.9 . . . . my“ :0
.1 3.6 . . A wan! :-
3“ ‘

n u. e.
I ...
v

9 v. 2'!

..1x1..

.
K1“..-
t
t
&.

.D
.u I vi . 3" VB
ﬁr. f-oi : 311.34 .~ 2.1.x...
90v 3 ‘96 .3. ‘13”:
larval. . N... .z. ..
f. :- vuJVJ'HKn‘

.Io. . _ . . I 37.37....
a , u. t... y ‘ 119.1»;
01.. .| I» .. . § flyi

.
c. Kw”... »
, i

. 7 10.00”). «4.1
- .. n..u.....‘...‘w....~...w\m;.$n..
‘ t r v .

amt...“ p. ,

.vErrK var: v. .575; ~: (“I

 

I

‘L- .\ ‘x .{

'19.:9

I

“it

Q

3%
v

5. i
I 5'7.

9

91' Pi
t”?!
f

7!

*K
IL

. li‘r;

.zL

 

THESlS

WE‘RS SITY LIBRARIES

211711!“ \lel

 

HI

 

 

 

 

 

 

11W ii 11111111

31293 0142

This is to certify that the
dissertation entitled
Antecedents and Consequences of Leader
Utilization of Staff Information in Decision
Making Teams: Addressing a Leadership Dilemma

presented by

Jean M. Phillips

has been accepted towards fulfillment
of the requirements for

 

 

Ph . D . degree if, Business Administrat ion

W (//¢//va:—

Major professor

 

Date January 3, 1997

 

MSU is an Afﬁrmative Action/Equal Opportunity Institution 0- 12771

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

MTE DUE I DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/” WWW“

 

ANTECEDENTS AND CONSEQUENCES
OF LEADER UTILIZATION OF STAFF INFORMATION
IN DECISION MAKING TEAMS:
ADDRESSING A LEADERSHIP DILEMMA
By

Jean M. Phillips

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Management

1997

ABSTRACT
ANTECEDENTS AND CONSEQUENCES OF LEADER
UTILIZATION OF STAFF INFORMATION IN DECISION MAKING
TEAMS: ADDRESSING A LEADERSHIP DILEMMA
By

Jean M. Phillips

This dissertation focused on the effectiveness of hierarchical decision making
teams with distributed expertise. Leaders of this type of group face a dilemma; making
accurate decisions often requires differentially utilizing staff members based on their
accuracy, but this differential utilization may lead to negative staff reactions. This
dissertation consisted of two experiments, which together allowed for the examination of
the antecedents and consequences of leader utilization of the information provided by
staff members in this type of team.

Participants in both studies learned a computerized decision making task requiring
the classiﬁcation of aircraft. In Experiment I, the presence or absence of staff judgment
conﬁdence and cumulative past performance feedback were crossed in a context in which
differential utilization of staff members was optimal. Hypotheses related to the effect of
providing leaders with staff past performance and/or staff judgment conﬁdence
information on the differential and accurate weighting of their staff were examined.

Experiment 11 tested hypotheses relevant to the consequences of leaders'
utilization of staff members for staff member reactions. Leader weighting strategy (equal

or differential) was crossed with team performance (low or high). The effect of team

performance and different types of leader utilization of staff member judgments on staff
member reactions was tested.

Results suggest that providing leaders staff past accuracy information is related to
greater variability in staff utilization and greater staff weighting accuracy. Leader
knowledge of staff members’ judgment conﬁdence did not lead to greater leader
weighting variability nor improved weighting accuracy, although staff judgment
conﬁdence level was positively related to leaders’ weighting of staff judgments. No
interaction effects between the availability of staff past judgment accuracy and staff
judgment conﬁdence were found.

Experiment 11 found that team performance was the primary determinant of staff
members’ reactions. The ﬁndings of previous decision inﬂuence research were found to
generalize to higher-performing, but not to lower-performing teams. Team performance
interacted with utilization level, utilization relative to the other staff members and
utilization accuracy in predicting some staff reactions. Implications and future research

directions are discussed.

Copyright by
Jean M. Phillips
1997

ACKNOWLEDGEMENTS

I would like to thank my family and friends who supported and encouraged me
throughout the completion of this project. I would also like to thank the members of my
committee, Dan Ilgen, John Hollenbeck, Alison Barber and Rick DeShon. I appreciate
the time and thought each of them put into this research and their willingness to develop
my skills as a researcher. I would also like to acknowledge the support of the Office of
Naval Research, Grant #N00014-93-O983, John R. Hollenbeck and Daniel R. Ilgen,
Principal Investigators. While I am grateful for their support, the ideas expressed in this
dissertation are not necessarily endorsed by the Navy.

I would particularly like to thank Dan Ilgen, whose quick return of drafts enabled
this project to maintain its momentum throughout the process. The quality and depth of
the comments on each draft continually challenged my thinking, and greatly improved the
quality of the research. Each draﬁ was a learning experience, and I want to thank him for
his many contributions. John Hollenbeck also deserves special mention as someone who
has contributed greatly over the last few years to my skills as a researcher and was a lot of
fun to work with.

Many other people have inﬂuenced my thinking in this area over the last ﬁve and

a half years, and I would be remiss not to mention them. Doug Sego and Debbie Major

were fantastic during the “early days” of the Team Effectiveness Research Lab, and I
learned a great deal from them about the research process. Jack Hunter also challenged
and expanded my thinking on the issues raised in this dissertation. I also appreciate the
time many fellow graduate students spent listening to my ideas and the comments they
provided that stretched my thinking and improved the quality of this research.

I would also like to acknowledge the work of Anders Johansen and Tom Peters in
their development and maintenance of the task and network used to collect the data.
They were quick with and solutions whenever problems arose, and are terriﬁc people.

I also appreciate the efforts of Spence Tower and Lori Sheppard in helping me to
keep my sanity during the many hours of data collection. Kevin and Melanie Ford were
also terriﬁc at helping me keep things in perspective throughout the process. I am also
indebted to Matt Taylor, who served as the confederate in the second study. His efforts
allowed data collection to run as smoothly as possible, and his positive attitude was
contagious throughout the study.

I would also like to thank Stan Gully for listening to, empathizing with, and
supporting me throughout the dissertation and throughout graduate school. Your
enthusiasm and willingness to help me track down answers to questions helped keep me
excited about this research. I will always welcome and appreciate your insights and
opinions.

In closing, I would like to thank everyone who contributed in any way to the
completion of this dissertation who I did not mention by name. This project was the

result of the efforts of many people, and could never have been accomplished alone.

vi

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................... x
LIST OF FIGURES ................................................................................................... xiii
CHAPTER 1
INTRODUCTION AND BACKGROUND ................................................................... 1
Lens Model ................................................................................................................. 7
Social Judgment Theory ........................................................................................... 10
Types of Leader Utilization of Staff Member Judgments ......................................... 12
Dyadic LUSJ .................................................................................................... 12
Dyadic LUSJ Accuracy .................................................................................... 12
Relative Dyadic LUSJ ...................................................................................... 12
Dyadic LUSJ Variability .................................................................................. 13
Team Effectiveness Consequences of Leader Utilization of Staff Judgments ......... 14
Decision Accuracy ........................................................................................... 14
Staff Development ........................................................................................... 17
Staff Members’ Reactions to the Leader’s Utilization of Their Judgments 1 8
Team Viability ................................................................................................. 21
Summary of Consequences of LUSJ ........................................................................ 23
Antecedents of Leader Utilization of Staff Judgments ............................................. 23
Leader Decision-Making Theories ................................................................... 24
Conﬂicting Findings Regarding the Antecedents of LUSJ .............................. 26
Resolving the Conﬂict: The Effects of Staff Past Accuracy
and Conﬁdence in Judgment ......................................................................... 27
Limitations ....................................................................................................... 31
Summary .................................................................................................................. 32
CHAPTER 2
RESEARCH PROBLEM AND HYPOTHESIS .......................................................... 34
Antecedents of LUSJ ................................................................................................ 35
Staff Member Past Judgment Accuracy and Judgment Conﬁdence ................ 3S
Consequences of LUSJ ............................................................................................. 41
Leader Utilization of Staff Member Judgments ............................................... 41
CHAPTER 3
EXPERIMENT I: ANTECEDENTS OF LUSJ .......................................................... 49

V11

Method

Experimental Design ........................................................................................ 49

Participants ....................................................................................................... 49

Task Description .............................................................................................. 50

Task Training ................................................................................................... 53

Measures .......................................................................................................... 53

Time .......................................................................................................... 53

Cognitive Ability ...................................................................................... 53

Past Accuracy Availability ....................................................................... 54

Judgment Conﬁdence Availability ........................................................... 54

Staff Member Past Judgment Accuracy Level .......................................... 54

Self-Report Staff Member Past Accuracy ................................................. 55

Staff Conﬁdence Level ............................................................................. 55

Dyadic LUSJ ............................................................................................. 55

Self-Report Weighting of a Staff Member ................................................ 57

LUSJ Accuracy ......................................................................................... 57

Dyadic LUSJ Variability ........................................................................... 58

Procedure ......................................................................................................... 59

Results .............................................................................................................. 61

Discussion ........................................................................................................ 81
CHAPTER 4

EXPERIMENT II: CONSEQUENCES OF LUSJ ...................................................... 85

Method ...................................................................................................................... 85

Experimental Design ........................................................................................ 85

Participants ....................................................................................................... 85

Task Description .............................................................................................. 86

Task Training ................................................................................................... 88

Measures .......................................................................................................... 88

Team Performance ...................................................................................... 88

Dyadic LUSJ ............................................................................................... 89

Self-Report LUSJ ........................................................................................ 89

Dyadic LUSJ Accuracy ............................................................................... 89

Relative Dyadic LUSJ ................................................................................. 90

Staff Member Willingness to Return ........................................................... 90

Staff Member Desire for Change on the Next Task ..................................... 91

Staff Member Satisfaction with Leader ....................................................... 91

Task Withdrawal .......................................................................................... 91

Staff Member Self-Efﬁcacy ......................................................................... 91

Procedure ......................................................................................................... 92

Results ....................................................................................................................... 93

Data Analysis .................................................................................................. 93

Discussion ............................................................................................................... 116

viii

CHAPTER 5
IMPLICATIONS, LIMITATIONS, AND AREAS FOR FUTURE RESEARCH....122

Antecedents of LUSJ .............................................................................................. 122
Consequences of LUSJ ........................................................................................... 126
Limitations .............................................................................................................. 129
FOOTNOTES ............................................................................................................ 131
LIST OF REFERENCES ........................................................................................... 132
APPENDIX A
TESTS OF LUSJ INDICES ....................................................................................... 142
APPENDIX B

MATERIALS USED IN BOTH EXPERIMENTS I AND II: CONSENT FORM,
DEMOGRAPHIC QUESTIONNAIRE, GENERAL TRAINING MANUAL AND
TRAINING SCRIPT .................................................................................................. 145

APPENDIX C
EXPERIMENT I MATERIALS: LEADER’S POSITION-SPECIFIC TRAINING
MANUAL, POST-SESSION QUESTIONNAIRE AND DEBRIEFING FORM ..... 153

APPENDIX D
EXPERIMENT II MATERIALS: POSITION-SPECIFIC TRAINING
MANUALS FOR STAFF, POST-SESSION QUESTIONNAIRES,

AND DEBRIEFING FORM ...................................................................................... 157
APPENDIX E

REPEATED MEASURES REGRESSION ANALYSES FOR EXPERIMENT

II USING EFFECTS CODING ................................................................................. 168

ix

LIST OF TABLES
Table 1a - Means, Standard Deviations and Intercorrelations of Variables in Table 1b .62

Table 1b - Repeated Measures Regression Analysis of Time and the Presence of
Cumulative Past Accuracy Information on LUSJ Variability ............................. 62

Table 2a - Means, Standard Deviations and Intercorrelations of Variables in Table 2b .63
Table 2b - Repeated Measures Regression Analysis of Time, the Level of Staff Member
Past Judgment Accuracy and the Presence of Cumulative Past Accuracy
information on Dyadic LUSJ ............................................................................... 65
Table 3a - Means, Standard Deviations and Intercorrelations of Variables in Table 3b .69
Table 3b - Repeated Measures Regression Analysis of Time and the Presence of Staff
Cumulative Past Accuracy Information on Leaders’ LUSJ Accuracy Across Staff
Members .............................................................................................................. 70

Table 4a - Means, Standard Deviations and Intercorrelations of Variables in Table 4b .71

Table 4b - Repeated Measures Regression Analysis of time and the Presence of Staff
Judgment Conﬁdence on LUSJ Variability ......................................................... 71

Table 5a - Means, Standard Deviations and Intercorrelations of Variables in Table 5b .72

Table 5b - Hierarchical Regression Analysis of Time, Staff Member Past Accuracy Level
and the Level of Staff Member Judgment conﬁdence on Dyadic LUSJ .............. 72

Table 6a - Means, Standard Deviations and Intercorrelations of Variables in Table 6b .74

Table 6b - Repeated Measures Regression Analysis of Time and the Presence of Staff
Judgment Conﬁdence on Leaders’ LUSJ Accuracy Across Staff Members ....... 74

Table 7a - Means, Standard Deviations and Intercorrelations of Variables in Table 7b .75
Table 7b - Repeated Measures Regression Analysis of Time, the Availability of

Cumulative Staff Past Accuracy Information and the Presence of Staff Judgment
Conﬁdence on Leaders’ LUSJ Variability ........................................................... 75

Table 8a - Means, Standard Deviations and Intercorrelations of Variables in Table 8b .79

Table 8b - Repeated Measures Regression Analysis of Time, the Availability of
Cumulative Staff Past Accuracy Information and the Presence of Staff Judgment
Conﬁdence on Leaders’ LUSJ Accuracy Across Staff Members ........................ 80

Table 9 - Means, Standard Deviations, and Intercorrelations of Variables Used in
Experiment 11 ....................................................................................................... 94

Table 10 - Repeated Measures Regression Analysis of LUSJ and Team Performance on
Willingness to Return .......................................................................................... 95

Table 11 - Repeated Measures Regression Analysis of LUSJ and Team Performance on
Desire to Change for Next Task ........................................................................... 98

Table 12 - Repeated Measures Regression Analysis of LUSJ and Team Performance on
Satisfaction With Leader ...................................................................................... 99

Table 13 - Repeated Measures Regression Analysis of LUSJ and Team Performance on
Task Withdrawal .................................................................................................. 102

Table 14 - Repeated Measures Regression Analysis of LUSJ and Team Performance on
Self-Efﬁcacy ........................................................................................................ 103

Table 15 - Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Willingness to Return ................................................................ 106

Table 16 - Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Desire to Change for Next Task ................................................ 107

Table 17 - Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Satisfaction With Leader ........................................................... 108

Table 18 - Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Task Withdrawal ........................................................................ 108

Table 19 - Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Self-Efﬁcacy .............................................................................. 109

Table 20 - Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Willingness to Return ................................................................ 110

Table 21 - Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Desire to Change for Next Task ................................................ 112

xi

Table 22 - Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Satisfaction With Leader ........................................................... 113

Table 23 - Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Task Withdrawal ........................................................................ 114

Table 24 - Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Self-Efﬁcacy .............................................................................. 116

xii

LIST OF FIGURES
Figure 1 - Brunswick’s (1956) Lens Model of Individual Decision Making .................. 8
Figure 2 - Brehmer and Hagafors’ (1986) Team Lens Model ......................................... 11
Figure 3 - Hypothesized Staff Member Past Judgment Accuracy Level by Availability of
Staff Cumulative Past Judgment Accuracy Information to Leader Interaction on

Dyadic LUSJ ........................................................................................................ 37

Figure 4 - Hypothesized Staff Member Dyadic LUSJ Level by Team Performance
Interaction on Staff Member Reactions ............................................................... 43

Figure 5 - Hypothesized Staff Member Relative Dyadic LUSJ by Team Performance
Interaction on Staff Member Reactions ............................................................... 45

Figure 6 - Hypothesized Staff Member Dyadic LUSJ Accuracy by Team Performance
Interaction on Staff Member Reactions ............................................................... 47

Figure 7 - Hypothesis la: Interaction Effect of Time and Cumulative Past Accuracy on
LUSJ Variability .................................................................................................. 64

Figure 8 - Hypothesis 1b: Interaction Effect of Time and Staff Past Accuracy Level on
LUSJ .................................................................................................................... 67

Figure 9 - Hypothesis 1b: Interaction Effect of Cumulative Past Accuracy Availability
and Staff Past Accuracy Level on LUSJ .............................................................. 68

Figure 10 - Hypothesis 3: Interaction Effect of Conﬁdence Availability and Cumulative
Past Accuracy Availability on LUSJ Variability ................................................. 77

Figure 11 - Hypothesis 3a: Interaction Effect of Time and Cumulative Past Accuracy
Availability on LUSJ Variability ......................................................................... 78

Figure 12 - Hypothesis 4a: Interaction Effect of Team Performance and LUSJ on
Willingness to Return .......................................................................................... 96

Figure 13 - Hypothesis 4d: Interaction Effect of Team Performance and LUSJ on Task
Withdrawal ........................................................................................................... 101

xiii

Figure 14 - Hypothesis 5a: Interaction Effect of Team Performance and Relative LUSJ
on Willingness to Return ..................................................................................... 105

Figure 15 - Hypothesis 6a: Interaction Effect of Team Performance and Dyadic LUSJ
Accuracy on Willingness to Return ..................................................................... 111

Figure 16 - Hypothesis 6e: Interaction Effect of Team Performance and Dyadic LUSJ
Accuracy on Self-Efficacy ................................................................................... 115

xiv

Chapter 1

INTRODUCTION AND BACKGROUND

A pervasive ﬁnding in the decision making literature is that individuals have a
limited capacity for processing information, resulting in difﬁculties in coping with
complex decision problems (Kahneman, Slovic, & Tversky, 1982; Slovic & Lichtenstein,
1971). Organizations often seek to improve their handling of complex decision situations
by assigning problems to groups (Brehmer & Hagafors, 1986; Salas, Dickinson,
Converse, & Tannenbaum, 1992). By increasing the number of information processors,
assigning a complex decision task to multiple people can improve effective decision
making by improving the acquisition, encoding, storage, and retrieval of information
(Michaelsen, Watson, & Black, 1989; Duffy, 1993).

Whether groups can process information better than individuals, however, is not
clear. Steiner (1972) recognized that groups have the potential to enhance as well as to
degrade individual decision making processes. Research has supported both positions
(see, e.g., Dyer, 1984; Tindale, 1993). The factors that make a group more likely to
perform effectively, or even to outperform individual decision makers, are far from well
understood.

There are many types of groups (e. g., hierarchical vs. non-hierarchical) and many

types of group tasks (e.g., decision making or production). Progress in understanding

2
groups and group processes has been most pronounced when research has focused on a

speciﬁc type of group performing a speciﬁc type of task. For example, juries have been
extensively studied (Davis, 1992; Thompson, 1993), resulting in a relatively solid
knowledge base on this type of small decision making group. This knowledge can be
used to generalize to non-jury groups that share a similar structure and face a similar
decision making task where every member has the same information about the decision
problem and an equal vote when it comes to reaching a ﬁnal decision.

This dissertation focuses on another type of small group which has been less well
studied: hierarchical decision making teams with distributed expertise (HTDE). This
type of group has several distinguishing characteristics. Status differences exist among
the members of such teams, with the responsibility for decisions distributed unequally
among team members. Speciﬁcally, responsibility for the ﬁnal decision lies at the top of
the hierarchy. This characteristic differentiates these teams from teams relying on
consensus in decision making (e. g., juries). Distributed expertise refers to the
characteristic that team members differ in the amount of knowledge and information each
brings to the decision problem (Hollenbeck, Ilgen, Sego, Hedlund, Major, & Phillips,
1995).

In such hierarchical teams, the decision problem becomes divided into a number
of subproblems, and each subproblem is the responsibility of an expert (Brehmer &
Hagafors, 1986). Staff members forward their interpretation of their particular
subproblem to the leader, and the leader's decision is usually based, at least in part, on the
information provided by the leader's staff of experts. The results of the decisions made

by the leader have consequences for both the leader and the staff.

3
This type of group is best characterized as a team rather than a set of independent

decision makers because of members' interdependence, common goals, and shared fate.
The fact that members of the team can inﬂuence each other in the course of making a
decision also makes this type of group a team rather than a set of independent decision
makers (Hollenbeck et al., 1995).

An HTDE can also be considered a special type of team due to its speciﬁc
features. The type of task confronting the team is strictly decision making (as opposed to
production). Although the leader often depends on the judgments of his or her staff, the
ultimate responsibility for the decision rests with the leader. The status hierarchy and
differential nature of each staff member's expertise also differentiates this type of team
from teams in general (Hollenbeck et al., 1995).

These teams are ubiquitous in military, medical, industrial and government
contexts because of their ability to process larger amounts of information and therefore
combat information overload. Most top executives of corporations and military
commanders would not be able to operate without appropriate staff support (Potter &

F iedler, 1981). Although these teams are prevalent in a variety of organizations, little is
known about the leader- and team-related processes that contribute to their effectiveness.

Leaders of HTDE face an important dilemma. On one hand good decisions
usually require that the leader pay closest attention to those staff members who make the
best judgments. In cases in which staff members vary in ability or in which the
information available to staff members is of different quality, this would lead to
differential utilization of staff members based on their past accuracy or ability. On the

other hand, staff members' perceptions that they are being listened to and utilized are

4
likely to affect their satisfaction with and commitment to the team. Research on decision

inﬂuence has consistently found that greater subordinate inﬂuence leads to greater
subordinate satisfaction (Graen & Cashman, 1975), greater satisfaction with the leader
(Deluga & Perry, 1991), greater organizational commitment (Wakabayshi, Minami, Sano,
Graen & Novak, 1980) and lower turnover (Graen & Ginsburgh, 1977; Graen, Liden &
Hoel, 1982). Because staff members often are aware of the inﬂuence they and others in
the team have on the leader, staff members' individual utilization as well as their relative
level of utilization by the leader as compared to their fellow staff members may similarly
inﬂuence their affect and commitment to the team. It may also affect other outcomes for
staff members such as satisfaction with the leader, self-efficacy and task withdrawal.
Despite the prevalence of this dilemma in HTDE, there has been little theory or
research aimed speciﬁcally at decision making in HTDE in general, or at the dilemma
described above, in particular. In 1986, Brehmer and Hagafors wrote:
One possible reason why there has been so little psychological research on staff
decision making may be that there has been no theory to guide research in this
area, nor even a pretheoretical framework. Indeed, there has not even been an
experimental paradigm for the study of staff decision making. (p. 182)
Unfortunately, advancements in our understanding of the conﬂict have been
limited since this statement was written in 1986. Brehmer and Hagafors (1986) were
among the ﬁrst to focus on HTDE. In an exploratory study they found that, in this
context, leaders struggled to make accurate decisions when staff members provided
judgments based on their subset of the available cues, particularly when the accuracy of

the staff members differed.

5
Sniezek and Buckley have presented a paradigm for J udge—Advisor decision

making focusing on the effect of a staff member's conﬁdence in his/her own judgment on
the leader's utilization of the staff member's judgment (Buckley & Sniezek, 1990 (as cited
in Sniezek & Buckley, 1995); Sniezek & Buckley, 1995). Hollenbeck et a1. (1995) also
focused on this type of hierarchical group, and proposed that the leader's appropriate
utilization of staff members' judgments (called hierarchical sensitivity) is one of three
core team-level constructs central to decision making accuracy.

Although Hollenbeck et a1.'s (1995) multilevel theory addresses the importance of
the leader's staff utilization policy on decision performance, it is a theory of team decision
making rather than of leader utilization of subordinate judgments (LUSJ). LUSJ is the
degree to which a subordinate's judgments are utilized by the leader in the team decision
making context', and is the focus of this dissertation.

While valuable for calling attention to LUSJ, the existing knowledge base is
limited by two factors. First, most existing approaches to the understanding of LUSJ
emphasize only the effects of LUSJ on team decision making accuracy (e. g., Brehmer &
Hagafors, 1986; Hollenbeck et al., 1995; Sniezek & Buckley, 1995). However, as
Hackman (1987) discusses, effectiveness has been shown to be a multidimensional
construct, particularly in groups. In this dissertation, different types of LUSJ are
addressed, as well as their implications for different aspects of team effectiveness.

Second, conﬂicting ﬁndings of studies investigating LUSJ regarding leaders'
propensity to differentially utilize staff member recommendations lend confusion rather
than clarity to our understanding of the antecedents and processes of LUSJ. This

confusion is important in that there are unique characteristics of these different research

6
approaches that may shed light on the antecedents of LUSJ. These characteristics were

explored further in this dissertation.

The purpose of this dissertation was to address the dilemma facing leaders of
HTDE: the conﬂict between decision accuracy requirements of differentially utilizing
staff members according to their accuracy and team viability costs associated with
differentiating among staff members. This dissertation will also extend the literature on
LUSJ by exploring the antecedents and consequences of LUSJ and by addressing the two
limitations discussed above. In the next section, two paradigms relevant to LUSJ will be
presented, the LUSJ construct will be developed, available literature addressing
antecedents to LUSJ will be reviewed, and the literature on the consequences of LUSJ for

multiple aspects of team effectiveness will be discussed.

7
BACKGROUND

Two decision making paradigms that are relevant to LUSJ are Brunswick's (1943;
1955; 1956) lens model of individual decision making and Brehmer and Hagafors' (1986)
adaptation of this model to teams. Each will be discussed in detail, then available
literature addressing the antecedents and consequences of LUSJ will be reviewed.

Lens Model

Brunswick (1943; 1955; 1956) proposed a widely used lens model to illustrate the
process by which an individual judge integrates n cues (predictors) into an overall
judgment (e.g., see Sniezek & Reeves, 1986). A schematic of this model is presented in
Figure 1.

The lens model of Figure 1 is based on the premise that individuals are rational
decision makers. The lens model proposes that decision makers obtain information on
the decision problem by consulting cues related to the decision criterion (Ye). These cues
have values that are intervally scaled, and have some linear relationship to Y,, which is
also intervally scaled. Leaders consult the relevant set of cues for any particular decision,
assign appropriate weights to the cues, combine the weighted cues in some fashion, and
reach an overall judgment. Overall system predictability is the ability of the set of cues to
predict the criterion and is the squared multiple correlation between Y, and the n cues.
Cue validig is rd, the product-moment correlation between cue Xi and the criterion
variable Y,. Cue utilization is the extent to which the judge's selection of a response
alternative, Y8, is (linearly) related to cue Xi, and is represented by rsi. The descriptive
model of the judge's strategy is therefore modeled by a least-squares multiple regression

equation for explaining responses (or judgments) as a function of the n cue variables:

Cues
. X1 \ Cue
I \ Utilization
,/// _// X2 \*~._ \\\\
Criterion Y, é? - A , X3 - e , - % Y,3 Decision
\‘~\‘ . "/,.. .- -”
\\ . //,_,_
X1 M”
Ecological \lalidity ; Utilization
Validities \re,i ' -
\

 

Figure 1. Brunswick's (1956) Lens Model of Individual Decision Making

Ys=b50+blel+'"+bsan (1)

where bso is the Y intercept and bsl through bsn are the b weights associated with the
judge’s utilization of each XI to Xn cue. The descriptive model can then be compared to

the normative model for predicting the criterion from the n cues:

Y e=bc0+belxl + +ben’Yn (2)

where bco is the Y intercept and bel through bcn are the b weights associated with the
optimal utilization of each Xl to Xn cue. The consistency index is the degree of linear
agreement between cue values and judgment responses (rs = rst), and the achievement
131;; is a measure of the linear correspondence between judgment responses and actual
criterion outcomes (r8 = rYeYs). The correlation between predicted values from the
descriptive model (Eq. (1)) and the normative model (Eq. (2)) is called the matching
index.

The results of studies utilizing the lens model have suggested that individuals
have greater difﬁculty learning and using cues when the cue-criterion relationship is non-
linear and non-positive than they do learning and using cues with positive linear cue-
criterion relationships (Brehmer, 1973; Deane, Hammond, & Summers, 1972; Slovic,

Fischhoff & Lichtenstein, 1977; Sniezek & Naylor, 1978).

1 0
Sorﬂ Judgment Theory

Brehmer and Hagafors' (1986) paradigm for the study of staff (or distributed
expertise) decision making discussed earlier is based on an extension of the general "lens
model" paradigm presented above. The general characteristics of this paradigm are
outlined in Figure 2.

As diagrarnmed, the team decision maker, or leader, has three experts on his/her
staff, each with specialized knowledge pertaining to a subset of the decision cues. Each
expert is responsible for two of the six cues and arrives at a judgment based on his/her
interpretation of these cues. The experts can therefore serve to reduce the six-cue
judgment task to a three-cue task for the team leader. This paradigm can easily be
extended to any number of cues or experts. The leader's task is to make the team decision
for the particular decision making problem using one of three strategies: relying on the
experts; ignoring the experts and coping directly with the six-cue task; or using some
combination of these two strategies. The team leader must therefore ﬁrst decide whether
or not s/he wants to rely on each expert at all, then determine what relative weight to give
to each expert.

Brehmer and Hagafors' (1986) paradigm focuses attention on the cognitive aspects
of leader decision making, and offers a preliminary framework from which to approach
decision making processes in teams. However, this framework does not directly address
the factors inﬂuencing the leader's ultimate utilization of the judgments provided by the
experts, nor does it address the consequences of LUSJ. These deﬁciencies will be

addressed later via a review of the literature on the antecedents and consequences of

LUSJ.

11

, Yo
Correct Leader’s
Decision Decision

 

Judgments
of Subordinates
A, B. and C

Figure 2. Brehmer and Hagafors' (1986) Team Lens Model

12
Types of Leader Utilization of Staff Member Judgments

At the core of LUSJ lies the actual weight a leader gives the judgments provided
by a subordinate to assist in the leader's decision making. These weights can be
conceptualized in several ways and at different levels of analysis. Different types of
LUSJ speak to different types of outcomes, which will be considered in detail later.

Dyadic LUSJ. The ﬁrst way of evaluating leader weighting of staff judgments
that will be considered in this dissertation is dyadic LUSJ. Dyadic LUSJ is the weight
the leader gives a staff member's judgments independent of the other staff members.
Dyadic LUSJ is the most basic type of LUSJ, and all other bases for evaluating the
leader’s utilization of staff judgments build upon it.

Dyadic LUSJ Accuracy, A second basis for evaluating the leader’s utilization of
staff member pertains to the accuracy of the LUSJ weights given to each staff member's
judgments. Dyadic LUSJ accuracy is the difference between a staff member's actual
dyadic LUSJ weight and the staff member's appropriate dyadic LUSJ weight. Because
each staff member can have a different accuracy of dyadic LUSJ score, this construct is
also at the dyadic level of analysis.

Team-level LUSJ accuracy, termed hierarchical sensitivity by Hollenbeck et a1.
(1995) is the average of the dyadic LUSJ accuracy scores for each of the leader’s staff
members. Higher hierarchical sensitivity scores reﬂect greater misweighting of the staff.
Because each leader (and therefore each team) will have only one team-level LUSJ
accuracy score, this construct is at the team level of analysis.

Relative Dyadic LUSJ. Another way to conceptualize leader utilization of staff

judgments concerns the weight a leader gives a subordinate's judgments relative to the

13
weight given to the other staff members' judgments. For example, does the leader utilize

the target staff member more than, about the same, or less than the average extent to
which s/he utilizes the other staff members? Staff members with higher relative dyadic
LUSJ scores have therefore been given greater inﬂuence by the leader in the leader's
decisions than staff members with lower relative dyadic LUSJ scores. As each staff
member has a unique relative dyadic LUSJ score, this construct is at the dyadic level of
analysis.

Dyadic LUSJ Variability. Another way to conceptualize leader utilization of staff
judgments is the variability in dyadic LUSJ that exists in a leader's dyadic LUSJ weights
across his/her staff members. For example, does the leader tend to weight all staff
members the same, or is there wide variation in his/her utilization of various staff
members? A leader who weights all staff members equally would have a low dyadic
LUSJ variability score. Alternatively, a leader who weights one staff member low,
another moderately, and a third high would have a high dyadic LUSJ variability score as
the dyadic LUSJ weights differ across staff members. Each leader will have only one
dyadic LUSJ variability score, making this a team level construct.

As illustrated by the fact that the same low value of dyadic LUSJ variability can
result from a leader giving the staff all high or all low weight, this is an index of
variability, not of level. The interaction of the average of a leader's dyadic LUSJ across
all staff members with the leader's dyadic LUSJ variability would reﬂect both the level
and variability of a leader's dyadic LUSJ weights across staff members, should this

information be desired.

14
Team Effectiveness Consequences of Leader Utilization of Staff Judgments

Effectiveness has been shown to be a multidimensional construct, particularly in
teams (Hackman, 1987). Hackman (1987) identiﬁed three effectiveness criteria of teams:
(a) output quality, (b) member need fulfillment, and (c) team viability, or the capability of
members to work together on subsequent team tasks.

Sundstrom, De Meuse and Futrell (1990) similarly deﬁned work team
effectiveness as both performance and viability, which includes at a minimum members'
satisfaction, participation, and willingness to continue working together. Hackman and
Oldham (1980) also recognized the possibility that unresolved conﬂict or divisive
interaction can leave members unwilling to work together. The consideration of social
and personal criteria in teams is becoming more common as their central role in long-
term team performance and viability is increasingly recognized.

It is important for theories of team effectiveness to consider multiple outcomes,
including performance and team viability as well as reactions on the part of individual
staff members such as satisfaction, turnover and commitment. The importance of each of
these effectiveness outcomes and the role that different types of LUSJ may play in
inﬂuencing each of these outcomes will be discussed in the following paragraphs.

Decﬁon Accurm Because of the detrimental effects of complexity and large
information loads on information processing (Kahneman, Slovic & Tversky, 1982;
Simon, 1978), relying on staff judgments to reduce the cognitive demands of the task
may lead to more accurate decisions even if the experts' accuracy is less than perfect

(Brehmer & Hagafors, 1986).

15
Hackman's (1987) normative model discussed earlier focuses on the potentially

malleable aspects of a group and its environment that are factors promoting effectiveness.
Among other things, this model addresses the process loss, or failure of a group to reach
its potential, that can result from the solicitation and weighting of member contributions
in a way that is inconsistent with members' expertise. Hackman proposes that to the
extent a group is able to weight members' contributions appropriately the group will take
better advantage of its resources and will perform more effectively.

While not explicitly addressing the processes by which a group can better identify
and utilize the contributions of its members, Hackman's (1987) model clearly highlights
the importance of appropriate LUSJ at both the dyadic and team levels in reducing
process losses and improving decision accuracy. Bottger and Yetton (1988) also
proposed and found that effective problem solving group performance depended on the
group's strategies in utilizing the group's resources, which include member abilities.

The ability of dyadic LUSJ accuracy to inﬂuence decision accuracy is likely to
depend on many factors. For example, if staff members each simply smnmarize all of the
relevant decision information for the leader (i.e., the team is not characterized by
distributed expertise), or if the judgments of one staff member are extremely highly
correlated with the criterion, the leader only needs to appropriately weight one staff
member to achieve high decision accuracy. Vroom and Yetton (1973) suggest that the
appropriateness of utilizing staff judgments is dependent in part on the staffs knowledge
of the decision problem, or ability to contribute to the decision at hand. If a staff
member's recommendations are grossly inaccurate, relying on them does not help the

leader accurately process larger amounts of decision related information. Thus, in all

16
types of teams, not utilizing poor staff members can be as critical to decision performance

as appropriately weighting good staff members. Also, if the leader cannot obtain or
process enough of the information relative to the decision by him/herself, even moderate
dyadic LUSJ accuracy can have signiﬁcant positive effects on decision performance.
Thus, as Brehmer and Hagafors (1986) describe, relying on staff judgments can lead to
more accurate decisions on the part of the leader even if staff accuracy is less than
perfect. It is the appropriate weighting of staff judgments based on each judgment's
accuracy that is likely to lead to higher decision performance.

Research has supported the proposition that the ability of equal status group
members to assess fellow group members' judgment accuracy affects group performance
(Bottger & Yetton, 1988; Libby, Trotman, and Zimmer, 1987). It seems likely that the
ability of a leader to appropriately assess and weight staff judgments will similarly affect
decision making accuracy.

A study by Hollenbeck et a1. (1995) found that staff validity can positively affect
the leader's decision making performance, and that a similar effect may exist for the
appropriateness of the staff weighting strategy used by the leader. Hollenbeck et a1.
(1995) also found that, in four-person teams, the interaction of greater staff validity and
more appropriate leader weighting of staff information in making the ﬁnal decision led to
higher leader decision making performance than the presence of either staff validity or the
appropriateness of leader weighting alone.

The implications of LUSJ for decision accuracy are that under high information
load conditions, decision performance may increase if leaders correctly utilize staff

judgments in arriving at a ﬁnal decision (dyadic LUSJ accuracy).

17
Staff Development. Heller (1992) presented a staff development approach to

decision inﬂuence. Heller posited that allowing subordinates greater decision inﬂuence
can increase organizational efﬁciency by making better use of staff members' existing
competence (experience and skill), developing new competencies and liberating dormant
motivation in staff members. The author is not aware of any data on the accuracy of this
proposition.

Also related to staff development are the Pygmalion and Golem effects. The
Pygmalion effect occurs when raising a leader's performance expectations of a
subordinate results in an increase in that subordinate's performance (Rosenthal &
Jacobson, 1968). The Golem effect refers to the negative effect of low leader
expectations on subordinate performance, or performance improvements lower than those
otherwise attainable (Babad, Inbar, & Rosenthal, 1982). Research on the Golem and
Pygmalion effects has consistently found the predicted effects: low leader expectations
result in restricted subordinate achievement (Oz & Eden, 1994; see Rosenthal, 1985,

1991 for reviews) and compromised staff development.

Because consistently low judgment weights can be reﬂective of a leader's low
performance expectations of that staff member, staff members who are weighted low by
the leader (low dyadic LUSJ) or even just lower than their teammates (low relative LUSJ)
might show decreased performance improvements or lower self-efficacy than if their
judgments were given higher weights by the leader. Receiving high judgment weights
might be interpreted by the staff member to mean that the leader has high expectations for
that staff member and might lead to greater performance improvements or self-efﬁcacy.

The role of LUSJ in staff development illustrates its importance to team effectiveness.

18
Staff Members' Reactions to the Leader's Utilization of Their Iggments.

Vertical dyad linkage theory (V DL), or leader-member exchange theory (LMX) describes
how role—making processes result in leaders developing different types of relationships
with different staff members, including different amounts of decision inﬂuence
(Cashman, Dansereau, Graen & Haga, 1976; Graen & Cashman, 1975; Graen &
Scandura, 1987). Findings concerning leadership differences across staff members have
suggested that the different relationships a leader establishes with different staff members
can inﬂuence staff member satisfaction, commitment (Graen, Liden & Hoel, 1982;
Katerberg & Horn, 1981; Vecchio, 1982) and turnover (Graen & Ginsburgh, 1977; Graen,
Liden & Hoel, 1982).

Much of the LMX research has investigated the relationship between satisfaction
and commitment and a global measure of LMX, of which decision inﬂuence is only a
part. Further research is necessary to identify the unique inﬂuence of various components
of LMX, such as decision inﬂuence, on staff member satisfaction and commitment in
hierarchical decision making teams. Because decision inﬂuence can differ across the
staff of a hierarchical team with distributed expertise, it is possible that the degree to
which a leader utilizes a staff member's judgments (dyadic, relative, and accuracy of
dyadic LUSJ) will also have implications for that staff member’s reactions that are
different from the reactions of the other staff.

Research on groups and individuals suggests that decision inﬂuence is positively
related to increased satisfaction and commitment (Drake & Mitchell, 1977; Vroom, 1964;
Wood, 1973). Preliminary research has found that greater upward staff member inﬂuence

effectiveness is related to greater staff member satisfaction with the supervisor (Deluga &

19
Perry, 1991; Scandura, Graen & Novak, 1986) and greater overall satisfaction (Drake &

Mitchell, 1977). Although research is limited on the relationship between decision
inﬂuence and satisfaction in HTDE, these results suggest that decision inﬂuence may
affect staff reactions in these teams as well.

Preliminary research has been done in the area of strategic management teams that
indicates that the manner in which leaders elicit, receive, and respond to team members'
input affects team members' attitudes toward the leader and other team members, and
their commitment to the decision made by the leader (Korsgaard, Schweiger & Sapienza,
1995). Although they did not examine the processes involved, Kim and Mauborgne
(1993) found a relationship between subsidiary managers' reactions to the multinational's
strategic decision processes and their cooperation in implementing the multinational's
strategic decisions. This research suggests that staff members who are dissatisﬁed with
the way their judgments are utilized by the leader may have negative reactions toward the
leader and the team, and, if possible, may interfere with the successful implementation of
the leader's decision.

Strategic management teams differ from the type of team of interest here in that
their members are responsible for the execution of the leader's decisions. In the type of
teams of interest in this dissertation, however, staff members' involvement in the decision
ends once the leader's ﬁnal decision is made; there is no staff implementation of the
decision as exists in strategic teams. This difference is important because the degree to
which the leader and the staff members have the same goals and the degree to which there

is a more than one correct solution might inﬂuence staff reactions to being utilized by the

leader.

20
In the case of strategic management teams, there is not necessarily a single correct

decision. Multiple alternatives may prove to be equally successful as decisions. The
nature of the decision made by leaders of strategic teams also affects the allocation of
resources _tp staff members. Staff members in strategic management teams may therefore
differ in their preferred leader decisions, and may have different goals as advisors. For
example, consider the case of a team consisting of the president of a multinational
company and a staff of advisors who are the heads of operations in different countries
facing the decision of how to allocate the next year's budget. Each advisor wants as large
a part of the budget as possible for his/her operation, leading to conﬂicting goals among
staff members. Money allocated to one operation (and thus one advisor) is money not
allocated to another. Advisors in strategic teams might therefore react more negatively
than advisors in teams in which the leader's decision does not differentially affect the
advisors when their judgments are utilized less by the leader than the judgments of the
other staff.

In the types of teams of interest here, however, the only differential treatment
across staff members that can occur is in dyadic LUSJ. In such teams, the only
implications of the leader's decision for the staff are the consequences of team decision
accuracy, which are shared by all team members. Unlike staff members of strategic
teams who may be motivated to inﬂuence the leader's decision for personal gain (e. g.,
receiving a larger share of the overall budget), staff members and leaders in HTDE have
the same goal: an accurate decision. Because of this shared goal (leader decision
accuracy), staff may not react as negatively as staff in strategic management teams to

leader underutilization of their judgments, as long as team goals are being met.

21
The literature on participative decision making also suggests that participation in

decisions improves satisfaction (Miller & Monge, 1986) and reduces turnover,
absenteeism, and conﬂict, although the size of the effect has been shown to depend on the
type of participation investigated (Locke & Schweiger, 1979; Wagner, 1994). The
greater the degree of actual inﬂuence, the stronger the effect.

Many investigations of the consequences of decision participation for staff
members have considered individuals' perceived rather than their actual inﬂuence,
resulting in the data being obtained from the same respondents using the same
questionnaire at the same time (percept-percept procedures). Attributed inﬂuence has
been shown to be only weakly correlated with actual inﬂuence (March, 1956), however.
A meta-analysis by Wagner & Gooding (1987a) found a correlation of .39 for studies
measuring both participation and its outcomes when the data were obtained from percept-
percept procedures and a correlation of .12 when multiple sources were used to gather the
data. Wagner and Gooding (1987a; 1987b) discuss the possibility that percept-percept
bias might inﬂate the effect size between participation and its outcomes, and suggest that
the actual effect size of participation may be substantially lower than frequently believed.

Team Viability. Team effectiveness involves more than just output or
performance. Equally important are the group's future prospects as a work unit
(Hackman, 1987; Hackman & Oldham, 1980; Sundstrom, De Meuse & Futrell, 1990), or
team viability. Sundstrom, De Meuse and Futrell deﬁne team viability as member
satisfaction, participation, and willingness to continue working together. A team that is
unable or unwilling to work together in the future cannot be said to be effective,

regardless of its performance level.

22
Components of team viability have been identiﬁed at both the team (cohesiveness,

teamwork, problem-solving, planning and communication) and staff member (job and
group satisfaction, participation, peer and leader relations, job clarity, absenteeism and
turnover) levels of analysis (Sundstrom, De Meuse & F utrell, 1990). The effects of
individual staff member reactions (e.g., turnover, job or group satisfaction, participation,
and absenteeism) can be seen at the team level (e. g., dissolving of the team, high member
replacement recruitment and training costs, poor quality decisions, low productivity, low
staff member satisfaction with the leader or group). Thus, although team viability is a
team level construct, it is highly dependent on individual staff member reactions.

In the case of HTDE, staff reactions to the leader's utilization of their judgments
are important as these reactions may affect future team processes as well as the future
existence of the team. Staff members' reactions to low LUSJ may include withdrawal
behaviors from the task (e.g., less participation) as well as from the team (e. g.,
absenteeism, turnover). For instance, a staff member who feels the recommendations
s/he is providing the leader are not being utilized may stop making them (Foushee, 1984).
Should the leader wish to utilize a recommendation of that staff member in the future, it
may not be available. If a dissatisﬁed staff member is absent from or quits the team,
additional resources must be expended by the team to recruit, select, and train a
replacement. If each team member holds a unique role in the team (e.g., in the case of
distributed expertise), team performance may suffer until the replacement is fully
knowledgeable about his/her task and role within the team. If several or all staff

members withdraw from the team or quit, the team may be dissolved. Team viability is

23
thus a critical aspect of team effectiveness, and is highly dependent on individual staff

member reactions, including staff member responses to LUSJ.
Summm of Consequences of LUSJ

Decision making teams are a common way of coping with the demands of
complex decision problems. Preliminary evidence has supported the proposition that
utilizing staff judgments according to their accuracy can improve leader decision
performance. The fact that increased decision participation and decision inﬂuence may
be related to greater satisfaction and commitment, however, creates a dilemma for
leaders. Differentially weighting staff members according to their accuracy might
increase decision performance at the expense of the satisfaction, commitment, and
development of the staff member(s) who have low LUSJ or low relative LUSJ. However,
if the team performs poorly, equally weighting staff members might also jeopardize team
viability. This potential tradeoff between team performance and team viability indicates
that research on LUSJ is important in understanding and improving team effectiveness.
More detailed investigations of the consequences of inﬂuence in teams need to be
performed before any real conclusions can be drawn, as differences in the nature of the
teams studied (e.g., decision making and implementation vs. decision making as an end in
itself, etc.) and differences in the role of staff judgments in team decision making
processes (e.g., central vs. peripheral) may limit the replicability of previous ﬁndings.
Antecedents of Leader Utilization of Staff Judgments

Although theory and research on LUSJ in teams is relatively limited, the literature

addressing the extent to which an individual is inﬂuenced by the judgments of others in

24
making a decision in both group and individual contexts will be discussed in the

following paragraphs.

Leader Decision Making Theories. Although several group, team, and leadership
theories mention the importance of LUSJ in inﬂuencing team effectiveness and team
processes (e.g., Blake & Mouton, 1964; Hackman, 1987; Heller, 1992; Heller & Yukl,
1969; Vroom & Yetton, 1973), few attempt to describe the process through which LUSJ
happens. Four theories or paradigms in this area introduced to date are Brehmer and
Hagafors' (1986) team lens model, leader-member exchange theory (Graen, Liden &
Hoel, 1982), Sniezek and Buckley's (1995) work on judge-advisor systems, and
Hollenbeck, et al.'s (1995) multilevel theory.

As discussed earlier, Brehmer and Hagafors (1986) introduced a paradigm based
on social judgment theory for the study of leaders of teams consisting of a staff of
experts. This paradigm suggests that over multiple decisions leaders analyze the
performance of their staff and incorporate their judgments about prior staff performance
in their weighting of the staff members’ current judgments. The only study that has been
reported using this paradigm found that leaders had difﬁculty accurately weighting their
staff unless staff validities called for an equal-weighting strategy across staff members
(Brehmer & Hagafors, 1986). These ﬁndings suggest that the norm of equal treatment
(equal weighting) may be one from which leaders have difﬁculty deviating, even when
the leader does not know the staff very well.

Leader-member exchange theory (LMX) (Graen, Liden & Hoel, 1982), an
extension of the vertical-dyad linkage model (Danserau, Cashman, & Graen, 1973;

Danserau, Graen & Haga, 1975; Graen & Cashman, 1975), posits that leaders select "in-

25
group" staff members to whom to allow greater decision inﬂuence and latitude. These in-

group subordinates have been shown to be chosen on the basis of their ability and
willingness to accept extra-role responsibilities (Graen & Scandura, 1987; Scandura,
Graen & Novak, 1986). The quality of the leader-member exchange reﬂects the degree of
inﬂuence and latitude allowed by the leader in performing job responsibilities.

LMX was designed to apply to situations in which a leader can select in-group
staff members to whom to allow greater job latitude and decision inﬂuence. These
situations do not necessarily involve decision making, however, and the groups may not
consist of members with distributed expertise. In the teams of interest in this dissertation,
the contributions of all staff members are important in that each is responsible for a
speciﬁc subproblem unique to his/her role. To the extent that a leader can still ignore or
differentially weight staff recommendations, a certain amount of discretion in terms of
the inﬂuence and latitude given to individual staff members is possible in the type of
teams of interest here. For example, the fact that the leader of a HTDE can choose not to
utilize a poorly performing staff member and can choose to heavily utilize a good
performing staff member, relationships of differential inﬂuence can be established.

Sniezek and Buckley (1995) have presented a paradigm on judge-advisor systems
(JASs) that addresses leader decision making in groups when at least one person is in the
role of advisor and formulates judgments or recommends alternatives that are then
communicated to the person in the role of judge. Experts in JASs do not necessarily
possess specialized knowledge or differential expertise, which differentiates them from
the teams of central interest here, but the judge has the responsibility for making the ﬁnal

decision and the decision has consequences for both the advisors and the judge.

26
Two assumptions of JASS are that conﬁdence in one's own judgment is a

mechanism of inﬂuence between advisors and the judge during the decision making
process, and that social inﬂuence is mediated by the subjective uncertainty of the leader
about the correct decision (Sniezek & Buckley, 1995). These propositions have received
preliminary support (Sniezek & Buckley, 1995).

As discussed earlier, Hollenbeck et al.'s (1995) multilevel theory addresses the
importance of the leader's staff weighting policy on decision performance, and is a theory
of team decision making rather than of leader utilization of subordinate judgments
(LUSJ). Constructs originating from the social system (e.g., group cohesion), roles (e. g.,
role conﬂict) and behavior settings (e.g., physical proximity between leaders and staff)
are said to affect constructs such as the leader's average dyadic LUSJ accuracy across all
staff members (termed "hierarchical sensitivity" by Hollenbeck et al.). The multilevel
theory incorporates a categorization scheme from McGrath (1976) rather than theorizing
about the determinants of and processes involved in LUSJ. Thus, while Hollenbeck et al.
suggest some causes of LUSJ, this theory's real contribution to the understanding of
LUSJ is its illustration of its critical role in decision accuracy.

Conﬂicting Findings Regarding the Antecedents of LUSJ. Although the
importance of recognizing and utilizing expertise in teams has been both recognized and
demonstrated, relatively little is known about conditions that affect LUSJ. Very limited
research has investigated the ability of leaders to differentially and accurately weight staff
judgments. Additionally, the literature that does exist on this topic has reported
conﬂicting ﬁndings regarding leaders' ability to differentially weight staff members. For

example, Brehmer and Hagafors (1986) found that leaders had difficulty differentially

27
weighting their staff as much as they should have, and tended to utilize an equal

weighting strategy. Research on VDL and LMX theory, however, has consistently
suggested that leaders quickly seek out and develop relationships of differential inﬂuence
with different subordinates, implying that leaders have little difﬁculty differentially
utilizing their staff (Danserau, Graen & Haga, 1975; Graen & Cashman, 1975; Scandura,
Graen & Novak, 1986). The research available on the JAS paradigm (Sniezek &
Buckley, 1995) discussed above also found that leaders differentially weighted their
advisors.

In sum, although some research suggests that leaders tend not to discriminate
across staff members (e.g., Brehmer & Hagafors, 1986), other research suggests that
leaders do discriminate across staff members (Sniezek & Buckley, 1995), and even seek
out relationships with subordinates involving different levels of decision inﬂuence (Graen
& Scandura, 1987; Scandura, Graen & Novak, 1986). Research has suggested that
factors such as a subordinate's ability and a leader's liking of that subordinate (Danserau,
Graen & Haga, 1975; Graen & Cashman, 1975; Scandura, Graen & Novak, 1986), the
subordinate's past performance (Bottger & Yetton, 1988; Hollenbeck et al., 1995), and
the leader's and subordinate's conﬁdence in the accuracy of their own judgments (Sniezek
& Buckley, 1995) can affect the decision inﬂuence a leader allows a staff member. This
confusion is important, as the ability of leaders to appropriately discriminate among staff
members can be crucial to decision accuracy.

Resolvigg the Conﬂict: The Effects of Staff Past Accuracy and Conﬁdence in
Judgment. In an effort to identify the appropriate weight to give each staff member's

judgment, leaders may rely on indicators of staff member accuracy in deciding how to

28
weight their staff. For example, a staff member's past accuracy might be considered a

reﬂection of a staff member's current judgment accuracy. Staff member judgment
conﬁdence might be considered an indicator of the staff member's perception of the
likelihood that the staff member's judgment will be accurate. The possibility that the
conﬂicting ﬁndings of Brehmer and Hagafors (1986), Sniezek and Buckley (1995) and
LMX research can be explained as a result of differences in the availability of the staff
accuracy indicators of staff member judgment conﬁdence and the past accuracy of each
staff member was explored in this dissertation.

Initial investigations have been made into the implications of conﬁdence in
decision making for the behavior of individuals. As Sniezek and Buckley (1995)
describe, however, "although empirical observation of conﬁdence in individual judgment
and choice has increased substantially in recent years, there has been little interest in
linking conﬁdence to behavior" (p. 106). Research in this area has found that a staff
member's conﬁdence in his/her judgment was strongly related to his/her ability to get the
leader to choose his/her recommendation regardless of its accuracy (Sniezek & Buckley,
1995). Judges given the conﬁdence assessments of their two advisors tended to accept
the recommendation of the more conﬁdent advisor when the two advisors disagreed.
Sniezek and Buckley's results provide support for their assumption that conﬁdence is a
mechanism of inﬂuence between advisor and judge, and suggest that leader knowledge of
subordinate judgment conﬁdence (assuming the conﬁdence level differs across staff
members) might stimulate leaders to discriminate across staff members.

Conﬁdence has been shown to enhance one's inﬂuence on others while

uncertainty makes one more susceptible to inﬂuence from others (Sniezek & Buckley,

29
1995). A staff member with a high level of conﬁdence in his/her own judgment has been

shown to be given greater inﬂuence by the leader (Buckley & Sniezek, 1990; Deutsch &
Gerard, 1955). Deutsch and Gerard's (1955) social decision making theory suggests that
a leader's conﬁdence about the correctness of staff members' judgments (which can be
inﬂuenced by staff members' own conﬁdence in their judgment accuracy) moderates the
relationship between a leader's conﬁdence in his/her own judgment and LUSJ. A staff
member's judgment conﬁdence may be interpreted by the leader as an indication of the
accuracy of the staff member's judgment. The availability of staff judgment conﬁdence
might therefore help to explain why Sniezek and Buckley's leaders showed a greater
propensity to discriminate across staff judgments than leaders in Brehmer and Hagafors'
(1986) study.

LMX research is typically characterized by ﬁeld studies in which the leader and
staff have worked together for a period of time and the leader has been able to observe
each staff member's performance over time and on a variety of tasks. The longitudinal
nature of the relationship between the leader and staff members implies that these leaders
have had multiple opportunities to observe staff members' performance over time and
across tasks. When available, knowledge of staff members' past performance might be
used by the leader as an indicator of current staff member accuracy, and inﬂuence the
differential weighting of staff members accordingly.

Research on the ability of leaders to identify and utilize their most competent staff
members has indicated that high past performance does serve to increase the weight that
person's input is given by the leader (Croner & Willis, 1967; Hollenbeck et al., 1995;

Kelman, 1950). Poor staff performance has been shown to lead to more autocratic

30
behavior on the part of the leader in general (Heller & Yukl, 1969) and less delegation

(Dewhirst, Metts, & Ladd, 1987, as cited in Yukl, 1989). Low staff member judgment
accuracy might similarly lead to lower utilization of that staff member's judgments by the
leader. In an analogous fashion, it seems likely that staff members who perform well on a
task may be more likely to be given increased decision inﬂuence by the leader in the
future. Vertical-dyad linkage theory has supported the proposition that staff members
who perform well on a task may be more likely to be given increased decision inﬂuence
by the leader in the future (Croner & Willis, 1967; Hollenbeck et al., 1995; Mausner,
1954a; 1954b).

Other research has supported the proposition that people tend to use a relative-
weight averaging model when combining the judgments or recommendations of others,
where the weights are primarily determined by source credibility (Bimbaum, Wong &
Wong, 1976; Birnbaum & Stegner, 1979). To the extent that conﬁdence and past
performance might be interpreted by the leader as a reﬂection of the accuracy of the
current recommendation, it is possible that a leader's knowledge of these factors will
enable him/her to discriminate across staff members in weighting staff judgments and
making a ﬁnal decision. Further theorizing and research is needed to determine the
effects of a leader's knowledge of subordinate judgment conﬁdence and past performance
on leaders' decisions to utilize a staff member at all (dyadic LUSJ), in addition to their
effects on leaders' discrimination across staff members in assigning weights to
subordinate judgments (variance in LUSJ), and to appropriately assign these weights

(dyadic LUSJ accuracy).

31
Steiner (1972) suggested that one reason why groups misweight the input of their

individual members is that proﬁcient members may have low conﬁdence in their own
ability to perform the task. On the other hand, researchers have consistently found that
humans tend to be overconﬁdent in the accuracy of their decisions (Fischhoff, Slovic &
Lichtenstein, 1977; Gigerenzer, 1991; Lichtenstein & Fischhoff, 1977; Sniezek, Paese &
Switzer, 1990). Interestingly, Sniezek and Buckley (1995) reported a nonsigniﬁcant
correlation of .15 between advisors' accuracy and their conﬁdence ratings. The fact that
advisor conﬁdence and performance may not be highly correlated suggests that while
they might each contribute to leaders' discrimination across subordinates in assigning
weights to staff judgments, they may contribute differently to leaders' ability to use this
information to accurately discriminate across staff members. Thus, the extent to which
the availability of both conﬁdence and past performance enable leaders to discriminate in
an m sense (LUSJ variance) as well as to appropriately weight staff members in such
a way that decision accuracy is maximized (LUSJ accuracy) was investigated in this
dissertation.

Limitations. Past research addressing the utilization of the information provided
by others in making decisions has suffered from several limitations. First, of the
literature that exists on social decision making, most is concerned with groups whose
members' roles are undifferentiated (Sniezek, 1992; Sniezek & Buckley, 1995). Yet role
differentiation is an integral aspect of teams, and has even been observed in ad hoc
experimental groups (Bales & Cohen, 1979 as cited in Sniezek and Buckley, 1995).

Second, research on LUSJ has ﬁ‘equently suffered from a methodological

limitation. Empirical efforts directed at LUSJ have not always controlled for the leader's

32
independent knowledge or personal judgment prior to receiving staﬁ judgments when

computing dyadic LUSJ (e.g., Brehmer & Hagafors, 1986; Hollenbeck et al., 1995). As
discussed earlier, this poses a potential problem in interpreting previous research results
as these factors can inﬂuence the results of the regression equation identifying the level of
dyadic LUSJ. To the extent that the leader's preliminary judgment prior to receiving staff
judgments or the information known by the leader is collinear with the information
available to a staff member, the staff member's dyadic LUSJ weight will be inﬂated if the
leader's initial judgment prior to receiving staff judgments is not controlled. In this
dissertation, this was controlled by giving leaders no one information other than the
summary judgments provided by staff members, making leaders completely dependent on
staff members for decision information.
Summm

Despite a vast amount of theory and research on teams and leadership, as well as
group and individual decision making, very little is known about how leaders of HTDE
utilize the information provided by staff members in making decisions. Theory and
research consistently point to the importance of the accurate weighting of staff judgments
in ultimate team and decision making effectiveness, but rarely have the factors and
processes through which this weighting is determined, nor the consequences of this
weighting, been addressed.

In fact, an important conﬂict exists in the literature addressing leaders' tendency
and ability to discriminate across staff members in weighting staff judgments.
Interestingly, however, unique elements of each study provide a means of generating and

testing potential antecedents to LUSJ. The past performance and judgment conﬁdence of

33
each staff member might inﬂuence leaders' weighting of individual staff members (dyadic

LUSJ) and ability to discriminate across staff members (variance in LUSJ), when some
variability exists across staff members in terms of past performance and judgment
conﬁdence.

The ability of staff conﬁdence and past performance information, both
individually and jointly, to enable a leader to accurately weight staff members' judgments
(dyadic LUSJ accuracy) also has yet to be determined. Sniezek and Buckley's (1995)
ﬁndings of a nonsigniﬁcant correlation between advisor conﬁdence and performance
suggest that the two constructs might have different effects on leaders' ability to
accurately discriminate across staff judgments. The unique and combined ability of staff
conﬁdence and past judgment accuracy information to affect a leader's dyadic LUSJ,

dyadic LUSJ variability, and LUSJ accuracy were investigated in this dissertation.

Chapter 2

RESEARCH PROBLEM AND HYPOTHESES

At the core of HTDE lies a leadership dilemma. On one hand, making accurate
decisions frequently requires differentially weighting staff members' judgments according
to their accuracy. On the other hand, there is reason to believe that differentially
weighting staff input may lead to negative reactions on the part of subordinates. If the
team performs poorly, however, even an equal weighting strategy may jeopardize team
viability. A review of the available literature on individual, group and team decision
making, as well as team effectiveness, identiﬁed a number of theories and issues that
have direct applicability to how leaders use staff member information, and the
consequences of LUSJ. The conclusions that can be drawn from this literature are
limited, however, due to the conﬂicting ﬁndings regarding the antecedents of LUSJ and
the limited research that has been done on the short- and long-term consequences of
LUSJ for staff members. Thus, the purpose of this dissertation was twofold. First, it
identiﬁed unique elements of the studies that have been performed on HTDE and tested
whether providing the leader with staff members' past accuracy and current judgment
conﬁdence levels affected different types of LUSJ. Second, the consequences of different

types of LUSJ for staff members were explored.

34

35
Antecedenth LUSJ

Staff Member Past Judgment Accurac_y_and Judgment Conﬁdence. Because the
primary objective of the type of team of interest in this dissertation is to maximize
decision accuracy, leaders may rely on indicators of staff accuracy in assigning LUSJ
weights. As suggested by the literature review, the staff member characteristics of past
judgment accuracy and judgment conﬁdence may inﬂuence the leader's ability to
accurately discriminate across staff members in assigning weights to staff judgments.
Bottger and Yetton (1988) found that members inﬂuence group decisions in proportion to
their ability. The past performance or competence of an individual on the same or on a
related task has been shown to increase the inﬂuence s/he is given by a decision maker
(Croner & Willis, 1967; Hollenbeck et al., 1995; Kelman, 1950; Mausner, 1954a;
Mausner, 1954b). LMX research has also identiﬁed a tendency of leaders to establish
relationships of differential inﬂuence with subordinates (Cashman, Dansereau, Graen &
Haga, 1976; Graen & Cashman, 1975; Graen & Scandura, 1987). As past judgment
accuracy can be considered a reﬂection of ability as well as of current judgment accuracy,
it might similarly inﬂuence the weight a staff member is given by the leader.

Leaders of HTDE may not always be aware of staff members' cumulative past
judgment accuracy, however. Leaders of HTDE are confronted with a large amount of
information, which might limit their ability to keep track of the performance of each staff
member. Information not available to the leader is unlikely to inﬂuence the leader's
decisions. Feeding back staff members' cumulative past judgment accuracy to the leader
gives the leader information that can serve as a basis on which to rate the likely accuracy

of current staff judgment accuracy and to weight staff members' judgments as a whole.

36
Because the availability of cumulative staff past judgment accuracy information enables

the leader to more easily assess the task ability of each staff member, when past judgment
accuracy differs across staff members, and differential weighting (as opposed to equal

weighting) is the appropriate staff weighting strategy, it is proposed that:

Hypothesis la: The availability of staff members' cumulative past judgment
accuracy will result in a wider variability of dyadic LUSJ weights (dyadic LUSJ

variability) than if staff judgment accuracy information is not available to leaders.

Hypothesis 1b: A staff member's cumulative past judgment accuracy level will
interact with the availability of the staff member's past judgment accuracy to the
leader to inﬂuence the leader's utilization of that staff member's judgments
(dyadic LUSJ) (see Figure 3 for a summary).
1: When the leader is provided a staff member's cumulative past judgment
accuracy, higher past judgment accuracy will be positively related to
higher dyadic LUSJ.
2: When the leader is not provided a staff member’s cumulative past
judgment accuracy level, higher past judgment accuracy will be less
positively related to dyadic LUSJ than when this information is provided

to the leader

Dyadic
LUSJ

High

Low

 

37

Staff Cumulative
Past Judgment
Accuracy
Available

 
    
   

Staff Cumulative Past
Judgment Accuracy
Not Available

 

Low High
Staff Past Accuracy
Level

Figure 3. Hypothesized Staff Member Past Judgment Accuracy Level by
Availability of Staff Cumulative Past Judgment Accuracy
lnforrnation to Leader Interaction on Dyadic LUSJ

38
Staff member past judgment accuracy can be considered an indicator of staff

member ability and the likelihood that the staff member's current judgment will be
accurate. Feeding back staff past judgment accuracy information to the leader therefore
gives the leader objective information that can be used to rate the likelihood that the staff
member will be accurate on the current decision. When staff member judgment accuracy
is relatively constant, and differential weighting (as opposed to equal weighting) is the
appropriate staff weighting strategy (either because of differences in the validity of the
information available to each staff member or because of differences in staff members’
ability), this information may therefore improve the accuracy of the leader's staff

weighting strategy. It is proposed that:

Hypothesis 1c: The availability of staff members’ cumulative past judgment
accuracy to the leader will be positively related to the accuracy of the leader's

utilization of staff members’ judgments (LUSJ accuracy).

The degree to which decisions are inﬂuenced by the judgments of others has also
been shown to depend largely on the amount of uncertainty associated with those
judgments (Buckley & Sniezek, 1990; Deutsch & Gerard, 1955). Because conﬁdence in
one's own judgment has been found to enhance one's inﬂuence on others (Buckley &
Sniezek, 1990; Deutsch & Gerard, 1955; Sniezek & Buckley, 1995), assuming there is

variation in staff members’ conﬁdence levels it is proposed that:

39
Hyp_othesis 2a: The availability of staff members' judgment conﬁdence will result

in a wider variability of dyadic LUSJ weights (dyadic LUSJ variabilig) than if

staff judgment conﬁdence information is not provided to leaders.

Hypothesis 2b: A staff member’s conﬁdence level will be positively related to the

leader’s utilization of that staff member's judgments (dyadic LUSJ ).

Staff member judgment conﬁdence can be considered a reﬂection of the staff
member's perception of the likelihood that the staff member's judgment will be accurate.
Providing staff judgment conﬁdence information to the leader therefore gives the leader
subjective information that can be used to rate the likelihood that the staff member will be
accurate on the current decision. Judgment conﬁdence information may therefore
improve the accuracy of the leader's staff weighting strategy. However, as Paese and
Sniezek (1991) discuss, although conﬁdence has been shown to be a means of social
inﬂuence, if the conﬁdence assessments associated with judgments are biased or
inconsistently related to judgment quality, decision quality is likely to suffer. The general
conclusion of the individual decision making and social judgment literatures concerning
conﬁdence is that people tend to be overconﬁdent in their decisions (Dunning, Grifﬁn,
Milojkovic & Ross, 1990; Lichtenstein, F ischhoff & Phillips, 1982; Sniezek & Buckley,
1995; Vallone, Grifﬁn, Lin & Ross, 1990). Thus, not enough is known to propose a
directional hypothesis relating the presence of staff member judgment conﬁdence to
LUSJ accuracy. The relationship of staff judgment conﬁdence to LUSJ accuracy will be

investigated.

40
It was proposed earlier that past judgment accuracy and staff judgment conﬁdence

information will independently lead to greater LUSJ variance when greater discrimination
is appropriate for maximizing decision accuracy because this information provides the
leader a basis from which to derive different LUSJ weights across staff members. But
past judgment accuracy and judgment conﬁdence provide different types of information.
While past judgment accuracy is objective and is more reﬂective of ability, judgment
conﬁdence is subjective and reﬂects the staff member's self-perceived accuracy on the
current decision.

The provision of both staff judgment conﬁdence and past judgment accuracy
information to the leader may allow greater and more accurate discriminability across
staff members than if either type of information was provided alone or if neither type of
information is provided. Because conﬁdence is a less objective indicator of ability or

judgment accuracy than past judgment accuracy, it is proposed that:

Hypothesis 3: The availability of both staff cumulative past judgment accuracy
and conﬁdence information to the leader will result in a wider variability of LUSJ
weights (greater dyadic LUSJ variabiligg) than if only one or neither type of

information is provided to the leader.

As discussed earlier, not enough is known to propose a directional hypothesis
relating the presence of staff member judgment conﬁdence to LUSJ accuracy, much less

in combination with the presence of staff past judgment accuracy. The combined

41
inﬂuence of the availability of staff past judgment accuracy information and staff

judgment conﬁdence information on LUSJ accuracy will be investigated.
Consequences of LUSJ

Leader Utilization of Staff Member Judgments. The literature review indicates
that higher LUSJ may lead to positive reactions for staff members, including satisfaction
and commitment (Bass, 1981; Drake & Mitchell, 1977; Locke & Schweiger, 1979).
Virtually all of the research performed to date on staff member reactions to decision
inﬂuence has been done on teams whose members' roles in the team are not unique,
however. Staff members' roles in the decision making teams addressed by this
dissertation require the provision of a judgment to the leader based on a unique subset of
available decision cues. As decision participation is thus required for this information to
be reﬂected in the team decision, it is possible that staff members will be more affected
by the degree of inﬂuence their judgments are ultimately given than individuals for whom
participation in decision making is not a formal part of their job or role.

It is therefore possible that the different types of LUSJ in a decision making team
context will lead to lower commitment for staff members than for staff members in non-
decision making teams because they hold a role in the team that is central to decision
making. Although the relationship between LUSJ and staff member responses in HTDE
has yet to be investigated, past ﬁndings will be proposed to generalize to these types of
teams. Also, much of the past research on the effects of decision participation has studied
teams that do not have a correct answer to the decision problem.

Workgroup performance has also been shown to be positively related to group

member satisfaction (Zeffane, 1994) and turnover (Jackofsky, 1984). When team

42
performance has reward consequences for both the leader and the staff, the decision

performance of the leader may inﬂuence staff members' reactions regardless of the level

of the staff member's dyadic LUSJ. It is proposed that:

Hypothesis 4: Team performance will moderate the relationship between a staff
member's dyadic LUSJ and the staff member's reactions to the leader and the
team. Speciﬁcally, for teams where performance is high, staff members' reactions
to the team will be high regardless of the level of dyadic LUSJ. For teams where
performance is low, staff reactions will be high only when dyadic LUSJ is high,
and low when dyadic LUSJ is low (see Figure 4 for a summary). This hypothesis
will hold for the following reactions of staff members:

a: Long-term withdrawal from the team.

p: Immediate withdrawal from the team.

: Satisfaction with the leader.

10

ID-

: Task withdrawal.

: Self-efﬁcacy.

In

The degree to which the leader utilizes the judgments provided by a staff member
relative to the other staff members might also affect the staff member's reactions to LUSJ.
Differences in reactions may exist across staff members who are weighted differently by

the leader. It is proposed that:

43

High High Team
(Positive) Performance

 
 
  
  

 

 

7

 

 

Staff
Member's
Reactions
Low Low Team Performance
(Negative)
Low High
Staff Member's Dyadic LUSJ
Level

Figure 4. Hypothesized Staff Member Dyadic LUSJ Level by
Team Performance Interaction on Staff Member
Reactions

44
Hypothesis 5: Team performance will moderate the relationship between a staff

member's relative dyadic LUSJ and the staff member's reactions to the leader and the
team. Speciﬁcally, for teams where team performance is high, staff members' reactions
will be high regardless of their level of relative dyadic LUSJ. For teams where team
performance is low, staff reactions will be high only when relative dyadic LUSJ is high,
and low when relative dyadic LUSJ is low (see Figure 5 for a summary). This hypothesis
will hold for the following responses of staff members:

a: Long-term withdrawal from the team.

p: Immediate withdrawal from the team.

: Satisfaction with the leader.

10

a: Task withdrawal.

a: Self-efficacy.

Team performance might also inﬂuence the relationship between dyadic LUSJ
accuracy and staff member satisfaction. For example, if a staff member's accuracy is
poor and the leader appropriately discounts the judgments of that staff member and the
team performs well as a result, satisfaction and commitment may result for that staff

member despite low LUSJ. It is proposed that:

Hypothesis 6: Team performance will moderate the relationship between a staff
member's dyadic LUSJ accuracy and the staff member's reactions to the leader

and the team. Speciﬁcally, for teams where performance is high, staff members'

45

 

 

 

High
High Team Performance
(Positive)
Staff
Member's
Reactions
Low Low Team Performance
(Negative)

 

 

Low High
Staff Member's Relative Dyadic LUSJ

Figure 5. Hypothesized Staff Member Relative Dyadic LUSJ
by Team Performance Interaction on Staff Member
Reactions

46
reactions will be high regardless of the level of dyadic LUSJ accuracy. For teams

where performance is low, staff reactions will be high only when dyadic LUSJ
accuracy is high, and low when dyadic LUSJ accuracy is low (see Figure 6 for a
summary). This hypothesis will hold for the following responses of staff
members:

a: Long-term withdrawal from the team.

p: Immediate withdrawal from the team.

9: Satisfaction with the leader.

a: Task withdrawal.

a: Self-efﬁcacy.

Two studies were performed to test these hypotheses. Experiment I will focus on
leaders and will examine the antecedents of dyadic LUSJ, LUSJ accuracy and LUSJ
variability. Experiment 11 will focus on staff reactions to dyadic LUSJ, dyadic LUSJ
accuracy and relative dyadic LUSJ. To maximize the beneﬁts of the laboratory setting,
Experiment 11 will involve the use of a confederate leader.

The experiments were conducted in a laboratory setting for several reasons. First,
the purpose of Experiment I is to determine if the availability of staff past judgment
accuracy and staff judgment conﬁdence affect leaders' ability to differentially and
accurately weight staff members when a differential weighting strategy is appropriate.
The laboratory is well suited for testing hypotheses of this "can it happen" nature (Ilgen,
1986; Mook, 1983). Second, research on the antecedents of LUSJ is difficult if not

impossible in the ﬁeld. Because of the necessity of minimizing the intercorrelation

47

High High Team
(Positive) Performance

    
  

 

 

7

 

 

Staff
Member's
Reactions
LOW Low Team Performance
(Negative)
Low High
Staff Member's Dyadic LUSJ
Accuracy

Figure 6. Hypothesized Staff Member Dyadic LUSJ Accuracy
by Team Performance Interaction on Staff Member
Reactions

48
among staff judgments, a great deal of control over the experimental task is required.

This type of control would rarely be possible in a ﬁeld setting. Third, ethical
considerations exist when studying the nature of staff reactions to LUSJ (Ilgen, 1986).
Because negative reactions are hypothesized on the part of staff members in Experiment
II, ethical concerns regarding the manipulation of LUSJ in the ﬁeld are extremely

relevant. The two experiments will be discussed in detail in the next two chapters.

Chapter 3
EXPERIMENT I: ANTECEDENTS OF LUSJ

Method

Experimental Design

Experiment I tested the ﬁrst three hypotheses regarding the antecedents of LUSJ.
The availability of staff member past judgment accuracy (fed back or not fed back to the
leader) and the availability of staff member judgment conﬁdence (fed back or not fed
back to the leader) were crossed in an environment in which differential staff weighting
was the appropriate staff weighting strategy (resulting from differences in the validity of
the information available to the staff members).
Participants

Eighty-four undergraduates enrolled in introductory management, psychology,
and communications classes at Michigan State University participated as leaders in this
study in exchange for course credit. Forty-ﬁve percent of the sample were male. The
mean age of the sample was 21.54. Anticipating moderate to large effect sizes, a power
analysis indicated a sample size of 84 would be adequate for a power of at least .80 across
all analyses at alpha < .05 (Cohen, 1988). Post hoc power analyses indicated that the

power was greater than .90 across all analyses at alpha < .05 (Cohen, 1988).

49

50
Task Description

The task was a team version of the TIDE2 (Team Interactive Decision Exercise for
Teams Incorporating Distributed Expertise; Hollenbeck, Sego, Ilgen, Maj or, Hedlund &
Phillips, 1995) computerized decision making task. TIDE2 is a software program for a
decision-task simulation. Research participants are presented with values on a number of
attributes (cues) about an object or situation and then make judgments and decisions
regarding that object (e.g., the medical status of a patient or the threat of an aircraft).
Alternatively, the team leader can be presented only with the summary judgments of staff
members and make decisions regarding the object on the basis of these summary
judgments. Participants are taught in pre-task training how to utilize the program and, if
the leader, how to interpret and combine the cues to make judgments and decisions about
the decision object.

In this experiment, the program was conﬁgured to represent a military decision
making situation where participants were leaders who, together with three staff members,
were charged with the responsibility of interpreting how to react to aircraft in their
airspace on a seven-point passive to aggressive scale. Each of the three staff members
(the commanding officers of a coastal air defense team (CAD), Air reconnaissance plane
(AWAC) and a Cruiser) measured, interpreted and combined a unique subset of three
cues about an incoming aircraﬂ, interpreted this cue information, sent to the leader a
summary aggressiveness judgment to the leader and, if the condition called for it, a
conﬁdence assessment regarding his or her self-perceived judgment accuracy. The leader
was not able to measure any cue information directly but relied on the three staff

members for summary judgments on their subset of the decision problem. The leader

5 1
was in the same room with the staff members and was responsible for combining the staff

recommendations into the team decision. The leader's decisions were the team’s
decisions, and were the decisions on which team decision accuracy (and any performance
rewards) was based.

The CAD’s responsibility was to summarize information relevant to the target’s
location, the AWAC’s was to summarize information about the target’s movement, and
the CRUISER summarized information relevant to the classiﬁcation of the target (what
type of plane it is). Each staff member therefore had something unique to contribute to
the decision. Each staff member saw three unique pieces of information, and their roles
did not overlap.

Teams addressed a total of 63 aircraft targets, presented one at a time. After
performing three practice trials lasting 300, 240, and 100 seconds, teams had 60 seconds
to perform each of the remaining 60 trials. Staff members measured cue attributes and
formulated and registered, via computer, a summary judgment to the leader on a seven
point aggressiveness scale (see Appendix B) along with a judgment conﬁdence
assessment if the condition called for it. When sent, each staff member's judgment
conﬁdence was received by the leader at the same time as the staff member's judgment.
Conﬁdence was communicated as a number reﬂecting the staff member's perception that
his/her judgment is accurate, in percentage terms. Thus, a communicated conﬁdence of
90 reﬂected that staff member's belief that they were 90% conﬁdent that their judgment is
accurate. Conﬁdence ratings were communicated on a 1% (reﬂecting low conﬁdence) to
100% (reﬂecting extreme conﬁdence) scale. The leader then registered the team decision,

and a feedback screen appeared for 5 seconds. This feedback screen contained decision

52
accuracy information for both the leader and the staff members on the previous target as

well as cumulative performance information for the team (based on the leader's
decisions).

For conditions requiring staff past judgment accuracy to be fed back to the leader,
a red validity bar appeared on the leader’s computer screen beneath the icon reﬂecting
each of the positions in the team. The bars were on the screen for the duration of each
target, and were removed only during the 5 second feedback period. Each bar reﬂected
the validity of that staff member's previous judgments (the correlation of that staff
member's judgments with the criterion), with a longer bar reﬂecting a higher correlation
between the staff member's judgments and the decision criterion. A number reﬂecting the
actual correlation between the staff member’s judgments and the decision criterion
appeared to the left of each red bar. The correlation and length of each bar were updated
after each trial, and varied naturally during the course of the experiment. The correlation
and colored bar were accurate beginning on the seventh target. When the condition called
for staff members to send the leader a judgment conﬁdence assessment, the assessment
appeared on the leader’s computer screen just below that station’s red validity bar.

Each staff member processed three unique cues, which combined additively to
predict the criterion. The cues were created such that the correlation between each staff
member's combination of three cues and the criterion was different (approximately .75 for
the CAD, .50 for the AWAC, and .25 for the Cruiser). It was stressed during training that

each staff member had something unique to contribute to the team’s decision.

53
Task Training

Two types of training were conducted for this experiment. The ﬁrst involved
asking participants to read a brief general training manual (see Appendix B) and a
position-speciﬁc training manual. The general training manual consisted of an overview
of the simulation and the roles in the team. The position-speciﬁc manual consisted of
information pertaining to the leader’s role in the team, instructions on how to combine
staff judgments into the team decision, and performance information. The second part of
the training was hands-on focusing on the mechanics of performing the task, where the
participants performed three practice trials as a team under the guidance of the researcher
(see Appendix B for a script of this training). Questions were answered only during this
initial training period.
Measures

Lima. The data were divided into two time periods: early and late. Target 7 was
chosen as the ﬁrst relevant target because this was the ﬁrst target for which the staff
members’ red validity bars were found to be stabilizing and valid in pilot work. The
early time period consisted of trials 7-34. The late time period consisted of trials 35-63.

Cognitive Ability. Participants’ college admissions scores on the American
College Test (ACT) were used as a measure of cognitive ability. The reliability of the
ACT is .96 (American College Testing Program, 1989). There is almost unanimous
consensus among researchers in the area of general cognitive ability that such tests are
highly g-loaded (Jensen, 1986; Gottfredson & Crouse, 1986; Hunter, 1986). Participants
authorized the researchers to obtain these scores from the university by signing the

consent form. One participant did not have ACT scores on record but did have Scholastic

54
Aptitude Test (SAT) scores. The ACT score was estimated from the participant’s SAT

score. Three ACT scores were obtained from the participant by phone or by mail because
the university did not have the ACT or SAT scores on record. For the eight participants
who had neither ACT nor SAT scores on record and who could not be contacted by
phone or by mail, the mean of participants’ ACT scores across the sample was
substituted.

Past Accuracy Availability. The availability of staff members’ past accuracy to
the leader was manipulated through the presence or absence of red validity bars and
accompanying correlation between the staff member’s past judgments and the correct
decision.

degment Conﬁdence Availability. Staff members’ judgment conﬁdence was
manipulated through the conﬁguration of the computer program. Staff members either
could or could not send a conﬁdence assessment to the leader along with their judgment,
depending on the condition.

Spaff Member Past Judgment AccuracLLevel. The past judgment accuracy of
each staff member was operationalized as the unstandardized b weight for each staff
member resulting from a regression of the correct decision on staff members' judgments.
Because the effects of time were investigated, staff member past judgment accuracy was
divided into accuracy on targets 7-34 (early) and targets 35-63 (late). To facilitate
computations, the decimal was dropped from the unstandardized b weight.

While the b weight operationalization of staff members’ past accuracy was used in
the analyses, during the simulation leaders were fed back staff past judgment accuracy in

the form of a red bar and accompanying number reﬂecting the correlation between the

55
staff member’s judgments and the correct decisions. The correlation between the

correlational and b weight operationalizations of staff past judgment accuracy was .81.
When the analyses were rerun using the correlational operationalization, the results were
consistent with those found using the b weights.

Self-Report Staff Member Past Accuracy. Aﬁer the simulation, leaders were
asked to indicate how accurate (on a scale of 0-100) they felt each of their staff members
had been in predicting the correct responses. Their response was used as their self-report
of that staff member’s past accuracy.

_S_taff Conﬁdence Level. Staff members in conditions requiring the provision of
staff member judgment conﬁdence to the leader entered a conﬁdence assessment on a
scale of 1 (reﬂecting low conﬁdence) to 100 (reﬂecting extreme conﬁdence) on each
target reﬂecting their conﬁdence that their judgment would be correct. Each staff
member's average conﬁdence level across targets 7-34 and across targets 35-63 was used
as his/her conﬁdence score for the early and late targets. As two of the four conditions
did not allow for the provision of conﬁdence information to the leader, only half of the
leaders have staff conﬁdence level data.

Dyadic LUSJ. Dyadic LUSJ was operationalized as the unstandardized b weight
for a staff member resulting from a regression of the leader's decision on all three staff
members' judgments entered as a block. Each leader therefore has three unique dyadic
LUSJ scores, one for each staff member. Dyadic LUSJ can be positive or negative, with
higher b weights reﬂecting higher dyadic LUSJ. A negative dyadic LUSJ score would
indicate that the leader tended to do the opposite of the recommendations of that staff

member.

56
There are two issues that cannot be overlooked with regard to dyadic LUSJ. First,

multicollinearity, or the existence of a substantial correlation among independent
variables (e.g., staff judgments), can lead to three main problems (Cohen & Cohen,
1983). First, the substantive interpretation of the partial coefﬁcients (dyadic LUSJ) will
be difficult. Because the independent variables (staff judgments), by deﬁnition, lay claim
to largely the same portion of the variance in the dependent variable (leader decisions),
individual staff judgments may not make much of a unique contribution to the prediction
of leader decisions. Second, multicollinearity will make the dyadic LUSJ b weights
unstable by increasing the standard error of the b weights (dyadic LUSJ). As the standard
error increases, the conﬁdence interval widens for individually predicted leader decisions.
This widened conﬁdence interval lessens the probability of rejecting the null hypothesis
that any partial correlation or regression coefﬁcient for a given predictor (e.g., staff
member A's judgments) is zero. Third, as the intercorrelation among staff judgments
approaches 1.0, errors associated with the computation of the regression weights (dyadic
LUSJ scores) can become potentially serious (Cohen & Cohen, 1983; Dillon &
Goldstein, 1984). In research involving LUSJ, then, it is important that staff judgments
be as unrelated as possible. In this dissertation this is accomplished by utilizing a
laboratory environment and retaining control over the relationships among the underlying
cues, and the cues available to each staff member. In this way, independent relationships
among staff judgments were built into the study, reducing the likelihood of
multicollinearity among staff judgments.

Second, to the extent that the leader's preliminary judgment prior to receiving staff

judgments or the information known by the leader is collinear with the information

57
available to a staff member, the staff member's LUSJ weight will be inﬂated if the effect

of the leader's initial judgment prior to receiving staff judgments is not controlled. The
fact that the leader may have a personal judgment prior to making the ﬁnal team decision
means that there is a factor other than staff judgments that may inﬂuence the actual
weight given these judgments by the leader. This initial personal judgment is information
that is critical to the calculation of the weight the leader then gives staff judgments, and
must be controlled before the weights given staff judgments can be calculated. In this
dissertation, this was controlled by limiting the leader's knowledge of the decision
situation to the information provided by staff members' summary judgments. Leaders in
this study had no decision information other than that provided by staff members'
judgments.

This approach was also employed by Hollenbeck et al. (1995), although they did
not partial the leader's initial judgment before entering staff judgments into the
regression. Again, to facilitate computations the decimal was dropped from the
unstandardized b weight.

Self-Report Waightingaf Staff Member. After the simulation, leaders were asked
to indicate how much they felt they weighted (utilized) the judgments of each of staff
member in making their decisions. They were told to divide 100 points across each of the
staff members in a manner that reﬂected how they thought they weighted the judgments
of each staff member during the simulation.

LUSJ Accuracy. The accuracy of a leader's dyadic LUSJ weights for staff
members was operationalized as the sum of the absolute values of the differences

between the unstandardized b weights for each staff member resulting from a regression

58
of the correct decision on staff members' judgments (appropriate weight) and each staff

member's dyadic LUSJ (actual weight). Higher scores reﬂect greater leader misweighting
of their staff members, while lower scores reﬂect more accurate staff weighting. The
closer LUSJ accuracy is to zero, the closer the match between appropriate and actual
leader weighting across all staff members. Again, the decimal was dropped from each
score to facilitate computations. Each leader has one accuracy of LUSJ score. This
approach was also used by Hollenbeck et al. (1995).

Dyadic LUSJ Variability Dyadic LUSJ variability is the degree of variability
across a leader's dyadic LUSJ scores. It was operationalized as the mean absolute
deviation, or the average of the absolute deviations of each staff member's dyadic LUSJ
from the average dyadic LUSJ weight for that leader across all staff members (dyadic
LUSJ average. Each leader therefore has only one dyadic LUSJ variability score. The
formula used for the computation of dyadic LUSJ variability was:

Dyadic 3

LUSJ = Z |Dyadic LUSJ Average-Dyadic LUSJSI

Variability 3:1
3

Higher scores therefore reﬂect greater dyadic LUSJ variability, or greater variability in
the dyadic LUSJ weights the leader gave to each of his/her four staff members. Lower
scores reﬂect less variability in the dyadic LUSJ weights used by the leader (or a more
equal weighting strategy). The decimal was dropped from the score to facilitate

computations.

59
Appendix A illustrates the application of all of the LUSJ indices across a variety

of possible data.
Procedure
Experiment I lasted approximately three hours. Participants signed up during
class time for a three-hour session. Upon arriving for the experiment, participants were
asked to sign in (to receive course credit). If other experiments were taking place
simultaneously (in separate rooms), participants were randomly assigned to one of the
experiments. Participants were given a consent form to read and sign (see Appendix B)
and told that the top three teams in each condition would receive a cash prize as detailed
in the consent form. To create a more realistic environment in which the leader could be
concerned that his/her behavior could have ramiﬁcations in terms of staff behavior,
participants were told that they would be performing two decision making tasks over the
next three hours. They were informed that the ﬁrst task would be done as a team, and
that the second may or may not be performed as a team or even with the same teammates.
It was stressed that while positions would be assigned for the ﬁrst task, they would be
given input as to how they wanted to perform the second task. Any questions were
answered, and participants were given an individual differences questionnaire to
complete. To enhance the credibility of the leader, participants were told that some of
their responses to this questionnaire would be used to assign them to one of the positions
in the team.
When all participants had ﬁnished, the questionnaires were collected and

participants were given 10 minutes to read a brief description of the task they would be

performing (see Appendix B) while the researcher went in the other room to “score” the

60
questionnaires. The researcher randomly assigned participants to the four positions in the

team, and gave participants a manual of position-speciﬁc training information (see
Appendix C). Participants completed a demographic questionnaire (see Appendix B) and
were given 10 minutes to study their position-speciﬁc training manual. Both the general
and speciﬁc training manuals were kept by the participants when they performed the
simulation. When this was completed, questions were again answered and participants
received approximately 15 minutes of hands-on computer training on the task (see
Appendix B for a script of this training). After completing the three practice trials,
participants began the experimental simulation.

At the conclusion of the simulation, participants were asked to complete a brief
questionnaire that assessed their perceptions of staff accuracy and their weighting of staff
members (see Appendix C). After performing a second, shorter decision making task as
individuals, participants were debriefed (see Appendix C), thanked, and released. All
participants were treated in accordance with the ethical standards of the American

Psychological Association (APA, 1992).

6 1
Results

Experiment I was designed to test hypotheses relevant to the antecedents of leader
utilization of staff judgments. Time was incorporated as a factor in the design, and, when
appropriate, leader cognitive ability was used as a covariate in the analyses. Unless
otherwise stated, repeated measures regression was used for the analyses (Cohen &
Cohen, 1983). The repeated measures regression technique divides the overall variance
in the dependent variable into within- and between-portions and systematically analyzes
each portion separately. Statistical inferences are then determined on the basis of the
relevant sources of variance (Gully, 1994). This approach was also used by Hollenbeck,
Ilgen and Sego (1994) in a longitudinal study of teams. Because sample sizes differed
across the analyses due to the variables used, tables of means, standard deviations and
intercorrelations are presented for each analysis.

Hypotheses 1a proposed that the presence of staff cumulative past judgment
accuracy would lead to greater LUSJ variability than the absence of this information.

The dependent variable for the analysis was LUSJ variability. The independent variables
for Hypothesis la were time (a within-leader variable), entered ﬁrst, followed by leader
ability and the presence or absence of cumulative staff past accuracy information (both
between-leader variables). The interaction of the availability of staff past accuracy and
time was entered last. The means, standard deviations and intercorrelations of the
variables used in testing Hypothesis 1a are presented in Table 1a. The results of

Hypothesis 1a are presented in Table 1b.

62
Table la. Means, Standard Deviations and Intercorrelations of Variables in Table 1b

 

Variable Mean SD 1 . 2. 3.
1. Time .50 .50
2. Ability 21.90 2.92 .00
3. Past Accuracy
(Present or Absent) .50 .50 .00 .20*

4. LUSJ Variability 12.15 6.82 .34“ -.02 .23**

 

Note. N=168
*p<.05. **p<.01.

Table 1b. Repeated Measures Regression Analysis of Time and the Presence of
Cumulative Past Accuracy Information on LUSJ Variability

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)

Time 4.59 .114“ .302" 35.86 (1,83)
Ability -.05 .001 .002 .13 (1,82)
Past Accuracy (present

or absent) 3.29 .056“ .090” 8.03 (1,81)
Past Accuracy X Time 3.49 .016 .042“ 5.29 (1,82)

Total R2 .187" .342" .092*

 

Note. The higher the score is, the greater the LUSJ variability. N=168. Between-subjects
variance=46.28 (62%), within-subjects variance=l7.49 (38%). Total df within-
subjects=84, total df between-subjects=83.

*p<.05. **p<.01.

Variance partitioning revealed that 62% of the total variance in LUSJ variability
was due to between-subjects variance and 38% was due to within-subjects variance. This
indicates that almost two-thirds of the total variance in LUSJ variability was due to
situational factors or individual differences of the leaders. The results show that time
accounted for 30% of the within-subjects variance. LUSJ variability increased from the

ﬁrst to the second half of the simulation. In support of Hypothesis 1, the provision of

staff member cumulative past judgment accuracy accounted for 9% of the between-

63
subjects variance. Leaders who were provided staff member cumulative past judgment

accuracy showed greater LUSJ variability. The interaction of the provision of staff past
judgment accuracy and time accounted for 4% of the within-subjects variance, and is
illustrated in Figure 7. This interaction shows that leaders provided staff past judgment
accuracy showed a larger increase in LUSJ variability over time than leaders without staff
past judgment accuracy.

Hypothesis 1b proposed that for leaders provided staff member cumulative past
judgment accuracy, higher staff member past judgment accuracy would be more
positively related to dyadic LUSJ than it would for leaders not provided this information.
First, dyadic LUSJ was regressed on the within-leader variables of time, the past
judgment accuracy level of staff members, and their interactions. The between-leader
variable of the presence of staff member cumulative past accuracy information was then
entered, and the within by between variable interactions were entered last. The means,
standard deviations and intercorrelations of the variables used in testing Hypothesis 1b

are presented in Table 2a. The results of Hypothesis lb are presented in Table 2b.

Table 2a. Means, Standard Deviations and Intercorrelations of Variables in Table 2b

Variable Mean SD 1 . 2. 3.

 

1. Time .50 .50
2. Past Accuracy Level 40.93 39.43 .12"
3. Past Accuracy

 

(Present or Absent) .50 .50 .00 .02
4. LUSJ 39.72 16.46 -.01 .42" -.03
Note. N=504

*p<.05. **p<.01.

LUSJ Variability

15

A -L
-a- w
—~—-~ {FF—w —-7 4 ___ ‘ *ﬁi——— - —-+» —ﬁ —- a

64

    

 

Eady

Time

—+— Past

Late

Accuracy
Not
Available

Past
Accuracy
Available

Figure 7. Hypothesis 1a: Interaction Effect of Time and

Cumulative Past Accuracy on LUSJ Variability

65

Table 2b. Repeated Measures Regression Analysis of Time, the Level of Staff Member
Past Judgment Accuracy and the Presence of Cumulative Past Accuracy
Information on Dyadic LUSJ

 

 

 

A in A in A in Incremental F

Variable b Total R2 Within R2 Between R2 (df,df)
Time -.44 .000 .000 0 (1,419)
Past Accuracy Level .18 .181" .199M 103.95 (1,418)
Time X Past Accuracy

Level .13 .022" .024” 13.00 (1,417)
Past Accuracy (present

or absent) -1.11 .001 .001 .91 (1,82)
Past Accuracy Level X
Past Accuracy
Availability .15 .031** .034" 19.11 (1,416)
Time X Past Accuracy
Availability -2.21 .001 .001 .62 (1,415)
Past Accuracy Level X

Past Accuracy

Availability X Time .10 .004 .004 2.47 (1,414)

Total R2 .240" .262" .001

 

Note. The higher the score is, the greater the LUSJ level. N=504. Between-subj ects
variance=24.66 (9%); within-subjects variance=245.87 (91%). Total (if within-

subjects=420; total df between-subjects=83.

‘p<.05. **p<.01.

66
Variance partitioning revealed that only 9% of the total variance in LUSJ weights

was due to between-subjects variance and 91% was due to within-subjects variance. This
indicates that almost none of the total variance in LUSJ was due to situational factors or
individual differences of the leaders. Given that leaders were completely dependent on
staff members for information on which to base their decisions, this distribution of within
and between variance in LUSJ is logical. The results for Hypothesis lb indicate that staff
member past accuracy level had a strong positive relationship with dyadic LUSJ
(accounting for 20% of the within-subjects variance), which increased with time. The
interaction between time and staff member past accuracy level accounted for an
additional 2% of the within-subjects variance in dyadic LUSJ and is illustrated in Figure
8. Over time, leaders’ weighting strategies changed such that less accurate staff members
received lower weight, and more accurate staff members received greater weight. The
largest difference over time is the leaders’ discounting of the less accurate staff members
rather than their increased weighting of the more accurate staff members.

The availability of staff member cumulative past accuracy also interacted with
staff member past accuracy level in inﬂuencing LUSJ, supporting Hypothesis 1b. This
effect accounted for 3% of the within-subjects variance in dyadic LUSJ and is illustrated
in Figure 9. Higher staff member past judgment accuracy was more positively related to
dyadic LUSJ for leaders provided staff member cumulative past judgment accuracy
information than for leaders not provided this information. When staff cumulative past
accuracy information was not available, leaders utilized a more equal weighting strategy,
although the less accurate staff members were discounted slightly, and the more accurate

staff members were given slightly greater weight. The availability of staff cumulative

LUSJ

55 4+-

50

40

67

+ Staff Past
Accuracy
-1 SD (1.5)

» -—I-- Staff Past
Accuracy
+1 SD (46)

 

25:

l
1
20+—— - —————— —

Eany

Late

Figure 8. Hypothesis 1b: Interaction Effect of Time and Staff

Past Accuracy Level on LUSJ

LUSJ

68

 

35 _T. + Cumulative Past
1 Accuracy Not
1 Available

30 T — --— Cumulative Past
' / Accuracy
f Available
1

25 +
l
I

20 +~——-—~ 4 am «A i A a —- ——1

-1 SD (1.5) +1 SD (80.4)

Staff Past Accuracy Level

Figure 9. Hypothesis 1b: Interaction Effect of Cumulative
Past Accuracy Availability and Staff Past Accuracy Level on
LUSJ

69
past accuracy information led to greater discounting of less accurate staff members and

increased weighting of more accurate staff members.

Hypothesis 1c proposed that the availability of staff member cumulative past
judgment accuracy to the leader would be positively related to LUSJ accuracy. To test
this hypothesis, LUSJ accuracy was regressed on time (a within-leader variable), leader
cognitive ability, the presence or absence of staff cumulative past accuracy information
and their interaction (between-leader variables), and the interaction of time and the
presence of staff cumulative past accuracy information. The means, standard deviations
and intercorrelations of the variables used in testing Hypothesis 1c are presented in Table

3a. The results for Hypothesis 1c are summarized in Table 3b.

Table 3a. Means, Standard Deviations and Intercorrelations of Variables in Table 3b

 

Variable Mean SD 1 . 2. 3.
1. Time .50 .50
2. Ability 21.90 2.92 .00

3. Past Accuracy
(Present or Absent) .50 .50 .00 .20*
4. LUSJ Accuracy 29.61 12.20 -.35** .05 -.20*

 

lﬁtp, N=168
*p<.05. **p<.01.

Variance partitioning revealed that 51% of the total variance in LUSJ accuracy
was due to between-subjects variance and 49% was due to within-subj ects variance. This
indicates that half of the total variance in LUSJ accuracy was due to situational factors or
individual differences of the leaders. The results for Hypothesis 1c indicate that leader

LUSJ accuracy improved over time and with the presence of staff cumulative past

70
judgment accuracy. Time accounted for 26% of the within-subjects variance and the

presence of staff cumulative past judgment accuracy information accounted for 9% of the
between-subjects variance in LUSJ accuracy. Hypothesis lc was therefore supported.

Table 3b. Repeated Measures Regression Analysis of Time and the Presence of Staff
Cumulative Past Accuracy Information on Leaders’ LUSJ Accuracy Across Staff

 

 

Members
A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)

Time -8.60 .125" .256” 28.56 (1,83)
Ability .20 .002 .004 .32 (1,82)
Past Accuracy (present

or absent) -5.20 .044" .086“ 7.65 (1,81)
Past Accuracy X Time -1 .71 .001 .002 .23 (1,82)

Total Rr .172" .258" .09*

 

 

Note. The higher the score is, the lower the leader’s LUSJ accuracy. N=168. Between-
subjects variance=75.73 (51%); within-subjects variance=72.25 (49%). Total (if within-
subjects=84; total df between-subjects=83.

*p<.05. **p<.01.

Hypothesis 2a proposed that the presence of staff judgment conﬁdence would
result in greater LUSJ variability. To test this proposition, LUSJ variability was ﬁrst
regressed on the within-leader variable of time. The between-leader variables of leader
ability and the presence or absence of staff judgment conﬁdence were then entered,
followed by the interaction of time and the presence or absence of staff judgment
conﬁdence. The means, standard deviations and intercorrelations of the variables used in
testing Hypothesis 2a are presented in Table 4a. The results for Hypothesis 2a are

summarized in Table 4b. The results indicate that the passage of time resulted in greater

LUSJ variability, but staff judgment conﬁdence availability had no effect. Thus,

71
Hypothesis 2a was not supported. Time accounted for 30% of the within-subjects

variance in LUSJ variability. It is possible that the negative skewness of judgment
conﬁdence (mean of 75, standard deviation of 10) detracted from leaders’ ability to use

this information as a means of determining how to weight staff judgments.

Table 4a. Means, Standard Deviations and Intercorrelations of Variables in Table 4b

 

Variable Mean SD 1 . 2. 3.
1. Time .50 .50
2. Ability 21.90 2.92 .00
3. Conﬁdence
(Present or Absent) .50 .50 .00 -.06

4. LUSJ Variability 12.15 6.82 .34" -.02 .07

 

Note. N=168
*p<.05. **p<.01.

Table 4b. Repeated Measures Regression Analysis of Time and the Presence of Staff
Judgment Conﬁdence on LUSJ Variability

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)

Time 4.59 .114“ .302" 35.86 (1,83)
Ability -.05 .001 .002 .13 (1,82)
Conﬁdence (present

or absent) .87 .004 .006 .52 (1,81)
Conﬁdence X Time -1.18 .002 .005 .63 (1,82)

Total R2 .121M .307“ .008

 

m, The higher the score is, the greater the LUSJ variability. N=168. Between-subjects
variance=28.79 (62%); within-subjects variance=l7.49 (3 8%). Total df within-
subjects=84; total df between-subjects=83.

*p<.05. **p<.01.

Hypothesis 2b proposed that a staff member’s conﬁdence level would be

positively related to that staff member’s dyadic LUSJ. Because all variables involved in

72

this regression are within-leader variables, hierarchical regression was used in the

analyses. Dyadic LUSJ was regressed on time, staff member past accuracy level, staff

member conﬁdence level and the interactions of these variables. The means, standard

deviations and intercorrelations of the variables used in testing Hypothesis 2b are

presented in Table 5a. The results of Hypothesis 2b are presented in Table 5b.

Table 5a. Means, Standard Deviations and Intercorrelations of Variables in Table 5b

 

 

Variable Mean SD 1 . 2. 3.
1. Time .50 .50
2. Past Accuracy Level 40.46 37.47 .12
3. Conﬁdence Level 74.73 10.01 .07 .07
4. Dyadic LUSJ 40.49 16.62 -.04 .32" .23"
Ma. N=252

*p<.05. **p<.01.

Table 5b. Hierarchical Regression Analysis of Time, Staff Member Past Accuracy Level
and the Level of Staff Member Judgment Conﬁdence on Dyadic LUSJ

 

 

 

A in Incremental F
Variable b Total R2 (df,df)
Time -1.26 .001 .36 (1,250)
Past Accuracy Level .14 .104" 28.84 (1,249)
Conﬁdence Level .36 .046" 13.50 (1,248)
Time X Conﬁdence
Level -.17 .003 .78 (1,247)
Past Accuracy Level X
Conﬁdence Level -.01 .008 2.47 (1,246)
Time X Past Accuracy
Level .09 .010 3.04 (1,245)
Time X Past Accuracy
Level X Conﬁdence
Level .00 .000 .10 (1,244)
Total R2 .172"

 

Note. The higher the score is, the greater the LUSJ level. N=252. Total df=251.

*p<.05. **p<.01.

73
The results support Hypothesis 2b and indicate that after partialling the effect of staff

member past accuracy level, which accounted for 10% of the variance in dyadic LUSJ,
staff member judgment conﬁdence level had a statistically signiﬁcant positive effect on
LUSJ. Staff member judgment conﬁdence level accounted for an additional 5% of the
variance in dyadic LUSJ. Leaders gave staff members reporting higher judgment
conﬁdence greater weight than staff members reporting lower conﬁdence when making
the team’s decisions. Leader utilization of staff member judgment conﬁdence did not
change as a result of time.

The effect of the availability of staff members’ conﬁdence levels on the accuracy
of staff members’ dyadic LUSJ weights was also investigated. In this analysis, LUSJ
accuracy was ﬁrst regressed on the within-leader variable of time. Ability and the
presence of staff member judgment conﬁdence (both between-team variables) were then
entered, followed by the interaction of time and the presence of staff member conﬁdence.
The results indicate that the presence of staff member judgment conﬁdence had no effect
on leaders’ LUSJ accuracy, although LUSJ accuracy did improve over time. Time
accounted for 26% of the within-subjects variance in LUSJ accuracy. The means,
standard deviations and intercorrelations of the variables used in the analysis are
presented in Table 6a. The results of this analysis are summarized in Table 6b.

Hypothesis 3 proposed that the availability of both staff member cumulative past
judgment accuracy and staff member judgment conﬁdence information would be related
to greater variability in leaders’ LUSJ weights across staff members than if only one or
neither type of information was available. This hypothesis was tested by regressing

LUSJ variability on time (a within-leader variable), leader ability (a between leader

74
variable), staff member judgment conﬁdence availability, staff member cumulative past

accuracy availability, and the interactions of time, conﬁdence availability and cumulative
past judgment accuracy availability. The means, standard deviations and intercorrelations
of the variables used in the analysis are presented in Table 7a. The results for this

analysis are summarized in Table 7b.

Table 6a. Means, Standard Deviations and Intercorrelations of Variables in Table 6b

 

 

Variable Mean SD 1 . 2. 3.
1. Time .50 .50
2. Ability 21.90 2.92 .00
3. Conﬁdence
(Present or Absent) .50 .50 .00 -.06
4. LUSJ Accuracy 29.61 12.20 -.35** .05 .02
l\_lat_e_. N=168

*p<.05. **p<.01.

Table 6b. Repeated Measures Regression Analysis of Time and the Presence of Staff
Judgment Conﬁdence on Leaders’ LUSJ Accuracy Across Staff Members

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)

Time -8.60 .125M .256* 28.56 (1,83)
Ability .20 .002 .004 .32 (1,82)
Staff Judgment

Conﬁdence (present

or absent) .61 .001 .002 .16 (1,81)
Conﬁdence (present or

absent) X Time .21 .000 .000 0 (1 ,82)

Total R2 .128" .256" .006

 

1% The higher the score is, the lower the leader’s LUSJ accuracy. N=168. Between-
subjects variance=75.73 (51%); within-subjects variance=72.25 (49%). Total df within-
subjects=84; total df between-subjects=83.

*p<.05. **p<.01.

75

Table 7a. Means, Standard Deviations and Intercorrelations of Variables in Table 7b

 

Variable Mean SD 1 . 2. 3. 4.

1. Time .50 .50
2. Ability 21.90 2.92 .00
3. Conﬁdence

(Present or Absent) .50 .50 .00 -.06
4. Past Accuracy

(Present or Absent) .50 .50 .00 .20* .00
5. LUSJ Accuracy 29.61 12.20 -.35** .05 .02 -.20

 

Note. N=168
*p<.05. **p<.01.

Table 7b. Repeated Measures Regression Analysis of Time, the Availability of
Cumulative Staff Past Accuracy Information and the Presence of Staff Judgment
Conﬁdence on Leaders’ LUSJ Variability

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Time 4.59 .114" .302W 35.86 (1,83)
Ability -.05 .001 .002 .13 (1,82)
Staff Judgment
Conﬁdence
(present or absent) .87 .004 .006 .52 (1,81)
Past Accuracy (present
or absent) 3.27 .056** .09“ 7.98 (2,80)
Conﬁdence X Past
Accuracy -4.90 .032” .051* 4.78 (1,79)
Time X Conﬁdence -1.18 .002 .005 .63 (1,82)
Time X Past Accuracy 3.49 .016 .042* 5.27 (1,81)
Conﬁdence (present or
absent) X Past
Accuracy X Time -5.24 .009 .024 3.04 (1,80)
Total R2 .234“ .373" .149*

 

Note. The higher the score is, the greater the leader’s LUSJ variability. N=168.
Between-subjects variance=28.79 (62%); within-subjects variance=l7.49 (3 8%). Total df
within-subjects=84; total df between-subj ects=83.

*p<.05. **p<.01.

76
The results support Hypothesis 3. Greatest LUSJ variability was shown by

leaders provided staff past accuracy but not staff judgment conﬁdence information.
Lowest LUSJ variability was shown by leaders provided neither staff judgment
conﬁdence nor staff past accuracy information. Although positive relationships were
found between LUSJ variability and time and LUSJ variability and the availability of
staff member cumulative past judgment accuracy, no relationship was found between
LUSJ variability and the provision of staff judgment conﬁdence. Time accounted for
30% of the within-subjects variance in LUSJ variability, and the presence of staff
member past judgment accuracy accounted for 9% of the between-subj ects variance.

An effect on LUSJ variability was found for the cumulative past accuracy
information by judgment conﬁdence availability interaction. This effect accounted for
5% of the between-subjects variance in LUSJ variability, and is illustrated in Figure 10.
The nature of this effect is such that the relationship between conﬁdence availability and
LUSJ variability is negative when staff past accuracy information is available, and
positive when staff past accuracy information is not available.

A signiﬁcant time by the availability of staff cumulative past accuracy interaction
was also found, and is illustrated in Figure 11. This effect accounted for an additional 4%
of the within-subjects variance in LUSJ variability, and mirrors the effect found in
Hypothesis 1a. LUSJ variability increased more over time when leaders were fed back
staff members’ cumulative past judgment accuracy than when this information was not
fed back to leaders.

The effect of the availability of staff members’ judgment conﬁdence levels on the

accuracy of staff members’ dyadic LUSJ weights was also investigated. In this analysis,

LUSJ Variability

77

   

+ Cumulative

11

1 Past
Accuracy

j Not

9 ' Available
1 1 - - Cumulative
I Past
l Accuracy

7 . Available
1
l

5 l_____ _ R. L _ . , ._ ._ _.

Not Available Available

Conﬁdence Availability

Figure 10. Hypothesis 3: Interaction Effect of Confidence
Availability and Cumulative Past Accuracy Availability on LUSJ
Variability

LUSJ Variability

78

  
 

+ Cumulative

Past
9 Accuracy
1 Not
T Available
'. - 9| - Cumulative
7 T Past
T Accuracy
5 Available
5 1%.... MAW..- A... _-._ __.__. m.
Early Late
Time

Figure 11. Hypothesis 3a: Interaction Effect of Time and
Cumulative Past Accuracy Availability on LUSJ Variability

79
LUSJ accuracy was ﬁrst regressed on the within-leader variable of time. Ability and the

presence of staff member judgment conﬁdence (both between-team variables) were then
entered, followed by the interaction of time and the presence of staff member conﬁdence.
The means, standard deviations and intercorrelations of the variables used in the analysis

are presented in Table 8a.

Table 8a. Means, Standard Deviations and Intercorrelations of Variables in Table 8b

 

Variable Mean SD 1 . 2. 3. 4.
1. Time .50 .50
2. Ability 21.90 2.92 .00
3. Conﬁdence
(Present or Absent) .50 .50 .00 -.06
4. Past Accuracy
(Present or Absent) .50 50 .00 .20* .00

5. LUSJ Variability 12.15 6.82 .34” -.02 .07 .23**

 

131% N=168
*p<.05. **p<.01.

The results, summarized in Table 8b, indicate that the presence of staff member
judgment conﬁdence had no effect on leaders’ LUSJ accuracy alone or in combination
with staff member past judgment accuracy. The presence of staff member past accuracy
did increase leaders’ LUSJ accuracy, and accounted for 9% of the between-subjects
variance in LUSJ accuracy. LUSJ accuracy also improved over time, which accounted
for 26% of the variance in LUSJ accuracy.

The ordering of cell means indicated that leaders provided only staff cumulative
past accuracy were the most accurate in weighting their staff (31.61), followed by leaders

provided both staff member judgment conﬁdence and staff cumulative past accuracy

80

Table 8b. Repeated Measures Regression Analysis of Time, the Availability of
Cumulative Staff Past Accuracy Information and the Presence of Staff Judgment
Conﬁdence on Leaders’ LUSJ Accuracy Across Staff Members

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Time -8.60 .125” .256” 28.56 (1,83)
Ability .20 .002 .004 .32 ( 1,82)
Staff Judgment
Conﬁdence (present or
absent) .61 .000 .000 0 (1,81)
Past Accuracy (present
or absent) -5.21 .044** .086" 7.56 (1,80)
Conﬁdence X Past
Accuracy -1 .49 .001 .002 .17 ( 1,79)
Time X Conﬁdence .21 .000 .000 0 (1,82)
Time X Past Accuracy -1 .71 .001 .002 .22 ( 1,81)
Conﬁdence (present or
absent) X Past
Accuracy X Time -4.00 .002 .004 .44 (1,80)
Total R2 .175M .262” .092

 

11(11- The higher the score is, the lower the leader’s LUSJ accuracy. N=168. Between-
subjects variance=75.73 (51%); within-subjects variance=72.25 (49%). Total (If within-
subjects=84; total df between-subjects=83.

*p<.05. **p<.01.

81
(32.27), leaders with neither (35.76) and leaders with only staff member judgment

conﬁdence (35.97).

Discussion

The results of Experiment I indicate that in an environment in which differentially
weighting staff judgments leads to greater decision accuracy, the provision of staff
member cumulative past accuracy to leaders is related to greater weighting variability and
weighting accuracy than if this information is not available. Leaders’ weighting accuracy
improved over time, and the variability of leaders’ weights across the staff members
increased over time. However, leaders who were fed back staff cumulative past accuracy
information used it in deciding how to weight staff judgments, and did so in a way that
led to more appropriate staff utilization and greater weighting variability. Although all
leaders saw feedback on the accuracy of each staff member after each target, leaders
relied more on staff member cumulative past accuracy in making their decisions when
this information was fed back to them than when it was not.

Staff member judgment conﬁdence information was also utilized by leaders in
making the team’s decisions, although its availability was unrelated to both LUSJ
variability and LUSJ accuracy. In fact, the correlation between staff member judgment
conﬁdence level and staff member judgment accuracy was only .07 (n.s.), yet the
correlation between staff member judgment conﬁdence level and leader weighting of that
staff member was .23 (p<.01). Comparable miscalibration of conﬁdence is also found
when the relationship between staff conﬁdence level and staff judgment accuracy is

examined by staff member. The correlation between judgment conﬁdence and judgment

82
accuracy for staff members with the most accurate cue information was .06 (p<.01); for

staff members with the least accurate cue information was .01 (n.s.), and for staff
members whose cue accuracy was in the middle it was -.01 (n.s.). Thus, although
leaders’ weighting variability and weighting accuracy improved with time, their
utilization of judgment conﬁdence, which was unrelated to judgment accuracy, detracted
from leaders’ appropriate weighting of staff members.

An additional ﬁnding concerns leaders’ self-reported perceptions of their staff
members’ accuracy levels. The higher a staff member’s judgment conﬁdence, the greater
the leader’s self-report of that staff member’s judgment accuracy after the simulation had
ended (r=.20; p<.01). This indicates that despite its poor calibration, staff member
judgment conﬁdence was not only used by the leader in making the team’s decisions, it
contributed to a positive bias in leaders’ recollections of staff member judgment
accuracy. This is supportive of the argument that leaders consider conﬁdence to be an
indicator of accuracy, and that this is why it is utilized by leaders in weighting staff input.
The availability of staff judgment conﬁdence and staff cumulative past accuracy to
leaders did not interact in affecting staff weighting accuracy nor staff weighting
variability.

Ability was not found to be a signiﬁcant predictor of how effectively leaders
utilized their staff. Effects for ability were not detected when ability was expanded in the
regression equations and interacted with the other variables, with the exception of a
statistically signiﬁcant (p < .05) three-way interaction effect of ability, past accuracy
availability, and conﬁdence availability on LUSJ variability. The nature of this effect

was such that conﬁdence availability was slightly negatively related to LUSJ variability

83
when past accuracy was available regardless of leader cognitive ability. For leaders

lower in cognitive ability, conﬁdence availability was negatively related to LUSJ
variability, although this relationship was positive among leaders higher in cognitive
ability. This may be due to the fact that higher ability leaders had staff members report
lower conﬁdence levels than lower ability leaders (r=-.l9, p < .01).

The relatively simple information processing requirements of the task for the
leader probably restricted the role played by cognitive ability in this study. As cognitive
ability is thought to be related to greater information processing capacity (Kanfer &
Ackerman, 1989), it is possible that a leader’s cognitive ability would be related to
his/her ability to appropriately discriminate among staff members in weighting their
judgments and making the team decision with a more complex decision task.

This study helps to shed light on the conﬂict in the existing literature on leaders’
tendency to differentially weight staff members. Leaders in this study exhibited greater
weighting variability and weighting accuracy when provided staff cumulative past
accuracy information. This is consistent with work on LMX and VDL (Dansereau, Graen
& Haga, 1975; Graen & Cashman, 1975; Scandura, Graen & Novak, 1986) that has
shown that leaders tend to seek out relationships of differential inﬂuence with staff
members based, at least in part, on staff member performance and ability. The fact that
leaders were positively inﬂuenced by the conﬁdence assessments of staff members in
making the team’s decisions is consistent with work on judge-advisor systems that has
shown that leaders are willing to differentially utilize staff members, and that conﬁdence

is a mechanism for this inﬂuence.

84
This study’s ﬁndings are also consistent with Brehmer and Hagafors’ (1986) work

that found that leaders without staff member judgment conﬁdence or staff member
cumulative past accuracy information were unable to adopt a differential weighting
strategy when it was appropriate. In this study, particularly on the ﬁrst half of the
simulation, leaders without staff past accuracy information utilized a more equal
weighting strategy than did leaders provided this information. Thus, it is reasonable to
conclude that staff member cumulative past judgment accuracy, when provided to the
leader, promotes the adoption of a more appropriate differential weighting strategy.
Judgment conﬁdence, on the other hand, is utilized by the leader, but the lack of
calibration of judgment conﬁdence makes this information unable to consistently
improve leaders’ staff weighting accuracy.

The implications of these ﬁndings, and directions for future research, will be

discussed in the last chapter. The next chapter will discuss Experiment II.

Chapter 4
EXPERIMENT II: CONSEQUENCES OF LUSJ
Mama

Experimental Design

Experiment II tested Hypotheses 4-6. The experimental design consisted of the
crossing of high and low team performance with an equal or differential leader weighting
strategy. A confederate played the role of leader, while participants were staff members
on four-person teams. In equal weighting conditions, the leader averaged the three staff
responses in making the team decision. In differential accuracy conditions, the leader
weighted the CAD judgment .5, the AWAC judgment .33 and the Cruiser judgment .17.
In low performance conditions, the confederate leader added a pre-determined random
error term to the decision.
Participants

Two hundred twenty-eight participants were recruited from introductory
management, psychology and communications classes and received course credit in
return for their participation in this study. Fifty-ﬁve percent of the sample were male,
forty-ﬁve percent were female. The mean age of the sample was 21.2. Anticipating a
moderate effect size, a power analysis indicated that a sample size of 228 would result in
a power of at least .80 across all analyses at alpha < .05 (Cohen, 1988). Post hoc power

analyses indicated that the power exceeded .80 for all analyses (Cohen, 198 8).

85

86
Task Description

The task was the same team version of the TIDE2 (Team Interactive Decision
Exercise for Teams Incorporating Distributed Expertise; Hollenbeck, Sego, Ilgen, Major,
Hedlund & Phillips, 1995) computerized decision making task discussed for Experiment
I. The computer program was again conﬁgured as a military decision making simulation,
but participants were one of three staff members (rather than the leader): either the C0 of
the Aegis cruiser (Cruiser), the Coastal Air Defense unit (CAD), or the AWACS
reconnaissance plane. Each of these three staff members was responsible for measuring,
interpreting and combining the same unique subset of three cues about an incoming
aircraft as in Experiment I. The leader (Carrier) was a confederate stationed in another
room who either equally or differentially weighted staff recommendations in making the
team’s decision. The same person was the confederate for all teams. The leader's
decisions were again the team’s decisions, on which team performance (and any
performance rewards) was based.

Participants were asked to make judgments about a total of 63 aircraft targets,
presented one at a time. The ﬁrst three targets were practice trials lasting 300, 240, and
100 seconds. Staff members measured cue attributes and formulated and registered, via
computer, a summary judgment to the leader on a seven point aggressiveness scale (see
Appendix B) along with a judgment conﬁdence assessment. The last 60 targets counted
toward the bonus money. Participants were given 60 seconds for each target to measure
cue attributes and to formulate and register, via computer, a summary judgment to the
leader. These judgments were made on a seven point scale representing increasing

amounts of force (see Appendix D).

87
After the leader registered the team decision, a feedback screen appeared for 5

seconds. This feedback screen contained decision accuracy information for both the
leader and the staff members on the previous target as well as cumulative performance
information for the team (based on the leader's decisions). Team performance was
determined by the accuracy of the leader's ﬁnal decision. The point scoring system is
described in Appendix D.

The level of the leader's past utilization (weighting) of their judgments were fed
back to staff members in both conditions. Information regarding the leader's past
utilization of each staff member was in the form of a green bar. The length of the green
bar reﬂected the weight the leader had given that staff member's judgments on all targets
encountered to that point (the correlation between that staff member’s judgments and the
leader’s decisions). The number reﬂecting the actual correlation between that staff
member’s judgments and the leader’s decisions was positioned immediately to the left of
each green bar. Each staff member saw the green bars and correlations for all three staff
members. These bars remained on the computer screen for the duration of each target,
and were removed only during the 5 second feedback period. The length of the bars was
updated after each trial, and were accurate beginning on the eighth target. The cues and
targets were the same as in Experiment I. Staff members sent conﬁdence assessments to
the leader as described in Experiment I, and were told that the leader’s screen displayed a
red bar reﬂecting the accuracy of each staff member’s previous judgments.

At the conclusion of the experiment, participants were debriefed (see Appendix

D), thanked, and released. Monetary awards earned by participants were delivered within

88
three weeks of the end of the experiment. All participants were treated in accordance

with the ethical standards of the American Psychological Association (APA, 1992).
Task Training

The training for Experiment 11 was the same as for Experiment 1. Participants
ﬁrst read a brief training manual (see Appendix B) containing general information
pertaining to the type of task the participants would be performing, their role as a staff
member of their team and performance information. They then read a position-speciﬁc
manual containing information on their role in the team and the cues for which they
would be responsible (see Appendix D). The second part of the training was hands-on
task training focusing on the mechanics of performing the task, where the participants
performed three practice trials under the guidance of the researcher (see Appendix B for a
script of this training). Questions were answered only during this initial training period.
Measures

_T_e_arn Performance. Team performance was manipulated in two ways. For teams
in the low accuracy condition the confederate leader added a pre-determined random error
to the decision, decreasing the team’s point total. Additionally, at the end of the
simulation participants in the low team performance condition were told by the researcher
that they had not performed well enough to be contenders for the bonus money.
Participants in the high team performance condition were told that they had done well and
had a good chance of earning bonus money. Team performance was operationalized as
the point total earned by the team over the 60 non-practice targets. The scoring system is

described in Appendix D. Team performance could range from -l20 to +120 points.

89
Dyadic LUSJ. Because the on-screen display of how the leader had weighted

staff members on all previous targets was displayed as the correlation between that staff
member’s judgments and the leader's decisions on all 63 targets, this correlation was
utilized as the dyadic LUSJ score. Each staff member therefore had a unique dyadic LUSJ
score. The correlation reﬂects the strength of association between the staff member’s
judgments and the decisions ultimately made by the leader. This approach has also been
used in previous research (e.g., Brehmer & Hagafors, 1986; Tucker, 1964). To facilitate
computations, the decimal was dropped from the dyadic LUSJ score.

Self-Report LUSJ. After the simulation, staff members were asked to indicate
how much they felt the leader weighted (utilized) the judgments of each of staff member
in making decisions over the course of the simulation. They were told to divide 100
points across each of the staff members in a manner that reﬂected how they thought the
leader had weighted the judgments of each staff member during the simulation. Their
rating of the leader’s weighting of their judgments was used as their self-report LUSJ.

Dyadic LUSJ Accurzgy, The accuracy of a leader's dyadic LUSJ was
operationalized as the absolute difference between the dyadic LUSJ score for a particular
staff member and the staff member's appropriate LUSJ score based on their judgment
accuracy. It was calculated by taking the absolute value of the result of subtracting the
correlation between the staff member’s judgments and the correct decisions from the staff
member’s correlational dyadic LUSJ (actual weight). A higher score therefore reﬂects
greater misweighting of a staff member. A score of zero reﬂects perfect utilization of that
staff member. Each staff member has one dyadic LUSJ score. The decimal was again

removed from the score to facilitate computations.

90
Relative Dyadic LUSJ. Relative dyadic LUSJ was operationalized as the target

staff member's correlational dyadic LUSJ score centered about the mean of all of the
dyadic LUSJ scores of that team. Centering is a technique that puts variables in deviation
score form so that their mean is zero (Aiken & West, 1991). The relative dyadic LUSJ
score is calculated by subtracting the mean of the staff member's leader's raw (not
absolute) dyadic LUSJ scores (leader's dyadic LUSJ mean) from the target staff member's
raw dyadic LUSJ score. Staff members with high relative dyadic LUSJ scores have
therefore been given greater inﬂuence by the leader in the leader's decisions than staff
members with lower relative dyadic LUSJ scores. Again, the decimal was removed from
the score.

By using raw, rather than absolute, dyadic LUSJ scores, this operationalization of
relative dyadic LUSJ preserves information about both the magnitude and the direction of
the target staff member's dyadic LUSJ relative to the leader's dyadic LUSJ across all staff
members. Because information about the direction of the dyadic LUSJ scores (positive or
negative) is important in determining a staff member's utilization by the leader relative to
the other staff members, it is important that raw dyadic LUSJ scores rather than the
absolute values be utilized (unless, of course, all dyadic LUSJ scores are positive).

Staff Member Willingness to Return. Staff member willingness to return to the
lab in the future for pay was measured with a six-item scale developed for this study.
Items reﬂect participants' willingness to work with the leader and staff members again in
the future, as well as their willingness to work on the task again. Scale items are listed in

Appendix D. The reliability of this scale was .95, with 6 items retained. The range of the

91
scale was 1, reﬂecting lower willingness to return, to 5, reﬂecting greater willingness to

return.

_S_t_aff Member Desire for Change on the Next TasL Staff members’ desire to
change the way they performed the next task was measured with a three-item scale
developed for this study. Scale items are listed in Appendix D. The reliability of this
scale was .58, with 3 items retained. The range of the scale was 0, reﬂecting less desire
for change on the next task, to 3, reﬂecting greater desire for change.

Staff Member Satisfaction with Leeﬁarp Staff member satisfaction with the leader
was assessed using a scale adapted from Seashore, Lawler, Mirvis and Carnmann (1982).

Scale items are presented in Appendix D. The reliability of this scale was .89, with 8
items retained. The range of the scale was 1, reﬂecting lower satisfaction with the leader,
to 5, reﬂecting greater satisfaction with the leader.

Task Withdrawal. Staff member task withdrawal was measured with an ll-item
scale adapted from Baker (1991) and Gilliland (1992). Scale items are listed in Appendix
D. The reliability of this scale was .82, with 9 items retained. The range of this scale was
1, reﬂecting lower task withdrawal, to 5, reﬂecting greater withdrawal.

Staff Member Self-Efﬁca_cy_. Staff member self-efficacy was measured with an 8-
item scale adapted from Locke, Frederick, Lee & Bobko (1984). Scale items are listed in
Appendix D. The reliability of this scale was .90, with 8 items retained. The range of

this scale was 1, reﬂecting lower self-efﬁcacy, to 5, reﬂecting greater self-efﬁcacy.

92
Procedure

Experiment 11 lasted approximately three hours. Participants signed up during
class time for a three-hour session. Upon arriving for the experiment, participants were
asked to sign in (to receive course credit). If other experiments were taking place
simultaneously (in separate rooms), participants were randomly assigned to one of the
experiments. Participants were given an appropriate consent form to read and sign (see
Appendix B) and told that the top three teams in each condition would receive a cash
prize as detailed in the consent form. To create a more realistic environment in which the
staff could react to what occurred in their team, participants were told that they would be
performing two decision making tasks over the next three hours. They were informed
that the ﬁrst task would be done as a team, and that the second may or may not be
performed as a team or even with the same teammates. It was stressed that while
positions would be assigned for the ﬁrst task, they would be given input as to how they
wanted to perform the second task. Any questions were answered, and participants were
given an individual differences questionnaire to complete. To enhance the credibility of
the leader, participants were told that some of their responses to this questionnaire would
be used to assign them to one of the positions in the team.

When all participants had ﬁnished, the questionnaires were collected and
participants were given 10 minutes to read a brief description of the task they would be
performing (see Appendix B) while the researcher went in the other room to “score” the
questionnaires. The researcher randomly assigned participants to the three staff member
positions in the team, and gave participants a manual of position-speciﬁc training

information (see Appendix D). Participants completed a demographic questionnaire (see

93
Appendix B) and were given 10 minutes to study their position-speciﬁc training manual.

Both the general and speciﬁc training manuals were kept by the participants when they
performed the simulation. When this was completed, questions were again answered and
participants received approximately 15 minutes of hands-on computer training on the task
(see Appendix B for a script). After completing the three practice trials, participants
began the experimental simulation.

At the conclusion of the simulation, participants were asked to complete a series
of questionnaires. After performing a second, shorter decision making task as
individuals, participants were debriefed, thanked, and released. All participants were
treated in accordance with the ethical standards of the American Psychological
Association (APA, 1992).

3.6%
Data Analysis

Experiment 11 was designed to test hypotheses relevant to staff reactions to
different forms of LUSJ. Study means, standard deviations and intercorrelations are
presented in Table 9. All hypotheses for Experiment 11 (Hypotheses 4a-6e) were tested
using repeated measures regression analysis. The dependent variable was the appropriate
staff reaction for the particular hypothesis, and variance partitioning was used in
determining statistical signiﬁcance. Appendix F contains tables of means, standard
deviations and intercorrelations for each analysis.

Hypothesis 4 proposed that leader decision accuracy (team performance) would

moderate the effect of LUSJ on ﬁve different staff reactions. To test the hypotheses,

94

.58.... .movm.

 

.v _ NHZ 523 he Ewe-_- onz :o owﬁﬁu Lou 260D 3088 833.5, =m .5.“ 3ng .082

 

 

1.0m.- mo. no. :3. 3. a3. 2. If. Iva. hm. 3+ SEEM-bow A:
*aovr .ILN. tam.- N—f not I. —N.- .I.mm.- .3..th v0. NTN _wBNHwﬁma xmmh d
IE..- 1.2. co. oo. 8. 1.2“. aamm. on. _N.m Sumo-H 5:5 cocoaﬁumm .w

8.- No. 5.- _ m m m- .3ri mm. as. xmﬂ. :82
co omSEU 8% “Emma .N.
co. we. no. *2. If. N: 36 530m 8 magma—=3 .o
3.- a... K. tam. no. 3.2 3.3” BSA toned-bum .m
no. mo. 3.- wod— v6: .8833. 334 omega .v
$3.. co. 3.2 cc. RDA Sega gum—om .m
1mm. v3: 3.? RDA .m
2.6— mmdm ouggotom Each.—

.a .w <- 6 .m .v .m .m ._ Om :82 033:3

= EoEtonm E “#53 333:; mo macaw—ohoobaﬁ use .mcocmSoD wee—O55 £532 6 033-

95
LUSJ (a within-team variable) was entered ﬁrst, followed by team performance (a

between team variable), followed by their interaction. Hypothesis 4a investigated staff

member willingness to return, and the results are summarized in Table 10.

Table 10. Repeated Measures Regression Analysis of LUSJ and Team Performance on

 

 

 

Willingness to Return
A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ .01 .026M .042“ 6.61 (1,151)
Team Performance .01 .013* .034 2.62 (1,74)
Performance X LUSJ .00 .014* .023 3.62 (1,150)
Total R2 .053" .065" .034

 

Mtg, The higher the score is, the greater the willingness to return. N=228. Between-team
variance=.48 (3 8%); within-team variance=.78 (62%). Total dfwithin-team=152; total df
between-team=75.
*p<.05. **p<.01.

Variance partitioning revealed that 38% of the total variance in staff member
willingness to return was due to between-team variance and 62% was due to within-team
variance. This indicates that just over one-third of the total variance in staff member
willingness to return was due to between-team factors such as team performance or
situational factors, and just under two-thirds of the total variance was due to within-team
factors such as LUSJ or individual differences of the team members.

The results of Hypothesis 4a indicate that the level of the leader’s weighting of a
staff member was positively related to that staff member’s willingness to return,
accounting for 4% of the within-team variance. Team performance accounted for 3% of

the between-team variance. Team performance also interacted with LUSJ in predicting

staff member willingness to return, accounting for 2% of the within-team variance

Willingness to Return

96

 

‘-
' ,, I
4.3 l
- +Team
- Performance
33 i -1 SD (14.6)
l -I — - Team
Performance
+1 SD (46)
I
2,-qu _ __ __ _ _ __ _ ,
-1 SD (24.8) +1 SD (61.3)

LUSJ

Figure 12. Hypothesis 4a: Interaction Effect of Team
Performance and LUSJ on Willingness to Return

97
supporting Hypothesis 4a. This effect is illustrated in Figure 12, and shows that when

team performance was low, LUSJ had a negative relationship with staff member
willingness to return. When team performance was high, LUSJ was positively related to
staff member willingness to return. The nature of this effect is slightly different from the
hypothesized effect. Rather than LUSJ being positively related to staff reactions for both
high and low team performance, LUSJ was negatively related to staff reactions when
team performance was low. Staff members whose judgments received higher weight but
whose teams performed poorly were the least willing to return, while staff members
whose judgments received greater weight but whose teams performed well were the most
willing to return.

The effect sizes changed slightly when team performance was entered into the
regression equation ﬁrst, followed by LUSJ and the interaction term. Team performance
accounted for 3% (p<.01) of the total variance in staff member willingness to return and
9% (p_<.OI) of the between-team variance. LUSJ accounted for 1% (n.s.) of the total
variance, and 1% (n.s.) of the within-team variance.

Hypothesis 4b investigated the moderating effect of team performance on the
relationship between LUSJ and staff member desire for change on the next task. The
results of the analyses are summarized in Table 11. Because N=214 for all analyses
involving staff member desire to change on the next task due to incomplete responses, the
mean for LUSJ in this analysis was 31.62 with a standard deviation of 10.66, and for
performance the mean was 30.85 with a standard deviation of 15.55.

Variance partitioning revealed that 43% of the total variance in staff member

desire to change for the next task was due to between-team variance and 57% was due to

98
within-team variance. This indicates that just under half of the total variance in staff

member desire to change for the next task was due to between-team factors such as team
performance or situational factors, and just over half of the total variance was due to
within-team factors such as LUSJ or individual differences of the team members.
Providing partial support for Hypothesis 4b, LUSJ accounted for 3% of the within-team
variance and team performance accounted for 15% of the between-team variance in staff
members’ desire to change for the next task. Staff were more interested in changing the
way the next task was performed (e.g., changing the leader, changing the team or working
as an individual) when the team either did poorly or when the staff member had lower
LUSJ. However, no moderating effect of team performance was found on the effect of
LUSJ on desire for change on the next task.

Table 11. Repeated Measures Regression Analysis of LUSJ and Team Performance on
Desire to Change for Next Task

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ -.01 .018* .032* 4.61 (1,141)
Team Performance -.02 .063M .146" 11.96 (1,70)
Performance X LUSJ .00 .000 .000 0 (1,140)
Total R2 .081 ** .032 .146"

 

Note. The higher the score is, the greater the desire to change for the next task. N=214.
Between-team variance=.37 (43%); within-team variance=.49 (57%). Total df within-
team=142; total df between-team=71.
*p<.05. "p<.01.

The effect sizes changed when the order of entry for LUSJ and team performance

was reversed. Team performance accounted for 8% (p<.01) of the total variance, and

19% (p<.01) of the between-team variance in staff member desire for change on the next

99
task. LUSJ did not account for any of the variance aﬁer controlling for team

performance.

Hypothesis 4c investigated the moderating effect of team performance on the
relationship between LUSJ and staff member satisfaction with the leader. The results of
Hypothesis 4c are summarized in Table 12.

Table 12. Repeated Measures Regression Analysis of LUSJ and Team Performance on
Satisfaction With Leader

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ .02 .137“ .349" 81.01 (1,151)
Team Performance .02 .170" .280" 28.75 (1,74)
Performance X LUSJ .00 .004 .010 2.39 (1,150)
Total RI .311” .359” .280“

 

Note. The higher the score is, the greater satisfaction with the leader. N=228. Between-
team variance=.35 (61%); within-team variance=.23 (39%). Total df within-team=152;
total df between-team=75.

*p<.05. **p<.01.

Variance partitioning revealed that 61% of the total variance in staff member
satisfaction with their leader was due to between-team variance and 39% was due to
within-team variance. This indicates that just under two-thirds of the total variance in
staff member satisfaction with their leader was due to between-team factors such as team
performance or situational factors, and just over one-third of the total variance was due to
within-team factors such as LUSJ or individual differences of the team members.

The results for Hypothesis 4c show that both LUSJ and team performance were

positively related to staff member satisfaction with the leader, providing partial support

for Hypothesis 4c. LUSJ accounted for 35% of the within-team variance in satisfaction

100
with the leader, and team performance accounted for 28% of the between-team variance.

No interaction effect between team performance and LUSJ on satisfaction with the leader
was observed.

The results for Hypothesis 4c changed when the order of entry for LUSJ and team
performance was reversed. Team performance accounted for 30% (p<.01) of the total
variance in staff member satisfaction with the leader, and 49%

(p<.01) of the between-team variance. LUSJ accounted for an additional 1% (n.s.) of the
total variance, and 2% (p<.05) of the within-team variance after controlling for team
performance.

Hypothesis 4d investigated the moderating effect of team performance on the
relationship between LUSJ and task withdrawal. Variance partitioning revealed that 44%
of the total variance in staff members’ self-reported task withdrawal was due to between-
team variance and 56% was due to within-team variance. This indicates that just under
half of the total variance in staff member self-reported task withdrawal was due to
between-team factors such as team performance or situational factors, and just over half
of the total variance was due to within-team factors such as LUSJ or individual
differences of the team members.

The results are summarized in Table 13, and indicate that LUSJ was negatively
related, and team performance was unrelated, to task withdrawal. LUSJ accounted for
20% of the within-team variance in LUSJ. In support of Hypothesis 4d, the interaction of
team performance and LUSJ was signiﬁcant, accounting for 4% of the within-team

variance in task withdrawal, and is illustrated in Figure 13. The nature of the interaction

Task Withdrawal

101

i +Team
, Performance
l -1 SD (14.6)
l — —-- -Team
l Performance
+ +1 SD (46)

 

 

 

2.4 l

l

l

l

2.2 «‘-

l
l ‘

l

l

2 l

l

l

18 j
. + , ——4

-1 SD (24.8) +1 SD (61.3)

LUSJ

Figure 13. Hypothesis 4d: Interaction Effect of Team
Performance and LUSJ on Task Vlﬁthdrawal

102
is slightly different than that proposed, in that LUSJ was more negatively related to task

withdrawal when team performance was higher than when team performance was lower.
Task withdrawal was lowest when team performance and LUSJ were both high. Having
a higher LUSJ weight was unable to compensate for the higher task withdrawal reported
by staff members on more poorly performing teams. Interestingly, the highest task
withdrawal was reported by the lower weighted staff members on high performing teams.
The results for Hypothesis 4d changed slightly when the order of entry for LUSJ
and team performance was reversed. Team performance accounted for 7% (p<.01) of
the total variance in task withdrawal, and 16% (p<.01) of the between-team variance
when entered into the regression equation ﬁrst. LUSJ accounted for an additional 5%

(p<.01) of the total variance, and 9% of the within-team variance (p<.01).

Table 13. Repeated Measures Regression Analysis of LUSJ and Team Performance on

 

 

 

Task Withdrawal
A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ -.01 .110** .196" 36.82 (1,151)
Team Performance -.01 .011 .025 1.90 (1,74)
Performance X LUSJ -.01 .021* .037W 7.32 (1,150)
Total R2 .142" .233” .025

 

Note. The higher the score is, the greater the task withdrawal. N=228. Between-team
variance=.18 (44%); within-team variance=.23 (56%). Total dfwithin-team=152; total df
between-team=75.
*p<.05. *"‘p<.01.

Hypothesis 4e proposed that team performance would moderate the effect of

LUSJ on staff member self-efﬁcacy. Variance partitioning revealed that 45% of the total

variance in staff members’ self-efﬁcacy was due to between-team variance and 55% was

103
due to within-team variance. This indicates that just over one-third of the total variance

in staff member self-efficacy was due to between-team factors such as team performance
or situational factors, and just under two-thirds of the total variance was due to within-
team factors such as LUSJ or individual differences of the team members.

The results for this hypothesis are summarized in Table 14, and show that both
LUSJ and team performance are positively related to staff member self-efﬁcacy,
providing partial support for Hypothesis 4e. LUSJ accounted for 6% of the within-team
variance, and team performance accounted for 7% of the between-team variance in self-
efﬁcacy. However, team performance and LUSJ did not interact in predicting self-

efﬁcacy.

Table 14. Repeated Measures Regression Analysis of LUSJ and Team Performance on

 

 

 

Self-Efﬁcacy
A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ .01 .033M .06“ 9.60 (1 ,151)
Team Performance .01 .031 ** .069* 5.55 (1,74)
Performance X LUSJ .00 .011 .02 3.25 (1,150)
Total R2 .075” .08“ .069*

 

Note. The higher the score is, the greater the self-efﬁcacy. N=228. Between-team
variance=.14 (45%); within-team variance=.18 (55%). Total dfwithin-team=152; total df
between-team=75.
*p_<.05. “p<.01.

When team performance was entered into the regression ﬁrst, followed by LUSJ,

team performance accounted for 6% (p_<.01) of the total variance in staff member self-

efﬁcacy, and 13% (p<.01) of the between-team variance. LUSJ accounted for less than

104
1% (n.s.) of the total and 1% (n.s.) of the within-team variance in staff member self-

efﬁcacy after team performance was partialled.

Hypothesis 5 investigated the moderating effect of team performance on the
relationship between staff members’ relative LUSJ and the same ﬁve staff reactions
tested in Hypothesis 4. To test the hypotheses, relative LUSJ (a within-team variable)
was entered ﬁrst, followed by team performance (a between-team variable) and the
interaction.

Hypothesis 5a proposed that the relationship between a staff member’s relative
LUSJ and the staff member’s willingness to return would be moderated by team
performance. The results for Hypothesis 5a are summarized in Table 15. Team
performance was positively related and relative LUSJ was unrelated to willingness to
return. Team performance accounted for 9% of the between-team variance in willingness
to return. Relative LUSJ and team performance interacted in affecting staff members’
willingness to return, accounting for 4% of the within-team variance. This effect is
illustrated in Figure 14. The nature of this effect was such that a positive relationship
existed between relative LUSJ and willingness to return among members of higher
performing teams, while the relationship was negative for members of lower-performing
teams. Staff members with higher LUSJ on higher-performing teams were the most
willing to return, while staff members with lower LUSJ on lower-performing teams were
the least willing to return. The results did not change when the order of entry for relative

LUSJ and team performance was reversed.

Wilingness to Return

105

 

4.8 T
I
l
4.3 f—
l - , - , -' "’ "I +Team
' ,- , ,, ' I Performance
1 .__ -, _. ~ -1 SD(14.6)
-—I--Team
Performance
+1 so (46)
3.3 T
'-
l
2.8 l_-,. T. T ,, - f -,v___- __ .
-1 SD (-13.6) +1 so (13.6)

Relative LUSJ

Figure 14. Hypothesis 5a: Interaction Effect of Team
Performance and Relative LUSJ on Willingness to Return

106

Table 15. Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Willingness to Return

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Relative LUSJ .01 .005 .008 1.23 (1,151)
Team Performance .01 .033” .087" 7.03 (1,74)
Performance X
Relative LUSJ .01 .025" .040* 6.36 (1,150)
Total R2 .063" .048* .087"

 

Mtg, The higher the score is, the greater the willingness to return. N=228. Between-
team variance=.48 (3 8%); within-team variance=.78 (62%). Total df within-team=152;
total df between-team=75.

*p_<.05. Mp<.01.

Hypothesis 5b proposed that team performance would moderate the effect of
relative LUSJ on staff member desire to change for the next task. Again, because N=214
for analyses involving staff member desire to change for the next task, the mean for
relative LUSJ was .03 with a standard deviation of 13.70, and the mean for performance
was 30.85 with a standard deviation of 15.55. Results for Hypothesis 5b are summarized
in Table 16, and do not support Hypothesis 5b. Although a negative relationship was
found between team performance and desire for change on the next task, accounting for
19% of the between-team variance, no relationship was found to exist between relative
LUSJ and desire for change on the next task. The interaction between the two variables
was not signiﬁcant. The higher the team performed, the less staff members wanted to

change in the way the next task was performed. The results did not change when the

order of entry for relative LUSJ and team performance was reversed.

107

Table 16. Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Desire to Change for Next Task

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Relative LUSJ .01 .012 .021 3.04 (1,141)
Team Performance -.02 .080" .185" 15.92 (1,70)
Performance X
Relative LUSJ .00 .003 .005 .76 (1,140)
Total R2 .095" .026 .185"

 

N91; The higher the score is, the greater the desire to change for the next task. N=214.
Between-team variance=.37 (43%); within-team variance=.49 (57%). Total df within-
team=l 42; total df between-team=71.

*p_<.05. **p<.01.

Hypothesis 5c proposed that team performance would moderate the effect of
relative LUSJ on staff member satisfaction with the leader. The results of Hypothesis 5c
are summarized in Table 17, and indicate that team performance had a strong positive
effect on staff satisfaction with the leader, accounting for 49% of the between-team
variance. Contrary to Hypothesis 5c, relative LUSJ and the interaction between relative
LUSJ and team performance had no effect on staff satisfaction with the leader. The
results did not change when the order of entry of team performance and relative LU SJ
was reversed.

Hypothesis 5d proposed that the effect of relative LUSJ on task withdrawal would
be moderated by team performance. The results are summarized in Table 18, and provide
partial support for this hypothesis. Relative LUSJ and team performance were both

negatively related to task withdrawal, but the interaction of the two was not signiﬁcant.

Relative LUSJ accounted for 8% of the within-team variance, and team performance

108
accounted for 16% of the between-team variance in task withdrawal. The results did not

change when the order of entry for relative LUSJ and team performance was reversed.

Table 17. Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Satisfaction With Leader

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Relative LUSJ .00 .000 .000 0 (1,151)
Team Performance .03 .298“ .49M 71.22 (1,74)
Performance X
Relative LUSJ .00 .003 .008 1.16 (1,150)
Total R2 .301 ** .008 .49"

 

N_o_t§, The higher the score is, the greater satisfaction with the leader. =228. Between-
team variance=.35 (61%); within-team variance=.23 (39%). Total dfwithin—team=152;
total df between-team=75.

*p_<.05. Mp<.01.

Table 18. Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Task Withdrawal

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Relative LUSJ -.01 .044" .078W 12.85 (1,151)
Team Performance -.01 .071 ** .162" 14.28 (1,74)
Performance X
Relative LUSJ .00 .006 .011 1.76 (1 ,150)
Total R2 .121" .089" .162M

 

Note. The higher the score is, the greater the task withdrawal. N=228. Between-team
variance=.18 (44%); within-team variance=.23 (56%). Total dfwithin-team=152; total df
between-team=75.
*p<.05. **p<.01.

Hypothesis 5e proposed that team performance would moderate the effect of

relative LUSJ on self-efﬁcacy. Table 19 summarizes the results. In partial support of

Hypothesis 5e, both team performance and relative LUSJ were positively related to self-

109
efﬁcacy. Relative LUSJ accounted for 3% of the within-team, and team performance

accounted for 13% of the between-team variance in self-efficacy. The interaction
between relative LUSJ and team performance did not have an effect. The results did not
change when the order of entry for relative LUSJ and team performance was reversed.

Table 19. Repeated Measures Regression Analysis of Relative LUSJ and Team
Performance on Self-Efﬁcacy

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
Relative LUSJ .01 .016* .029* 4.51 (1,151)
Team Performance .01 .060" .134" 11.44 (1,74)
Performance X
Relative LUSJ .00 .001 .002 .28 (1,150)
Total RT .077" .031 .134"

 

Note. The higher the score is, the greater the self-efﬁcacy. N=228. Between-team
variance=.14 (45%); within-team variance=.18 (55%). Total dfwithin-team=152; total df
between-team=75.

*p<.05. "p<.01.

Hypothesis 6 tested the proposition that the accuracy of the leader’s weighting
strategy would affect the ﬁve staff reactions explored in Hypotheses 4 and 5, moderated
by team performance. To test these hypotheses, dyadic LUSJ accuracy (a within-team
variable) was entered into the regressions ﬁrst, followed by team performance (a between
team variable) and their interaction.

Hypothesis 6a proposed that the effect of dyadic LUSJ accuracy on staff member
willingness to return would be moderated by team performance. Table 20 summarizes

the results for this hypothesis. Team performance was found to be positively related to

staff member willingness to return, accounting for 9% of the between-team variance.

110
Although dyadic LUSJ accuracy had no direct effect, the interaction between dyadic

LUSJ accuracy and team performance accounted for 7% of the within-team variance in
staff member willingness to return. The interaction is illustrated in Figure 15. The nature
of the interaction is such that dyadic LUSJ accuracy was positively related to willingness
to return in lower-performing teams, but slightly negatively related to willingness to
return in higher-performing teams. Receiving a more accurate weight by the leader led to
a fairly high willingness to return regardless of team performance, while being
inaccurately weighted on a lower-performing team led to the lowest willingness to return.
The results changed slightly when the order of entry for LUSJ accuracy and team
performance was reversed. Team performance accounted for 3% (p<.01) of the total
variance in staff member willingness to return, and 9% (p<.01) of the between-team
variance. Dyadic LUSJ accuracy accounted for less than 1% (n.s.) of the total and 1%
(n.s.) of the within-team variance in willingness to return after partialling the effect of
team performance.

Table 20. Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Willingness to Return

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ Accuracy .00 .001 .002 0 (1,151)
Team Performance .01 .035M .092" 6.56 (1,74)
Performance X LUSJ
Accuracy -.01 .042” .068W 11.47 (1,150)
Total R2 .078" .07** .092”

 

Egg; The higher the score is, the greater the willingness to return. N=228. Between-
tearn variance=.48 (3 8%); within-team variance=.78 (62%). Total dfwithin-team=152;
total df between-team=75.

*p<.05. **p_<.01.

VWlingness to Return

111

4.3

  
 

—9— Team

3'8 Performance
-1 SD (14.6)
» «I— -Team
T Performance
l +1 SD (46)
3.3 (r
l
l
l
l
l
2.8 L— ___- T T _ «___-.- WA. A.___ .3
-1 SD (5.1) +1 SD (30.4)

Dyadic LUSJ Accuracy

Figure 15. Hypothesis 6a: Interaction Effect of Team
Performance and Dyadic LUSJ Accuracy on VWllingness to
Return

112
Hypothesis 6b proposed that dyadic LUSJ accuracy’s effect on staff member

desire to change for the next task would be moderated by team performance. Because
N=214 for this analysis, the mean for LUSJ accuracy was 17.48 with a standard deviation
of 12.91, and the mean for performance was 30.85 with a standard deviation of 15.55.
The results of the analyses are presented in Table 21. The results do not support
Hypothesis 6b. Only team performance was negatively related to staff member desire to
change for the next task, accounting for 19% of the between-team variance. The results
did not change when the order of entry for dyadic LUSJ accuracy and team performance
was reversed.

Table 21. Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Desire to Change for Next Task

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ Accuracy .00 .000 .000 0 (1,141)
Team Performance -.02 .081" .188M 16.16 (1,70)
Performance X LUSJ
Accuracy .00 .002 .004 .49 (1,140)
Total R2 .083" .004 .188"

 

Note. The higher the score is, the greater the desire to change for the next task. N=214.
Between-team variance=.37 (43%); within-team variance=.49 (57%). Total df within-
team=142; total df between-team=71.
*p<.05. **p<.01.

Hypothesis 6c proposed that the effect of dyadic LUSJ accuracy on staff member
satisfaction with the leader would be moderated by team performance. The results are

presented in Table 22, and do not provide support for the hypothesis. Only team

performance was positively related to staff member satisfaction with the leader,

113
accounting for 50% of the between-team variance. Dyadic LUSJ accuracy and its

interaction with team performance had no effect on staff satisfaction with the leader.

Table 22. Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Satisfaction With Leader

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ Accuracy .00 .000 .000 0 (1,151)
Team Performance .03 .301" .495" 72.64 (1,74)
Performance X LUSJ
Accuracy .00 .001 .003 .38 (1,150)
Total R2 .302" .003 .495"

 

Note. The higher the score is, the greater satisfaction with the leader. N=228. Between-
team variance=.35 (61%); within-team variance=.23 (39%). Total dfwithin-team=152;
total df between-team=75.

*p<.05. **p<.01.

The results changed slightly when the order of entry of team performance and
dyadic LUSJ accuracy was reversed. Team performance accounted for 30% (p<.01) of
the total and 49% (p<.01) of the between-team variance and LUSJ accuracy accounted for
less than 1% (n.s.) of the total and 1% (n.s.) of the within-team variance.

Hypothesis 6d proposed that team performance would moderate the effect of
dyadic LUSJ accuracy on task withdrawal. The results of the analyses are presented in
Table 23. Contrary to Hypothesis 6d, only team performance had a statistically
signiﬁcant negative effect on task withdrawal, accounting for 17% of the between-team

variance. Neither dyadic LUSJ accuracy nor the interaction of dyadic LUSJ accuracy and

team performance had a signiﬁcant effect.

114
Table 23. Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team

 

 

 

Performance on Task Withdrawal
A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ Accuracy -.01 .006 .011 1.63 (1,151)
Team Performance -.01 .076* .173** 15.50 (1,74)
Performance X LUSJ
Accuracy .01 .010 .018 2.75 (1,150)
Total RI .092" .029 .173"

 

N_ot_;e._ The higher the score is, the greater the task withdrawal. N=228. Between-team
variance=.18 (44%); within-team variance=.23 (56%). Total df within-team=152; total df
between-team=75.

*p<.05. **p<.01.

The results changed slightly when the order of entry of dyadic LUSJ accuracy and
team performance was reversed. Team performance accounted for 7% (p<.01) of the
total variance in task withdrawal, and 16% (p<.01) of the between-team variance. Dyadic
LUSJ accuracy did not account for any signiﬁcant amount of variance in task withdrawal
once the effect of team performance had been partialled.

Hypothesis 6e proposed that the effect of dyadic LUSJ accuracy on staff member
self-efﬁcacy would be moderated by team performance. The results are presented in
Table 24. Providing partial support for Hypothesis 6e, both team performance and its
interaction with dyadic LUSJ accuracy had statistically signiﬁcant effects on self-
efﬁcacy. Team performance accounted for 14% of the between-team variance, and its
interaction with dyadic LUSJ accuracy accounted for 3% of the within-team variance in
self-efﬁcacy. Dyadic LUSJ accuracy did not have a direct effect on self-efﬁcacy.

The interaction effect of dyadic LUSJ accuracy and team performance in

predicting staff member self-efﬁcacy is illustrated in Figure 16. The nature of this effect

Self-Efficacy

115

  
 

+ Team

 

Performance
-1 SD (14.6)
3.7 : ——I--Team
i Performance
l +1 SD (46)
l
3.5 45%—___---“ __. ._____‘,____ __,.V___ __,
'1 30 (5-1) +1 so (30.4)

Dyadic LUSJ Accuracy

Figure 16. Hypothesis 6e: Interaction Effect of Team
Performance and Dyadic LUSJ Accuracy on Self-Efficacy

116
is such that the relationship between dyadic LUSJ accuracy and self-efﬁcacy was slightly

negative for staff members on higher-performing teams, and positive for staff members
on lower-performing teams. Less accurately weighted staff members on higher-
performing teams reported the highest self-efﬁcacy, and less accurately weighted staff
members on lower-performing teams reported the lowest self-efﬁcacy.

Table 24. Repeated Measures Regression Analysis of Dyadic LUSJ Accuracy and Team
Performance on Self-Efﬁcacy

 

 

 

A in A in A in Incremental F
Variable b Total R2 Within R2 Between R2 (df,df)
LUSJ Accuracy .00 .005 .009 1.38 (1,151)
Team Performance .01 .064M .143” 12.33 (1,74)
Performance X LUSJ
Accuracy -.01 .015* .027* 4.23 (1,150)
Total R2 .084" .036 .143”

 

Note. The higher the score is, the greater the self-efﬁcacy. N=228. Between-team
variance=. 14 (45%); within-team variance=. 18 (55%). Total dfwithin-team=152; total df
between-team=75.
*p<.05. **p<.01.

The results changed when the order of entry of team performance and dyadic
LUSJ accuracy was reversed. Team performance accounted for 6% (p<.01) of the total
variance in self-efficacy, and 13% of the between-team variance. Dyadic LUSJ accuracy
accounted for an additional 1% (n.s.) of the total variance, and 2% (n.s.) of the within-
team variance after controlling for the effect of team performance.

Discussion

The results for Experiment 11 indicate several ﬁndings. The level of the leader’s

weighting of a staff member’s judgments interacted with team performance in affecting

117
only two of the ﬁve staff reactions investigated: willingness to return and task

withdrawal. The pattern of the moderating effect of team performance on the relationship
between LUSJ and task withdrawal indicated that being weighted more heavily by the
leader had less inﬂuence on task withdrawal than did team performance. In fact, contrary
to previous research on the effects of decision inﬂuence, in poorer-performing teams
being weighted more heavily was negatively related to willingness to return. In teams
that performed well, however, LUSJ was related to greater willingness to return and
lower task withdrawal. This may be because staff members in lower-performing teams
who were being weighted more heavily by the leader felt responsible for the team’s poor
performance and this inﬂuenced their reactions.

LUSJ had statistically signiﬁcant direct effects on all ﬁve outcomes studied when
the effects of team performance were not controlled for. Because LUSJ is signiﬁcantly
correlated with team performance (.53, p<.01), the effects on task withdrawal and
satisfaction with the leader were weakened when team performance was entered in to the
regression ﬁrst. The effects disappeared for willingness to return, desire to change for the
next task and self-efﬁcacy once team performance was controlled for. These results
highlight the dominant role of team performance in affecting participants’ reactions.

Staff members’ relative LUSJ only interacted with team performance in affecting
staff member willingness to return, although staff members’ relative LUSJ did have a
positive direct effect on self-efﬁcacy and a negative direct effect on task withdrawal.

The nature of the interaction is such that the relationship between relative LUSJ and
willingness to return was positive in higher-performing teams, and negative in lower-

performing teams. Receiving a high weight from the leader relative to the other team

118
members led to the greatest willingness to return in higher-performing teams, and the

lowest willingness to return in lower-performing teams. Interestingly, the correlation
between staff members’ self-report of the weight given their judgments by the leader and
their actual dyadic LUSJ was .58 (p<.01). The correlation between staff members’ self-
report LUSJ and their relative dyadic LUSJ was a slightly higher .71 (p<.01).

The accuracy of a staff member’s dyadic LUSJ weight did not directly affect any
of the outcomes studied. LUSJ accuracy did interact with team performance in affecting
willingness to return and self-efficacy. In both of these interactions the effect was carried
by the positive relationship between LUSJ accuracy and the reaction among members of
lower-performing teams. In other words, the accuracy of staff members’ LUSJ weights
was not as positive a factor in inﬂuencing willingness to return or self-efﬁcacy in higher-
performing teams as it was in lower-performing teams.

In terms of the ﬁve staff reactions that were investigated, for the staff reaction of
being willing to return in the future, team performance had a direct effect and interacted
with dyadic LUSJ level, relative dyadic LUSJ and dyadic LUSJ accuracy. Once team
performance was controlled for, the effects of dyadic LUSJ on willingness to return were
greatly reduced. Relative dyadic LUSJ and dyadic LUSJ accuracy did not inﬂuence
willingness to return directly. The ﬁndings suggest that high team performance leads to
the highest staff willingness to return. In higher-performing teams, relative LUSJ and
dyadic LUSJ have a strong positive relationship with willingness to return, but dyadic
LUSJ accuracy is slightly negatively related to willingness to return. In lower-
performing teams, dyadic LUSJ accuracy is positively related while dyadic LUSJ and

relative LUSJ are negatively related to willingness to return. In other words, staff

119
willingness to return was affected negatively by weighting level, but positively by

weighting accuracy in lower-performing teams. In higher-performing teams, staff
willingness to return was affected positively by weighting level and only slightly
negatively by weighting accuracy.

Only dyadic LUSJ level and team performance had an effect on staff desire to
change for the next task. After controlling for team performance, dyadic LUSJ level had
no effect. Dyadic LUSJ level, relative dyadic LUSJ and dyadic LUSJ accuracy did not
interact with team performance in predicting staff desire to change for the next task.
Team performance was the main inﬂuence on staff desire to change.

Team performance was also the most important factor inﬂuencing staff
satisfaction with the leader. The better the team performed, the more satisﬁed the staff
were with the leader. Dyadic LUSJ also had a small positive effect even after controlling
for the effects of team performance, but the bulk of the variance in staff satisfaction with
the leader was accounted for by team performance. Dyadic LUSJ, relative LUSJ and
dyadic LUSJ accuracy did not interact with team performance in predicting staff
satisfaction with the leader.

Staff member task withdrawal was affected by dyadic LUSJ, relative LUSJ, and
team performance. Withdrawal was lower when dyadic LUSJ or relative LUSJ was
higher, and when the team had higher performance. The interaction of dyadic LUSJ and
team performance also contributed to the prediction of task withdrawal. Higher team
performance led to generally lower task withdrawal regardless of the level of LUSJ. Staff
members given more weight by the leader reported lower task withdrawal than staff

members given lower weight regardless of team performance. Weighting accuracy was

120
unrelated to task withdrawal, and did not interact with team performance in predicting

staff reports of task withdrawal.

Staff member self-efﬁcacy was positively related to dyadic LUSJ, relative dyadic
LUSJ and team performance. Although dyadic LUSJ accuracy did not have a direct
effect on self-efﬁcacy, it interacted with team performance to contribute to the prediction
of staff self-efﬁcacy. On higher-performing teams, dyadic LUSJ accuracy was negatively
related to self-efﬁcacy. On lower-performing teams, dyadic LUSJ accuracy was
positively related to self—efﬁcacy.

The contribution of weighting level over team performance in predicting the staff
reactions is blurred by the fact that the error manipulation created a correlation of .53 (g <
.01) between LUSJ and team performance. To obtain a clearer indication of the role of
the leader’s staff utilization strategy in predicting staff reactions, effects coding of both
team performance and leader utilization of staff members was used to evaluate the effects
of the team performance and leader weighting strategy manipulations on the ﬁve staff
reactions after partialling team performance. The results of the regression analyses are
presented in Appendix E.

Statistically signiﬁcant main effects (2 < .05) of the weighting manipulation were
found for the staff reactions of desire for change on the next task and task withdrawal.
The nature of the main effects were such that higher-weighted staff members were more
desirous of change on the next task compared to the other team members (p < .01).
Higher-weighted staff members were less likely to withdraw from the task (2 < .01), and
lower-weighted staff members were more likely to withdraw from the task (9 < .05) than

the other team members.

121
The interaction of the team performance manipulation with the leader weighting

manipulation reached statistical signiﬁcance (p < .05) only for staff member willingness
to return. While the slope of the relationship between team performance and willingness
to return was positive for all team members, the slope was less positive for lower-
weighted staff members (1; < .05) than the other team members. Although the interaction
of the team performance manipulation with the leader weighting manipulation did not
reach statistical signiﬁcance for withdrawal from the task, the effect for lower-weighted
staff members was statistically signiﬁcant (p < .05). This effect was such that the
relationship between team performance and task withdrawal was negative for all team
members, and the slope was less negative for lower-weighted staff members than for
anyone else. No statistically signiﬁcant effects were found for the leader weighting
manipulation on self-efﬁcacy or satisfaction with the leader, although the team
performance by leader weighting interaction approached signiﬁcance for staff self-
efﬁcacy (2 < .08). These ﬁndings indicate that leader utilization of a staff member does
inﬂuence staff member willingness to return, desire for change on the next task, and
withdrawal after controlling for the effects of team performance.

The implications of these ﬁndings, and directions for future research, will be

discussed in the next chapter.

Chapter 5
IMPLICATIONS, LIMITATIONS, AND AREAS FOR FUTURE RESEARCH
Antecedents of LUSJ

Experiment I found that staff member judgment conﬁdence did not improve
leader decision making accuracy when provided alone or in combination with staff
cumulative past judgment accuracy. Staff members’ judgment conﬁdence was poorly
calibrated with judgment accuracy, yet these assessments still inﬂuenced leaders’ staff
weighting strategies. Additionally troublesome, leaders’ self-reports of staff member
judgment accuracy after the simulation had ended were positively biased by the level of
the staff member’s judgment conﬁdence. Clearly, the existence of this bias could have
negative implications for performance evaluations that rely on the leader’s perceptions of
staff accuracy. If these performance evaluations are later used as indications of staff past
accuracy or ability, the usefulness of past accuracy information could also be
compromised.

This study’s ﬁndings that staff judgment conﬁdence is at best weakly related to
judgment accuracy and that conﬁdence serves as a mechanism of inﬂuence between staff
member and leader replicates the ﬁndings of Sniezek and Buckley (1995). Sniezek and
Buckley also found that a staff member’s conﬁdence in his/her judgments was strongly
related to his/her ability to get the leader to choose his/her recommendation regardless of

its accuracy.

122

123

These ﬁndings are also consistent with the literature on leader-member exchange
theory that has found that leaders tend to identify “in-group” staff members to whom to
allow greater input and decision latitude on the basis of staff member ability (Graen &
Scandura, 1987; Scandura, Graen & Novak, 1986). When given information about the
performance of staff members, leaders tended to use this information in weighting future
staff member input.

Past ﬁndings that that leaders increase the weight given the judgments of higher
past performing staff members (Croner & Willis, 1967; Hollenbeck et al., 1995; Kelman,
1950) were also replicated. Providing leaders with staff member cumulative past
judgment accuracy information was extremely useful to leaders in this study. In
particular, leaders more readily discounted the judgments of staff members with low past
accuracy when this information was fed back than when it was not available, and LUSJ
accuracy improved as a result. As long as staff members’ accuracy does not change over
time, as was the case in this study, providing leaders staff member cumulative past
accuracy information can be a positive intervention in improving leader weighting
accuracy and encouraging the adoption of an appropriate differential weighting strategy
across staff members.

The ﬁndings of Brehmer and Hagafors (1986) were also replicated in this study.
Brehmer and Hagafors found that leaders tended to adopt a more equal weighting strategy
regardless of the distribution of appropriate staff weighting levels. In this study, where
differential staff weighting was the appropriate strategy, leaders not fed back staff
member cumulative past accuracy information tended to adopt a much more equal

weighting strategy than leaders fed back this information.

124
The conﬂicting ﬁndings in the literature regarding the antecedents of leader

utilization of staff judgments thus appear to be resolved once the inﬂuence of the
availability of staff member judgment conﬁdence and staff member cumulative past
performance information is taken into account. In general, leaders tend to utilize an equal
weighting strategy in the absence of information that helps them discriminate among their
staff members. Leaders tend to use staff member judgment conﬁdence in weighting staff
judgments when it is available, leading to greater discrimination among staff members
but not contributing to improved weighting accuracy. Leaders also tend to use staff
member cumulative past judgment accuracy in more appropriately weighting staff
judgments, and are able to make even better use of this information when staff
cumulative past performance is fed back to them than when it is not.

The ﬁnding that providing leaders with staff member cumulative past accuracy
information improves weighting accuracy while conﬁdence is unrelated to weighting
accuracy suggests that in the interest of more accurate decision making only staff member
cumulative past judgment accuracy information should be provided to leaders. The
availability of staff member judgment conﬁdence did not improve leaders’ LUSJ
accuracy or LUSJ variability alone or when paired with the feedback of staff member
cumulative past judgment accuracy information. This ﬁnding indicates that staff member
judgment conﬁdence information should not be provided to nor sought out by leaders of
hierarchical teams with distributed expertise.

The ﬁnding that judgment conﬁdence appears only to impair leader decision
making accuracy is likely due to the small or absent relationship between conﬁdence

level and accuracy. This is an effect that has consistently been demonstrated in the

125
literature (Sniezek & Buckley, 1993). Because conﬁdence has been consistently shown

to be a mechanism of inﬂuence, future research efforts might beneﬁt from a focus on
improving individuals’ calibration of their judgment conﬁdence. Should interventions be
discovered that improve the calibration of judgment conﬁdence, it is likely that
confidence could become beneﬁcial information for leaders interested in improving their
decision accuracy.

The ﬁndings for the effects of conﬁdence on leader decision making despite the
poor calibration of conﬁdence to judgment accuracy are not surprising when other
literature on conﬁdence is considered. Research on jury decision making, for example,
has consistently shown that jurors rely on eyewitness conﬁdence in making a verdict
(Moore & Gump, 1995; Penrod & Cutler, 1995), despite the consistently poor calibration
of conﬁdence with lineup identiﬁcation accuracy (Juslin, Olsson & Winman, 1996;
Sporer, Penrod, Read & Cutler, 1995).

Additional theorizing and research might prove fruitful in the area of identifying
other interventions that promote accurate differential weighting of staff on the part of
leaders. For example, although leaders’ LUSJ variability and LUSJ accuracy increased
over time in this study, interventions might be developed that further speed up the
learning curve for leaders learning how to best utilize their staff.

For the sake of experimental control, leaders in Experiment I did not have access
to any decision-related information other than staff member judgments. Future
investigations into the effects of providing leaders their own unique decision information

or information redundant with staff members’ information should also be investigated.

126

Future investigations into individual differences of leaders may also shed light on
leader utilization of staff information. Just as leaders higher in cognitive ability better
utilized their staff, different information processing styles might make leaders more or
less likely to utilize the information provided by aids such as staff past judgment
accuracy or staff judgment conﬁdence. Highly dogmatic individuals, for instance, tend to
make decisions quickly and tend not to utilize all available information. Individuals
higher in conscientiousness, on the other hand, might prove to be more adept at
identifying and utilizing the most beneﬁcial information.

Searches for environmental factors that inﬂuence the nature of leaders’ weighting
strategies are also encouraged. Differing reward structures, for example, might help or
inhibit leaders from differentially utilizing their staff. Leaders concerned about staff
turnover or decreased commitment to the resulting decision might be less likely to engage
in a differential weighting strategy even when it would lead to higher performance. If the
consequences for poor team decision accuracy were severe, on the other hand, leaders
might be more willing to differentially weight staff judgments for the sake of improved
accuracy.

Consequences of LUSJ

Experiment 11 found that team performance consistently had a strong effect on
staff member reactions to the team. If the team performed well, staff members tended to
have more favorable reactions to the team than if the team performed more poorly
regardless of how the leader utilized their judgments in making the team’s decisions.
This ﬁnding is consistent with past research on the effects of workgroup performance on

group member satisfaction (Zeffane, 1994) and turnover (Jackofsky (1984).

127
The strong team performance effect may be attributable to the fact that teams

competed for bonus money that was dependent solely on team performance. The fact that
the teams were assembled for three hours solely for the purposes of this study may have
served to intensify the effects of team performance. The fact that effects were still found
for different types of leader utilization of staff judgments despite this limitation indicates
that the hypothesized effects for LUSJ might be even greater in more natural settings.

The ﬁnding that the pattern of results for the effect of dyadic LUSJ on staff
reactions was opposite in higher relative to lower-perfonning teams for staff member
willingness to return and greatly reduced for task withdrawal was surprising. This
indicates that past research on the positive effects of decision inﬂuence and participation
(e. g., Bass, 1981; Drake & Mitchell, 1977; Locke & Schweiger, 1979) may generalize to
higher-performing, but not to lower-performing teams in which there is a single correct
decision. In a context with such heavy team reward implications, it is logical that team
performance tends to drive staff reactions rather than the level of the leader’s utilization
of their judgments. If team rewards are made less salient, or if individual consequences
such as ﬁnancial rewards or promotion are present, it is possible that staff members will
react more strongly to how their input is utilized by the leader.

Future research should continue to explore the moderating role of team
performance and the presence of a single correct answer in the relationship between
inﬂuence level and staff reactions. For example, if the task has a correct answer, team
performance is likely to play a much stronger role in affecting staff reactions than if no

accuracy assessment can be made.

128

Future theorizing and research should try to identify other causes of staff reactions
to LUSJ, in addition to investigating methods of managing more unfavorable staff
reactions to lower LUSJ weights. For the task used in this study, lower staff accuracy for
some team members was part of the design. The ability of staff members to perform well
was strongly determined by the design of the task and had much less to do with a staff
member’s ability. Yet the more negative reactions of the less-accurate and lower-
weighted staff indicate that leaders do indeed face a dilemma when differential staff
weighting is required for higher team performance. The fact that the team’s function is to
make accurate decisions and that team performance consistently accounted for greater
variance in staff reactions than any type of leader weighting of the staff member indicates
that team performance should be the top priority of leaders. Again, replications of these
ﬁndings in more natural settings are needed before any conclusions can be drawn.

Unexpectedly, dyadic LUSJ accuracy was a positive factor on lower-performing
teams and a slightly negative factor on higher-performing teams in inﬂuencing self-
effrcacy and staff willingness to return. The more accurately the leader weighted the
judgments of a staff member on lower-performing teams, the more willing the staff
member was to return and the greater the staff member’s self-efficacy.

Taken together, these ﬁndings highlight the existence of a real dilemma for
leaders of this type of decision making team. The strongest factor affecting staff
reactions in this study was team performance. Staff in higher-performing teams reported
more favorable reactions than did staff in lower-performing teams. When differential
utilization of staff judgments is required for high team performance, as was the case in

this task, some staff members must receive a higher or a lower LUSJ weight than other

129

staff members if the team is to perform well. In the higher-performing teams, however,
receiving a lower LUSJ weight led to less favorable reactions than did receiving a higher
weight. As discussed earlier, the withdrawal or turnover of even lesser-accurate team
members can threaten team viability, particularly when it is the nature of the task cues
that creates the differential staff validity.

Limitations

These two studies were not without limitations. The participants were college
undergraduates, did not know each other when they performed the simulation, and knew
that the study would last no more than 3 hours. Investigations of teams that work
together longer and for whom the consequences of performing well, or even of being
heavily weighted by the leader, have greater implications for team members need to be
performed. Additionally, the military-theme computer simulation was unfamiliar to
participants. Future research should attempt to replicate these ﬁndings with teams
interacting on more familiar tasks, and to teams in which the leader has unique decision-
related information.

Also, team viability is a complex phenomenon. Only ﬁve different staff reactions
were investigated in this study. Some of the staff reactions that were investigated were
also found to be moderately intercorrelated, and the reliability of the staff member desire
for change scale was lower than would be hoped. Future investigations should
incorporate a better measure of this construct, in addition to exploring other types of staff

reactions and team viability outcomes.

130
Research of this type is often criticized because of the artiﬁciality of the

laboratory setting. In fact, the laboratory is an ideal environment for testing theory
pertaining to groups and teams. As stated by Driskell and Salas (1992):

The primary criterion for designing an empirical setting to test theory is that it

provide a clear and robust test of that theory, not that it resemble the outside

world. However, when we attempt to apply the theory to a real-world setting,
then the realism of the setting in which the application or intervention is

conducted is critical. (p. 110)

The more the research setting contains only those variables relevant to the theory
being tested, and excludes extraneous variables (and thus the greater the artiﬁciality of
the research setting), the better the setting provides a clear test of the hypothesis (Webster
& Kervin, 1971; Mook, 1983). The artiﬁciality offered by laboratory settings is thus a
beneﬁt, not a liability when it comes to studying complex phenomena such as teams.

The fact that these results were found suggests that much remains to be learned
about managing this dilemma that faces leaders of hierarchical teams with distributed
expertise. Given the prevalence of these types of teams in organizations today, and the
relative paucity of knowledge that exists concerning the promotion of their effectiveness,

it is hoped that the results of these studies stimulate future research in this important area.

FOOTNOTES

1 For the sake of consistency, the term judgment will be used to refer to the interpretation

of cues by either the leader or a staff member, while the term decision will reﬂect the

 

ﬁnal decision ultimately registered by the team leader upon which team decision accuracy
or performance is based. Thus, a leader's initial judgment of the situation may differ from

the decision the leader ultimately registers for the team.

131

LIST OF REFERENCES

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting
interactions. Newbury Park: Sage.

American Psychological Association (1992). Ethical principles and codes of
conduct. American Psychologist. 47. 1597-1611.

American College Testing Program (1989). Preliminary technical manual for the
enhanced ACT assessment. Iowa City, IA: ACT Publications.

Babad, E. Y., Inbar, J ., & Rosenthal, R. (1982). Pygmalion, Galatea, and the
Golem: investigations of biased and unbiased teachers. Journal of Educatiorﬂ
Psychology. 74. 459-474.

Baker, CV. (1991). The effects of inter-positional uncertainty and workload on
team coordination skills and task performance. Doctoral dissertation, University of South
Florida, Department of Psychology, Tampa, FL.

Bales, R. F., & Cohen, S. P. (1979). SYMLOG: A system for the multiple level
observation of groups. New York: Free Press.

Bass, B. M. (1981). Stogdill's @dbook of leadership (Rev. ed.). New York: Free
Press.

Bimbaum, M. H., & Stegner, S. E. (1979). Source credibility in social judgment:
Bias, expertise, and the judge's point of view. Journ_al of Personality and Social
Psychology. 37. 48-74.

Bimbaum, M. H., Wong, K, & Wong, L. (1976). Combining information from
sources that vary in credibility. Memory and Cognition. 4. 330-336.

Blake, R. R., & Mouton, J. S. (1964). The managerial gg'd. Houston: Gulf
Publishing.

Bottger, P. C., & Yetton, P. W. (1988). An integration of process and decision

scheme explanations of group problem solving performance. Organizational Behavior
and Humjan Decision Performgce. 42. 234-249.

132

133

Brehmer, B. (1973). Single-cue probability learning as a function of the sign and
magnitude of the correlation between cue and criterion. Organizational Behavior and
Human Decision Processes. 9. 377-395.

Brehmer, B., & Hagafors, R. (1986). Use of experts in complex decision making:
A paradigm for the study of staff work. Organizational Beh_avior and Hum_an Decision
Processes. 38. 181-195.

Brunswick, E. (1943). Organismic achievement and environmental probability.
Psychological Review. 47. 69-78.

Brunswick, E. (1955). Representative design and probabilistic theory in a
functional psychology. Psychological Review. 50. 255-272.

Brunswick, E. (1956). Perception and representative design of expgriments.
Berkeley: University of California Press.

Buckley, T., & Sniezek, J. A. (1990). Conﬁdence as inﬂuence in a no feedback
choice task. Annual meeting of the Judgment and Decision Making Society, New
Orleans.

Cashman, J ., Dansereau, F. Jr., Graen, G., & Haga, W. J. (1976). Organizational
understructure and leadership: A longitudinal investigation of the managerial role-making
process. Organizational Behavior gd Humﬂ PerformanceLIS. 278-296.

Cohen, J. (1988). Statistical power aﬂlysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ: Lawrence Erlbaum Associates.

Cohen, J ., & Cohen, P. (1983). Applied multiple reggession/correlation analysis
for the beh_avioral sciences. Hillsdale, NJ: Lawrence Erlbaum Associates.

Croner, M. D., & Willis, R. H. (1967). Perceived differences in task competence

and asymmetry of dyadic inﬂuence. Journ_al of Abnormi and Social Psychology. 62.
705-708.

 

Danserau, F., Cashman, J ., & Graen, G. (1973). Instrurnentality theory and equity
theory as complementary approaches in predicting the relationship of leadership and
turnover among managers. Organizational Beh_avior and Hum Perform_ance. 10. 184-
200.

Danserau, F. Jr., Graen, G., & Haga, W. J. (1975). A vertical dyad linkage
approach to leadership within formal organizations: A longitudinal investigation of the
role making process. Organizational Behavio; and Human Perform_ance. 13. 46-78.

134

Davis, J. H. (1992). Some compelling intuitions about group consensus decisions,
theoretical and empirical research, and interpersonal aggregation phenomena: Select
examples, 1950-1990. Organizational Behavior and Hum_an Decision Processes. 52. 3-
38.

Deane, D. H., Hammond, K. R., & Summers, D. A. (1972). Acquisition and
application of knowledge in complex inference tasks. J ourn_al of Experimental

Psychology, 92, 20-26.

Deluga, R. J ., & Perry, J. T. (1991). The relationship of subordinate upward
inﬂuencing behavior, satisfaction and perceived superior effectiveness with leader-
member exchanges. Jom of Occupational Psychology, 64, 239-252.

Deutsch, M. & Gerard, H. B. (1955). A study of normative and informational
social inﬂuences upon individual judgment. Journal of Abnormal and Social Psychology.
51, 629-636.

Dewhirst, D., Metts, V., & Ladd, R. T. (1987). Exploring the delegation decision:
Managerial responses to multiple contingencies. Paper presented at the Academy of
Management Meetings, New Orleans.

Dillon, W. R., & Goldstein, M. (1984). Multivariate apalysis: Methods and
applications. New York: John Wiley & Sons.

Drake, B., & Mitchell, T. (1977). The effects of vertical and horizontal power on
individual motivation and satisfaction. Academy of Management Journal, 20, 573-591.

Driskell, J. E., & Salas, E. (1992). Can you study real teams in contrived settings?
The value of small group research to understanding teams. In R.W. Swezey and E. Salas,
(Eds), Teams: Their Training and Performance (pp. 101-124). Norwood, NJ: Ablex.

Duffy, L. (1993). Team decision-making biases: An inforrnation-processing
perspective. In G.A. Klein, J. Orasanu, R. Calderwood, and CE. Zsambok (Eds.),
Decision Making in Action: Models and Methods (pp. 346-359). Norwood, NJ: Ablex.

Dunning, D., Griffin, D. W., Milojkovic, J. D., & Ross, L. (1990). The
overconﬁdence effect in social prediction. Journg of Person_ality aﬂ Soclrl Psychology.
3, 568-581.

Dyer, J. L. (1984). Team research and team training: A state-of-the-art review. In
F.A. Muckler (Ed.), Human factors review: 1984 (pp. 285-323). Santa Monica, CA:
Human Factors Society.

F ischhoff, B., Slovic, P., & Lichtenstein, S. (1977). Knowing with certainty: The
appropriateness of extreme conﬁdence. Journ_al of Experimental Social Psychology:
Human Perception and Performance 3 552-564.

 

135

F oushee, H. C. (1984). Dyads and triads at 30,000 feet: Factors affecting group
process and aircrew performance. Amerigan Psychologist, 39, 885-893.

Gigerenzer, G. (1991). Probabilistic mental models: A Brunswikian theory of
conﬁdence. Psychological Review. 98. 506-528.

Gilliland, SW. (1992). The perceived faimess of selection systems: An
organizational justice perspective. Doctoral dissertation, Michigan State University,
Department of Psychology, East Lansing, MI.

Gottfredson, L.S., & Crouse, J. (1986). Validity versus utility of mental tests:
Example of the SAT. Jourpal of Vocational Behavior 29 363-378.

 

Graen, G., & Cashman, J. (1975). A role-making model of leadership in formal
organizations: A developmental approach. In J. Hunt & L.L. Larsen (Eds.), Leadership
Frontiers (pp. 143-165). Kent, OH: Kent State Univ. Press.

Graen, G., & Ginsburgh, S. (1977). Job resignation as a function of role
orientation and leader acceptance: A longitudinal investigation of organizational
assimilation. Organizational Beh_avior and Hum_an Performance. 19. 1-17.

Graen, G., Liden, R. C., & Hoel, W. (1982). Role of leadership in the employee
withdrawal process. Journal of Applied Psychology, 67, 868-872.

Graen, G., & Scandura, T. A. (1987). Toward a psychology of dyadic organizing.
In B. Staw & L.L. Cummings (Eds.), Research in organizational behavior, vol. 9 (pp.
175-208). Greenwich, CT: JAl Press.

Gully, S. M. (1994). Repeated measures regression analysis: A clariﬁcation with
illustrative examples. Paper presented at the 9th Annual Conference of the Society for
Industrial/Organizational Psychology, Nashville, TN.

Hackman, J. R. (1987). The design of work teams. In J .W. Lorsch (Ed.),
Handbook of organizational behavior (pp. 315-342). Englewood Cliffs, NJ: Prentice-Hall.

Hackman, J. R., & Oldham, G. R. (1980). Work redesigg New York: Addison-
Wesley.

Heller, F. A. (1992). Decision-making and the utilization of competence. In F .A.
Heller (Ed.), Decision-making and leadership. New York: Cambridge University Press.

Heller, F. A., & Yukl, G. (1969). Participation, managerial decision making, and
situational variables. Organizational Beﬂwior and Human Performance 4 227-241.

 

136

Hollenbeck, J. R., Ilgen, D. R. & Sego, D. J. (1994). Repeated measures
regression and mediational tests: Enhancing the power of leadership research. Leadership

Quarterly, 5, 3-23.

Hollenbeck, J. R., Ilgen, D. R., Sego, D. J., Hedlund, J., Major, D. A. & Phillips,
J. (1995). Multilevel theory of team decision making: Decision performance in teams
incorporating distributed expertise. Journal of Applied Paychology, 80, 292-316.

Hollenbeck, J. R., Sego, D. J., Ilgen, D. R., & Major, D. A., Hedlund, J., &
Phillips, J. (1995). Team decision making accuracy under difﬁcult conditions: Construct
validation of potential manipulations using the TIDE2 simulation. In M. T. Brannick, E.
Salas, & C. Prince (Eds), Team performance assessment and measurement: Theory.
research and applications. Hillsdale, NJ: Erlbaum.

 

Hunter, J .E. (1986). Cognitive ability, cognitive aptitudes, job knowledge and job
performance. Journal of Vocational Behavior. 29. 340-362.

Ilgen, D. R. (1986). Laboratory research: A question of when, not if. In E. A.

Locke (Ed.), Generalizing from laboratog to ﬁeld settings (pp. 257-267). Lexington,
MA: Lexington Books.

Jackofsky, E. F. (1984). Turnover and job performance: An integrated process
model. Academy of Mﬂagement Review. 9. 74-83.

Jensen, AR. (1986). g: Artifact or reality? Journal of Vocational Behavior 29
301-331.

 

Juslin, P., Olsson, N., & Winman, A. (1996). Calibration and diagnosticity of
conﬁdence in eyewitness identiﬁcation: Comments on what can be inferred from the low
conﬁdence-accuracy correlation. Journal of Experimental Psychology: Learning,
Memory. & Cogpition, 22, 1304-1316.

Kahneman, D., Slovic, P., & Tversky, A. (1982). Judgment under uncertainty:
Heuristics an; biases. Cambridge, MA: Cambridge University Press.

Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive abilities: An
integrative/aptitude-treatment interaction approach to skill acquisition [Monograph].

Journal of Applied Psychology, 74, 657-690.

Katerberg, R., & Hom, P. W. (1981). Effects of within-group and between-groups
variation in leadership. Jourpal of Applied Psychology. 66. 218-223.

Kelman, H. C. (1950). Effects of success and failure on "suggestibility" in the
autokinetic situation. Journal of Abnormal and Social PsychologY. 45. 267-285.

137

Kim, W. C., & Mauborgne, R. A. (1993). Procedural justice, attitudes, and
subsidiary top management compliance. Apademy of Management J ourn_a_lé 6. 502-526.

Korsgaard, M. A., Schweiger, D. M., & Sapienza, H. J. (1995). Building
commitment, attachment, and trust in strategic decision-making teams: The role of
procedural justice. Academy of Mgagement Jomal. 38. 60-84.

Libby, R., Trotrnan, K. T., & Zirnmer, I. (1987). Member variation, recognition of
expertise, and group performance. Journal of Applied Psychology. 72. 81-87.

Lichtenstein, S., & F ischhoff, B. (1977). Do those who know more also know
more about what they know? Organizational Behavior and Human Performance, 20, 159-
183.

Lichtenstein, S., F ischhoff, B., & Phillips, L. D. (1982). Calibration of
probabilities: The state of the art to 1980. In D. Kahneman, P. Slovic, & A. Tversky
(Eds), Judgment under uncertainty: Heuristics and biaiesa Cambridge, MA: Cambridge
University Press.

Locke, E. A., Frederick, E., Lee, C., & Bobko, P. (1984). Effect of self-efﬁcacy,
goals, and task strategies on task performance. Journal of Applied Psychology. 69. 241-
251.

Locke, E. A., & Schweiger, D. M. (1979). participation in decision making: One
more look. In B. M. Shaw (Ed.), Research in organizational behavior vol. 1 (pp. 265-
339). Greenwich, CT: JAI Press.

 

March, J. G. (1956). Inﬂuence measurement in experimental and semi-
experimental groups. Sociometry. 19. 260-271.

Mausner, B. (1954a). The effect of prior reinforcement on the interaction of
observer pairs. Journal of Abnormal Social Psychology. 49. 65-68.

Mausner, B. (1954b). The effect of one partner's success or failure in a relevant

task on the interaction of observer pairs. Journal of Abnormal and Social Psychology. 49.
557-560.

McGrath, J. E. (1976). Stress and behavior in organizations. In M. D. Dunnette

(Ed.), Handboon inclasﬁd and organizational psychology (pp. 1310-1367). Chicago:
Rand McNally.

Michaelsen, L. K., Watson, W. E., and Black, R. H. (1989). A realistic test of
individual versus group consensus decision making. Journal of Applied Psychology.7$
834-839.

138

Miller, K. 1., & Monge, P. R. (1986). Participation, satisfaction, and productivity:
A meta-analytic review. Academy of Management Journal, 29, 727-753.

Mook, D. G. (1983). In defense of external invalidity. American Psychologist. 38.
379-3 87.

Moore, P. J ., & Gump, B. B. (1995). Information integration in juror decision
making. Journal of Applied Social Psychology, 25, 2158-2179.

Oz, 8., & Eden, D. (1994). Restraining the golem: Boosting performance by
changing the interpretation of low scores. Journal of Applied Psychology, 79, 744-754.

Paese, P. W., & Sniezek, J. A. (1991). Inﬂuences on the appropriateness of
conﬁdence in judgment: Practice, effort, information, and decision-making.
Organizational Behavior and Human Deci§ion Processes. 48. 100-130.

Penrod, S., & Cutler, B. (1995). Witness conﬁdence and witness accuracy:

Assessing their forensic relation. [Special issue: Witness memory and law]. Psychology,
Public Policy. & Law. 1. 817-845.

Potter, E. H., & Fiedler, F. E. (1981). The utilization of staff member intelligence
and experience under high and low stress. Academy of Maaagement Journal. 24. 361-
376.

Rosenthal, R. (1985). From unconscious experimenter bias to teacher expectancy
effects. In J.B. Dusek, V.C. Hall, & W.J. Meyer (Eds), Teacher expectationa (pp. 37-65).
Hillsdale, NJ: Erlbaum.

Rosenthal, R. (1991). Teacher expectancy effects: A brief update 25 years after
the Pygmalion experiment. Journal of Research in Education 1 3-12.

 

Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom. New York:
Holt, Rinehart, and Winston.

Salas, E., Dickinson, T. L., Converse, S. A., & Tannenbaum, S. l. (1992). Toward
an understanding of team performance and training. In R. W. Swezey & E. Salas (Eds),
Teams: Their training and performance (pp. 3-29). Norwood, New Jersey: Ablex
Publishing Corporation.

Scandura, T. A., Graen, G. B. & Novak, M. A. (1986). When managers decide not
to decide autocratically: An investigation of leader-member exchange and decision

inﬂuence. Journal of Applied Psychology, 71, 579-584.

139

Seashore, S.E., Lawler, E.E., Mirvis, P., and Cammann, C. (eds) (1982).
Observing and measuring organizational change: A guide to ﬁeld practice. New York:
Wiley.

Simon, H. A. (1978). Rationality as process and as product of thought. American
Economic Review. 68. 1-16.

Slovic, P., F ischhoff, B., & Lichtenstein, S. (1977). Behavior decision theory.
Annual Review of Psychology, 28, 1-39.

Slovic, P. & Lichtenstein, S. (1971). Comparison of Bayesian and regression
approaches to the study of information processing in judgment. Organizational Behavior
and Hum_an Performance 6 649-744.

 

Sniezek, J. A. (1992). Groups under uncertainty: An examination of conﬁdence in
group decision making. Special issue of Organizational Beh_avior and Human Decision
Processes. 52. 124-155.

Sniezek, J. A., & Buckley, T. (1995). Cueing and cognitive conﬂict in judge-
advisor decision making. Organizational Behavior and Humg Decision Processes. 62.
159-174.

Sniezek, J. A., & Buckley, T. (1993). Becoming more or less uncertain. In N. J.
Castellan, Jr. (Ed), Individual and group decision making: Current issues. (pp. 203-218).
Hillsdale, NJ: Lawrence Erlbaum Associates.

Sniezek, J. A., & Naylor, J. C. (1978). Cue measurement scale and functional
hypothesis testing in one probability learning. Organizational Behavior and Human
Decision Processes. 22. 366-374.

Sniezek, J. A., & Paese, P. W., & Switzer, F. S. (1990). The effect of choosing on
conﬁdence in choice. Organizational Behavior and Human Decision Processes, 46, 264-
282.

Sniezek, J. A., & Reeves, A. P. (1986). Feature cues in probability learning: Data

base information and judgment. Organizational Beh_avior and Hum_an Decision Processes.
3_7_, 297-315.

Sporer, S. L., Penrod, 8., Read, D., & Cutler, B. (1995). Choosing, conﬁdence,
and accuracy: A meta-analysis of the conﬁdence-accuracy relation in eyewitness

identiﬁcation studies. Psychological Bulletin. 118. 315-327.

Steiner, I. D. (1972). Group process and productivig. New York: Academic
Press.

140

Sundstrom, E., DeMeuse, K. P., & Futrell, D. (1990). Work teams: Applications
and effectiveness. American Psychologist, 45, 120-133.

Thompson, W. C. (1993). Research on jury decision making: The state of the

science. In N. J. Castellan, Jr. (Ed), Individual and grog) decision making: Current
issues, (pp. 203-218). Hillsdale, NJ: Lawrence Erlbaum Associates.

Tindale, R. S. (1993). Decision errors made by individuals and groups. In N. J.
Castellan, Jr. (Ed), Individual and group decision making: Current issues, (pp. 203-218).
Hillsdale, NJ: Lawrence Erlbaum Associates.

Tucker, L. R. (1964). A suggested alternative formulation in the developments by
Hursch, Hammond, and Hursch, and by Hammond, Hursch, and Todd. Psychological
Review 71 528-530.

 

Vallone, R. P., Griffin, D. W., Lin, 8., & Ross, L. (1990). Overconﬁdent
prediction of future actions and outcomes by self and others. Journal of Personality a_n_d
Social Psycholgg. 58. 582-592.

Vecchio, R. P. (1982). A further test of leadership effects due to between-group
and within-group variation. Journal of Applied Psycholggy, 67. 200-208.

Vroom, V. A. (1964). Work and motivation. New York: Wiley.

Vroom, V. H., & Yetton, P. W. (1973). Leadership and decision making.
Pittsburgh: University of Pittsburgh Press.

Wagner, J. A. III (1994). Participation’s effect on performance and satisfaction: A
reconsideration of research evidence. Academy of Management Review. 19. 312-330.

Wagner, J. A. III, & Gooding, R. Z. (1987a). Effects of societal trends on
participation research. Administrative Science Quarterly, 32, 241-262.

Wagner, J. A. III, & Gooding, R. Z. (1987b). Shared inﬂuence and organizational
behavior: A meta-analysis of situational variables expected to moderate participation-
outcome relationships. Academy of Management Journal. 30. 524-541.

Wakabayshi, M., Minami, T., Sano, K., Graen, G., & Novak, M. (1980).
Management progress: Japanese style. International Journal of Cultural Relations. 4. 391-
420.

Webster, M., & Kervin, J. B. (1971). Artiﬁciality in experimental sociology.
Canadian Review of Sociology and Anthropology, 8, 263-272.

141

Wood, M. T. (1973). Power relationships and group decision making in
organizations. Psychological Bulletin. 79. 280-293.

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

Zeffane, R. M. (1994). Correlates of job satisfaction and their implications for
work redesign: A focus on the Australian telecommunications industry. Public Personnel

Management, 23, 61-75.

APPENDICES

APPENDIX A

TESTS OF LUSJ INDICES

APPENDIX A

TESTS OF LUSJ INDICES
A monte carlo simulation involving three staff members' judgments and a leader's
decisions over 100 decisions was performed under six different conﬁgurations of leader
staff weighting strategies. In all scenarios the intercorrelations of staff judgments were
kept low. The four LUSJ constructs (Dyadic LUSJ, Accuracy of Dyadic LUSJ, Relative
Dyadic LUSJ, and Dyadic LUSJ Variability) are examined below in each of these six
scenarios. Dyadic LUSJ Accuracy was calculated assuming a correct judgment B weight
of .2 for staff member A, .4 for staff member B and .6 for staff member C.
Scenario 1:
A's judgments are given high weight by the leader
B's judgments are given moderate weight by the leader
C's judgments are given low weight by the leader
Correlation Matrix:
A B C
B .01
C -.07 -.01
Leader Decision .70 .48 .21

LUSJ Indices:
Staff Dyadic Dyadic LUSJ Relative Dyadic LUSJ
Member LUSJ Accuracy Dyadic LUSJ Variability

A .68 .48 .22 .14

B .46 .06 .00 .14

C .25 -.35 -.21 .14
Scenario 2:

A's judgments are given low weight by the leader

B's judgments are given high weight by the leader

C's judgments are given high weight by the leader
Correlation Matrix:

 

A B C
B -.01
C -.02 .00
Leader Decision .21 .63 .58
LUSJ Indices:
Staff Dyadic Dyadic LUSJ Relative Dyadic LUSJ
Member LUSJ Accuracy Dyadic LUSJ Variability
A .20 .00 -.27 .18
B .63 .23 .16 .18
C .59 -.01 .12 .18

142

143

Scenario 3:
A's judgments are given high weight by the leader
B's judgments are given low weight by the leader
C's judgments are given low weight by the leader

 

Correlation Matrix:
A B C
B -.02
C -.02 .03

Leader Decision .72 .17 .21

 

LUSJ Indices:
Staff Dyadic Dyadic LUSJ Relative Dyadic LUSJ
Member LUSJ Accuracy Dyadic LUSJ Variability
A .76 .56 .37 .25
B .17 -.23 -.22 .25
C .23 -.37 -.16 .25
Scenario 4:

A's judgments are given moderate weight by the leader
B's judgments are given moderate weight by the leader
C's judgments are given moderate weight by the leader

Correlation Matrix:

A B C
B -.01
C -.04 -.01

Leader Decision .33 .36 .31

 

LUSJ Indices:
Staff Dyadic Dyadic LUSJ Relative Dyadic LUSJ
Member LUSJ Accuracy Dyadic LUSJ Variability
A .34 .14 .00 .01
B .35 -.05 .01 .01

C .32 -.28 -.02 .01

144

Scenario 5:
A's judgments are given high weight by the leader
B's judgments are given high weight by the leader
C's judgments are given high weight by the leader

Correlation Matrix:

 

A B C
B -.01
C -.05 -.04
Leader Decision .48 .53 .48
LUSJ Indices:
Staff Dyadic Dyadic LUSJ Relative Dyadic LUSJ
Member LUSJ Accuracy Dyadic LUSJ Variability
A .50 .30 -.01 .01
B .53 .13 .02 .01
C .51 -.09 .00 .01
Scenario 6:

A's judgments are given low weight by the leader
B's judgments are given low weight by the leader
C's judgments are given low weight by the leader

Correlation Matrix:

A B C
B .06
C -.04 -.04

Leader Decision .12 .14 .15

 

LUSJ Indices:
Staff Dyadic Dyadic LUSJ Relative Dyadic LUSJ
Member LUSJ Accuracy Dyadic LUSJ Variability
A .12 -.08 -.O2 .02
B .15 -.25 .01 .02

C .16 -.44 .02 .02

APPENDIX B

MATERIALS USED IN BOTH EXPERIMENTS I AND II:
CONSENT FORM, DEMOGRAPHIC QUESTIONNAIRE,
GENERAL TRAINING MANUAL AND TRAINING SCRIPT

APPENDIX B

EXPERIMENTS I AND 11
CONSENT FORM

This set of two studies was designed to investigate team decision making effectiveness. If
you choose to participate in this study, you will be asked to learn a computer-simulated
aircraft- identiﬁcation task, operate the simulation task as part of a four-person team, and
complete a series of questionnaire items. Also, by signing below you will indicate that
you choose to participate in this study and that you authorize the researchers to have
access to your SAT/ACT scores.

Your participation in the simulation should take about three hours. You will receive
course credit in exchange for your participation in this study. Your participation in this
research is completely voluntary. You are free to decline to answer any questions or to
terminate your participation at any time. Your participation in this study will be totally
conﬁdential. Your data will be included in a summary report along with the data from
others. The report will not include any information that will allow anyone to identify any
of your individual responses. If you have any questions or concerns regarding this study,
you may contact Jean Phillips in the Management Department at 353-7116.

Participant Statement

1 agree to participate in the Team Decision Making Study. By signing below I authorize
the researchers to use my SAT/ACT scores, and I recognize that I must provide my
student number (PID) to do this. It is my understanding that these materials will be
strictly conﬁdential and will not be seen by anyone other than the research team. I
consent to having these materials used for research purposes. I also understand that I will
learn to operate a computer simulation and perform the simulation with other individuals,
and that I will complete a series of questionnaires before and aﬁer the simulation.

I understand that the top teams in each condition will receive cash prizes on the following
basis: First place $20/ person; Second place $15/ person; Third place 310/ person.

I understand that my participation is voluntary, that I may discontinue participation at any
time without penalty, that all of my individual responses will be kept strictly conﬁdential,
and that I will not be identiﬁed in any report of this study.

 

 

 

 

Printed Name Date
Signature

Student Number (PID)

Course/T A Section #

 

 

145

O‘M-ﬁWN-ﬂ

\l

146

EXPERIMENTS 1 AND 11
DEMOGRAPHIC QUESTIONNAIRE

PARTICIPANT #

. DATE:

.NAME:

. STUDENT NUMBER:

. MANAGEMENT 302 TA:

.SEX (CIRCLE ONE): MALE FEMALE

.YEAR (CIRCLE ONE): FRESHMAN SOPH JUNIOR SENIOR OTHER

 

 

 

. AGE:
. Have you ever participated in a study in this lab before? Y N

. Approximately how often do you use a personal computer?

(Circle a whole number using the scale below)

<--l 2 3 4 5-->
Monthly Weekly Daily
or Less

 

10. Approximately how often do you play video games?

(Circle a whole number using the scale below)

 

<-- l 2 3 4 5-->
Monthly Weekly Daily
or Less

11. Approximately how well do you know each of the people you will be working with today (don't rate

 

 

 

 

yourself):
CARRIER
<--1 2 3 4 5-->
Not at Casual Very Good
All Acquaintance Friends
CAD
<--1 2 3 4 5-->
Not at Casual Very Good
All Acquaintance Friends
AWAC
<--1 2 3 4 5-->
Not at Casual Very Good
All Acquaintance Friends
CRUISER
<--l 2 3 4 5-->
Not at Casual Very Good

All Acquaintance Friends

147

EXPERIMENTS I AND 11
GENERAL TRAINING MANUAL

INTRODUCTION

The year is 1996 and you are a part of a US. naval Carrier group's command and control
team stationed in the Middle East. A regional conﬂict between two nations in this area
has recently broken out, and your mission is to protect seagoing commercial traffic in the
area ﬁ'om accidental or intentional attacks. As history indicates, this is a highly sensitive
task. For example, in 1987, failure by a command and control team to quickly and
accurately identify a plane as threatening, allowed an Iraqi jet to accidentally ﬁre two
Exocet missiles into the Frigate U.S.S. Stark, killing 37 American servicemen and
crippling the vessel. One year later, a command and control team error resulted in the
USS. Cruiser Vincennes accidentally shooting down an Iranian passenger plane killing
290 innocent civilians. Any repeat of mistakes of this kind will probably lead to a
withdrawal of American forces from the area. Such a withdrawal would have disastrous
economic and political ramiﬁcations that would spread well beyond this region.

THE TASK FORCE

Your naval Carrier group is an array of ships, planes, and other supporting units with the
purpose of protecting approximately 196,000 square miles of ocean. In order to control
such a large area, radar surveillance is necessary so that the Carrier group is not surprised
by the enemy. Four units provide the bulk of radar coverage over a Carrier group. These
units are linked together by an electronic data network so that they can supply bits and
pieces of critical information concerning possible enemy planes to each other. These four
units are sometimes called a command and control team.

Essentially, these four units communicate and coordinate what they see on their
individual radars, so that the team commander, located on the aircraft Carrier, ends up
seeing an accurate overall 'big-picture'. This accurate picture is necessary so that the
commander can make appropriate decisions concerning possible enemy targets (aircraﬁ
that are being tracked are called targets).

The ﬁrst station of the command and control team consists of an Air Force AWAC S
(Airborne Warning and Control System) reconnaissance plane which ﬂies overhead using
radar to identify targets far off in the distance. The second station is a land-based Marine
CAD (Coastal Air Defense) unit which supplies radar coverage from a beach. The third
station is a fast, highly maneuverable navy ship called a Cruiser (which supplies radar
coverage from the sea). Finally, the Carrier provides leadership and integrates the
information gathered by the other three stations into the team's ﬁnal decision for any
offensive or defensive tactical actions.

148

TEAM MISSION - Monitoring Air Space

The team, of which you are a part, will role play the Commanding Officers of the stations
which compose the Carrier group’s command and control team. Your mission is to
monitor the airspace surrounding the Carrier group, making sure that neutral ships are not
attacked. In performing this role, you must make certain that you do not allow loss of life
resulting from accidental or intentional attacks on ships in the task force. At the same
time, it is also of paramount importance that you do not inadvertently shoot down
ﬁiendly military aircraft or any civilian aircraft. Many passenger ﬂights move in and out
of the region, and friendly military aircraﬁ from nations not involved in the conﬂict also
patrol the area. In 1994 two US. F-15 ﬁghters shot down two ﬁiendly helicopters in
Northern Iraq killing 26 people. Another occurrence such as this, or of the USS. Stark or
U.S.S. Vincennes variety, will diminish public support for the current mission, and in
turn, jeopardize peace in this region.

OVERVIEW OF ROLES

There are four roles in this simulation, one for each member of a four person team. The
leader is the Commanding Oﬁicer (CO) of the Aircraft Carrier. The other team members
include the C0 of an AWACS air reconnaissance plane, the CO of a Cruiser, and the CO
of a CAD unit (Coastal Air Defense unit located on a beach). The team’s task is to decide
what response the Carrier group should take toward incoming air targets. The COs of the
AWACS, Cruiser, and CAD will make recommendations to the Carrier CO, who will
then make the ﬁnal decision for the team. Team members base their decisions on data
they collect by measuring characteristics of targets that enter the Carrier group’s area of
responsibility. These measures are obtained from sophisticated radar and other electronic
devices. Each staff member has something that is unique to contribute to the
decision. There are seven possible choices to make for each incoming target. These
responses are graded in terms of their aggressiveness and there is one correct response for
each aircraﬁ. Each of these is described on the next page, moving from least to most
aggressive.

1 49
SEVEN POSSIBLE DECISIONS

1) IGNORE: This means that no further attention should be devoted to the target and
instead focus should be directed on other possible targets in the area. Never ignore a
target that might possibly attack. This would most assuredly lead to loss of lives.

2) REVIEW: This means attention can be shifted away from this target momentarily.
After a short period of time this target should be returned to in order to update its status.
A large number of targets can be in review status, however, reviewing targets decreases
the amount of team resources that can be spent addressing other targets.

3) MONITOR: This means that the target should be continuously tracked. The systems
that do this tracking are capable of monitoring fewer targets than can be reviewed, and
thus monitoring diminishes overall patrol capacity.

4) WARN: This means that a message is sent to the target ordering it to turn away.
Warning targets that should be ignored detracts from the importance of legitimate
warnings. Warning targets that intend to attack is also bad, since the warning makes it
easier for the attacker to locate the ship.

5) READY: This means to get into a defensive posture and to set defensive weapons on
automatic. A ship in a readied position is rarely vulnerable to attack. This stance should
not be taken to non-threatening targets since weapons set to automatic and ﬁre mistakenly
at innocent targets that ﬁre too closer to the Carrier group. A ship in this position cannot
readily take offensive action toward other targets.

6) LOCK-ON: This synchronizes radar and attack weapons so that the weapons ﬁx
themselves on the target. A ship at Lock-On position can take offensive action at a
moment’s notice. The capacity to track other targets is severely constrained once there is
Lock-On to a single target, however. Thus, this should be reserved for targets that are
almost certain to be threatening.

7) DEFEND: This is “weapons away” and means to attack the target with missiles or
depth charges. A defend decision cannot be aborted once initiated and thus must only be
used when enemy attack is imminent.

150

EXPERIMENTS I AND II
INTERACTIVE TRAINING SCRIPT

Explain to the participants, "The interactive training involves the ﬁrst three targets.
The ﬁrst three targets are extremely long to allow for questions and to ensure that
we cover everything you need to know. You will then see sixty targets that will count
toward the bonus money. Each of these sixty targets will be sixty seconds long. The
ﬁrst three targets are just practice to allow you to get used to the game - they will
not count at all toward the bonus money. I will be available during all three
practice targets to answer questions, but once the real targets start I can not answer
any questions."

"Please follow along closely and do not get ahead of me and the rest of your team.
This will ensure that we cover everything."

The researcher will then call up the ﬁrst target and begin the hands-on training. (For
Experiment 11, explain that the team leader, the Carrier, is in another room). Make sure
that everyone is on the blue icon screen.

1. Point out and quickly explain the icons, game #, time clock, and menu bar. Mention
that the 60 targets that count toward the bonus money will each last 60 seconds.

If called for by the condition, explain the conﬁdence and past accuracy portions of
the screen. Explain that the past accuracy bars on the leader’s screen, and the
utilization bars on the staff members’ screens will not be accurate until the
seventh target because the statistic takes that long to calibrate.

2. Explain that the staff members will now learn how to measure attributes. Say, “You
Measure attributes by using the mouse to open the Measure menu. To measure,
use the mouse and point to the desired cue you want to measure and click on it.
The gray box on the lower leﬁ portion of the screen showing the attribute value
will disappear in 3 seconds or when enter is pressed again. Measure another
attribute.”

3. Point out that each staff member can only measure the attributes on which they have
been trained and that are within their area of expertise. Remind participants that
each staff member sees three unique pieces of information, and that each has
something unique to contribute to the team’s decision. Ask participants to click
on Measure again and then click on the Measure Summary. This is a summary
box that will display all of the attributes measured by that staff member on the
current target. Explain that it will stay open for about 3 seconds, then disappear.

151

4. Ask participants to hit F2. Explain that this is an even faster way to open the Measure
Summary box. Comment that they must have measured an attribute before it
appears in this box, but once they’ve measured it it stays in the summary box until
the next target comes up.

5. When each staff member has measured all 3 attributes and is ready to send a judgment
to the leader, explain that at 30 seconds left, the clock will start beeping,
indicating that judgments must be sent soon by the outlying stations (CAD,
AWAC and Cruiser). This judgment must reach the Carrier with enough time
to make a team decision.

Have everyone click on Judgment. Brieﬂy explain the seven decision options,
and tell participants to click on their judgment choice. Explain that if they don't do
anything else, the judgment will register in about 3 seconds. If they made a
mistake and want to resend their judgment, they can click on Cancel and resend
their judgment. If the judgment is the one they want, they can also hit OK and
their judgment will immediately be sent to the leader.

6. Explain that the leader in the other room will receive their judgments and will be
shown by another researcher how to register the team decision. Ask if there are
any questions.

7. When the feedback screen comes up, explain the previous decision's feedback
information. Explain that timing the simulation, this information will stay on the
screen for about 5 seconds, then the next target will automatically come up. Stress
that there is nothing they can do to make the next target come up any faster, and to
avoid hitting anything on the keyboard to prevent them from being locked out of
the next target. Explain that if anything ever seems wrong, they should get a
researcher immediately.

8. When the next target comes up, explain that you will stay in the room for this and the
following target to answer any questions participants might have. Point out that
the game number has changed (it is the second target), and that they will see a
total of 60 targets during the experimental session. Remind them that they will
also be doing a second decision making task near the end of the session, and that
they will be given more input as to how the second task will be completed (e.g.,
keeping things the same, changing the leader, changing the team, or doing it as
individuals).

9. Near the end of the second target, tell participants that after they have registered their
judgment, they should not open any new menus on their computer screen to
prevent being locked out when the leader makes the team decision. Explain that
open windows can prevent the reset signal from resetting the computer.

152

10. At the end of the third practice target, save the practice data and load the experimental
simulation. Do not start the experimental session until you tell participants:

"At this time, I want to stress two very important points:

1. With 20 seconds it is very important that you make judgments relatively
quickly to leave time for the Carrier to receive them and to make the team
decision.

2. When time is running down and you have already made a judgment, clear
your menu bar (hit ESCAPE in the top left corner) to ensure that you
proceed to feedback and to the next target. If you notice negative time on the
clock or if you notice that you are still in feedback when the other stations
have moved on to the next target, contact a researcher immediately.

3. When you are in a text message box, the clock will appear frozen, but it is
actually counting down. Be aware!!

Let them know that they will be monitored through an intercom in case someone gets
locked out. Remind them not to talk during the simulation, and tell them that everything
they type is recorded by the computer.

APPENDIX C

EXPERIMENT 1 MATERIALS: LEADER’S POSITION-SPECIFIC
TRAINING MANUAL, POST-SESSION QUESTIONNAIRE
AND DEBRIEFING FORM

APPENDIX C

EXPERIMENT I
LEADER’S POSITION-SPECIFIC TRAINING MANUAL

The CARRIER is a large ship that is the core of the naval command group. As the
Commanding Ofﬁcer (C0) of the Carrier, you are responsible for making the naval
group’s decisions of how to respond to incoming aircraft. Your staff members, the COs of
the CAD, AWAC, and CRUISER, are responsible for summarizing information about the
incoming aircraft and sending you a judgment relevant to their area of expertise. The
CAD’s responsibility is to summarize information relevant to the target’s location, the
AWAC’s is to summarize information about the target’s movement, and the CRUISER
summarizes information relevant to the classiﬁcation of the target (what type of plane it
is). Each staff member therefore has something unique to contribute to the decision --
their roles do not overlap.

CONFIDENCE (if called for by the condition)

In addition to sending you their judgment for the target, you will also receive an
indication of how conﬁdent each staff member is (on a scale of 1%, reﬂecting a guess, to
100%, reﬂecting extreme conﬁdence) in the accuracy of their judgment. For example, in
addition to recommending an IGNORE decision, the CRUISER may tell you that s/he is
80% conﬁdent that this judgment is correct, based on the information s/he has acquired.
This will become more clear during training.

PAST JUDGMENT ACCURACY (if called for by the condition)

During each active target, there will be a red bar on your screen near each of the icons
reﬂecting your staff members. This bar reﬂects how accurate that staff member’s
judgments have been in the past (in terms of the strength of the relationship between their
judgments and the correct decision). These bars will be seen by all team members. The
longer the red bar, the more accurate the person’s judgments have been.

COMBINING THE THREE JUDGMENTS INTO A TEAM DECISION

As the C0 of the Carrier you are responsible for combining the CO’s judgments of the
target’s standing into the team decision. Each staff member has information that is unique
to contribute to the decision. An incoming target could therefore look different to each of
the three staff members in terms of its threat. It is up to the leader to combine this
information into the team decision. ONLY THE DECISION REGISTERED BY THE
LEADER is considered to be the team’s decision.

OUTCOMES OF DECISIONS

According to the accuracy of your decision, there are ﬁve possible evaluative outcomes
(scoring is done automatically by the computer). ONLY THE DECISION
REGISTERED BY YOU, THE LEADER is considered to be the team’s decision. The
ﬁve possible outcomes are:

153

154

 

 

 

 

 

 

 

OUTCOME DEFINITION EXAMPLE SCORE

(1) HIT The decision was Carrier said defend, correct 2
exactly correct answer was defend

(2) NEAR MISS The decision was off Carrier said defend, correct 1
by one level answer was lock-on

(3) MISS The decision was off Carrier said defend, correct 0
by two levels answer was ready

(4) INCIDENT The decision was off Carrier said defend, correct -1
by three levels answer was warn

(5) DISASTER The decision was off Carrier said defend, correct -2
by more than three answer was either monitor,
levels review, or ignore

 

 

 

 

 

 

155

EXPERIMENT I
LEADER POST-SESSION QUESTIONNAIRE

Please indicate how much you feel you weighted (utilized) the judgments of each of your
staff members. Divide 100 points across each of the staff members in a manner that
reﬂects how you think you weighted each staff member during the simulation.

Example: If you felt you weighted the CAD the most, followed by the AWAC, and
tended to ignore the Cruiser, you would ﬁll in the blanks like this:

CAD: 65
AWAC: 35
Cruiser: 0

Your response:
CAD:

AWAC:

Cruiser:

Subordinate Accuracy

Please indicate how accurate (on a scale of 0-100) you feel each of your staff members
was in predicting the correct decision:

CAD:

AWAC:

Cruiser:

156

EXPERIMENT I
DEBRIEFING FORM

The purpose of this study was to investigate factors that affect a leader's ability to
accurately utilize information provided by subordinates. Judgment conﬁdence
information (e. g., the CO would recommend an IGNORE decision and add that s/he was
80% conﬁdent), staff member past judgment information, neither, or both were provided
to different participants to test whether this information improved team performance and
the appropriateness of leaders' strategies for combining the judgments of the four COs.
People's ability to accurately weight staff members in conditions in which differential vs.
equal weighting strategies are appropriate and no staff past judgment accuracy or
judgment conﬁdence information is available was also investigated. If you have any
questions about this study, please contact Jean Phillips in the Management Department at
353-7116.

To avoid affecting the results of this study, IT IS VERY IMPORTANT that you do not
discuss your experience with this study with other students who might participate in the

future.

Thank you for participating!

APPENDIX D

EXPERIMENT 11 MATERIALS:
POSITION-SPECIFIC TRAINING MANUALS FOR STAFF,
POST-SESSION QUESTIONNAIRES,

AND DEBRIEFING FORM

APPENDIX D

EXPERIMENT II
POSITION-SPECIFIC STAFF TRAINING MANUALS

[NOTE: PARTICIPANTS WILL ONLY RECEIVE INFORMATION ABOUT
THEIR POSITION AND THE ATTRIBUTES AND COMBINATION RULE
RELEVANT TO THEIR PARTICULAR AREA OF RESPONSIBILITY]

The CAD is the Coastal Air Defense unit (located on a beach). As the Commanding
Ofﬁcer (CO) of the CAD, you are responsible for providing the Carrier (the team leader)
with a judgment summarizing a target’s standing in terms of it’s location.

 

DETERMINING THE LEVEL OF THREAT FOR A TARGET

In general, the degree to which an incoming target is threatening depends on its standing
on nine attributes, three of which you are responsible for interpreting. These nine
attributes combine into three simple rules which in combination are used to determine the
danger associated with any target. The commanding officers of the CAD, AWACS, and
Cruiser are each responsible for combining three different attributes into one of these
rules. The commanding ofﬁcer of the Carrier is responsible for combining staff
summaries of these three rules into a correct overall team decision.

CHARACTERISTICS OF AIRBORNE TARGETS
The three attributes of targets for which you, as the C0 of the AWAC, are responsible are
listed below along with the ranges of possible values for these attributes:

 

 

CAD
(1) Altitude Lower targets are more threatening. 35,000 to 5,000 ft.
(2) Corridor A corridor is a 20 mile wide "safe lane" open to commercial
Status air trafﬁc. Targets in the center of the corridor are less
threatening than those farther away from the center of the

corridor.
0 miles (in the middle of it) to 30 miles (way out of it)

 

(3) Range Distance of the aircraft from the Carrier. 200 miles to 1 mile

 

 

 

 

 

Location Rule (CAD):

ALTITUDE, CORRIDOR STATUS, and RANGE go together to determine the location
of the aircraft. Aircraft are threatening only if they are low (low value on altitude),
outside commercial trafﬁc lanes (high value on corridor status), and close (low value on
range) to the Carrier. If any one of these three values are non-threatening, then the aircraft
is to be considered none-threatening in terms of the location rule

157

158

The AWAC is an air reconnaissance plane. As the Commanding Ofﬁcer (C0) of the
AWAC, you are responsible for providing the Carrier (the team leader) with a judgment
summarizing a target’s standing in terms of its movement.

DETERMINING THE LEVEL OF THREAT FOR A TARGET

In general, the degree to which an incoming target is threatening depends on its standing
on nine attributes, three of which you are responsible for interpreting. These nine
attributes combine into three simple rules which in combination are used to determine the
danger associated with any target. The commanding ofﬁcers of the CAD, AWACS, and
Cruiser are each responsible for combining three different attributes into one of these
rules. The commanding ofﬁcer of the Carrier is responsible for combining staff
summaries of these three rules into a correct overall team decision.

CHARACTERISTICS OF AIRBORNE TARGETS
The three attributes of targets for which you, as the C0 of the AWAC, are responsible are
listed below along with the ranges of possible values for these attributes:

 

 

AWAC
(1) Speed Faster targets are more threatening.
100 to 800 m.p.h.
(2) Angle Descending targets are more threatening - the sharper the

descent, the greater the threat +15 degrees (rapid ascent) to
-15 degrees (rapid descent)

 

(3) Direction Targets headed directly at the Canier are more dangerous
than those passing far to the left or right +30 degrees
(passing far to the left or right of the Carrier) to 00
degrees (coming straight into the Carrier)

 

 

 

 

Movement Rule (AWACS):

SPEED, ANGLE, AND DIRECTION go together to determine the movement of the
aircraft. Aircraft are threatening only if they are going fast (high value on speed),
descending (low value on angle), and coming straight in to the Carrier (low value on
direction). If any one of these three values are non-threatening, then the aircraft is to be
considered non-threatening in terms of the movement rule.

 

159

The CRUISER is a large ship that provides support from the ocean. As the Commanding
Officer (C0) of the Cruiser, you are responsible for providing the Carrier (the team
leader) with a judgment summarizing a target’s standing in terms of its classiﬁcation.

DETERMINING THE LEVEL OF THREAT FOR A TARGET

In general, the degree to which an incoming target is threatening depends on its standing
on nine attributes, three of which you are responsible for interpreting. These nine
attributes combine into three simple rules which in combination are used to determine the
danger associated with any target. The commanding ofﬁcers of the CAD, AWACS, and
Cruiser are each responsible for combining three different attributes into one of these
rules. The commanding ofﬁcer of the Carrier is responsible for combining staff
summaries of these three rules into a correct overall team decision.

CHARACTERISTICS OF AIRBORNE TARGETS
The three attributes of targets for which you, as the C0 of the Cruiser, are responsible are
listed below along with the ranges of possible values for these attributes:

 

 

CRUISER
(1) Size Smaller targets are more threatening. 65 to 10 meters
(2) IFF IFF stands for "Identiﬁcation Friend of Foe," this is a radio

signal that identiﬁes whether an aircraft is civilian,
para-military or military. .2 MHz (civilian) to 1.8 MHz
(ﬁghter)

 

(3) Radar Type The kind of radar possessed by the aircraft. Class 1 (weather
radar only) to Class 9 (weapons radar)

 

 

 

 

 

Category Rule (Cruiser):

SIZE, IF F, and RADAR TYPE go together to determine the category of aircraft. Aircraft
are threatening only if they are small (low value on size), military (high value on IFF) and
carrying weapons radar (high value on radar). If any one of these three values are
non-threatening, then the aircraft is to be considered non-threatening in terms of the
category rule.

160

(THIS PAGE GIVEN TO ALL POSITIONS)

CONFIDENCE (11' called for by the condition)

In addition to sending the leader your judgment of the target, you will also be asked to
send an indication of how conﬁdent you are (on a scale of 0% reﬂecting a guess to 100%
reﬂecting extreme conﬁdence) in the accuracy of your judgment. For example, in
addition to recommending an IGNORE decision, you will inform the leader that you are
80% conﬁdent that this judgment is correct, based on the information you have acquired.
This will become more clear during training.

PAST JUDGMENT ACCURACY (if called for by the condition)

During each active target, there will be a red bar on the leader’s screen near the icon that
represents your station. This bar reﬂects how accurate your judgments have been in the
past (in terms of the strength of the relationship between your judgments and the correct
decision). There will also be a number to the left of the red bar on a ~100 to +100 scale.
The number and bar will only be seen by the leader. The longer the red bar and the more
positive the number, the more accurate your judgments have been.

COMBINING THE THREE RULES INTO A TEAM DECISION

The CO of the Canier is responsible for combining the CO's judgment of the target's
standing on each of three rules into the team decision. ONLY THE DECISION
REGISTERED BY THE LEADER is considered to be the team's decision.

OUTCOMES OF DECISIONS

Your decisions regarding each target are to be made based upon the information on the
dimensions listed above. According to rules described in this section, there are ﬁve
possible evaluative outcomes associated with the accuracy or the team's decisions
(scoring is done automatically by the computer). ONLY THE DECISION
REGISTERED BY THE LEADER is considered to be the team's decision. The ﬁve
possible outcomes are:

161

 

 

 

 

 

 

 

OUTCOME DEFINITION EXAMPLE SCORE

(1) HIT The decision was Carrier said defend, correct 2
exactly correct answer was defend

(2) NEAR MISS The decision was off Carrier said defend, correct 1
by one level answer was lock-on

(3) MISS The decision was off Carrier said defend, correct 0
by two levels answer was ready

(4) INCIDENT The decision was off Carrier said defend, correct -1
by three levels answer was warn

(5) DISASTER The decision was off Carrier said defend, correct -2
by more than three answer was either monitor,
levels review, or ignore

 

 

 

 

 

 

162

EXPERIMENT II
WILLINGNESS TO RETURN QUESTIONNAIRE

Please circle one of the three options for each of the following questions. If you respond "yes" or
"maybe" for any of the questions, you may be contacted later this term regarding ﬁrrther
opportunities to participate in this or other research projects. You may change your mind and
decline to participate at any time, but if you think that you might be interested, please indicate
this below.

1. Would you be willing to return for $7.50/hour later this term to participate in further research,
doing the SAME type of task with the SAME team members?

YES NO MAYBE

2. Would you be willing to return for $7.50/hour later this term to participate in further research,
doing a DIFFERENT task but with the SAME teammates and SAME leader?

YES NO MAYBE

3. Would you be willing to return for $7.50/hour later this term to participate in further research,
doing the SAME type of task with the SAME leader but with DIFFERENT teammates?

YES NO MAYBE

4. Would you be willing to return for $7.50/hour later this term to participate in further research,
doing a DIFFERENT task with the SAME leader but with DIFFERENT teammates?

YES NO MAYBE

5. Would you be willing to return for $7.50/hour later this term to participate in further research,
doing the SAME type of task with the SAME teammates but with a DIFFERENT leader?

YES NO MAYBE

6. Would you be willing to return for $7.50/hour later this term to participate in further research,
doing a DIFFERENT task with the SAME teammates but with a DIFFERENT leader?

YES NO MAYBE

Name:

 

(please print)

Student Number: Phone Number:

 

163

EXPERIMENT II
DESIRE TO CHANGE FOR NEXT TASK QUESTIONNAIRE

The next thing you will be doing is a second decision making task. The top performers on
this task will receive a cash bonus: First Place will receive $10, Second and Third Place
will each receive $5.
For the next decision making task, would you like to:
l. Remain working with the same leader? (circle one)

Keep the Same Leader Change
2. Remain working with the same team? (circle one)

Keep the Same Team Change

3. Do the next task as an individual? (circle one)

Work as an Individual Work With a Team

164
EXPERIMENT II

STAFF SATISFACTION WITH LEADER QUESTIONNAIRE
Adapted from Seashore, Lawler, Mirvis and Cammann (1982)
Please use the following scale in responding to the statements below. There are no right
or wrong answers; please answer honestly. Fill your response in the corresponding circle

on the computer scorable answer sheet.

1 = Strongly Disagree

2 = Disagree
3 = Neither Agree nor Disagree
4 = Agree

5 = Strongly Agree
1. Overall, I am satisﬁed with my leader.
2. In general, I don't like my leader.
3. I would be willing to work with this leader again in the future.
4. I think my leader did a poor job in making decisions.
5. If we were to perform another set of targets, 1 would deﬁnitely want to change leaders.
6. I think I could have done a better job than my leader did.
7. I am satisﬁed with my team's performance.

8. I think my ﬁiends would be interested in applying for this project.

165

EXPERIMENT II
WITHDRAWAL FROM TASK QUESTIONNAIRE

Adapted from Baker (1991) and Gilliland (1992)

Please use the following scale in responding to the statements below. There are no right
or wrong answers; please answer honestly. Fill your response in the corresponding circle
on the computer scorable answer sheet.

1 = Strongly Disagree

2 = Disagree
3 = Neither Agree nor Disagree
4 = Agree

5 = Strongly Agree
1. I often daydreamed while working on the task.
2. I was frequently bored while working on the task.
3. At the end of the session, I felt as though I had accomplished something.
4. I felt that I was working below my abilities most of the time.
5. When I was doing the task I wished I was anyplace else.
6. I often thought of quitting the task.
7. If I hear of other projects like this, I would be interested in participating.

8. If I knew in advance what this project would entail, I would not have chosen to
participate.

9. I would recommend this project to my classmates.

166

EXPERIMENT II
SELF-EFFICACY QUESTIONNAIRE

Adapted from Locke, Frederick, Lee & Bobko (1984)
This set of questions asks you to describe how you feel about your capabilities to perform

the task if you were toaperform the simulation again. Please use the scale shown below to
make your ratings.

 

Strongly Strongly
Disagree Disagree Neutral Agree Agree
<_-| | I I l >

I l | I
(1) (2) (3) (4) (5)
1. I can meet the challenges of my role in this simulation.

2. I am conﬁdent in my understanding of how information cues are related to the
decisions I have to make.

3. I can deal with decisions under ambiguous conditions.

4. I am certain that I an manage the requirements of my position for this task.

5. I believe I will fare will in this task even if the workload is increased.

6. I am conﬁdent that I can cope with my role if the simulation becomes more complex.
7. I believe I can develop methods to handle the requirements of my task and my role.

8. I am certain I can cope with task components competing for my time.

167

EXPERIMENT II
DEBRIEFING FORM

The purpose of this study was to investigate factors that affect decision making processes
in hierarchical decision making teams, as well as the consequences of different decision
making processes. Staff member reactions to leaders differentially vs. equally utilizing
staff member judgments in making the team decision were investigated. If you have any
questions about this study, please contact Jean Phillips in the Management Department at
3 5 3 -7 l 1 6.

To avoid affecting the results of this study, IT IS VERY IMPORTANT that you do not
discuss your experience with this study with other students who might participate in the
future.

Thank you for participating!

APPENDIX E

REPEATED MEASURES REGRESSION ANALYSES
FOR EXPERIMENT II USING EFFECTS CODING

APPENDIX E

REPEATED MEASURES REGRESSION ANALYSES
FOR EXPERIMENT II USING EFFECTS CODING

Repeated Measures Regression Analysis of LUSJ and Team Performance on Willingness
to Return

 

 

 

A in A in A in Incremental F
Variable Total R2 Within R2 Between R2 (df,df)
Team Performance .021* .055" 4.32 (1,74)
LUSJ .021 .034 1.74 (3,149)
Performance X LUSJ .043"‘ .069* 3.77 (3,146)
Total R2 .085" .103* .055*

 

lira; The higher the score is, the greater the willingness to return. N=228. Between-team
variance=.48 (38%); within-team variance=.78 (62%). Total df within-team=152; total df
between-team=75.
*p<.05. **p<.01.

Repeated Measures Regression Analysis of LUSJ and Team Performance on Desire to
Change for Next Task

 

 

 

A in A in A in Incremental F
Variable Total R2 Within R2 Between R2 (df,df)
Team Performance .118" .273“ 27.83 (1,74)
LUSJ .040* .070* 3.76 (3,149)
Performance X LUSJ .018 .032 1.72 (3,146)
Total R2 .176" .102* .273"

 

w; The higher the score is, the greater the desire to change for the next task. N=214.
Between-team variance=.37 (43%); within-team variance=.49 (57%). Total df within-
team=142; total df between-team=71 .

*p<.05. *"‘p<.01.

168

169

Repeated Measures Regression Analysis of LUSJ and Team Performance on Satisfaction
With Leader

 

 

 

A in A in A in Incremental F
Variable Total R2 Within R2 Between R2 (df,df)
Team Performance .283" .466" 64.51 (1,74)
LUSJ .002” .005 .25 (3,149)
Performance X LUSJ .016 .041 2.08 (3,146)
Total R2 .301" .046 .466"

 

N91; The higher the score is, the greater satisfaction with the leader. N=228. Between-
tearn variance=.35 (61%); within-team variance=.23 (39%). Total dfwithin-team=152;
total df between-team=75.

*p<.05. "p<.01.

Repeated Measures Regression Analysis of LUSJ and Team Performance on Task
Withdrawal

 

 

 

A in A in A in Incremental F
Variable Total R2 Within R2 Between R2 (df,df)
Team Performance .079** .180“ 16.24 (1,74)
LUSJ .046" .082M 4.44 (3,149)
Performance X LUSJ .019 .034 1.86 (3,146)
Total R2 .144" .116" .180M

 

N_o_t§_. The higher the score is, the greater the task withdrawal. N=228. Between-team
variance=.18 (44%); within-team variance=.23 (56%). Total dfwithin-team=152; total df
between-team=75.

*p<.05. **p<.01.

170

Repeated Measures Regression Analysis of LUSJ and Team Performance on Self-
Efﬁcacy

 

 

 

A in A in A in Incremental F
Variable Total R2 Within R2 Between R2 (df,df)
Team Performance .038" .085* 6.86 (1,74)
LUSJ .013 .024 1.20 (3,149)
Performance X LUSJ .031 .056* 2.97 (3,146)
Total R2 .082* .080* .085*

 

& The higher the score is, the greater the self-efficacy. N=228. Between-team
variance=. 14 (45%); within-team variance=.18 (55%). Total dfwithin-team=152; total df
between-team=75.

*p<.05. **p<.01.

"llllllllllllllllill-l“