Mental Models and Decision Making
In a Dynamic Health Care Environment
Frank C. Schultz, Pradyumna Dutta, Paul E. J ohnson
University of Minnesota
321 19" Ave S
Minneapolis, MN 55455-0430
Phone: (612) 626-0320/ Fax: (612) 626-1316
pjohnson@ csom.umn.edu
ABSTRACT
Forty-two senior managers (20 MD trained and 22 MBA trained) from two large managed care
organizations were asked to make a series of strategic business decisions in a simulated (and
dynamic) health care environment (Risky Business - developed by High Performance Systems).
One half of the managers in each group (MD and MBA) were given a stable environment,
characterized by constant levels of competition and stable economic conditions. The other half
were given a turbulent environment characterized by aggressive pricing and quality
improvements by competitors plus changing economic conditions. Each manager in each
group was asked to make decisions that maximized the goals of financial performance and
quality of care. In the stable environment, eight of 21 managers (2 MDs and 6 MBAs) using
primarily feedforward decision control strategies (Brehmer, 1990) were able to stay in
business for the full 20-year period of simulated time. In the turbulent environment, two of 21
managers (1 MD and 1 MBA) using financial mental models, were able to stay in business for
the 20 years of simulated time. Mental models (financially-focused, quality-focused, balanced)
of both successful and unsuccessful managers constrained task performance depending upon
the degree of turbulence and the control strategy employed.
Background
This research investigates the thinking of health care executives as they engage in a dynamic
decision making task. Typically three factors have been shown to influence executive decision
processes in dynamic environments: 1) mental models (Lane, 1999; Doyle et. al., 1998), 2) use
of feedback/feedforward decision control strategies (Brehmer, 1990), and 3) changing
(exogenous) conditions in the decision environment (Dormer, 1997). This study extends
previous research by examining the relationship between each of these factors and task
performance in the health care industry.
The US health care industry is characterized by constant pressures to improve both the quality
of care delivered to patients as well as the financial performance of individual healthcare
organizations. While some believe that an inherent tradeoff exists between quality of care and
financial performance, healthcare organizations have been seeking ways to achieve both goals
without sacrificing either (e.g. System Dynamics Review Special Issue on Health Care, 1999).
Simulations, such as the one employed in this research, have been developed to aid senior
executives from a variety of fields in making long term strategic decisions that simultaneously
improve both the quality and the financial performance of organizations.
Study Design
The research we report employed the iThink simulation, Risky Business: Mastering the New
Business of Health, developed by High Performance Systems in cooperation with The
Healthcare Forum, Breakthrough Leaming, Inc., the 3M Foundation and twelve major U.S.
managed healthcare organizations. The simulation was developed as a training tool for
healthcare executives and was altered for research purposes (help facilities were removed and a
new interface was developed to facilitate data collection).
The simulation was administered individually via laptop computer to forty-two senior
healthcare managers (twenty MD trained and twenty-two MBA trained - no joint degree
managers participated) from two large Midwestern integrated healthcare companies. The task
of each study participant was to assume the role of a Chief Executive Officer (CEO) of an
integrated’ healthcare organization and make resource allocation decisions over a 20 year
period of simulated time (using nine potential decision variables provided by the simulation,
e.g. pricing, infrastructure investment). Participants were instructed to pursue the goal of
maximizing their company's net income and customer satisfaction.
After entering their response on each decision variable for a given year, participants were given
access to fifteen performance feedback items. In order to view a specific item (for example,
customer satisfaction) participants had to select that item from an information board (Payne &
Braunstein, 1978). The information board displayed information only when the item was
selected (by using a mouse to click the on-screen button associated with each type of
information).
The study design yielded three sources of data: 1) decisions made by each participant on the
nine decision variables, as well as results on the fifteen feedback items for each year of
simulated time, 2) a process trace created by a mouse-tracking program that provided a record
of each piece of information as it was viewed (clicked on with a mouse) and how long each
piece of information was viewed by a participant, 3) a demographic questionnaire that was
administered to participants after the completion of the simulation.
Three performance outcomes were assessed: 1) Years in Business 2) Net Income and 3)
Customer Satisfaction. Years in Business was calculated based on the number of years a
participant was able to maintain a positive case balance. Net Income and Customer Satisfaction
were provided by the Risky Business simulation based on the decisions made by each
participant.
Two versions of the Risky Business simulation were developed - one representing a stable
exogenous environment and another representing a turbulent exogenous environment. The
stable and turbulent environments were the same in all respects, including the presence of four
simulated competitors, except that over the twenty years of simulated time, the turbulent
environment reflected: 1) increasing levels of service quality and decreasing prices by
simulated competitors, 2) changing numbers of patients with insufficient resources to pay for
their healthcare services, 3) changing capitation rates’ and 4) a more rapidly aging patient
population. Participants within each group were randomly assigned to one of these two
versions of the simulation (10 MD Stable, 10 MD Turbulent, 11 MBA Stable and 11 MBA
Turbulent). Three mental models (Financial, Quality of Care, and Balanced) and two Decision
Control Strategies (Feedback and Feedforward) were assessed for each participant.
2
Results
Ten of 42 participants (7 MBA’s and 3 MD’s) successfully completed all twenty years of the
simulated task*. Of the 32 unsuccessful participants, ten exited by year seven (2 Turbulent
MDs, 1 Stable MD, 5 Turbulent MBAs, 2 Stable MBAs). Among these ten, it was primarily
the influence of extreme decisions in the first years that lead to their early departure. While the
influence of turbulence generally speeded departure (7 of the 10 were in the turbulent
environment), mental models and decision controls strategies were not significant
distinguishing factors.
A second group of 18 unsuccessful participants (4 Turbulent MDs, 7 Stable MDs, 4 Turbulent
MBAs, 3 Stable MBAs) were in business for 8 to 15 years. These 18 relied predominately on
financial mental models (13 of the 18) and feedback control strategies (13 of the 18). While
financial mental models were also characteristic of the ten successful participants, feedforward
control (as opposed to feedback control) predominated among successful participants.
A third group of unsuccessful participants (3 Turbulent MDs and 1 Turbulent MBA)
successfully negotiated the initial impacts of turbulence (they stayed in business for 16 to 18
years). As turbulent forces increased in later years, however, their reliance on a feedback
control strategy (3 of the 4 used a feedback strategy) prevented them from preparing for an
increasingly hostile competitive environment.
Figure 1 presents graphs of two participants who were unsuccessful in completing all twenty
years of the simulation. The graph shows characteristic changes over time for the participants
on the two performance measures of Net Income and Customer Satisfaction. The top graph
shows an MBA participant in a stable environment who initially generates minor declines in
performance but is able to show gains in year seven. While this MBA is able to avoid large
fluctuations in performance, she generates small performance declines each year (negative
percentage changes). By year thirteen, these initial small losses have compounded and the
MBA goes out of business. This MBA used a financial mental model and a feedback control
strategy.
The lower graph, for an MD in a stable environment, shows dramatic oscillations in
performance which eventually become irreversible and lead to the MD’s departure in year
fifteen. This MD used a financial mental model and a feedback control strategy.
Figure 2 presents graphs for two of the ten successful participants (one MBA and one MD).
Both of these participants generate fluctuations in performance over time, but avoid large
variations (usually caused by frequent short-term increases and decreases in investments and
pricing) in either performance measure and successfully complete the twenty years of simulated
time. The MD used a financial mental model with a feedback control strategy while the MBA
used a financial mental model with a feedforward control strategy.
Figure 3, summarizes the mental models and decision control strategies used by participants in
stable and turbulent environments. Feedforward strategies were used by 70 percent (7 of 10) of
successful participants (both environments combined) while only 34% (11 of 32) of
unsuccessful participants used this strategy. Similarly, 70% (7 of 10) of the successful
participants employed a financial mental model (both environments combined) while only 59%
(19 of 32) of the unsuccessful participants used this model.
Five successful (71%) participants with financial mental models used a feedforward control
strategy while two successful (29%) participants with financial mental models used a feedback
control strategy. In contrast, six (31%) of unsuccessful participants with financial mental
models used a feedforward strategy while 13 (69%) used a feedback strategy. Thus the
successful participants tended to use a financial mental model with a feedforward decision
control strategy while unsuccessful participants who used a financial mental model did so in
combination with a feedback decision control strategy.
Conclusion
Our results suggest that MDs who have entered management positions within healthcare
organizations appear to shift their mental models away from the quality of patient care to the
financial viability of organizations when making strategic decisions. This shift has positive
performance implications. MBAs, who already prefer financial mental models, shift from
feedback to feedforward decision control strategies in order to achieve high levels of
performance.
| Within the healthcare industry, an integrated organization is one that includes hospitals,
clinics and some form of insurance plan within a single organizational structure.
* Capitation refers to a payment method whereby healthcare providers are reimbursed a fixed
payment per month per patient. Increasing capitation shifts the management for risk away
from healthplans and onto healthcare providers and can therefore influence major strategic
decisions such as capital spending.
3 Participants were judged to have "gone out of business" in the decision year in which their
cash balance dropped below zero.
References
Brehmer, Berndt (1990). Strategies in Real-Time, Dynamic Decision Making. In Robin
Hogarth (Ed.), Insights in Decision Making. Chicago, IL: University of Chicago Press
Dorner, Dietrich (1997). The Logic of Failure: Recognizing and Avoiding Error in Complex
Systems. Reading, MA: Addison-Wesley.
Doyle, James K. & David N. Ford, (1998). Mental models concepts for system dynamics
research. System Dynamics Review, 14(1), 3-29.
Lane, David C. (1999). Friendly amendment: A commentary on Doyle and Ford’s proposed re-
definition of ‘mental model’. System Dynamics Review, 15(2), 185-194.
Payne, J. W. & M. L. Braunstein (1978). Risky choice: An examination of information
acquisition behavior. Memory & Cognition 5, 554-561.
System Dynamics Review (1999). Special issue on health and health care dynamics. 15(3).
Figure 1 - Performance of Two Unsuccessful Participants
Change in Net Income % and Customer
Satisfaction Performance for Participant MBAS3
30
20
10
0 (ORES SSE
-10
-20 ~*
-30
-40
0 5 10 15 20
Year
Change in Net Income % and Customer
Satisfaction Performance for Participant MDS7
30
10 ie C
0 “ = LZ ence
“10 W ey
-20 SS
-30
-40
0 5 10 15 20
Year
—¢@—Customer Satisfaction —™—NetIncome Percentage
Figure 2 - Performance of Two Successful Participants
Change in Net Income % and Customer
Satisfaction Performance for Participant MBAT1
5 10 15 20
Year
Change in Net Income % and Customer
Satisfaction Performance for Participant MDT1
5 10 15 20
Year
—¢— Customer Satisfaction —®— Net Income Percentage
8
Stable
Environment
Turbulent
Environment
FIGURE THREE
Feedforward
Strategy
Successful
Participants
©
;
Feedback
Strategy
_ > [Financial ania er
6 [Quality Mental Model)
a
> [FFnaneat Menta Moder
Feedforward
Strategy
Unsuccessful
Participants
13
Feedback
Strategy
Feedforward
Strategy
2 y [Quality Mental Wodel_}1
[Balanced Mental Modelo
+ [Fina Weis Hose
5 [ Qualy Mental oder Jo
eT
__— > [Franca ental Woe]
8 [Quality Mental Model”)
[Balanced Mental Model 2
— Financial Mental Model |1
Quality Mental Model |0
Successful
Participants
;
A
N
Feedback
Strategy
— Balanced Mental Model |0
— Financial Mental Model ]1
Quality Mental Model |0
Feedforward
Strategy
Unsuccessful
Participants
“Nt
19
Feedback
Strategy
[Balanced Mental Modet Jo
+ [Fanta Meta eae
ad [Quality Mental Wodel_]2
[Balanced Mental Model|2
__— > [Financial Wentat ode ]a
13 ,- (_Qualiy Mental Model ]3
SS Balanced Mental Model |2
