t: . 1x I :
t :4.

.24 .5.

bu Al.

:x.
E

 

in
. .udhuzé
.fiti...3mm....«.:ow ‘ 1.: an“... :.. .
u H fl
Gufif? ‘u.
.Vifiummflhr.

 

 

1.01 «a.
ii; ‘
. b:

.3: ramp...

 

 

 

 

 

 

 

 

“Jammy”;

92:113.!
1:... :4,

 

 

I
'II'

 

 

 

 
 

Dunlrfimmmm
.I. I o

:1

'Qvl’
£17.. ,

I"; 5»: r a! u

1 1
. ha.
. fix. ..
I. ‘\’
£33.14.

gt

3.3 O: I e":
w!) J. at
5 .1 u. 41!? 11.4%!
. i .l .
:3: 31’1””

.. :. 51:11: ...
Irauxt

  

er. in?
awning?
we; I

(1.51:... 4
.519va v
‘13.:

a... (I.

.%‘:J..m¥&zzzsfi§
it n I "l. .o

. .i

‘1 4.3.6.:

Shutalxllv

to. .

, i. a
1.35»...

 

 

valiaua...‘v.

. ' :2. 3!.

i

. 3‘.)

.16!!!
a

142.3...

\‘tils

 

x: : .‘nltilt:
. 3 :31... 3.....59. 3
I0. n.131l.5t.|v¢. 1
1 {0'13 .
2.

6.:
. .4? 1.

    

‘ .,
.x.-t..t$d.s~%«h«fi .
1.1.: «

 
  

This is to certify that the
dissertation entitled

THE EFFECT OF WORKLOAD ON INDIVIDUAL AND TEAM
LEARNING, AFFECT, AND PERFORMANCE

presented by

MICHAEL D. JOHNSON

has been accepted towards fulfillment
of the requirements for the

PhD degree in Management

 

 

/flZ/ZJZ—d

 

Major Professor’ 3 Signature

Date

MSU is an Affirmative Action/Equal Opportunity Institution

UbHARY
Michigan State
University

 

‘ '-—-~-a-o-o---a--------.---.-u---v-I--o-------n---.-o---- ---

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 p:/ClRC/DateDue.indd-p.1

THE EFFECT OF WORKLOAD ON INDIVIDUAL AND TEAM LEARN ING.
AFFECT. AND PERFORMANCE

By

Michael D. Johnson

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Management

2006

ABSTRACT

THE EFFECT OF WORKLOAD ON INDIVIDUAL AND TEAM
LEARNING, AFFECT. AND PERFORMANCE

By

Michael D. Johnson

This experiment examined the relationships between workload on the one hand, and
emotional arousal, learning, and performance on the other, among both individuals
working alone and work teams. Through integrating various cognitive and motivational
theories, I show that there is reason to believe that the relationships will vary across
levels. Results indicated that the effect of workload on emotional arousal for individuals
was moderated by neuroticism and extraversion, but in teams the effect was moderated
by team structure and transactive learning. Workload was negatively related to learning
in teams, and the effect persisted even when team workload decreased. In contrast, for
individuals working alone, the effect of workload on learning was moderated by the
individual’s level of neuroticism. Previous workload had no direct effects of subsequent
performance for individuals working alone, but was marginally positively related to

subsequent speed of performance in teams.

ACKNOWLEDGEMENTS

I would like to thank my dissertation committee for their hard work and rigorous reviews
of this research: John R. Hollenbeck (chair), Daniel R. Ilgen, Frederick P. Morgeson, and
Remus Ilies. I would also like to offer special thanks to the following people who assisted
selflessly in data collection: Jeff Audretsch, Alex Barelka, Christopher Barnes, Scott
DeRue, Nikos Dimotakis, Stephen Harmon, Dustin J undt, Alicia Luzkow, Jennifer
Nahrgang, Kelly Schwind, Kyle Smith, Matthias Spitzmuller, David Wagner, and Harold

Willaby.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES ......................................................................................................... viii
INTRODUCTION ............................................................................................................... I
INDIVIDUAL WORKLOAD AND PERFORMANCE ..................................................... 4
Emotional Arousal ........................................................................................................ 13
Individual Learning ...................................................................................................... 16
TEAM WORKLOAD AND PERFORMANCE ................................................................ 28
Collective Arousal and Emotional Contagion .............................................................. 28
Team learning .............................................................................................................. 32
MODERATORS OF THE EFFECTS OF WORKLOAD ................................................. 39
Personality .................................................................................................................... 39
Structure ........................................................................................................................ 44
METHOD .......................................................................................................................... 48
Research Participants and Task ................................................................................... 48
Manipulations and Measures ........................................................................................ 50
Data Analysis ................................................................................................................ 57
RESULTS .......................................................................................................................... 59
Descriptive Statistics ..................................................................................................... 59
Hypothesis Testing ........................................................................................................ 60
Additional Post-hoe Analyses ....................................................................................... 68
DISCUSSION .................................................................................................................... 75

iv

Findings for Individuals Working Alone ...................................................................... 75

F indings for Teai'ns ....................................................................................................... 79
Limitations .................................................................................................................... 79
Post-hoc Inferences ....................................................................................................... 84
Theoretical Implications ............................................................................................... 89
Practical Applications .................................................................................................. 9 1
REFERENCES .................................................................................................................. 93
APPENDIX ...................................................................................................................... l 07

LIST OF TABLES

Table 1. Means and standard deviations .......................................................................... 107

Table 2. lntercorrelations between experimental manipulations and pregame measures

.......................................................................................................................................... 108
Table 3. lntercorrelations between Game 1 measures ..................................................... 109
Table 4. lntercorrelations between Game 2 measures ..................................................... 110

Table 5. Correlations between experimental manipulations. pregame measures, and Game
1 measures ........................................................................................................................ I II
Table 6. Correlations between experimental manipulations, pregame measures, and Game
2 measures ........................................................................................................................ 112
Table 7. Correlations between Game 1 measures and Game 2 measures ........................ 1 13
Table 8. Repeated measures regression of proposed mediators on workload and time for
individuals working alone in Game 1 .............................................................................. I 14
Table 9. Repeated measures regression of performance dimensions on workload and time
for individuals working alone .......................................................................................... 115
Table 10. Repeated measures regression of emotional arousal indicators on workload and
game for teams in Game 1 ............................................................................................... l 16
Table 11. Repeated measures regression of performance dimensions on workload and
time for teams .................................................................................................................. 1 17
Table I2. Repeated measures regression of learning indicators on workload and game for

teams in Game I ............................................................................................................... l 18

vi

Table 13. Moderated regression of Game 1 emotional arousal on workload and
personality for individuals working alone ....................................................................... 119
Table 14. Moderated regression of Game 1 emotional arousal on workload and
personality for teams ........................................................................................................ 120
Table 15. Moderated regression of Game 1 learning on workload and personality for
individuals working alone ................................................................................................ 121

Table 16. Moderated regression of Game 1 learning on workload and personality for

teams ................................................................................................................................ l 22

Table 17. Moderated regression of emotional arousal indicators on manipulations for

teams ................................................................................................................................ 123

Table 18. Moderated regression of learning indicators on manipulations for teams ....... 124

vii

LIST OF FIGURES

Figure l. Hypothesized model ......................................................................................... 125
Figure 2. Interactive effects of workload and time on speed of performance ................. 126

Figure 3. Interactive effects of workload and Extraversion on individual emotional

arousal .............................................................................................................................. 127

Figure 4. Interactive effects of workload and Neuroticism on individual emotional arousal

.......................................................................................................................................... 128
Figure 5. Interactive effects of workload and Neuroticism on individual learning ......... 129
Figure 6. Interaction of workload and type of track on friendly fire kills in teams ......... 130
Figure 7. Interaction of workload and type of track on good attacks in teams ................ 131
Figure 8. Post-hoe team model ........................................................................................ 132

viii

INTRODUCTION

In response to the increased use of team-based structures in organizations,
research on team effectiveness has greatly expanded (Sanna & Parks, 1997). Two broad
characteristics appear to be evident in recent teams research. First, there is a greater
recognition that teams do not perform in isolation, but rather are embedded within larger
organizations, and are themselves made up of smaller units (i.e., individual members). In
this respect, teams engage in a dynamic interplay with both larger and smaller levels of
analysis, affecting and being affected by both their members and their organizations
(Arrow, McGrath, & Berdahl, 2000; Ilgen, Hollenbeck, Johnson, & Jundt, 2005). Second,
whereas previous theory and research largely conceptualized teams as static entities
engaging in single-cycle performance contexts, an emerging body of research examines
teams in dynamic contexts that perform across time. This research has found that the past
history of teams often reaches forward and affects their future processes and performance
(Johnson et al., 2006; Moon et al., 2004).

The first characteristic is most evident in multi-level theory and research
(Kozlowski & Klein, 2000). Although multi-level theories can take many forms, one of
the most basic is the homologous model. Homologous models suggest that the
relationships between constructs are isomorphic at more than one level of analysis. For
example, Chen et a1. (2002) found that the relationship between achievement motivation
and performance was mediated by efficacy beliefs at both the individual and team levels.
Tests of homology are critical in understanding organizational behavior, because they
highlight the similarities and differences between phenomena at the individual. team. and

organizational levels (House, Rousseau, & Thomashunt. 1995).

The second characteristic is evident in research that takes multiple measures of
team processes and performance. Rather than examining phenomena with between-team
designs, these studies examine within-team change over time. This research recognizes
that “a group’s activity (is) in part a function of its own past history” (Arrow, McGrath,
& Berdahl, 2000, p. 29), often examining team performance in dynamic contexts. For
example, Johnson et a1. (2006) found that teams’ past reward systems affected their
future communication and performance, even after the reward system changed.

Drawing on both of these broad trends in teams research, this research tested a
model of performance at the individual and team levels in a dynamic context.
Specifically, in a laboratory setting, this experiment tested the effects of past workload on
the subsequent performance of both teams and individuals working alone. I propose that
although the relationship between past workload and subsequent performance is likely to
be homologous across individual and team levels in terms of speed (or quantity), it is
likely to vary in terms of accuracy (or quality). As will be shown, this is because the
effect of workload on accuracy depends to a large extent upon learning, which operates
differently at the individual and team levels. In contrast, I propose that the effect of
workload on speed is largely a function of emotional arousal, which develops similarly
across levels. Figure l graphically depicts the entire hypothesized model. I note that this
figure is not a structural model, but rather a heuristic for examining the relationships
between the various constructs.

Prior to describing the rationale for this model. boundary conditions must be
established. First, the model assumes that veridical feedback is being provided on one‘s

task performance. and thus does not apply in situations where no feedback, inaccurate

Is)

feedback, or equivocal feedback is provided. Feedback is an important factor in
influencing one’s performance (Kluger & DeNisi, 1996). It may be provided by a
supervisor, by coworkers. or may be derived from the task itself. Second, the model
applies only in situations where workers have at least some degree of latitude over the
way they perform their jobs. This applies both to the behaviors that the workers engage
in, as well as the opportunity to make decisions regarding their work. Karasek’s (1979)
job demands/control theory (discussed in later parts of the manuscript) suggests that
many of the effects of workload depend upon whether workers have some control over
their jobs, and my model is expected to operate at the higher end of the control
continuum. Third, the model is expected to apply in situations where the task is
somewhat novel. The model suggests that learning is one of the mediators of the
relationship between workload and performance, and therefore the task must be one that
is not already well-leamed. Fourth. the team portion of the model only applies to teams
without well-established communication and coordination patterns. Previous research has
established that teams with such patterns adapt more effectively to high-stress situations
than teams without such patterns (Entin & Serfaty, 1999). Therefore, this model best
applies to teams that are relatively new. Finally, the team portion of the model is also
expected to apply only to teams that work together relatively regularly and have at least a
moderate level of interdependence. It is not expected to apply to teams where the
individuals typically work alone and only come together sporadically to discuss their

work.

INDIVIDUAL WORKLOAD AND PERFORMANCE

Although the effect of workload on various outcomes has received consistent
attention in the psychology literature for the past thirty years, it has become a particularly
salient applied issue in recent years. due to downsizing, business expansion, and
outsourcing (Schwartz, 2004). Both objective and subjective measures indicate that
employee workloads have increased over the past few decades. According to the
Economic Policy Institute, the average annual hours worked by Americans increased
from 1,720 hours in 1973 to 1.898 hours in 1998 (Schor, 2002); an even more recent
survey showed that 62% of workers said their workload had increased over the past six
months, and more than half claimed their workload left them “overtired and
overwhelmed” (Schwartz, 2004). Excessively high workloads can lead to stress-related
health issues, including heart disease (Vahtera et al., 2004) and mental health problems
(Ganster, Fox, & Dwyer, 2001). These health problems are harmful for the organization
as well as the employee, as the organization incurs costs from increased employee sick
leave (Westerlund et al., 2004) and may be legally liable for injury if they could have
reasonably foreseen the risk ("The psychiatric injury liabilities of excessive workloads",
2005). Additionally, excessive workload may cause workers to commit errors; recent
cases have included a medical center that failed to inform patients about the results of
their cancer screening tests (Santora, 2005) and a train crash that caused the death of four
people ("Dispatcher's errors caused fatal crash, NTSB rules", 1998).
Workload as a Construct

Workload is a deceptively complex construct. Although the word seems to be

such an established part of the lexicon, it did not appear in most dictionaries until the

1970’s (Huey, Wickens, & National Research Council Commission on Behavioral &
Social Sciences & Education, 1993). At first blush. it appears to refer simply to how
much work one has to do. It is not clear, however, whether this refers to objective
workload or subjective perceptions of workload; whether the amount of work is simply
quantitative or contains a qualitative dimension; or whether it is a characteristic of the
job, the worker, or both. Indeed, theorists have had a difficult time both defining and
agreeing upon the definition or workload. Huey et a1. (1993, p. 54) note that “operational
definitions (of workload) proposed by psychologists and engineers continue to disagree
about its source(s), mechanism(s), consequence(s), and measurement.” Of the related
construct of mental workload, Xie and Salvendy (2000, p. 76) noted, “The simple fact is
that nobody seems to know what mental workload is.” Thus. I wade into turbulent waters
as I discuss the construct of workload. First, I discuss what workload is and is not; then I
distinguish it from the related concepts of job demands, job challenge, and job stress; and
finally, I briefly discuss the two major lines of research that have dealt with workload.

What workload is and is not. The first issue that must be resolved is whether
workload is an objective characteristic of the work environment, or a subjective
perception of the worker. Shaw and Weekley (1985, p. 88) made the case for treating it as
an objective characteristic by stating that managers “are, for the most part, limited to
manipulating the objective characteristics of the work environment. . Several empirical
studies have adopted this objective approach to workload, including ones where workload
was experimentally manipulated in laboratory (Froggatt & Cotton, 1987; Perrewe &

Ganster, 1989; Shaw & Weekley, 1985) and field settings (Parkes. 1995), as well as

observed in naturalistic field settings (Hurst & Rose, 1978; Kakimoto, Nakamura, Tarui,
Nagasawa, & et al., 1988; Rose & Fogg, 1993).

Other research has adopted a more subjective approach to workload. This research
suggests that it is an individual’s perceptions of workload, rather than the objective
amount of work itself, that best represents the construct. This perspective is reflected in
the work of Daniel Gopher, who defined workload as “the interaction between a person
and a task” in the sense that people have different maximum capacities for handling
different levels of work (Gopher & Donchin, 1986). Gopher likens workload to electrical
current, in that current is a joint property of both the voltage supplied by the battery and
the resistance provided by the circuit. Thus, two people may perceive objectively
identical workloads differently, because they provide different “resistances” to their
tasks. Empirical studies using subjective perceptions of workload have been conducted
primarily in field settings (Greenglass, Burke, & Moore, 2003; Jacobs & Dodd, 2003;
Markel & Frone, 1998; Spector, Dwyer, & Jex, 1988; Yeung, Genaidy, Deddens, &
Sauter, 2005), although subjective perceptions of workload have also been used in
laboratory settings (Fisher & Ford, 1998). Drawing upon this idea that both the demands
of the task and the capacity of the individual determine workload, I view workload as an
objective construct that is determined both by the objective demands of the task and the
objective capacity of the individual. Empirically, however, it is difficult to determine an
individual’s workload capacity (particularly in non—physical tasks).

A second issue that must be resolved in the definition of the construct of workload
is whether it is strictly quantitative or if it has a qualitative dimension as well. Most

workload research has at least implicitly viewed workload as quantitative. as evidenced

by the way the construct has been operationalized. The two most commonly used
subjective measures of workload, the NASA Task Load Index (NASA-TLX; Hart &
Staveland. 1988) and the Subjective Workload Assessment Technique (SWAT; Reid &
Nygren. 1988) use items like, “How much time pressure did you feel due to the rate or
pace at which the task occurred?” (NASA-TLX) and, “Often have spare time.
Interruptions or overlap among activities occur infrequently or not at all” (SWAT). These
representative items clearly tap into the idea that workload deals more with “how much”
work one has to do rather than “what type” of work one is doing.

Far less research has examined workload as a qualitative construct. Shaw and
Weekley (1985) discussed the concepts of “qualitative overload” and “qualitative
underload." Qualitative overload exists when the individual cannot complete their tasks
regardless of the amount of time they have to do them; qualitative underload exists when
the tasks assigned to the individual are far below their ability, “such that the tasks are
completed with boring ease” (p. 87). Because of its reference to ability, qualitative
workload could also be conceptualized as a fit construct; one’s qualitative workload is
determined by how closely the nature of one’s tasks matches one’s ability. Although this
is an intriguing concept that warrants further examination, it is beyond the scope of the
current study. Therefore, in keeping with the vast majority of workload research, I view
workload as solely quantitative, reflecting how much one has to do rather than what type
of work one is doing.

Hence, with these two issues resolved. the definition of individual workload used
in this study is the degree to which the amount of work assigned to an individual is within

that individual '3 capacity within the specified duration of time. The definition also

applies to the team level with “team” substituted for “individual.” This is in keeping with
at least one existing definition of team workload; Bowers, Braun, and Morgan (1997)
defined team workload as “the relationship between the finite performance capacities of a
team and the demands placed on the team by its performance environment.” Thus, like
Bowers et al. (1997). I assume that teams, like individuals, vary in their capacity for
work. This capacity is likely not simply the sum of the individual team members’
capacities, partly because teams are subject to process losses and gains (Steiner, 1972),
but also because the type of tasks assigned to teams affects their capacity. For example,
in conjunctive tasks, team performance is dependent upon the ability of the worst
performing member, whereas in disjunctive tasks, team performance is dependent upon
the ability of the best performing member. Therefore, team workload capacity depends
upon the type of tasks facing the team in a way that is qualitatively different from the
individual level. Nevertheless, the conceptualization of workload itself is isomorphic
across levels.

Related constructs. Having defined workload, I now turn to distinguishing it from
the related constructs of job demands, job challenge, and stress. The term job demands
derives from Karasek’s (1979) job demands/controls theory. In short, this theory holds
that jobs vary meaningfully both in terms of the demands they place on the employee and
the latitude (control) with which the employee can make decisions. According to this
theory, jobs that are high in demands but low in control place a great deal of strain on the
employee (high strain jobs); jobs that are high in demands and high in control (active
jobs) lead to high productivity and high job satisfaction; job that are low in demands but

high in control (low strain jobs) are relaxing and leisurely: and jobs that are low in

demands and low in control (passive jobs) lead to apathetic behavior. In this theory, job
demands refers specifically to psychological pressures on employees, of which workload
is just one of many: Karasek suggests that job demands include “work load demands,
conflicts or other stressors which place the individual in a motivated or energized state of
‘stress'” (p. 287). Thus, job demands may be considered a superordinate category of job
stressors that includes workload.

The construct of job challenge arose out of application of role theory to
organizational settings (Berlew & Hall, 1966). The notion was that people are given high
challenges early in their careers (or their employment in a particular organization), they
will rise to the challenge and fill that role with all of its expectations for good
performance. Moreover, job challenge had a carryover effect, such that high job
challenge was associated with better performance later in one’s career (e.g., Berlew &
Hall, 1966; Orpen, 1974). As a construct, job challenge appears to be somewhat similar
to job demands, in that the early literature on job challenge referred to “relatively
demanding jobs” and “less demanding tasks” (Berlew & Hall, 1966). It does not,
however, include the other informal stressful aspects of jobs (such as conflict) that
characterize job demands; instead, job challenge appears to refer to the formal aspects of
assigned work. In this respect, job challenge may also be considered to be a superordinate
category that includes workload, but may also include other job characteristics. For
example, a job challenge that may not involve greater amounts of work could be
responsibility for major decisions or a major project.

Finally, workload must be distinguished from the general term job stress. These

terms are closely linked in the research literature, as workload has often been considered

to be a “job stressor” (Spector et al., 1988). Indeed, both inordinately high and
inordinately low levels of workload can be stressful if they are sustained for long periods.
Gaillard (1993) noted that because of this, the constructs often confound each other in
empirical research. They are similar in that both are determined by demands of the
environment and the capacity of the individual to meet them and that both result in the
mobilization of energy to meet the environmental demands (Gaillard, 1993). Gaillard
(1993) suggests that they are distinct, however, in three ways. First, the energy released
to meet one’s workload is constructive and focuses attention, whereas the energy released
as a result of stress is destructive and distracting. Second, the energy activation caused by
workload is limited to the time of meeting the load; in contrast, the energy released by
stress tends to persist outside of the task situation in the form of generalized anxiety.
Third, workload causes attention to be focused on the task, whereas stress causes
attention to be focused on self-protection. Thus, although related, workload and job stress
are distinct constructs; sustained inordinately high or low workload may lead to job
stress, but so may many other factors, such as those discussed above under job demands
and job challenge.

Lines of workload research. Workload research has generally followed two
streams: one cognitive, and one motivational. The cognitive stream of research on
workload derives primarily from the theoretical work of Kahneman (1973) and Gopher
and colleagues (Gopher & Donchin, 1986). Kahneman’s (1973) theory of mental effort
suggests that people have limited attentional capacities and that they develop “allocation
policies” based on their dispositions, momentary intentions, evaluation of the demands on

their attention, and degree of arousal. This capacity model is structural in nature, in that it

10

holds that a single cognitive mechanism processes information, and that when individuals
have more than one possible stimulus to which they can attend, they must allocate their
attention to only one of the stimuli. The competing demands for attention cause
interference that exceeds individuals' available cognitive capacity. Although there is
some debate in the cognitive psychology literature over whether the interference limits
the perception of multiple stimuli or the processing of multiple streams of information
(as well as whether there are single or multiple channels for processing information), the
primary model of the theory has received considerable support (e. g., Kahneman, Ben-
Ishai, & Lotan, 1973; Ninio & Kahneman. 1974).

Gopher and colleagues (Gopher & Donchin, 1986; Gopher & Koriat, 1999)
explicitly integrated the concept of workload into attentional capacity theory. Indeed,
they discussed the “close affinity between the literature concerned with workload and the
literature that focuses on attention” (Gopher & Donchin, 1986, p. 4). The key element in
Gopher’s research is the “central processing mechanism” that determines where and how
attention is allocated, and which information is processed. This notion is also addressed
by Baddeley (2002), who referred to individuals’ “executive control,” where attention is
shifted either automatically or in a controlled fashion between cognitive tasks.

The basic message of these theories is that when individuals’ workload is within
their attentional and information processing capacity, they can perform adequately, but
when the central processing mechanism is overwhelmed by an inordinately high
workload. attention is divided and overall performance declines. Empirical research has
borne out this theoretical proposition in numerous arenas. Kahneman and colleagues

found that the divided attention caused by high cognitive load was significantly related to

II

the accident rate of bus drivers (Kahneman et al., 1973) and significantly reduced
reaction time and increased errors of omission in an experimental task (Ninio &
Kahneman. 1974). Aviation studies have consistently found that high cognitive demands
reduce pilot performance. sometimes with tragic results (Tsang & Vidulich, 2003).

The motivational stream of research derives primarily from the work on burnout
by Maslach and Jackson (1984). This theory is affective in nature and suggests that an
inordinately high workload (among other antecedents) leads to three components of
burnout: emotional exhaustion, depersonalization, and reduced personal accomplishment.
Like the theories of attentional capacity, burnout theory also suggests that performance
suffers under high workload, but differs in that it specifies the causal mechanism as being
an affective, rather than a cognitive, state. This theory has been empirically studied less
than attentional capacity theory, which arose during the cognitive revolution in
psychology. Some suggest that we are currently undergoing an “affective revolution” in
organizational behavior (Barsade, Brief, & Spataro, 2003), so it is likely that this theory
will receive more attention in the future. The theory has been supported, however, in a
handful of field studies with various samples like nurses (Greenglass et al., 2003), high
school principals (Somech & Miassy-Maljak, 2003) bank employees and teachers
(Houkes. Janssen, J onge, & Nijhuis, 2001), and social workers (Himle. Jayaratne, &
Thyness. 1991).

Long-term stress outcomes like burnout due to inordinately high workload,
however, develop gradually over a long period of time (Ganster, Fox, & Dwyer, 2001).
Thus, burnout theory does not lend itself to the types of laboratory studies that

characterize the empirical testing of attentional capacity theory. Kinicki, McKee. and

Wade (1996) also note that these types of long-term outcomes are seldom assessed by
organizational researchers, who tend to focus on stress outcomes that are more short—
term. Ganster et al. (2001) made perhaps the only attempt in the organizational literature
to draw a connection between short- and long-term stress outcomes when they showed
that cortisol levels (a hormone related to stress) mediated the relationship between job
demands and long-term health care costs.

This connection between short- and long-term outcomes of workload in the form
of job demands may actually provide a way to link the dominant cognitive and
motivational theories related to workload. I suggest that the elevated cortisol readings
that Ganster et a1. (2001) found in workers with high job demands were the result of a
more general experience of emotional arousal, and that extended periods of high
emotional arousal due to high workload lead to the experience of bumout.

Emotional Arousal

Emotional arousal refers to “how wide awake the organism is. of how ready it is
to react” (Berlyne, 1960). Arousal captures the “continuum from slumber to activity”
(Gopher & Donchin, 1986, p. 14). Schachter and Singer’s (1962) cognition-arousal
theory proposed that emotions are the product of two interrelated factors: physiological
arousal and cognitions about the causes of that arousal. In their view, arousal determined
the intensity of the emotion, while cognition determined the quality of that emotion. That
is, people may feel aroused, but attribute that arousal to a specific emotion (e. g., anger,
fear, excitement) based on their cognitions about what caused that arousal. Russell’s
(1980) and Larsen and Diener’s (1985) circumplexes of emotion expanded upon this

theory, suggesting that emotions can be placed in a two-dimensional space represented by

13

hedonic tone and arousal. Positive (pleasant) emotions can be ones of high arousal (e.g.,
joy, excitement) or low arousal (peace, contentment). Similarly, negative (unpleasant)
emotions can be of high arousal (fury, terror) or low arousal (sadness, despair). Early
research established a “general” arousal factor involving both the reticular ascending
activation system and the limbic system (Berlyne, 1964). Like most constructs in
psychological research. it is a hypothetical construct in the sense that it cannot be
measured directly, but only through its effects. Various studies have operationalized
arousal by means of heart rate, blood pressure, electrodermal activity, and self-report
measures.

Cognitive and social psychological research have extensively examined emotional
arousal in a variety of domains, but organizational research has largely neglected the
study of emotional arousal in favor of examining hedonic tone. This may be due in part to
the methodological difficulties associated with assessing arousal levels in field studies. as
well as being due to the factor analytic properties associated with the most commonly
used scale for assessing emotions (PANAS; Watson, Clark, & Tellegen, 1988), which
typically only shows two factors associated with positive and negative affect (the hedonic
tone dimension of the circumplex)'.

This neglect of arousal is unfortunate, because it figured prominently into both
Kahneman’s and Gopher’s treatments of attentional capacity theory. As noted above,
Kahneman (1973) suggested that arousal affects individuals’ attention allocation policies.

Indeed, Kahneman proposed that individuals’ “arousal and capacity both increase or

 

' It should be noted that the emotional adjectives used in the PAN AS tend to tap into aroused positive and
negative emotions. For example. positive affect is measured with such adjectives as “excited” and
“enthusiastic.” and negative affect is measured with such adjectives as "scared” and "irritable." Thus. the
PAN AS differentiates between aroused positive affect and aroused negative affect.

l4

decrease according to the changing demands of current activities” (p. 10). This is because
people tend to exert more effort when they are emotionally aroused, which allows them
to “pay more attention” to the task. In contrast, when people are in low states of arousal,
they exert less effort in their tasks and, according to Kahneman, “cannot work as hard”
(p. 14) as when they are more highly aroused. Thus, according to Kahneman, individuals’
attentional capacities actually vary according to their degree of arousal, which is in part
determined by the activities in which one is currently engaging. Although an individual’s
maximum potential capacity remains constant, their momentary capacity is determined
by their level of arousal. Gopher and Donchin (1986) echoed this notion, suggesting that
the relationship between workload and attention is in part determined by the energy
available to individuals as a result of their arousal level.

Drawing on these rich theoretical traditions, I suggest that emotional arousal
should have a positive linear relationship with workload; the higher the workload, the
more aroused should be the affective state of individuals. Higher workloads provide more
stimuli to which individuals can potentially direct their attention, which will naturally
heighten individuals’ arousal levels. As noted above in the discussion of the distinction
between workload and job stress, both result in the mobilization of energy in response to
environmental stimuli (Gaillard, 1993). In other words, both workload and job stress are
emotionally arousing, preparing the body for action; the difference is that workload
focuses attention on the task, whereas job stress focuses attention on self-protection.

The literature on job demands also supports this hypothesis. Karasek’s (1979) job
demands model suggests that the more demanding a job is. the more strain it places on

the individual, which is manifested in aroused emotions such as anxiety or nervousness.

15

He found support for this relationship in large samples in both the US. and Sweden.
Perrewe and Ganster (1989) found similar results when physiological indicators of
arousal (heart rate and skin temperature) were regressed on job demands. Although, as
noted above, job demands and workload are not identical constructs, they are sufficiently
close to suggest that their relationship with arousal may be similar. Indeed, direct effects
of objective quantitative workload on physiological arousal have been found in air traffic
controllers (Hurst & Rose, 1978; Rose & Fogg, I993), aircraft pilots and co-pilots
(Kakimoto et al., 1988), manufacturing employees (Parker & Sprigg. 1999), and driving
test examiners (Parkes. 1995).

H I: For individuals working alone. workload is positively related to emotional

arousal.
Individual Learning

Decades of both educational and psychological research have examined the
factors that influence Ieaming. both in terms of knowledge acquisition and transfer of that
knowledge (Mayer, 2004). A largely neglected area, however, has been the impact of
workload on Ieaming. This is not to say that the educational literature has entirely
ignored workload, however. A number of studies have tested the hypothesis that
workload was related to student evaluations of their instructors, generally finding that
workload and evaluations were positively related, contrary to the proposed hypothesis
(e.g., Marsh, 2001; Marsh & Roche, 2000).

Intuitively, workload is likely to affect Ieaming. and two theories have been
advanced that outline its effects. First, Karasek’s updated dynamic model of job demands

and control (Karasek & Theorell, 1990) suggests that the demands of a job affect not only

16

strain, but also the degree to which the employee learns. Specifically, their “active
learning hypothesis” suggests that jobs with high demands and high control lead to the
most learning, whereas jobs with low demands and low control lead to the least learning
(jobs with high demands/low control or low demands/high control lead to moderate levels
of learning). This is because learning requires that the situation not be “unchallengingly
simple (thus unimportant)” (p. 92). but rather be one that forces the individual to seek
new knowledge or test new hypotheses. In jobs where demands are low. individuals do
not have the opportunity to learn; such undemanding jobs may even be “associated with
the reverse process of skill atrophy and unleaming” (p. 94). Several empirical studies
have supported this hypothesis (Holman & Wall, 2002; Kauffeld, Jonas, & Frey, 2004;
Parker & Sprigg. 1999; Taris & Feij, 2004; Vahtera. Pentti, & Uutela, 1996). Again,
although job demands and workload are not identical constructs, their similarity warrants
a test of the effect of workload on learning.

Second, Kahneman’s (1973) theory of mental effort suggests that the effect of
workload on learning could be mediated by arousal. If attentional capacity and arousal
level covary as suggested by Kahneman and Gopher, higher workload should lead to both
higher emotional arousal and higher attentional capacity. Kahneman (1973) discussed the
notion of processing resources, which can determine whether a system fails to perform
under high demands. The extent to which individuals are emotionally aroused, then,
enhances their ability to process information. In their discussion of Kahneman’s theory,
Gopher and Donchin (1986) suggested that “the concept of resources is closely related to
the concept of with arousal” (p. 14, italics in original), and failures of individuals to

process information can be attributed at least in part to their arousal level.

17

This notion is not a new one. Berlyne’s (1964) review suggested, “Among the
internal factors that determine whether an organism is or is not in a state conducive to
learning are apparently those that are classifiable as ‘attentional,’ including level of
arousal” (p. 132). A certain level of arousal is a necessary (though not sufficient)
condition for the processing of information involved in learning to take place. But how
high must that arousal level be? The classic “Yerkes-Dodson law” (Yerkes & Dodson.
1908) suggested that there is an inverted U-shaped relationship between arousal and
performance, with a moderate arousal level leading to the highest level of performance.
Performance improves up to a point where the individual becomes overaroused, leading
to a decline in performance.

McGrath (1976), however, called this theory into question, noting that (a) most
studies have found no evidence of the downward slope of the curve at high levels of
arousal; (b) the theory is incomplete in its explanation of the downward slope, relying on
“the assumption of the right-hand side of the stress-performance curve” (p. 1360, italics
in original); and (c) the theory specifies the inverted U only within-persons, and thus we
cannot expect to detect it in between-person tests of the relationship. He proposed that
tests of this relationship have been confounded with task difficulty; that is, studies that
found performance decline at high levels of arousal did so only because the task was
more difficult at these levels. When task difficulty is kept constant or statistically
controlled, however, the relationship between arousal and performance is positive and
linear.

Similarly, LePine, LePine, and Jackson (2004) noted that although the inverted U

relationship has intuitive appeal, it has received very little empirical support. To clarify

18

past inconsistent findings, they decomposed arousal into two types of stress: challenge
stress—stress that is appraised as leading to future gains—and hindrance stress—stress
that is appraised as hindering future gains. In a study of learning performance. they
showed that challenge stress was linearly positively related to learning, whereas
hindrance stress was negatively related to Ieaming. Notably, for the purposes of this
study, they suggested that workload leads to challenge stress: “For example, learners who
experience stress associated with high workload and difficult learning content will exert
more energy trying to learn because they believe that by doing so they will eventually
come to understand and master the material” (p. 885). Interestingly, this mirrors the
discussion above on how workload is distinguished from stress; both lead to the
mobilization of energy, but only the energy mobilized by workload focuses attention on
the task. LePine et al.’s (2004) conceptualization of hindrance stress is roughly equivalent
to what Gaillard (1993) referred to generically as stress, which focuses attention on self-
protection. Their conceptualization of challenge stress, however, refers to the
constructive mobilization of energy that can arise due to an increased workload. Task
demands that exceed an individual’s capacity, however, could induce hindrance stress;
this sort of inordinately high workload would be more likely to mobilize energy toward
self-protection, rather than toward task performance or learning. Assuming the task
demands do not exceed the individual’s capacity, on the basis of this theory and empirical
research, workload should be positively related to learning, and this effect should be

mediated by emotional arousal.

l9

H2: For individuals working alone, workload is positively related to Ieaming.]

H3: For individuals working alone. emotional arousal mediates the effect of

workload on learning.

To this point, I have only considered the effects of workload in static conditions;
now I turn to consideration of how changes in workload affect performance. Although
workloads that are inordinately high may hamper individual Ieaming. it is not clear
whether decreasing the workload will enable an individual to achieve higher performance
or whether there will be residual effects of the previous workload. Conversely, the effects
of increasing an individual’s workload are not entirely clear, and may differ from
maintaining a consistently high workload.

Several studies suggest that increases or decreases in an individual’s workload
may indeed affect their task performance. In a laboratory study that examined changes in
workload at the task level, Campbell and [1 gen (1976) administered chess problems
where participants had to make decisions of one, two, or three moves, and then
subsequently presented participants with all three types of problems. Those who had just
completed two- or three-move problems performed significantly better than participants
who had completed one-move problems. Similarly, Taylor (1981) found that high prior
task challenge was positively related to performance standards, task-related attitudes, and
perceptions of skill competence. These studies suggest that there are carryover effects of
previous job experiences (i.e., degree of previous challenge) on future performance.
Notably, both of these studies examined changes in workload over a very short period of

time (the duration of a laboratory session), but the phenomenon appears to apply not only

 

3 As noted above. individuals do indeed have maximum potential capacities. and thus reach a limit where
increases in workload cannot bring about any higher increases in learning. In most work settings (and in
this research). this extreme high end of workload is not examined.

20

to the task level, but to the job level as well. Studies on job challenge have found that
having high initial job challenge in one’s career resulted in higher salary growth and
performance (Berlew & Hall, 1966; Orpen, 1974, 1994), and job learning (Morrison &
Brantner, 1992) many years later. Thus, workload may affect later performance and other
outcomes over both short and long time periods, at both the task and job levels.

Neither attentional capacity theory nor burnout theory specifically address how
changes in workload should affect performance. but perhaps I can extrapolate from them
based on the results of the empirical studies cited above. A long tradition of research has
demonstrated that performance is not a unidimensional construct. but rather consists of at
least two dimensions. Drawing on a theory first articulated by Woodworth (1899) over
100 years ago, Beersma et a1. (2003) and Johnson et al. (2006) showed that task
performance is often best conceptualized along the dimensions of speed and accuracy.
This roughly parallels the quantity/quality distinction made by others (e.g., Erez, 1990;
Gilliland & Landis, 1992). Speed represents the rate at which people accomplish their
tasks, but can also be measured by total output (i.e., faster individuals produce more than
slower individuals). Accuracy represents the number of errors people make in the
performance of their tasks, and is often conceptualized in terms of quality.

I first consider the effects of previous workload on speed of performance. The
well-known “fight or flight” response to threat has often been viewed as a vestigial
evolutionary adaptation in humans that is maladaptive in modern settings (Carruthers,
1981), but decades of research on both animals and humans have found that
environmental stressors increase performance on a variety of tasks (Dienstbier, 1989).

This research has outlined the physiological and neurological mechanisms responsible for

the improved performance. Although these specific mechanisms are beyond the scope of
this research, the general point is that environmental stressors serve to “toughen” people
(and animals) so that they handle future stressors more readily. Dienstbier’s (1989) model
suggests that this is similar to the way that aerobic exercise increases body tolerance to
physical stress; previous stressors act like “workouts” that toughen people and allow
them to act more quickly and with more endurance when facing future stressors.
Dienstbier (1989) also suggests that early experiences are vital in this process; although
his neurological model addresses this in terms of aging, it may be that early experiences
within a given context may have similar effects. That is, early difficult experiences of
individuals in a job inure them to later experiences in that job.

Another line of research suggests that environmental stressors can actually speed
up the functioning of the body’s internal clock. Based on the idea that humans possess a
“temporal pacemaker” that regulates individuals’ perceptions and speed of activity, this
research has shown that various stimuli cause changes in one’s subjective perception of
time (Treisman, Cook. Naish, & MacCrone, 1994). In a series of four studies, Penton-
Voak, Edwards, Percival, and Wearden (1996) found that a mild environmental stimulus
(audible clicks) caused individuals to overestimate the duration of subsequent gaps
between stimuli (a series of tones independent from the clicks) by an average of 10%
(and as high as 19%). Changes in body temperature have shown similar results, with rate
of subjective time increasing by as much as 30% (Wearden & Penton-Voak, 1995). This
phenomenon may underlie the perception of individuals under great stress that time slows

down, causing them to perceive the events as happening in slow motion.

IQ
Is)

Taken together, these two lines of research suggest that the emotional arousal that
accompanies higher workloads may actually increase individuals’ subsequent speed of
performance. First. based on Dienstbier’s (1989) research. the higher previous workload
could act as a form of exercise that allows individuals to respond more quickly in the
future. and will increase their stamina to maintain higher levels of performance over time.
Second, based on Treisman’s temporal pacemaker notion, individuals with higher
workloads should be likely to be faster in their subsequent performance due to the
speeding up individuals’ internal clocks. The magnitude of these effects is likely to be
commensurate with both the previous workload itself, and with the duration of time that
the individual worked under that workload. That is, very high previous workloads are
likely to have greater effects on subsequent speed than only moderately high workloads.
Similarly, operating under high workloads for longer periods of time is likely to exert an
effect on speed for a longer subsequent period of time. For example, an individual who
works under a high workload for only a short period of time will likely not have long-
term improvement on speed.

H4: For individuals working alone, higher previous workload is associated with

better subsequent speed of perfomtance.

This hypothesis does not assume, however, that emotional arousal carries over
into subsequent performance situations. Indeed, as noted above in the discussion on the
distinction between stress and workload, the energy that is mobilized in response to
workload dissipates after the task demand is met (Gaillard, 1993). Arousal is a transient
state that comes and goes based on situational characteristics (Kahneman. 1973), and thus

should only be affected by current workload. The effects of arousal. however, may be

more permanent. First, as noted above. arousal is associated with the speed of
individuals’ temporal pacemakers. Highly aroused people tend to underestimate the
amount of time that has passed, and this is associated with a faster pace of behavior
(Brown & Boltz, 2002). Similarly. research on entrainment theory (McGrath, Kelly, &
Machatka. 1984) has shown that people’s rhythms of behavior can be set by external
pacers, and furthermore, these rhythms can be relatively enduring even after the external
pacers are no longer present. Thus. the speed with which people perform tasks because of
the way their levels of arousal affect their temporal pacemakers is likely to carry over,
even when the workload level has changed. Second. the “toughening” of individuals
associated with the increased levels of arousal under conditions of high workload should
also carry over to subsequent performance situations, in the same way that physical
exercise strengthens the body for future physical tasks. In other words, individuals who
have been exposed to high levels of workload should handle subsequent workloads more
quickly and efficiently than individuals who have been exposed to low levels of
workload. Thus, individuals’ previous workload affects previous levels of arousal, which
in turn affect subsequent speed of performance.

H5: For individuals working alone. the relationship between previous workload

and subsequent speed is mediated by previous arousal.

Furthermore, previous workload should be associated with higher subsequent
accuracy in individuals. Campbell and Ilgen (1976) explained their findings from the
chess study on the basis of task skills; players who received the more difficult problems
first had presumably developed more skills than those who had received the easy

problems. Similarly. Monison and Brantner’s (1992) study of naval department heads

24

found that the primary effect of prior job challenge was on learning knowledge and skills
necessary to perform in future jobs. This is presumably because as I proposed above,
people learn more under higher workloads than under lower workloads. As noted above,
arousal is associated higher attentional capacity, which allows individuals to learn more.
Although arousal and attentional capacity are both transient states, learning is more
enduring. Learning from a previously high workload is likely to increase accuracy, then,
because the development of job-related knowledge and skills should lead individuals to
make fewer errors.

Although this explanation for why previous workload should be positively related
to subsequent accuracy focuses on cognitive factors, one must consider whether
motivational factors have a countervailing effect. Specifically, the concept of progressive
mastery (Bandura & J ourden, 1991) that derives from social cognitive theory (Bandura,
1986) may be relevant here. Social cognitive theory holds that people self-regulate their
effort based on their self-efficacy; when individuals believe that they are competent at
performing their tasks, they set higher goals (Locke & Latham, 1990) for their
performance than individuals who have less confidence in their ability to perform their
tasks. As a specific application of social cognitive theory, the concept of progressive
mastery suggests that individuals who progressively improve their performance (and
upgrade their self-efficacy beliefs) will end up performing better overall than those with
consistently high or low self-efficacy, and those who experience a decline in mastery.

Bandura and J ourden (1991) indeed found this effect. In an experimental setting,
they manipulated feedback provided to the participants, such that some were given

feedback that they consistently outperformed the other participants (superior

capabilities); others were given feedback that they had consistently achieved average
performance compared to the other participants (similar capabilities); others were given
feedback where they were initially told that they were underperforrning, then that they
had achieved average performance, and finally that they had outperformed the other
participants (progressive mastery); and others were given feedback that they initially
outperformed the other participants, then that they had achieved average performance,
and finally that they underperforrned compared to the other participants (progressive
decline). Those in the progressive mastery condition achieved higher overall performance
than those in the other three conditions.

When applied to the concept of changing workloads. an initial interpretation of
the phenomenon of progressive mastery could be that one should give people lighter
workloads when they are initially learning a task, in order to allow them to improve their
self-efficacy on their tasks. Later, one increases the workload when the individual has
higher self-efficacy beliefs. This may, however, result in a condition more similar to the
“superior capabilities" condition from Bandura and lourden (1991): the individual may
consistently maintain high self-efficacy beliefs over time. which was found to be
associated with lower performance than those in the progressive mastery condition. In
contrast, assigning high workloads early on and then reducing them later is likely to
mimic the progressive mastery condition; individuals are likely to have low perceptions
of self-efficacy under conditions of high workload, perhaps because they believe that the
reason they cannot keep up with the workload demands is due to their low competence in
the task. When workload is reduced, their self-efficacy beliefs are likely to increase, as

they now feel capable of handling their current workload. Thus. the motivational effect of

26

progressive mastery when shifting from conditions of high workload to low workload is
likely to enhance the cognitive effect of task learning on subsequent accuracy.
H6: For individuals working alone, higher previous workload is associated with
better subsequent accuracy of performance.
H7: For individtutls working alone. the relationship between previous workload

and subsequent accuracy is mediated by previous learning. ‘

TEAM WORKLOAD AND PERFORMANCE

So far, this discussion has only considered the effect of workload on individual
cognition, affect, and performance. A key question of multi-level theory (Kozlowski &
Klein, 2000) is one of homology, or. “Do the relationships found at one level of analysis
generalize to other levels?“ In this case. then. the question may be whether teams or even
higher levels of analysis (e. g., departments, organizations) also perform better overall
when they have high previous workloads.

Examining phenomena at higher levels of analysis (e.g., teams), however,
introduces greater levels of complexity due to the social relationships that are inherent
within them. Individuals working alone do not need to coordinate their actions with
others as much as those working in teams (Ilgen et al., 2005). Aside from the difficulties
of coordinating their behaviors, teams must also coordinate their thoughts about their
tasks. As Moreland. Argote, and Krishnan (1996, p. 58) note, “Information processing by
groups requires socially shared cognition, that is. collaboration among members who
seek to encode, interpret, and recall information together rather than apart.” Additionally,
recent research has suggested that team members also coordinate their emotional states
with each other (i.e., emotional contagion; Barsade, 2002; Bartel & Saavedra, 2000).
Thus, it appears that at the team level, as at the individual level, there are both cognitive
and motivational approaches to the effects of workload on performance. The following
section reviews the literature relevant to specific hypotheses regarding the effects of
previous workload on teams.

Collective Arousal and Emotional Contagion

One approach to the study of workload at the team level that is motivational in
nature is the theory of affect regulation (George, 2002). According to this theory, team
members’ emotions and moods are affected by their membership in the team, such that
teams tend to converge in their affective states; these affective states in turn influence
team member behavior. George (2002) suggests that affect is transmitted throughout
teams in three ways. all of which appear to have relevance to the study of workload
history.

F irst, a great deal of research has documented “emotional contagion” in groups,
mobs, and crowds. Dating back at least as far as Le Bon (1895), social scientists have
recognized that emotions can be “caught” from other group members in a quite primitive
way. The general thinking on emotional contagion is that it is an automatic process that
accompanies the deindividuation of members as they are absorbed into the group (Diener,
Lusk, DeFour, & Flax, 1980). Members naturally mimic the emotional expressions of
other group members, leading them to experience the emotions themselves. This effect
has been documented not just in mob settings, but also in less affectively intense
contexts, including workgroups (Barsade, 2002; Bartel & Saavedra, 2000).

Second, emotions may be transmitted through exposure to common tasks,
outcomes, and events. Rather than emotions being passed from member to member, in
this mechanism the emotions are generated in all of the team members by an outside
source. For example, it a team is engaging in tedious work, feelings of boredom may
develop throughout the entire team as a result of the work itself. This is also the primary

thrust of Affective Events Theory (Weiss & Cropanzano. 1996): work-related events

stimulate certain affective states, which in turn influence work-related judgments, affect-
driven behaviors, and judgment-driven behaviors.

Third, emotions may be transmitted through vicarious processes. In this mode of
transmission, team members observe the emotional expressions of their team members
which can directly stimulate the same emotions in themselves through empathic
processes. Although this process is similar to the more primitive emotional contagion
process, vicarious transmission is usually more consciously engaged. George (2002)
suggests that they are particularly distinct when it comes to the transmission of negative
emotions. When negative emotions are transmitted through emotional contagion, team
members are motivated to alleviate their own negative states; when negative emotions are
transmitted through vicarious empathic processes, however, team members are motivated
to alleviate the other’s negative state.

This theory fits within the broader Input-Process-Output (I-P-O) model of group
emotion outlined by Kelly and Barsade (2001 ). In their model, both affective context
(e. g., organizational emotion norms, group emotion norms, group emotional history) and
nonaffective context (intergroup relations, physical environment, technological
conditions) act as inputs that impact both individual-level and group-level emotional
processes. Like George (2002), Kelly and Barsade (2001) suggest that emotions converge
within groups via processes like the ones mentioned above, giving rise to “group
emotions (that) are phenomenologically experienced as real by group members” (p. 117).
If group emotions are legitimate constructs, then, the question arises as to how similar
they are to individual emotions. According to Morgeson and Hofmann (1999), collective

constructs can be compared to individual-level constructs in terms of structure and

30

function. If they are structurally similar. then the construction of the phenomenon is the
same at both levels; if they are functionally similar. then the outcomes or effects of the
phenomenon are the same at both levels. Group emotion, like individual emotion. is
experienced intrapersonally by the group members; its transmission, however. is subject
to the interpersonal dynamics outlined above. Thus. group emotion is to some extent
structurally dissimilar to individual emotion. Functionally, however. group emotion is
likely to cause the same effects as individual emotion. Groups experiencing positive or
negative emotions are likely to generate the same types of behavioral outcomes as
individuals experiencing positive or negative emotions, as are groups experiencing
varying levels of arousal likely to generate similar behaviors to individuals experiencing
varying levels of arousal.

In terms of workload, the affective states generated from having varying levels of
workload are likely to converge among team members via the processes outlined by
George (2002) and Kelly and Barsade (2001). Team members may feel similar emotions
as a result of their shared experiences, which converge further through primitive
emotional contagion and vicarious empathy. Specifically, team members may experience
similar levels of arousals due to high workloads, and these may converge further as they
transmit this emotional experience to other members. Similarly, team members who
experience boredom due to low workloads may transmit a different form of negative
affect to other members. Thus, the relationship between workload and emotional arousal
in teams should be similar to that in individuals; if the relationship differs at all, it should
be stronger in teams, because team members’ emotional states are likely to converge.

H8: For teams. workload is positively related to emotional arousal.

31

Similarly, the relationship between previous workload in teams and their
subsequent speed is likely to take the same form as that found in individuals, because the
constructs are functionally similar (Morgeson & Hofmann. 1999). That is, although
individual and team workload differ in structure, they are likely to lead to a faster pace of
behavior at both levels. The entrainment effect discussed above should affect teams (and
other higher levels of analysis) in the same way that it affects individuals (Aneona &
Chong, 1996). Indeed, this relationship may be stronger than that found in individuals,
because teams develop norms of behavior (Bettenhausen & Mumighan, 1985) that are
often mutually enforced (Barker, 1993). Thus, the pace of work that teams develop in
their early experiences as a result of their previous emotional arousal becomes the teams’
expectations for future behavior. When teams begin working at a fast pace due to high
workload and high emotional arousal, this will carry over to their subsequent
performance, even if their workload decreases. Conversely, teams that begin working
slowly due to low previous workload are likely to struggle to speed up when their
workload increases.

H9: For teams. higher previous workload is associated with better subsequent

speed of performance.

H10: For teams. previous emotional arousal mediates the relationship between

previous workload and subsequent speed.
Team Learning

The cognitive approach of the study of group phenomena derives from the social
cognition movement (Fiore & Schooler, 2004), in the sense that it examines “those social

processes... that relate to the acquisition, storage, transmission. manipulation and use of

32

information for the purpose of creating a group-level intellective product” (Larson &
Christensen, 1993, p. 6). This theoretical approach has outlined the effects of both shared
and distributed cognition on the process and performance of work teams.

Interest in team-level information processing and learning has burgeoned in recent
years and several theoretical models have appeared (Argote, Gruenfeld, & Naquin, 2001 ).
Although a consensus model of team learning has not yet emerged, the theoretical and
empirical treatments of the subject appear to agree that team learning is qualitatively
different from individual Ieaming. Interestingly, in outlining the unique characteristics of
team-level learning, these conceptualizations draw upon individual-level models as a
basis for understanding the similarities and differences between learning at each level
(Hinsz, Tindale, & Vollrath, 1997). Unfortunately, because these models are based on
individual-level cognition, they tend to ignore the social factors that affect learning at the
team level. Argote, Gruenfeld, and Naquin (2001) suggested that when thinking about
team learning, one must consider these social factors, examining team learning from a
systems perspective. The actions of team members affect the team system whether or not
they intend to, which can either help or inhibit team Ieaming. In keeping with
conceptualization. I suggest that team leaming—like team—level emotional arousal—has
both intrapersonal and interpersonal elements, and thus differs in structure from
individual-level cognition (Morgeson & Hofmann, 1999). As noted in Ellis et al.’s (2003,
p. 822) empirical examination of team learning, “...teams can process information not
only within. but also between the minds of team members.” Unless information is

transmitted between team members. true team-level learning does not take place.

33

Groups inevitably possess more knowledge together than any one team member
does individually and thus possess an advantage over individuals (Argote et al., 2001).
Indeed. in many cognitive activities, research has shown that teams perform better than
individuals working alone. Teams tend to recall more information than individuals
working alone (Hill, 1982; Hinsz, 1990). and tend to be more accurate in the information
they recall (Hill, 1982). Teams also tend to perform better than individuals on induction
tasks, where the goal is to find general principles or explanations (Laughlin,
VanderStoep. & Hollingshead, 1991), and on general problem solving (Laughlin, Bonner,
& Miner. 2002). As noted by Argote, Gruenfeld, and Naquin (2001, p. 377), however, the
advantage that teams have over individuals “is dependent upon the degree to which the
knowledge of individuals within the group can be effectively shared, or pooled. during
group discussion.”

Hinsz, Tindale, and Vollrath’s (1997) review also noted the importance of
communication—in terms of sharing information, ideas. and cognitive processes—to
team level Ieaming. In order to effectively learn at the team level, team members must
communicate their ideas to the rest of the team. MacMillan, Entin. and Serfaty (2004, p.
61) call this “the hidden cost of team cognition,” noting that communication between
team members has no analogue at the individual level. Similarly, Larson and Christenson
(1993, p. 12) note:

Groups too must cope with the reality of limited cognitive resources—of

the social as well as the individual variety. Consider, for example, the time

and effort that group members have to devote to problem solving. These

34

are limited resources that. like attention, can be channeled in one direction

or the other for the purpose of acquiring problem-relevant information.

Teams, like individuals, do not possess unlimited cognitive capacity, and their
limited sets of social cognitive resources may be allocated to the task or to other factors
that concern the team. In other words, team may choose to focus their cognitive resources
on taskwork—their interactions with equipment, tools or systems—or on teamwork—
their interactions with each other (Morgan, Salas, & Glickman, 1993). Although some
teamwork is simply directed at the maintenance of the team as a social system (e. g.,
maintaining social order, providing social support), other teamwork serves to facilitate
future taskwork (e. g., sharing information. observing others in order to better coordinate
actions). In most situations, teams often switch their focus between these two, alternating
between focusing on performing their task and attending to other team members.

In effect, then, working on a team is a dual-task paradigm where team members
must do both taskwork and teamwork (Bowers et al., 1997). The dual-task paradigm has
been utilized extensively in the workload literature, and it involves assigning individuals
two tasks to perform simultaneously. In general, research has revealed that performance
declines when individuals must switch their attention between two tasks. even when they
are proficient in both; Navon and Gopher (1979) referred to this as “concurrence cost.”
Bowers, Braun, and Morgan (1997) suggested that when team members must constantly
switch their attention between taskwork and teamwork. their performance in one or the
other will decline.

The degree to which this decline in performance occurs is almost certainly

dependent upon the team’s level of workload. As task workload increases, the social

35

cognitive resources of teams are increasingly taxed. inevitably resulting in a decline in
performance on teamwork. Conversely. as workload decreases, teams have a surfeit of
social cognitive resources and can reallocate more resources to teamwork. A wealth of
empirical research supports this assertion. Bowers et al. (1998) reported the results of a
study that found that flight crews in automated cockpits communicated more than flight
crews in traditional cockpits. The explanation for this finding was that in the automated
cockpit, crew members were able to devote more resources to communication, whereas in
the traditional cockpit, crew members allocated those resources to performing their tasks.
Entin and Serfaty (1999) found that teams that were given training on efficient
communication strategies performed significantly better under high workload conditions
than teams without such training. Similarly, Cannon-Bowers, Salas, Blickensderfer. &
Bowers (1998) found that teams that were cross-trained performed better under high
workload conditions than teams that were not cross-trained, because of “the more
efficient communication strategy (i.e., volunteering more information) observed in cross-
trained teams” (p. 99). Urban. Weaver, Bowers, & Rhodenizer (1996) found that teams
under high workload communicated less and performed worse than teams under low
workload.

These findings appear to hold not just for workload in general, but also for
information load. Stasser and Titus (1987) found that only groups with low information
load communicated unshared information with the group; groups with a high load of
information tended to discuss only that information which was shared by all group
members. Thus, despite the fact that high workload conditions provide more information

to which team members can attend. they tend to limit their discussions to information that

is already shared by other members of the team. rather than communicating information
they hold uniquely.

As workload increases. then. teams have less time, effort, and attentional
resources to devote to communicating with each other, and teamwork suffers. The effect,
then. is that unlike at the individual level—where there is likely to be a positive
relationship between workload and Ieaming—the team level relationship between
workload and learning should be attenuated, and may in fact be negative.

H11: For teams. the relationship between workload and learning is lower than
the relationship for individuals working alone.

As with leaming at the individual level. team learning is likely to have a carryover
effect on the team’s subsequent performance. Kozlowski, Gully, Nason. and Smith’s
(1999) model of team development suggests that there is a reciprocal causal relationship
between team processes and performance, such that team processes at time tn affect team
performance at time tn, and that this performance then becomes an input to team process
at time tn... Additionally, there is likely also a direct relationship between processes at
time tn and processes at time tn“, such that the team’s interaction behaviors tend to
become institutionalized. In the case of the current research. it is likely that the degree to
which teams learn will affect not simply their current performance. but also their
subsequent learning and performance.

As at the individual level, accuracy is the most likely dimension of performance
to be affected by team Ieaming. Although it is possible that the lack of learning will also
slow teams down due to difficulties in coordination, the degree to which the team makes

errors is more likely to be affected. This is because the extent to which members share

37

what they have learned tends to reduce the number of errors made by the team
(Edmondson, 1996). Taken together. I expect:
H12: For teams. higher previous workload is associated with lower subsequent
accuracy.
H13: Team learning mediates the relations/zip between previous workload and

subsequent accuracy.

38

MODERATORS OF THE EFFECTS OF WORKLOAD

To this point, I have only considered one influence on arousal, learning, and
performance—that of the individual or team workload. In work settings, however.
individuals and teams have multiple influences on their behavior, including
characteristics employees bring to the job and aspects of their work environment. In this
section, I propose that: (a) individual differences between employees—specifically,
personality characteristics—act as moderators of the effect of workload at both the
individual and team levels; and (b) structural characteristics—in terms of resource
allocation and learning responsibilities, which exist only at levels higher than the
individual—act as moderators of the effects of workload at the team level.

Personality

The study of personality has a long history in many areas of psychology, and has
greatly informed research in organizational settings. The “Big Five” model of personality
factors (Costa, Busch, Zonderman. & McCrae, 1986) has been a particularly fruitful
avenue for assessing the effects of individual differences on various outcomes. In terms
of personality effects on arousal and learning (and ultimately performance), both theory
and prior empirical research suggest that extraversion and neuroticism. respectively.
should show interactive effects with workload.

Extraversion. Along with Agreeableness, Extraversion has been considered to be
an interpersonal personality trait; that is, it manifests itself in social situations (Wiggins
& Pincus, 1992). In contrast, the other three traits in the five-factor model of
personality—conscientiousness, neuroticism. and openness to experience—are generally

considered to be intrapsychic; that is, their manifestations are not limited to social

39

situations, but are also manifested when one is alone. This solely interpersonal
characterization of extraversion. however, ignores a great deal of the rich theory and
empirical work on this trait. much of which was focused on how Extraversion relates to
emotional arousal.

Eysenck’s (I967) three-factor theory of personality focused on the traits of
extraversion, neuroticism. and psychoticism (which, according to Eysenck. is at least
partially comprised of agreeableness and conscientiousness). More than anything else,
Eysenck’s theory is one of arousal, as he delineates how each of these traits is related to
different arousal mechanisms. Extraversion, he proposed, is related to arousal through the
reticulo-cortical system; neuroticism is related to arousal through the reticulo-limbic
system; and psychoticism is related to arousal either through serotonin or dopamine
(Eysenck. 1997). Although it is not the purpose of this paper to delve into
neuropsychology. the point is that extraversion was proposed to be directly related to
cortical arousal. The theory suggested that people high on extraversion seek external
stimulation because they have lower baseline levels of arousal than those low on
extraversion.

Empirical research. however, has generally not borne out this main effect theory
of the effects of extraversion on arousal. Instead, tests of Eysenck’s theory show that
rather exerting a main effect, extraversion tends to show interactive effects with the
arousing effects of the experimental task itself. For example. in his review of the
literature on electroencephalographic (EEG) activity, Gale (1973) concluded that the
effects of extraversion depended upon the arousal inducing properties of the testing

environment. In testing environments that did not induce much arousal. extraversion

40

exerted no effect on EEG readings; in testing environments that induced moderate
arousal, however, those high on extraversion showed higher levels of EEG activity than
those low on extraversion. Similarly, Mangan and Hookway (1988) found that those high
in extraversion showed higher heart rates and electrodermal activity when viewing
aversive film clips than those low in extraversion (but not when viewing neutral clips).

The picture that seems to be emerging is one of a person by situation interaction;
in situations that are arousal inducing, people high in extraversion tend to become more
aroused than those low in extraversion. In situations that are not very arousal inducing,
extraversion appears to exert no effect. In the context of this study, then, I would expect
extraversion to interact with workload, such that under low workload conditions,
extraversion is unrelated to arousal, but under high workload conditions, extraversion and
arousal are positively related.

This effect is likely to be homologous across levels. As Hofmann and Jones
(2005) noted, the concept of personality refers to regularities in behavior; individuals
tend to engage in routines or habits that are in keeping with their personality traits. They
note that these sorts of behavioral regularities are just as observable in groups as they are
in individuals, giving rise to “collective personality.” They also note that collective
personality is functionally isomorphic with individual personality; that is, individual and
collective personality produce similar outcomes for individuals and teams, respectively.
Although Hofmann and Jones (2005) suggest that collective personality is an emergent
property that arises out of extended group interaction, a great deal of other research has
shown that simply aggregating the individual personalities of group members is also a

meaningful conceptualization of collective personality (e.g., Barrick, Stewart. Neubert, &

41

Mount. 1998; LePine, 2003; LePine, Hollenbeck, Ilgen, & Hedlund, 1997). These studies
have shown that teams exhibit behaviors in keeping with the mean level of personality
traits of the team members. Because the current research applies to teams that have not
had extensive interaction (see the boundary conditions noted above). this latter
compositional model is more appropriate. Thus, I suggest that team extraversion.
assessed as the aggregation of individual team member Extraversion. will manifest the
same types of behaviors as individual extraversion. The moderating effect of extraversion
on the workload-arousal relationship, then, should be similar across levels of analysis.
H14: For both individuals and teams. extraversion moderates the eflect of
workload on arousal. such that the highest levels of arousal will befound in
individuals and teams high in extraversion under conditions of high workload.
Neuroticism. Besides extraversion, neuroticism is the second personality trait that
appears in both Eysenck’s (1967) three-factor model and the currently popular five-factor
model of personality (Costa et al., 1986). Although the term was popularized by Carl
Jung, the concept had been considered at least as far back as the second century A.D..
when the Greek physician Galen described the “four temperaments,” with melancholics
being described as “anxious” and “worried,” and cholerics being described as “quickly
roused” and “histrionic” (Eysenck. 1967, p. 35). Current descriptions of people who are
high on neuroticism (or low on emotional stability) include adjectives such as frustrated,
depressed, hopeless, and anxious (Barrick & Mount, 1991; Costa & McCrae, 1992).
Neuroticism is relevant to discussions of learning because of its anxiety
component. M.W. Eysenck’s (1981) theory of anxiety draws on attentional capacity

theory in suggesting that the task-irrelevant information generated by anxiety competes

with task-relevant information for cognitive attention. When individuals are in a state of
anxiety, they use up cognitive resources by attending to the anxiety-provoking stimuli
and their coping mechanisms for dealing with their anxiety. Consequently, fewer
cognitive resources are available for processing task-relevant information. Because
neuroticism is associated with higher reactivity to anxiety-producing stimuli (Eysenck.
1967), this should hamper the performance of people high in neuroticism under
conditions of high workload.

Empirically, a few studies lend support to this notion at the individual level.
Jensen (as discussed in Eysenck. 1967) found that learning was significantly predicted by
an interaction between neuroticism and the rate at which the stimulus was presented.
When the stimulus was presented at a relatively fast rate, those high in neuroticism made
significantly more errors than those low in neuroticism. Presumably, the cognitive load
associated with the rapid pace of presentation was anxiety-producing, and overtaxed the
cognitive capacity of the neurotic participants. Interestingly, when the stimulus was
presented at a relatively slow rate, those high in neuroticism made somewhat fewer
errors. Similarly, Wallace and Newman (1998) found that women who were high in
neuroticism were more impaired by a distractor stimulus than those who were low in
neuroticism.

I expect, then, that neuroticism moderates the relationship between workload and
Ieaming. Although the proposed main effect of workload on learning is different at the
individual (positive) and team (negative) levels, the moderating effect of neuroticism on
the relationship should be similar. That is. because anxiety-producing stimuli overtax the

cognitive resources of people who are high in neuroticism, individuals and teams that are

43

high in neuroticism should learn less under conditions of high workload. For individuals,
this should attenuate the positive correlation between workload and leaming; for teams,
this should strengthen the negative correlation between workload and Ieaming.

H15: For both individuals and teams. neuroticism moderates the effect of

workload on learning. such that individuals and teams high in neuroticism learn

less under conditions of high workload than those low in neuroticism.
Structure

Because the effects of workload on teams involve interpersonal as well as
individual dynamics. it is likely that the structure of the team exerts effects on the
mediating processes discussed above and on team performance. Team structure refers to
how the actions of the team members are formally coordinated (Hollenbeck et al., 2002).
It can include horizontal elements—how tasks or resources are distributed within the
team—or vertical elements—the team’s hierarchical leadership or authority structure. In
this study, I focus on the horizontal dimension, and suggest that the distribution of
resources within the team should affect the team’s emotional arousal, and distribution of
responsibilities within the team should affect team Ieaming.

Resource allocation. Past research on the arousal hypothesis in the social
facilitation literature has established that the mere presence of others increases arousal,
particularly when one senses that one is being evaluated by others (Geen, 1991; Zajonc,
1965). One might expect, then, that individuals working in teams would show higher
levels of arousal than individuals working alone. Subsequent research by Jackson and
colleagues, however, has shown that being in groups can actually reduce arousal. Jackson

and Latane (1981) found that when they had their participants perform on a stage. those

who performed as part of the group felt significantly less stage fright—in the form of
tension and nervousness—than those performing alone. Similarly, Jackson and Williams
(1985) showed that people felt less pressure when working on an experimental task in
groups rather than working alone. The causal mechanism underlying this effect is that
people feel safer in groups if the other members of the group are facing the same threat or
challenge. Interestingly, in a direct test of this effect versus the social facilitation effect,
Jackson and Williams (1985) found that people felt the most pressure when they worked
with just one other person—heightening the sense that they may be evaluated by their
partner. as predicted by social facilitation theory, but the least when working in a group,
as predicted by arousal reduction theory (Seta & Schkade, 1976).

Small group research has shown that groups vary in their entitativity, or the
degree to which they constitute a group that is recognized by its members and outsiders
(Campbell, 1958). For example, a group of people waiting in line for a bus is lower in
entitativity than a formal work team. Similarly, research on team structure has
demonstrated that teams vary in the degree to which they work together or separately.
Teams that are structured functionally (rather than divisionally). are centralized in terms
of decision-making authority. and/or are rewarded cooperatively engage in more
interdependent behaviors (e. g., helping, information sharing) than teams that have the
opposite structures (Johnson et al., in press). In essence, groups that are structured these
ways are more interdependent and have a greater sense of possessing a common fate
(Kramer & Brewer. 1984). Although the research on arousal reduction cited above
compared individuals alone to individuals in groups. it may be that it applies to varying

types of team structure.

45

In particular, the horizontal dimension of structure—in terms of resource
allocation—is likely to affect the degree to which the arousal reduction effect applies.
Members of teams that are structured functionally—possessing unique resources and thus
having specialized areas of expertise—work more cooperatively with each other than
members of teams that are structured divisionally—possessing the same set of resources
as other members and being responsible for different areas of the team’s task. This is
because members of teams that are structured divisionally are not required to depend
upon the other team members to accomplish tasks, because they already possess the “full
complement of resources to handle whatever tasks they face. In contrast, members of
teams that are structured functionally must depend upon the other team members to
accomplish the parts of the team’s task that require resources that the individual team
member does not possess (Hollenbeck, 2000).

I suggest that members of functionally-structured teams, then, are more likely to
be subject to the arousal reduction effect than members of divisionally-structured teams.
The members will have a greater sense that accomplishment of the team’s task is jointly
shared by the other members, and this will provide a sense of “safety in groups.”
Members of divisionally-structured teams, however, are more like “individuals working
alone,” (Hollenbeck, 2000), and thus are less likely to be affected by the arousal
reduction effect.

H l 6: In teams. resource allocation structure moderates the relationship between

workload and arousal, such that the positive relationship between workload and

arousal is attenuated in functionally structured teams as compared to divisionally

structured teams.

46

Transactive learning. As noted above, recent research has examined the
importance of transactive memory for team performance. An unexamined notion,
however. is the idea of transactive learning; just as teams can distribute their knowledge
among team members. they also have the ability to distribute responsibility for learning
among team members. I suggest that transactive learning may be a way to alleviate the
problem of lower team accuracy following higher workloads. In their empirical study of
team Ieaming. Ellis et al. (2003) found that assigning members specialized roles within
the team enhanced Ieaming. Their explanation centered on how distributing cognitive
information processing within the team enhances team performance because it takes
better advantage of individual member cognition and enhances communication processes
within the team (Hyatt & Ruddy, 1997; Stasser. Stewart, & Wittenbaum. 1995; Stewart &
Stasser, 1995). Ellis et al. (2003) manipulated roles in terms of resource allocation. but it
is likely that simply assigning responsibility for learning different aspects of the task
could accomplish the same objective. By distributing responsibility for learning about
their task. teams reduce the cognitive load on the individual members because the
members are not attempting to learn all aspects of the task.

H17: 1n teams. transactive learning moderates the relationship between workload

and learning, such that the negative relationship between workload and learning

is attenuated when learning responsibilities are effectively distributed.

47

METHOD
Research Participants and Task

Four hundred forty-six undergraduate students in an upper-level management
course at a large Midwestem university participated; 364 were arrayed into 91 four-
person teams, and 82 participated as individuals working alone. The expected power for
finding a population r of .30 is .82 for the team sample size, and .78 for the individual
sample size (p < .05). Assuming main effects of this size and a two-way interaction
accounting for 5% incremental variance. the power to detect this interaction was over .60
at these sample sizes. Participants signed up for a research session at their discretion, and
were randomly assigned either to teams or to work as individuals within their session.
Teams and individuals were also randomly assigned to conditions in which they
participated in two 30-minute simulations. Students received course credit for
participation, and they also had a chance to win a cash prize contingent upon their
performance on the simulation.

Participants engaged in a dynamic and networked computer simulation. The
simulation is a modified version of the Distributed Dynamic Decision-making (DDD)
simulation developed for the Department of Defense for research and training (Miller.
Young, Kleinman, & Serfaty, I998). The version of the simulation used here was
developed for teams of four members with little or no military experience (MSU-DDD).

Team task. The teams’ mission in this simulation was to monitor air and ground
space of a hypothetical geographic region, keeping unfriendly forces from moving into
the restricted areas, while at the same time. allowing friendly forces to move about freely.

Radar representations of these forces moving through the geographic space monitored by

48

the team are known as “tracks.” In monitoring the geographic space, each team member’s
base had radar capacities that covered only a portion of the area that needed to be
monitored. Any track outside the radar range was invisible to the team members from
their base. If the participants wished to determine the nature of a track outside this ring,
they could launch a vehicle and move it near the track, or they could ask their teammates
to share that information with them.

Each team member had control of four vehicles that could be launched and moved
to different areas of the screen. In total, each team had control of four AWACS planes.
four tanks, four helicopters, and four jets. Each of these vehicles varies in its capacities
on four different dimensions: (a) range of vision. (b) speed of movement. (c) duration of
operability. and ((1) weapons capacity. The distribution of these vehicles varied according
to the resource allocation structure manipulation. as noted below.

There were eight types of “standard tracks” that were known a priori to have
specific characteristics, and these were taught in the training session prior to the start of
the simulation. There were also four types of novel tracks, or “unknown tracks” that were
not encountered during training. Thus, team members did not know whether the novel
tracks were friendly or unfriendly, or what power was required to disable them if they are
unfriendly. The team members were only able to learn the nature of these tracks via
deductive trial and error experience. The teams’ overall objective, then, was to disable
enemy tracks as quickly as possible if they enter the restricted airspace. while avoiding
the errors of disabling friendly tracks or wasting resources by attempting to engage tracks
with less power than was required. Each team participated in two sessions on this

simulation, which I will refer to as Game 1 and Game 2.

49

lmlividual task. Participants participating as individuals played a modified version
of the team task, such that they defended one quadrant of the restricted area. Their track
set was identical to the track set faced by individuals in one quadrant of the restricted
area, and they possessed the same array of vehicles as those in the team version. Thus,
the simulation was identical in all respects to the team game except one: they were
completely responsible for their performance on the task (whereas in the team version,
team members could pick up the slack for underperforrning members).

Manipulations and Measures

Research design. The design of the team portion of the study was a 5 x 2 x 2
factorial design, with five levels of initial workload, two levels of resource allocation
structure. and two levels of transactive learning, as outlined below. These manipulations
were fully crossed, creating twenty cells with 4-5 teams in each cell. The individual
portion of the study consisted of the five workload conditions only.

Workload. Workload was experimentally manipulated by adjusting the number of
tracks faced by the teams/individuals in Game 1. Five levels of workload were
introduced: (a) very high initial workload, where teams faced 140 tracks; (b) high initial
workload, where teams faced 120 tracks: (c) moderate initial workload, where teams
faced 100 tracks; low initial workload, where teams faced 80 tracks; and (e) very low
initial workload, where teams faced 60 tracks. In Game 2, all teams faced 100 tracks.
Those engaging in the individual task faced 35 tracks in the very high workload
condition. 30 tracks in the high workload condition. 25 tracks in the moderate workload
condition, 20 tracks in the low workload condition. and 15 tracks in the very low

workload condition. In Game 2. all individuals faced 25 tracks. Although this

50

manipulation created five distinct conditions. the increments between the conditions were
equal, and there is a meaningful zero point; therefore. this manipulation was analyzed as
a continuous variable.

As a manipulation check. perceptions of workload were measured with four items
from the NASA Task Load Index (NASA-TLX; Hart & Staveland. 1988). Coefficient
alpha was .73 for Game 1 and .84 for Game 2 in the individual portion of the study,
and.76 for Game 1 and .84 for Game 2 in the team portion of the study. The results from
this scale were compared with the objective levels of workload in Game 1. and the
pattern of correlations between the workload manipulation and subjective workload
varied between teams and individuals. These two variables showed a significant positive
correlation in Game 1 for individuals (.34, p < .01 ), but not for teams (.08, us). The
pattern in Game 2 was the opposite, however; they showed a significant negative
correlation for teams (-.28, p < .01), but not for individuals (.05, us). The relationships
found for individuals was expected, as workload was only manipulated in Game 1. For
the teams. however, it appears that the contrast between the objective workloads in
Games I and 2 drove their subjective perceptions of workload.

By employing both an objective manipulation and a subjective perception of
workload, I was able to get at workload from multiple perspectives. As noted above. I
view workload as a function of both the situation (the demands placed upon the
individual) and the person (an individual’s workload capacity). Because participants were
randomly assigned to workload conditions. workload capacities should be equal across
conditions. This means that the objective workload manipulation should still capture the

person/situation conceptualization of workload (albeit with noise from randomization).

51

The subjective perception of workload provided by the NASA-TLX gets at the
person/situation conceptualization more directly. but is subject to common method and
source bias when examined with other self-report measures.

Resource allocation structure. Two levels of resource allocation structure were
experimentally manipulated for the team portion of the study. Functionally structured
team members were allocated all of one type of vehicle (four tanks. fourjets, four
helicopters, or four AWACs). Divisionally structured team members were allocated one
of each vehicle type (each member had one tank. one jet. one helicopter, and one
AWACs).

Transactive learning. Two levels of transactive learning were experimentally
manipulated (treatment and control conditions) for the team portion of the study in Game
1. The treatment condition consisted of assigning each of the team members
responsibility for learning the power of one of the four unknown tracks (discussed below
under the learning measure). In the control condition, no such assignment was given.

Personality. Extraversion and neuroticism were measured with the NEO-PI-R.
which assesses these personality factors with twelve items each. Coefficient alpha for
extraversion was .86 in the individual portion of the study and .78 in the team portion of
the study. Coefficient alpha for neuroticism was .85 for the individual portion of the
study and .84 for the team portion.

Emotional arousal. Emotional arousal was measured in two ways. First.
participants self reported arousal through the Current Mood Questionnaire developed by
Feldman-Barrett and Russell (1998), which includes two statement measures and one

adjective measure each for both the arousal and hedonic tone dimensions of affective

states. The two statement measures are largely redundant, so only one was assessed. The
arousal statement measure includes seven items on which participants indicate the degree
to which they think the statement describes them: (a) “I’m full of energy and tension,” (b)
“I’m keyed up.” (c) “I am stirred up,” ((1) “I’m feeling placid, low in energy” (reverse
coded), (e) “My internal engine is running slow and smoothly” (reverse coded), (f) My
body is in a quiet. still state” (reverse coded), and (g) “My mind and body are resting,
near sleep” (reverse coded). Responses were given on a five-point scale from “Describes
me not at all,” to “Describes me very well.” The arousal adjective scale contains six
words on which participants indicate the degree to which they are feeling them at the
moment: (a) aroused, (b) alert. (c) activated, (d) sleepy, (6) still. and (f) quiet. Reponses
were given on a five-point scale from “Not at all” to “Extremely.”

These measures showed good convergent validity with a previously validated
semantic differential scale of emotional arousal from Russell and Mehrabian (1974), with
zero-order correlations of .82, and .69 respectively in one study, and .64, and .65 in
another. They also showed good discriminant validity from similar measures of hedonic
tone, with zero-order correlations ranging from .00 to .26 (mean r = .10) across the two
studies. These results have been replicated with similar findings across two samples (Yik,
Russell, & Barrett, 1999). Neither the original study nor the replication provided
reliability data in the form of coefficient alpha, but both showed high factor loadings on
their intended factors and excellent fit in their structural equation models. Participants
completed this questionnaire three times: when they first arrived at their laboratory
session and after both games. Coefficient alpha for the statements measure was .86 at

pregame. .90 after Game 1, and .92 after Game 2 in the individual portion of the study.

53

and .84 at pregame. .81 after Game 1. and .86 after Game 2 in the team portion.
Coefficient alpha for the adjectives scale was .65 at pregame, .85 after Game 1, and .80
after Game 2 for the individual portion of the study, and .70 at pregame, .74 after Game
1. and .73 after Game 2 for the team portion.

Second, arousal was assessed via the noninvasive Nellcor-200 heart rate monitor,
which was attached by a clip to each participant’s fingertip. In this simulation, only one
hand is used for mouse operations, so the monitor was attached to the other hand. The
monitor logged the participant’s heart rate every five seconds.

The idea that emotions can be measured via physiological indicators dates as far
back as William James (1884). The advantage of physiological indicators is that they
bypass self-report measures, tapping into the physiological component of affective states.
In particular, functions of the autonomic nervous system (ANS) such as heart rate. blood
pressure. and finger temperature, have been commonly used as measures of emotion
since the 1950’s (Cacioppo, Klein, Bemtson, & Hatfield, 1993). According to Levenson
(1988, p. 40), the role of the ANS as an indicator of emotion is “the centerpiece of its
evolutionary value in emotion.” There has been some debate among emotion researchers
as to whether ANS indicators can effectively discriminate among discrete emotions (i.e.,
anger vs. disgust. happiness vs. fear). Fortunately, for the purposes of this study.
discriminating among discrete emotions was not necessary; instead, I only needed to
measure the arousal dimension of emotional states. On this, emotion scholars largely
agree that ANS measures are good indicators of arousal (e.g., “...the psychophysiology of
emotion literature indicates that ANS measures are better indexes of the arousal

dimension than of affective valence” (Guglielmi, 1999. p. 148); “Most measures of

54

physiological reaction give direct indications only of the extent of arousal” (Cook &
Selltiz, 1964. p. 53)).

Of the various ANS measures, heart rate is the most reliable measure of emotional
arousal (Cacioppo et al.. 1993; Guglielmi, 1999). Arena et al. (1983) showed high test-
retest reliabilities of heart rate from week to week, with correlations of .66 while resting,
and .53 and .79 under two stress conditions (performing mental arithmetic and imagining
a stressful event). Additionally, they showed good test-retest reliabilities of change in
heart rate (from resting to stress conditions), with correlations of .43 and .52. measured
one week apart. Thayer (1970) showed a correlation of .33 between heart rate and self-
reports of emotional activation, and concluded that heart rate and skin conductance were
the two best physiological indicators of emotional arousal. Other studies have shown
heart rate increases under experimentally-induced conditions of fear (Chessick. Bassan,
& Shattan, 1966; Grossberg & Wilson, 1968; Tourangeau & Ellsworth, 1979) anger
(Chessick et al., 1966; Frodi, 1978), and social anxiety (Knight & Borden, 1979), all of
which are considered to be high-arousal emotions. Although these studies provide some
evidence for the construct validity of heart rate as a measure of arousal, this evidence is
certainly not overwhelming. Thus. an additional contribution of the present study was
assessing the validity of heart rate as a measure of emotional arousal.

Because individual resting heart rates vary, studies using heart rate as a measure
of arousal typically use the change score between resting heart rate and heart rate during
the experimental manipulation. Difference scores. however. can be problematic because

one cannot tell whether the effect is due to the starting or ending values (Edwards &

55

Parry. 1993). Therefore, the effects of workload on emotional arousal were analyzed
through repeated measures regression (as noted below under Results).

Learning. Learning was assessed via a multiple-choice test, and participants
completed this test three times: when they arrived at their laboratory session and after
both games. The test consisted of two parts. F irst, it assessed the degree to which
individuals and teams correctly identified the four types of novel tracks in terms of the
power levels of each type of novel track. A similar measure was previously used on this
task by Ellis et al. (2003; although they used behavioral measures rather than a
knowledge test). Second, it quizzed teams on the best strategy for playing the game. I
wrote four questions for this portion of the test, and tested them on seven people who
were experts on the DDD simulation (current and former Research Assistants). All seven
agreed on the correct answers for two of the items. Six agreed on the answer for the third
item, and five agreed on the fourth. The intraclass correlation was over .99, indicating
strong inter-rater agreement. The observed reliability of the test for the experiment was
assessed using the Kuder-Richardson formula for dichotomously scored items. KR-20 on
the pre-game test was .11, after Game 1 was .35, and after Game 2 was .38. The low
reliability for the pregame test was expected, because participants guessed at the answers.
The reliabilities after both games were considerably higher than the pregame test. but
were still quite low. Therefore, I did not proceed with testing any of the learning
hypotheses using this test.

Perfonnance: Speed and Accuracy. Speed of performance was operationalized as
a combination of attack speed and identification speed. Attack speed measures the

elapsed time between when an enemy track enters the restricted area and when a team

56

member engages it. Identification speed measures the elapsed time between when a track
enters the screen and when a team member identifies it. The two variables were
standardized and averaged to create the speed composite. which was obtained in both
Game 1 and Game 2.

Accuracy of performance was operationalized as a combination of friendly fire
errors and missed opportunities. Friendly fire errors reflect a count of the number of
times a friendly track is disabled. Missed opportunities is a count of the number of times
an enemy track is engaged, but the vehicle used to engage the track does not have enough
power to disable it. Both of these variables represent errors made by the team, and were
standardized and averaged to create the accuracy composite, which was also obtained in
both Game 1 and Game 2.

Data Analysis

The research design in this study incorporated both within-units (individuals or
teams) and between-units components. Workload, resource allocation structure, and
transactive learning structure were between-units components, and time was a within-
units component. Because of this mixed design, I utilized repeated measures regression to
analyze the data. Repeated measures regression partitions the variance in the dependent
variable into two orthogonal sources: variability between units, due to differences in
overall performance, and variability within units. due to differences in time (Cohen &
Cohen, 1983). Performance is then regressed onto the predictors. and the variance
explained by the predictors is compared to the appropriate variance partition (either the
between-units or within-units variance). This allows one to see how much of the

appropriate variance component is explained by the predictors.

57

For example. workload was a between-units manipulation (i.e., each individual or
team experiences only one sequence of workload), and thus any variance explained by
workload should be compared to the between-units variance, rather than to the total
variance in performance. Time was a within-units variable (i.e., each individual or team
provides a measure of the mediators and dependent variables in both games). This
analytical method also allows one to test interactions between within-units components
with between-units components. Significant interactions would indicate support for
contingencies, like the ones predicted. I note that this analysis is identical to one in which
the measures for each game were treated as separate dependent variables, followed by a
test of the difference between the regression coefficients. I chose repeated measures

regression for parsimony and clarity of presentation.

58

RESULTS

Descriptive Statistics

Table 1 presents the means and standard deviations for all of the measured and
manipulated variables. Examination of this table reveals that in terms of the pregame
variables (personality, arousal, and knowledge), those who participated in the individual
portion of the experiment were virtually identical to those who participated in the team
portion of the experiment. They also did not differ in terms of their subjective perceptions
of workload in either Game 1 or 2. In both games, however, those who worked in teams
reported being significantly more emotionally aroused both in terms of the CMQ
statements (Game I: M = 3.50, SD = .36: Game 2: M = 3.50, SD = .39) and the CMQ
adjectives (Game 1: M = 3.38, SD = .37; Game 2: M = 3.31, SD = .39) than those who
worked alone (Game 1 statements: M = 3.04, SD = .90, p < .01; Game 2 statements: M =
2.94. SD = .98, p < .01; Game 1 adjectives: M = 2.83, SD = .93, p < .01; Game 2
adjectives: M = 2.87, SD = .92, p < .01). They were not significantly different, however,
in terms of heart rate in either game. This may reflect the social facilitation hypothesis,
that peeple are more aroused when interacting with others than when they are alone.

Interestingly, they also did not differ significantly in terms of either of the
learning variables in either game. This indicates that on this task at least, having the
additional information provided by team members did not result in more learning than
working alone. Also notable is the fact that in terms of both of the learning variables
measured before they played the game. the participants achieved scores that would be
expected by chance. This was expected. as they simply guessed at the answers to these

questions. because they could not have known them without playing the game.

59

Table 2 presents the correlations of all of the measured and manipulated variables.
Although some of the relationships are discussed below in testing the hypotheses, one
other relationship bears mentioning. Despite random assignment, the workload
manipulation correlated significantly with both neuroticism (.29. p < .01) and
extraversion (.-.25. p < .05) in teams. For individuals working alone, however, these
relationships were not significant. The size of the correlations is due in part to the fact
that aggregating the personality measures to the team level almost invariably reduces the
standard error. The correlations prior to aggregation were .15 for workload and
neuroticism, and -.13 for workload and extraversion: the significance levels were the
same as with the aggregated variables. Neuroticism and extraversion did not correlate
significantly with the other two manipulated team variables.

Hypothesis Testing

Hypothesis 1, which predicted a positive relationship between workload and
emotional arousal in individuals in Game 1, was tested via repeated measures regression.
The arousal measures were regressed on workload, time (baseline arousal measured when
they arrive at their session. and after Game 1), and their interaction term. For the
hypothesis to be supported, there would be a positive relationship between workload and
arousal in Game 1, but no relationship with the baseline measure. Table 8 shows the
results of this test. None of the arousal measures (C MQ statements, CMQ adjectives, or
heart rate) were significantly affected by the workload manipulation. and thus Hypothesis
1 was not supported. Subjective workload, however, was significantly associated with
both of the C MQ scales. When controlling for pregame arousal levels, the NASA-TLX

scale was positively associated with both the CMQ statements scale ([3 = .23. p < .05) and

60

the CMQ adjectives scale (13 = .27. p < .01). Although these relationships may suffer from
common method and source bias. the fact that pregame arousal levels were included in
the regression equation lessens the concern about this bias. Any shared variance because
of common source or method would presumably also be found in the self-report pregame
measure as well, and thus would not be reflected in the relationship between subjective
workload and Game I arousal.

Game 1 objective workload did, however, have an effect on Game 2 self-reported
arousal. For the CMQ statements, controlling for pregame arousal, Game 1 workload
exerted a negative effect on Game 2 arousal (,6 = -.20. p = .05), accounting for 4.1% of
the variance. The implication seems to be that the contrast between the workload they
experienced in Game 1 (which varied across conditions) and the workload they
experienced in Game 2 (which was the same across conditions) created differing levels of
arousal. Those who had higher workloads in Game 1 reported lower levels of arousal in
Game 2 (when their workload decreased), whereas those who had lower workloads in
Game 1 reported higher levels of arousal in Game 2 (when their workload increased). For
the CMQ adjectives and heart rate. the coefficients were in the same direction, but were
not statistically significant.

Hypothesis 2 predicted that there would be a positive relationship between
workload and learning in individuals. This was also tested via repeated measures
regression, and Table 13 also shows these results. As with arousal, workload did not
significantly affect either of the learning variables; instead, time showed a positive main
effect on the participants’ identification of the unknown tracks ([1 = .47, p < .01). This

was expected, as playing the game gave the participants a chance to identify the tracks

61

(whereas they simply guessed at them prior to the game). I also examined the relationship
between workload and learning by regressing Game 1 learning on the workload variable
and its squared term (controlling for the baseline learning measure), in order to determine
whether the relationship is best characterized as linear (as hypothesized) or quadratic (as
predicted by the Yerkes-Dodson law). Neither workload nor its squared term exerted a
significant effect on learning. Thus, neither Hypothesis 2 nor the Yerkes-Dodson law
were supported in these data. Hypothesis 3 predicted that emotional arousal would
mediate the relationship between previous workload and individual learning. Because
workload was not significantly related to Ieaming. Hypothesis 3 was not supported.
Hypothesis 4 predicted a positive relationship between previous workload (Game
1) and subsequent speed (Game 2) for individuals. I tested this both by examining the
bivariate correlation and through repeated measures regression, where Game 2
performance was regressed on Game 1 workload, time (Game 1 or 2), and their
interaction term. The bivariate correlation was not significant (-. 1.2). As can be seen in
Table 9. the interaction term of the repeated measures regression was significant, but this
was due to the strong relationship between workload and speed in Game 1. and not due to
a carryover effect of previous workload on subsequent speed. Thus, Hypothesis 4 was not
supported. Hypothesis 5 predicted that emotional arousal mediates the relationship
between previous workload and subsequent speed. Because the relationship between
workload and subsequent speed was not significant, Hypothesis 5 was not supported.
Hypothesis 6 predicted a positive relationship between previous workload and
subsequent accuracy in individuals. As with speed, the bivariate correlation was not

significant (-.05), but the interaction term of the repeated measures regression equation

62

was. as can be seen in Table 9. Again, this was due to the relationship between workload
and accuracy in Game 1. and not due to a carryover effect of workload on subsequent
accuracy. Thus, Hypothesis 6 was not supported. Interestingly. however, the bivariate
correlation between subjective workload in Game 1 and accuracy in Game 2 were
negatively related (r = -.28. p < .05). This means that individuals who perceived the
workload to be high in Game 1 made more errors in Game 2 than those who perceived
the workload in Game 1 to be low. Hypothesis 7, which predicted that individual learning
mediates the relationship between previous workload and subsequent accuracy, was not
supported, as neither the relationship between workload and Ieaming. nor between
workload and accuracy, was significant.

Hypothesis 8 predicted that workload is positively related to emotional arousal in
teams. and like Hypothesis 1. was tested through repeated measures regression. As can be
seen in Table 10, the interaction term was not significant for any of the emotional arousal
measures, and thus Hypothesis 8 was not supported. As with the individuals working
alone. however, subjective workload did have a significant effect on both the CMQ
statements scale ([1 = .41, p < .01) and the CMQ adjectives scale (,6 = .33. p < .01), when
controlling for pregame arousal levels. Again, the concern about common method and
source bias is somewhat lessened because of the inclusion of the pregame scales in the
regression equation.

Also similar to individuals working alone, Game I objective workload exerted an
effect on Game 2 self-reported arousal. For the C MQ statements, the effect of Game 1
workload on Game 2 arousal was significant and negative when controlling for pregame

arousal (,6 = -.24. p < .05). accounting for 4.6% of the variance. Again, this pattern held

63

even when controlling for Game 1 arousal levels. For the C MQ adjectives, the effect was
again in the same direction, but not significant. Again, this implies that the contrast
between the workloads in Games 1 and 2 affected arousal levels. Despite the fact that all
teams played the same workload in Game 2. teams that had higher workloads in Game 1
reported lower arousal levels in Game 2 (when their workload decreased), whereas teams
that had lower workloads in Game 1 reported higher arousal levels in Game 2 (when their
workload increased).

Hypothesis 9 predicted that teams will show a positive relationship between
previous workload and subsequent speed. Like Hypotheses 4 and 6, I tested this both by
examining the bivariate correlation and through repeated measures regression. The
correlation was positive (.18) but did not reach traditional levels of statistical significance
(p = .10). In the repeated measures regression equation, the interaction term was
significant ([1 = .35, p < .01), and can be seen in Table 11. Figure 2 shows the nature of
the interaction; there is a negative slope for workload and speed for Game 1, but a slight
positive slope for Game 2. Because of the lack of bivariate significance, however.
Hypothesis 9 was not supported. Hypothesis 10 predicted that emotional arousal mediates
the relationship between previous workload and subsequent speed. Because the bivariate
relationship between previous workload and subsequent speed was not significant,
however. this hypothesis was not supported.

Hypothesis 11 predicted that teams will show an attenuated relationship between
workload and Ieaming. The bivariate correlation was significantly negative for
identification of the unknown tracks (-.3 l , p < .01). Similarly, as can be seen in Table 12,

the interaction term was significant in the repeated measures regression for identification

64

of the unknown tracks (,6 = -.56. p < .05). Thus, Hypothesis II was supported. This
finding for the unknown tracks is particularly interesting because the negative effect of
workload on learning the values of the unknown tracks occurred despite the fact that the
higher workloads provided more opportunities to learn the values of the unknown tracks.
Each progressively higher workload condition had four more unknown tracks than the
condition below it. Thus, the very high workload condition had sixteen more unknown
tracks (one of each type) than the very low workload condition, yet they were much
worse in determining the values of these tracks.

This team finding appeared to be different than the null finding for individuals
working alone. so I tested the differences between the workload-leaming correlations for
teams and individuals. This test revealed. however, that the correlations were not
significantly differently from each other (p = .16. two-tailed). Thus, the data do not
support the notion that high workloads are harmful for team learning but not for
individual Ieaming.

Interestingly, the effect of workload on team learning continued to hold in Game
2. The bivariate correlation between Game 1 workload and Game 2 identification of the
unknown tracks was -.20 (p = .06), and when controlling for pregame guesses of the
unknown tracks. the effect was significant ([3 = -.23, p < .05). This implies that the higher
the team’s workload, the worse they were at determining the values of the unknown
tracks, and this had a carryover effect into the future. Even though their workload
declined. they were not able to make up for their previous lack of Ieaming.

Hypothesis 12 predicted a negative relationship between previous workload and

subsequent accuracy in teams. The bivariate correlation was not significant (-.02). nor

65

was the interaction term of the repeated measures regression displayed in Table 11, and
thus Hypothesis 12 was not supported. Hypothesis 13 predicted that team learning
mediates the relationship between previous workload and subsequent accuracy, but
because the relationship between previous workload and subsequent accuracy was not
significant, this hypothesis was not supported.

Hypotheses 14 and 15 predicted that the personality traits of extraversion and
neuroticism moderate the relationship between workload on the one hand, and arousal
and learning on the other, at both the individual and team levels. These hypotheses were
tested through moderated regression, where Game 1 arousal and learning (with pregame
levels as a control) were regressed on workload and personality (at the team level, this
was the mean level of the personality traits for the team), and their interaction terms.
Tables 13 and 14 present the results of Hypothesis 14, which predicted that the highest
levels of arousal would be found in individuals and teams high in Extraversion and under
the highest levels of workload. For individuals working alone, extraversion did indeed
moderate the effect of workload on emotional arousal for both the CMQ statements scale
([9 = -2.22, p < .05) and the CMQ adjectives scale (5 = ~2.07, p < .05). The nature of the
interaction, however, was contrary to the hypothesis. As can be seen in Figure 3, which
displays the results for the CMQ statements scale. it was a fully crossed interaction where
the highest levels of emotional arousal were found in individuals high in Extraversion
and low in workload, or low in extraversion and high in workload. Although not
graphically displayed, the CMQ adjectives scale had the same interaction form. For
teams, extraversion did not significantly moderate the effect of workload on any of the

arousal measures. As a result. Hypothesis 14 was not supported.

66

Although it was not hypothesized, neuroticism also moderated the effect of
workload on emotional arousal for individuals working alone. As with extraversion,
neuroticism moderated the effect of workload on the CMQ statements scale (,8 = -1 .06, p
< .05) and the CMQ adjectives scale ([1 = -l.10. p < .05), but had no effect on heart rate
([1: .63. ns). Figure 4 displays the results of the interaction for the CMQ statements
scale. showing that at low levels of workload. neuroticism had no effect on emotional
arousal. At high levels of workload, however. individuals low on neuroticism had higher
levels of arousal than those who were high on neuroticism. Interestingly, the moderating
effects of both extraversion and neuroticism carried over into Game 2 with almost
identical patterns. despite the fact that the workload had changed.

Hypothesis 15 predicted that neuroticism moderates the relationship between
workload and learning at both the individual and team levels. and the results of these tests
are displayed in Tables 15 and 16. For individuals working alone, neuroticism exerted a
negative main effect ([1: -.29, p < .05), and also moderated the effect of workload on
learning the unknown tracks ([1 = - 1.30. p < .05). The pattern of the interaction for the
unknown tracks is displayed in Figure 5 and is consistent with the hypothesis. For
individuals high in neuroticism, workload exerted a negative effect on Ieaming. and the
lowest level of learning was for individuals who were high in neuroticism and had high
workloads. Interestingly, workload exerted a positive effect on learning for individuals
who were low in neuroticism. Thus. the main effect hypothesis that workload would be
positively related to learning in individuals working alone, held only for those who were

low in neuroticism. For teams, however, neuroticism did not significantly moderate the

67

effect of workload on leaming the unknown tracks. Thus, Hypothesis 15 was partially
supported.

Hypothesis 16 predicted that resource allocation structure moderates the
relationship between workload and emotional arousal in teams, and was tested through
moderated regression. Table 17 displays the results of this test. The two-way interaction
between workload and resource allocation was not significant for any of the arousal
measures. and thus Hypothesis 16 was not supported.

Hypothesis 17 predicted that transactive learning moderates the relationship
between workload and Ieaming. and the results of this test are displayed in Table 18. As
noted earlier, workload exerted a negative main effect on learning the unknown tracks.
Structure also exerted a main effect, in that divisionally structured teams showed better
learning than functionally structured teams. This is no doubt due to the fact that each
team member in the divisional structure had all of the resources they needed to
independently determine the values of the unknown tracks, whereas team members in the
functionally structured teams had to depend on information from their team members to
collectively determine the values of the unknown tracks. The two-way interaction
between workload and structure was not significant for learning the unknown tracks,
however. and therefore Hypothesis 17 was not supported.

Additional Post-hoe Analyses

Because few of the hypotheses were supported, I conducted additional post-hoe
analyses to determine (a) whether support for the proposed model could be found in other
operationalizations of the constructs, and (b) whether other non-hypothesized

relationships could account for significant variance in the proposed mediators and

68

dependent variables. Below. I describe these analyses. first for teams, and then for
individuals working alone.

Teams. First, I examined different operationalizations of the dependent variables
as captured by the simulation. Because I proposed that learning the unknown tracks
would be affected by workload (and indeed it was for teams), I was able to separate parts
of the dependent variables into whether or not they involved unknown tracks. It could be
that rather than previous workload affecting total subsequent speed and accuracy. it could
differentially affect speed and accuracy involving unknown tracks versus speed and
accuracy involving known tracks.

For teams, workload had a direct negative effect on learning the unknown tracks.
Therefore. one may expect that team accuracy toward the unknown tracks would be most
affected by workload. I was able to separate friendly fire kills of unknown tracks and
friendly fire kills of known tracks in order to test this. Repeated measures regression
(with known and unknown being the repeated measure) revealed that this was indeed the
case. As illustrated in Figure 6, previous workload greatly affected friendly fire kills of
the unknown tracks, but had almost no effect on friendly fire kills of the known tracks.

Because I had proposed that learning would mediate the relationship between
previous workload and subsequent accuracy, I tested this using learning the unknown
tracks as the mediator and friendly fire kills of unknown tracks as the dependent variable.
As noted in Table 5. workload and learning were significantly related (r = -.31, p < .01).
Learning and subsequent friendly fire kills of unknown tracks were also significantly
related (r = -.36, p < .01). With learning controlled, the effect of workload on subsequent

friendly fire kills of unknown tracks dropped from .32 (p < .01) to .24 but remained

69

significant (p < .05). Sobel’s test revealed that these were significantly different (test
statistic: 2.08. p < .01). Therefore, learning did not fully mediate, but appears to have
partially mediated the effect of workload on subsequent accuracy with these
operationalizations.

Neither of the variables used to compose the speed composite showed significant
differences between the correlations of workload with known and unknown tracks. but a
related variable did. Good attacks represents the total number of times the team
successfully disabled an enemy track; as such it is a quantity of performance variable,
and as noted above, quantity and speed are related performance factors. Although I
expected the accuracy related to unknown tracks to be more affected by workload than
the accuracy related to unknown tracks, I expected the opposite for speed/quantity. I had
expected teams with higher previous workloads to be faster than teams with lower
previous workloads. Because these teams did not learn the unknown tracks as well as
teams with lower workloads, their advantage in quantity. then, should be more evident
with the known tracks. That is, because teams with higher previous workloads did not
learn the power levels of the unknown tracks, they would not be able to attack them
quickly in the subsequent game. Thus, I would expect that they would attack more known
tracks than teams that had lower workloads, but may not be different in their number of
attacks on unknown tracks. Repeated measures regression revealed that this was indeed
the case. As illustrated in Figure 7, workload was positively related to the number of
good attacks of known tracks, but had no relationship with the number of attacks of

unknown tracks.

70

Second. I examined possible relationships among other measured variables. These
included both variables captured by the simulation and variables measured through self-
report. and I examined both main effects and interactive effects with the measured
personality variables. Workload had a significant main effect on a Game 2 self-report
variable in teams. Mental overload (Campion & McClelland. 1993) in Game 2 was
negatively related to previous workload (r = -.35, p < .01). This relationship reflects the
contrast between the teams’ workload in Game 1 and their workload in Game 2. Teams
whose workload decreased reported lower levels of overload, and teams whose workload
increased reported higher levels of overload. Notably, mental overload in Game 1 was
also related to Game 1 workload (r = .26, p < .05).

I tested whether mental overload mediated the relationship between previous
workload and subsequent good attacks of known tracks. Controlling for mental overload
in Game 2, the beta for workload dropped to .14 (from .24) and became non-significant.
Sobel’s test also indicated that these were significantly different (test statistic: 2.11, p <
.05). This indicates that mental overload at least partially mediated the effect of previous
workload on subsequent quantity of performance.

Regarding personality variables, extraversion showed no interpretable main or
interactive effects with workload on any team variables. Neuroticism did not show any
significant main effects on team variables, but it did interact with workload in predicting
Game 2 friendly fire kills in teams. The nature of the interaction was such that for teams
low in neuroticism. previous workload had almost no effect on their subsequent friendly
fire kills. For teams high in neuroticism, however, previous workload had a strong

positive effect on their subsequent friendly fire kills. This is in keeping with the original

71

model. which suggested that the teams that learned least would be those high in
neuroticism with high workloads. Subsequently. one would expect that the low levels of
learning would lead to lowest accuracy in Game 2. Although the learning hypothesis was
not directly supported in the test (perhaps due to the strong main effect of workload on
learning), the effect was evident in the teams’ behavior in Game 2.

I also measured the other three factors of the “Big 5” model of personality:
agreeableness. conscientiousness, and openness to experience. I had no theoretical
reasons to test the effects of these factors on the mediators and dependent variables, but I
did so purely for inductive reasons. Like extraversion, agreeableness yielded no
significant main or interactive effects on team variables. Conscientiousness exerted a
significant interactive effect with workload on one captured team variable in Game 2.
Attack speed was one of the original variables comprising the speed composite.
Workload had no effect on attack speed for teams high in conscientiousness, but was
positively related to attack speed for teams low in conscientiousness. This may suggest
that the proposed main effect of previous workload on subsequent speed is attenuated in
teams high in conscientiousness. Because conscientiousness is associated with fulfilling
one’s responsibilities and duties (Costa & McCrae, 1992). it may be that highly
conscientious teams work the same way regardless of their workload history.

Openness interacted with workload to predict the total number of times the team
attacked tracks (these were not just “good” attacks as discussed above, but all attacks.
including missed opportunities and friendly fire kills). This was a fully crossed
interaction. such that teams high in openness that had low previous workloads, and teams

low in openness that had previous workloads. showed the highest number of attacks in

72

Game 2. This may suggest that teams high in openness see increases in workload as a
challenge to be met, and thus work harder than teams low in openness. This does not,
however. result in an overall performance improvement, because although they are more
productive (more attacks), they also make more mistakes (more friendly fire kills and
missed opportunities). When their workload decreases, however, teams high in openness
may lose interest and not work as hard as teams low in openness.

Individuals working alone. Workload did not show any main effects on any of the
captured or self-report Game 2 variables for individuals working alone. In terms of
personality, however, neuroticism showed numerous main effects on captured individual
variables captured in Game 2. Neuroticism was negatively related to the number of tracks
identified (r = -.28. p < .05), the number of good attacks (r = -.29. p < .05), the number of
launches (r = -.25, p < .05). and attack speed of unknown tracks (r = -.32, p < .01). It was
also positively related to the number of missed opportunities (r = .28, p < .05). Taken as a
whole, these indicate significantly worse performance by individuals high in neuroticism,
as they were less productive overall, slower in attacking unknown tracks. and were more
likely to attack tracks with assets that did not have enough power to disable them. As
noted in Table 5. neuroticism also showed a negative effect on learning the unknown
tracks (r = -.27, p < .05), which explained much of the variance in attack speed of the
unknown tracks (the effect of neuroticism dropped to .21 and became non-significant
when learning was controlled) and missed opportunities (dropped to .16 and became non-
significant when learning was controlled). In terms of interactive effects, neuroticism
showed no significant interactions with workload on any of the individual variables in

Game 2.

73

Extraversion. agreeableness, and conscientiousness showed no interpretable main
or interactive effects with workload on captured individual variables. Openness to
experience, however, showed a number of significant main effects. Openness was
positively related to the number of tracks identified (r = .22. p < .05) of good attacks (r =
.25. p < .05), especially of the unknown tracks (r = .34. p < .01); and was negatively
related to the number of missed opportunities (r = -.29, p < .05) and the number of times
a track was disabled using more power than was needed (wastes; r = -.23, p < .05). Thus.
people high on openness appear to have been better performers overall, as they were

more productive, accurate, and efficient than those low on openness.

74

DISCUSSION

This research set out to test a model of the effects of previous workload on
subsequent performance in teams and individuals. The results advance multi-level theory
regarding the affective and cognitive processes that accompany varying levels of
workload in both teams and individuals. This research demonstrated how a relevant past
experience—their previous workload—affects future performance through individual and
team thoughts and feelings as they perform their task. Thus, the research continues the
current trend of not simply identifying which variables affect individual and team
performance, but how and why they affect performance (Ilgen, Hollenbeck, Johnson, &
Jundt, 2005). In this discussion, I first summarize the findings and limitations of the
study. then proposed a revised theoretical model, and finally offer theoretical
implications and practical applications of the study.
F indin gsfor Individuals Working Alone

For individuals working alone, workload had no main effects on arousal, Ieaming.
or subsequent performance. Instead. the effect of workload on arousal and learning
depended upon the personality traits of the individual. Individuals who were extraverts
exhibited high levels of arousal under conditions of low workload, whereas individuals
who were introverts exhibited high levels of arousal under conditions of high workload.
These results appear to paint a more complex picture of the relationship between
extraversion and arousal than that indicated by the theories of Hans Eysenck (1967), who
suggested that extraverted people have lower baseline levels of arousal and are more
affected by external stimuli. In the individual portion of this study, extraverts did not

report significantly lower levels of arousal than introverts when they arrived at their

75

experimental sessions (although this could be due to the stimulating effect of anticipation
of participating in an experiment, which would have raised extraverts’ baseline arousal
levels). Then in what appears to contradict Eysenck’s theory. extraverts did not become
more aroused under conditions of high workload, but rather became less aroused,
compared to those under conditions of low workload. In contrast. introverts became more
aroused under conditions of high workload, and less aroused under conditions of low
workload.

It must be noted that although much research has found support for Eysenck’s
extraversion-arousal theory, a number of published studies have not supported the theory.
Matthews and Gilliland (1999) note that three factors in particular have accounted for the
seemingly anomalous findings in this area. F irst, time of day has moderated the effect of
extraversion on arousal. because extraverts tend to be more aroused in the evening than in
the morning (Revelle et al., 1980). Although I found no time of day effects in these data,
this could be because the range of workload was restricted in the morning and afternoon
laboratory sessions, with only the evening sessions having the full range of workload.
Second, there is some evidence that instead of extraversion leading to higher arousal
under higher stimulus conditions, extraversion and arousal interact to predict
performance. I also found no extraversion by arousal interactive effects in these data.

The third factor suggested by Matthews and Gilliland (1999) may provide the best
post—hoe explanation for these unexpected results. The nature of the task and the
laboratory setting in which the experiment took place may have affected arousal levels.
The extraversion-arousal effect “is largely restricted to simple tasks requiring encoding of

easily-perceived stimuli or to more complex tasks with a routine encoding component”

76

(Matthews & Gilliland, 1999, p. 610), but is not found in tasks that are attentionally
demanding. The attentional demands of this simulation increased as workload increased,
and the unknown tracks created additional demands that required more than “routine
encoding.” Therefore, this task may not have been simple enough to find the expected
effect.

It may also be that extraverts are only emotionally aroused by external stimuli that
involve interaction with other people. Due to space limitations, multiple participants in
the individual portion of the study were sometimes working at computers in the same
room. Under the low workload conditions. it is possible that the extraverts interacted with
the other people in the room, which would have raised their arousal levels. Under the
high workload conditions. however, the task was so demanding that they would not have
had the time or cognitive resources to devote to interacting with the other participants,
and thus would experience lower arousal levels. In contrast, the introverts reported higher
arousal levels when the workload was higher, possibly because the individual nature of
the task was more stimulating to them when it was more demanding.

Neuroticism also moderated the effect of workload on emotional arousal for
individuals working alone, and again, it appears to be in contrast to Eysenck’s (1967)
theory. Eysenck proposed that people high on Neuroticism become more emotionally
aroused by external stimuli than people low on Neuroticism. The results of this study,
however, showed that people high in Neuroticism were less aroused under conditions of
high workload than they were under conditions of low workload. In contrast. people low
in Neuroticism were more aroused under conditions of high workload than they were

under conditions of low workload.

77

Two post-hoe explanations may account for this unexpected finding. First,
Eysenck (1994) postulated a protective mechanism he called “transmarginal inhibition”
(TMI). TMI occurs when individuals reach high levels of stimulation: rather than
continuing to become more aroused, increasing stimuli can actually cause people to
become less aroused. This effect has been supported empirically (LeBlanc, Ducharme. &
Thompson. 2004). It may be that the participants who were high in Neuroticism found the
higher workload conditions too stimulating, which activated the TMI mechanism.
Second. positive affect (PA) may be driving the finding. Although not reported above. I
also measured positive and negative affect, and found a very strong correlation between
PA and the CMQ statements scale (.48 at pretest, .74 after Game 1, and .69 after Game
2). Neuroticism moderated the effect of workload on PA in an almost identical pattern as
the one found for arousal. It may be. then, that people high in Neuroticism simply found
higher levels of workload unpleasant, and mentally disengaged from the task. In contrast.
people low in Neuroticism found higher workloads to be more pleasant and arousing.

Neuroticism also moderated the effect of workload on Ieaming. as hypothesized.
Individuals who were low on Neuroticism learned more under conditions of high
workload than under conditions of low workload, whereas those who were high in
Neuroticism showed the opposite pattern. This is consistent with Eysenck’s (1967) notion
that that people high in Neuroticism are more reactive to anxiety-producing stimuli, and
use up cognitive resources to cope with the anxiety rather than devoting them to Ieaming.
When individuals are low in neuroticism, however. they can take advantage of the
additional information afforded by higher workloads. and learn more than when their

workload is lower.

78

Findings for Teams

In contrast to its effect on individuals working alone. personality exerted almost
no influence on team outcomes. Instead. workload exerted a negative main effect on team
Ieaming. regardless of the personality characteristics of the team. This supports the notion
of team performance as a dual-task paradigm, where team members must attend both to
taskwork and teamwork. When workload is high. teams cannot do both of these
simultaneously. and teamwork is more likely than taskwork to suffer. Team learning is
primarily a function of teamwork. because team members must attend to the information
communicated by their teammates.

Communication among team members, however, suffers under conditions of high
workload. Team members focus on their individual tasks—their taskwork—and neglect
to either transmit or receive vital information that would help the team learn. This study
revealed that teams under high workloads learned less than teams under low workloads.
Even more troubling. this had a carryover effect such that teams facing initially high
workloads did not “catch up” to the teams facing initially low workloads, even when their
workload decreased. This may reflect communication norms that were developed early
and carried over into the teams’ later experiences. Teams under high workloads

developed norms of low communication, and thus were hindered in their learning in the

future.
Limitations

There are at least eight aspects of this study that limit its findings: four related to
both the individual and team portions of the study, two specifically related to the

individual portion of the study. and two specifically related to the team portion of the

79

study. F irst, this study examined a limited range of workload. It is not clear what the
results would be at more extreme levels of workload. For example, at inordinately high
levels of workload may cause participants to simply give up and not even attempt to work
at the task. I suspect that this would not be limited to laboratory tasks like the one in this
study, but would also hold true in field settings. Employees who are assigned so much
work that it is overwhelming may strike or otherwise walk off the job. knowing that they
are not able to accomplish the workload assigned.

Second, this study examined a specific conceptualization of workload: an
objective quantitative increase in the number of tasks assigned. As noted in the
introduction. Shaw and Weekley (1985) have discussed the idea of workload having both
quantitative and qualitative elements. It is unclear what the effects of previous workload
would be on future performance if the workload changed qualitatively. A different type
of work—rather than simply more or less of the same type of work—may yield different
results than the ones found here.

Additionally, at least two factors limit the overall generalizability of the findings
in this study. First, the experimental task is clearly not representative of all tasks teams
undertake, and thus the findings may vary slightly for different tasks. The task has a great
deal of both “mundane realism” and “psychological realism” (Berkowitz & Donnerstein,
1982), however, that enhances my confidence in the external validity of the findings.
Mundane realism exists when a laboratory environment bears superficial similarity to a
real-world task. The nature of this task—in terms of collecting information and making
decisions using computer technology—is somewhat similar to military command-and-

control settings and some civilian settings (e.g., air traffic control, emergency dispatch).

80

Psychological realism exists when a laboratory setting induces similar psychological
states to real-world situations. Although the rewards for good performance and the
consequences for errors were not as significant as in real work settings. the participants
were generally highly engaged by the task and were visibly upset when they made errors.

Second, the sample consisted of undergraduate university students and thus may
not be representative of the population of people in work settings. I note, however, that in
many military settings, tasks like this are performed by lieutenants, who are
approximately the same age as the participants in this study. Third, in this study
participants performed over a short period of time. Thus, it is not clear whether the results
would hold for individuals or teams that perform over longer periods. It may be that the
negative effect of workload on learning found in teams would be lessened over time. On
the other hand, however. it may be possible that if a team performed under a high
workload over a longer period of time, the effect would be even greater. That is, as teams
perform over long periods of time, they develop ingrained communication patterns that
are resistant to change. Teams that had previously high workloads for a long period of
time may not be able to overcome these patterns and may never learn effectively
together.

Two aspects of the individual portion of this study greatly limit its findings. First,
as noted above. participants in the individual portion of the study were in the same room
with up to three other individuals who were also participating in the study. Although they
were instructed to work on the task alone, it is possible that they shared information with
each other about the task. Thus, participants could have acted as a de facto “team” and

assisted each other with learning about the unknown tracks. This is somewhat unlikely. as

81

participants in the individual portion of the study were instructed that their chance of
winning a cash prize was dependent upon them being in the top half of all individual
performers. Thus. assisting another participant would decrease their own chances of
winning a prize. A more likely scenario is that participants may have simply chatted with
each other but not provided assistance. Because the participants were all drawn from a
pool of students in the same course, they may have known each other or talked with each
other about course-related subjects. This may have affected their performance on the
simulation in unpredictable ways.

Second, the individual simulation was constructed to mirror the team simulation
as much as possible. Because of this, I left in all tracks that did not enter the quadrant of
the restricted area that the participant was to defend. This allowed the individual
participant to see the tracks in other quadrants in the same way that participants in the
team portion of the study could see. This was to ensure that the task “looked” the same
way to participants in the individual and team portions of the study. The downside of this
was that participants in the individual portion of the study could send vehicles to the
other quadrants and attack those tracks if they wished. They were specifically instructed
not to do this and to only defend their own quadrant. Nevertheless, 73% of the
participants did attack at least one target outside of their quadrant, with an average of
2.84 attacks in Game 1 and 3.44 attacks in Game 2. This was less than the amount for
participants in teams for both Game 1 ([1 = 4.76) and Game 2 (,u = 6.65), but in teams. the
total number of tracks was not affected. That is, teams as a whole did not face more
tracks when team members attacked tracks outside of their own quadrants. For

individuals working alone. however. attacking tracks outside of their quadrants increased

82

the total number of tracks they faced. Indeed, one participant attacked 25 tracks outside
of his quadrant in Game 2. effectively doubling his workload. It is unclear what effect
this had on the performance of individuals in the study.

- At least two factors limit the findings in the team portion of the study as well.
First. personality was operationalized as the mean level of the team members’ self-
reported personalities. This operationalization is quite common in the teams literature
(e.g., Barrick, Stewart. Neubert, & Mount. 1998; Boone et al., 2005; Ellis et al., 2003;
LePine, 2003; Porter et al., 2003; Taggar, 2002), and Steiner (1972) advocated this
approach for additive tasks like the one in this study. This aggregation method, however,
is not without its critics. For example, Kozlowski and Klein (2000) suggested considering
the configuration of individual personalities rather than simply taking the average of
individual personalities. This method considers the statistical variance of individual
personality scores in addition to the mean, and has been used by some (e. g., Barrick,
Stewart, Neubert, & Mount. 1998; Boone et al., 2005). Testing the effects of personality
variance, however, implies theoretical questions about the diversity of personalities in the
team, which were not a focus of this study.

Additionally, Hofmann and Jones (2005) question whether team personality is a
valid construct for short-term ad hoc teams like the ones in this study. They suggest that
team personality is an emergent construct that develops as team members interact over
extended periods of time. It is possible that personality has very different effects on team
affect. Ieaming. and performance for teams that have interacted enough for a true team

personality to emerge.

83

Second, the task represented specific levels of task interdependence among the
team members. The level of interdependence was partly a function of the simulation itself
and partly a function of team structure. Teams in the functional structure were much
more interdependent than teams in the divisional structure. because each team member
did not have all of the assets necessary to disable every enemy track. Thus, certain team
members had to rely on other members to disable enemy tracks in their quadrants of the
restricted area. Nevertheless. it is not clear how other forms of task interdependence (e. g.,
sequential. pooled, reciprocal; Thompson. 1967), or other forms of team interdependence

( e. g., goal interdependence. outcome interdependence; Wageman, 2001) would affect the

results.
Post-hoe Inferences

On the basis of the hypothesis testing and additional post-hoe analyses, I revised
the proposed model into one that appears to be more in keeping with the empirical
results. Workload did not affect any of the outcomes in the individual portion of the
study, and this may be due to the limitation mentioned above—that individuals attacked
tracks outside of their quadrants and thus changed their workloads in unpredictable ways.
Therefore. I am reluctant to develop a post-hoe theoretical model for individuals working
alone on the basis of these null results. I do, however, offer a revision of my team model
that is both grounded in theory and consistent with my empirical findings. This post-hoe
model is illustrated in Figure 8; I note as well that it is not drastically different from the
original team model I proposed.

One of the primary changes in the model is the reconceptualization of

performance in terms of quantity and quality, rather than speed and accuracy. These

84

dimensions are conceptually very similar, but these dimensions more precisely fit the
empirical results. The quantity/quality distinction is well-established in organizational
behavior/industrial-organizational psychology research (e.g., Anderson, 1947; Muscio,
1912; O’Hara, Johnson. & Beehr, 1985). Quality represents the degree of productivity on
a task. and quality represents the extent to which errors are made on the task.

These dimensions are not entirely orthogonal; at the upper range of performance.
one must sacrifice one of the dimensions in order to achieve higher performance in the
other (Ilgen & Moore, 1987). This model, however, suggests that the dimensions can be
pulled apart by previous workload, even at moderate performance levels. When their past
workload has been high, teams will generally be highly productive but also make many
errors. When their past workload has been low, however, teams will generally be less
productive but will also make fewer errors.

The explanation for this phenomenon can be found in the two theoretical
mediators: learning and mental overload. Team Ieaming. as proposed in the original
model, suffers under conditions of high workload because of the difficulty of
communicating effectively under such conditions. When workload is high, teams cannot
focus on both taskwork—what the team members must accomplish individually—and
teamwork—what the team members must accomplish together. Team learning is
particularly problematic under these conditions because it requires the team members to
both share information they have gleaned and attend to information communicated by
other team members. This “dual-task paradigm”—attempting to focus both on one’s tasks
and on communicating with teammates—is extremely difficult to accomplish under

conditions of high workload, and team members will generally sacrifice communication

85

in order to accomplish their individual taskwork. When workload is low. however, team
members can accomplish both their individual taskwork and their collective teamwork.

Subsequent performance quality is a function of previous Ieaming. When teams
under conditions of low workload learn about their task and the environment in which
they are operating, they can convert this knowledge into better quality in the future. They
learn how to avoid mistakes on their tasks, and thus reduce their errors in the future.
Teams under high workload conditions, however, do not learn about their task and
environment, and thus are more likely to make errors in the future.

Mental overload is new to my model, but it has been examined in previous
workload research as “mental workload” or just “mental load.” Mental workload
concerns “the cost of mental operations, and the constraints that are imposed by these
costs on the ability of a performer to cope with the demands of a task that he or she is
given to perform” (Gopher, 2000, p. 197). More simply, it is “the amount of mental effort
needed to perform a task” (Brown & Boltz, 2002, p. 600). Mental workload has been
found to be negatively related to such outcomes as test performance (Croizet et al., 2004),
driving performance (Recarte & Nunes, 2003), and flight performance (Hardy &
Parasuraman, 1997).

In this model, however, the concern is not so much on the effects of workload on
concurrent mental overload, but with the effects of previous workload on subsequent
mental overload. Although team workload should affect concurrent mental overload (and
indeed it did in these data), the effect is stronger after the team’s workload has changed.
Increases in workload lead to higher levels of mental overload. and decreases in workload

lead to lower levels of mental overload. In effect, the team members’ degree of mental

86

overload is more affected by the contrast between their previous workload and their
current workload than by the raw level of workload. The team members get used to a
certain workload level. and establish their work patterns and interaction routines on the
basis of that level of workload. When the workload changes, these patterns and routines
are disrupted. If the workload increases, the team is overwhelmed and cannot be as
productive. In this way, low previous workload leads to higher subsequent mental
overload. which in tum leads to lower performance quantity. If the workload decreases,
however, team members have less work to which they need to devote their cognitive
resources, and hence experience less mental overload. They can continue working at the
highly productive pace from their previously high workload, and thus maintain high
performance quantity.

The relationships between previous team workload on the one hand, and Ieaming.
mental overload, quality and quantity of performance on the other, are moderated by
another new construct in my model: task novelty. Previous research has examined the
effects of task novelty on attentional processes (e.g., Ball & Zuckerrnan, 1992) with the
idea that people pay more attention to novel stimuli than to stimuli to which they have
become habituated. My use of the term, however, does not carry the connotation of a
novel task being more interesting than a familiar one; rather, task novelty simply captures
the degree to which a task presents learning opportunities to the team. Novel tasks, then,
are tasks that are not well-Ieamed by the team. and as the team Ieams about a novel task,
they make fewer errors (improve their performance quality). A task that is familiar to the
team. however. presents few learning opportunities; the team cannot learn much more

about the task and thus cannot reduce their error rate any further.

87

Thus, task novelty moderates the relationship between workload and Ieaming. in
that teams can learn a great deal about novel tasks but not about familiar ones. The
interactive effect of workload and task novelty on learning would be represented by a flat
slope when tasks are familiar (workload has little effect because there is little to be
learned), and a negative slope when the task is novel (higher workloads lead to less
learning). In tum, the relationship between previous workload and subsequent
performance quality would take the same form: no relationship for familiar tasks, but a
negative slope for novel tasks. This was represented in the data in this study by the
relationship between previous workload and friendly fire kills. Teams that had high
previous workloads committed more friendly fire kills of unknown (novel) tracks than
teams with low workloads, but there was no relationship between previous workload and
friendly fire kills of known (familiar) tracks.

Task novelty is also likely to moderate the relationships on the bottom half of my
model: the relationships of previous workload with mental overload and performance
quantity. 1 was not able to test the first of these relationships with the current data
because I only had a global measure of mental overload, but I was able to test the second
(performance quantity). Teams that had high previous workloads had more good attacks
of known (familiar) tracks than teams that had low previous workloads; for the unknown
(novel) tracks. however. there was no relationship between workload and good attacks.
The notion here is that teams under high workloads carry over their high rate of
productivity only if they are familiar with the task; when they have to work on novel
tasks. however. they cannot be as productive because they are not certain how to go about

the task.

88

Previous workload and task novelty are also likely to show an interactive effect
on mental workload. As noted above. when teams experience a decrease in workload,
they also experience lower levels of mental overload. If the team is facing a novel task,
however. their levels of mental workload are likely to be higher than if they are facing a
familiar task. The stress of having to perform a novel task is likely to negate the effect of
their previous workload. Thus. the nature of the interaction should be that previous
workload has a negative effect on mental workload for familiar tasks, but no relationship
for novel tasks.

Theoretical Implications

This study extends personality, affect, and teams research. In terms of personality,
this experiment demonstrated that two personality characteristics—extraversion and
neuroticism—interact with workload to affect both individuals’ emotional arousal and
Ieaming. Consistent with Mischel and Shoda’s (1995) cognitive-affective personality
system (CAPS), this study revealed that these personality characteristics did not exert a
deterministic main effect on emotions and Ieaming. Instead, there was a person by
situation interactions in terms of people’s responses to varying levels of workload. This
appears to be the kind of “if. . .then” dynamic Mischel and Shoda (1995) suggested, in
that personality traits have differential effects depending upon the situation.

The study also found disconfirming evidence for Eysenck’s (1967) theory of the
arousing effects of personality. Eysenck proposed that extraversion and neuroticism (and
a third trait he termed “psychoticism”) are related to emotional arousal. Specifically. he
suggested that people high in extraversion and neuroticism are more affected by

stimulating situations than people who are low on these traits. This study found. however.

89

that people high on these traits showed negative relationships between workload and
emotional arousal, whereas people low on these traits showed positive relationships.
Although this finding may be due to characteristics of the task and the experimental
environment, it at least suggests that the relationship between the traits and arousal is
more complex than Eysenck proposed. It may be that extraverts are only sensitive to
stimulating social situations, rather than to stimulating situations in general. Introverts, on
the other hand. may be more sensitive to stimulating non-social situations, like the one in
the individual portion of this study. People high in Neuroticism may be affected by
stressful situations by disengaging from the task and decreasing their arousal levels,
whereas people low in neuroticism see these situations as a challenge and increase their
arousal levels accordingly.

In terms of team structure, as noted by Hollenbeck et al. (2002), it is important to
match individuals to structure (internal fit), and structure to the environment (external
fit). Although this research did not directly examine internal fit, it did examine external
fit in terms of matching structure to workload demands. Extending the research on
resource allocation structures, this study examined the notion that distributing resources
among team members in a functional manner reduces the emotional arousal of the team
under conditions of high workload. As noted above, however, the results indicated that
this was only true when combined with another manipulation. Thus, it does not appear
that functional structures are necessarily associated with lower arousal levels.

Extending the transactive memory literature in teams by examining the concept of
transactive learning, the experiment tested a simple method for overcoming the negative

effects of high workload on team learning through the assignment of responsibility for

90

learning about different aspects of the task. On the assumption that this would distribute
the cognitive load throughout the team and thus effectively reduce the load on any
individual team member. I expected that these teams would learn more effectively than
teams where learning responsibility was not distributed. The results did not support this
for teams as a whole, but a post-hoe analysis of specific responses by team members
revealed that the manipulation did have some significant effects. Thus. the notion that
teams can learn more effectively by distributing responsibility for learning among the
team members may bear more examination.

In terms of affect. two findings from this study may have implications for theory.
First, although arousal and hedonic tone have been conceptualized as somewhat distinct
dimensions of affect (Russell, 1980), these data showed a strong correlation between
measures of these dimensions. This casts doubt either on the dimensional perspective or
on the construct validity of the measures. Second, although previous research has found
heart rate to be a valid measure of emotional arousal, this study found that heart rate was
not related to the self-report measures of arousal, nor was it affected by any of the
experimental manipulations. It is possible that the sedentary nature of the task (sitting at a
computer) contributed to the null effects, but other seemingly sedentary tasks have found
heart rate effects (e. g., Arena et al.. 1983). At the very least. the results of this study
suggest that heart rate may not be a construct valid measure of arousal across types of
tasks.
Practical Applications

Two practical applications are evident from the results that may provide

information for managers on how much of a workload to assign to new individuals and

91

teams. For teams. if learning about the task is an important and desired outcome,
managers would be advised to give teams low early workloads. This would allow the
teams to take the extra time and cognitive resources needed for communicating
information to each other that will enable them to learn about their task. The evidence
from this study suggests that high workloads negatively affect teams in a lasting manner.
such that they are not able to learn well even when their workload decreases. Starting
with a low workload may be particularly critical for teams that have just formed, like the
teams in this study. Previous research on entrainment in teams (Aneona & Chong. 1996)
has suggested found that teams establish norms of communication that carry over and
affect their future performance (Johnson et al. 2006: Moon et al.. 2004). Thus, if newly
formed teams start out with a high workload. they may develop communication norms
that are inadequate for effective team Ieaming.

For individuals who are working alone, however, the amount of workload a
manager should assign depends upon the personality of the employee. Specifically,
managers would be advised to ascertain the level of neuroticism of their employees prior
to assigning new tasks. Employees that are low in neuroticism appear to be engaged by
high workloads (as evidenced by their emotional arousal levels) and learn well due to the
increased information available. Employees that are high in neuroticism. however, appear
to disengage from the task and do not learn well when workload is high. For these

employees. lower workloads may be advisable until they gain mastery on the task.

REFERENCES

Dispatcher’s errors caused fatal crash. NTSB rules. (1998, May 21). Wall Street Journal.

The psychiatric injury liabilities of excessive workloads. (2005. May 8). Human
Resources Magazine.

Ancona. D.. & Chong, C. L. (1996). Entrainment: Pace. cycle. and rhythm in
organizational behavior. In B. M. Staw & L. L. Cummings (Eds). Research in
organizational behavior: An annual series of analytical essays and critical
reviews (Vol. 18, pp. 251-284). Stamford, CT: JAI Press.

Arena. J. G.. & et al. (1983). Reliability of psychophysiological assessment. Behaviour
Research & Therapy. 21, 447-460.

Argote, L., Gruenfeld, D., & Naquin, C. (2001). Group learning in organizations. In M.

E. Turner (Ed.), Groups at work: Theory and research (pp. 369-411). Mahwah,
NJ: USum Associates.

Arrow, H., McGrath. J. E., & Berdahl, J. L. (2000). Small groups as complex systems:

F ormation, coordination. development. and adaptation. Thousand Oaks, CA:
Sage.

Baddeley, A. D. (2002). Is working memory still working? European Psychologist. 7. 85-
97.

Ball, S. A.. & Zuckerrnan, M. (1992). Sensation seeking and selective attention: Focused
and divided attention on a dichotic listening task. Journal of Personality and
Social Psychology. 63. 825-831.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall.

Bandura. A.. & Jourden, F. J. (1991). Self-regulatory mechanisms governing the impact
of social comparison on complex decision making. Journal of Personality &
Social Psychology. 60. 941-951.

Barker, J. R. (1993). Tightening the iron cage: Concertive control in self-managing
teams. Administrative Science Quarterly. 38, 408-437.

Barrick, M. R., & Mount. M. K. (1991). The big five personality dimensions and job
performance: A meta-analysis. Personnel Psychology, 44. 1-26.

Barrick, M. R.. Stewart, G. L., Neubert, M. J., & Mount. M. K. (1998). Relating member

ability and personality to work-team processes and team effectiveness. Journal of
Applied Psychology. 83. 377-391.

93

Barsade. S. G. (2002). The ripple effect: Emotional contagion and its influence on group
behavior. Administrative Science Quarterly. 47. 644-675.

Barsade. S. G.. Brief. A. P.. & Spataro, S. E. (2003). The affective revolution in
organizational behavior: The emergence of a paradigm. In J. Greenberg (Ed.).
Organizational behavior: The state of the science (2nd ed.). Mahwah, NJ:
Lawrence Erlbaum Associates.

Bartel. C. A.. & Saavedra, R. (2000). The collective construction of work group moods.
Administrative Science Quarterly. ~15. 197-231.

Beersma, B.. Hollenbeck. J. R., Humphrey, S. E., Moon, H., Conlon, D. E., & Ilgen. D.
R. (2003). Cooperation. competition, and team performance: Toward a
contingency approach. Academy of Management Journal. 46. 572-590.

Berkowitz. L.. & Donnerstein. E. (1982). External validity is more than skin deep: Some

answers to criticisms of laboratory experiments. American Psychologist, 37, 245-
257.

Berlew. D. E., & Hall. D. T. (1966). The socialization of managers: Effects of
expectations on performance. Administrative Science Quarterly. 11. 207-223.

Berlyne. D. E. (1960). Conflict. arousal. and curiosity. New York, NY, US: McGraw-
Hill.

Berlyne. D. E. (1964). Emotional aspects of Ieaming. In P. R. Farnsworth, O. McNemar
& et al. (Eds.), Annual review of psychology: XV (pp. 115- 142): Oxford, England.

Bettenhausen, K.. & Mumighan, J. K. (1985). The emergence of norms in competitive
decision-making groups. Administrative Science Quarterly, 30. 350-372.

Boone. C., Van Olffen, W., & Van Witteloostuijn, A. (2005). Team locus-of-control
composition, leadership structure, information acquisition. and financial

performance: A business simulation study. Academy of Management Journal. 48,
889-909.

Bowers. C.. Thornton, C.. Braun, C.. Morgan, B. B.. Jr., & Salas. E. (1998). Automation.
task difficulty, and aircrew performance. Military Psychology, 10, 259-274.

Bowers, C. A.. Braun, C. C., & Morgan, B. B., Jr. (1997). Team workload: Its meaning
and measurement. In M. T. Brannick, E. Salas & et al. (Eds), Team performance

assessment and measurement: Theory. methods. and applications (pp. 85-108):
Mahwah, NJ. USum Associates. Publishers. 1997. xii, 370.

Brown, S. W., & Boltz, M. G. (2002). Attentional processes in time perception: Effects of

mental workload and event structure. Journal of Experimental Psychology:
Human Perception & Performance. 28. 600-615.

94

Cacioppo, J. T., Klein, D. J., Bemtson. G. G.. & Hatfield. E. (1993). The
psychophysiology of emotion. In M. Lewis & J. M. Haviland (Eds), Handbook of
emotions (pp. 119-142): New York, NY. US, 1993, xiii, 653.

Campbell. D. J., & Ilgen. D. R. (1976). Additive effects of task difficulty and goal setting
on subsequent task performance. Journal oprplied Psychology. 61, 319-324.

C ampbell. D. T. (1958). Common fate. similarity, and other indices of the status of
aggregates of persons as social entities. Behavioral Science, 3. 14-25.

Campion, M. A. & McClelland, C. L. (1991). Interdisciplinary examination of the costs
and benefits of enlarged jobs: A job design quasi-experiment. Journal of Applied
Psychology, 76, 186-198.

Cannon—Bowers. .l. A.. Salas, E., Blickensderfer, E., & Bowers, C. A. (1998). The impact
of cross-training and workload on team functioning: A replication and extension
of initial findings. Human Factors. 40, 92-101.

Carruthers, M. (1981). Field studies: Emotion and beta-blockade. In M. J. C. P. G.

Mellett (Ed.), Foundations of psychosomatics (pp. 223-241). Chichester, England:
Wiley.

Chen. G., Webber. S. S.. Bliese, P. D., Mathieu. J. E., Payne, S. C., Born. D. H.. et al.
(2002). Simultaneous examination of the antecedents and consequences of
efficacy beliefs at multiple levels of analysis. Human Performance. 15, 381-410.

Chessick, R. D.. Bassan, M., & Shattan, S. (1966). A comparison of the effect of infused

catecholamines and certain affect states. American Journal of Psychiatry. 123,
156-165.

Cohen, J ., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the
behavioral sciences (2nd ed.). Mahwah, NJ: Lawrence Erlbaum Associates.

Cook, S. W., & Selltiz, C. (1964). A multiple-indicator approach to attitude
measurement. Psychological Bulletin. 62, 36-55.

Costa, P. T., Busch. C. M., Zonderman, A. B.. & McCrae, R. R. (1986). Correlations of
MMPI factor scales with measures of the five factor model of personality. Journal
of Personality Assessment. 50. 640-650.

Costa, P. T., & McCrae, R. R. (1992). Revised NEO personality inventory. Odessa. FL:
Psychological Assessment Resources.

Croizet, J .-C., Despraes. G., Gauzins. M.-E., Huguet, P., Leyens. J .-P.. & Mfieot. A.
(2004). Stereotype threat undermines intellectual performance by triggering a
disruptive mental load. Personality and Social Psychology Bulletin. 30. 721-731.

95

Diener. E., Lusk, R.. DeFour. D.. & Flax. R. (1980). Deindividuation: Effects of group
size, density. number of observers. and group member similarity on self-
consciousness and disinhibited behavior. Journal of Personality & Social
Psychology. 39, 449-459.

Dienstbier. R. A. (1989). Arousal and physiological toughness: Implications for mental
and physical health. Psychological Review. 96, 84-100.

Edmondson. A. C. (1996). Learning from mistakes is easier said than done: Group and
organizational influences on the detection and correction of human error. Journal
oprplied Behavioral Science. 32, 5-28.

Edwards. J. R.. & Parry. M. E. (1993). On the use of polynomial regression equations as
an alternative to difference scores in organizational research. Academy of
Management Journal. 36, 1577-1613.

Ellis, A. P. J., Hollenbeck, J. R.. Ilgen. D. R.. Porter, C. O. L. H., West, B. J., & Moon.

H. (2003). Team learning: Collectively connecting the dots. Journal of Applied
Psychology. 88, 821-835.

Entin. E. E., & Serfaty. D. (1999). Adaptive team coordination. Human Factors. 4]. 312-
325.

Erez, M. (1990). Performance quality and work motivation. In U. Kleinbeck, H.-H. Quast
& et al. (Eds), Work motivation (pp. 53-65). Hillsdale, NJ: Englandiates, Inc.

Eysenck, H. J. (1967). The biological basis of personality.Thomas: Spring-field, Ill.

Eysenck, H. J. (1994). Personality: Biological foundations. In P. A. Vernon (Ed.), The
neuropsychology of individual differences. San Diego, CA, US: Academic Press.

Eysenck. H. J. (1997). Personality and experimental psychology: The unification of

psychology and the possibility of a paradigm. Journal of Personality & Social
Psychology, 73, 1224-1237.

Eysenck. M. W. (1981). Learning, memory, and personality. In H. J. Eysenck (Ed.), A
model for personality (pp. 169-209). Berlin. Germany: Springer-Verlag.

Feldman Barrett, L.. & Russell, J. A. (1998). Independence and bipolarity in the structure
of current affect. Journal of Personality & Social Psychology. 74, 967-984.

Fiore, S. M., & Schooler, J. W. (2004). Process mapping and shared cognition:
Teamwork and the development of shared problem models. In E. Salas & S. M.
Fiore (Eds), Team cognition: Understanding the factors that drive process and
performance (pp. 133-152). Washington, DC: USical Association.

Fisher. S. L., & Ford. J. K. (1998). Differential effects of learner effort and goal
orientation on two learning outcomes. Personnel Psychology. 51, 397-420.

96

Frodi, A. ( I978). Experiential and physiological responses associated with anger and
aggression in women and men. Journal of Research in Personality. 12, 335-349.

Froggatt. K. L.. & Cotton. J. L. (1987). The impact of type a behavior pattern on role
overload-induced stress and performance attributions. Journal of Management.
13. 87-98.

Gaillard. A. W. (1993). Comparing the concepts of mental load and stress. Ergonomics.
36. 991-1005.

Gale. A. (1973). The psyt'hophysiology of individual differences: Studies of extraversion
and the EEG. New York: Wiley.

Ganster, D. C., Fox. M. L.. & Dwyer. D. J. (2001). Explaining employees' health care
costs: A prospective examination of stressful job demands, personal control, and
physiological reactivity. Journal of Applied Psychology. 86. 954-964.

Geen. R. G. (1991). Social motivation. In M. R. Rosenzweig & L. W. Porter (Eds.),
Annual review ofpsychology (Vol. 42, pp. 377-399). Palo Alto, CA: USnC.

George. J. M. (2002). Affect regulation in groups and teams. In R. G. Lord. R. J.
Klimoski & R. Kanfer (Eds.), Emotions in the workplace (pp. 183-218). San
Francisco: Jossey-Bass.

Gilliland. S. W., & Landis, R. S. (1992). Quality and quantity goals in a complex
decision task: Strategies and outcomes. Journal of Applied Psychology. 77, 672-
681.

Gopher. D. 2000. Mental workload. In A. E. Kazdin (Ed.), Encyclopedia ofpsychology,
Vol. 5: 197-198. Washington, DC: American Psychological Association.

Gopher, D.. & Donchin. E. (1986). Workload: An examination of the concept. In K. R.
Boff, L. Kaufman & et al. (Eds.), Cognitive processes and performance (Vol. 2.
pp. 1-49): Oxford, England.

Gopher, D., & Koriat, A. (1999). Attention and performance X VII: Cognitive regulation

ofperformance: Interaction of theory and application. Cambridge, MA, US: The
MIT Press.

Greenglass, E. R.. Burke, R. J., & Moore. K. A. (2003). Reactions to increased workload:

Effects on professional efficacy of nurses. Applied Psychology: An International
Review. 52. 580-597.

Grossberg. J. M., & Wilson, H. K. (1968). Physiological changes accompanying the

visualization of fearful and neutral situations. Journal of Personality & Social
Psychology. 10, 124-133.

97

Guglielmi, R. S. (1999). Psychophysiological assessment of prejudice: Past research,
current status, and future directions. Personality & Social Psychology Review, 3.
123- 157.

Hardy, D. J., & Parasuraman, R. (1997). Cognition and flight performance in older pilots.
Journal of Experimental Psychology: Applied. 3, 313-348.

Hart, S. G.. & Staveland. L. E. (1988). Development of NASA-TLX (task load index):
Results of empirical and theoretical research. In P. A. Hancock & N. Meshkati
(Eds.), Human mental workload (pp. 139-183): Oxford. England, 1988, xvi, 382.

Hill. G. W. (1982). Group versus individual performance: Are n + 1 heads better than
one? Psychological Bulletin. 91, 517-539.

Himle. D. P., J ayaratne. S., & Thyness. P. (1991). Buffering effects of four social support
types on burnout among social workers. Social Work Research & Abstracts. 27,
22-27.

Hinsz, V. B. (1990). Cognitive and consensus processes in group recognition memory
performance. Journal of Personality & Social Psychology. 5 9, 705-718.

Hinsz, V. B., Tindale, R. S., & Vollrath, D. A. (1997). The emerging conceptualization of
groups as information processes. Psychological Bulletin. 121, 43-64.

Hofmann, D. A.. & Jones. L. M. (2005). Leadership, collective personality, and
performance. Journal of Applied Psychology. 90, 509-522.

Hollenbeck, J. R. (2000). A structural approach to external and internal person-team fit.
Applied Psychology: An International Review, 49, 534-549.

Hollenbeck. l. R., Moon, H., Ellis, A. P. J., West, B. J., Ilgen, D. R.. Sheppard, L.. et al.
(2002). Structural contingency theory and individual differences: Examination of
external and internal person-team fit. Journal of Applied Psychology. 87, 599-606.

Holman, D. J., & Wall, T. D. (2002). Work characteristics, leaming-related outcomes,
and strain: A test of competing direct effects. mediated. and moderated models.
Journal of Occupational Health Psychology. 7. 283-301.

Houkes, I.. Janssen, P. P. M., Jonge, J. d., & Nijhuis, F. J. N. (2001). Specific
relationships between work characteristics and intrinsic work motivation, burnout
and tumover intention: A multi-sample analysis. European Journal of Work &
Organizational Psychology. 10, 1-23.

House, R.. Rousseau, D. M., & Thomashunt, M. (1995). The meso paradigm - a
framework for the integration of micro and macro organizational-behavior. In
Research in organizational behavior: An annual series of analytical essays and
critical reviews (Vol. 17. pp. 71-114).

98

Huey, B. M., Wickens, C. D.. & National Research Council Commission on Behavioral
& Social Sciences & Education. (1993). Workload transition: Implicationsfirr
individual and team performance. Washington, DC, US: National Academy
Press.

Hurst. M. W., & Rose. R. M. (1978). Objective job difficulty, behavioural response, and
sector characteristics in air route traffic control centres. Ergonomics. 21. 697-708.

Hyatt. D. E., & Ruddy. T. M. (1997). An examination of the relationship between work

group characteristics and performance: Once more into the breech. Personnel
Psychology. 50, 553-585.

Ilgen, D. R.. Hollenbeck. J. R.. Johnson, M., & Jundt, D. (2005). Teams in organizations:
From input-process-output models to IMOI models. Annual Review of
Psychology, 56. 517-543.

Ilgen, D. R. & Moore. C. F. 1987. Types and choices of performance feedback. Journal
of Applied Psychology, 72(3): 401—406.

Jackson, J. M., & Latanae. B. (1981). All alone in front of all those people: Stage fright
as a function of number and type of co-perforrners and audience. Journal of
Personality & Social Psychology. 40, 73-85.

J ackson. J. M., & Williams. K. D. (1985). Social loafing on difficult tasks: Working
collectively can improve performance. Journal of Personality & Social
Psychology. 49, 937-942.

J acobs, S. R.. & Dodd, D. K. (2003). Student burnout as a function of personality, social
support. and workload. Journal of College Student Development, 44. 291-303.

J arnes. W. (1884). What is an emotion? Mind. 9, 188-205.

Johnson. M. D., Hollenbeck, J. R.. Humphrey, S. E., Ilgen, D. R.. J undt. D., & Meyer, C.
J. (2006). Cutthroat cooperation: Asymmetrical adaptation to changes in team
reward structures. Academy of Management Journal, 49. 103-119.

Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice Hall.

Kahneman, D., Ben-Ishai. R.. & Lotan, M. (1973). Relation of a test of attention to road
accidents. Journal oprplied Psychology. Vol. 58, 113-115.

Kakimoto, Y., Nakamura, A., Tarui, H., Nagasawa, Y., & et al. (1988). Crew workload in
JASDF C-l transport flights: I change in heart rate and salivary cortisol. Aviation.
Space. & Environmental Medicine, 59, 511-516.

Karasek. R. (1979). Job demands. job decision latitude. and mental strain: Implications
for job redesign. Administrative Science Quarterly, 24. 285-308.

99

Karasek, R.. & Theorell, T. (1990). Healthy work: Stress. productivity. and the
reconstruction of working life. New York: Basic Books.

Kauffeld, 5.. Jonas, E., & Frey. D. (2004). Effects of a flexible work-time design on
employee- and company-related aims. European Journal of Work &
Organizational Psychology. I 3. 79- 100.

Kelly. J. R.. & Barsade. S. G. (2001). Mood and emotions in small groups and work
teams. Organizational Behavior & Human Decision Processes. 86, 99-130.

Kinicki, A. J., McKee. F. M., & Wade. K. J. (1996). Annual review. 1991-1995:
Occupational health. Journal of Vocational Behavior. 49. 190-220.

Kluger. A. N.. & DeNisi. A. (1996). Effects of feedback intervention on performance: A

historical review. a meta—analysis. and a preliminary feedback intervention theory.
Psychological Bulletin. 119, 254-284.

Knight. M. L., & Borden, R. J. (1979). Autonomic and affective reactions of high and

low socially-anxious individuals awaiting public performance. Psychophysiology.
16. 209-213.

Kozlowski, S. W. J., Gully, S. M., Nason. E. R.. & Smith, E. M. (1999). Developing
adaptive teams: A theory of compilation and performance across levels and time.
In D. R. 11 gen & E. D. Pulakos (Eds. ) The changing nature of performance (pp.
240-292), San Francisco: Jossey-Bass.

Kozlowski. S. W. J ., & Klein, K. J. (2000). A multilevel approach to theory and research
in organizations: Contextual, temporal, and emergent processes. In K. J. Klein &
S. W. J. Kozlowski (Eds.), Multilevel theory. research. and methods in

organizations: Foundations. extensions. and new directions. San Francisco. CA,
US: Jossey-Bass.

Kramer. R. M., & Brewer. M. B. (1984). Effects of group identity on resource use in a
simulated commons dilemma. Journal of Personality & Social Psychology. 46.
1044-1057.

Larsen, R. J ., & Diener. E. (1985). A multitrait-multimethod examination of affect

structure: Hedonic level and emotional intensity. Personality & Individual
Differences. 6. 631-636.

Larson, 1. R.. & Christensen. C. (1993). Groups as problem-solving units: Toward a new
meaning of social cognition. British Journal of Social Psychology. 32, 5-30.

Laughlin, P. R.. Bonner, B. L.. & Miner, A. G. (2002). Groups perform better than the
best individuals on letters-to-numbers problems. Organizational Behavior and
Human Decision Processes. 88, 605-620.

100

Laughlin. P. R.. VanderStoep. S. W., & Hollingshead, A. B. (1991). Collective versus
individual induction: Recognition of truth, rejection of error, and collective
information processing. Journal of Personality & Social Psychology. 61 , 50-67.

LeBlanc. J .. Ducharme. M. B.. & Thompson, M. (2004). Study on the correlation of the
autonomic nervous system responses to a stressor of high discomfort with
personality traits. Physiology & Behavior. 82, 647-652.

Le Bon. G. (1895). The crowd. a study of the popular mind ([13th impression] ed.).
London: T.F. Unwin.

LePine, J. A. (2003). Team adaptation and postchange performance: Effects of team
composition in terms of members’ cognitive ability and personality. Journal of
Applied Psychology. 88, 27-39.

LePine, J. A.. Hollenbeck. J. R.. Ilgen, D. R.. & Hedlund, J. (1997). Effects of individual
differences on the performance of hierarchical decision-making teams: Much
more than g. Journal oprplied Psychology. 82, 803-811.

LePine, J. A., LePine, M. A., & J ackson, C. L. (2004). Challenge and hindrance stress:
Relationships with exhaustion. motivation to learn. and learning performance.
Journal oprplied Psychology. 89, 883-891.

Levenson, R. W. (1988). Emotion and the autonomic nervous system: A prospectus for
research on autonomic specificity. In H. L. Wagner (Ed.), Social
psychophysiology and emotion: Theory and clinical applications (pp. 17-42).
Oxford. England.

Locke, E. A.. & Latham, G. P. (1990). A theory of goal setting & task performance.
Upper Saddle River, NJ . US: Prentice-Hall, Inc.

MacMillan, J .. Entin, E. E., & Serfaty, D. (2004). Communication overhead: The hidden
cost of team cognition. In E. Salas & S. M. Fiore (Eds.), Team cognition:
Understanding the factors that drive process and performance (pp. 61-82).
Washington, DC: USical Association.

Mangan, G. L., & Hookway, D. (1988). Perception and recall of aversive material as a
function of personality type. Personality & Individual Differences. 9, 289-295.

Markel, K. S., & Frone, M. R. (1998). Job characteristics, work-school conflict, and
school outcomes among adolescents: Testing a structural model. Journal of
Applied Psychology. 83, 277-287.

Marsh. H. W. (2001). Distinguishing between good (useful) and bad workloads on

students' evaluations of teaching. American Educational Research Journal. 38,
183-2 12.

101

Marsh, H. W., & Roche, L. A. (2000). Effects of grading leniency and low workload on
students' evaluations of teaching: Popular myth. bias. validity. or innocent
bystanders? Journal ofEducational Psychology. 92. 202-228.

Maslach. C., & Jackson, S. E. (1984). Burnout in organizational settings. Applied Social
Psychology Annual. 5, 133-153.

Matthews, G. U.. & Gilliland. K. (1999). The personality theories of H. J. Eysenck and J.

A. Gray: A comparative review. Personality and Individual Differences. 26, 583-
626.

Mayer, R. E. (2004). Teaching of subject matter. Annual Review of Psychology. 55. 715-
744.

McGrath. J. E. (1976). Stress and behavior in organizations. In M. D. Dunnette (Ed.),

Handbook of industrial and organizational psychology (pp. 1351-1395). Chicago:
Rand McNally.

McGrath, J. E., Kelly, J. R.. & Machatka. D. E. (1984). The social psychology of time:
Entrainment of behavior in social and organizational settings. Applied Social
Psychology Annual. 5. 21-44.

Miller. D. L.. Young, R. Kleinman, D., & Serfaty, D. (1998). Distributed Dynamic

Decision-making simulation phase I release notes and user's manual.Woburn,
MA: Aptima, Inc.

Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality:
Reconceptualizing situations, dispositions. dynamics, and invariance in
personality structure. Psychological Review. 102, 246-268.

Moon, H., Hollenbeck, J. R.. Humphrey. S. E., Ilgen, D. R.. West, B. J ., Ellis, A. P. J .. et
al. (2004). Asymmetric adaptability: Dynamic team structures as one-way streets.
Academy of Management Journal. 47, 681-696.

Moreland, R. L.. Argote, L., & Krishnan, R. (1996). Socially shared cognition at work:
Transactive memory and group performance. In J. L. Nye & A. M. Brower (Eds.),
What’s social about social cognition? Research on socially shared cognition in
small groups (pp. 57-84). Thousand Oaks, CA: USical Associates.

Morgan, B. B., Salas. E., & Glickman, A. S. (1993). An analysis of team evolution and
maturation. Journal of General Psychology. 120, 277-291.

Morgeson, F. P., & Hofmann. D. A. (1999). The structure and function of collective
constructs: Implications for multilevel research and theory development. Academy
ofManagement Review. 24, 249-265.

Morrison, R. F.. & Brantner. T. M. (1992). What enhances or inhibits learning a new job?
A basic career issue. Journal of Applied Psychology. 7 7 , 926-940.

102

Navon. D.. & GOpher. D. (1979). On the economy of the human-processing system.
Psychological Review. 86, 214-255.

N inio. A., & Kahneman. D. (1974). Reaction time in focused and in divided attention.
Journal of Experimental Psychology. 103, 394-399.

Orpen, C. (1974). The effect of expectations of managerial job performance: A four-year
longitudinal study with South African business managers. Joumal of Social
Psychology. Vol. 93. 135-136.

Orpen, C. (1994). The effect of initial job challenge on subsequent performance among

middle managers: A longitudinal study. Psychology: A Journal ofHuman
Behavior. 31, 51-52.

Parker, 5. K.. & Sprigg. C. A. (1999). Minimizing strain and maximizing learning: The
role of job demands. job control, and proactive personality. Journal of Applied
Psychology. 84. 925-939.

Parkes. K. R. (1995). The effects of objective workload on cognitive performance in a

field setting: A two-period cross-over trial. Applied Cognitive Psychology. 9.
$153-$171.

Penton-Voak. I. S.. Edwards, H., Percival, A., & Wearden, J. H. (1996). Speeding up an
internal clock in humans? Effects of click trains on subjective duration. Journal of
Experimental Psychology: Animal Behavior Processes. 22, 307-320.

Perrewe, P. L.. & Ganster, D. C. (1989). The impact of job demands and behavioral

control on experienced job stress. Journal of Organizational Behavior, 10, 213-
229.

Porter. C. O. L. H., Hollenbeck. J. R.. Ilgen. D. R.. Ellis, A. P. J., West. B. J., & Moon.
H. (2003). Backing up behaviors in teams: The role of personality and legitimacy
of need. Journal oprplied Psychology. 88. 391-403.

Recarte. M. A., & Nunes, L. M. (2003). Mental workload while driving: Effects on visual
search, discrimination, and decision making. Journal of Experimental
Psychology: Applied. 9, 119-137.

Reid, G. B., & N ygren. T. E. (1988). The subjective workload assessment technique: A
scaling procedure for measuring mental workload. In P. A. Hancock & N.

Meshkati (Eds.), Human mental workload (pp. 185-218): Oxford. England. 1988.
xvi, 382.

Revelle. W., Humphreys, M. 8.. Simon. L., & Gilliand. K. (1980). The interactive effect

of personality. time of day. and caffeine: A test of the arousal model. Journal of
Experimental Psychology: General. 109. 1-31.

103

Rose, R. M.. & Fogg, L. F. (1993). Definition of a responder: Analysis of behavioral.
cardiovascular, and endocrine responses to varied workload in air traffic
controllers. Psychosomatic Medicine. 55, 325-338.

Russell. J. A. (1980). A circumplex model of affect. Journal ofPersonality & Social
Psychology. 39. 1161-1178.

Russell, J. A.. & Mehrabian. A. (1974). Distinguishing anger and anxiety in terms of
emotional response factors. Journal of Consulting & Clinical Psychology. Vol. 42,
79-83.

Sanna, L. J., & Parks, C. D. (1997). Group research trends in social and organizational

psychology: Whatever happened to intragroup research? Psychological Science.
8. 261-267.

Santora. M. (2005. May 19. 2005). Clerk’s workload blamed for hospital test mix-up.
New York Times. p. l.

Schachter, S., & Singer. J. (1962). C ognitive, social. and physiological determinants of
emotional state. Psychological Review, 69, 379-399.

Schor. J. (2002, September 2. 2002). Why Americans should rest. New York Times. p. 15.

Schwartz. J. (2004). Always on the job. employees pay with health. New York Times. p.
l.

Seta, J. J., & Schkade. J. K. (1976). Effects of group size and proximity under
cooperative and competitive conditions. Journal ofPersonality & Social
Psychology. 34, 47-53.

Shaw, J. B.. & Weekley, J. A. (1985). The effects of objective work-load variations of

psychological strain and post-work-load performance. Journal of Management.
I I. 87-98.

Somech. A., & Miassy-Maljak, N. (2003). The relationship between religiosity and
bumout of principals: The meaning of educational work and role variables as
mediators. Social Psychology of Education. 6, 61-90.

Spector, P. E., Dwyer. D. J., & Jex, S. M. (1988). Relation of job stressors to affective.
health. and performance outcomes: A comparison of multiple data sources.
Journal oprplied Psychology. 73, 11-19.

Stasser. 6., Stewart. D. D., & Wittenbaum. G. M. (1995). Expert roles and information

exchange during discussion: The importance of knowing who knows what.
Journal of Experimental Social Psychology. 31. 244-265.

104

Stasser. G.. & Titus, W. (1987). Effects of information load and percentage of shared
information on the dissemination of unshared information during group
discussion. Journal of Personality & Social Psychology. 53 . 81-93.

Steiner, I. D. (1972). Group process and productivity. New York: Academic Press.

Stewart. D. D.. & Stasser. G. (1995). Expert role assignment and information sampling
during collective recall and decision making. Journal of Personality & Social
Psychology. 69, 619-628.

Taggar, S. (2002). Individual creativity and group ability to utilize individual creative
resources: A multilevel model. Academy of Management Joumal. 45, 315-330.

Taris, T. W., & Feij, J. A. (2004). Learning and strain among newcomers: A three-wave
study on the effects of job demands and job control. Journal of Psychology:
Interdisciplinary & Applied. 138, 543-563.

Taylor, M. S. (1981). The motivational effects of task challenge: A laboratory
investigation. Organizational Behavior & Human Performance. 27 , 255-278.

Thayer, R. E. (1970). Activation states as assessed by verbal report and four
psychophysiological variables. Psychophysiology. Vol. 7, 86-94.

Thompson, J. D. (1967). Organizations in action: Social science bases ofadrninistrative
theory. New York, NY, US: McGraw-Hill.

Tourangeau. R.. & Ellsworth, P. C. (1979). The role of facial response in the experience
of emotion. Journal ofPersonality & Social Psychology, 3 7, 1519-1531.

Treisman, M., Cook, N., Naish, P. L. N., & MacCrone, J. K. (1994). The internal clock:
Electroencephalographic evidence for oscillatory processes underlying time

perception. Quarterly Journal of Experimental Psychology: Human Experimental
Psychology. 47, 241-289.

Tsang, P. S.. & Vidulich, M. A. (2003). Principles and practice ofaviation psychology.
Mahwah, NJ , US: Lawrence Erlbaum Associates.

Urban, J. M., Weaver. J. L.. Bowers, C. A.. & Rhodenizer. L. (1996). Effects of workload
and structure on team processes and performance: Implications for complex team
decision making. Human Factors. 38, 300-310.

Vahtera. J ., Kivimeaki. M.. Pentti, J .. Linna. A.. Virtanen. M., Virtanen. P., et al. (2004).
Organisational downsizing, sickness absence, and mortality: lO-town prospective
cohort study. British Medical Journal. 328, 555.

Vahtera. J ., Pentti, J .. & Uutela, A. (1996). The effect of objective job demands on

registered sickness absence spells: Do personal. social and job-related resources
act as moderators? Work & Stress. 10. 286-308.

105

Wageman, R. 2001. The meaning of interdependence. In M. E. Turner (Ed.), Groups at
work: Theory and research. Mahwah, NJ . US: Lawrence Erlbaum Associates.

Wallace. J. F., & Newman. J. P. (1998). Neuroticism and the facilitation of the automatic
orienting of attention. Personality & Individual Differences. 24, 253-266.

Watson. D.. Clark, L. A.. & Tellegen, A. (1988). Development and validation of brief
measures of positive and negative affect: The PANAS scales. Journal of
Personality & Social Psychology. 54, 1063-1070.

Wearden, J. H., & Penton-Voak, I. S. (1995). Feeling the heat: Body temperature and the
rate of subjective time, revisited. Quarterly Joumal of Experimental Psychology:
Comparative & Physiological Psychology. 48, 129-141.

Weiss, H. M., & Cropanzano. R. (1996). Affective events theory: A theoretical
discussion of the structure, causes and consequences of affective experiences at
work. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational

behavior: An annual series of analytical essays and critical reviews (Vol. 18, pp.
1-74). Stamford, CT. US: JAI Press.

Westerlund, H., Ferrie. J ., Hagberg, J ., J eding. K.. Oxenstiema, G., & Theorell, T.
(2004). Workplace expansion. long-term sickness absence, and hospital
admission. Lancet, 363. 1193-1197.

Wiggins, J. S.. & Pincus, A. L. (1992). Personality: Structure and assessment. Annual
Review of Psychology. 43 , 473-504.

Woodworth. R. S. (1899). The accuracy of voluntary movement. Psychological Review,
3. 1-119.

Xie, B., & Salvendy, G. (2000). Review and reappraisal of modelling and predicting

mental workload in single- and multi-task environments. Work & Stress. 14. 74-
99.

Yerkes, R. M., & Dodson, J. D. (1908). The relation of strength of stimulus to rapidity of
habit formation. Journal of Comparative Neurology & Psychology. 18, 459-482.

Yeung, S. S.. Genaidy, A., Deddens, J .. & Sauter, S. (2005). The relationship between
protective and risk characteristics of acting and experienced workload, and
musculoskeletal disorder cases among nurses. Journal ofSafety Research. 36, 85-

95.

Yik. M. S. M., Russell, J. A.. & Barrett. L. F. (1999). Structure of self-reported current
affect: Integration and beyond. Journal ofPersonality & Social Psychology. 7 7 .
600-619.

Zajonc, R. B. (1965). Social facilitation. Science. 149, 269-274.

106

APPENDIX

Table 1. Means and standard deviations.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Teams Individuals

M 5.0. M 5.12.

1. Workload 3.03 1.46 2.99 1.45

2. Team structure .51 .50 - -

3. Transactive learning .53 .50 - -

4. Neuroticism 2.46 .29 2.55 .62

5. Extraversion 3.70 .25 3.70 .57

q, 6. CMQ statements 2.89 .38 2.86 .79
5.5. 7. CMQ adjectives 2.67 .33 2.60 .69
8 8. Heartrate 75.53 6.55 75.50 8.80
D“ 9. Unknown tracks 1.06 .65 .93 .80
10. NASA-TLX 3.05 .43 3.01 .76

l l. CMQ statements 3.50 .36 3.04 .90

:5 12. CMQ adfirctives 3.38 .37 2.83 .93
g 13. Heart rate 76.14 6.10 74.64 9.39
O 14. Unknown tracks 2.44 1.05 2.23 1.44
15. Sfied .00 .94 .00 .90

16. Accuracy .00 .83 .00 .85

17. NASA-TLX 2.93 .45 2.92 .86

18. CMQ statements 3.50 .39 2.94 .98

C; 19.CMQ adjectives 3.31 .39 2.87 .92
g 20. Heart rate 75.15 5.87 73.71 9.47
D 21.Unknown tracks 3.03 .88 2.80 1.43
22. Speed .00 .87 .00 .84

23. AccuracL .00 .85 .00 .79

 

Individual N = 82; team N = 91.

107

 

Eva: H5v}... Hnave...
.5 M Z Emu. ”mm M 2 726365 .Ecowflc 05 323 2a 388 ”.mcomfle 2: 30.3 25 39636:—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mo.. 2.- 3. B. S. 2.- 2.- oc. wEEfiJ .o
9.- .3. ac. EN: 5. C. 3. mo. 88 :3: .w
3. 5.- Iwm. 2. oo. 3.- no. .5..- 2628088Eases—2258mm .h
I. 3.- 3.2-. law. no. No: 3.- S.- GEQEBSmVEmacs—acozofim .o
m_. 00. E. .8. **mm.. no. .2. *mm.- 5288be .m
.50. Em. 2.- 5. 3.3..- mo. .8. tam. :Zozoesoz .v
I I I I I I co.- 8. wEEafloEomflae-r .m
.- I I I I I I .o. 8:82; 83-.- .N
me. me. vor .8.- me: we: I I 93283 636930 ._
o w h c w. v m .6. ~

 

 

£2332: 2:.“on 95 2.233952: 3:08:35 .6953 30.58.59.25 .N 033.

108

c_.v§:o.vm._.._nmo.v§

.5 M Z :32 ”mm M Z 3:23—6:— ._ncowa€ 05 323 2a .352 23036 05 30% 2a £96365

 

 

 

 

 

 

 

 

 

ram. *8. 8.- 2. 5.- E. 3882.2
:2. $5.. .57 2. 6o. 2.. 825.2
19.. 2. 8.- *8. am. :. 3585.3
3.- 8. mo. 8. S. 5.- 2.: :3: .2
E. 8. .5. mo. :8. :6»: $638.8338,522.95.2
*mm. 8. :N. :. 3mm. 3.3.. 25522:58.222559:
as; 5.- 8.- :. :8. *8. 26:53 6:86.382
2 2 3 2 S : o.

 

 

 

 

 

 

 

 

 

.8532: _ 6:50 562:3 mcocfloboeuzz .m 22:-

109

Evfi 5v}... ”no. va...
.3 N Z :52 ”mm N Z 5:239: .Ecowflu 2: 322. 05 5:58 gasowfln on. 325 05 3.3239:

 

 

110

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ton .656. no. :3. 5.. .5.- 5233.8
2. 5.3-. 55. 5. 5. :- 8898
:64. 5.- 5.- tom. tom. 5. 5:58-75
S. 2. 5.- 5.- t. 8. 68._ee:.5~
ram. 2. EN. :. .35». 5.3. $658.8328222255.2
as. *5. ta. 2. 2.5. .354. sausage588.2885.2
.57 5.- 5.8..- 5. 13.. :2. 58:53 258.35.:
mm 5 mm .8 5 cm 2

 

 

.mQu—amflofiz N “Ugo 2003.05 .nCOCEDtOUhQZS .V 223%-

2v: .5.vm** Hove...
.3 N Z Emu. ”mm M Z 5:22....— .momo€=2«m o... E 05 20.5.2.8 =52 ”momochEn. o... 282. 05 25.5.2.8 5:223:—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a... so.-. Gm.-. E... 55.-. .5.-. :8...

5.. o..- N... No. 5. 2.- E... .12..-. Rm.- 5282

.2... EN.-. 8..-. E..-. .2. 5m.-. $2.5.-.

2. 5. 5.- 2.- 2..- 2..- E... .2. 1mm- 58%

5.... 55.-. 5. .-. E... :8... E... E... .18.-. .13.. .558 38:85

E..-. .5.-. E... 55... 5... .2.-. .5. E... 5. .-. .558 3...;

.4. .. E... :5.-. .2... .2... E... .15.-.

SN. 2.- co. .5.- o.. *8: E... 5.5.. .5.- 2:534

3....-. 58.. 5... .2. E.-. a... E...

5.. it... .~.- N..- .5. 4.. 2.... 5... 8.- 2.: :3:

.2.-. E... :5. .2. E... E... .2.-. 2.4.5.

*5. 8.- 3.2. to... 8. *2.- 25. .. E... E..- .288 Eases-n.

5..-. 2... 3.5.. :3. E... 2.2.. 2..-. 2.5;.

*2. 5.- Io... .32.. w .. 3.5m.- .N . .. E... 8. .58.5. Ecosoem

.2... :5.-. 3... Am... 35... A..-MN. E...

e...- m .. 5.. .5. co. 5. .2 .. 55.-. :5. 58:53 228.3%
7.5... 2:53 . . 25.0

“.5. 5255 52.05 :02? E29. sz-52 222:.» 9.2.4.0.: l
wEEauA :3: ..oEm_ ..o:.m -a..xm -5502 02.2.25...- an-r 03.8.30 2:..on

 

 

.mDuZmflD-fi m UENO UCQ flea—.3609: UCEWDLQ .mCOCfl—Snrcme EHCDFSCOQKD :DOBHOD mflOSfl—DEOU .W 0—33;-

111

EVE. ”EVE; .895.
.5 n Z Emu. ”mm N 2 1.3.3:...— .momo:.:o.am 2.. E 8.. 25:22.09 E3. ”8355.9. on. 282. 2.. 25:22.50 1.3.3:...—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

icm. 20.. A..-w... .2... .2... Ave.-. 3c.-.

in. we: 2. E. Km. *3...- C-m... A..-ON.-. me.- 55.53;.-

3... £5... :2... fmm.-. 3..-. 50.. 8...

MN. we. 3. NOV 8.. No. 2o. fan—m... N..- Baum

8o. :5 -. .2.-. Go. A. ..-. So... So. fix... .2... .EEou 3.5.35

3:... 3.... So... Co.-. 2c. .2... .2... .2.-. Am... .888 25.;

far-M.. 20.-. 30... .8... 2o. 8.... CON.-.

MN. KN: m... we. *2... inn: 30.. CO. e... mic-:54

50.-. Ainmo. 2.... AN... 3;.-. 3... CA...

no. .13.. x..- R..- Z..- 2. Am... EON. 3.- 2.: :3:

8c. .2... Aim. 8c. Q... 3:... :5... 3a...

*Om. 8.. .31.... tam. m... *mm.. 6... 80.. 3.- 12:0... 3.2.9.1..

30.-. A..-Cm. A..-ON. Am... 5... 3c.-. A..-mm.-. 3.83.

*mm. 8. iom. 1.2. 2. m..- 2 _.. Ame. EN..- 22:03 3:238”..—

.NC.-. 3...... Aficm. A..-om. fnm. Aim. A....."wm.-.

D..- 5. *mm. 3. 3. X..- 2 _.. :3... no. 95253 u>_.--.o.E:m
7...... 62:5. . N 25.0

0:... 2.1.30... 2.2.9.. col-5.» E2... wEEau. 2325... 953.03 +|
wEEfiJ :3: .SEM ..oEm Azim. .252 3.83:8.- Eno-r 02.8.30 Ecumen—

 

 

.mu..:m3:. N 25.0 9... £23.38 25.35 .20....3958 55:...ng 503.2. 25:30:00 .0 93....-

112

0.95209315095
.0 N Z :50. "mm M Z .9535... .m0m0...=0..x. 0... c. 0... 25:20.50 E00. $302.02.... 0... 05.0.. 0... 80:20.50 7.3.3.0...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7.31.4... AN... Tcwnm. SA... A*NN.. AN... Awe.

3.. .0. no. 2... $3.. in... .5. M... in. *wmr 38:004.

3... fmm. Aim-M.. :o.-. .0... .0... CA...

*mm. **.mm. .00.-. .13.. 0.. m.. SUN. *3. 00. 0025

£3.00. 50.. .05. .2... 3....-. A..-MN.-. Am... .2... 30.. .8800 0.02.005

5.... ....nm.-. 2... A... .N. A.....Nm.-. A0... .00.. .8. So.-. 5:50 .5..0>

A..-*9”. film. $3.30.. ANC.-. $3.5”. :NN. Av...

3.3.. co. .0... far-N. 1.00. no.- *wm. tum. v..- w...E..0-.

.00.-. Am... .8... Aims. :c.-. .. ... 3....-.

N0.- 0... 3..-. 2 ... I. :3. 00. 00. 00. 0.... :8:

.2. a... .2. :0.-. 50.... 5%.. :2. 2......

m.. 0c. .8... :5... .-mm. 2. 3.3.. 3N». *nm. .220... .a:0..0Em.

$5.. 3... 2.... .0... ........0.. $30.. A»... 3.2:...

4.0... m .. .mo.-. .00.. 0.. 0.. 3mm. 10w. *mm. .025... .0:0..0Em.

3.... .2... GO. :0. TERM. few—M.. Tel-r...

.5. - me.- .00.-. 6.... m..- a... **0m. *mm. **m0. 000.253 02803.6
3.3... 62E... . m 0:30

8:50 .5200 0.... .200... .930... 0.5....03 T
m0...500< 000mm 0.000.005 .3.0> 95:001. :00: .55.. .00-m. 0>..00.3:m . 0:50

 

 

£052.08 M 0:80 0—... 0052.08 . 0:30 0003.00 £5......0....00 .5 0.0....-

113

62.2-03: _c. v a .2.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

awuz

cc. mm. cc. cc. cc. 8. cc. 5. .0:-

wc. 23.2 2. 2. 0c. 22 5. mm. ,0

c. .- _ _.- 8.- ac. 0:50 g 95:53

in it» _c. 8.- 2. 3. cc. 2. 255

cc. .8.- oc. cc. cc. 8.- 8.- _c.- c8253
N whim _ QBW N QBW — 9.3m N a5 _ QBW N mgm .dw.m
wEFEo-H 0.5.. two: mo>touflua OED 3:05.05; OED

 

 

.. 0:50 0. 25.1. mix-53 £000.32: 5.. 05: 0:0 05.0.53 :0 058.005 00.5005 -.0 0070.00.00. 0057.008 00.00001 .w 030..-

114

Table 9. Repeated measures regression of performance dimensions on workload and time

for individuals working alone.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Speed Accurac

Step I Step; Step 1 Stg) 2
Workload -34“ 4.03% -.16* .5?
Time .00 -.48** .00 -.24
Workload x time .86” .43
F 10. 14** 9.56** 2.08 2.11
JR: . 12 .05 .03 .01
N = 82.

* p < .05 (two-tailed)
** p < .01 (two-tailed)

l15

8255: S. v m 1
823.53: 8. v a *
9023-03: of v A +

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.3 M Z
cc. co. cc. 5. No. R. 5. .0. ow. mks-
mo. 33(- wm. co. 3; imméi hm. hm: **_m._m_ k
5.. E. M2. 25% x wEan. 03865.8;-
mc. cor VN. 25w x 2:825
mo. 3. o_. oan x 95353
E. 2:. 2.. cc. 2.- 3. mesa. 2:825
m... +2. be. 5. MN:- 5.- BBQ—Em
5. me. 3 .- : _.- 3.- 5.- 1.5253
we. no. no. “3&0. ”.315. ".3." _ h. inmv. "flue. "Ina-c. 0:30
m cogm m mam _ 92m m 56 m 92m _ cam m 93m m mam _ norm

BE :3: 33303:“ 020 5:253; OED

 

 

._ 2:50 E £58 5.“ 25% 25 «52.4.53 :0 £238.: 3.95.3 3:25:74 .8 :07.»ng 3.532: venom—om .3. 033-.-

116

Table l 1. Repeated measures regression of performance dimensions on workload and
time for teams.

 

 

 

 

 

 

 

 

 

 

 

 

 

Speed Accuracy

Step 1 Step 2 Step 3 Step 1 Step4fi2_ Step 3 ‘
Game .00 .00 - l . 19** .00 .00 -.43*
Workload -.32** -l.87** -.15* -.53*
Structure .21** -.10 -.32** -.7l**
Transactive learning -.02 -.07 .13 .00
Workload game 195’” .47
Structure x game .34 .42
Transactive learning x gpme .07 .15
F .00 l0.06** 25.70** .00 9.23“ 2.18
JR: .00 .15 .27 .00 .14 .03

 

 

 

 

 

 

 

 

N = 91.
* p < .05 (two-tailed)
** p < .01 (two-tailed)

ll7

Table 12. Repeated measures regression of learning indicators on workload and game for

teams in Game 1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Learning

Step 1 Step; Step 3 ‘
Game .62** .62** .68**
Workload -. 15* .29
Structure .10 -.38
Transactive learning -.01 -.24
Workload game -.56*
Structure x game 53*
Transactive learning x game .25
F 104.62** 3.15* 5.23**
JR3 .38 .03 .05
N = 91.

* p < .05 (two-tailed)
** p < .01 (two-tailed)

118

 

825.25 5. v a .2.
625-3: 8. v a *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.Nw M Z
06. CC. CC. on. co. v0. co. 5N. no. 0O. 8. _N mkfi
* **

Om. Om. wQ "wowfim * :.m \- _ .N ON. mmdm *hofi **wo.v co. **ow.N k

Efiéaso:

mo. *o~._- .36;- x 9.5353

:o_m._o>a:xo

hm. *hod- *NN.N- x two—x52,

.7”.- VO.. wN. *MNF —N. *ONI swimming—=02

3.- 8. mm. o. .- *3. 8. 822,35

:0.- co.. 0c: "EON 0O: v0.- **©o.m No: .0. 253.53

Inwx. from. **ww. inhw. clam»! fan. **Nm. **Nm. **NV. .39». “Ismv. «Iamv. .r._o>v_ 05a obn—
flfljfllflfljfilflfl m new. N mew _ New a. 9% m 95 N 2m _ mgm
2.: :3: _ 3350?.” 0.20 3:0an3 0—20

 

 

.25—s wEioB £56385. 5“ 23:8th 95 95353 :0 .3503 35:25 _ 2:30 we soap—mo. 8.3252 .2 03...;-

119

Ave—«3-03; ~O. V R .3.
625.3: 8. v a ..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.Nw M 2
CO. CO. CO. hm. NO. _c. _c. we. we. mo. 8. no. ~29.
0?. mm. 0— . **aw.mmm co; em. on. 01-0.? mm; om; we. “3.0.0 k
3“. on;- NON- Sam-0:830: x 95353
N _ .- owr E..- :o_.£o>abxo x 95333
3.- mo. ov. me. *3. 2. 8.083252
cc. #3. — _. Nor mo. vor :c_.x.._u>w.:xm
@- .- NO. NO. 3 .N m _ .- 00.- VON we: NOV 650353
.335. *iuo. .33-0. ".315. MN. MS. ©_. in. scam. *mm. *ON. *ON. £o>2 oanohm
V. brim m- GUHW N QDHW y QDHW V qoum M Qmim N Qwam ~ Guam V vam M GDHW N QDHW — abdm
UHQH :60: nU>CUDHUw OED man—Oeounzm OED

 

 

 

 

 

5:32 5.“ Eggnog 95 303.53 so 3.30.8 3.23:5 _ 98:0 .3 573%“... 3.2252 .3 035,—-

120

Table 15. Moderated regression of Game 1 learning on workload and personality for
individuals working alone.

 

 

 

 

 

 

 

 

 

 

Learning

SteLl Step ,2_ Ste 3
Workload -.04 m? .83
Extraversion -.O4 -. 10
Neuroticism -.29* .35
Workload x extraversion .25
Workload x neuroticism - 1.30*
F .09 2.43 2.88
JR: .00 .07 .08

 

 

 

 

 

N = 82.
* p < .05 (two-tailed)

Table 16. Moderated regression of Game 1 learning on workload and personality for
teams.

 

 

 

 

 

 

 

 

 

 

 

Leamino

Step 1 Smog! Steg 3
Workload -.3 l ** -.32** .76
Extraversion -.02 .04
Neuroticism .Ol .19
Workload x extraversion -.39
Workload x neuroticism -.75
F 9.51** .04 .26
JR: .10 .00 .01

 

 

 

 

N = 91.
** p < .01 (two—tailed)

625.95: S. v a i.
€25-53: no. v m *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.3 M 2
so. cc. co. hm. co. No. 8. 3. no. we. me. be. HS-
".35

5. mm. mm. wfimm ivd oc. mm; *noé inns mm; co. 2.3.5 R

95.32 93535:.

as. *8; *3...— x 8:53; x E5253

* *2 ._ wEEfi:

7c. mo. *5: 5.- - mm: 035355 x 2:555

3:22.3—

co. 5. #37 cm: .155.- cm: 353:5: x E5253

E. _ _. cw: on. 3.. mm. 2382:» x @5353

v3.- co: 5.- .1. 5. av. om. **mo._ 3‘. m _. wEEafl 0353.5:-

wc. - 5:.- wc. em. mm.. 8.. mm. .8.- N_. 05525

mo.. 3.- m3. w _. o_ .- E .- mm. we: no: 35353

:3. .315. 3.3. .535. ON. ON. cm. *3... IR. *5. 13m. *8. £32 9:5qu
[3&1 m mam N 2m _ mam v mam m moi m mam _ moi v 2m m mam m mam _ moi
2.: :3: 33.3.55 OED £=oEBSm OED

 

 

.253. 5a 55:23:52: :0 $2855 3253 35:95 .5 5.29%“: 623252 .2 29d-

l23

Table 18. Moderated regression of learning indicators on manipulations for teams.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Leamina

Stefl Step 2 Step 3
Workload -.3 l ** -.34** -.39'l'
Structure .27** .09 -.Ol
Transactive learning .06 .07 .01
Workload x structure .17 .30
Workload x transactive -.l l .00
learning
Structure x transactive .07 .23
learninL
Workload x structure x -. l9
transactive learning
F 5.93** .29 .I9
JR: .17 .01 .00
N = 91.

+ p < .10 (two-tailed)
** p < .01 (two-tailed)

124

 

._o>o_ Sam. 2: a nosing—8%: Eco $230328 38 22:3ch 2865 3:: @2750

 

 

 

Beam

 

 

 

 

.3303
Eggs—m

 

 

 

cozoiuim

  

 

 

8.0830502

 

  

 

 

 

 

 

5883‘

 

 

 

 

 

 

8:323
ouSOmom

 

 

 

 

93253
£5305
.------~r ......... u
.
“ wig“: "
. .
n 0283:“; u

425:. Bififiom»: . _ Bzwi

5

Figure 2. Interactive effects of workload and time on speed of performance.

0.5 r

 

 

Speed
0

-o.5 -~ -~—- —-

 

 

 

 

-1.5

 

Very low w orkload Very high w orkload

|—+—— Game1-+-Game2;

 

 

..__.._,»g

126

Figure 3. Interactive effects of workload and Extraversion on individual emotional

arousal.
4 .
3.8 i , , ,, , ,- .___---___
3.5 ._._- t-.-_____ ._,- . ,, ,,,-.-_, H.-- . --__-._.-__

 

 

Emotional arousal

 

 

 

 

 

24 — ,,_ — ~ ~ ——————————~-————— -_ %
2.2 { ---— ___,____.--___ -— -~—-——-~—-~--—— 7- _._--_

l 2 l

l Very low w orkload Very high w orkload

l " ““ ' “’“‘— '—“" _._- — "‘1

J .—O—— Low extraversion - -l- — High extraversion l l

 

Extraversion is plotted 1 SD above and below the mean.

Figure 4. Interactive effects of workload and Neuroticism on individual emotional

arousal.

 

 

 

 

 

Emotional arousal

 

 

 

 

 

 

 

 

l Very low w orkload Very high w orkload

I» . . . . . 1
l --O—- Low neuroticusm - -I- - Hugh neurotlcnsm .
l “fl M-.... .

 

 

 

 

Neuroticism is plotted 1 SD above and below the mean.

Figure 5. Interactive effects of workload and Neuroticism on individual Ieaming.

 

 

 

 

Learning

 

 

 

 

 

 

 

Very low w orkioad Very high w orkload

 

—0— Low neuroticism — -l- — High neuroticism 3
-._. _M_ _ _._..__.___ _ -.-__-.__.__ __________ J l

 

 

Neuroticism is plotth 1 SD above and below the mean.

Figure 6. Interaction of workload and type of track on friendly fire kills in teams.

 

Friendly tire kills

 

 

 

Low w orkload High w orkload

l+Unknown --I--Knownl

_ , i, _._—_._- __ .__._-..._-__ ,, . _____-..-.__._-.____-.,~ .,_.____. .-,4_ J

130

Figure 7. Interaction of workload and type of track on good attacks in teams.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I
50 ———~ ~ ~ —~—- ~—— ’1— H,__
I
I i
I
I l
40 - _,___/!-..
/
2 ,’
5 , l
i 8 ,’ i
l o I
‘ 0 I .
20 e-.-_,--- i . -_ W__ .
v;
A l
10 A - . -~ i
l
l
0 ‘ 1
Low workload High workload

 

l—O— Unknow n — -I— — Know—r17 i

131

 

 

 

ESSSO

 

 

 

98:95
5:02

 

 

 

>:o>o:

v.5.

 

I32

 

98353
3235

 

 

 

 

 

 

 

 

5:30

 

 

 

wEEfiJ

 

 

 

425:. :32 032-20; .w 833”.

   

iiiiiiiijiiiiiiiiiiiiii