NSF grant

Jeff Vancouver 940110

CSGnet, below is a draft of a grant I am submitting to NSF. It is a
little rough in spots and the cites have not been completed, but if anyone
is interested, I would appreciate their feedback. It is long for a post,
so I understand if you do not want to read it. The first section
summarizes it. =20

I also suspect I will get some unhappiness regarding the study.=20
Unfortunately, it is not an argument directed to PCTers, so my position is
not well reflected. That is, I am using NSF appreciated language. I want
eventually to engage in the debate
more, but now is not the time. Bottom line, if you do not have anything
constructive to say ...

Thanks Jeff Vancouver

                                Project Summary

      This project investigates dynamic decision making and
behavior associated with managerial effectiveness. Focus is
placed on how individuals maintain prioritize and resolve
conflicts between numerous goals from multiple constituencies
(e.g., personal or self, subordinates, administration, and
regulatory agencies) with feedback from multiple sources.
      Unfortunately, methods for rigorously studying the processes
involved in these contexts are scarce (Rasmussen, Brehmer,
LePlat, 1991; Stevenson, Busemeyer, & Naylor, 1990). To that
end, a computer-based simulation of a nurse manager's job in a
hospital is developed as a research tool for studying the
interactions between human and organizational systems on human
performance (i.e., managerial effectiveness). The simulation
interacts dynamically with the participant such that decisions at
one time affect what the participant sees at a future time.=20
Features of the simulation include the administration of
questionnaires (e.g., self-esteem) throughout the role playing,
control over the information available and over the complexity
and difficulty of the task, and the tracking of information
sought, decisions made, perceptions, espoused goals.
      Three study sets are proposed to better understand and model
the behavior of managers in the simulation. A living system
theory framework is used as a guide for prediction and modeling
the manager's behavior. The first set examines a series of
hypotheses regarding effective managerial behavior in a complex,
multi-constituency environment. The second set of studies is
used to develop a hypothetical model of the manager's goal
hierarchy and to test the hypotheses. The third set investigates
the modes of decision making and operation of the manager with
the simulation given variations in the environment.
      Specifically, study 1.1 looks at the relationship between
feedback seeking and managerial effectiveness when a constituency
definition of effectiveness is used (). Study 1.2 examines the
effect of individual differences on effectiveness and feedback
seeking over time. Studies 1.3, 1.4, and 1.5 manipulate
constituency power, task ambiguity, and amount of spontaneous
feedback. The effects of these manipulations on effectiveness
and goal monitoring will be assessed across time.
      Study 2.1 uses protocol analysis and action identification
theory to develop a model of the manager's goal hierarchy. Study
2.2 uses the test for a controlled variable to test the goal
hierarchy. The test for a controlled variable is essentially an
examination of stability where their should be instability.
      Studies 3.1 through 3.4 examine the role of task complexity,
task uncertainty, task difficulty, and time pressure
respectively. These variables are expected to provide clues
regarding underlying system properties (e.g., lag and load) that
the manager may be controlling. Goal planfulness (Frese & ??)
will be an individual difference variable examined in the last
study. =20
                              Project Description

A. Objectives and Significance

      Managerial effectiveness is key to the productivity and
quality of service in organizations (Tsui & Ashford, 1994).=20
Recent models of self-regulation in a managerial context (e.g.,
Manz, 1986; Tsui & Ashford, 1994) speak to the importance of
maintaining and monitoring numerous goals emanating from numerous
constituencies using various means. In these situations, which
define most managerial environments, individuals must prioritize
and resolve conflicts between goals. They must do this under
limited information processing conditions during changing
environmental circumstances, which they have affected. Decisions
involving allocating information processing resources (Kanfer &
Ackerman, 1989), seeking information from the environment (Tsui &
Ashford, 1994), and acting (Frese & Zapf, 1994) occur constantly.=20
Furthermore, as individuals become more familiar with situations,
tasks become less resource sensitive (i.e., more automatic) and
mental models of the environment begin to emerge that allow
greater, faster, and more accurate prediction of environmental
circumstances (Reither & St=84udel, 1986). Yet, with these changes
inappropriate actions are more likely to occur as a result of
insufficient monitoring of the environment, inadequate models, or
habit errors that lead to ineffectiveness or disastrous
consequences (e.g., Heckhausen & Beckmann, 1990). Thus,
understanding these processes are critical to improving the
productivity of humans, a primary goal of the Human Capital
Initiative and specifically to the Productivity in the Workplace
      To understand these environments, cybernetic or control
theory models of interacting systems have arisen to prominence
(Ashford & Taylor, 1990; Beach, 1990; Campbell & Lee, 1988;
Cropanzano, James, & Citera, 1993; Edwards, 1992; Fellenz, 1994;
Flamholtz, Das, & Tsui, 1985; Ford, 1987; Frese & Zapf, 1994;
Gollwitzer, 1990; Green & Welsh, 1988; Hershberger, 1989; Hyland,
1988; Kanfer & Ackerman, 1989; Klein, 1989; Kuhn, 1986; Lord &
Levy, 1994; Manz, 1986; Nelson, 1993; Senge, 1990; Seeman, 1989;
Tsui & Ashford, 1994; Vancouver, submitted for review; Wright &
Snell, 1991). These models focus on goal directed behavior in
actual settings and fall under the general rubric of living
systems theory (LST; Vancouver, submitted for review). LST
describes individuals and social units as nested systems that act
on and choose environments to achieve multiple, dynamic goals (D.
Ford, 1987; M. Ford, 1992; Lord & Levy, 1994; Miller, 1978;
Vancouver, submitted for review). It incorporates features of
sociology (e.g., role theory), motivation (e.g., control theory
and goal-setting theory), cognitive science (e.g., information
processing and decision models), and individual differences
(e.g., skill acquisition and cognitive ability theories). The
individual (human) model incorporates and builds from action
theory (e.g., Frese & Zapf, 1994), perceptual control theory
(PCT; Powers, 1973; 1989; 1992), and the self-construction living
systems framework (LSF; D. Ford, 1987; Ford & Ford, 1987). Much
of the research on errors on dynamic decision making has occurred
under action theory (see Frese & Zapf, 1994 for a recent review)
and much of the modelling described below derives from PCT (e.g.,
Marken, 1992).
      Although many would agree that the complexity of the
environment and the nature of one's interaction with it are as
described above, few empirical programs of research are developed
around this description (see Ford & Ford, 1987; Frese & Sabini,
1986; Marken, 1992; and Rasmussen, Brehmer & LePlat, 1991 for
exceptions). As a systems model, LST explicitly considers the
interaction across time among other systems and between systems
and the environment at multiple levels of analysis (within the
individual, between individuals and between groups of individuals
- i.e., social units). The cybernetic or control quality focuses
on the goals of the system and the means for achieving and
monitoring goal progress. Thus, LST requires collecting data on
systems purposes, perceptions, and processes as they interact
with other systems over extended periods of time. Specifically,
data must span person, occasions, and variables to adequately
capture the problem space (Nesselroade & Ford, 1987). Further,
observation settings must include a linkage between the actor
system and the environmental system to adequately represent
reality (Stevenson, et al., 1990). This type of data collection
paradigm is achieved in certain kinds of field research designs
or simulations. The current proposal focuses on the latter.
      Specifically, the set of studies described here are designed
to investigate the effects of different person and situation
variables on the person, behavior and the situation in the
complex, albeit simulated, environment of the manager. The
simulated environment allows for precise control and measurement
of behavior and information flow between the subject and the
simulated environment. A second set of studies is designed to
explore the perceptual goal hierarchy used by the manager while
interacting in the simulated environment. A final set looks at
modes of interacting with the environment and internal system
goals likely to affect the modes.

B. Relation to P.I.'s Longer-Term Goals

      To understand the behavior of living systems (i.e., humans,
groups, & organizations) models derived from general principles
and specific theories need to be expanded. The P.I.'s primary
interest is to integrate the various psychological and social
science disciplines and subdisciplines within a meta-theory to
facilitate interaction between the members of the scientific
community. To that end, living systems theory (LST) has been
used to integrate psychological and sociological perspectives
(Vancouver, submitted for review) and the goal construct within
LST has been used to integrate cognitive, personality, and
motivational perspectives in psychology (Austin & Vancouver,
submitted for review). Others have used LSF and control theory
within LST to integrate various motivational theories (M. Ford,
1992; Klein, 1989). Action theory (as well as control theory)
has been used to address the gap between the objective
environment and cognition (i.e., interpretations of the
environment), and cognition and action (Frese & Sabini, 1986;
Lord & Levy, 1994).
      This project is the beginning of a long-term research
program to create a model of higher-order (i.e., cognitive and
social) processes of humans in complex work settings. The
modelling is based on the hierarchical control system developed
by Powers and his colleagues (Powers, 1973, 1989, 1992; Marken,
1992). Much of the current modelling occurs at the lower levels
in the hypothesized hierarchy (Marken, 1992). This research
program focuses on the mental models and higher-order goals
(e.g., Emmons, 1989; Klinger, 1977; Little, 1989, Markus & Wurf,
1987 & Steele, 19??). Further, these mental models are often
models of other hierarchical control systems (e.g., other humans
and social units) which simplifies the modeling process. A
simulated environment populated with hierarchical control systems
at multiple levels (humans, organizations) is being developed to
study the actions and cognitions of the focal system (the
participant "manager"). The communication between these systems
is via another control process called the role episode (Berlo,
19??; Katz & Kahn, 1978). The immediate goal is to assess the
impact of changes to the messages received and available to the
focal manager. Eventually, the PI hopes to study the formation,
use, and abuse of mental models (i.e., heuristics, schemata) of
the other systems in the environment. Because the other systems
are completely described by the simulation, their attributes are
known and controllable.=20
      After an initial emphasis on the development of a
descriptive model where actual behaviors serves as the standard
of comparison, more prescriptive models can be derived where we
attempt to change actual behaviors. The results of the studies
described here along with other research (e.g., Reither &
St=84udel, 1986) will be used to define problem areas in human
functioning. Armed with this information, training or other
environmental changes can be operationalized into the simulation.=20
For example, Reither and St=84udel (1986) reported success from
training self-reflective or meta-cognitive skills (see also,
Karoly, 1993).

C. Relation to State of Knowledge and Work in Progress

      LST and its sub-theories (action theory and perceptual
control theory) integrate the motivational perspective found in
the organization and management literatures (e.g., Locke &
Latham, 1990; Klein, 1989; Frese & Zapf, 1994), the social
psychological perspective found in role theory and organizational
theory (e.g., Biddle, 19??; S&S. 19??, K&K, 1978), the cognitive
perspective found in the information processing, decision making,
and social cognitive literatures (e.g., Beach, 1990; Brehmer &
A?? 1991; Fiske, 1993; Kanfer & Ackerman, 1989), and the more
general systems theory (e.g., Ashby, 19??; Boulding, 1956;
Deustch, 19??; von Bertalanffy, 1950). The relevant aspects of
these perspectives are briefly reviewed below and the research
questions that each highlights are presented. More specific
literature is presented in the description of each study set.

      Living Systems Theory. The basic observation that informs
LST is that systems maintain regularity despite irregularity in
the environment. Systems theorist (Ashby, 1952; von Bertalanffy,
1950; Wiener, 1948) have attributed this regularity to control
systems that monitor variables and attempt to maintain them
according to some standard or goal. Further, the dynamic
homeostatus indicative of living systems can be modelled via
hierarchical combinations of these control systems (Miller,
Gallanter, & Pribrum, 1960; Powers, 1973). The nesting of
relatively simple systems (e.g., control systems) into more
complex systems (e.g., humans) extends to even more complex
systems (e.g., social units) (Boulding, 1956). The up side of
this description of humans and social units is that a
parsimonious building block (i.e., the control system) underlies
the structure of complex systems. The down side is that an
understanding of the interaction among the building blocks and
other complex systems is not straightforward. =20
      Fortunately, the middle-range theories prominent in many
subdisciplines of psychology and management science either
implicitly or explicitly use a systems perspective (Vancouver,
submitted for review). These can be used to narrow research
questions and facilitate the study design.

      Goal-setting and the motivational literature. The dominant
motivational theory in applied psychology is goal setting theory
(Locke & Latham, 1990). The theory revolves around and much
evidence supports the notion that specific, challenging goals
enhance performance (Locke, Shaw, Saari, & Latham, 1981). Yet,
this theory was developed and the evidence collected under
contexts with single goals for relatively simple tasks. Only
after numerous applications of the goal setting principles in
complex environments did the importance of goal acceptance,
commitment, strategy, and feedback emerge as important factors
(Chesney & Locke, 1991; Erez, 19??; Hollenbeck, ???; Locke,
Smith, Erez, Chah, & Schaffer, 1994; Wood & Locke, 1990). Recent
theoretical and empirical work has turned to self-regulating and
cybernetic models of goal-setting and performance processes
(e.g., Latham & Locke, 1991; Lord & Hanges, 1987). As the
relevance of complex and dynamic processes has emerged, the use
of computer simulations has also emerged as a method of studying
goal-setting processes (Chesney & Locke, 1991; Wood & Bandura,
1991;…). Much of this work has demonstrated the importance of
strategies and the role of self-efficacy on performance, but not
always in expected directions (). What needs further elaboration
is the consideration that strategies are sub-goals and therefore
modelled via control system processes (Austin & Vancouver, 1994),
that self-efficacy and other behavioral choice variables may play
a role only early on in the self-regulation process (Mitchell, et
al., 1994), and that complex behaviors, often relating to the
configuration of one's environment, are likely when individuals
attempt to balance numerous resources among numerous
constituencies' demands.

      Social psychology and role theory. When human systems were
considered as interacting within larger systems of social and
organizational units, role theory emerged as a means of modeling
the interaction or communication among the social units (Katz &
Kahn, 1978). Role theory describes a cybernetic process of
communication where role senders monitor the behaviors or results
of role receivers and adjust their communication based on their
observations. The role senders for a focal individual (i.e., the
role receiver) are referred to as the role set. The role set
includes the numerous constituencies referred to in the previous
section on goal-setting. Most of the literature on
organizational role theory has focused on the conflict and
ambiguity found among the messages received from the role set
(Jackson & Schuler, 198?). Thus, role theory research has
focused on one element of the complex environment.=20
Unfortunately, most of the research is cross-sectional and
therefore not focused on the dynamic interaction among the human
and social systems involved (Jackson & Schuler, 198?).
      Meanwhile, more sociological versions of role theory have
included the dynamic aspects (S&S, 19??). For example, as a
means of facilitating interaction, role theorists note that the
focal individual begins to develop a mental model of members of
their role set (called role-taking)(S&S, 19??). Through this
interaction a shared meaning is developed which help defines
social units (Weick, 19??). Mutual goals and well-defined
procedures (i.e., sub-goals and behaviors) emerge based on past
interaction and interpretations of events (). Thus, higher-order
entities can be modeled in the role-taking process, not just
other individuals. =20

      Decision and information processing issues. Acting in an
environment of multiple constituencies with multiple goals over
time and where the actions affect the environment creates a heavy
demand on decision making and information processing. According
to Brehmer and Allard (1991), individuals have three alternatives
in dynamic environments: a) develop a mental model, b) develop
heuristic rules, or c) rely on feedback and modify behavior
gradually. LST uses its model of human system to break this down
even further (Vancouver, submitted for review). Individuals a)
observe the reactions of the environment without actually
engaging it (what Bandura, 1986, calls modeling), b) develop
intentions without actually observing or engaging the environment
(what Powers, 1988, calls the thinking, planning, or memory
mode), d) interact with the environment and observe the
consequences (what Powers, 1988, calls the behaving mode), and d)
act on the environment without observing the consequences (what
Vancouver, submitted for review, calls the assumption mode).=20
These four alternatives are a function of two types of gates in
the control systems that determine whether memory structures
(e.g., schemata) are evoked or information and energy is
exchanged with the actual environment. One type of gate
determines if input to the system is from memory stores or the
actual environment. The other gate determines if outputs reach
the environment or merely enter memory. Thus, the mode of
observing without actually engaging with the environment is when
the gates for inputs are set to the environment and gates for the
outputs are set to memory. This mode is useful for efficiently
creating a mental model of the environment (i.e., more efficient
than trial and error). =20
      In LST, decision making is more closely related to the
thinking mode. Stevenson et al., (1990) define problem solving
as the development of alternatives and decision making as the
process of selecting among alternatives. LST is careful to
identify point-of-view. Decision making from an external point-
of-view is where someone defines the alternatives which the focal
person may or may not need to think about selecting. From the
individuals' point-of-view selection among alternatives occurs
only when goals conflict, resulting in a need to reorganize the
goal hierarchy. Often this is accomplished by thinking through
alternatives and choosing the alternative that best meets the
desired consequences according to the memory stores (Powers,
1988). Hypothesis testing by actually interacting with the
environment can be used to supplement insufficient memory. This
construction of the decision (Payne, et al, 1992) ...

      The actions that result from using modes that require using
the memory stores (i.e., planning and assuming) are highly
dependent on the accuracy of the mental models. It is
hypothesized that higher-order control systems (meta-cognitions)
often monitor this dependency (Karoly, 1993) and may promote more
interactive modes. Further, as the memory structures are
developed, limited information processing capacities are required
(Schrank & Abelson, 197?). Further, in planning mode (i.e., when
developing intentions), the use of these memory stores may
require more or less information processing capacities depending
on the nature of the memory stores and the context presented by
the environment (Mitchell & Beach, 1990). Larger discrepancies
between goals and perceptions or anticipated perceptions (based
on memory stores) may require more intense thinking and problem
solving (Mitchell & Beach, 1990; Vallacher & Wegner, 19??). On
the other hand, smaller discrepancies or tolerances for
discrepancies will lead to more interactive (e.g., automatic)
modes of operation. One of the basic objectives of the research
proposed here is to assess the factors that effect the use of the
various modes.
      Another line of research that overlaps the role theory and
cognitive perspectives is social cognition. The social cognition
literature focuses on the specific goals one monitors while
interacting with others (see Schwenger & Weigold, 1992, for
example), the affective and behavioral reactions to discrepancies
from those goals (see Higgins, 19?? for example), the formation
of impressions of others (referred to as role taking above) and
the processes involved in maintaining the goals and impressions
(see, Fiske, 1993 for a review). Recent attention has been paid
particularly to the cognitive resources applied to these
processes and how they are distributed among goals (Gollwitzer &
Bargh, 1995). The specific application of this work to
understanding the social demands on the manager is of concern

      Systems theory. Most of the work on systems theory was done
in the middle of this century (Boulding, Ashby, Weiner, ).=20
However, systems and cybernetic models are seeing a resurgence in
psychology via self-regulation models (e.g., Kanfer, 1990;
Karoly, 1993) as well as management (e.g., Wright & Snell, 199?).=20
Besides the overall perspective of interacting parts within
wholes, the proposed studies will focus on the role of lag and
load in the processes examined. Lag is the delays in information
as the pass through the system and load is the capacity of the
system to process the information. Not only does lag play a
critical role in how a system operates (Brehmer & Allard, 1991),
it also provides clues about its structure (Lord & Levy, 1994).=20
Load has the same dual function in that resource allocation
processes greatly affect performance (Brehmer & Allard, 1991),
and information collection is often a clue regarding underlying
structure (Reither & Str=84udel, 1986). =20

D. General Plan of the Proposed Work

      The primary method of data collection will be a simulation
of a nurse manager's job in a medium-sized hospital. Details of
the simulation are described below. The series of studies using
this simulation are designed to a) assess the effects of
individual difference and situation variables on managerial
effectiveness and b) develop an understanding of the focal
managers' (i.e., the subjects') internal control structure.=20
the perceptual goals the focal manager is attempting to maintain
are assessed via verbal protocols and application of the test for
a controlled variable (TCV; Powers, 1973; Runkel, 1990a; 1990b).=20
The notion behind the TCV is to look for the basic observation of
LST: the lack of change to a variable despite disturbances to
the variable. Further support is attained when blocking the
ability of the system to monitor the variable results in change
to the variable when disturbances are applied. Internal goals
hypothesized to control the use of modes of interaction will also
be assessed. In these cases no direct method of blocking inputs
is available. A primary issue through these investigations is
the source of variance in the behavior of the manager. The
relative effects of multiple persons, variables (e.g., situation
manipulations), and occasions will be assessed to support
assumptions of aggregation (Nesselroade & Ford, 1987). When
multiple levels of analysis are required (e.g., within and
between person), hierarchical linear models will be used to
analyze the results (e.g., Vancouver, Millsap, & Peters, 1994).

                               The Major Method

      The primary method for stimuli presentation and data
collection is the SimNurse computer program [Marti, I should
probably trade mark this]. The SimNurse is a developing Visual
Basic program which attempts a) to create a realistic simulation
of the job of nurse manager in a medium size hospital's Intensive
Care Unit, and b) allow for the general analysis of the
characteristics of dynamic decision environments. Thus, the
simulation endeavors to provide a realism that facilitates
generalizations of findings typical of field settings while
providing the control and measurement typically of laboratory
studies (Vancouver, Williams, Liu, 1995). The simulation is
similar to D=94rner's (1991), and Brehmer and Allard's (1991)
simulations of complex and uncertain environments. They differ
in that the context is a managers job and the elements that can
be manipulated and measured only partially overlap the elements
examined in those simulations.
      Specifically, greater focus is given to the nature and
availability of the information the individual uses to monitor
progress on goals and decisions. Further, the information is
mostly related to social units (i.e., individuals,
constituencies, and organizations). The primary changes in the
environment that require anticipation and action are via other
people. This is particularly true for the manager who must
coordinate the activities of numerous people to keep everyone
happy. Thus, when creating a simulation of the dynamic
environment facing managers the underlying model must attempt to
model the other individuals and social systems in the manager's
environment. The more accurate this underlying model, the more
confidence one can have in the results emanating from the
simulation (Amy need cite). The same general model the research
in this proposal is designed to refine is also used to model the
simulated systems in the simulation. Actions and information
emanate not only from participants, but also from the modelled
systems within the simulations. Developing the complexity of
these action and information, and the actions and information
available to the participants is a main task for the work needed
to test the research questions presented here.
      The simulation, as it currently stands, is populated with 12
Registered Nurses (RNs) and an administration. The RNs represent
the subordinate constituency and the administration a supervisory
constituency. It is the focal managers job to schedule the RNs
in order to staff the ICU floor. Each schedule covers a two week
period in the unit. The RNs "monitor" the schedules for their
amount of overtime, shift rotation and frequency of changes in
rotations, number of consecutive shifts, number of consecutive
weekends, number of shifts in a day, length of shifts, and total
hours. Each nurse has preferences regarding these factors and
discrepancies are reported to the participant (the focal manager)
should he/she seek feedback on the schedule. The values for
these elements also effect the subsequent preferences and
consequences for later rounds of scheduling. Other consequences
include the RNs stress levels, propensity for illness,
performance, propensity for absenteeism, and their intention to
quit. That is, the past schedules affect the RNs' subsequent
"behavior", which affects the environment in which the manager
makes subsequent schedules. The RN's desires and consequences
are also related to four higher-order goals: a) non-work role, b)
patient care, c) the value of money, and d) the commitment to the
hospital. Each goal, both for the schedules and the higher
goals, have four relevant properties: desired goal level, goal
importance, means of monitoring, and means of affecting. All
these elements are equal across the RNs accept desired level of
schedule goals, which differ as a function of hospital policy
(e.g., senior nurses are allowed and often desire 12 hour shifts)
and random variations.
      The administration monitors the total amount of overtime
scheduled, the cost of the schedule in terms of the personnel
budget, the number of nurses scheduled for a shift and average
experience level of the nurses scheduled for each shift. The
administration has goals for each of these elements and
performance vis-a-vie those goals for a schedule or proposed
schedule is available upon request. Currently, all these goals
can be met simultaneously (if one ignored the RN's goals). Like
with the RNs, discrepancies from past schedules affect the levels
of the goals in subsequent schedules. For example, if too much
overtime is scheduled in round one, less overtime will be allowed
in subsequent rounds of scheduling. The overall goals driving
the administration's schedule goal are monetary and regulatory;
that is, related to the fiscal state of the hospital and the
rules of regulatory agencies. Properties related to these goals
are invariant in the current version of the simulation.
      As mentioned, the primary aspect of the simulation is to
study the actions and use of information by the focal managers.=20
Information in the form of feedback and messages from the role
set is available and can be tracked by the computer program. In
addition, like in real life, costs are associated with seeking to
information (Vancouver & Morrison, 1995). Currently, the primary
cost is the time it takes to seek feedback on potential
schedules. Again, under the guise of realism, feedback from the
RNs and/or administration takes 24 hours of virtual time.=20
Virtual time is implemented in the simulation to facilitate the
passage of time while keeping the amount of real time needed to
complete several rounds of scheduling to a minimum.
      The simulation was developed in part from a grant from New
York University (The Challenge Fund [#??]) and a fellowship (The
New York University Goddard Fellowship). The program was written
as a Windows program using the Visual Basic programming
environment. The PI was training in Visual Basic by the system's
analyst so that the program could be developed independent of
availability of the a systems analyst. The Windows environment
promotes a user friendly atmosphere missing from many complex
simulations. It also supports graphics and multi-media, which
enriches the experience for the participants and allows greater
flexibility in programming.
      Given the current state of the program numerous variables
can be manipulated with little programming required. These
including 1) task difficulty: by changing the desired levels for
the RNs and administrative goals the ability of the manager to
please all the constituencies and their members varies; 2) task
complexity: one or more constituency goals can be turned off to
simplify the context of the scheduling task; 3) goal difficulty:
the desired level of individual goals can be manipulated to
change goal difficulty; 4) goal specificity: tolerances in the
simulation models (via increasing the importance variables) can
be loosened or tightened to change the specificity of the goals
the manager must meet. These changes can be accompanied with
changes in assessable information regarding their levels or
learnable only through trial and error (e.g., multiple cue
learning paradigm). More critically, there can be difference in
the "espoused" goals and the actual goals used by the model.=20
This is a common phenomenon in organizations (Argyris & Schon,
197?) and individuals (Kuhl, 1991).

                     Study Set 1: Managerial Effectiveness

      Tsui and Ashford (1994) developed arguments for the modeling
of managerial effectiveness as the regulation of the goals of the
members of manager's constituencies (e.g., subordinates,
supervisors, clients). Managerial effectiveness is defined as
the minimizing of discrepancies from these constituencies goals.=20
Thus, difference constituencies and constituency members may
define effectiveness differently depending on the goals they seek
and maintain. Tsui and Ashford (1994) hypothesized that
monitoring the discrepancies in their constituencies' goals is
necessary and predictive of managerial effectiveness. Further,
they hypothesized that individual differences such as self-esteem
and self-efficacy related to meeting constituencies' goals will
influence monitoring of discrepancies and performance. Further,
they argued that attributes of the constituencies (e.g., power)
will influence which discrepancies are monitored and addressed.=20
They also hypothesized that attributes of the task and context
will affect monitoring. Ambiguous tasks and lack of spontaneous
feedback from the environment will prompt more active monitoring
from the manager. These hypotheses will be tested in the first
set of studies.

      Study 1.1: Managerial effective will be defined as amount
of discrepancies in RNs and administrations goals and the
schedules produced by the managers. Monitoring constituencies'
discrepancies will be assessed by the amount of feedback sought
while creating schedules for the ICU. Pilot studies will be used
to determine the costs of seeking feedback needed to achieve
variance on the amount feedback sought. Based on Tsui and
Ashford (1994), we hypothesize that managers who seek more
feedback will be more effective.

      Study 1.2: Using amount of feedback seeking as the
dependent variable, the effects of individual difference in self-
esteem, self-efficacy, and personal goals will be assessed. A
longitudinal study design will be used to control for ability and
strategy factors and to track the changes in predictive ability
of the individual difference factors over time. Self-esteem is
expect to initially relate to feedback seeking, but drop off over
time. Self-efficacy is expected to be more predictive after
initial feedback than before, and to become less predictive than
personal goals over time (Mitchell, et al., 1994).

      Study 1.3: The perception of the power of constituencies
will be assessed as well as manipulated via descriptions of the
hospital culture after an initial set of practice rounds of
scheduling. Initial constituency discrepancy reduction will be
compared to discrepancy reduction after the power manipulation.

      Study 1.4: Task ambiguity will be manipulated by
specification of goal levels and priority for the scheduling
task. In task ambiguous conditions, levels and priority (i.e.,
relative importance weight) will not be communicated. In task
unambiguous conditions, goal levels and priority will be
communicated in instructions following practice. It is
hypothesized that managers in more task ambiguous setting will
seek more feedback, but that managers in task unambiguous
conditions will do better at reducing constituency discrepancies.

      Study 1.5: The amount of spontaneous feedback will be
manipulated via feedback from posted schedules (i.e., the
schedule from the previous round). It is hypothesized that
initially managers receiving more spontaneous feedback will
engage in more feedback seeking during the schedule creation
phase than those with less spontaneous feedback. However, over
time the reverse effect should occur as predicted by Tsui and
Ashford (1994). To conduct this study, a spontaneous feedback
module must be developed for the simulation computer program.

             Study Set 2: Assessing the Perceptual Goal Hierarchy

      In the previous set of studies, the goals of the manager are
assigned or assessed from a perspective outside the manager.=20
That is, they are defined by the constituencies with which the
manager interacts. In this set of studies the goals the manager
controls are assessed more directly. A priori hypotheses of
personal goals come from the goal-setting and social psychology
literature. Specifically, it is expected that the managers will
assume at least some of the goals communicated by the role set
(i.e., the constituencies). However, the exact nature of the
internalization of these goals (Sussman & Vecchio, 19??) and the
relationship of the goals to other goals the manager may be
maintaining need to be assessed in order to create a model of the
managers goal hierarchy. This set of studies attempts to assess
the goals and perceptions the manager is regulating in this

      Study 2.1: Verbal protocols will be used to develop a list
of possible goals the focal managers may be maintaining.=20
Participants will be asked periodically to report what they are
doing, how they are doing it, and why they are doing it. This is
the methodology Vallacher and Wegner (1987) developed under the
rubric of action identification theory. It is expected that the
actions identified will reach higher levels of abstraction as the
task becomes more automatized. From this preliminary
investigation, a hypothesized hierarchical goal structure can be
created. Further, the learning curve for subjects can be

      Study 2.2: Based on the verbal protocol analysis, goal
taxonomies in LSF (e.g., Ford & Nichols, 1987), the social
psychological literature regarding impression regulation (e.g.,
Schlenker & Weigold, 1992), and the nature of the task (e.g.,
role descriptions and messages), potential controlled variables
will be hypothesized and tested using the TCV method.=20
Disturbances to hypothesized controlled variables via changing
program and simulated system parameters. Blockage of inputs can
be manipulated via elimination of feedback.

          Study Set 3: Assessing Internal Goals and Their Effects on
Decision Making

      The degree of observing, thinking, interacting, and assuming
is hypothesized to be a function of lag, load, and the perceived
quality of the mental models of environment (e.g., of the role
set). These modes of operation and decision making can be
assessed via a) the use and seeking of information, b) the acts,
and c) response timing (Lord & Levy, 1994). Further, hypotheses
regarding the use of the modes can be tested using manipulations
to load, lag, and the importance of each. For example, in a
highly dynamic environment, lag becomes a more important quantity
to control (). To examine these processes, task complexity, task
uncertainty, task difficulty, and time pressure are manipulated.=20
As the more internal control systems are effected, emotional
responses are anticipated that may give clues regarding
underlying processes (Powers, 1992).

      Study 3.1: Task complexity is operationalized as the number
of goals monitored by the manager's role set that the manager's
behaviors disturb. That is, a manager schedules for the nurses
in the ICU affect variables the nurses and administration
monitor. As the number of variables increases, the focal
managers job becomes more complex. Disturbances to those
variables are translated as potential messages to the focal
manager (i.e., role sending). Although the manager has some
control over the receipt of some of the messages from their role
set, more messages will spontaneously reach the manager. Thus,
as the complexity increases, it is hypothesized that the load on
the focal manager would increase. However, as a means of
maintaining load at an internally desired level, it is
hypothesized that information seeking and hypotheses testing will
decrease. Further, the emotional response to increases in task
complexity is expected to be perceived stress.

      Study 3.2: Task uncertainty is operationalized as the
degree of change in the role sets' goals. Therefore, as time
passes and new schedules are required, different schedules are
required to meet the desires of the nurses and the
administration. Thus, to meet the desires of the role set under
greater task uncertainty, more information gathering is required
and planning becomes less useful. To maintain an efficient use
of resources, focal managers will more likely seek information
and engage in an interactive mode as task uncertainty increases.=20
Mental models will focus on the stability of the members of the
role set and less on the level of goals. Finally, anxiety will
most likely be the emotional response.

      Study 3.3: Task difficulty is operationalized as the degree
to which all the desires in the role set can be optimized. A
perfectly optimized schedule would meet all the desires of the
entire role set. Setting higher levels and conflicting desires
increases the amount of discrepancy from the best schedule and
the difficulty of the scheduling task facing the focal manager.=20
Lowering desired levels and reducing conflicts increases the
number of schedules that might meet all the desires of the role
set, and thus decrease difficulty. As the task becomes more
difficult, conflicts increase and thus greater need arises for
cognitively intense decision making. On the other hand, load
increases as well. It is hypothesized that as task difficulty
increases, the more important goals a manager is maintaining will
receive greater resources and the less important goals less
attention. Thus, providing evidence for the importance levels of

      Study 3.4: To manipulate the need for planning over
interacting with the environment, the cost of seeking feedback
will be changed. For example, if schedules are due in 2 days in
virtual time, a 24 hour turnaround time will only allow one draft
of the schedule to be assessed. As a result, one would expect
more time would be used to think through the plan before seeking
feedback and before posting. Secondly, one with a state
orientation may expect to perform better in this situation than
one with an action orientation (Kuhl, 1991). Further, the more
dynamic the environment, the better performance one could expect
with more time planning and for a more state oriented individual.


Austin, J. T., & Vancouver, J. B. (submitted for review).=20
      Critical analysis of the goal construct in Psychology.=20
      Psychological Bulletin.

Ford, D.H. (1987). Humans as self-constructing living systems: A
      developmental perspective on behavior and personality.
      Hillsdale, NJ: Lawrence Erlbaum.

Ford, M.E. (1992). Motivating humans: Goals, emotions, and
      personal agency beliefs. Hillsdale, NJ: Erlbaum.

Frese, M., & Zapf, D. (1993). Action as the core of work
      psychology: A German approach. In H.C. Triandis, M.D.
      Dunnette, & L.M. Hough (Eds.), Handbook of Industrial and
      Organizational Psychology (2nd ed., Vol 4, pp. 271-340).=20
      Palo Alto, CA: Consulting Psychologists' Press.

Heckhausen, H., & Beckmann, J. (1990). Intentional action and
      action slips. Psychological Review, 97, 36-48.

Kanfer R. & Ackerman, P.L. (1989). Motivation and cognitive
      abilities: An integration of developmental, differential,
      and cognitive perspectives. Journal of Applied Psychology,
      74, 657-690.

Lord, R.G., & Levy, P.E. (1994). Moving from cognition to
      action: A control theory perspective. Applied Psychology:=20
      An International Review, 43, 335-398.

Marken, R.S. 1992. Mind readings: Experimental studies of
      purpose. Gravel Switch, KY: Control Systems Group.

Miller, J.G. (1978). Living Systems. New York: McGraw-Hill.

Rasmussen, J., Brehmer, B., & LePlat, J. (1991). Distributed
      Decision Making: Cognitive models for cooperative work.=20
      John Wiley & Sons.

Stevenson, M.K., Busemeyer, J.R., & Naylor, J.C. (1990).=20
      Judgment and decision-making theory. In M.D. Dunnette & L.M.
      Hough (Eds.) Handbook Industrial and Organizational
      Psychology, Vol 1. 283-374.

Vancouver, J. B. (submitted for review). Living systems theory
      as a paradigm for organizational behavior: Understanding
      humans, organizations, and social processes. Academy of
      Management Review.

Vancouver, J. B., Larson, P., & McDonald, D. (1992, August).=20
      Feedback timing effects on the performance of two goals. In
      J. B. Vancouver (Chair), The effects of multiple goal
      systems in work settings. Symposium conducted at the
      meeting of the Annual Convention of the American
      Psychological Association, Washington DC.

Vancouver, J. B., & Millsap, R., Peters, P. A. (1994).=20
      Multilevel Analysis of Organization Goal Congruence.=20
      Journal of Applied Psychology, 4, 666-679.

Vancouver, J. B., & Morrison, E. W. (in press). Feedback
      inquiry: The effects of source attributes and individual
      differences. Organizational Behavior and Human Decision

Vancouver, J. B., & Schmitt, N. (1991). An exploratory
      examination of person-organization fit: Organizational goal
      congruence. Personnel Psychology. 44, 333-352.