Вы находитесь на странице: 1из 5

PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING—2003 449

SITUATION AWARENESS APPLICATIONS TO EXECUTIVE DASHBOARD DESIGN

Marc L. Resnick
Florida International University
Miami, FL
resnickm@fiu.edu

With the advent of mass customization, global financial markets, and Internet-based business models,
executive decision making has an increased need for speed, scope, and accuracy. To meet these needs, IT
vendors have created executive dashboards, systems that display visualizations of critical data via
interfaces that pull from corporate data warehouses. While the visualization techniques they use take
advantage of the latest technology and may support complex data analysis, these systems often fall short
because they do not match the schema of the decision maker nor do they support situation assessment or
awareness. A design process that considers a more naturalistic decision making perspective and uses
cognitive engineering techniques such as ecological interface design would provide a significant
improvement to the design of executive dashboard interfaces. However there are several challenges to
applying current cognitive engineering methods to the executive decision making domain. This paper will
present the executive decision making domain, illustrate how cognitive engineering principles are
necessary for the design of executive dashboards, and present some methodological challenges to applying
these principles.

INTRODUCTION come to a distressing conclusion. Often, the cost of


acquiring knowledge exceeds its value toward executive
Naturalistic decision making (NDM) has been decision making. The implications of raw data are
studied in a variety of domains. Some involve the use of seldom obvious and need to be interpreted within the
complex interfaces, such as air traffic control (Rodgers, context of collection circumstances and possible uses.
Mogford and Strauch, 2000), nuclear power plants They find that the key requirement of executive
(Roth, 2001) and commercial and military pilots information systems is to provide a clear and consistent
(Amalberti and Deblon, 1992). Some involve very little framework from which the executive can create a
technology, such as fireground commanders (Klein, schema of the environment. As with pilots, fireground
2000a). But regardless of the technological differences, commanders, and other complex decision makers,
these domains have many things in common, including identifying the general nature of the problem is key to
complexity, uncertainty, significant consequences, selecting the proper approach.
competing goals, and time constraints (Zsambok and Sutcliffe and Weber (2003) directly correlated
Klein, 1997). decision making with macro indicators of business
An additional domain that shares many of these performance and found that the accuracy of the
characteristics is executive decision making. The speed executive’s knowledge of particular details is less
and accuracy with which business executives’ decisions important than having a structural schema that
must be made has increased substantially over the past maximizes the executive’s situation awareness and his or
decade. Reasons for this acceleration vary for each her ability to improvise and adjust quickly. Becoming
industry, but include a decrease in the typical product “lost” in the raw data was often worse than not having
development cycle, integration of processes such as the data in the first place. This contrasts with the
production and development into enterprise systems, a approach of many executive dashboard designers which
shift to Internet-based business models, and the is to maximize the amount of data users can access.
sensitivity of the stock market to fluctuations in a Errors on the opposite end of the spectrum are also
company’s financial position. The similarities between common. Bonabeau (2003) reports that executives
executive decision making and other NDM domains frequently use intuitive decision making styles. He
suggest that similar techniques can be used to evaluate suggests that this stems from the natural pattern
the decision making processes and design support recognition capabilities of the human information
systems and interfaces to accommodate them. processing system. The lack of specific training in
Sutcliffe and Weber (2003) provide a detailed decision making reduces the ability of the executive to
analysis of the knowledge needed by executives and recognize and overcome these processes. However, in
450 PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING—2003

the complex environment of business decisions, intuition businesses (Kaplan and Norton, 2001). Each quadrant is
is frequently wrong. Notorious cases of successful divided into components , the number of which depends
intuition (such as Jack Welch of General Electric Corp.) on the particular strategy selected by the company.
often mask the more numerous cases of failures of Certain parameters are immediately visible, based on the
intuition in the business environment. Premature color of the display, and within each component there
commitment to a particular understanding of the are a number of indicators. Additionally, more detailed
environment narrows an executive’s thinlung and data, historical analysis and other advanced features are
reduces the alternatives that will be considered. available by drilling down into each component.
Ecological models can be used to identify the The kinds of decisions executive dashboards are
information sources that should be considered for each designed to support vary widely in scope and depth.
type of decision environment and focus the user more Some decisions focus on details from just one function
directly on the most diagnostic sources. This can while others require integration of data from several
increase the likelihood that natural pattern recognition functions. Hung (2003) also includes trend analysis and
processes and anchoring bias the user towards to the multi-dimensional analysis as typical capabilities of an
correct interpretation. executive dashboard; thus the interface must support
integrating substantial quantities of data in complex
ways. However, the decisions are not the same for each
EXECUTIVE DASHBOARDS
business. Depending on the strategy, industry, market
position and other factors, the information required to
In order to assist executives in making these
make decisions will vary. In order to support this
complex decisions, a new kind of computer system has
considerable breadth of requirements, executive
been developed - the executive dashboard. Executive
dashboards must be customizable for various types of
dashboards are systems that allow company executives
users and tasks.
to view key business facts, providing a circumspective
Current designs often have powerful data analysis
view that is designed to support effective decision
functionality, but are not based on cognitive engineering
making. Functionality can include completely
principles, either regarding the information content that
configurable displays based on ERP-connected data
is provided, how this information is organized for the
warehouses and analytic modules. Executive dashboards
user, or the design of specific displays and navigation
are intended to provide managers and executives with
controls. They often are used by executives with a great
the business intelligence they need to benchmark and
deal of business knowledge, but perhaps very little
improve business processes, motivate employees,
experience with executive dashboards or other decision-
increase organizational learning, and improve decision-
support technologies (Hung, 2003). And with the high
making to keep pace with or out-perform the
turnover among corporate executives that has become a
competition (Gledhill, 2002).
rule rather than a trend, experience with a particular
As with traditional dashboards for automobiles and
implementation of an executive dashboard and
aircraft, executive dashboards are composed of several
familiarity with specific data structures cannot be
gauges that represent various aspects of the system to be
assumed. However current executive dashboard designs
managed. Representations can include current status,
do not always support the prospective users’ task
history, rates of change, warnings and indicators, and
specific needs.
other formats depending on the data type and source.
The depiction of priority can be accomplished using
SITUATION AWARENESS
typical screen design techniques such as location, size
and color.
It would be much more effective if executive
However, unlike typical dashboards, the decisions
dashboards were developed to maximize the situation
that must be supported are much more complex and the
awareness (Endsley, 2000a) of the executive user. It is
amount of data that must be accessible can be enormous.
critical that the interface draws attention to the
Data from functions such as finance, operations,
parameters that are most influential in the current
marketing, sales, distribution, competitive positioning,
situation, facilitates the integration of these data into an
and others may be needed for certain kinds of high level
assessment of the business’ current situation, and assists
decisions, and thus must all be available to the executive
in the prediction of what is likely to occur next. These
in an intuitive and intelligible fashion.
correspond to the three levels of situation awareness
For example, Figure 1 presents a hierarchical
described by Endsley (2000b) as perception,
dashboard that is divided into quadrants based on the
comprehension, and prediction. Considering the
Balanced Scorecard method commonly used to manage
tremendous amount of data that is contained within an
PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING—2003 451

Figure 1. Sample screen from an executive dashboard (Qualitech Solutions, Inc.)

executive dashboard, the interface must support a that the retailers had with its distributors, publishers,
filtration process whereby executives’ attention is led to franchisees, employees and other partners did not
the appropriate data. One serious implication of this change. During the period between 1998 and 2003, the
need is for the system to help the executive limit his or relative performance of Amazon and online retailing in
her consideration to those data that are most relevant general went through many stages. As this new data
(Pew, 2000). became available, the implications for bricks and mortar
Wickens’ (2000) use of SA to refer to a ‘continuous retailers needed to be constantly reevaluated.
extraction’ of information from the environment is The extraction of information from data sources is
perhaps the most relevant to the executive decision often driven by the user’s goals and expectations
making domain. But while the research cited by (Endsley, 200 1). Considering the extended durations of
Wickens generally refers to tasks that take minutes and executive decision making situations, expectations are
hours, the executive is extracting information over more prevalent and more strongly ingrained into the
months and years and must maintain a consistent schema executive’s schema, thus creating more opportunity for
of the competitive environment that evolves from some bias when the expectations are unfounded. Any support
events but persists despite extreme shocks from others. system must help executives identify these biases and
For example, the situation awareness of executives perhaps overcome them. This is particularly relevant
at major book retailers such as Barnes and Noble when a potentially large shift in the market occurs.
regarding their competitive landscape went through For example, when Amazon began selling books
major changes when Amazon.com entered the market. over the Internet, the traditional retailers were forced to
Amazon priced its books well below the prices found at deal with this completely new model of competition.
bricks and mortar stores and made a much wider However, the ingrained marketing channels, distribution,
selection of books available. However the relationships vendor relationships, inventory control and other
452 PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING—2003

business processes they had were not going to disappear post-hoc methods (Rodgers, Mogford, and Strauch,
either from reality or from the strong decision making 2000) are the most appropriate.
schema that executives had developed over thousands of The personalities of the executive users also may
repetitions (Zsambok, 1997). Executive dashboards can challenge the use of SA methods in ways that are not as
help executives make better strategic decisions by problematic in traditional SA domains. Subjective
focusing attention on those factors that are relevant to methods require participants to reflect on how confident
each kind of decision - some of which must be radically they are in their SA (Jones, 2000). However executives
reconfigured in the new Internet-based market while may be less able to introspect their SA than pilots and
others should not. commanders, in part because of the personalities that
tend to obtain executive level positions and in part
SA METHODS FOR EXECUTIVE DASHBOARD because of an unwillingness to concede a lack of
DEVELOPMENT awareness. The long duration of each decision unit may
also create challenges. SA may vary greatly between the
Unfortunately, many of the methods currently used beginning and end of a single decision, both in reality
to evaluate SA in other domains are not feasible in the and in subjective perception.
executive decision making domain, both from content
and methodological perspectives. Companies are very NEW SA METHODS
different from each other in terms of strategic goals,
organizational structures, operational processes, and the As a result, there is a need to extrapolate the findings
products and services they offer. Within each company, of SA research from other areas to gain insight into
there are only a few individuals who are responsible for executive dashboard design. SA knowledge can be
executive level decisions. Even within each functional drawn from past research into the design of systems and
area (ie. marketing), there are only a few people who are displays for pilots and air traffic controllers, training of
involved in the highest level strategic decisions. individuals and teams, and others to design executive
Therefore creating an evaluation scenario with a large dashboards that facilitate the development of SA for
enough sample population to collect meaningful data is business executives. But this extrapolation must be done
rarely possible. Whereas 747 pilots interact with very carefully by experts in both SA and the business domain
similar challenges using similar interfaces regardless of to insure that it is applied appropriately.
what airline they work for, this cannot be assumed for Reising and Sanderson (2002) describe a
marketing executives selecting a brand strategy. methodology using ecological design to create an
Therefore new techniques that can be used with much interface for an industrial decision support system that
smaller sample sizes must be developed. Pritchett and accurately maps the semantics of the applied
Hansman (2000) describe the advantages and environment directly to the display. Similar techniques
disadvantages of several SA methods, but they do not can be used specifically for the creation of executive
consider the sample sizes required. dashboards.
A second difference is the timing of the decision Testing of executive dashboard systems to evaluate
making process. Although in many cases the actual their support of SA must also evolve to fit the needs of
decision must be made quickly, the development of the the business domain. One cannot test a generic
situation awareness occurs over months and years. implementation of an executive dashboard across several
Events that occurred in past times may not only companies because the interface must be significantly
contribute to general domain knowledge but also may be customized for each one in order to be relevant. The
directly relevant to the current decision. effectiveness of these systems is limited out of context.
c
-- Again considering the domain of book retailing, the Therefore more of a case study method that evaluates
,JILranceof Amazon into the market affected pricing and qualitative data must be used. Effective methods do not
delivery, but the marketing of each genre of book to currently exist and must be developed specifically for
particular user groups did not change. Selecting an this domain.
online marketing strategy for each user group requires
combining the new and old information into one
integrated schema. CONCLUSION
This is problematic because running simulations that
take months to complete is not a feasible analysis The executive decision malung domain is a rich
method. Direct measurement methods such as those environment for cognitive engineering because of the
described in Endsley (2000~)would not be feasible. complexity of the decision making environment and the
Perhaps critical incident reports (Klein 2000b) and other importance of these decisions to the performance of the
PROCEEDINGS of the HUMAN FACTORS AND ERGONOMICS SOCIETY 47th ANNUAL MEETING—2003 453

company and often to national economies. However the Kaplan R.S. and Norton D.P. (2001). The Strategy-Focused
current architecture of executive dashboards fails to take Organization. Harvard Business School Press: Cambridge,
advantage of what cognitive engineering can provide. MA.
Cognitive engineering principles such as situation Klein G. (2000a). Sources of Power. MIT Press: Cambridge,
MA.
assessment and awareness are critical for successful
Klein G. (2000b). Analysis of situation awareness from
executive decision making and can be used to improve critical incident reports. In M.R. Endsley and D.J. Garland
executive dashboard design. But SA research methods (eds). Situation Awareness Analysis and Measurement.
must be modified to investigate executive decision Lawrence Erlbaum Associates: Mahwah, NJ.
making effectively. Executive decision making is a new Pew R.W. (2000). The state of situation awareness
and challenging domain for SA research and practice. measurement: heading toward the next century. In M.R.
Endsley and D.J. Garland (eds). Situation Awareness
Analvsis and Measurement. Lawrence Erlbaum Associates:
REFERENCES Mahwah, NJ.
Pritchett A.R. and Hansman R.J. (2000). Use of testable
Amalberti R. and Deblon F. (1992). Cognitive Modeling of response for performance-based measurement of situation
fighter aircraft process control: a step towards an intelligent awareness. In M.R. Endsley and D.J. Garland (eds).
on-board assistance system. International Journal of Man- Situation Awareness Analysis and Measurement. Lawrence
Machine Studies. 36,639-671. Erlbaum Associates: Mahwah, NJ.
Bonabeau E. (2003). Don't trust your gut. Harvard Business Reising D.V.C. and Sanderson P.M. (2002). Ecological
Review, May, 116-122. Interface Design for Pasteurizer 11: A process description of
Endsley M.R. (2000a). Theoretical underpinnings of situation semantic mapping. Human Factors, 44,2,222-241.
awareness: A critical review. In M.R. Endsley and D.J. Rodgers M.D., Mogford R.H. and Strauch B. (2000). Post hoc
Garland (eds). Situation Awareness Analysis and assessment of situation awareness in air traffic control
Measurement. Lawrence Erlbaum Associates: Mahwah, NJ. incidents and major aircraft accidents. In M.R. Endsley and
Endsley M. R. (2000b). Situation Assessment Analysis and D.J. Garland (eds). Situation Awareness Analysis and
Measurement. Lawrence Erlbaum Associates: Mahwah, NJ. Measurement. Lawrence Erlbaum Associates: Mahwah, NJ.
Endsley M.R. (2000~).Direct measurement of situation Roth E.M. (2001). Situation Assessment Panel. Presented at
awareness: validity and use of SAGAT. In M.R. Endsley the Human Factors and Ergonomics Society 45thAnnual
and D.J. Garland (eds). Situation Awareness Analysis and Meeting. Human Factors and Ergonomics Society, Santa
Measurement. Lawrence Erlbaum Associates: Mahwah, NJ. Monica, CA.
Endsley M.R. (2001) Designing for situation awareness in Sutcliffe K.M. and Weber K. (2003). The high cost of accurate
complex systems. Proceedings of the Second International knowledge. Harvard Business Review, May, 75-82.
Worhhop of Symbiosis of Humans, Artifacts and Wickens C.D. (2000). The trade-off of design for routine and
Environments, Kyoto, Japan. unexpected performance: implications of situation
Gledhill J. (2002). Create value with IT investments: how to awareness. In M.R. Endsley and D.J. Garland (eds).
generate a healthy ROI across the enterprise Food Situation Awareness Analysis and Measurement. Lawrence
Processing, Sept 2002 v63 i9 p76-81. Erlbaum Associates: Mahwah, NJ.
Hung S.Y. (2003). Expert versus novice use of the executive Zsambok C.E. (1997) Naturalistic decision making:
support systems: an empirical study. Infirmation & Where are we now? In Naturalistic Decision Making.
Management, 40, 3, 177-189.
C.E. Zsambok and G. Klein (eds). Lawrence Erlbaum
Jones D.G. (2000). Subjective measures of situation
awareness. In M.R. Endsley and D.J. Garland (eds). Associates: Mahwah, NJ.
Situation Awareness Analysis and Measurement. Lawrence Zsambok C.E. and Klein G. (1997). Naturalistic Decision
Erlbaum Associates: Mahwah, NJ. Making. Lawrence Erlbaum Associates: Mahwah, NJ.

Вам также может понравиться