Strategic Planning And Forecasting Fundamentals
From Kenneth Albert (ed.), The Strategic Management Handbook. New York: McGraw Hill, 1983, pp. 2-1 to 2-32.
J. Scott Armstrong
Individuals and organizations have operated for hundreds of years by planning and forecasting in an intuitive manner. It was not until the 1950s that formal approaches became popular. Since then, such approaches have been used by business, government, and nonprofit organizations. Advocates of formal approaches (for example, Steiner, 1979) claim that an organization can improve its effectiveness if it can forecast its environment, anticipate problems, and develop plans to respond to those problems. However, informal planning and forecasting are expensive activities; this raises questions about their superiority over informal planning and forecasting. Furthermore, critics of the formal approach claim that it introduces rigidity and hampers creativity. These critics include many observers with practical experience (for example, Wrapp, 1967).
This chapter presents a framework for formal planning and forecasting which shows how they interact with one another. Suggestions are presented on how to use formal planning for strategic decision making. (For simplicity, references to planning and forecasting in this chapter will mean formal strategic planning and forecasting.) Planning is not expected to be useful in all situations, so recommendations are made on when planning is most useful. Descriptions of forecasting methods are then provided. Finally, suggestions are made on which forecasting methods to use when developing plans for a company.
Where possible, the advice on planning and forecasting is supported by relevant research. In some areas much research exists. (For a review of the psyc hological literature on forecasting and planning, see Hogarth and Makridakis, 1981.) In many areas, however, little research has been done.
Various aspects of formal planning and forecasting are illustrated here by using the strategic decision by Ford to introduce the Edsel automobile in 1957. In this situation, formal planning and forecasting would have been expected to be useful. Judging from published accounts by a participant at Ford (Baker, 1957) and an observer (Brooks, 1969), Ford did not use formal planning and forecasting for the strategic decisions involved in the introduction of the Edsel. (Of course, having decided intuitively to proceed, they did carry out operational planning for the production of the car.) The introduction of the Edsel is regarded as one of the largest business errors of all time. Ford itself lost $350 million. Their dealers also lost a substantial amount. Is it possible that formal planning and forecasting might have protected Ford from such a large strategic error?
? With acknowledgments to Richard C. Hoffman IV, Spyros Makridakis, Deepak Mehta and Robert Fildes, who provided useful comments on various drafts of this chapter. Support for this paper was provided by IMEDE in Lausanne, Switzerland.
Figure 2-1 provides a framework to conceptualize strategic planning within a company. A scanning of the environment yields relevant data for the “Data Bank.” This data bank (or information system) would contain such data as government regulations, demographic indicators, industry sales, the resources of the company and of its competitors, and information on available technologies for production. Ideally, these data would be assembled in a central location, such as in a filing cabinet, chart room, or computer.
The left-hand side of Figure 2-1 examines planning. A variety of planning processes can be used. These will be described in more detail below. The planning processes draw upon information from the data bank (evidence on the current situation) and also upon the forecasts evidence on what will happen in the future). The two-way arrow from “Data Bank” to “Planning Processes” indicates that the planning process, to a large extent, dictates what information is required. It is recommended that formal planning start with the planning process rather than with the data.
The planning process produces a set of plans. These describe objectives and alternative strategies. One strategy is selected as a basis for action. In practice, the actions actually taken by the company can deviate substantially from the intended strategy. The actions lead to results, both intended and unintended. A record of these results is kept in the data bank.
The right-hand side of Figure 2-1 examines forecasting. To make forecasts for a company, it is necessary to have information about the company's proposed strategies (thus the arrow from “Plans” to “Forecasting Methods”). An examination of the forecasting methods, then, will help determine what data are required (thus the two-way arrow from “Data Bank” to “Forecasting Methods”). The forecasting methods, to be described in more detail below, yield a set of forecasts. What will happen if the company attempts strategy A and environment X occurs? How likely is environment X? How much confidence can we have in the forecast? These forecasts are then used as inputs to the planning process.
Note the distinctions between forecasting and planning. Planning provides the strategies, given certain forecasts, whereas forecasting estimates the results, given the plan. Planning relates to what the firm should do. Forecasting relates to what will happen if the firm tries to implement a given strategy in a possible environment. Forecasting also helps to determine the likelihood of the possible environments.
The remainder of this chapter discusses the items in the two circles on Figure 2-1, the Planning Process and Forecasting Methods.
DESCRIPTION OF THE STRATEGIC PLANNING PROCESS
Formal strategic planning calls for an explicit written process for determining the firm's long-range objectives, the generation of alternative strategies for achieving these objectives, the evaluation of these strategies, and a systematic procedure for monitoring results. Each of these steps of the planning process should be accompanied by an explicit procedure for gaining commitment. This process is summarized in Figure 2-2. The arrows suggest the best order in which to proceed. The need for commitment is relevant for all phases. The specification of objectives should be done befo re the generation of strategies which, in turn, should be completed before the evaluation. The monitoring step is last. The dotted line indicates that, to some extent, the process is iterative. For example, the evaluation may call for going back to the generation of new strategies, or monitoring may require a new evaluation of strategies.
The various steps of the planning process are described below along with some formal techniques that can be used to make each step explicit. (Although commitment is the first step, it is easiest to discuss this last.) This discussion is prescriptive; it suggests how planning should be done. Numerous accounts are available of how formal strategic planning is done (for example, see Wood, 1980, and the extensive review of the descriptive research by Hofer, 1976).
Formal planning should start with the identification of the ultimate objectives of the organization. Frequently, companies confuse their objectives (what they want and by when) with their strategies (how they will achieve the objectives). For example, suppose that a company desires to make money for its stockholders. To do this, it decides to build a tunnel through a mountain in order to charge tolls to automobiles. They plan to complete the tunnel in five years. On the way through the mountain, they strike gold. To mine the gold, activities on the tunnel must be suspended. Does the company pursue its objective of making money or does it stay with its strategy of tunnel building? What would your organization do?
The analysis and setting of objectives has long been regarded as a major step in formal strategic planning. Informal planners seldom devote much energy to this step. For example, in Baker's (1957) summary of the Edsel, less than 1 percent of his discussed concerned objectives. Unfortunately, the identification of objectives is a difficult step for organizations. It is even difficult for individuals. The simplest way to demonstrate this is the following: Ask yourself to set objectives for your use of this chapter. Write your objectives. Be specific. Find measurable objectives. Set time deadlines for implementing changes. It is possible (for example, you could have as an objective that you will take action within the next month on at least one technique to improve the strategic planning of your organization), but it is stressful.
The difficulties in setting objectives have led some observers to recommend that formal planners ignore this step. The recommendation here is just the opposite. Significant time and money should be allocated to the analysis of objectives. This difficult step might be aided by use of an outside consultant to help the group focus only upon the objectives. The question can also
be attacked by asking what results would define successful performance by the company over the next twenty years. At this stage, no concern should be given as to how to achieve the objectives.
Companies pursue many objectives and planners should explicitly recognize all of the important objectives of the system. One way to help ensure that the analysis of objectives is comprehensive is to use the stakeholder approach. This calls for a listing of all groups that contribute resources to the firm. Then a description is provided of the objectives of each of these stakeholders.
Applying the stakeholder approach in the Edsel case, the following groups would be included: creditors, stockholders, employees, consumers, suppliers, dealers, and the local community. In many cases, these groups will have conflicting objectives. The planners would write out the objectives for each group, for example, return on investment (ROI) for stockholders; stability, good wages, and good working conditions for employees; safe and reliable products at a low price for consumers; ROI for the dealers. Specific measures would then be established for each objective (for example, ROI should exceed 10 percent per year after taxes in real dollars). In contrast to this stakeholder approach, Ford's informal approach led to a narrow objective: “to obtain 3.3 percent to 3.5 percent of the auto market” (Baker, 1957). Explicit consideration was not given to other stakeholders.
A strengths and weakness analysis should then be conducted. This calls for an inventory of the organization's resources (such as financial, marketing, production). What do they have now and what do they plan to have? The objectives would then be drawn from what is desired (stakeholder analysis) and what is feasible (strengths and weakness analysis).
The written statement of objectives should start with the ultimate objectives. These general objectives would then be translated into more specific objectives so that each decision maker can see how to contribute to the overall objectives. In addition to being specific, the objectives should be measurable (Latham and Kinne, 1974). The objectives would include statements on what is desired and when. Thus, the marketing department can refer to the planning manual to determine its role in meeting the overall company objectives.
One danger in planning is that the objectives may become confused with the strategies. For example, a company might decide that one strategy, to better meet the needs of its stakeholders, is to increase its market share during the next five years. But this strategy might falsely be regarded as an objective by the marketing department. Five years later, the department might still pursue market share—even if it is detrimental to the company's objectives. (They continue to build the tunnel and ignore the gold.)
Advocates of informal planning argue that specific written objectives create political problems within the organization. Vague objectives allow for the greatest flexibility in actions. Politically oriented leaders often prefer that the objectives be unstated. But evidence from studies in organizational behavior suggests that explicit and specific objectives are of substantial benefit, especially when used in conjunction with the other planning steps (see reviews of this research in Latham and Yukl, 1975; Tolchinsky and King, 1980; and Locke et al., 1981).
Once the objectives have been specified, the planners can proceed to the generation of strategies. If the objective setting was successful, the remaining steps will be easier.
Generate Alternative Strategies
A strategy is a statement about the way in which the objectives should be achieved. Strategies should be subordinate to objectives. That is, they are relevant only to the extent that they help to meet the objectives.
This advice is obvious but often ignored. The generation of alternative strategies helps to avoid this problem. It recognizes explicitly that the objectives may be achieved in many different ways.
Strategies should first be stated in general terms. The more promising strategies should be explained in more detail.
The planning process is not complete until the company has at least one (and preferably more than one) operational strategy. An operational strategy describes:
1. What tasks must be done 2. Who is responsible for each task 3. When each task must be started and completed 4. The resources (time and money) available for each task 5. How the tasks relate to one another
This operational strategy becomes the basis for action by various functions in the firm: finance, personnel, production, and marketing.
Alternative strategies can improve the adaptability of the organization in two ways. First, by explicitly examining alternatives, it is likely that the organization will find some that are superior to their current strategy. Second, the environment might change; if alternative (contingency) plans have been prepared, the organization is in a better position to respond successfully. Alternatively, they can select a strategy that performs well even if the environment changes.
Organizations sometimes have difficulty developing alternative strategies to deal with unfavorable environments (threats). One technique that can help organizations with this problem is the use of scenarios. This involves having decision makers write stories about the future of their company. They can write a scenario describing what will happen to their company if the threat occurs, given their current strategy. Then, they could write about a desired future. What would they want the company to be like? The question then becomes, “What must we do to achieve this type of future?” Consideration can be given to changing the organization's resources or to the use of alternative strategies.
The development of scenarios calls for creativity within the organization. To bring out this creativity, it is helpful to use brainstorming. Key stakeholders for the organization can be asked to consider alternative strategies, alternative resources, and alternative environments by following these rules for brainstorming:
1. Gain agreement within the group to use brainstorming. 2. Select a facilitator. The facilitator:
For a more complete pp. 24-29), and Chapter 10.
a. b. c. d. e.
Records ideas as they are mentioned Encourages quantity of ideas Reminds the group not to evaluate (either favorably or unfavorably) Encourages wild ideas Does not introduce ideas
description of scenarios, see Armstrong (1978a, pp. 38-43), Ackoff (1970,
It is difficult to say how many alternative strategies should be listed. Certainly more than one! But the number could quickly get out of hand considering the vast number of possible combinations. Try to list strategies to deal with dramatically different yet likely environments. After this larger list has been developed, screen the list to determine which strategies should be developed in more detail.
Two guidelines appear to be of particular importance for the development of a strategy. The strategy should be comprehensive and it should provide slack.
To ensure that strategies are comprehensive, planners have typ ically suggested the use of flow charts. These list each of the key tasks that must be accomplished and show how each task relates to the others. Numerous publications have offered advice in this area (for example, Ansoff, 1965; Steiner, 1979). Slack means that resources (time, money, facilities) should not be fully committed to the recommended strategy. Some resources should be held in reserve; these can be used to relieve stress if parts of the plan break down. Slack is analogous to the use of inventories. The use of slack adds flexibility to the plan.
The Edsel case illustrated the informal approach to strategic planning. Ford decided to build a large, powerful, and ornate automobile. They did not report that they examined alternatives. Their plan did no t appear to be comprehensive, and no mention was made of provision for slack. The environment changed prior to the introduction. Ornate cars were not so popular, small foreign cars were capturing a growing market segment, large powerful engines were the subject of much criticism, and a small recession was under way when the first Edsels came onto the market. But Ford had no contingency plans. In retrospect, low-cost contingency plans could have been introduced. For example, the distribution of the cars could have been done primarily through existing dealers rather than through the new Edsel dealer network. (This recommendation was proposed by management students who developed a plan for a disguised version of the Edsel case. Most of these students, who had been asked to try formal planning, decided to use the existing dealer network.)
Evidence from studies in organizational behavior suggests that, in general, the generation of ideas should be separated from the evaluation of ideas (Mater, 1963); they cannot be done together with much effectiveness. Thus, this step of generating alternatives should be completed before the next step is begun.
Evaluate Alternative Strategies
Once sufficient strategies have been proposed, the evaluation of alternatives can begin. This requires a procedure by which each alternative plan is judged for its ability to meet the objectives of the organization. Such a process is not simple, because conflicting objectives usually exist among stakeholders. Furthermore, the presence of uncertainty complicates the choice of a strategy. For example, one should consider not only how well the strategy does for the most likely situation, but also how well it does against other possible situations, especially those that are dramatically different.
One procedure for the evaluation of alternatives is the Delphi technique. Various strategies (for a given environment) are presented to the key stakeholders. Each person works independently to rank these alternatives. A summary of the group rankings is then presented to these same stakeholders, and they are asked to provide a second ranking, still working independently. This procedure can be repeated for a number of "rounds" As a variation, group discussion can be used to exchange information between rounds. The Delphi technique provides a more efficient and less biased way to use the information held by the key decision makers than that provided by informal methods (for more on Delphi see Linstone, 1975).
The use of scenarios is also relevant to evaluation, particularly when dealing with negative evidence from the environment. Much research suggests that organizations avoid unpleasant information. As an example of this tendency to reject negative evidence, Griffith and Wellman (1979), in a study of expansion plans in six hospitals, found that forecasts of decreasing demand were ignored. As a result, the hospitals overbuilt. The use of scenarios might have identified the reactions to unfavorable forecasts prior to investing money on these forecasts. The hospitals could then have canceled the proposed expenditures on forecasting if they could not decide how the forecasts might affect their decision making.
Other formal procedures for evaluation can also be used. For example, structured rating sheets can be used to evaluate the general strategies against the stakeholders' objectives and to gauge the extent to which negative information was considered. Also, one could rate each operational strategy on the extent to which it succeeded in the following areas: provided adequate resources, allowed adequate slack, set reasonable time deadlines, presented a comprehensive strategy, and presented an operational strategy. The use of the devil's advocate, when a person argues against a favored alternative, can help to ensure that both sides of a plan are considered (Cosier, 1978).
The major point for evaluation is to use formal procedures and to not use informal ones, such as the traditional group meeting. The latter provides one of the poorest ways to evaluate strategies. Janis (1971) examined a number of major failures in strategy evaluation, such as the
Bay of Pigs, and concluded that much of the blame was due to the lack of formal processes for evaluation. He provided a checklist that groups can use to improve their ability to generate and evaluate alternative strategies.
The evaluation step concludes with the selection of an operational strategy. This is the strategy the company will attempt to implement. (This strategy should contain contingency plans also.) But will the strategy really meet the objectives? To assess this, the next step of the planning process, monitoring results, is taken. This step is prepared prior to the implementation of the strategy.
The value of feedback has been well established in laboratory studies, especially when combined with the setting of objectives (Tolchinsky and King, 1980; Ilgen, Fisher, and Taylor, 1979). Field studies have also demonstrated the value of explicit feedback (for example, Becker, 1978). It seems important, then, to provide feedback to the organization on how well they are meeting their objectives. In other words, specific procedures should be developed to “monitor results.”
The monitoring system should allow for corrective action. To do this, the following items should be measured in a systematic way:
1. Changes in the environment (sometimes called “environmental scanning”) 2. Changes in the organization's capabilities (and in their competitors' capability) 3. Actions that were actually taken by the organization (did they implement the
desired strategy?) 4. Actions by major competitors 5. Results
Planning involves a trade-off between consistency and flexibility. Formal planners try to develop a strategy so that a complex organization can operate in a coordinated manner. The members of the organization must sacrifice flexibility in order to follow a consistent strategy. However, changes in any of items 1 to 5 above could suggest a change in strategy. Thus, the monitoring system should signal when a change in strategy should be considered.
Fixed review times should be selected in advance. Many firms conduct a review once a year. At these times, decisions should be made whether to continue with the original strategy, revise the strategy, or switch to a contingency plan. For very large changes it is best to view the strategy as being experimental and to schedule more frequent review periods, perhaps quarterly. In addition to fixed review times, the monitoring system should also have control limits. These would be upper and lower bounds for each of the above five areas. When the system goes outside of these limits, a planning review would be conducted whether or not it was time for the fixed review.
The monitoring of outcomes should relate back to the objectives for each stakeholder. This should allow for a comparison to be made between results and objectives in order to decide whether the strategy is successful for each stakeholder.
The monitoring system is expected to have a greater impact if it is tied into the organization's incentive system. This helps to ensure that the participants are committed to the objectives described in the plan. Companies sometimes develop comprehensive plans, but then focus solely on the stockholders or the managers. The monitoring system should focus on the long-range impact of the plan on all of its stakeholders. For example, to recognize the interests of its customers, IBM uses consumer-satisfaction surveys to help determine management's compensation.
In the Edsel case, no monitoring procedure had been developed. Substantial confusion seemed to occur when the initial results were examined. What results constituted a failure? This had not been defined in advance. Some months after what seemed to be a disastrous introduction, Ford told its dealers that there was no cause for alarm (Brooks, 1969). Apparently, Ford was unable to respond rapidly to evidence that their strategy was failing.
The Edsel monitoring procedure, or lack of it, is apparently not unusual. Horovitz (1979), in a survey of the planning practices of 52 large firms in Great Britain, France, and West Germany, found that virtually none of them had a formal procedure for monitoring results of their long-range plans.
One way to improve the monitoring of results is to have an evaluation performed each year by an independent auditor. The following questions could be addressed: Is the monitoring system comprehensive? Is the planning process adequate? Is the forecasting process adequate? (A procedure for the auditing process for forecasting is provided in Armstrong, 1982a.)
Business plans and forecasts are frequently ignored; at other times they are used to rationalize a course of action previously decided. What can be done to develop commitment to the planning process? What can be done to ensure that the various stakeholders will cooperate and try to implement the chosen strategy? Attention should be given to commitment throughout each of the above steps in planning.
Formal planning calls for an explicit procedure for gaining commitment to the plan. A first condition is that key stakeholders should be evolved in the planning process. This would mean, at least, that information should be obtained from these stakeholders.
Publicly stated objectives are a requirement if the objectives are expected to have an impact on behavior. Each stakeholder group and each key decision maker should be aware of the objectives. This can help to achieve consensus.
Commitment to objectives is expected to be higher if those who are affected by the strategic decisions participate in the objective-setting process. In other words, self-set objectives are more likely to be attained than objectives set by others. This generalization is based on laboratory studies (for example, Bass, 1977) and on field studies. Participation is not necessary in all situations; however, it generally helps, and seldom does it make things worse.
Participation by stakeholders is also helpful in the generation and evaluation of alternative strategies (Van de Ven, 1980). This is most important where the strategy involves large changes, because the threat to the various stakeholders is reduced if they have some control over these changes.
Commitment can be maintained more effectively if the monitoring system provides quantitative feedback on success in meeting each objective. Key decision makers can then use this feedback to make tactical changes. Stakeholders can see how the strategy is meeting their objectives.
Rather than seeking commitment to the plan, top management sometimes uses planning as a way to gain control over others. They may use it to reduce the authority of subordinate managers and unilaterally to reduce the ability of these managers to act. This may help to explain why planning is more popular among top management. For example, in a survey done by Ang and Chua (1979), 80 percent of top management reported that they were “very favorable” toward long-range planning; 30 percent of the operating managers agreed. If plans are imposed on others, their impact might be detrimental. Operating management could feel less responsible for the success of strategic decisions. They might even feel threatened by the strategic decisions and attempt to reduce their effectiveness.
To avoid having the monitoring system used to control others, it is best to provide managers with information about how their group has performed, not the individuals within their group. Their subordinate managers, in turn, would receive information only about their group. Overall, then, sufficient feedback is received, but it is used to guide one's own actions as a manager. "How can I help my group to perform?" is the issue, not "How can I control the managers under me?"
SITUATIONS FAVORING THE USE OF STRATEGIC PLANNING
Consider a simple example: planning for lunch on a workday. Generally, you need not do much. When you are hungry, you may send for a sandwich, or you may walk to the nearest restaurant. What happens, however, when a group of managers from corporate headquarters comes for a meeting that includes lunch? You obtain input to decide what type of atmosphere and food would be appropriate. After deciding on a restaurant, you ask your secretary to make reservations and arrangements for transportation to and from the restaurant. Obviously, you will charge the luncheon to the company's expense account.
As implied by this example, some everyday activities can benefit from planning. In particular, planning is most helpful in situations that involve more complexity (a fancy meal),
change (a large increase in the number of people to be fed), uncertainty (what do our guests like? how will everybody get to the restaurant?), and an inefficient market (expense account).
Planning is also expected to be very useful for organizations facing major strategic decisions as these generally involve high task complexity, change, uncertainty, and inefficient markets. These characteristics are summarized below:
1. High complexity of the task means that there is a greater need for explicit plans to ensure that the various bits and pieces fit together. The production and marketing of an automobile, for example, is a complex task.
2. Large changes create a need for planning because organizations are designed to deal primarily with repetitive situations. The changes could come from the environment (an economic recession), from competitors (foreign competition in automobiles), or from the firm itself (a decision to introduce a new line of automobiles). For large changes, the standard bureaucratic responses would be less useful. Large changes call for planning rather than merely reacting.
3. Uncertainty can lead to a waste of resources. Organizations must be prepared to meet different environments. Planning can address “what if” questions so that the firm can develop ways to respond. As uncertainty increases, the need for planning also increases. Ford faced an uncertain economy when it introduced the Edsel.
4. Inefficient markets call for planning because the price system does not dictate the organization's actions. The organization has much flexibility in how it acts. Thus, planning is expected to be more relevant to government organizations, nonprofit organizations, regulated sectors, and protected industries. Ford, for example, received some protection from foreign imports. (Managers in competitive markets may feel that planning is more important as competition increases. This is because poor planning could lead to the failure of the company. However, failure is a natural event in competitive markets. An efficient market would inform stakeholders and would help to ensure that their needs are met, no matter what an individual company does. If they plan poorly, another company will replace them.)
Planning is expected to be most relevant when all four of these conditions hold. Ford faced this situation when introducing the Edsel. A more extreme example of a company that meets all four conditions would be a utility deciding whether to build an atomic reactor. It has a complex task, large changes are involved, uncertainty is high (for example, what if the law is changed so that the company must bear the Full costs of waste disposal?), and the market is inefficient (huge subsidies are paid by the government and the local community bears the costs of disasters).
One industry that has been moving toward the above four conditions is banking. According to Wood (1980), change and uncertainty have increased in this industry during the 1970s. During this period, the use of formal strategic planning increased from 6 percent of the banks prior to 1970 to 80 percent by the end of 1977.
Another example that met the above four conditions was Ford's introduction of the Edsel. The market inefficiencies in this case, however, were not large.
At the other extreme would be a company that meets none of the conditions. Here formal planning would be of little value. An example would be the normal operations for an existing middle-priced restaurant n New York City.
An investment in formal planning might be considered like an insurance policy: It might be needed. But in situations where the risk is small' the investment in insurance may not be necessary.
The above conditions are inferred from research in organizational behavior (see review in Armstrong, 1982b). Perhaps there are other conditions that are more important. A survey of the empirical field research on the value of strategic planning yielded twelve studies: Van de Ven (1980), Ansoff et al. (1970), Thune and House (1970), Herold (1972), Wood and LaForge (1979), Karger and Malik (1975), Harju (1981), Kudla (1980), Leontiades and Tezel (1980), Grinyer and Norburn (1975), Kallman and Shapiro (1978), and Fulmer and Rue (1974). A systematic analysis of results from these studies concluded that the evidence was consistent with the position that planning is useful for organizations (Armstrong, 1982b). But the studies provided little useful data on “how to plan” and on “when to plan” because few of them provided adequate information on the planning processes used or on the situations in which the planning was used.
THE FORECASTING METHODS
Forecasting methods, as defined here, are explicit procedures for translating information about the environment and the company's proposed strategy into statements about future results. What would be the results if the environment were favorable and we did A? What if it were unfavorable and we did A? What if it were unfavorable and we did B?
Before discussing how the forecasting methods can be used in strategic planning, a general description is provided here on the various methods that can be used in forecasting. A number of schemes exist for classifying forecasting methods (see, for example, Chisholm and Whitaker, 1971; Chambers, Mullick, and Smith, 1974). These schemes are based upon the type of data used, the type of people doing the forecasting, or the degree of sophistication of the methods used to analyze data. The scheme used below is based upon the methods used to analyze the data.
Research on methods for analyzing data has historically been organized along three continuums: subjective versus objective, naive versus causal, and linear versus classification methods. The discussion below considers the fictitious end points of each continuum.
Subjective versus Objective Methods
Subjective methods are those in which the processes used to analyze the data have not been well specified. These are also called implicit, informal, clinical, or intuitive methods. They may be based on simple or complex processes. They may use objective data or subjective data as inputs. Subjective methods may be supported by much formal analysis or by none. But the critical point is that the analyst makes the forecast in his or her head. For example, executives could be asked to make annual forecasts of automobile sales for the next five years. They would be provided with any information they request, but they would produce the final forecasts by thinking.
Objective methods are those that use well-specified processes to analyze the data. Ideally, they have been specified so well that other analysts can replicate them and obtain identical forecasts. These are also called explicit, statistical, or formal methods. They may be based on simple or complex processes. They may use objective data or subjective data as inputs. They may be supported by much formal analysis or none. But the critical factor is that the inputs are translated into forecasts using a process that can be replicated by other analysts. Furthermore, the forecasting process could be programmed on the computer. An example would be an econometric model to forecast industry automobile sales.
The choice between subjective and objective methods is an important one. Most forecasts are made using subjective methods (Rothe, 1978). It also seems that the more important the forecast, the greater is the likelihood that subjective methods will be used. (But the popularity of a method is a poor guide in determining which method is most useful.)
Naive versus Causal Methods
A continuum of causality exists in forecasting models. At the naive end, no statements are made about causality (automobile sales can be plotted against time and the trend can be projected); at the causal end, the model may include many factors (the real income per capita, the real price of gasoline, the real price of automobiles, the population, and the real price of substitute forms of transportation).
Causal methods are more complex than naive methods. First, data must be obtained on the causal factors. Estimates of causal relationships are obtained from these data. These estimates of the causal relationships should be adjusted so that they are relevant over the forecast horizon. Next, one must forecast the changes in the causal variables. Finally, the Forecasts of the causal variables and the relationships are used to calculate the overall forecast.
Causal methods are of more obvious value in planning. They can be used in any phase of planning. However, naive methods can be used in some phases. For example, naive methods can provide forecasts of environmental factors.
Linear versus Classification Methods
Methods that are objective and that rely upon causality can be categorized according to whether they use linear or classification methods. This decision generally has only a small impact on accuracy. It depends mostly upon convenience and the availability of data (classification methods typically require much data).
The linear method is based upon the usual way we think about causality: “If X goes up, this will cause Y to go up by so much.” An attempt is made to find linear relationships between X and Y. Linear methods are used because it is easier to work with models where the terms can be combined by using simple arithmetical operations. Thus, one might try to predict automobile sales by forecasting changes in income and price, and then multiplying by the relationships of these factors to auto sales.
The classification approach groups similar behavioral units. These groups or segments would be expected to respond in a similar fashion. For example, to forecast automobile sales, one segment might be “family size of two, age of head of household 65 to 75, low income, living in apartment in a large city, near mass transportation.” Another segment might be “family size of five, age of head of household 25 to 35, high income, living in house in a suburb, not near mass transportation.” The people within each segment would be expected to have similar behavior with respect to the purchase of automobiles, but the segments differ substantially from one another (low automobile purchases in the first group and high in the second group). To make a prediction using the classification method, forecasts would be made of the population of each segment and also of their behavior. These are then combined to get a forecast of auto sales for each segment. By summing across segments, an overall forecast is obtained (for example, total industry sales). The classification approach is most useful when the groups differ substantially from one another.
The Methodology Tree
The methodology tree (Figure 2-3) is used to summarize the above discussion on the choice of a forecasting method. The first decision to be made is whether it is most appropriate to use a subjective or objective method. The subjective branch leads to the “judgment” leaf. An extension of this is called “bootstrapping.” This involves the development of an objective method to replicate the judgmental forecasts. This can be done by asking the judges to specify the rules they used to make forecasts. Alternatively, one can statistically analyze the judgmental forecasts and the data used by the judges to infer what rules were used. The bootstrapping model can then be used to make the forecasts.
The objective branch offers a number of approaches to forecasting. One must decide whether it is most appropriate to use naive or causal methods. The naive branch leads to the “extrapolation method.”
Use of the causal branch requires an additional decision. Should you use linear or classification methods? The linear branch leads to “econometric methods” and the classification branch leads to “segmentation methods.”
The thickness of the branches of the methodology tree indicates which decisions are most important in the selection of a forecasting method. The “leaves” of the tree (boxes) can be used as a checklist for selecting 3r method.
The methods will be discussed in a somewhat more detailed fashion below. For a more in-depth description, many sources exist (e.g., Wood and Fildes, 1976; Wheelwright and Makridakis, 1980).
MATCHING THE FORECASTING METHOD TO THE SITUATION
Formal forecasting methods help to improve planning in two ways. First, they can increase accuracy over what would occur with informal methods and, thus, reduce uncertainty. Second, they can provide better estimates of the degree of uncertainty (risk).
Improved accuracy and better estimates of risk are needed for various Phases of the forecasting and planning processes. These needs are described below, starting with the environmental forecast.
Environmental forecasts are useful as an input to strategic planning. The identification of possible states of the environment and a forecast of their likelihood can provide ideas on what strategies should be considered by your company.
Environmental forecasts also can help to provide better industry forecasts (the total demand for a product class in a given market). Industry forecasts can be made for each of the possible states of the environment and also for various assumptions about the future behavior of the companies in this industry. Forecasts would be required for each of the company's major products and markets.
The company then can forecast what actions it will actually take. Ideally, the company's optimal strategy would be translated directly into actions. However, the actual strategy (actions) frequently departs from the proposed strategy due to communication problems, lack of interest, resistance to the strategy by those in the company, insufficient resources within the company, or a decision to abort the strategy because of environmental changes (such as a change in available technology).
Actions by the company are also influenced by the actions of its competitors. Thus, it is helpful to forecast how the competitors will react to environmental changes. For industries that are not highly competitive (that is, for most situations), you should also try to forecast how competitors will react to major strategy changes by your company.
Forecasts of the actions (and reactions) by a company and its competitors can help to forecast the company's market share. Sales forecasts can then be calculated by multiplying the forecast market share times the industry forecast. This should be done for each major product market.
Costs should also be forecast. These depend primarily upon environmental changes, the actions taken by your company, and the level of your company's sales.
The company is then in a position to forecast results. The sales and cost forecasts allow for a forecast of profits. It would also be possible to examine the forecast costs and benefits to each of the company's stakeholders.
This list of the forecasting needs in company planning is summarized in Figure 2-4. The exhibit elaborates on the “Forecasts” box of Figure 2-1. It starts with environmental forecasts and then proceeds downward through the other areas until forecasts are obtained from results in the organization. The larger arrowhead indicates the preferred sequence. The smaller arrowhead indicates that some backtracking will probably be needed. Below, a discussion is provided on techniques that can be used for each of the areas listed in Figure 2-4. In some cases, existing knowledge on the most useful techniques is scarce. Thus, an open-minded use of the methodology tree (Figure 2-3) is advised.
Environmental forecasts are needed to help the company formulate its strategy. It is important that the forecasting methods first identify the possible states of the future. For this,
brainstorming among a variety of experts would be useful. Particular attention would be given to the more important of these possible states. Importance should be judged not only by the likelihood of the environmental change, but also by its potential impact on the company if it does occur. The use of structured judgmental methods provides an obvious starting point to assess the likelihood of the events. However, extrapolation from analogous events in history can also be useful.
An environmental change can directly affect company actions (e.g., a change in export laws), or it can indirectly affect the company by its impact on the industry (e.g., increased energy costs).
Surprisingly, the accuracy of industry forecasts is not highly sensitive to the accuracy of the environmental forecast (evidence on this. point is summarized in Armstrong, 1978a, pp. 219,
241, 378). It is expected that this generalization will not hold for extremely large environmental changes such as wars, depressions, shortages, government controls, or major technological innovations. But, generally, highly accurate environmental forecasts are not required for industry forecasts.
It is important to determine which are the important factors in the environment that might affect the industry. It is also important to predict the direction of change in the important factors, and to then get “approximately correct” predictions of the magnitude of the changes in these factors. For the direction of change in environmental factors, only general trends, not cycles, should be considered. Other than recurrent events owing to the seasons of the year (seasonality), cycles have been of little value for improving the accuracy of forecasts. The reason? One must also predict the phases (timing) of the cycles. If the timing is off, large errors can occur.
Ample data exist on trends in the environment. The more important factors are published in magazines, newspapers, and financial newsletters. The problem is not a lack of data; rather, it is how to use the data. Companies often spend much time and money seeking information from the environment that will confirm their beliefs. Frequently they ignore negative or "disconfirming" information that is easily available. It seems useful, therefore, to severely limit the budget for the collection of environmental data. Seldom is the additional information expected to have a strong positive impact on decision making. (Most of the evidence in this area is from studies in psychology; Goldberg, 1968, provides a summary of this research.)
This advice on environmental forecasts is counterintuitive. People typically expect that better environmental forecasts are of great value to the company. Thus, much time and money are spent by firms to obtain “better forecasts.” For example, many companies purchase econometric forecasts to obtain short-range forecasts of GNP, inflation rates, and unemployment. This practice is widespread despite the fact that little evidence exists to suggest that these forecasts are superior to other, cheaper alternatives such as extrapolations or forecasts by a panel of experts (Armstrong, 1978b).
After preparing the environmental forecasts, the company should prepare industry forecasts. In some cases, these have been prepared by others. For example, Predicasts, Inc. of Cleveland, Ohio, summarizes the U.S. forecasts in a quarterly publication called Predicasts. They publish forecasts for other countries in their Worldcasts. The disadvantages of using forecasts prepared by others are that:
1. They may not use the product-market definitions that are relevant to your company
2. A time lag exists from the time the forecast was made until it was published 3. Forecasts are not updated frequently 4. The original sources (as cited in Predicasts and Worldcasts) often do not provide
sufficient information on the assumptions behind the forecasts
For these reasons, medium-sized and large firms are best advised to develop their own industry forecasting models.
Much of the error in industry forecasts is due to errors in estimating the current status. What are the industry sales now? Thus, some useful though often ignored advice is to break the forecasting problem into two subproblems. First, estimate the current sales level. Then use methods to forecast change over the forecast horizon. The forecast is the sum of the current sales plus the change in sales over the forecast horizon.
Judgmental methods are often appropriate for estimating current sales. Experts, such as sales people, are likely to have up-to-date information on the current sales. In contrast, objective data are often reported after much delay.
To obtain judgmental estimates of current sales, use structured methods such as the following: First, provide those concerned with up-to-date information in an easy-to-read format (such as tables or graphs). Then, replace the group meeting with a survey. After each person makes his or her best estimate, an average of their forecasts is calculated. A refinement, helpful when there may be ambiguities in the question, is the Delphi technique, described earlier in this chapter.
Although judgment might be useful for estimating current sales, it is not so relevant in forecasting change. For this task, objective methods are more appropriate (see Armstrong 1978a, pp. 363-372 for a summary of the evidence leading to this conclusion).
If experts are used to forecast change, there is no need to obtain the “best” experts (Armstrong, 1980). According to the research, sufficient expertise in the area of interest can be obtained in a few months. Thus, it is advisable to obtain inexpensive experts.
Of the objective methods available to forecast change, econometric methods are perhaps the most useful. The econometric model should aim at two desirable, but conflicting, goals in industry forecasting: (1) include all important factors, and (2) keep it simple. Research in this area suggests that little complexity is needed. Often, “near optimal” results have been obtained with the use of only two or three variables.
The magnitude of the causal relationships in an industry-forecasting model can be estimated judgmentally. However, for most situations it is safer to obtain estimates from historical data or from experiments. Regression analysis provides a common and useful way of estimating these relationships. In some cases, these estimates can be obtained from published studies using regression analyses of similar products. Surprisingly, accurate estimates of regression coefficients are often not necessary (Dawes 1979), although exceptions to this generalization do occur (Remus and Jenicke 1978). A reasonable approach, recommended in econometrics, is to start with judgmental (à priori) estimates of causal relationships, then update these by use of regression analysis.
Another objective method relevant to long-range market forecasting is the segmentation model. This approach is expected to be accurate, but it requires much data. Furthermore, it is difficult to use when examining changes in the company's strategies or in competitive responses.
The resulting industry forecasts can be used as an input to the planning process. For example, different strategies might be required depending on whether the organization is in a growing or declining industry.
Forecasts for the environment and the industry can then be considered along with the company's proposed strategies to predict what actions would actually be taken. In other words, what will the strategy look like in practice? Forecasting is aided if the company considers well- defined and operational strategies, if the people in the company will be firmly committed to implementing the strategy, and if the company will have adequate resources.
But will all of the stakeholders be successful at implementing the strategy? One way to forecast the actual actions by the company is to survey these stakeholders. They would be presented with a description of a strategy and would be asked how successful they would be in carrying out their part. Perhaps a given strategy is not realistic from their viewpoint.
In some situations, such as when negative effects would be encountered by the company, it may be difficult to forecast stakeholder actions by direct questioning. Here, “group depth interviews” may be useful. To do this, groups of key decision makers in the organization meet with a consultant. The consultant presents scenarios with different strategies and environments. The decision makers are then asked how they would act in these situations. The reason for meeting in a group is because the decisions of these people are interdependent. A similar procedure could be followed with the stakeholders.
Forecasts should also be made of the company's resources. Will financial resources, supplies of raw materials, and personnel be adequate for a given strategy? Forecasts of labor- management relations might be important at this stage.
A company’s strategy is often dependent on the actions of its competitors. But the competitors are unlikely to tell you about their intentions. In many cases, it may be sufficient to forecast the competitors' actions using expert judgment. In doing this, it may be helpful to consider a forecast of the competitors’ resources.
If a substantial amount of historical data exists, it may be possible to find analogous situations. Summaries of how competitors reacted in past situations may allow for a forecast of how they will respond to changes in the environment or to different strategies that your company might use. For example, how do your major competitors react to new product introductions?
One technique that is useful in forecasting competitors’ actions is role-playing. This involves having some members of your management team act as if they were in their current role while others play the role of competitors. Role-playing can be used to test various strategies. It is
especially useful for analyzing unusual strategies when secrecy is important (such as with new products). It is also useful when it would be impractical to test out a strategy with a field experiment.
Role-playing is relatively easy to carry out. The following rules are suggested:
1. Assign people to roles of key decision makers for the company and its major competitors. Provide a short description of the role, the environment, the firm's capabilities, and the selected strategy. Use four pages, one for each of these four topics.
2. The role players should not step out of their roles; that is, once they meet, they should “be” that person at all times. Ask the role players to prepare individually and then return to the meeting place when they are ready to stay with their roles.
3. The players should improvise as needed. 4. The players should act as they themselves would act in that role, asking
themselves, “What would I do, given this strategy?”
Role-playing has been popular in political science and in the military, where it is called gaming. The use of role-playing in business has been limited. Busch (1961) said that the Lockheed Corporation used role-playing to forecast the behavior of their customers. IBM used a form of role-playing to forecast the reactions of a jury in a trial (The Wall Street Journal, February 3, 1977). Armstrong (1977) used it to forecast the actions by members of the board of directors of the Upjohn Corporation in a case in which the government tried to force Upjohn to remove one of its drugs from the market.
Forecasts from role-playing may differ greatly from those provided by other methods. Four studies have contrasted the accuracy of role-playing with judgmental forecasts. In three studies, role-playing was superior, and there was no difference in the fourth. Armstrong (1978a, pp. 118-121) summarized this evidence.
In the Edsel case, it would have been useful to identify the key competitors of the Edsel and to conduct role-playing to predict their reactions. For example, one of the key competitors was Ford's Mercury division. Role-playing might have predicted what happened. (What actually happened, according to Brooks, 1969, was that Mercury launched a large advertising campaign in retaliation against the introduction of the Edsel. Furthermore, Mercury production workers apparently sabotaged the Edsel cars that were being produced in the same plants.)
Given the forecast actions by the company and by its competitors, what market share can the company expect? Research in this area has provided few generalizations on which methods are most effective. However, a number of techniques seem reasonable.
For small changes in strategy, it may be sufficient to extrapolate the company's market share. Alternatively, you might employ the judgment of a group of experts using structured
methods to obtain these forecasts. In most situations it is better to first obtain forecasts independently with both extrapolation and judgment methods, and then use the average of these forecasts.
For large changes in strategy, the use of econometric methods is desirable. This assumes that one has data on the dependent variable (for example, sales) and on the key aspects of the strategy. If significantly different strategies were used in the past (such as the use of different prices) and if this led to substantial differences in sales, the econometric model may be useful in identifying the effects due to changes in the strategy.
In some cases (such as with new products), data are not available for the sales variables. Here the use of bootstrapping can be considered. The bootstrapping model is developed from management's judgment and from the data they use. Typically, the model is estimated by regression analysis. An example of such a model would be:
M = c + bpP + baA
where M is management's forecasted market share c is a constant
P is the product's price relative to its competition A is the product's advertising relative to its competition bp and ba are coefficients reflecting the relationships used by management
The bootstrapping model offers some advantages. First, it applies management's beliefs in a consistent manner. Thus, it can evaluate a large set of alternative strategies in a consistent way. Second, bootstrapping can provide insight to management's current forecasting beliefs, and this may foster learning. Finally, bootstrapping is slightly more accurate than the judges themselves (evidence is summarized in Armstrong, 1978a, pp. 251-259, and in Camerer, 1981). Thus the name: It is like lifting oneself up by the straps on one’s boots. An application of this model for new product forecasting is described by Montgomery (1975).
The company is now in a position to forecast its costs. An explicit forecast should be made of the costs of a given strategy to each of the important stakeholders. For example, in the Edsel case, how much would Ford’s strategy cost the employees, the dealers, the customers, the local community, and the stockholders? The formal approach is expected to produce better forecasts of the costs to each member of the system than would be obtained by informal forecasting methods.
Forecasts of costs can be used in the evaluation of the strategies (thus the two-way arrow from “Costs” to “Company Actions” in Figure 2-4). Where will the company's future costs be low relative to its competitors?
The cost forecasts depend upon the environment, the actions taken by the company, and its sales. Thus, forecasts of these other areas should be made before making the cost forecasts.
The best methods for forecasting costs will depend upon the industry he strategy, and stakeholders. It is difficult to provide generalizations on what method is best in which situation. A logical starting point, however, is the use of extrapolations. For example, learning curves can be used to forecast decreases in manufacturing costs. The simplest way to do this is to estimate a straight line on log paper to reflect what is typically a constant percentage decrease in manufacturing costs as volume increases. Experience curves attempt the same thing using the decrease in total costs as volume rises. But large changes in technology or in supplies can lead to significant departures from this extrapolation. In view of the potential for errors, it seems reasonable to base the forecast on two or more different methods.
Completion of the market share and industry forecasts allows for a calculation of the sales forecast. If uncertainty has been estimated for both the industry and market share forecasts, these can be used to estimate the uncertainty in the sales forecast.
By using forecasts from the preceding steps, calculations can then be made of the forecasted costs and benefits of each strategy for each of the major stakeholders. For example, profit and loss statements could be prepared for the stockholders, environmental impacts could be summarized for the local community, and the effects of product usage could be summarized from the consumer's point of view.
The forecasts should examine not only the expected results for a given strategy, but also the uncertainty. What are the most favorable and least favorable results that might occur? Furthermore, they should describe how the strategy performs in different environments. These forecasts of results provide the basis for selecting among the various strategies. How do they each perform against the original objectives? If none of the strategies is acceptable, it is useful to go back to the planning process for the generation of additional strategies.
A framework was presented to show the relationship between strategic planning and forecasting (Figure 2-1). To make forecasts, it is necessary to consider the strategies; to plan, it is helpful to have good forecasts of the environment and of the impact of various strategies.
A description of the planning process was then provided (see Figure 2-2). This stressed the need for commitment-seeking in all steps of the planning process. Some of the major guidelines for planning were:
1. For objective setting, start with the ultimate objectives for each stakeholder. Translate these into specific, challenging, and measurable objectives by considering the comparative strengths and weaknesses of the company.
2. Develop alternative strategies for each of the more important possible states of the environment. Particular attention should be given to unfavorable situations; scenarios and brainstorming can be used here to increase creativity and openness within the organization. The plans should be comprehensive and they should contain slack. The more promising strategies should be specified in operational terms. Contingency plans should also be prepared for alternative environments.
3. Evaluate alternative strategies explicitly. The Delphi technique can be used here as well as group depth interviews, scenarios, rating sheets, and the devil's advocate.
4. Establish a monitoring system to obtain information on: a. Environmental changes
b. Changes in the company's capabilities and in the capabilities of its competitors
c. Actions taken by the organization d. Actions taken by major competitors e. Results
This information should be compared with predetermined standards to indicate when the strategy should be reexamined. Furthermore, fixed review periods should be scheduled.
Participation of the company's stakeholders in each of the above four steps should help to increase the commitment of these stakeholders to making the strategy a success.
This advice on formal strategic planning is summarized in the checklist of Figure 2-5. Although it may help to use some of the planning techniques separately, it seems best to use them in combination with one another.
Formal planning is expected to be most useful in situations where:
1. The task is complex 2. Large changes occur (in the environment, by competitors, or by the company
itself) 3. Uncertainty is high 4. The market is inefficient
When one or more of these conditions do not hold, an investment in formal planning would be expected to yield a smaller return.
Various forecasting methods were described. They were classified using the forecasting methodology tree (see Figure 2-3). This described he methods as subjective or objective, naive or causal, and linear or classification. The tree can be used as a checklist in selecting a method or a given forecasting problem.
Suggestions were made for the use of forecasting methods for the various needs in the company's planning (see Figure 2-4). This included methods for forecasting the environment, industry, company actions, competitors' actions, market share, and costs. These forecasts allow one to calculate sales forecasts and then to examine the costs and benefits for each stakeholder.
Some of the more important suggestions were:
1. Use structured judgmental methods to forecast the possible environmental states, their likelihood, and their potential effect on the company and the industry. As an input to the industry forecast, concentrate on identifying the important factors and their direction of change. Generally, only approximately correct predictions are needed for the magnitude of change. Do not forecast long-range cycles.
2. To obtain industry forecasts, first estimate current sales, and then forecast changes. Judgmental forecasts are often appropriate for estimating current sales. Simple econometric and segmentation methods are useful to forecast changes.
3. Organizations do not always have operational strategies. When they do have such strategies, they do not always follow them. To forecast what actions the company
will actually take, use surveys or group depth interviews with the key decision makers and stakeholders. Group depth interviews seem especially useful for dealing with situations that could be unfavorable to the organization.
4. Competitive reactions to large changes in the environment or to changes in your company's strategy can be forecast by historical analogies or by role-playing.
5. Market share can be forecast by extrapolation or by judgment. If large changes in strategy are considered, econometric models are appropriate. However, when data are lacking on actual results, bootstrapping can be used.
6. Forecast the costs for each stakeholder by using forecasts of the environment, company actions, and sales levels. Try to obtain forecasts using different methods in order to compensate for errors inherent in single method.
This chapter has summarized the current state of the art in formal planning and forecasting. The evidence to date suggests that formal planning and forecasting are valuable for organizations.
Ackoff, Russell L.: A Concept of Corporate Planning, John Wiley, New York, 1970. Ang, James S., and Jess H. Chua: “Long-Range Planning in Large United States Corporations—
A Survey,” Long-Range Planning, vol. 12, 1979, pp. 99-102.
Ansoff, H. Igor: Corporate Strategy. McGraw-Hill, New York, 1965.
Ansoff, H. Igor et al.: “Does Planning Pay? The Effect of Planning on Success of Acquisitions in American Firms,” Long-Range Planning, vol. 3, 1970, pp. 2-7.
Armstrong, J. Scott: “Social Irresponsibility in Management,” Journal of Business Research, vol. 5, 1977, pp. 185-213.
Armstrong, J. Scott: Long-Range Forecasting: From Crystal Ball to Computer, John Wiley, New York, 1978a.
Armstrong, J. Scott: “Forecasting with Econometric Methods: Folklore versus Fact,” Journal of Business, vol. 51, 1978b, pp. 549-564.
Armstrong, J. Scott: “The Seer-Sucker Theory: The Value of Experts in Forecasting,” Technology Review, vol. 83,June-July, 1980, pp. 18-24.
Armstrong, J. Scott: “The Forecasting Audit,” in Spyros Makridakis and Steven C. Wheelwright (eds.), The Handbook of Forecasting: A Manager's Guide, John Wiley, New York, 1982a.
Armstrong, J. Scott: “The Value of Formal Planning for Strategic Decisions: Review of Empirical Research,” Strategic Management Journal 1982b.
Baker, Henry G.: “Sales and Marketing Planning of the Edsel,” in Marketing's Role in Scientific Management, American Marketing Association, Chicago, 1957, pp. 128-144.
Bass, Bernard M.: “Utility of Managerial Self-Planning on a Simulated Production Task with Replications in Twelve Countries,” Journal of Applied Psychology, vol. 62, 1977, pp. 506- 509.
Becker, Lawrence J.: “Joint Effect of Feedback and Goal Setting on Performance: A Field Study of Residential Energy Conservation,” Journal of Applied Psychology, vol. 63, 1978, pp. 428-433
Brooks, John N.: Business Adventures, Weybright and Talley, New York, 1969. Busch, G. A.: “Prudent-Manager Forecasting,” Harvard Business Review, vol. 39, 1961, pp. 57-
Camerer, Colin: “General Conditions for the Success of Bootstrapping Models,” Organizational Behavior and Human Performance, vol. 27, 1981, pp. 411-422.
Chambers, John C., S. Mullick, and D. D. Smith: An Executive's Guide to Forecasting. John Wiley, New York, 1974.
Chisholm, Roger K., and Gilbert R. Whitaker,Jr.: Forecasting Methods, Irwin, Homewood, III., 1971.
Cosier, Richard A.: “The Effects of Three Potential Aids for Making Strategic Decisions on Prediction Accuracy,” Organizational Behavior and Human Performance, vol. 22, 1978, pp. 295-306.
Dawes, Robyn M.: “The Robust Beauty of Improper Linear Models in Decision Making,” American Psychologist, vol. 34, 1979, pp. 571-582.
Fulmer, Robert M., and Leslie W. Rue: “The Practice and Profitability of Long-Range Planning,” Managerial Planning, vol. 22, May-June, 1974, pp. 1-7.
Goldberg, Lewis R.: “Simple Models or Simple Processes? Some Research on Clinical Judgments,” American Psychologist, vol. 23, 1968, pp. 483- 496.
Griffith, J. R., and B. T. Wellman: “Forecasting Bed Needs and Recommending Facilities Plans for Community Hospitals,” Medical Care, vol. 17, 1979, pp. 293-303.
Grinyer, P. H., and D. Norburn: “Planning for Existing Markets: Perceptions of Executives and Financial Performance,” Journal of the Royal Statistical Society (A), vol. 138, 1975, pp. 70-97.
Harju, Paavo: Attitude of Strategic Managers toward Formalized Corporate Planning, School of Economics, Turku, Finland, 1981.
Herold, David M.: “Long-Range Planning and Organizational Performance: A Cross Validation Study,” Academy of Management Journal, vol. 15, 1972, pp. 91-102.
Hofer, Charles W.: “Research on Strategic Planning: A Survey of Past Studies and Suggestions for Future Efforts,” Journal of Economics and Business, vol. 28, 1976, pp. 261-286.
Hogarth, Robin M., and Spyros Makridakis: “Forecasting and Planning: An Evaluation,” Management Science, vol. 27, 1981, pp. 115-138.
Horovitz, J. H.: “Strategic Control: A New Task for Top Management,” Long-Range Planning, vol. 12, 1979, pp. 2-7.
Ilgen, Daniel R., C. D. Fisher, and M. S. Taylor: “Consequences of Individual Feedback on Behavior in Organizations,” Journal of Applied Psychology, vol. 64, 1979, pp. 349-371.
Janis, Irving L.: “Groupthink,” Psychology Today, November 1971, pp. 43-77.
Kallman, Ernest J., and H. Jack Shapiro: “The Motor Freight Industry—A Case Against Planning,” Long-Range Planning, vol. 11, February 1978, pp. 81-86.
Karger, Delmar W. and Zafar A. Malik: “Long-Range Planning and Organizational Performance,” Long-Range Planning, vol. 8, 1975, pp. 60-64.
Kudla, Ronald J.: “The Effects of Strategic Planning on Common Stock Returns,” Academy of Management Journal, vol. 23, 1980, pp. 5-20.
Latham, Gary P., and Sydney B. Kinne: “Improving Job Performance through Training in Goal Setting,” Journal of Applied Psychology, vol. 59, 1974, pp. 187-191.
Latham, Gary P., and G. A. Yukl: “A Review of Research on the Application of Goal Setting in Organizations,” Academy of Management Journal, vol. 18, 1975, pp. 824-845.
Leontiades, Milton, and Ahmet Tezel: “Planning Perceptions and Planning Results,” Strategic Management Journal, vol. 1, 1980, pp. 65-75.
Linstone, Harold: The Delphi Method: Techniques and Applications, Addison-Wesley, Reading, Mass., 1975.
Locke, Edwin A., K. N. Shaw, L. M. Saari, and G. P. Latham: “Goal Setting and Task Performance: 1969 - 1980,” Psychological Bulletin, vol. 90, 1981, pp. 125 - 152.
Maier, Norman R. F.: Problem-Solving Discussions and Conferences, McGraw-Hill, New York, Montgomery, David B.: “New Product Distribution: An Analysis of Supermarket Buyer
Decisions,” Journal of Marketing Research, vol. 12 1975, pp. 255-264. Remus, William E., and Lawrence O. Jenicke: “Unit and Random Linear Models in Decision
Making,” Multivariate Behavioral Research, vol. 13, 1978, pp. 215-221. Rothe, James T.: “Effectiveness of Sales Forecasting Methods,” Industrial Marketing
Management, vol. 7, 1978, pp. 114-118.
Steiner, George A.: Strategic Planning, Free Press, New York, 1979.
Thune, Stanley S., and Robert J. House: “Where Long-Range Planning Pays Off,” Business Horizons, vol. 13, 1970, pp. 81 - 87.
Tolchinsky, Paul D., and Donald C. King: “Do Goals Mediate the Effects of Incentives on Performance?,” Academy of Management Review, vol. 5, 1980, pp. 455-467.
Van de Ven, Andrew H.: “Problem Solving, Planning, and Innovation. Part 1: Test of the Program Planning Model,” Human Relations, vol. 33, 1980 pp. 711-740.
Wheelwright, Steven C. and S. Makridakis: Forecasting Methods for Management, 3d ea., John Wiley, New York, 1980.
Wood, D. Robley, Jr.: “Long Range Planning in Large United States Banks,” Long-Range Planning, vol. 13, 1980, pp. 91-98.
Wood, D. Robley, Jr. and R. L. LaForge: “The Impact of Comprehensive Planning on Financial Performance,” Academy of Management Journal, vol. 22, 1979, pp. 516-526.
Wood, Douglas and Robert Fildes: Forecasting for Business—Methods and Applications, Longman, London, 1976.
Wrapp, H. Edward, “Good Managers Don't Make Policy Decisions,” Harvard Business Review, vol. 45, 1967, pp. 91-99.
Published in Group & Organization Studies, Vol. 7, No. 4 (1982) 457-475. Strategies for Implementing Change:
An attitude survey and a role-playing case were used to identify the typical approaches people use to implement important changes in organizations. This typical strategy, suggested or used by over 90% of the subjects, was not successful in producing change in any of the fourteen role-playing trials. However, with ten minutes of instruction in the ”Delta Technique,” 86% of the subjects were successful in introducing change in another fourteen role-playing trials. The ”Delta Technique” consists of simple rules drawn from half a century of research.
An Experiential Approach
J. Scott Armstrong
Partial funding of this project was provided by IMEDE, Lausanne, Switzerland
Previous research has provided useful guidelines on how to implement change successfully. Despite the published research, popularized restatements, and education courses that provide guidelines for change, managers typically use intuitive strategies to introduce change. These intuitive strategies are seldom effective in situations calling for important behavioral changes.
It appears that one more study on how to implement change would have little effect on the situation. This is consistent with research in social psychology such as that by Nisbett and Borgida (1975). They found that people were unwilling to generalize from research on others. For example, evidence of what worked for over 95 of 100 people was seldom used by individuals to change their own behavior. But, they were often willing to generalize from their own experience to say how they would act in the future and to predict how others would act.
Instead of providing one more study, then, this article focuses on an experiential approach to strategies for change. It begins with a short exercise for the reader. (This approach could be easily extended to a classroom situation). Following this, two strategies are discussed. The first is the one most commonly used; the second, called the Delta Technique, is developed from a substantial body of published research.
One aspect of the experiential approach is that you are asked to solve a problem before you are given n technique for solving it. Thus, I will ask you to solve the COMPU-HEART case before I present the Delta Technique.
Another aspect of the experiential approach relates to your orientation as a reader. It is suggested that the Delta Technique is a proven method – for others. But it may differ from your current approach to change. How would you react to this information? There are other possibilities, but check the item closest to your expected reaction:
(A) I will disagree with the Delta Technique. This article will not provide enough information to convince me.
(B) I will decide that it was really a communication problem – that I already use the Delta Technique but have different words for it.
(C) I will feel that I had read an exciting article and agree that the Delta Technique is excellent. (D) I will take action by experimenting with the Delta Technique.
Alternative D, to experiment, creates stress. Yet D has the most value to you and your organization. Those selecting A will feel bad, Bs will feel O.K., and Cs will feel good: But D provides an opportunity to benefit from this article. (If you would like to compare your responses with others, a sample of MBAs from two organizational behavior courses at Wharton answered this question before reading this article. Of the 23 respondents, 65% selected alternative D (to experiment), 18% selected A (to disagree), 13% selected C (exciting article), and one person selected B (communication problem).
To examine your own approach to change, consider the COMPU-HEART Case.
THE COMPU-HEART CASEAssume you are on the administrative staff at Rosemont Hospital. What strategy would you use to implement change in this problem?
To provide space for its outpatient facilities, Rosemont Hospital had examined its inpatient program. When considering the whole system they concluded that there were clear-cut benefits in reducing the space allotted to patients with heart problems. In fact, they were excited about a program called COMPU-HEART, which provides home treatment for heart patients. This could be done by paramedics who visit the patient, perform some simple tests, and ask questions of the patient. The paramedics then call a time-sharing computer system that suggests what steps should be taken (e.g., ”do nothing,” ”send patient to hospital,” ”contact personal physician,” ”eliminate drug X,” and so on). This program was developed by modeling the decisions used by a committee composed of ten of the best heart specialists in the world. The computer is already used by Rosemont for other purposes, so the startup costs would not be large.
An experimental study run at a hospital in Los Angeles found that use of COMPU-HEART resulted in:
(1) fewer fatalities, (2) faster recovery rate for patients, (3) more satisfaction among patients, and (4) much lower costs.
However, you realize that when previous suggestions of a similar nature had been made, the doctors were skeptical. So, even though the doctors are currently overworked and this proposal would save time for them, you expect them to resist this change. The doctors prefer to have the patients in the hospital where they can see them and talk with them. They do not trust paramedics or computers to handle their patients for them.
You have scheduled a meeting with the doctors’ decision making committee. You would like to convince them to accept the COMPU-HEART program.
Before going on, you will find it helpful to take at least ten minutes to write your procedure for introducing change in the COMPU-HEART Case.
I presented the above written description of the COMPU-HEART case to a class of seventeen seniors in a Wharton course in September 1981. These students, who had all taken courses in organizational behavior, were given fifteen minutes to write a strategy that they would use to convince the doctors to adopt the COMPU-HEART system. They were asked to provide enough detail so that someone else could follow their strategy. They were also told that a strategy would be selected from someone in the group to use in a role-playing trial.
I coded the strategies. (Copies of these strategies can be obtained from the author. However, this task can be easily replicated.) Two months after the initial administration, I returned the strategies to the participants and asked them to code them. Sixteen of their replies were capable of being coded. Both codings produced nearly identical results, as shown in Table 1. Only one of the participants used the Delta Technique.
Rational Mixed Delta
Table 1. Coding of Strategy
11 12 4 3 1 1
I also presented the COMPU-HEART Case as a role-playing exercise. It was administered as a teaching vehicle on fifteen occasions from May 1978 to June 1980 in a two-day seminar on health care marketing. Most participants were high-level executives in hospitals and other health care institutions; a few came from companies that supply goods to health care institutions.
Two volunteers from each group acted as staff members. An additional two volunteers played the roles of the doctors. The role players were given about ten minutes to prepare, using only the information provided above. The role playing itself required about ten minutes to reach an outcome. In fourteen of the fifteen trials, the ”staff’ used primarily rational arguments. This strategy was never successful in gaining agreement to change by the doctors. The only staff group that was successful in introducing change did not use the rational approach; instead, their strategy was similar to the Delta Technique.
Two additional members in each group had volunteered to act as staff members in case the first staff group was unsuccessful. These staff members were given instructions for the Delta Technique (see Appendix) and were asked to follow them when meeting with the doctors. This second role play took place prior to discussion of the first role play. Staff members in two of the fourteen groups had difficulty applying the Delta strategy and met with no success. All other groups achieved success by gaining agreement in principle for an experiment and by scheduling another meeting. Despite the short time available for this role play (generally less than ten minutes), most groups achieved success as judged by the instructor and the class. This increase in effectiveness from 7% success without training in Delta, to 86% with training, was dramatic and statistically significant (p < .001 using X2 test). Over the past ten years, I have used two other versions of this case; one dealing with a change in a hiring policy and the other dealing with the adoption of a new forecasting method. Often the role players were given a few days to prepare. The role playing has been conducted with undergraduates and MBA students as well as executives. The results were similar to those in the COMPU-HEART Case.
Delta also worked in real life. COMPU-HEART is a disguised version of a study published by Marrow and French (1945). During World War II, the Harwood Manufacturing Company, a manufacturer of pajamas, was having difficulty in hiring workers. ”Older women” were available, but management believed that they were not good workers. A number of empirical studies indicated that, in fact, they were excellent workers.
Harwood used an approach similar to Delta. Top management was in control of the change process and they conducted an experiment to test their beliefs. This experiment provided disconfirming evidence for their current practice and they changed: They decided to hire older women. But before making a final decision, they used the Delta Technique with the other key stakeholders in their organization.
From a research viewpoint, there were limitations. Both the survey and the role playing were administered by me, and I have become an advocate of the Delta Technique. Furthermore, the judgments of the outcomes were made by me. On the positive side, the differences in outcomes between the rational approach and the Delta Technique were so large that there was little need for precise measurement. Most important, however, is the fact that this result can be easily replicated by the reader, using the information provided here.
Below, I discuss the rational approach and why it fails to produce important behavior change. Then I discuss the Delta Technique in more detail.
THE RATIONAL APPROACHMosteller (1981) provides a brief history on the rational approach to an important problem. At one time, scurvy would contribute to the deaths of perhaps half of the sailors on long voyages. In 1601, an Englishman, James Lancaster, conducted an experiment and found that lemons were effective in preventing scurvy. This was the first in a series of rational steps. But the British Navy did not adopt the use of lemons. About 150 years later, James Lind conducted an experiment that showed citrus fruits to be effective in preventing scurvy. The British Navy did not adopt citrus fruits until 1795, nearly 50 years later. The British Board of Trade followed 70 years later, in 1865. In other words, given rational and convincing evidence on the life-saving capability of citrus fruits, it required 264 years for the findings to be adopted by the British merchant marine.
Consider the rational approach to current problems. What if I presented evidence to managers of large corporations showing that personal interviewing of potential management employees is a poor investment because it is expensive, reduces the quality of the people hired, and introduces prejudice into hiring? (I believe research supports all of these statements.) How would the managers involved react to such information?
To examine the rational approach, Batson (1975) asked members of a church youth group to indicate on a questionnaire how strongly they believed that Jesus Christ was God. They were then told that documents had recently been discovered by archaeologists. These documents were said to be correspondence that took place among the apostles after Christ died and the theme was similar to a Watergate cover up:
“Now that Christ has died, people will realize that he is not God, so all is lost. ”
“Perhaps not. People want to believe that Christ is God, so maybe we can continue with this deception.”
The members of this youth group were told that The New York Times, under pressure from the World Council of Churches, was withholding the story. The group was led to think that the reactions of various people were being studied to see what would happen if and when The New York Times released the story. After learning about the documents, the youth group members completed a second questionnaire that again asked how strongly they believed that Christ was God.
Here is what happened. Nonbelievers lessened their belief that Christ was God, but those with a strong belief that Christ was God reported a strengthening of their belief. This strengthening occurred only for those believers who thought the information was authentic.
How can we explain Batson’s results? Certainly they are not based on ”common sense.” I presented a description of this study, on a self-administered questionnaire, to sixteen Wharton School seniors, none of whom had previously heard of the Batson study. None of the sixteen respondents was able to predict the complete results of Batson’s studies. Furthermore, only three of the sixteen were correct in their prediction of the direction of attitude change for the group of primary concern, the youth group members who believed that Christ was God and who also believed the information to be authentic.
Similar results were obtained from a group of executives attending a strategic planning course at INSEAD, in Fontainebleau, France, in April 1981. Only three of fourteen respondents correctly predicted the increased belief among the believers who thought they had received authentic evidence. Batson had designed his study as a further test of the theory of cognitive dissonance. This theory is widely known and is routinely included in programs on management education. Another way to view it is in terms of balance theory (see Brown, 1965, for a description). Balance theory can be used to examine the key elements of this study from the viewpoint of the changes, that is, the youth group member who believed Christ is God. These elements are the youth group member, the researchers, and the message about Christ. They are illustrated in Figure 1, where the signs represent evaluations. Thus, the minus
sign indicates that the youth group member does not like the message that ”Christ is not God.” According to balance theory, when you multiply the signs, the product should be positive. If not, the situation creates dissonance. To obtain a positive product in Figure 1, a youth could change the “?” to a “-” by deciding that the researchers are not competent or credible. Such a change creates balance, thus requiring no further change. (This is alternative A in the introduction to this article).
An alternative solution is for the youth group member to decide he or she heard things incorrectly; that is, the link between the researchers and the message was really negative. Then the youth group member could be favorable toward the researchers (+). (This could lead to either strategy B or C described in the experiential introduction.) Or the youth could try to change the researchers’ opinion on this matter.
Still another solution is for the youth to decide he or she is wrong and the researchers are right. Despite many years of belief in Christ as God, and many activities in support of that belief, he or she would change his or her opinion on the basis of a ten-minute communication. A key point emerges from the above analysis. The rational approach is rational only for the change agent. For the changee, change seems irrational. Should we change important beliefs each time someone thrusts disconfirming evidence on us? It is not surprising that “people are resistant to change.”
The rational approach implies that the target of the change is irrational.
Figure 1: Use of Balance Theory for the Jesus Christ Study
Balance theory can also be applied to the COMPU-HEART Case. The doctors’ viewpoint differed from the staff s. It seemed rational for the doctors to try to change the staff s viewpoint or to decide they did not like the staff. These reactions showed up clearly in the role playing.
THE DELTA TECHNIQUE The Delta Technique (named for the Greek letter commonly used to represent a small change) is my name
for the following synthesis of previously published research.
The Delta Technique draws on the three phases of the change process described by Lewin (1952). These are analogous to the phases involved in changing the shape of a piece of ice: unfreezing, changing, and refreezing. Rather than going directly to the change, Lewin stressed the need to devote time and energy first to helping the client to unfreeze. In view of its importance, the unfreezing phase is discussed here in some detail.
Youth Group Member
Message: “Christ is not God”
The unfreezing phase is critical, yet often ignored. It creates stress to examine one’s current beliefs. This stress frequently leads to the complaint: ”We’re wasting time. Let’s get on with the change.” But without the unfreezing phase, change efforts are likely to conflict with existing behavior. Referring back to the analogy, ice will shatter if enough pressure is applied to change its shape without going through the unfreezing phase.
Unfreezing can begin if a person is willing to tolerate the suspicion that his or her current procedure is not optimal. He or she then must search for disconfirming evidence (negative feedback). This is easier said than done; people generally seek evidence to support rather than disconfirm their beliefs.
The tendency for intelligent adults to avoid disconfirming evidence was demonstrated by Wason (1968). He provided three numbers”2 – 4 – 6” to subjects, and they were asked to determine what rule had been used to generate the numbers. To gain information, the subjects were allowed to generate other series of three numbers to see if each new series was in agreement with the rule. The typical subject thought of a rule and then generated series that were co-w-tent with his or her rule. For example, if he or she thought the rule was ”add 2 to each number,” the series 6- 8 – 10 could provide confirming evidence to him or her. It was unusual for a subject to use a series that was inconsistent with his or her own rule(e.g., the series 5 – 6 – 7 would violate the rule ”add 2 to each number”). In short, the subject avoided evidence that might disconfirm his or her belief.
Subjects who were told that their rules were incorrect in Wason’s study were allowed to generate additional series of numbers. The majority of these subjects kept the same rule that they had used previously, but they stated it in different words. Many people cling to their hypotheses even when given clear evidence that they are wrong.
The avoidance of negative feedback seems as common among scientists as among lay people, despite the fact that scientists receive formal education on the use of disconfirming evidence. Mahoney and DeMonbreun (1977) found that physical scientists and psychologists did no better than Protestant ministers on the 2 – 4 – 6 problem. Overall, only 13% of these subjects’ trials sought disconfirming evidence.
When disconfirming evidence is thrust on people, they may use it irrationally as did the youth group members in the Jesus Christ study: The believers increased their belief in God. Or people may remember incorrectly: Fischhoff and Beyth (1975) found that subjects remembered their predictions differently when the eventual outcome was in conflict with their prediction. Or they may get mad: Janis and Mann (1965) found that smokers were annoyed when people told them about evidence on the dangers of smoking, Most likely, however, they will ignore the information: Dawes (1979) provides a perfectly rational argument in support of some simple quantitative decision models, but the models were not used despite the overwhelming evidence that they would improve on the current decision process.
The Delta Technique addresses the unfreezing phase by encouraging the client actively to seek disconfirming evidence. The client should suggest alternatives to his or her current procedure and then define what information would lead to an adoption of one of the alternatives.
Clients sometimes have difficulty in defining what information would disconfirm the current procedures. If this occurs, the change agent can present the client with a series of possible research outcomes and ask what decision would be made given each possibility. Would the new evidence affect decision making?
The change agent should assist the client in developing a low-risk approach to change. The client should maintain control over the change process while the change agent should provide support.
Once the change is accomplished, it is important to obtain periodic feedback on the change. If the change is a desirable one, it helps to restructure the organization to ensure that the new behavior is rewarded.
APPLYING THE DELTA TECHNIQUEThis section outlines how you can apply the Delta Technique in a step-by-step fashion. Throughout this procedure, your role as a change agent is not to be the source of change, but to be a supporting actor in the client’s change efforts.
As change agent, you first work with the clients to define the problem. Ask them how they perceive the problem and ask them to state the problem in different ways. (This helps to keep open alternative solutions because a problem statement often implies a solution.) Review what you heard with the clients, to be sure that you have heard them correctly. Such a summary also helps clients to clarify their own viewpoints. Include your own opinions on the problem after this.
Your clients include the people who are paying you. But clients also include other groups that are affected by the change. These ”stakeholders” should be asked for their perspectives on the problem.
Given the problem statements, the change agent can help find alterative solutions. Once again, the clients are a major source of ideas. Clients may choose to maintain the current solution, but they should also experiment with at least one other feasible solution, even though this alternative can (and most likely will) appear to be inferior. In COMPU-HEART, the alternative involved the use of paramedics and computers, approaches that did not appeal to the doctors.
This use of alternative solutions, although often recommended, seems to be widely ignored. Even studies that are thought to represent the best in management science ignore this: Only 22% of the articles in the journal Management Science consider reasonable alternatives (Armstrong, 1979). By generating alternative solutions, management can change its role from “defender of the current solution” to “evaluator of future solutions.” Instead of disconfirming the current solution, evidence can confirm another of management’s alternatives. This is the second step of the Delta Technique and it completes the unfreezing phase.
Those affected must control each of the alternatives considered. To provide control, a three-step procedure is recommended: Experimentation, participation, and feedback.
Sell an experiment, not a change. The experiment should make sense to the people involved. It should provide a rational way to assess each of the possible solutions (e.g., it should allow for a comparison of alternatives to the current approach). The change agent provides expertise on experimental design and handles administrative details. An experiment can reduce risk by reducing the scope of the change and by putting a time limit on the change.
As the change agent, you should try to get the experiment started. Do this by asking for a commitment to act, however, small. That is, get your ”foot in the door.” The foot-in-the-door technique suggests that you ask initially for a small commitment. Preferably this commitment should be operational, it should be written, and it should have a time deadline. (For more on foot-in-the-door research, see Snyder and Cunningham, 1975.) In COMPU-HEART, users of the Delta Technique could ask the doctors to experiment with COMPU-HEART. To keep a foot in the door, a typical commitment was to ”have a meeting next Tuesday to discuss the details of an experiment.”
Encourage the stakeholders to participate in the design of the experiment (to the extent they desire). The clients should define the evidence that they would accept as sufficient to adopt an alternative solution. This may lead to suggestions on the experimental design. If the clients are not able to describe their change criteria, but are interested in the project, you can create some numbers and present ”possible outcomes” of the project. The clients can then discuss how each outcome would affect their decisions.
The doctors who went through the Delta Technique in the COMPU-HEART case often were able to outline what type of experiment would lead to change. These ”experiments” varied greatly from one group to another.
Evidence on the value of participation has been provided in studies since the 1930s. A review of research to the mid-1960s is provided in Blumberg (1968). The findings were almost always the same: When employees were given control over the situation, they were often willing to make changes. Once the decision to change was made, implementation was rapid and successful. This contrasts with the long delays involved in traditional change processes.
Numerous additional studies have been published since Blumberg’s review. They also show that participation helps in many situations. Lawler and Hackman (1969) demonstrated the value of participation in the adoption of a new incentive pay plan among building cleaners. French, Kay, and Meyer (1966) found that participation in goal setting by a subordinate manager increased the likelihood of successful changes by that manager. Bass (1977), using over 1400 managers from 12 countries in a simulated production task, found that teams that developed their own plans were more successful and more satisfied than those who were using an assigned plan.
Enable those involved to monitor the results of the experiment. Fixed checkpoints should be established so that clients can terminate the experiment if they desire. Control over the change process implies that those involved have the right not to change.
The feedback should summarize the success or failure for each alternative over a number of trials. In general, avoid case-by-case feedback because few people use such data in a rational way (Ward and Jenkins, 1965). 1nstead, wait until a series of events has been generated and then present a summary so users can understand the relationships. Positive feedback at this stage helps to refreeze the change.
Role-playing subjects using the Delta Technique proposed that the COMPU-HEART program would be summarized to show success rates by treatment. These summaries would be presented to the doctors on a regular schedule. (The value of actually providing the feedback was not assessed in the COMPU-HEART study.)
Research has demonstrated the value of feedback, especially when the people involved have control over this feedback (see review by Ilgen, Fisher, & Taylor, 1979). In n recent study, Becker (1978) showed that feedback was important in getting people to reduce their residential energy consumption.
When Delta Should Not Be Applied
The Delta Technique is not relevant for all situations. It does not seem to be as important for new employees, in short-term tasks (White, Mitchell, & Bell, 1977), or in other situations in which group norms have not been established (Hillery & Wexley, 1974). It is not useful if disagreement exists on the statement of the problem, a situation that is probably common in organizations.
One benefit of the Delta Technique is that it helps to avoid efforts that will not contribute to change. Beware of clients who say only, ”We are open to change and would like you to obtain more information.” If you cannot agree on the problem or if you cannot obtain commitment on what information could possibly be disconfirming to the present solution, it is best not to do the study. Certainly, many aspects of our lives (or the lives of our organizations) are not open to change. Do not spend time and money to collect information on these areas. If you ignore this advice, do not be surprised when you find that the study produces no change. Slovic and Fischhoff (1977) found that outcomes of research studies seldom surprised subjects, no matter what the reported results. Disconfirming evidence is most useful when the client is committed to finding such evidence.
THE DELTA TECHNIQUE AND YOU
The Delta Technique contradicts common sense. It is contrary to the years of experience during which we learned that, as change agents, we should use rational arguments and tell others how to change. We interpret our
own actions as the source of change in others. This phenomena was demonstrated by Strickland (1958). Subjects in his experiments were asked to act as managers for two workers whom I will call Stan and Ned. The experiment was designed so the manager could easily communicate with Stan. In contrast, communication was not possible with Ned. Actually, Stan’s and Ned’s work followed a fixed pattern that had no relation to their manager. Furthermore, they each produced the same amount and quality of work. Whom did the manager trust?
He trusted Ned. He thought Stan required constant supervision and encouragement from him to produce the same output.
The illusion that we control others is not restricted to managers in business. It applies in roles such as being a government bureaucrat, a parent, an athletic coach, or a teacher. For research on the illusion of control, see Langer (1975). Mark Twain said it well in describing a fight
Thrusting my nose firmly between his teeth, I threw him heavily to the ground on top of me.
Before trying the Delta Technique on others, it seems fitting that you use it on yourself. To do this, first find some disagreement between the Delta Technique and your beliefs. Then suspend evaluation and answer the following question ”What information could possibly convince me that this aspect of the Delta Technique is superior to my current behavior?” It helps to be specific and to write your answer to this question. (This is alternative D from the introduction.)
If you cannot find any ideas with which to disagree, or if you find an idea with which to disagree, but no information will change your mind, this article may be of little value. If you find an idea with which to disagree and can identify what information will change your mind, you must then decide how to obtain the information. To make sure you do this use the foot-in-the-door technique on yourself. First, find an idea from the Delta Technique. Then select a small step to get started. Set a time deadline for the completion of this step. Write:
• a description of the idea, • the situation in which you will use it, and • your time deadline.
(Based on responses from previous readers, you can make more effective use of this article if you invest at least five minutes now to write a ”foot-in-the-door” plan.)
EVIDENCE ON THE DELTA TECHNIQUEStudies on the various components of the Delta Technique have spanned half a century, although much of this research is recent. Note that over half of the studies cited in this article were published after 1970. The studies provide strong support for the Delta Technique. (The term ”Delta Technique” was not used in these studies; it refers to my summary of the change strategy.) Whether this section of the article is relevant to you depends not on the evidence cited here, but on the evidence for which you are looking. Three important and recent studies are briefly summarized here.
(1) Lonnstedt (1975) examined 107 operations research projects in twelve Swedish companies. He conducted personal interviews with the head of each of the operations research departments and telephone interviews with users. When the clients participated in the problem definition, the likelihood of implementation of the change increased from 38% with no client participation to 82% with client participation.
(2) Zand and Sorensen (1975) examined change projects in industry by use of a mail survey completed by 154 management scientists. In the successful projects, the clients were in control during the
”unfreezing, moving [change], and refreezing” phases of change. This seldom occurred for the unsuccessful projects.
(3) In a study of the implementation of 29 computer systems, Ginzberg (1979) examined the use of the Lewin change process. User perception of success was greater when the unfreezing, change, and refreezing phases were carried out successfully.
As noted at the beginning of this article, the most important way for you to gain evidence is to run your own experiment. The study by Nisbett and Borgida (1975) implied that personally generated evidence is more important to people than is evidence from studies done on others. This article allows the reader to replicate the COMPU-HEART role-playing case. Furthermore, the cited literature on change, with controlled studies, is easily available. In some cases, the change might not turn out as the change agent hoped: Those involved may prefer a “lower-quality solution.” Of course, it is not just the quality of the solution that determines its overall effectiveness; acceptance is also important. A lower-quality decision that is accepted is often preferable to a higher quality decision that meets resistance. The Delta Technique is aimed at improving overall effectiveness.
A major disadvantage of the Delta Technique is that management may feel that its power is eroding. If the managers give up control now, they feel they will have less influence in the future. Some people have argued the opposite, claiming that the mutual sphere of influence grows when we allow others to influence us. In any event, the loss of control represents a concern for some managers.
USE OF AN EXPERIENTIAL ARTICLEThe format of this article departs from tradition in order to give the reader an opportunity to use elements of the Delta Technique. How did you react? Did you... (choose closest response):
(A) disagree with the Delta Technique? (B) decide that a communication problem exists? (C) feel that you read an exciting article? (D) make a promise to experiment with the Delta Technique?
For those readers selecting A, B, or C, I hope that this article has at least created dissonance. Some readers, I hope, selected alternative D. Self-administered questionnaires were administered to nineteen Wharton MBA students who had read an earlier draft of this article: 32% selected D (experiment),47% selected C (exciting article),21% selected B (communication problem), and none selected A (disagreed with Delta).
CONCLUSIONSAn attitude survey and a role-playing experiment provided results consistent with naïve empiricism: namely, the typical approach to change is the rational approach. The rational approach is rational only from the change agent’s viewpoint. It provides an outside force on the target with the implication that the target has been acting irrationally. Thus, this approach is irrational from the changee’s viewpoint.
The Delta Technique removes the outside forces that are thrust on the client. The steps in the Delta Technique (illustrated in Figure 2) can help the client to go through the unfreezing, changing, and refreezing phases. The client should have control over each step.
The Delta Technique can help to identify feasible change projects and implement useful changes. It is appropriate for situations in which management sees a problem, but in which the desired change conflicts with their current beliefs. It is not needed in areas in which the change is consistent with the organization’s perceptions of its mission.
Evidence of the value of the Delta Technique was drawn from prior research and was supported by a role- playing experiment that can be easily replicated.
APPENDIX: INSTRUCTIONS FOR CHANGE AGENTSAssume that you had previous meetings with the doctors and that the problem you discussed (e.g., shortage of space in this exercise) was one that they view as a problem. Also assume that the solution, to use COMPU- HEART, was one that had been raised as a possibility in those previous meetings. The first thing you do is to summarize the results of those previous meetings. This is crucial. You are here to assist them with their problem and to examine a possibility that they had asked you to look at.
I. PROBLEM DEFINITION: Include all key stakeholders Obtain alternative statements of the problem Add your perspective as the change agent
II. ALTERNATIVE SOLUTIONS: Include all key stakeholders Develop list of reasonable alternatives Identify disconfirming evidence (e.g. assess reactions to possible new evidence)
III. EXPERIMENTATION (WITH PARTICIPATION) Use foot-in-door Reduce risk Work towards a rational design
Clients set criteria and constraints Clients influence design
IV. FEEDBACK Establish a schedule for feedback Provide summary data
Figure 2: Checklist for the Delta Technique
To address the change itself, the following guidelines should be used:
“Don’t sell the change, sell an experiment.”
Recognize that it is important to see how things will work in the particular situation. The experiment should allow the new method to be compared with the current method.
Allow the managers to participate in the experiment.
Put yourself in a position of being there to provide a service to the managers. If they want you to carry out the experiment that’s okay, but they will need to say how the decision will be made.
Therefore, you must:
a) Get them to specify the criteria (i.e., what will represent a success or a failure for a new method?).
b) Let them make suggestions about how the experiment should be conducted (e.g., What types of changes would they like to see examined?).
The key question is what information do they regard as sufficient before they make the change.
Let them know that they will be kept up to date on the experimental outcomes. This would allow them to terminate the experiment at any time.
Successful completion of this meeting will be in terms not of a promise to change, but rather of a specific promise to take the next small step (the “foot in the door”). For example, you could work “to schedule a meeting with top management to discuss the details for an experiment.”
REFERENCESArmstrong, J. S . (1979), “Advocacy and objectivity in science,” Management Science, 25, 423-428. Bass, B. M. (1977), “Utility of managerial self-planning in a simulated production task with replications in twelve
countries,” Journal of Applied Psychology, 62, 606-609. Batson, C. D. (1976), “Rational processing or rationalization? The effect of disconfirming information on a stated
religious belief,” Journal of Personality and Social Psychology, 32, 176-184. Becker, L. J. (1978), “Joint effect of feedback and goal setting on performance: A field study of residential energy
conservation,” Journal of Applied Psychology, 63, 428-433.
Blumberg, P. (1968), Industrial Democracy: The Sociology of Participation. New York: Schocken.
Brown, R. (1965), Social Psychology. New York: Free Press.
Dawes, R. M. (1979), “The robust beauty of improper linear models in decision making,” American Psychologist, 34, 571-582.
Fischhoff, B., & Beyth, R. (1975), “I knew it would happen: Remembered probabilities of once-future things,” Organizational Behavior and Human Performance, 16.
French, J.R.P., Kay, E., & Meyer, H. H. (1966), “Participation and the appraisal system,” Human Relations, 19, 3- 20.
Ginzberg, M. J. (1979), “A study of the implementation process,” in Robert H. Doktor, R. L. Schultz, and D. P. Slevin (Eds.), The Implementation of Management Science. New York: North-Holland.
Hillery, J. M., & Wexley, K. N. (1974), “Participation effects in appraisal interviews conducted in a training situation,” Journal of Applied Psychology, 9, 168-171.
Ilgen, D. R., Fisher, C. D., & Taylor, M. S. (1979), “Consequences of individual feedback on behavior in organizations,” Journal of Applied Psychology, 64, 349-371.
Janis, I. L., A Mann, L. (1965), “Effectiveness of emotional role playing in modifying smoking habits and attitudes,” Journal of Experimental Research in Personality, 1, 84-90.
Langer, E. S. (1975), “The illusion of control,” Journal of Personality and Social Psychology, 32, 311-328. Lawler, E. E., & Hackman, J. R. (1969), “The impact of employee participation in the development of pay incentive
plans,” Journal of Applied Psychology, 53, 467-471. Lewin, K. (1952), “Group decision and social change,” in G. E. Swanson et al. (Eds.), Readings in Social
Psychology. New York: Holt, Rinehart & Winston.
Lonnstedt, L. (19750, “Factors related to the implementation of operations research solutions,” Interfaces, 5, 23-30.
Marrow, A. J., & French, J.R.P. (1945), “Changing a stereotype in industry,” Journal of Social Issues, 1, 33-37.
Mahoney, M. J., & DeMonbreun, B. G. (1977), “Psychology of the scientist: An analysis of problem-solving bias,” Cognitive Therapy and Research, 1, 229-238.
Mosteller, F. (1981), “Innovation and evaluation,” Science, 211, 881-886.
Nisbett, R. E., & Borgida, E. (1975), “Attribution and the psychology of prediction,” Journal of Personality and Social Psychology, 932-943.
Slovic P., & Fischhoff, B. (1977), “On the psychology of experimental surprises,” Journal of Experimental Psychology: Human Perception and Performance, 544-551.
Snyder, M. & Cunningham, M. R. (1975), “To comply or not to comply: Testing the self perception explanation of the ‘foot-in-the-door’ phenomenon,” Journal of Personality and Social Psychology, 31, 64-67.
Strickland, L. H. (1958), “Surveillance and trust,” Journal of Personality, 26, 200-215. Ward, W. C., & Jenkins, H. M. (1965), “The display of information and the judgment of contingency,” Canadian
Journal of Psychology, 19, 231-241. Wason, P. C. (1968), “On the failure to eliminate hypotheses – a second look,” in P. C. Wason and P. N. Johnson-
Laird (Eds.), Thinking and Reasoning. Baltimore: Penguin.
White, S. E., Mitchell, T. R., & Bell, C. H., Jr. (1977), “Goal setting, evaluation apprehension, and social cues as determinants of job performance and job satisfaction in a simulated organization,” Journal of Applied Psychology, 62, 665-673.
Zand, D. E., & Sorensen, R. E. (1975), “Theory of change and the effective use of management science,” Administrative Science Quarterly, 20, 532-545.
Underlined terms take you to other definitions in the dictionary. References are provided at the end of each definition where relevant.
Abbreviations and Acronyms
Complete List of Defined Terms
Complete Dictionary References
This dictionary defines terms as they are commonly used in forecasting. The aims, not always met, are to:
- provide an accurate and understandable definition of each term,
- describe the history of the term,
- demonstrate how the term is used in forecasting,
- point out how the term is sometimes misused, and
- provide research findings on the value of the term in forecasting.
Note: If you are viewing a translation of this page, you must enter the search term in English.
Â© J. Scott Armstrong, The Wharton School, University of Pennsylvania
Programmed by Simon Galperin and Jason Cotton with support from the International Institute of Forecasters.