Dis/talk2

INTERFACES Copyright © 1979, The Institute of Managemenl Sciences Vol, 9, No. 5, November 1979 0O92-2102/79/O9O5/0121$O1.25

Management f 3 Science

PROCESS

ROBERT J. GRAHAM

Editor's Note

Many Management Scientists think that the Management Science process begins with constructing a model. The editor of this column believes that the successful process begins long before the first equation is written, at a stage called problem identification. The guest paper given below presents and contrasts three methods of problem identification that Management Scientists may find helpful in identifying the correct problem — as opposed to elegantly solving the wrong problem.

Robert J. Graham

METHODS FOR MANAGERIAL PROBLEM CAUSE ANALYSIS

John C. Anderson and Marius A. Janson

Graduate School of Business Administration, University of Minnesota, Minneapolis, Minnesota 55414

ABSTRACT. The importanl managerial processes of problem identifi- cation, formulation, and solution are often approached experientially. There are. however, several formalized approaches which have been developed and are currently the subject of many managemenl developmeni programs. These approaches altempl lo structure the eslablishmcnl of a cause and effect relationship between a problem and its causets) for purposes of taking cor- rective action. This paper summarily presents the cause and effect methods, critically evaluates thetn. explores their differences, similarities, and the kind of problems for which each method is appropriate.

Introduction Solving problems may be regarded as tbe essence of the management process.

This process consists of problem identification and formulation, design of solutions for the problem identified, and the implementation of the solution deemed most appropriate.

The supposition behind this process is that a cause and effect relationship exists between a problem and its cause(s) and that the proper intervention in the relationship

SIMULATION—SYSTEMS DYNAMICS

INTERFACES November 1979 121

will result in a resolution of the problem. The correct identification of the problem's cause(s) is thus crucial to the success of the management process.

Problem identification is frequently done in an intuitive manner; in fact Pounds states " . . . managers rarely, if ever, understand or analyze the causes of their prob- lems" [5, p. 1]. That is not to say that management does not engage in cause-finding activity, but rather that this activity is often ill structured, experiential, and subjec- tive.

The authors feel that a more formalized problem identification approach would be beneficial and subsequently lead to better solutions. This paper examines some alternative formal methods of problem identification and critically evaluates their differences and similarities. The methods examined are; Cause and Effect diagrams, the Kepner-Tregoe approach, and Control Data Corporation's Alpha-Omega method. The methods differ in (1) the type of problems they help to identify, (2) the extent to which they support the development and implementation of a solution, and (3) the degree to which they use verbal and graphical techniques of problem structuring.

This paper briefly illustrates how each of the three methods would be employed in a particular problem situation and proceeds to compare and contrast the methods along a number of characteristics.

The problem The common problem used in the illustration of application and for comparison

of effectiveness of tbe three methods is the "Blackened Filament Problem" (Kepner and Tregoe, pp. 25-38), as summarized below;

These events look place in a large, well managed plant making plastic filament for textiles. The plant had six huge machines that extruded viscose raw material through tiny nozzles into acid-hardening baths where the viscose streams become plastic strands. As each filament moved through the acid bath it was supported by lead pulleys and ferrules. Each machine had an exhaust fan evacuating fumes from the enclosed acid environment through roof vents and a large duct with an air intake at the rear of the building.

Each machine mounted 480 nozzles, and each of the 480 gossamer strands of filament produced was at one point in the process spun into a revolving hard rubber bucket. Cen- trifugal force of the spinning bucket threw each strand against the side of its bucket and thus buill up layers of filament from the outside in toward ihe center. Every eight hours the case of titament strands in each bucket had to be emptied, and the process was timed so that the six doffers. the men who tended the buckets, could each empty a bucket on his machine every minute or 480 in an eight-hour shift. This tightly scheduled operation ran like clockwork 24 hours a day. wilh a relief doffer on call should there be any trouble.

Early one morning, trouble came. One of the doffers on the midnight shift emptied bucket No. 232 out on the work table and noticed something strange. Inside the core of the filament cake he saw that the last filament that had come off the machine was dirty black instead of translucent. He didn't stop to wonder about it, however, and went on to empty the next bucket a minute later. Again he saw there was even more blackened plastic in the core of the filament cake, and again he went on to handle the next bucket, it was found that the blackening was caused by carbon being deposited on the filaments. Possible carbon sources were identified as; a locomotive moving through the company yard, a coal burning boilerhouse. and carbon cars stockpiled in the yard.

The above problem is a good one for illustration in that most managers when faced with the situation would engage in some form of cause and effect analysis in an effort to resolve the blackened filaments. The reader is invited to think about how he or she would approach structuring this problem.

122 INTERFACES November 1979

What follows will be the way this problem Is diagnosed using the three methods examined. First, the formalized structure of each method will be illustrated and then they will be compared and contrasted.

Cause and Effect Diagram The Cause and Effect Diagram, as developed by Inoue and Riggs [3], essentially

is a graphical method, employing vectors indicating the relation between cause and effect. The cause and effect diagram for the blackened filament problem is presented in Figure 1. The exhibit divides into two sections. To the left of the goal or problem box are the principal causes, indicated by major arrows which end in the horizontal main shaft.

FIGURE 1.

Minor arrows directed into the major arrows constitute principal cause control parameters. Faced with a problem one starts by putting down major arrows which are believed to be determinants of the cause and effect relationship. Figure I denotes these as "process" and "environment." The graph is expanded by adding paramet- ers which are thought to control the causes. For example, carbon is a control parame- ter of the environment and can itself come from various sources. New insights occur as the cause and effect diagram is being constructed, which can be expanded without restriction. To the right of the problem or goal box are tbe main effects resulting from the principal causes. For example. Figure I shows filament quality as a main effect. Basic effects are the result of main effects, and are shown by basic effect arrows placed on main effect arrows. In this case basic effects of quality are rejects and

INTERFACES November 1979 123

filament clarity. By adjusting control parameters one influences the principal causes, which in turn influences the main effects, finally influencing basic effects.

In addition to providing insight into relationships between causes and effects, the diagram also serves as a list for things to check. Putting it differently, the cause and effect diagram provides structure in the search for problem causes. The search is clustered around the control parameters, i.e.. once carbon has been identified as the blackening agent, all possible carbon sources are investigated. When the source has been positively identified, it is either removed from the environment or its effect eliminated.

Alpha Omega Alpha Omega [ I ] , developed by Control Data Corporation, is a bit more exten-

sive in that it provides a formal structure to the entire problem solving process. The particular phases employed are; problem identification, search for possible causes, development of alternative solutions, solution selection, and implementation. As in the previous model, a cause and effect relationship is assumed between a problem and its cause. Problem cause identification takes place in the context of the system's approach, a system being defined in terms of its elements and the relationships between these elements as shown in Figure 2 for the "blackened filament" problem. To begin, system variables which are likely to influence the problem are identified. The investigator then searches for possible differences in the actual and normal value of these variables. One then goes through a well defined process to ensure that all likely variables or causes are included in the system. A theory or framework is then developed which attempts to explain the problem and variables causing it. This theory is tested by the methods of logical induction, covariation, empirical verifica- tion, and time lines. Time lines specify the time of occurrence and duration of a problem and serve to rule out possible causes which do not agree with problem time lines. Problem identification and system design are viewed as parallel activities so that the system can be expanded to include more variables as the problem identifica- tion and formulation progresses.

The exhibit divides into cause and effects sections. The causes are given + and — signs indicating the direction of effect change in response to a cause change. For example, filament translucence increases when the distance between the locomotive and air intake increases. Causes may influence filament quality in an indirect way. This is the case where acid purity and ferrule conditions affect acid bath conditions while the latter, in turn, affects filament quality. The effect of a cause may, in turn, be a cause of another effect. Figure 2 looks similar to the cause and effect diagram of Figure 1, but has the distinction of focusing on a more specific problem. Figure 2 shows a detailed relationship between cause and effect and therefore is less flexible than a cause and effect diagram during the early stages of problem identification, when cause linkages are quite tentative.

Alpha Omega has a comprehensive structure leading into problem solution and further a formalized disciplined structure for identifying and verifying the existence of cause and effect.

124 INTERFACES November 1979

FIGURE 2,

DISTANCE LOCOMOTIVE AND AIR INTAKE

FILAMENT TRANSIUCENCIES

ACID PURITT

ROFIT

• l .

Kepner-Tregoe method The Kepner-Tregoe method as developed in The Rational Manaf-er [4], like

Alpha Omega, consists of a number of phases which include problem identification, information gathering, cause identification, alternative solution evaluation, and final solution selection.

A problem cause is identified by searching for differences in certain variables under before and after problem conditions. This implies that there is an established desired state and that the actual state is deviant. In other words, one is attempting to reach rather than alter some desired goal.

The process of diagnosing problem cause is structured around concepts of specifying location of problem occurrence, placing it in time, and measuring tbe extent of differences between actual and desired states. Problem characteristics are divided into " I S " and "IS NOT" classes. Kepner and Tregoe provide a tabular tool embodying above concepts. Table I illustrates this tool for the blackened filament problem. Tbis table shows the problem as a deviation from the desired goal of producing translucent filament, A distinction between " I S " and "IS NOT" defines what is different between them. Finding this distinction and the change from which it resulted directs the search for possible causes. Possible causes are verified and a final problem cause is selected using the distinction criteria.

INJERFACES November 1979 125

TABLE 1.

Whai deviation object

Where on objeci observed

When on object observed

Exieni How much How many

Possible causes fc7 test

IS

Black deposit — caibon filamenl on machine # 1

On .surface On machine # 1

Al what stage in manu- facturing process —

in acid bath At what time did il occur

Heavy deposii. unifonn # filaments affecled

— ail

TSNOT

Any other material Any other machine

Wiihin filament On any niher machine

Nol befon; acid bath When did it occur

Slight deposit, inierminani

# rdaments not affecled — none

Location of locomotive ai time of probkm

Distinction of t h e - I S "

Carbon deposii

Machine #1 ax intake

Location of locomotive

All filament on machine #1 were

aflected — unifonnly

Change

-

—

Locomotive moves ihmugh the yarrf

The Kepner-Tregoe method allows for multiply-caused problems; however, it has strong orientation towards problems of a technical nature which are assumed to have a single cause. This is evident from Table 1, which becomes very unwieldy for complex, multiply-caused problems.

As does Alpha Omega, Kepner-Tregoe has a comprehensive structure leading to problem solution. Kepner-Tregoe is a tabular way of attaining many of the goals of Alpha Omega but, in addition, uses the categories of what, where, when, to what extent, and distinction as a focus for diagnosing the problem.

Comparing the methods

This paper has briefly illustrated the features of three cause and effect analysis methods. We must now compare and contrast the methods in an effort to clarify effective use. Table 2 provides a helpful summarization of the characteristics of each of the methods. Some summary statements of each of the methods and overall assessment follows.

Cause and Effect Diagrams are most useful in the stages of problem formulation, problem cause identification, and the design of alternative problem solutions. They do not, however, provide structure in the selection of a solution implementation strategy. Cause and Effect Diagrams are appropriate for problems of a technical and nontechni- cal nature, and effective in expressing one's growing insights into a problem with a minimum of effort. The usual approach allows expanding or altering the diagram in one area without disturbing other nonrelated diagram sections. The nature of the problem may be either singly or multiply caused. The relative lack of a disciplined structure is an advantage to the Cause and Effect Diagram approach in that it allows ease of evolution which other methods do not possess. Cause and Effect Diagrams can be used in conjunction with other problem solving methods such as Alpha Omega.

126 INTERFACES November 1979

Dimension

Goal Seeking Goal Changing Problem Idcniificalion Cuusc Itlcntillcation Solution Search Decision Analysis Decision Implementation Ease of Mtxlel Building Timed Needed to Apply Method Technical Problem Nontechnical Problem Single/Multiple Cause

TABLE 2.

C & E Diagram

graphical

yes yes + +

+ -\- + -t-

- - - - + +

short yes yes

Multiple

Alpha Omega

graphical/ descriptive

yes yes

+ + + + + + -t--i--i-

+ + + + +

+ + + moderate

yes yes

Multiple

Kepner-Tregoe

descriptive

yes

no + + + + -H-H

+ + + +

long yes ?

Single

Nole: The larger the number of + signs, the greater the degree to which the characteristic is present.

Alpha Omega offers the decision maker assistance in problem formulation, cause identification, and design of and selection hetween alternative solutions, and it sup- ports the formulation of a solution implementation strategy. Problems may be of a technical and nontechnical nature, multiply or singly caused. Problem cause and effect are graphically expressed in a vector diagram. This diagram may not be as flexible as the Cause and Effect Diagram. For the latter, relations between possible cau.ses and effects may be quite ill defined or speculative which is very helpful in diagnosing and verifying problem causes.

Kepner-Tregoe is one of the early problem solving methods. Its features consist of problem identification, design of alternative solutions, selection of an appropriate solution, and potential problem analysis resulting from the solution implementation. A strong case is made for the idea that a problem has a single cause. Problem cause identification is given focus by completing an " I S " and "IS NOT" table. This table is quite time consuming to fill out and the least flexible of all three methods. Kepner- Tregoe is oriented towards problems of a technical nature for which relatively simple solutions are thought to exist.

A further difference between the three methods exists in terms of the problem type for which they are appropriate. Problems may be divided into goal seeking and goal changing problems [2]. A goal seeking problem occurs when the actual state has drifted away from the desired goal. In the case of a goal changing problem, the manager has set a new goal which is different from the old goal. This classification is important in that it determines the appropriateness of problem diagnosing methods. Both Cause and Effect Diagrams and Alpha Omega are appropriate for goal seeking and goal changing problems.

INTERFACES November 1979 127

A final comment

The foregoing has been what we hope is a helpful presentation and clarification of some alternative methods of problem formulation which structure cause and effect determination — an area of modeling that is all too often overlooked in Management Science literature. It is believed that each of the methods can potentially aid man- agement problem diagnosis and search for problem causes — and perhaps one could conclude with such a statement. However, in a column on process, it seems appro- priate to conclude with a comment on implementation.

To varying degrees, each of the methods provides a systematic, structured ap- proach to problem formulation which could be successfully implemented where management-personalized experiential methods fall short. Two of the approaches have been widely presented within the context of management development semi- nars. Kepner-Tregoe, in particular, has been presented in many large corporations in some form or another since the middle l960's.

It is fair to say that formal implementation is very low. We know of very few people who have formally implemented the entire method in problem analysis. Cer- tainly, more complete applications may exist. However, the interesting observation is that practically all managers who have studied the approaches seem to integrate some of the di.scipline into their problem diagnosis repertoire of personalized methods. For example, it is not uncommon for managers to use the "what is, what is not the problem." or "distinction," ideas in Kepner-Tregoe — or the "Theory test" in Alpha Omega — or the "hierarchy of control parameters" in Cause and Effect Diagrams.

Like many other "models," these methods have opportunity for complete im- plementation where time, degree of risk, and personal knowledge permit, and further they possess discipline within their construction that potentially can become a very important part of individual management problem analysis.

REFERENCES

[1] Alpha Omega '-Problem Analysis and Decision Making," Control Dala Corporation (1976). [2] Chen. C . '"What is Ihe Systems Approach." Interfaces. Vol. 6. No. I (November 1975). [3] Inoue. M.S. and Riggs. J L , , "Cause and Effect Diagrams." Industrial Engineering {April 1971),

pp. 26—31. [4] Kepner. C. and Tregoe, B., The Rational Manager. McGraw-Hill (1965)- [ 5 ] P o u n d s . W . F . , " T h e Pmce>ii.of P r o b l e m F i n d w g . " lndu.strial Management R e v i e w . V o l . I I . N o . 1

(Fall 1969). pp. 1—19.

128 INTERFACES November 1979