Training Decision Making in Organizations:
Dealing with Uncertainty, Complexity, and Conflict
di Scott Middleton

Abstract

Decision making is an employee skill that is critical to the success of all organizations. This paper discusses the relationship between decision making and problem solving, and details many strategies from the research literature that have been shown to enhance decision making for individuals and groups. Training these decision making aids to develop employee's decision making abilities, both individually, and in groups, is a wise investment in any organization's human resources.

Training Decision Making in Organizations Definition

Decision making has been defined as "the processes of thought and action that culminate in choice behavior"(MacCrimmon, 1973). Decision making and problem solving are interrelated topics both involving use of judgment, bridging thought and action. The research literature sometimes describes decision making as a subset of problem solving, dealing with evaluation and choice from a set of alternatives. Problem solving is seen as "the whole process of problem formation, alternative generation, and information processing which culminates in choice"(MacCrimmon & Taylor,1976, p. 1397). Other literature sees problem solving as a subset of decision making, since it deals primarily with "simple solutions that often have correct solutions, while decision making encompasses broader, more important contexts" (MacCrimmon & Taylor,1976, p. 1397). Obviously, decision making abilities of employees play a key role in the performance and success of an organization, so training that improves decision making is highly valuable investment in human resources development. This training can focus on two levels, individual decision making skills and group decision making skills. These are both critical in the modern, learning organization. Because such organizations have the goal of empowering employees by delegating decision making authority to the lowest levels, it would be wise to ensure those employees are well prepared for the task. Also, the ever-increasing use of work groups to make decisions makes the process more complex because of the social, interpersonal factors at play. So you need to educate employees on decision making strategies both on the individual and group level.

Components of the Decision Problem
Information processing theorists such as Reitman (1964) conceptualize decision problems by a three component vector. The first component is the initial or current state the decision maker, or the resources he/she has available. The second is the terminal or desired state he/she would like to achieve, the target or goal. The third are called transformations, decision alternatives, or operators, but essentially are the processes or steps that the decision maker must take to get from the initial condition to the desired condition. This manner of viewing decision problems shows how they are subjective and relative to the perceptions of the decision maker. If he/she doesn't perceive the gap between the two conditions, he doesn't see the need to make a decision. Even if he/she perceives the gap, he/she must be motivated to resolve it. Also, the decision maker has to have the abilities or resources available, or it is futile.

Well-Structured vs. Ill-Structured Decision Problems
To the extent that the decision maker is familiar with the three components in a decision problem, it is well-structured. Well-structured decision problems allow the decision maker to apply past experience by using transformations that have worked for similar situations in the past. Routine decisions have standard operating procedures (SOPs) that can be used. These standard responses can be algorithms that are a finite number of logical steps, that like a computer program, lead to the same output. They also can be heuristics, or rules of thumb that are procedures or outlines for seeking solutions. They increase effectiveness, but do not guarantee success. To the extent to which the components are unfamiliar to the decision maker, the decision problem is ill-structured. The decision maker will use the component he is most familiar with as a tool to solve the problem. If he only knows the current state, he will tend to make incremental moves, testing to see if it is in the right direction. If he knows the desired state, he will try to deduce backwards to derive transformations that will lead back to the initial state. If the decision maker is familiar with a certain set of transformations, he will try to specify the initial and final state to allow him to use what he knows.

Conditions Constibuting to Ill-Structuredness
Typically, most critical decisions in organizations tend to be ill-structured. Decision making strategies can be categorized by the three environmental conditions that contribute most strongly to ill-structuredness: uncertainty, complexity, and conflict. (MacCrimmon, 1970).

Uncertainty
The decision maker can try to take account of the uncertainty by trying to model it. People try to cope with uncertainty by attempting to identify patterns and fit occurrences that change in systematic ways into a model. They can plug recent observations into formulas to make time-series forecasts. They can also attempt to assign subjective probabilities to uncertain events. There are various biases and various methods for assessing subjective probabilities that can be taught to decision makers.

Alternative Generation and Testing
There also are strategies for reducing uncertainty by alternative generation and testing. Brainstorming, a well-known technique created by Osborn (1941), uses "group participation to facilitate the generation of creative decision alternatives." Brainstorming can also be done by individuals working in isolation, and then pooling their results. This produces even superior results.

The synectics approach (Gordon, 1961; Prince, 1968), is a more structured way to generate and test alternatives. It employs a carefully selected and trained group led by an experience leader through eight phases. It highlights generating creative alternatives by forcing participants to use analogies to free participants from thinking in the usual way. Once alternatives are created, they can be tested by trying them out and simulating their results. This again, reduces uncertainty by modeling and simulation.

Acquiring Additional Information
Uncertainty can also be reduced by acquiring and processing additional information. Bayesian probability revision is a strategy for "prescribing the optimal impact that additional information should have on a decision maker's judgment of initially uncertain decision outcomes"(MacCrimmon & Taylor, 1976, 1400). Research has shown that decision makers tend to be more conservative than the optimal Bayesian revision. The Probabilistic Information Processing System (PIP) can be used to help decision makers apply the Bayesian approach more accurately. Decision makers can also calculate the economic cost of uncertainty by multiplying the probability of each possible event that might occur by the cost of the best action to take in that situation. By doing this they can calculate the optimal amount of information to collect to reduce uncertainty. The expense of collecting the uncertain information can be compared to the cost of not having it.

Complexity
Experimental research and field studies of decision making have demonstrated that "decision makers find complexity a major barrier to effective decision making"(MacKinnon, 1976, p 1402). Unfortunately, the decisions typically encountered in organizations are quite complex, with many interconnected factors to be considered. The first strategy is to model the complexity. This can be accomplished by using input-output models, developed by Leontief (1966), and Decomposable Matrices, as suggested by Simon (1969).

Decision Alternatives
Complexity has been found to "hamper the ability of the decision maker to formulate decision alternatives" (MacCrimmon & Taylor, 1976, p.1404). Two strategies, using morphology and the relational algorithm, help reduce complexity by formulating alternatives. Using morphology is based on identifying key dimensions of the solution. For each dimension, values are identified, which then are taken in all possible combinations. This exhaustive strategy formulates "alternatives with very little likelihood of neglecting important elements in the decision" (MacCrimmon & Taylor,1976, p.1405). Using the relational algorithm is another systematic procedure involving the generation of combinations. An association between elements of the solution is created by using relational words.

Diagnose and Specify the Problem
Strategies that diagnose and specify the problem can be used to reduce complexity. The first approach is focusing on what is or is not part of the problem (Kepner and Tregoe, 1965). The decision maker can analyze what the characteristics of the problem are and discriminate them from the characteristics of what is not the problem. By focusing on the factors differing between the two, they can diagnose what is causing the problem. A second strategy is to focus on changes. To find causes of a problem, look at changes that occurred prior to the occurrence of the problem. If your heater is malfunctioning just prior to a gas leak, the leak is probably on the heater.

Factor into Subproblems
A complex decision problem can be factored into smaller, more manageable subproblems to reduce complexity. These subproblems then can be divided among decision makers in a group, and solved in parallel (Simon, 1969). Subproblems can be delegated to group members with relevant strengths in that area. One proviso in using this strategy is that it is not effective for highly interrelated subparts, because then the coordination problems outweigh the advantages gained.

Means-End Analysis
The means-end analysis is a useful strategy for coping with complexity in goal hierarchies or multiple goals. Decision makers analyze what means can be used to close the gap between the current situation and the desired goal state, the end. Then successive means are found to further close the gap and applied step by step. This again models the way computer programs chip away in logical steps in problem solving and decision making. Newell, Shaw, and Simon (1960) discuss a computer program called the General Problem Solver, that employs this strategy. Notice the relationship to the concept of creative tension.

Controllability
To reduce complexity, and increase manageability, a useful strategy is to classify factors of a decision problem into controllable and uncontrollable, and focus on those that are controllable. Find solutions to the controllable factors that you can do something about (MacCrimmon & Taylor, 1976).

Working Forward and Backward
To solve complex decision problems, decision maker can work in two directions, forward and backward. Most problems are attacked in a forward direction, trying some method of attack, and checking progress. Some problems are more easily solved working backwards from the desired end state. A combination of both strategies can also be used (MacCrimmon & Taylor, 1976).

Aggregating Information
Another group of strategies work to reduce complexity by aggregating information. Psychology tends to evaluate decision making from an information processing standpoint (Reitman, 1964). Both individuals and groups' information processing capacities can be severely overwhelmed by the information processing demands placed on them by complex problems typically seen in organizational settings.

Chunking
The strategy of "chunking," which helps decision makers reduce complexity by organizing and grouping information into categories or "chunks," and arranging them by order of importance (Simon, 1969). This strategy serves to effectively enhance the decision maker's information processing capacity. You can see the cognitive psychology theory behind this strategy, and its relationship with a similar technique for increasing the capacity of short-term memory.

Aggregation
Marschak (1964), suggests partitioning information effectively by using an optimal "level of aggregation" for decisions. Some information is too specific to be helpful, while other information is too aggregated and general to be useful. If detailed information is not retained during aggregation, it will not be able to be disaggregated to use later. While aggregation reduces complexity, the specifics of detailed information tend to get lost when it is aggregated, and may be needed for later decisions.

Simultaneous vs Successive Scanning
Bruner, et al. (1956) did research concerning four conjunctive strategies that reduce complexity in decision making. Simultaneous scanning is testing many hypotheses simultaneously, while successive scanning tests the hypotheses one at a time. Conservative focusing is finding one positive instance and varying one attribute at a time, while focus gambling varies more than one attribute at a time (MacCrimmon & Taylor, 1976). Conservative focusing and focus gambling are the two strategies with the most benefit in dealing with information load. If the decision maker lacks information, they should use the conservative strategy, and with more information, can adopt the more risky strategy.

Reliability and Validitity of Information
A commonsense strategy in reducing decision making complexity that people seem to automatically use in experimental settings is selecting information that they deem as valuable based on its relevance and reliability. When aggregating information, they select information that is relevant to the decision, and comes from a reliable source, while ignoring information from other sources. Decision makers seem to have well established preferences for sources that have provided information that have led to successful decision making in the past. In this regard, decision makers are rational and learn from experience as psychologists would predict. However, resistance to sources and information perceived as lower quality can be problematic, if the bias is unfounded.

Delphi Method
An effective strategy for aggregating information from the expertise of decision makers in a group is the Delphi technique (Dalkey & Helmer, 1963). This technique employs interaction among group members that are isolated from each other to preclude "forceful group members from dominating the discussion and stifling contributions of other group members" (MacCrimmon & Taylor,1976, p. 1420). The experts are separated and are given questionnaires soliciting their opinions and reasons for them. These questionnaires are then circulated anonymously to each other group member. After each round of questionnaires, information is consolidated and again circulated anonymously among group members. This is a strategy to maximize decision making benefits of a group, while limiting some of its weaknesses. With current computer technology, this technique would be especially easy to implement.

Communication Networks
The use of appropriate communication networks of decision making groups maximizes accurate and efficient exchange of communication. This reduces complexity by increasing the amount of information. In complex tasks, decentralized communication networks have been found to be more efficient (Lawson, 1965; Mulder, 1960; Shaw, 1954) and to produce more satisfaction among group members (Cohen, 1961; Lawson, 1965; Leavitt, 1951; Shaw, 1954). The down side to this communication network is that leadership is less likely to emerge in it than in more centralized ones.

Aggregated Preferences
Decision making not only benefits by complexity reduction through aggregated information, but also through aggregated preferences for alternative choices. The first decision making strategy for aggregating preferences is bootstrapping with a linear model (MacCrimmon & Taylor, 1976). This is useful for situations in which a decision maker makes a number of similar decisions. linear regression can model the decision maker's behavior. In fact, these linear regression models are more reliable than the decision maker, because it does not have as variable attention and other similiar human inconsistencies. This statistical backup can be an aid to decision making, ensuring high reliability of decisions.

Subjective Weighting
A subjective weighting model would work the same way, but instead of deriving the preferences the decision maker was using by his behavior, it would have the decision maker specify preferences and plug them into the formula as coefficients. The weighting of importance of attributes can be related to their instrumentality in achieving the goal. This is another way of explicitly describing the decision making process statistically. Notice the similiarity to multiple regression.

Mathematical Programming Methods
If available alternatives can be defined by mathematical constraints, and the decision problem is oriented towards design rather than choice, mathematical programming methods can be used to program a computer algorithm. The decision maker inputs preferences for trade-offs for incremental changes in attribute values from a reference point. This computer program will then calculate new alternatives optimally to solve the problem (MacCrimmon & Taylor, 1976).

Satisfice
A final strategy for using preferences to reduce complexity is to set constraints and satisfice. The decision maker decides preference constraints and then searches for an alternative that satisfies the constraints without trying to determine if a better alternative exists. Consumer behavior has been analyzed using this strategy (MacCrimmon & Taylor, 1976).

Span Voting Strategy
In groups of decision makers, each individual has their own preferences for constraints. Using the SPAN voting strategy of MacKinnon and MacKinnon (1969), these preferences can be taken into account without aggregating them into a social choice function. Each group member has a fixed number of votes that he allocates to alternatives based on his preferences, or to other group members to use if he feels they have more expertise for this situation.

Complexity
With decision making groups, conflict between group members can be a problem for decision making. There are several strategies to manage and possibly reduce conflict to aid decision making in group settings. These can prevent groupthink and the Abilene paradox from occurring.

Information Processing
There are five related strategies for managing conflict through information processing. Acquiring and using information concerning the conflicting opinions of other members of the group serves to manage the level of conflict and aid decision making. The first is acquiring information about one's own and other's outcomes of the decision making situation. This arises from the game-playing research literature that suggests that knowing the outcomes of opponents and the game playing strategies they use to maximize those outcomes is valuable to a decision maker. The second is to identify and utilize social motives of group members. The third is using maxi-min rules, choosing the action that maximizes each member's minimum payoff.

The fourth is to use conflicting opinions. This comes from Mason's (1969) two approaches, the devil's advocate and the Hegelian dialectic, to improve the quality of the decision by using conflict to reveal any hidden biases or invalid assumptions in the preferred alternative. The devil's advocate is a well-known strategy of presenting the opposition's side of the issue. The Hegelian dialectic requires decision makers to examine the situation completely from two opposing points of view. The fifth strategy is to use position papers to summarize each decision maker's position. This can communicate to all members of the group where everybody stands and why. Groups can write position papers as well, which has the benefit of clarifying the group's position and solidifying its goals, values, etc. (MacCrimmon & Taylor, 1976).

Barganing
Conflict can also be managed through bargaining which is based on interpersonal influence attempts. Face-to-face discussions, persuasion tactics, bluffing strategies, threatening and promising, conceding reluctantly, and responding tit-for-tat are all bargaining techniques from the social psychology research literature that can be used to reduce conflict (MacCrimmon & Taylor, 1976).

Developming Joint Agreements
The development of joint agreements among conflicting parties to reduce disparity between their desired outcomes serves to effectively reduce conflict. A disinterested third party can offer an objective viewpoint and lessen conflict, even without serving as a mediator or arbiter. Lindholm's (1965) strategy of mutual adjustment, allows moving toward a mutually preferable point by deferring judgment until the end and not requiring agreed on preemptory rules at the beginning. Forming stable coalitions such as the minimum winning size coalition and collusion and merging are other strategies of forming joint agreements and reducing conflict (MacCrimmon & Taylor, 1976).

Structural Mechanisms
Structural mechanisms such as rules, procedures, or environmental features, can assist decision makers in reducing conflict. Redefining the conflict situation in reference to a common, higher goal is a strategy to reduce conflict that happens quite often in organizations. Strategies to restructure the environment by changing communication channels or placing buffers between conflicting parties or isolating them from each other. Units of authority can issue directives to reduce conflict, such as governmental directives. Mediation and arbitration are strategies that can reduce conflict, mediation by enhanced bargaining, and arbitration by directive bargaining. Voting is a common democratic form of settling conflict, and there are various forms, ranging from majority to unanimity. Finally, logrolling is a procedure where groups can trade off votes on issues they consider unimportant, for future ones on issues they do (MacCrimmon & Taylor, 1976).

Summary
These are the major decision making strategies and aids that can be taught to employees to enhance their decision making abilities. Many major companies have established training programs in decision making (Geber, 1988). Human Resource departments are the ones who have to take established knowledge of decision making and decide how to train their company's employees in this critical skill. They also are the ones responsible for implementing the philosophy of the learning organization and serve as facilitators of employees' development (Solomon, 1994). Hopefully this review of pertinent material will be a useful source for accomplishing both these goals.

References

Bruner, J. S., Goodnow, J. J., & Austin, G. A. (1956). A study of thinking. New York, N. Y.: Wiley.

Cohen, A. M. (1961). Changing small group communication networks. Journal of Communication, 11, 116-124.

Dalkey, N., & Helmer, O. (1963). An experimental application of the Delphi method to the use of experts. Management Science, 9, 458-467,

Geber, B. (1988) Decisions, Decisions. Training, 25, 52-61.

Gordon, W. J. J. (1961). Synetics. New York, N. Y.: Harper and Row.

Kepner, C. H. & Tregoe, B. B. (1965). The rational manager. New York, N. Y.: McGraw-Hill.

Lawson, E. D. (1965) Change in communication nets, performance, and morale. Human Relations, 18, 139-147.

Leontief, W. (1966). Input-output economics. Oxford, England: Oxford University Press.

Leavitt, H. J. (1951). Some effects of certain communication patterns on group performance. Journal of Abnormal Social Psychology, 46, 38-50.

Lindholm, C. E. (1965). The intelligence of democracy: Decision making through mutual adjustment. New York, N. Y.: The Free Press.

MacCrimmon, K. R. (1970). Elements of decision making. In W. Goldberg (Ed.), Behavioral approaches to modern management, Vol. 1. (Pp. 15-44). Gothenburg, Sweden: BAS.

MacCrimmon, K. R. (1973). Managerial decision making. In J. W. McGuire (Ed.), Contempory management: Issues and viewpoints. Chapter 15. Englewood Cliffs, N. J.: Prentice-Hall.

MacCrimmon, K. R. & Taylor, R. N. (1976). Decision making and problem solving. In M.D. Dunnette (Ed.), Handbook of Industrial and Organizational Psychology (pp. 1397-1453). Chicago, Il: Rand McNally College Publishing Company.

MacKinnon, W. J., & MacKinnon, M. J. (1969). The decisional design and cyclic computation of SPAN. Behavioral Science, 14.

Marschak, J. (1964). Actual versus consistent decision behavior. Behavior Science, 9, 103-110.

Mason, R. O. (1969). A dialectical approach of strategic planning. Management Science, 15, 403-414.

Mulder, M. (1960). Communication structure, decision structure, and group performance. Sociometry, 23, 1-14.

Newell, A., Shaw, J. C., & Simon, H. A. (1960). Report on a general problem-solving program. Proceedings of the International Conference on Information Processing. Paris, France: UNESCO.

Osborn, A. F. (1941). Applied imagination: Principles and procedures of creative thinking. New York, N. Y.: Scribner's.

Prince, G. M. (1968). The operational mechanism of synectics. The Journal of Creative Behavior, 2, 1-13.

Reitman, W. R. (1964). Heuristic decision procedures, open constraints, and the structure of ill-defined problems. In M. Shelley & G. Bryan (Eds.), Human judgements and optimality. New York, N. Y.: Wiley.

Shaw, M. E. (1954). Some effects of problem complexity upon problem solution efficiency in different communication nets. Journal of Experimental Psychology, 48, 211-217.

Simon, H. A. (1969). The sciences of the artificial. Cambridge, M. A.: M.I.T. Press.

Soloman, C. M. (1994) HR Facilitates the Learning Organization Concept. Personnel Journal, 73, 56-66.