive
vieWpoint. In G. R. Ferris and K. M. Rowland (Eds.),
Pezfpfnenee eveiuation, goal setting, and feedpacn.

Greenwich, Conn: JAI Press.

Feldman-Summers, S., & Kiesler, S. B. (1974). Those who
are number two try harder. Journal of Personality and
Sepia; Psychoiogy, Pp, 846-855.

Fidell, L. S. (1970). Empirical verification of sex
discrimination in hiring practices in psychology.

Aneriean Psychologist, Pp, 1094-1098.

Fiske, S. T. (1993). Controlling other people: The impact

of power on stereotyping. American Psychologist, eE,
621-628.

Fiske, S. T., Bersoff, D. N., Borgida, E., Deaux, K., &
Heilman, M. E. (1991). Social science research on
trial: The use of sex stereotyping research in Price
Waterhouse v. Hopkins. American Psychologist, Ag,
1049-1060.

Fiske, S. T., & Neuberg, S. L. (1990). A continuum of
impression formation, from category-based to
individuating processes: Influences of information and
motivation on attention and interpretation. Advances
in Exnerimental Social Psyghology, PE, 1-74.

Fiske, S. T., & Taylor, S. E. (1984). Social cognition.
Reading, MA: Addison-Wesley.

Folger, R., & Greenberg, J. (1985). Procedural justice:
An interpretive analysis of personnel systems.
Research in Personnel and Human Resources Mana ement,
1, 141-183.

Folkes, V. S., & Marcoux, R. E. (1984). Beauty and the
attributions of the beholder. Journal of Experimental

Eocial Psychology, Pp, 514-530.

Fontaine, G. (1972). Some situational determinants of
causal attributions. Unpublished doctoral
dissertations, University of Western Australia.

Ford, J. D. (1985). The effects of causal attributions on
decision makers' responses to performance downturns.
Emir of Manaqw. 10. 770-786-

Frieze, I. H. (1976). Causal attributions and information
seeking to explain success and failure. Journel pf
Peseareh in Pensonality, T9, 293-305.

 

 

363
& Price, K. H. (1977). Attitudes toward women

Garland, H.,
in management and attributions for their success and
failure in a managerial position. Joupnal of Applied

Psycnology, pg, 29-33.
Self-serving bias and

Gioia, D. A., & Sims, H. P. (1985).
actor-observer differences in organizations: An

empirical analysis. Journal of Applied Social
Psychology, T§(6), 547-563.

Gioia, D. A., & Sims, H. P. (1986). Cognition-behavior
Attribution and verbal behavior in

connections:
Organizational
197-229.

leader-subordinate interactions.
Eenevio; end Euman Eecision Proeesses, 11(2),
Situational influences

Goodstadt, B., & Kipnis, D. (1970).
Journal of A lied Ps cholo ,

on the use of power.
55, 201-207.

Gordon, R. A. (1990).

white-collar crime:
defendant race on simulated juror decisions.

of Applied Social Psychology, PQ(12), 971-983.

Green, S. G., & Liden, R. C. (1980). Contextual and
attributional influences on control decisions.

of Applied Peychology, dd, 453-458.
(1979). Attributional

Attributions for blue-collar and

The effects of subject and
Journal

Journal

S. G., & Mitchell, T. R.

Green,
processes of leaders in leader-member interactions.
Qrganizational Behavior and Human Performance, Pd,
429-458.
A taxonomy of organizational justice

(1987).
9-22.

Greenberg, J.
Academy of Management Review, TE,

theories.
Greenberg, J. (1990). Employee theft as a reaction to
The hidden cost of pay cuts.

underpayment inequity:
dournal of Applied Psychology, 1;, 561-568.
Why procedural

Greenberg, J., & Tyler, T. R. (1987).
justice in organizations. Social Justice Researcn, T,

127-142.
eases task of

Greenhouse, L. (1989). Court, 6-3,
plaintiffs in job-bias suits. New York Timee, 2 May,

1.

Gully, S., & Rodin, M. (1990) The effects of perceived
eontroliability and stability of performance trend on
Paper presented at the Western

jpp feward.
Psychological Association in Los Angeles, CA, 1990.

 

364

Hamilton, D. L. (Ed.). (1981). Cognitive processes in

 

epepeppypingiand intergroup behavior. Hillsdale, NJ:
Erlbaum.

Hamilton, D. L., Sherman, S. J., & Ruvolo, C. M. (1990).
Stereotype-based expectancies: Effects on information
processing and social behavior. Journal of Social

Iespes, Td(2), 35-60.

Hamner, W. C., Kim, J. S., Baird, L., & Bigonness, W. J.
(1974). Race and sex as determinants of ratings by
potential employers in a simulated work sampling task.

dournal of Applied Psychology, pg, 705-711.

Harshman, R. A. (1976). PARAFAC: Methods of three-way
factor anaTyses and multidimensional scaling according

to the principle of_proportionalTprofiles. Unpublished
doctoral dissertation, University of California, Los
Angeles.

Harvey, J. H., & Weary, G. (1984). Current issues in
attribution theory and research. Annual Review of

Psychology, TE, 427-459.

Heider, F. (1958). The psyghology of interpersonei
pelations. New York: Wiley.

Heilman, M. E. (1983). Sex bias in work settings: The
lack of fit model. In L. L. Cummings & B. M. Staw
(Eds.), Research in organizational behavior (Vol. 5,
pp. 269-298). Greenwich, CT: JAI Press.

Heilman, M. E., Block, C. J., Martell, R. F, & Simon, M. C.
(1989). Has anything changed? Current
characterizations of men,women, and managers. Journal

pf Applied Psychology, 1A(6), 935-942.

Heilman, M. E., & Guzzo, R. A. (1978). The perceived cause
of work success as a mediator of sex discrimination in
organizations. Organizational Behavior and Human
Performance, PT, 346-357.

Heilman, M. E., & Martell, R. F. (1986). Exposure to
successful women: Antidote to sex discrimination in
applicant screening decisions? Organizational Behavior
and Human Decision Processes, T1(3), 376-390.

Heilman, M. E., Martell, R. F., & Simon, R. C. (1988). The
vagaries of sex bias: Conditions regulating the
undervaluation, equivaluation, and overvaluation of

female job applicants Organizational Eehaviof end

Human Decision Processes, AT, 98-110.

365

Heilman, M. E., & Stopeck, M. H. (1985a). Attractiveness
and corporate success: Different causal attributions

for males and females. Journal of Applied Psychology,
19, 379-388.

Heilman, M. E., & Stopeck, M. H. (1985b). Being
attractive, advantage or disadvantage?
Performance-based evaluations and recommended personnel
actions as a function of appearance, sex, and job type.

Qfganizapional Behavior and Human Decision Pnoeesses,
5, 202-215.

 

Heneman, R. L., Greenberger, D. B., & Anonyuo, C. (1989).
Attributions and exchanges: The effects of
interpersonal factors on the diagnosis of employee

performance. Academy of Management Journal, TE,
466-476.

Hewstone, M. (1990). The 'ultimate attribution error'? A
review of the literature on intergroup causal
attribution. European Journal of Social Psychology,
29. 311-335.

Hewstone, M., & Ward, C. (1985). Ethnocentrism and casual
attribution in Southeast Asia. Journal of Personality
end Social Psychology, T§(3), 614-623.

Hofmann, D. A., Jacobs, R., & Baratta, J. E. (1993).
Dynamic criteria and the measurement of change.
Journal of Applied Psychology, 1E, 194-204.

Hogan, E. H. (1987). Effects of prior expectations on
performance ratings: A longitudinal study. Academy of
Management Journal, dd, 354-368.

Hollenbeck, J. R., Ilgen, D. R., & Sego, D. J. (in press,
1993). Repeated measures regression and mediational
tests: Enhancing the power of leadership research.
LeadershipTOuarterly.

Hunter, J. E. (1987). Multiple dependent variables in
program evaluation. In M. M. Mark and R. L Shotland
(Eds.), New Directions for Program Evaluation: Number
35, Multiple methods in program evaluation San
Francisco, CA: Jossey-Bass.

Ilgen, D. R. (1986). Laboratory research: A question of
when, not if. In E. A. Locke (Ed.), Generalizing from

iaporatory to field settings (pp. 257-267). Lexington,
MA: Lexington Books.

Ilgen, D. R., & Klein, H. J. (1988). Organizational
behavior. Annual Review of Psychology, ed, 327-351.

366

Ilgen, D. R., & Knowlton, W. A., Jr. (1980). Performance
attributional effects on feedback from superiors.
Organizational Behavior and Human Performance, Pp,
441-456.

Ilgen, D. R., & Youtz, M. A. (1986). Factors affecting the
evaluation and development of minorities in
organizations. Pesearch in Personnel and Human
Bespurces Management, (Vol. 4, pp. 307-337).

Jackson, L. A., Sullivan, L. A., & Hodge, C. N. (1993).
Stereotype effects on attributions, predictions, and
evaluations: No two social judgments are quite alike.
dpurnal of Personality and Social Psychology, dd, 69-
84.

Jones, E. E. (1979). The rocky road from acts to
dispositions. American PsyghoTpgist, TA, 107-117.

Jones, E. E., & Davis, K. E. (1965). From acts to
diSpositions: The attribution process in person
perception. In L. Berkowitz (Ed.), Advances in

experimental social psychology (Vol. 2). New York:
Academic Press.

Jones, E. E., Rock, L., Shaver, K. C., Goethals, G. R., &
Ward, L. M. (1968). Pattern of performance and
ability attributions: An unexpected primacy effect.

Journal of Personality and Social Psychology, T9, 317-
340.

Jussim, L., Coleman, M., & Lerch, L. (1987). The nature of
stereotypes: A comparison and integration of three
theories. Journal of Personality and Social

Psychology, ET, 536-546.

Kelley, H. H. (1971). Attribution in social interaction.
In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E.
Nisbet, E. S. Valins, & B. Weiner (Eds.), Attribution:
Penceiving fne causes pf behavio;. Morristown, NJ:

General Learning Press.

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

Kinicki, A. J., & Griffeth, R. W. (1985). The impact of
sex-role stereotypes on performance ratings and causal
attributions of performance. Journal of Vocational
Behavior, 31(2), 155-170.

Kipnis, D., & Cosentino, J. (1969). The use of leadership
powers in industry. Journal of A lied Ps chol , Ed,
460-466.

 

367

Knowlton, W. A., Jr., & Mitchell, T. R. (1980). Effects of
causal attributions on a supervisor's evaluation of
subordinate performance. Journal of Applied
Peychology, pp, 459-466.

Kraiger, K., & Ford, J. K. (1985). A meta-analysis of
ratee race effects in performance ratings. J ur of
Applied Psychology, 19(1), 56-65.

Landy, F. J., & Farr, J. L. (1980). Performance rating.
Psychological Bulletin, d1, 70-107.

Lefkowitz, J., & Battista, M. (1992). Bias in supervisor
ratings: Sex. ethnicityi,ability. and supervisorTe
liking. Paper presented at the 7th Annual Conference
of the Society for Industrial & Organizational
Psychology, Montreal, Quebec, Canada.

Leventhal, G. S. (1980). What should be done with equity
theory? New approaches to the study of fairness in
social relationships. In K. J. Gergen, M. S.
Greenberg, & R. H. Willis (Eds.), Social exchange:
Advances in theory and research (pp. 27-55). New York:
Plenum.

Leventhal, G. S., & Michaels, J. W. (1971). Locus of cause
and equity motivation as determinants of reward
allocation. Journal of Personality,and Social

Psychology, T1, 229-235.

L'Heureux-Barrett, T., & Barnes-Farrell, J. L. (1991).
Overcoming gender bias in reward allocation: The role

of expectations of future performance. Psychology of
Women Quarterly, Td(l), 127-139.

London, M. (1989). Managing the training enterprise. San
Francisco, CA: Jossey-Bass Publishers.

Lord, R. G., & Maher, K. J. (1991). In s. Zedeck (Ed.),

People and organizations: Vol. 1. Leadership and
information processing. Boston, MA: Unwin Hyman.

Martell, R. F. (1991). Sex bias at work: The effects of
attentional and memory demands on performance ratings

of men and women. Journal of Applied Social
Peychology, 1T, 1939-1960.

Martinko, M. J., & Gardner, W. L. (1987). The
leader/member attribution process. Academy of
Managemenf_89xiew. 12(2). 235-249.

 

368

Medway, F., & Lowe, C. (1976). The effect of stimulus
person valence on divergent self-other attributions for
success and failure. dournal of Research in
Personalify, Td, 266-278.

Mendel, R. M., & Shoenfelt, E. L. (1992). Gender and
somatotype bias in ratings of physically demanding
performance. Paper presented at the 7th Annual
Conference of the Society for Industrial &
Organizational Psychology, Montreal, Quebec, Canada.

Meyer, J. P. (1980). Causal attribution for success and
failure: A multivariate investigation of
dimensionality, formation, and consequences. Journal
of PersonalityTand Social Peychology, T§(5), 704-718.

Mitchell, T. R., Green, S. G., & Wood, R. E. (1981). An
attributional model of leadership and the poor
performing subordinate: Development and validation.
In L. L. Cummings & B. M. Staw (Eds.), Research in
cfganizational behavior (Vol. 3, pp 197-234).
Greenwich, Conn: JAI Press.

 

Mitchell, T. R., & Kalb, L. S. (1982). Effects of job
experience on supervisor attributions for a
subordinate's poor performance. Journal of Applied

Psychology, e1, 181-188.

Mitchell, T. R., & O'Reilly, C. A., III. (1990). Managing
poor performance and productivity in organizations. In
G. R. Ferris and K. M. Rowland (Eds.), Perfopnance
evaluation, goal setting. and feedback. Greenwich, CT:
JAI Press.

Mitchell, T. R., & Wood, R. E. (1980). Supervisor's
responses to subordinate poor performance: A test of
an attributional model. Organizational Behavior and
Human Performance, 1;, 123-138.

Morrison, A. M., & Von Glinow, M. A. (1990). Women and
minorities in management. American Psychologist,
g§(2), 200-208.

Murphy, K., Herr, B. M., Lockhart, M. C., & Maguire, E.
(1986). Evaluating the performance of paper people.

Jeurnal of Applied Psychology, 1T, 654-661.

Nieva, V. F., & Gutek, B. A. (1980). Sex effects on

evaluation. Academy of Management Review, §(2),
267-276.

369

Nicholls, J. G. (1976). Effort is virtuous, but it's
better to have ability: Evaluative responses to

perceptions of effort and ability. Jonrnal pf Peseefcn
in Personality, Td, 306-315.

Olian, J. D., Schwab, D. P., & Haberfeld, Y. (1988). The
impact of applicant gender compared to qualifications
on hiring recommendations: A meta-analysis of
experimental studies. Organizational Behavior and
Ennen peeision Processes, AT, 180-195.

Pazy, A. (1986). The persistence of pro-male bias despite
identical information regarding causes of success.

Qrganizational Behavior and Human Decision Processes,
.;§(3), 366-377.

Pence, E. C., Pendleton, W. C, Dobbins, G. H., & Sgro, J. A.
(1982). Effects of causal explanations and sex
variables on recommendations for corrective actions
following employee failure. Organizational Behavio;
and Human Performance, 12, 227-240.

Pettigrew, T. F. (1979). The ultimate attribution error:
Extending Allport's cognitive analysis of prejudice.

Personality and Social Psychology Bulletin, p, 461-476.

Pheterson, G. I., Kiesler, S. B., & Goldberg, P. A. (1971).
Evaluation of the performance of women as a function of
their sex, achievement, and personal history. Journal
of Pereonality and Social Peychology, TP, 114-118.

Plake, B. S., Murphy-Barman, V., Dersheid, L. E., Gerber, R.
W., Miller, S. K., Speth, C. A., & Tomes, R. E.
(1987). Access decisions by personnel directors:
Subtle forms of sex bias in hiring. Psychology of

flomen Quarterly, TT(2), 255-263.

Renwick, P. A., & Tosi, H. (1978). The effects of sex,
marital status, and educational background on selection
decisions. Academy of Management Journal, 1T, 93-103.

Rest, S., Nierenberg, R., Weiner, B., & Heckhausen, H.
(1973). Further evidence concerning the effects of
perceptions of effort and ability on achievement

motivation. Journal of Personality and Social
Psychology, 1E, 187-191.

Rosen, B. & Jerdee, T. H. (1973). The influence of
sex-role stereotypes on evaluation of male and female

supervisory behavior. Journal of Applied Psychology,
e1, 44-48.

370

Rosen, B., & Jerdee, T. H. (1974). Influence of sex role

stereotypes on personnel decisions. Journal pf Applied
Psycnelogy, d2, 9-14.

Rosenwasser, S. M., a Dean, N. G. (1989). Gender role and
political office: Effects of perceived
masculinity/femininity of candidate and political
office. Annual Meeting of the International Society
of Political Psychology (1987, San Francisco, CA).

Psyehology of Women Quarterly, TT(l), 77-85.

Ruble, T. L., Cohen, R., & Ruble, D. N. (1984). Sex
stereotypes: Occupational barriers for women.

American Behavioral Scientist, 11(3), 339-356.

Russell, D. (1982). The Causal Dimension Scale: A measure
of how individuals perceive causes. Journal of
Personality and Social Psychology, 31(6), 1137-1145.

Schein, V. E. (1973). The relationship between sex role
stereotypes and requisite management characteristics.

Journal of Applied Psychology, E1, 95-100.

Schein, V. E. (1975). Relationships between sex-role
stereotypes and requisite management characteristics

among female managers. Journal of Applied Psychology,
pp, 340-344.

Schein, V. E. (1978). Sex role stereotyping, ability and
performance: Prior research and new directions.
Personnel Psychology, 1T, 259-268.

Schein, V. E., & Mueller, R. (1992). Sex role stereotyping
and requisite management characteristics: A cross
cultural look. Journal of Organizational Behavior, T1,
439-447.

Scott, W. E., & Hamner, W. C. (1975). The influence of
variations in performance profiles on the performance
evaluation process: An examination of the validity of
the criterion. Organizational Behavior and Human
EQEIQEEQDES, 14. 360-370-

Spencer, B. A., & Taylor, G. S. (1988). Effects of facial
attractiveness and gender on causal attributions of
managerial performance. Sex Rolee, TP(5-6), 273-285.

SPSS, Inc. (1988). SPSS-X advanced statistics guide (2nd
edition). Chicago, IL: SPSS, Inc.

 

 

371

Stangor, C., Lynch, L., Duan, C., & Glass, B. (1992).
Categorization of individuals on the basis of multiple
social features. Journal of Personality and Secial
ESYQQQIQQY:.§ZI 207-218-

Stevens, G. E., & DeNisi, A. S. (1980). Women as managers:
Attitudes and attributions for performance by men and
women. Academy of Management Journal, 11, 355-361.

Swim, J., Borgida, E., & Maruyama, G. (1989). Joan McKay
versus John McKay: Do gender stereotypes bias
evaluations? Psychological Bulletin, 105(3), 409-429.

Tajfel, H., & Turner, J. C. (1986). The social identity
theory of intergroup behavior. In S. Worchel and W.
Austin (Eds.), Psychology in intergroup relations (2nd
ed., pp. 7-24). Chicago: Nelson-Hall.

Taylor, D. M., & Jaggi, V. (1974). Ethnocentrism and
causal attribution in a South Indian context. Journal
of Crose3Cultural Psyghology, E, 152-171.

Taylor, D. M., & Moghaddam, F. M. (1987). Theoriee_of
intergrogplrelations: International andleocial

psychological perspectives (Chapter 4). NY: Praeger.

Taylor, S. E. (1981). A categorization approach to
stereotyping. In D. L. Hamilton (Ed.), Cognitiye

processes in stereotyping and intergroup behavior (pp.
83-114). Hillsdale, NJ: Erlbaum.

Taylor, S. E., Fiske, S. T., Etcoff, N. L., & Ruderman, A.
J. (1978). Categorical and contextual bases of person
memory and stereotyping. Journal of Personality and

Social Psychology, 1d, 778-793.

Terborg, J. R., & Ilgen, D. R. (1975). A theoretical
approach to sex discrimination in traditionally
masculine organizations. Orgenizational Behavior and
Human Performance, T1, 352-376.

Terborg, J. R., Peters, L. H., Ilgen, D. R., & Smith, F.
(1977). Organizational and personal correlates of
Attitudes Toward Women as Managers. Academy of
Management Journal, 19(1), 89-100.

Tosi, H., & Einbender, S. (1985). The effects of the type
and amount of information in sex discrimination
research: A meta-analysis. Academy of Management
Jenrnel. 28. 712-723.

372

Tversky, A., & Kahneman, D. (1973). Availability: A
heuristic for judging frequency and probability.
Eggnifive Psyenology, 9, 207-232.

Tversky, A., & Kahneman, D. (1974). Judgment under
uncertainty: Heuristics and biases. Science, 185,
1124-1131.

U.S. Department of Labor. (1987, May). Workfofce 1999.
Washington, D. C.: Employment and Training
Administration, U. S. Department of Labor.

Valle, V. A., & Frieze, I. H. (1976). Stability of causal
attributions as a mediator in changing expectations for

success. Journal of Personality and Social Psycholegy,
39, 579-587.

Von Glinow, M. A., & Krzyczkowska-Mercer, A. (1988,
Summer). Women in corporate America: A caste of
thousands. New Management, 9, 36-42.

Weiner, B. (1972). Theories of motivation: From mechanism
to cognition. Chicago: Morkham.

Weiner, B. (1974). Achievement motivation as
conceptualized by an attribution theorist. In B.
Weiner (Ed.), Achievement motivation and attribution
theory. Morristown, N.J.: General Learning Press.

Weiner, B. (1979). A theory of motivation for some
classroom experiences. Journallof Educational

Psychology, 1T, 3-25.

Weiner, B. (1985). An attributional theory of achievement
motivation and emotion. Peychological Reviey, 91,
548-573.

Weiner, B. (1986). An attributional theory of motivation
end emotion. New York: Springer-Verlag.

Weiner, B., & Kukla, A. (1970). An attributional analysis
of achievement motivation. Journal of Personality and

Social Psychology, T9, 1-20.

Weiner, B., Frieze, I. H., Kukla, A., Reed, L., Rest, S., &

Rosenbaum, R. M. (1971). Perceiving the causes of
success and failure. Morristown, NJ: General Learning

Press

Weiner, B., & Kukla, A. (1970). An attributional analysis
of achievement motivation. Journal of Personality and

Eecial Psyenolpgy, TE, 1-20.

 

373

Weiner, B., Perry, R. P., & Magnusson, J. (1988). An
attributional analysis of reactions to stigmas.

donrnal of Personality and Social Psychology, 99,
738-748.

Winer, B. J. (1971). Statistical principles in
eXperimental design. New York: McGraw-Hill.

Wood, R. E., & Mitchell, T. R. (1981). Manager behavior in
a social context: The impact of impression management
on attributions and disciplinary actions.

Qrganizational Behavior and Human Performance, 19,
356-378.

Wyer, R. S., & Srull, T. K. (Eds.) (1984). Handbook of
eocial cognition (Vols. 1-3). Hillsdale, NJ: Erlbaum.

Zoeller, C., Mahoney, G., & Weiner, B. (1983). Effects of
attribution training on the assembly task performance
of mentally retarded adults. American Journal of
Mental Deficiency, 99, 109-112. (Cited in Dugan,
1989).

Zuckerman, M. (1979). Attribution of success and failure
revisited, or: The motivational bias is alive and well
in attribution theory. Journal of Personality, 91,
245-287.

 

 

 

 

 

 

 

 

 

MICHIGQN STQTE UNIV LI

i "IIHIHILII'HItilglii

l
HI
02 805

aRIES
H ‘
I: 1
NH

 

£12930