Today’s worldwide marketplace provides not only more customers, suppliers and competitors, but also increased complexity for the decision-making process. Simultaneously, the speed of communications makes the environment less stable and predictable and reduces the available time for examining data and relationships [10]. As a result, necessary data often is unavailable to the decision-maker for analysis or the requisite analyses are infeasible. Not surprisingly, managers are increasingly dissatisfied with established procedures for making decisions [1].
This does not mean we should abandon models or the use of decision support systems (DSS) to support decisions. Rather, DSS should blend analytical tools with intuitive heuristics to improve managers’ insights about factors too complex to build into models. In particular, DSS must facilitate the modification of results of analytical tools when they contradict intuition [2]. This is consistent with the basic goals of model use as noted by Jones [5], “developing insight into model behavior is ultimately a process of discovery, of finding trends, surprising behaviors, and comparing the behavior of the model to what is expected or observed in the real system.” Alternatively, analytical tools can test and verify intuition before applying it to the decision-making process.
Decision-making
Managers have four types of decision-making styles: left-brain, right-brain, accommodating, and integrated. The left-brain style stresses analytical and quantitative techniques and employs rational and logical methods of reasoning. Decision-makers decompose problems, approaching each subproblem sequentially using logic and data. Quantitative analyses of database-stored information lend themselves to this style of decision-making [1].
The literature documents the benefits of this approach, and designers know how to build necessary support into DSS. However, analytic methods do not always provide the best support for decisions. Left-brain style works best when all relevant variables can be controlled or predicted, measured, quantified, and when complete information is available. These conditions often are not met, and hence the exclusive use of analytic methods is inappropriate. Models do not address critical but unmeasurable considerations such as values and morals. In fact, since the analytic thought process stresses the “bottom line,” such factors are often ignored. Furthermore, since the analytic style accentuates individual criteria, holistic evaluation of complementary factors is not considered. Problem solvers are encouraged to seek single causes when many (or no predictable) causes may exist—reducing uncertainty by ignoring the unpredictable, and simplifying the complex. Finally, past data is given too much importance without adequately considering the appropriateness of the assumption that the future will imitate the past.
An alternate decision-making style is the right-brain approach. This style uses intuitive techniques, often placing more importance on feelings than facts. Right-brained decision-makers use an unstructured and spontaneous procedure of considering the whole rather than its parts, even when information is inadequate. Brainstorming and emergent trends projection are characteristic examples of appropriate use of this style.
This intuitive thought process is vastly different from the analytical approach. Analytic thought involves explicitly defining the problem, deciding on exact solution methodologies, conducting an orderly search for information, increasingly refining the analysis, aiming for predictability and a minimum of uncertainty. Intuitive thought, on the other hand, avoids committing to a particular strategy. The problem-solver acts without specifying premises or procedures, experiments with unknowns to get a feel for what is required, and considers many alternatives and options concurrently, while keeping the total problem in mind. While this approach addresses some shortcomings of the left-brain style, it has its faults, most obvious of which is the absence of data-tested theories and the use of a methodology that cannot be duplicated.
In the first two cases, decision-makers have a strong preference for a given style (either left- or right-brained) and implement it unless there are significant disadvantages to doing so. By contrast, accommodating decision-makers have dominant styles, but have learned from experience decisions for which “opposite-brainedness” is more appropriate. In those circumstances, these decision-makers adopt that alternate decision-making style.
Finally, the integrated style combines the first two, taking advantage of their obvious symbiosis. The analytical thought process filters information, and intuition helps decision-makers contend with uncertainty and complexity. Decision-makers reason, analyze, and gather facts that trigger intuition. If intuition leads the thought process in a different direction, decision-makers reason and analyze again to verify and elaborate upon it. These additional facts and analyses again trigger intuition, and the process repeats. Decision-makers can also start with an intuitive hunch, and then analyze it to determine its appropriateness. They can also apply intuition at the end of the process to reveal false premises, invalid inferences and faulty conclusions. In this way, the integrated style of decision-making utilizes both right and left-brain styles, using both facts and feelings, depending upon which is available and appropriate at the time [1]. Before discussing this further, it is prudent to define intuition.
What Is Intuition?
Intuition refers to “a sense of feeling of pattern or relationships” and vague feelings. Intuition also can be “holistic thinking, immediate insight, seeing the answer without knowing how it was reached” [12] or as “compressed expertise, a way of rapidly accessing chunks and patterns of knowledge formed from previous experience” [10]. Decision-makers might sense feelings and visual clues or see a pattern in apparently unrelated facts. To the observer, intuitive decision-making appears to include instant information processing and choice [1].
When most people think of intuition, they think of the form called illumination—a sudden awareness of information. Using this form of intuition, the decision-makers know facts or relationships, without knowing why. In reality, intuition can be triggered in various ways, many of which are tightly coupled with traditional analytic methods. Another form, detection, usually occurs when the mind is directed toward something other than solving the problem. While apparently at work on another problem, the mind reveals verifiable facts, supplies answers to questions or problems, or provides insights into the real nature of the problem. The decision-maker suddenly can draw relationships among facts or components previously appearing to have no relationship. However, such intuition can only happen after rational thought sets the groundwork and provides data and analyses as the basis for detection.
A third form of intuition, evaluation, facilitates choice among alternatives. It involves a feeling of certitude, such as “it just doesn’t feel right” in response to data analyses or reports. Such intuition can help users decide if analytical-based information is sufficient, or if inconsistent measures exist. Of course, as with the use of any form of intuition, a danger inherent in evaluation is confusing feeling with emotions or confusing it with habit. Rather, intuition is the “sudden” appearance of something new.
Similarly, prediction is a form of intuition. Prediction involves developing hypotheses without first analyzing data. It may happen alone or may accompany quantitative model building to evaluate the appropriateness of the historical data for predicting the future.
Operative intuition guides and provides a sense of direction. It suggests something that needs investigation. Sometimes operative intuition can pinpoint events that seem to have no relationship but, in reality, have a meaningful relationship.
Creative intuition is similar to detection, but it revolves around alternatives, options, or possibilities. Creative intuition supplements detection by generating ideas. The answers are discovered in a detection phase and alternatives are generated in a creative phase, or vice versa.
How Do Decision-makers Use Intuition?
When and how managers use intuition varies because people can be intuitive in some situations and not in others. As managers gain experience in certain situations, they acquire expertise by internalizing certain activities and making them automatic. The decision-maker develops an overview approach to problem-solving that guides information gathering, and provides “tricks” to link information in nonobvious ways. This overview can trigger intuitive approaches that are different from what one has seen in the past.1
While intuition can promote innovation, it can also be negative. Intuitive managers become impatient with routine, details, or repetition. They may reach conclusions too quickly, ignore relevant facts, or follow an inspiration when it is clearly bad [1].
There are several available methods to manage these tendencies. Decision-makers guided by intuition must understand their strengths, weaknesses, and vulnerabilities. They must evaluate all intuitively acquired information using appropriate analytical tests, and weigh all factors carefully without bias. Since decision-makers often seek information selectively to confirm their beliefs (and often ignore disconfirmatory evidence), particular effort must be placed upon validation of assertions. Causation and probability must be analyzed thoroughly. DSS can encourage intuition while helping users guard against inappropriate use of intuition.
Implications for Decision Support Systems
Virtual experience. Good managers are similar to chess players—over time they learn to recognize patterns of conditions for which particular tools or strategies will most likely work. The reason is that experience encourages intuition. If managers began as apprentices, and worked in the same organizations with the same products for their entire lifetimes, they would experience many decision points. This background would allow decision-makers to gain experience about the important factors in the organization and the role these factors played in creating a favorable outcome. Such experience allows decision-makers to reflect more on information provided to them, to imagine creative options, and to seek historical evidence with which to evaluate hypotheses. Their decision-making generally would be more open-ended, involving speculation about unstated possibilities. In other words, their intuition would increase.
Managers of today often do not have such intuition because they do not have longevity with organizations, products, or individuals. An alternative is to allow managers to experience those decision processes vicariously. This can happen if DSS provide convenient, quick access to databases and analysis tools so that the decision-makers can “rummage around” to extract and manipulate database fragments in ways that mesh well with individuals’ normal ways of viewing and resolving situations [9].
These managers need to access data reflecting multiple perspectives of an organization. Recent advances in data warehousing simplify this process and give decision-makers access to richer information. Without the data warehouse, DSS can only access data available from regular operations. Not only is this data insufficient in content, it is inefficient to use. Further, the data represents only a frozen slice of operations, containing factors at one particular point in time. With the data warehouse, DSS can provide nonvolatile, subject- and time-variant data to support a variety of analyses consistently. This allows decision-makers to see how factors have changed over time and how circumstances affect the issues under consideration. Such analyses help decision-makers vicariously or intellectually experience more aspects of their organizations and therefore help them to develop better intuition about what “works” and what does not “work.”
Tracking experience. Even as managers gain experience with an organization and with decision-making, they need a record of those experiences, and a mechanism for organizing the data to trigger intuition. This may include results of applying rules of thumb. Further, they may have data about past decisions, including the processes and the results. Finally, they may have data they have collected privately that they can use to obtain a strategic advantage in their corporations. Sometimes, they simply keep notes of political processes in their organizations and how they might influence or be influenced by a particular decision.
Actual decision-makers use this supplementary data to facilitate the choice process. For example, some hotels provide general managers with DSS that utilize information about profits, transactions, and physical facilities, and may maintain information collected during their decision-making processes. This information might include a database of upcoming events, such as changes in tourist attractions, changes in room availability, or conventions that might influence decisions about special promotions. Alternatively, decision-makers might keep records about special abilities of employees that would influence scheduling decisions.
DSS must simplify the development and maintenance of these private databases. Systems need to help the decision-maker generate and populate these databases, to provide easy access to the data, and to possess a range of retrieval and reporting capabilities. Whether the system resides locally, on a mainframe, or on a distributed network, it is possible to maintain private databases on one’s PC. In any case, DSS must provide sufficient security to ensure that only the decision-maker can access the information.
When making decisions, a manager considers values, ethics, morals, goals, and plans [8]. Allowing DSS users to enter this information into the system or allowing the system to deduce relevant factors based on past decisions could facilitate intuition. The system could analyze personal tendencies to determine guidance and presentation needs. Decision-makers often approach problems similarly and try to frame current problems based upon the success or failure of similar problems of the past [8]. The DSS should provide a means of locating and displaying previous problems, the decisions made, and the consequences of those decisions. This capability would support managers in their decision-making processes and stimulate intuition.
Data mining and intuition. Sometimes, data warehouses can provide so much information that users become lost in the possibilities. Data mining tools simplify analyses by employing filters based upon specific user-defined, qualifying criteria (such as a list of employees who have held specific job titles), percentile and rank filtering (such as the top 10% of their raw materials used). Users can specify information to be found regarding a particular business unit and compare it to that of multiple business units or to the company as a whole. Sometimes scanning all relevant data can help decision-makers extract similarities among events and hence inspire hypotheses.
Data mining tools often use artificial intelligence. If these tools work well, they can identify relevant data and patterns in the data, which should spur intuition or help test intuitive-based hypotheses. The patterns and rules might provide guidelines for decision-making, or they might identify the issues upon which the decision-maker should focus during the choice process [4].
Tools for analysis. Data is only useful if, once identified, it is summarized and analyzed. A wide range of models and guidance for using these models will encourage decision-makers to experiment freely and perform sensitivity analyses. DSS need to provide a library of models that give decision-makers easy access to the models.
The types of models provided will have an impact on the use of intuition. The availability of descriptive modeling tools, such as statistical tools, helps decision-makers develop intuition. Measures of central tendency and dispersion can help users get the “feel” of their data. Similarly, measures of correlation and association can suggest how variables might be associated. Providing trend analysis capabilities is important for analyzing visual representations of trends that can lead to intuitive flashes that would not otherwise occur.
Presentation for intuition. Not only must DSS perform computations, but they must present results so that decision-makers understand them; simply reporting numbers is not enough. The availability of other presentation tools can ensure that decision-makers grasp the full implications of their data. For example, graphs and charts can help decision-makers see patterns among phenomena they might not otherwise notice. To illustrate how the graphical and pictorial tools can help decision-makers, consider the data in Table 1. A regression analysis of the data indicates that maintenance costs increase as machines age and the relationship is statistically significant. But there is more that the decision-makers need to know.
It is important that DSS not simply report raw data, but that they help inspire intuition by illuminating trends, patterns or anomalies that are apparent only in graphical representations of the data. For example, Figure 1 shows that maintenance costs increase with age but at midlife there is an important drop in costs followed later by a major increase in costs (for example, due to extensive rebuilding). The figure motivates a search for a cause that the tabular data does not even identify. Therefore, if the goal is to provide improved intuition support in the DSS, then providing data in graphical formats is crucial for motivating the intuitive process. Moreover, because most individuals have difficulty visualizing in advanced dimensions, these graphs generally will be restricted to two dimensions. Further, if there is a dominant format convention in an organization, there might be further restrictions on the type of graph, such as a preference for bar charts over line graphs.
Not only might such tools generate intuitive breakthroughs, they also help verify intuition. The decision-maker may have an intuitive thought while browsing through the available data. The models in the DSS should allow the manager to test these intuition-based hypotheses using standard analytical tools. In some cases, it may be possible to test the hypotheses, while in others the analysis can only suggest the appropriateness of the hypotheses.
DSS can help decision-makers by prompting them to consider important issues, such as those associated with data mining tools. For example, one system used neural networks to analyze credit card data and provide hypotheses to decision-makers about credit card theft. The system returned with a unique insight; credit card thieves were charging low amounts to a card, such as $1 at a gas pump, to test the cards before using them for higher-cost purchases. This insight was complementary to those provided by humans, which tended to focus on large, unusual purchases.
Stimulating and testing intuition. Models and guidance should be present to ensure the decision-maker does not rely upon invalid assumptions. Geoffrion [3] states that “[o]ne must know not only what the optimal solution is for a given set of input data, but also why” [emphasis added]. Further, Jones [5] adds, we need to know where the expected and observed phenomena match and differ and why. Finally, Sharda and Steiger [11] add that decision-makers need to perceive the “inner nature of the tradeoffs inherent in any complex business situation.” In other words, simply providing the answer to a problem is not sufficient for decision support. There are three aspects of the solution that are critcal for understanding. First, DSS should help decision-makers understand what they know. This may require some predigestion of information. Sometimes it is necessary to have prepared analyses available for the manager. It may also be necessary to provide additional details that decision-makers can request. For example, the DSS might provide access to a database of position papers or in-depth discussions of analyses for decision-makers to consult.
The models in the DSS should allow the manager to test intuition-based hypotheses using standard analytical tools.
Decision-makers also need system-based intelligence that recognizes traps of analysis and brings overlooked, relevant information to the attention of the user. The system should question conclusions reached too quickly, and encourage further analysis. Tedious calculations and time-consuming analysis should be performed automatically for intuitive decision-makers because they might not otherwise be bothered with details.
A second aspect is helping decision-makers to understand the underlying assumptions by providing enough appropriate information for decision-makers to understand the issue without overloading them with unnecessary or unwanted details. DSS should provide predefined information and analyses, which allow decision-makers to identify the analyses that generated a particular result. Alternatively, DSS should provide information about promising additional analyses. This option encourages users to develop original analyses and recommends analyses, but allows the user to select desired analyses. This option allows unknowledgeable decision-makers to explore the decision environment and allows knowledgeable users to pursue subtle clues.
Finally, DSS must help users test assumptions, especially those that differ from decision-makers’ preconceived ideas. DSS can illuminate how a current context is similar to one faced previously, and why similar strategies might work; or they can help decision-makers to understand why the current context is different, and therefore why different strategies might work. Specifically, this means a DSS should have decision aids that support users’ abilities to recognize trends. This might include development of databases with which to track options, relevant factors, and outcomes of choices.
This might also require the need for alternative-generating options that might use solutions from past problems [6, 7]. If this capability is included, however, there must be some manner for considering and experimenting with these strategies in a solitary and secure manner. Decision-makers need to be able to store alternatives (with annotations) in a retrievable and searchable format, and they need to be able to consider these options and discard them (if necessary) without a record of their use. Otherwise, the highly competitive environments (both internal and external) of most managers will discourage their use.
Sharda and Steiger [11] suggest using various inductive technology tools. They illustrate how the classical statistical toolkit, statistical cluster analysis, correlation analysis and rank regression analyses can help decision-makers test assumptions. Further, they suggest neural networks to develop similarity maps with which decision-makers can glean new insights.
The DSS should also encourage users to challenge model results, especially those deviating from decision-makers’ intuition. Decision-makers can use the “stray bullet drill” to learn what might go wrong when everything seems to be going as planned [1]. Sensitivity analyses that help decision-makers answer “what if” questions should accompany all models, and the models themselves should be able to generate possible scenarios.
DSS should provide the results to the user in an understandable form. Most models provide information to the user that is not understandable by people who do not use the package frequently. One task of the model base management system in DSS is to help decision-makers understand the implications of using a model. This is not always easy because decision-makers may not be inclined to ask questions. Of course, artificial intelligence tools can incorporate expertise that would help decision-makers use models more effectively. Heuristics could be incorporated to test assumptions and guide analyses by providing suggestions. However, many managers are skeptical of information programmed through artificial tools. Therefore, the DSS must provide sufficient explanation of the reasoning process; this is especially true before decision-makers develop confidence in the quality of the system’s recommendations.
Electronic memory. Intuitive thought can disappear as quickly as it appears, and so capturing the thought and what caused it can be critical. DSS help the user re-create the process to recapture the thoughts. Re-creation of events requires storage of input screens, the models used, the input and output of the models and information viewed, and mechanisms to step through changes in the screens temporally. Stepwise analysis allows users to review concepts, alternatives, and the flow of information as it was compiled in order to better understand the process and allow identification of lost ideas. Not only can decision-makers get the general impression of ideas, they can re-create the process leading to the final positions, to help them understand the “why” behind the “what,” potentially generating even more ideas. Designers must show care in providing a complete representation of the data and to preserve the richness of the information associated with the process.
Most managers do not accept information solely because the source is unbiased. If the information were provided by a human (rather than the DSS), then who presents the information makes a difference. If managers have confidence in the people presenting an option, they have confidence in the option. This has implications for the design of DSS. It means that there must be some way to identify the source of information in the DSS. In addition, there must be a way by which users can obtain, store, access, and aggregate others’ opinions and analyses. Further, the system should include electronic access to magazines, newspapers, wire services, and other forms of media.
Conclusion
Intuition is becoming increasingly important in decision-making. Managers need tools to assist in generating intuitive thoughts and to verify the intuition once it occurs. DSS can be designed to facilitate such a goal by providing models, guidance and warning devices, and even incorporating intuition into the models themselves that support, encourage, and verify intuition.
Join the Discussion (0)
Become a Member or Sign In to Post a Comment