Euclid: Strategic AlternativeAssessment Matrixtavana.us/publications/EUCLID.pdf · Euclid: Strategic AlternativeAssessment Matrix MADJIDTAVANA* Management Department, La Salle University,

JOURNAL OF MULTI-CRITERIA DECISION ANALYSIS

J. Multi-Crit. Decis. Anal. 11: 75–96 (2002)

Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/mcda.318

Euclid: Strategic AlternativeAssessmentMatrix

MADJIDTAVANA*Management Department, La Salle University, Philadelphia, Pennsylvania, USA

ABSTRACT

The vast amount of information that must be considered to solve inherently ill-structured and complex strategicproblems creates a need for tools to help decision makers (DMs) recognize the complexity of this process anddevelop a rational model for strategy evaluation. Over the last several decades, a philosophy and a body of intuitiveand analytical methods have been developed to assist DMs in the evaluation of strategic alternatives. However, theintuitive methods lack a structured framework for the systematic evaluation of strategic alternatives while theanalytical methods are not intended to capture intuitive preferences. Euclid is a simple and yet sophisticatedmultiobjective value analysis model that attempts to uncover some of the complexities inherent in the evaluation ofstrategic alternatives. The proposed model uses a series of intuitive and analytical methods including environmentalscanning, the analytic hierarchy process (AHP), subjective probabilities, and the theory of displaced ideal, to plotstrategic alternatives on a matrix based on their Euclidean distance from the ideal alternative. Euclid is furthercompared to the quantitative strategic planning matrix (QSPM) in a real world application. The informationprovided by the users shows that Euclid can significantly enhance decision quality and the DM’s confidence. Euclid isnot intended to replace the DMs, rather, it provides a systematic approach to support, supplement, and ensure theinternal consistency of their judgments through a series of logically sound techniques. Copyright # 2003 JohnWiley & Sons, Ltd.

KEY WORDS: strategic decision making; environmental scanning; analytic hierarchy process; subjectiveprobabilities; theory of displaced ideal

1. INTRODUCTION

Complexity forms the essence of strategic decisionmaking. The process of strategic decision makinginvolves so many dimensions that the solutionsand simplifications in one dimension cause pro-blems and complexities in another. Strategicdecision making is by its nature ill-structured,complex and yet very important for the successand survival of an organization (Schwenk, 1988).It is often difficult to cope with large volumes ofpotentially important information during strategyevaluation. Strategic decision making tends to becharacterized by high levels of complexity, puttingenormous cognitive demands on decision makers(DMs). The vast amount of information that mustbe considered for solving these inherently ill-structured strategic problems creates a need fortools to help DMs recognize the complexity of thisprocess and develop a rational model for strategy

evaluation (Fahey et al., 1981; Klein and Newman,1980; Weigelt and Macmillan, 1988). Hofer andSchendel (1978) note that most DMs perform farbetter when they separate this process into distinctsteps, address each step separately, and thencombine the results at the end.

Over the last several decades, a philosophy anda body of intuitive and analytical models havebeen developed to assist DMs in the evaluation ofstrategic alternatives. Some of the intuitive modelsinclude dialectic policy analysis (Mitroff, 1982);vulnerability analysis (Hurd, 1979); corporatesimulation models (Ginter and Rucks, 1984;Naylor and Schauland, 1976; Nees, 1983); portfo-lio models (Cardoza and Wind, 1985; Naylor,1982); dialectical inquiry, devil’s advocacy, and theconsensus approach (Schweiger et al., 1986, 1989:Schweiger and Sandberg, 1989). Although thesetechniques have a theoretical appeal, their applica-tion seems to be ad hoc, and they lack a structuredframework for the systematic evaluation of stra-tegic alternatives. Some of the analytical modelsinclude strategic program evaluation (Grant andKing, 1982a; King, 1980), quantitative strategicplanning matrix (David, 1985, 1986, 1993), Electre

Copyright # 2003 John Wiley & Sons, Ltd.

*Correspondence to: Professor M. Tavana, Manage-ment Department, La Salle University, Philadephia, PA19141, USA. E-mail: [email protected]

II (Godet, 1987), the McKinsey matrix (Hill andJones, 1989), competitive strength assessment(Thompson and Strickland, 1998), the scenario-strategy matrix (Naylor, 1983), and decisionsituation outcomes evaluation (Radford, 1980).These techniques have made contributions to thestrategy evaluation process, but are not intendedto capture intuitive preferences or to modelenvironmental processes (Edwards et al., 1988).Other more recently developed methods blendintuitive and analytical approaches (Tavana andBanerjee, 1995). However, these methods lacksimplicity and are primarily used by highly trainedmanagers. Simplicity helps the DM implement thetools that have been developed and communicatethe results of this utilization (Edwards et al.,1988).

Schoemaker and Russo (1993) describe fourgeneral approaches to decision making rangingfrom intuitive to highly analytical. These methodsinclude intuitive judgments, rules and shortcuts,importance weighting, and value analysis. Theyargue that analytical methods such as importanceweighting and value analysis are more complex butalso more accurate than the intuitive approaches(Schoemaker and Russo, 1993). Euclid is a simpleand yet sophisticated multiobjective value analysismodel that attempts to uncover some of thecomplexities inherent in the evaluation. Theproposed approach uses a series of intuitive andanalytical methods to plot strategic alternatives ona matrix based on their Euclidean distance fromthe ideal alternative. Using the theory of displacedideal to grasp the extent of the emerging conflictbetween means and ends, the DM explores thelimits attainable with each environmental benefitand risk. The highest achievable scores with allcurrently considered benefits and risks form acomposite, an ‘ideal alternative’ (Zeleny, 1982).

Euclid captures the DM’s beliefs through aseries of sequential, objective, and structuredprocesses. Organizational members demand asound rationale for strategic decisions. Unless thatrationale is seen to be at least somewhat structuredand objective, the commitment required may notexist or be sustained (Moore, 1995). An objectiveand structured strategic decision making proce-dure can be very useful in clarifying the elementsof a decision and arriving at the best decision(Kirkwood, 1997). Strategic decision making canbe improved using a formal decision makingprocess (Ansoff, 1980; Camillus, 1975). Quantita-tive data often clarifies the elements of strategic

decisions, and it forces DMs to be explicit abouttheir reasoning. This improves decision making,and it also aids communication about the basis fora decision (Kirkwood, 1997).

Euclid promotes comprehensive scanning of theenvironment by decomposing the environmentalforces into internal, transactional, and contextualbenefits and risks as suggested by many research-ers (Daft et al., 1988; Grant and King, 1982a,b;Rhyne, 1985; Schmid, 1986). The open systemapproach to organizations recognizes that there isan identifiable, if semi-permeable, layer betweenthe organization and its environment. It is vital toview the internal context of the organizations asbeing embedded in its external environment. Thecontext of an organization (both internal andexternal) refers generally to the environment fromwhich it has emerged and in which it exists and canreasonably be seen as the fundamental base fromwhich the success and failure of strategy can beunderstood.

Euclid uses intuitive and analytical methodssuch as environmental scanning, the analytichierarchy process (AHP), subjective probabilities,and the theory of displaced ideal to enhancedecision quality and the DM’s confidence inevaluating a set of strategic alternatives. The nextsections present a step-by-step explanation of theprocedure, the model, a practical application,managerial implications, and conclusion.

2. THE PROCEDURE

Euclid uses a six-step procedure to systematicallyevaluate potential strategies by plotting them in afour-quadrant matrix based on their Euclideandistance from the ideal alternative. The six stepsused in Euclid are:

(i) Generate strategic alternatives;(ii) Identify the relevant benefits and risks and

group them into internal, transactional, andcontextual sets of environmental factors;

(iii) Define environmental weights and the im-portance weight of each benefit and risk;

(iv) Develop subjective probabilities for eachalternative;

(v) Identify the ideal probabilities and calculatethe total Euclidean distance of each strategicalternative; and

(vi) Select the most attractive alternative usingvisual and numerical information.

M. TAVANA76

Copyright # 2003 John Wiley & Sons, Ltd. J. Multi-Crit. Decis. Anal. 11: 75–96 (2002)

Each of these steps is described below.

(i) Generate strategic alternatives: Alternativesare the set of potential means by which thepreviously identified objectives may be attained.There must be a minimum of two mutuallyexclusive alternatives in the set to permit a choiceto be made (Zeleny, 1982). In well-structuredproblems, alternative generation may be relativelyroutine. However, in most strategic problems thistask is anything but routine because this type ofproblem is normally ill-structured. Alternativescan be generated using various brainstorming andintuitive methods. Keller and Ho (1988) discussseveral alternative generation approaches thatcould be used in a formal strategic decisionmaking model.

(ii) Identify the relevant benefits and risks andgroup them into internal, transactional, andcontextual sets of environmental factors: Next,environmental scanning is used to assemble andanalyse potential benefits and risks within theinternal, transactional, and contextual environ-ments. The internal environment consists ofbenefits and risks within the functional areas of afirm such as finance, sales, personnel, and manu-facturing. The transactional environment includesfactors that directly affect and are affected by anorganization’s major operations. Some of thesefactors include benefits and risks associated withcompetitors, customers, creditors, suppliers, laborunions, trade associations, and special interestgroups. The contextual environment includesmainly uncontrollable factors such as socio-cultural, technological, political, and legal benefitsand risks. When more than one DM is involved inthe decision process, all the DMs participatecollectively in identifying the benefits and risks.Aguilar (1967) presents a detailed discussion ofenvironmental factors that could be consideredduring the strategic decision making process.

(iii) Define environmental weights and the im-portance weight of each benefit and risk: Multi-criteria decision making (MCDM) techniquesrequire the determination of a set of weights thatreflect the relative importance of various compet-ing objectives. Several approaches such as pointallocation, paired comparisons, trade-off analysisand regression estimates could be used to specifythese weights (Kleindorfer et al., 1993). Euclidutilizes AHP developed by Saaty (1977, 1980,1990a) to estimate environmental weights forbenefits ðWbt Þ and risks ðWri Þ. AHP is also used

to estimate the importance weight of each benefitðFbij Þ and risk ðFrij Þ. The process is simplified byconfining the estimates to a series of pairwisecomparisons. The measure of inconsistency pro-vided by AHP allows for the examination ofinconsistent priorities.

Shoemaker and Waid (1978) have comparedAHP with three commonly used multicriteriadecision making techniques including multipleregression, the multi-attribute utility approach ofKeeney and Raiffa (1976) and simple directassessment. These methods differ in several ways:(1) they require different types of judgments, (2)they require different response modes, and (3) theyhave different application domains. Shoemakerand Waid (1978) show that all four methodsproduce similar results, but each has advantagesover the others. AHP does not require consistencyamong preferences, while the construction of autility function by the multi-attribute utilityapproach requires a transitive preference relation.In addition, AHP provides more detailed informa-tion from its pairwise comparison process, and it isapplicable in domains where non-measurablefactors exist. For repetitious decision-makingsituations, the multi-attribute utility approach ismore advantageous. However, in practice, utilityfunctions change rapidly and need to be re-evaluated frequently. Thus, the multiple attributeutility approach does not do better operationallythan AHP. AHP is also preferred to multipleregression and direct access for non-repetitivedecision making situations because these situationsdo not allow an easy derivation of measurablefactors. Euclid does not use AHP conventionallyto determine the relative importance of eachalternative in terms of the decision factors.Instead, probabilities of occurrence are used tocapture the relative performance of each alter-native. These probabilities are used to identifyideal probabilities and form the ideal alternativediscussed later. AHP has been a very populartechnique for determining weights in MCDMproblems (Shim, 1989; Zahedi, 1986). The advan-tage of AHP is its capability to elicit judgmentsand scale them uniquely using a procedure thatmeasures the consistency of these scale values(Saaty, 1989).

Euclid is a normative MCDM model withmultiple factors representing different dimensionsfrom which the alternatives are viewed. When thenumber of factors is large, typically more than adozen, they may be arranged hierarchically (Saaty,

EUCLID: A VALUE ANALYSIS MODEL 77


1980; Triantaphyllou, 2000; Triantaphyllou andMann, 1995). Euclid assumes a hierarchicalstructure by initially classifying decision factorsinto opportunities and threats. Opportunities andthreats are further classified into internal, transac-tional, and contextual categories. This hierarchicalstructure allows for a systematic grouping ofdecision factors in large problems. The classifica-tion of different factors is undoubtedly the mostdelicate part of the problem formulation (Bouys-sou, 1990) because all different aspects of theproblem must be represented while avoidingredundancies. Roy and Bouyssou (1987) havedeveloped a series of operational tests that canbe used to check the consistency of this classifica-tion.

There has been some criticism of AHP in theoperations research literature. Harker and Vargas(1987) show that AHP does have an axiomaticfoundation, the cardinal measurement of prefer-ences is fully represented by the eigenvectormethod, and the principles of hierarchical compo-sition and rank reversal are valid. On the otherhand, Dyer (1990a) has questioned the theoreticalbasis underlying AHP and argues that it can leadto preference reversals based on the alternative setbeing analysed. In response, Saaty (1990b) ex-plains how rank reversal is a positive feature whennew reference points are introduced. In Euclid, thegeometric aggregation rule is used to avoid thecontroversies associated with rank reversal (Dyer,1990a,b; Harker and Vargas, 1990; Saaty, 1990b).

(iv) Develop subjective probabilities for eachalternative: The probability of occurrence of eachbenefit ðpmbij Þ and risk ðpmrij Þ for each alternative isestimated. Subjective probabilities are commonlyused in strategic decision making because theyrequire no historical data (observation of regularlyoccurring events by their long-run frequencies) (DeKluyver and Moskowitz, 1984; Schoemaker, 1993;Schoemaker and Russo, 1993; Vickers, 1992;Weigelt and Macmillan, 1988). Subjective prob-abilities can be measured by asking a DM for theodds on an event. If the DMs are familiar withprobability concepts, they can be asked directly forthe required probability. If not, some sort ofmeasuring instrument is required. Some research-ers suggest using verbal phrases such as ‘‘likely,’’‘‘possible,’’ ‘‘quite certain,’’ etc., to elicit therequired information and then converting theminto numeric probabilities (Brun and Teigen, 1988;Budescu and Wallsten, 1985; Tavana et al., 1997).Other commonly used approaches include reason-

ing (Koriat et al., 1980), scenario construction(Schoemaker, 1993) and cross-impact analysis(Stover and Gordon, 1978). In this study, theverbal probabilistic phrases in Table I were used toelicit numeric probabilities as suggested by Tavanaet al., (1997). Alternatively, the DM may usenumeric probabilities instead of the probabilisticphrases. Merkhofer (1987) and Spetzler and Staelvon Holstein (1975) review some probabilityelicitation procedures that are used in practice.

The probabilities associated with the decisionfactors are assumed to be binomial. Binomialprobabilities are commonly used in strategicdecision making so that the decision maker cansimplify the problem by analysing possible out-comes as either occurring or not occurring. Forexample, Schoemaker (1993) assigns binomialprobabilities to factors such as ‘‘Dow JonesIndustrial Average falling below 1500 mark by1990’’ or ‘‘Election of a Democrat as US presidentby 1990.’’ Vickers (1992) also assigns binomialprobabilities to similar factors such as ‘‘Japanesecar manufacturers gain at least 30% of theEuropean market share’’ and ‘‘The incorporationof East Europe into Europe by 1993’’ in order toexamine the future of European automobileindustry. The main motivation for using thebinomial probabilities is to reduce the complexityof the model and allow DMs use event-drivenfactors.

(v) Identify the ideal probabilities and calculatethe total Euclidean distance of each strategicalternative: Euclid is a weighted-sum, multicriteriadecision model (Triantaphyllou, 2000). Manyweighted-sum models such as strategic programevaluation (Grant and King, 1982a,b; King, 1980)and QSPM (David 1985, 1986, 1993) have been

Table I. Verbal probabilistic expressions and perceivedprobability estimates

Verbal Expression Probability

Impossible 0.00

Small possibility 0.10

Small chance 0.20

Somewhat doubtful 0.30

Possible 0.40

Toss-up 0.50

Somewhat likely 0.60

Likely 0.70

Very likely 0.80

Quite certain 0.90

Certain 1.00

M. TAVANA78


developed to help DMs deal with the strategyevaluation process. The weighted-sum scores inEuclid are used to compare potential alternativesamong themselves and with the ideal alternative.The concept of ideal alternative, an unattainableidea, serving as a norm or rationale facilitatinghuman choice problem is not new. See for examplethe stimulating work of Schelling (1960), introdu-cing the idea. Subsequently, Festinger (1964)showed that an external, generally non-accessiblealternative assumes the important role of a pointof reference against which choices are measured.Zeleny (1974, 1982) demonstrated how the highestachievable scores on all currently considereddecision criteria form this composite ideal alter-native. As all alternatives are compared, thosecloser to the ideal are preferred to those fartheraway. Zeleny (1982, p. 144) shows that theEuclidean measure can be used as a proxy measureof distance. Using the Euclidean measure sug-gested by Zeleny (1982), Euclid synthesizes theresults by taking the square root of the sum of thesquares of the benefit and risk ratings.

As suggested by Zelany (1982), the idealprobability of each benefit and risk is identified.The ideal probability for a benefit ðP�

bijÞ is the

highest probability among the set of subjectiveprobabilities developed in the previous step whilethe ideal probability for a risk ðP�

rijÞ is the lowest

probability among the set. As shown in Equations(1) and (2), two composite scores called the totalEuclidean distance from the ideal benefit ðBmÞ and

total Euclidean distance from the ideal risk Rmð Þare developed for each strategic alternative. First,the Euclidean distance of each probability from itsrespective ideal probability is calculated. ThisEuclidean distance is the ideal probability minusthe probability squared. Next, the compositescores which are the sum of the Euclideandistances for each benefit and risk times itsimportance weight and the weight of the environ-ments are determined.

(vi) Select the most attractive alternative usingvisual and numerical information: The strategicalternatives are next plotted on a matrix similar toFigure 1 along with the mean Euclidean distanceof benefits ðBÞ and risks ð %RRÞ. The horizontaldimension (x-axis) indicates Bm while the verticaldimension (y-axis) shows Rm. With the idealalternative ðB� ¼ 0Þ; ðR� ¼ 0Þ as the origin, themean Euclidean distance of benefits and risksdivide the matrix into four quadrants: Exploita-tion, Challenge, Discretion, and Desperation Zones.

Exploitation Zone: In this quadrant, benefits arestrong ðBm4 %BBÞ and risks are weak ðRm4 %RRÞ be-cause the strategy is close to both the ideal benefitand risk points. This area represents the greatestpossible advantage for a firm. Strategies falling intothis zone should be considered seriously by the firm.

Challenge Zone: In this quadrant, benefits arestrong ðBm4 %BBÞ and risks are strong ðRm > %RRÞbecause the strategy is closer to the ideal benefitpoint than the ideal risk point. This zone requiresfull use of the firm’s abilities and resources.

Total Euclidean Distance from the Ideal Benefit (Bm)

Tot

al E

uclid

ean

Dis

tanc

e fr

om t

he I

deal

Ris

k (R

m)

Challenge Zone(Strong Benefits & Strong Risks)

(_

BB m ≤ ) and (_

RR m > > >

>

)

Desperation Zone(Weak Benefits & Strong Risks)

(_

BB m ) and (_

RRm )

Exploitation Zone(Strong Benefits & Weak Risks)

(_

BB m ≤ ) and (_

RRm ≤ )

Discretion Zone(Weak Benefits & Weak Risks)

(_

BB m ) and (_

RRm ≤ )

( **,RB )−

B

−R

Figure 1. The four quadrants and their characteristics.



Discretion Zone: In this quadrant, benefits areweak ðBm > %BBÞ and risks are weak ðRm4 %RRÞ sincethe strategy is closer to the ideal risk than theideal benefit. This zone represents the area wherethe firm has freedom or power to act or judge on itsown.

Desperation Zone: In this quadrant, benefits areweak ðBm > %BBÞ and risks are strong ðRm > %RRÞbecause the strategy is far from both the idealbenefit and risk points. Strategies falling into thiszone should be undertaken as a last resort.

The value determination procedure has beenspecified so that an alternative that has the mostpreferred score on both dimensions will have avalue of zero for ðBmÞ and ðRmÞ. Similarly, analternative that has the least preferred score onthese two dimensions will have a value of 1 forðBmÞ and ðRmÞ. Typically, there will not be anyactual alternatives that are this good or bad, butthe value number for the actual alternatives can beinterpreted by comparing these actual alternativeswith the ideal alternative having values of zero forðBmÞ and ðRmÞ. The value number for a particularalternative gives the proportion of the distance, ina value sense, that an alternative is from the idealalternative. In cases where more than one strategyfalls into one of these quadrants, the one with thelowest overall Euclidean distance ðDmÞ is thepreferred strategy.

Consider hypothetical alternative A withBA=0.40 and RA=0.30. No specific meaning canbe given to value numbers without considering theranges of the evaluation measures that are beingused ð04Bm41 and 04Rm41Þ. In this context,the value of 0.40 means that alternative A is 40%away from the ideal benefit and 60% away fromthe alternative with the worst possible benefit.Similarly, the value of 0.30 means that alternativeA is 30% away from the ideal risk and 70% awayfrom the alternative with the worst possible risk.Equation (3) calculates the overall Euclideandistance of alternative A from the ideal alternative

ðDA ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffið0:40Þ2 þ ð0:30Þ2

q¼ 0:50Þ. The overall dis-

tance of alternative A from the worst possiblealternative is ð

ffiffiffi2

p� 0:50Þ which is 0.92. This

means, alternative A is relatively closer to the idealalternative than the worst possible alternative.

Once the model is developed, sensitivity analysescan be performed to determine the impact on theranking of alternatives for changes in variousmodel assumptions. Some sensitivity analyses thatare usually of interest are on the weights and

probabilities of occurrence. The weights represent-ing the relative importance of the environments,benefits, and risks are occasionally a point fordiscussion among the various DMs. In addition,probabilities of occurrence which reflect the degreeof belief that an uncertain event will occur aresometimes a matter of contention.

3. THE MODEL

To formulate the model algebraically, let usassume:

Bm = Total Euclidean distance from the idealbenefit for the mth strategic alternative;(m=1,2, . . ., M)

Rm = Total Euclidean distance from the idealrisk for the mth strategic alternative;(m=1,2, . . ., M)

Dm = Overall Euclidean distance of the mthstrategic alternative; (m=1,2, . . ., M)

%BB = Mean Euclidean distance of the benefits%RR = Mean Euclidean distance of the risksWbi = The ith environment weight for benefits;

(i=1,2, and 3)Wri = The ith environment weight for risks;

(i=1,2, and 3)Fbij = The importance weight for the jth benefit

in the ith environment; (j=1,2, . . ., Jbi ; andi=1,2, and 3)

Frij = The importance weight for the jth risk inthe ith environment; (j=1,2, . . ., Jri ; andi=1, 2, and 3)

Pmbij

= Probability of occurrence of the jth benefitin the ith environment given the choice ofmth strategic alternative; (m=1,2, . . ., M;j=1,2, . . ., Jbi ; and i=1,2, and 3)

Pmrij

= Probability of occurrence of the jth risk inthe ith environment given the choice ofmth strategic alternative; (m=1, 2, . . ., M;j=1, 2, . . ., Jri ; and i=1, 2, and 3)

P�bij

= The ideal probability of occurrence of thejth benefit in the ith environment; (j=1, 2,. . ., Jbi ; and i=1, 2, and 3)

P�rij

= The ideal probability of occurrence of thejth risk in the ith environment; (j=1, 2, ...,Jri ; and i=1, 2, and 3)

Nbi = Number of benefits in the ith environment(i=1, 2, and 3)

Nri = Number of risks in the ith environment(i=1, 2, and 3)

M. TAVANA80


Assuming that i=1–3 represent the internal,transactional, and contextual environments.

Bm ¼X3i¼1

Wbi

XJbij¼1

Fbij P�bij� Pm

bij

� �2� � !ð1Þ

Rm ¼X3i¼1

Wri

XJrij¼1

Frij P�rij� Pm

rij

� �2� � !ð2Þ

Dm ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiBm2 þ Rm2

pð3Þ

B ¼XMm¼1

Bm=M ð4Þ

R ¼XMm¼1

Rm=M ð5Þ

where

P�bij

¼ Maxm

Pmbij

P�rij¼ Min

mPmrij

X3i¼1

Wbi ¼ 1

X3i¼1

Wri ¼ 1

XJbij¼1

Fbij ¼ 1

XJrij¼1

Frij ¼ 1

04Pmbij41

04Pmrij41

4. A PRACTICAL APPLICATION

From its inception in the early 1950s as adesigner and manufacturer of custom machinery,Semicony has become one of the leading US

suppliers of assembly equipment to the interna-tional semiconductor industry. Over the past 15years the semiconductor industry has undergonemajor structural changes that have had profoundconsequences for Semicon. In the mid 1970s USproducers of semiconductors controlled about65% of the world market while Japan controlledonly 25%. Today the positions of both countries’companies have almost reversed. Semicon’s man-agement has decided to expand its presence inJapan from a sales and service office located inTokyo to one of the four alternatives listed below(step (i)):

Alternative A}increase sales presence in Japan:Currently, Semicon has a sales and service office inTokyo with eight salespeople covering the majorsemiconductor houses, most of which are locatedin Tokyo. While their presence in front ofcustomers’ purchasing and corporate personnelhas been acceptable, it is the process engineers andmanufacturing people in the outlying factorieswho form the basic support for any equipmentacquisitions. These factories are usually located inother parts of Japan, and the management feelsthat their presence needs to be increased at theselocations. The challenge would be in hiring topquality Japanese personnel to fill these positions. Itis still difficult for an American company to recruittop Japanese university graduates and industrypeople. Most of them either work for, or want towork for, the top Japanese companies.

Alternative B}design a machine especially forJapanese customers: One reason for the smallmarket share of Semicon in Japan is the lack ofappropriate machinery. A development projectcould be undertaken to design and manufacture amachine specifically for the Japanese market. Anew machine designed correctly would competesuccessfully with Oshinaway in the most importantareas in Japan. Consequently, R&D expenseswould need to be increased beyond their alreadyhistorically high level, or other projects critical tosupporting core markets would need to be scaledback. This would be a major project in terms ofengineering manpower as well as cost.

Alternative C}jointly manufacture existingequipment in Japan: Japanese customers’ confi-dence in a supplier is directly proportional to thesupplier’s presence in Japan. The same personnelproblems as those discussed in increasing the salesforce would be encountered, only on a much largerscale. A joint venture with a Japanese manufac-turer of precision equipment could eliminate the

yThe name of the company has been changed to protect itsanonymity.



personnel issues, and would allow a much fasterand less costly implementation curve. In return forsharing the benefits of this enterprise with anothercompany, Semicon would have a good chance ofputting pressure on Oshinawa in its core market,something the company has been unable to do.Putting Oshinawa in a defensive position wouldcause it to allocate more of its resources to defendits base and would leave Semicon a larger share ofthe non-Japan market. In addition, the increaseOshinawa would experience in its operating costswould put upward pressure on its prices and wouldmake Semicon even more competitive. Finally, bymanufacturing and selling in Japan, most of therisk from exchange rate fluctuations would beeliminated.

Alternative D}Drop prices: Lower base pricescoupled with the weaker dollar would allowJapanese customers to view Semicon equipmentas superior in price/performance analyses. If theJapanese economy does not improve soon, puttingsuch pressure on in the short term may haveenough of a negative impact on Oshinawa to causethem serious problems for several years. In themeantime, Semicon could be making inroads inJapan. The obvious concern here is that such aprice action will be a drag on margins, and may setpricing precedents that will be difficult to change inthe future. Also, the Japanese may not desertOshinawa solely because of lower prices unlessSemicon’s price/performance ratio is at least anorder of magnitude better.

In step (ii), the DM identified the benefits andrisks for the internal, transactional, and contextualenvironments. They are presented in Table II.

4.1. Internal benefitsABP}Automation of business processes: In

order to become more competitive on a coststructure basis, it is imperative that Semiconupgrade and consolidate their order processingand manufacturing systems.

ITK}increase technical knowledge base: Anincreased presence in Japan, in any of the formsdiscussed above, would provide a better under-standing of Japanese customer needs. By doing abetter job of designing solutions into theirmachines, Semicon would be providing unsur-passed customer service while gaining valuabletechnical knowledge. Increased technical knowl-edge for Semicon could translate into greatermarket share and margins.

RQI}Reaching 585% quality index level infactory: Japanese customer requirements coulddrive Semicon to greater levels of quality in theirequipment. One good measure of quality shippedout the door is the quality index level used bySemicon. Although the current 75% level is a bigimprovement from past performance, higher levelswould be needed to compete directly with Oshi-nawa in their home market. It is a customerrequirement that a machine be uncrated and run‘‘out of the box’’ so that machine utilization ismaximized.

OTD}Reaching actual delivery against commit-ment of � 3 days: Another area Semicon wouldneed to improve in order to gain market share inJapan is meeting delivery commitments. Japanesemarkets are volatile, and it is difficult to forecastspecific product mixes and shipment quantities fora particular period of time with a high degree ofaccuracy. However, to compete in Japan, Semiconwould need to assure that delivery occurs within 3days of the committed delivery.

IES}Improve employee skills: All of the pre-viously stated objectives require a concerted effortto increase the skill levels of Semicon’s employees.Engineering, manufacturing, marketing, and salespeople will not only need to know how to improvethe company’s performance, but will also need athorough understanding of the Japanese environ-ment. Japan has economic, cultural, and socialdimensions that are diverse and must be addressedif Semicon is to be successful.

4.2. Transactional benefitsICR}Improve company reputation: If Semicon

can compete successfully in Japan, it will haveaccomplished something that very few non-Japa-nese companies have done. Success is really not anoption for Semicon. Success in the Japanesemarkets would bring a special status that, whileintangible, would be very valuable in othermarkets. Semicon would no longer suffer theperception problems it now encounters with someof its other customers who use Japanese equip-ment. Neutralizing some of these quality andcustomer service issues could translate directlyinto increased sales.

ICS}Improve competitive cost structure: This isan important competitive issue. Re-engineeringprocesses and systems in the company will have apositive effect on costs. Manufacturing costswould be most directly affected, while R&D costswould be more dependent upon the marketing

M. TAVANA82


decisions made by Semicon. Even so, there wouldbe some benefit in this area just from being able toaccess the automated and integrated systems inother parts of the company. If Semicon can reduceits costs and approach Oshinawa’s cost structure,price competitiveness would improve, especially inJapan.

PCB}Protect core business: A Semicon assaulton the Japanese market, Oshinawa’s core business,with high quality, price competitive products,would require Oshinawa to enact defensive coun-termeasures. These measures require Oshinawa tomake significant expenditures and to reallocatesome of their resources, particularly in sales,engineering, and marketing. If Oshinawa tries todefend their positions in both the US and Japanmarkets simultaneously, their current cost andresource problems would most likely be exacer-bated further.

KJC}Key Japanese components more availableto semicon: As Semicon increases its presence inJapan, it would not only learn more about itscustomers, but it would also have the ability to dobusiness with leading Japanese vendors of keyprecision machine components. In some areas, theJapanese offer superior technology than thatavailable in the US. Incorporating Japanesecomponents in these instances would improvemachine performance and reliability.

4.3. Contextual benefitsSGL}Special government legislation to help

semiconductor trade: For the past decade, tradebetween the US and Japan has been a source ofacrimony between both countries. The US claimsthat Japan’s huge trade surplus is due, in largepart, to non-tariff trade barriers that createdvirtually closed markets. Japan claims that the

Table II. Environmental benefits and risks

Benefits

Internal environment

ABP Automation of business processes

ITK Increase technical knowledge base

RQI Reaching 85% quality index level in factory

OTD Reaching actual delivery against commitment of� 3 days

IES Improve employee skills

Transactional environment

ICR Improve company reputation

ICS Improve competitive cost structure

PCB Protect core business

KJC Key Japanese components more available to Semicon

Contextual environment

SGL Special government legislation to help semiconductor trade

FER Falling exchange rate

JER Japanese economic rebound within next 2–3 years

Risks

Internal environment

IRD Increase In R&D expenses

LPM Lower prices and margins

RTW Resistance to increased workloads required to compete

HET Higher employee training and educational expenses

Transactional environment

JPI Japanese personnel hiring and training issues

HCB High cost of doing business in Japan

CCS Competitor’s counter strategies could cause pain

LOF Loss of focus on core business

Contextual environment

BTE High barriers to entry

JCD Japanese culture is significantly different than US

CJR Continued Japan recession would hurt prospects

PTF Possible trade friction between the US and Japan



US’ problems are due mostly to its huge budgetdeficits that ultimately cause a reduction incompetitiveness. However, the US has been ableto get Japan to agree to specific targets for sales ofUS semiconductors in Japan. While the merits ofthis agreement could be debated in a larger,macro-economic context, further developments ofthis nature would be favorable to Semicon. Just asthe US equipment industry followed the USsemiconductor industry in lost market share, it isreasonable to expect that the reverse could occur.

FER}Falling exchange rate: Whether the dollarwill continue to weaken against the yen remains tobe seen. A weaker dollar allows the Japanese tospend fewer yen than they would otherwise haveused to purchase Semicon’s equipment. Thiswould allow Semicon to reduce prices and putfurther pressure on Oshinawa.

JER}Japanese economic rebound within next2–3 years: Along with the current competitivedisadvantages Semicon faces in Japan, the weakJapanese economy has also had negative impacton Semicon’s business prospects. The managementbelieves that this should change radically in thenear future. Japanese companies are movingtowards restructuring through layoffs, early retire-ment, etc. In addition, capital spending, which hasalso been curtailed for replacement equipment,should rise as it becomes impossible to producenew, complex semiconductors on older machines.

4.4. Internal risksIRD}Increase in R&D expenses: Management

believes it is very important for Semicon tocontinue their leadership in the US market. Thisis likely to require an increase in R&D spending.

LPM}Lower prices and margins: In Japan,prices for the same type of equipment as that soldin the rest of the world tend to be lower. This isdue to the relationship Japanese customers havewith their vendors, and manufacturers have withtheir subcontractors. In return for getting a lowerprice, a vendor is assured of a long-term relation-ship with a customer. A major foray into Japanmeans that Semicon will see its current marginserode to some extent. If costs are kept under tightcontrol, then a 3–5% reduction in gross margincan be expected, otherwise a 5–9% reduction ispossible. This is a significant risk because healthymargins are required to fund the development oftechnologically sophisticated products.

RTW}Resistance to increased workloads re-quired to compete: A major increase in activity

across most of the company will be required tosuccessfully expand Semicon’s market share inJapan. Management will be challenged to motivateits employees to not only accept this large job, butalso to perform at the high level required forsuccess.

HET}Higher employee training and educationalexpenses: It is obvious that increasing employeeskills does not come without a cost. Althoughincreasing the skills needed to succeed in Japan isessential, the costs will have to be considered whenevaluating Semicon’s cost structure versus Oshi-nawa’s. Again, some trade-offs may be required.

4.5. Transactional risksJPI}Japanese personnel hiring and training

issues: Japanese personnel would be needed tomanage and run some of Semicon’s operations inJapan. The goal of any Japanese joining the laborforce is to join a top Japanese firm. This is not justan economic desire but also one of great pride.Consequently, top Japanese companies have ac-cess to the best and brightest people in the laborforce. It would be very difficult for Semicon torecruit and hire good Japanese managers andworkers.

HCB}High cost of doing business in Japan: It iswell known that Tokyo, and to a lesser extent therest of Japan, is one of the most expensive placesto live and work in the world. The salaries ofJapanese workers would not be prohibitive.However, the cost of expanding in Japan wouldbe felt primarily in additional rents and leases, andin any costs incurred by US expatriates living inJapan.

CCS}Competitor’s counter strategies couldcause pain: Semicon’s management believes thatthere is a possibility that Oshinawa and its workerscould make the sacrifices necessary to keep costs inline as they confront the need to cut their prices.Oshinawa might even cut its prices to the pointwhere Semicon would find its margins eroding toan unacceptable degree. Any number of otheractions Oshinawa could take including enhancedproducts, more sales coverage or more attractiveterms would hurt Semicon’s bottom line.

LOF}Loss of focus on core business: In pursu-ing the Japanese market, Semicon would need tobe very careful not to neglect its US market. Eventhough succeeding in Japan is essential to longterm growth, losing dominance in the US marketcould have harmful consequences.

M. TAVANA84


4.6. Contextual risksBTE}High barriers to entry: Much of the trade

friction that Japan encounters with other countriesis a consequence of its non-tariff barriers. Forexample, the separate standards and regulationsthat foreign companies must meet for access toJapan’s consumers are more onerous than thosefaced by Japanese companies. Another example isthat success, in many cases, requires a workingknowledge of, and access to, Japan’s Byzantinedistribution networks. In Semicon’s case, equip-ment evaluation requirements that are longer andmore stringent than those that are required ofOshinawa could be a problem. These barriers arecertainly areas of risk that will add cost toimplementing the strategy.

JCD}Japanese culture is significantly differentthan US: One problem that will need to beaddressed in a major expansion effort in Japan ishow to run an operation, negotiate contracts, anddeal with employees playing under rules that areoften unfamiliar. An important part of Japanesebusiness culture is that a solid relationship andlocal presence must be established before anyoutside vendor can win large orders.

CJR}Continued Japan recession would hurtprospects: A return on the large investment inJapan will not be possible until Japan’s economybegins to recover. The key is to time theimplementation in Japan to occur as closely aspossible to the end of the recession. This may bedifficult to time exactly, but a gross error couldseriously harm the company.

PTF}Possible trade friction between the US andJapan: There is always the possibility that a smalltrade skirmish could escalate into a trade war. Thisis not unlikely given the often contentious relation-ship between the two countries. If this occurs,retaliatory measures by the Japanese could makeSemicon equipment cost prohibitive in Japan.Obviously, this is detrimental to achieving thestrategic objectives specified above.

In step (iii), the DM estimated the relativeimportance of the internal, transactional andcontextual environments through a series ofpairwise comparisons required by AHP. UsingExpert Choice (1990), an AHP-based software, theDM provided the pairwise comparisons presentedin the environmental comparisons section ofTables III and IV. Expert Choice synthesizedthese judgments and provided the DM with therelative weights and the inconsistency ratios thatare also presented in Tables III and IV. The DM

was asked to re-evaluate his judgments if theinconsistency ratio was greater than 10%. It isclear that for the Benefits, the internal environ-ment, with a weight of 0.659, outweighs otherenvironments; for the Risks, the transactionalenvironment is most important (0.696).

Expert Choice is used again by the DM in stepiv to evaluate all the sets of environmental benefitsand risks. For example, 10 pairwise comparisonswere made between the internal benefits (ABP,ITK, RQI, OTD and IES) presented in Table III.Expert Choice used these judgments to estimatethe relative weight associated with each internalbenefit. If the inconsistency ratio was greater than10%, the DM was asked to reconsider theirjudgments in the pairwise comparison matrix.One pairwise comparison matrix is needed foreach set of benefits and risks in the internal,transactional and contextual environments. There-fore, the process of entering the judgments wasrepeated by the DM six times. All mathematicalmanipulations were performed by Expert Choice.The pairwise comparisons and weights for benefitsare given in Table III and for the risks in Table IV.

Step (v) requires the DM to estimate theprobability of occurrence for each benefit and riskunder each potential alternative. These probabil-ities represent numerically the expectation of afactor’s occurring. Using the verbal probabilityscale in Table I, the DM estimated the occurrenceprobabilities presented in Table V for all thebenefits and risks associated with each alternative.For example, the DM estimated a 90% chance ofreaching 85% quality index level (RQI) if a newmachine is designed (Alternative B). However, theDM expects only 20% chance of achieving thistarget if prices are dropped (Alternative D).

In step (vi), the ideal probabilities presented inTable V were identified for each benefit and riskfrom the subjective probabilities within thetable. For example, the ideal probability for RQIbenefit (reaching 85% quality index level infactory) is 90% because it is the best (highest)among the set (70%, 90%, 20%, and 20%).However, the ideal probability for LPM risk(lower prices and margins) is 20% since it is thebest (lowest) among the set (20%, 50%, 70%, and60%).

The last step (step (vii)) is to calculate the totalEuclidean distance of each strategic alternativefrom the ideal benefit and the ideal risk. Thespreadsheet in Table VI was used to perform theoperations specified in equations (1) and (2).



After calculating the total Euclidean distancesof each alternative from the ideal alternative, themean Euclidean distance of the benefits ð %BB ¼0:162Þ and risks ð %RR ¼ 0:108Þ were determined andplotted on a graph. The placement of the twomeans on the graph results in the formationof the four quadrants. Strategic choices with Bm

40:162 and Rm40:108 fall in the exploitationzone while those with Bm40:162 and Rm > 0:108fall in the challenge zone. Furthermore,strategic choices with Bm > 0:162 and Rm40:108fall in the discretion zone and those with Bm >0:162 and Rm > 0:108 fall in the desperationzone. As it is shown in Figure 2, alternativeA ðBA ¼ 0:023; RA ¼ 0:027Þ with strong benefitsand weak risks is the most preferred alternativefalling in the exploitation zone. Alternative C ðBC

¼ 0:267; RC ¼ 0:159Þ with weak benefits andstrong risks is the least preferred alternative fallingin the desperation zone. Alternatives B and D fallin the challenge zone and discretion zone respec-tively.

5. FROM QSPM TO EUCLID

Solving MCDM problems is not searching forsome kind of optimal solution, but rather helpingDMs master the (often complex) data involved intheir problem and advance toward a solution(Roy, 1990). As often happens in applied mathe-matics, the development of multicriteria models isdictated by real-life problems. It is therefore notsurprising that methods have appeared in a ratherdiffuse way, without any clear general methodol-ogy or basic theory (Vincke, 1992). Euclid wasdeveloped after attempting to address a real-lifestrategic decision making problem using QSPM, aMCDM technique similar to the multiobjectivevalue analysis proposed by Keeney (1992) andKirkwood (1997). We found the results fromQSPM to be unsatisfactory, then, our search fora better approach resulted in Euclid.

The author was hired as an external consultantto assist the strategic assessment committee ofSemicon select one of the four previously defined

Table III. Pairwise comparison matrices (benefits)

Environmental comparisons

Inconsistency ratio=0.030

Internal Transactional Contextual Relative weight

Internal 1 3 7 0.659

Transactional 1/3 1 4 0.263

Contextual 1/7 14

1 0.078

Internal factors


ABP ITK RQI OTD IES Relative weight

ABP 1 1/5 1/9 1/7 1/5 0.033

ITK 5 1 1/5 1/4 1/3 0.088

RQI 9 5 1 2 4 0.452

OTD 7 4 12

1 2 0.267

IES 5 3 14

1/2 1 0.160

Transactional factors


ICR ICS PCB KJC Relative weight

ICR 1 2 4 9 0.501

ICS 1/2 1 3 7 0.310

PCB 1/4 1/3 1 6 0.149

KJC 1/9 1/7 1/6 1 0.040

Contextual factors


SGL FER JER Relative weight

SGL 1 1/2 9 0.374

FER 2 1 8 0.571

JER 1/9 1/8 1 0.055

M. TAVANA86


alternatives. The committee was appointed by theboard of directors and comprised of twelvemembers, mostly senior managers with graduateeducation in business and engineering. The boardinvited the committee to work with the facilitatorat a three-day decision conference outside thecorporate settings and apart from daily interrup-tions and distractions. Decision conferencing isgeared towards on-the-spot development of com-puter based decision models that incorporates thediffering perspective of participants (McCartt andRohrbaugh, 1989). The committee, with theauthor acting as facilitator and analyst, wereexpected to build a multicriteria decision modelsimilar to QSPM and examine the implications ofthe model.

The assessment process began with a brain-storming session focusing on the question ‘‘whatfactors should be used in deciding whether toincrease sales presence in Japan, design a machineespecially for Japanese customers, jointly manu-facture existing equipment in Japan, or drop

prices?’’ Moore (1994) suggests several approacheswhich could be used to facilitate this process. Thecommittee chose the flip-chart approach in whichthe participants listed their decision factors alongwith a brief justification on separate sheets of flip-chart paper. Once everyone completed this ex-ercise, all flip-chart papers were posted for every-one to see. With the help of the facilitator, thecommittee discussed these factors and constructeda list on which they agreed. As suggested byKeeney (1992), the list was examined to assure thechosen factors were complete, operational, non-redundant, concise, measurable, understandable,and relevant.

As the conference continued, the committee wasinstructed to classify the decision factors intoexternal opportunities/threats and internalstrengths/weaknesses according to QSPM. Duringthis classification process, participants were con-fused about the semi-controllable factors. Themain difficulty was whether they should beincluded in the internal or external factors. The

Table IV. Pairwise comparison matrices (risks)

Environmental comparisons


Internal Transactional Contextual Relative weight

Internal 1 1/4 4 0.229

Transactional 4 1 7 0.696

Contextual 1/4 1/7 1 0.075

Internal factors


IRD LPM RTW HET Relative weight

IRD 1 1/2 2 7 0.278

LPM 2 1 5 8 0.529

RTW 1/2 1/5 1 5 0.149

HET 1/7 1/8 1/5 1 0.044

Transactional factors


JPI HCB CCS LOF Relative weight

JPI 1 1 1/8 1/5 0.066

HCB 1 1 1/8 1/5 0.066

CCS 8 8 1 2 0.557

LOF 5 5 1/2 1 0.311

Contextual factors


BTE JCD CJR PTF Relative weight

BTE 1 1 1/7 5 0.134

JCD 1 1 1/7 5 0.134

CJR 7 7 1 9 0.692

PTF 1/5 1/5 1/9 1 0.040



facilitator suggested a two-level classificationschema. The first level consisted of benefits(maximizing criteria) and risks (minimizing criter-ia). The second level consisted of internal (con-trollable), transactional (semi-controllable), andcontextual (uncontrollable) factors. The commit-tee agreed on the new hierarchy and six groups ofdecision factors were created: internal, transac-tional, and contextual benefits and internal,transactional, and contextual risks.

The second day, the committee members beganby assigning importance weights to their decisionfactors. During this weight assignment, someparticipants argued for a more scientific weightelicitation procedure. The facilitator suggestedother techniques such as paired comparisons,trade-off analysis, and regression estimates assuggested by Kleindorfer et al., (1993). After

discussing the advantages and disadvantages ofeach method, participants agreed on using pairedcomparisons and AHP to determine the impor-tance weight of their decision factors. ExpertChoice (1990), an AHP-based software wasdemonstrated to the participants. After a briefintroductory session and a couple of hands-onexercises, Expert Choice was used to capturecommittee member’s judgments.

The standard procedure in QSPM required thedetermination of the attractiveness scores.QSPM defines an attractiveness score as anumerical value that indicates the relativeattractiveness of a criterion for a given alternative.The range for attractiveness scores in QSPM is1=not attractive to 4=highly attractive. Most ofthe participants thought this limited scoringrange was unrealistic. Similarly, they perceived as

Table V. Probabilities of occurrence and ideal probabilities

Factors Increase Design Joint venture to Drop Ideal

sales force (A) new machine (B) manufacture (C) prices (D) probability

Benefits

Internal ABP 0.90 0.50 0.50 0.50 0.90

ITK 0.50 0.70 0.70 0.20 0.70

RQI 0.70 0.90 0.20 0.20 0.90

OTD 0.90 0.70 0.20 0.20 0.90

IES 0.60 0.70 0.40 0.20 0.70

Transactional ICR 0.90 0.80 0.60 0.30 0.90

ICS 0.60 0.40 0.40 0.50 0.60

PCB 0.80 0.80 0.50 0.50 0.80

KJC 0.20 0.60 0.90 0.10 0.90

Contextual SGL 0.40 0.50 0.30 0.30 0.50

FER 0.40 0.60 0.40 0.20 0.60

JER 0.50 0.60 0.50 0.60 0.60

Risks

Internal IRD 0.40 0.90 0.40 0.10 0.10

LPM 0.20 0.50 0.70 0.60 0.20

RTW 0.40 0.60 0.60 0.30 0.30

HET 0.60 0.70 0.40 0.30 0.30

Transactional JPI 0.60 0.70 0.70 0.10 0.10

HCB 0.30 0.30 0.90 0.10 0.10

CCS 0.30 0.70 0.70 0.70 0.30

LOF 0.50 0.70 0.50 0.40 0.40

Contextual BTE 0.60 0.70 0.70 0.10 0.10

JCD 0.30 0.30 0.90 0.10 0.10

CJR 0.70 0.70 0.50 0.70 0.50

PTF 0.50 0.70 0.50 0.40 0.40

M. TAVANA88


unreasonable the constraint that if a factor has noeffect upon an alternative, then no attractivenessscore can be assigned to the remaining alterna-tives. In addition, the committee felt morecomfortable using probabilities of occurrencebecause the factors were defined as events.

The third day, the importance weights andthe probabilities of occurrence were aggregatedfor each participant using a weighted summethod. Once the committee members werepresented with their overall scores, they questionedthe meaning of the data in the absence of areference point. This motivated the facilitator todevelop the ideal alternative (Schelling, 1960;Festinger, 1964; Zeleny, 1974, 1982). Using

the ideal alternative and Euclidean distance,participants were able to visualize their choiceson a two-dimensional graph and find theirpreference order for the alternatives. Once theindividual rankings were identified, MAH wasused to identify a consensus ranking of thealternatives for all the committee members.MAH forms consensus orderings that reflectcollective DM agreement given ordinal or cardinalrankings. Normative averaging procedurescould have been used to combine individualrankings. However, normative aggregationprocedures (see, e.g., Aczel and Saaty, 1983; Steeband Johnson, 1981) provide little stimulus for theexchange of opinions. Additionally, MAH was

Table VI. Overall analysis

Alternatives

Environmental

weight

Criteria

weight

A B C D Ideal

alternative

Benefits

Internal environment 0.659

ABP 0.033 0.90 0.50 0.50 0.50 0.90

ITK 0.088 0.50 0.70 0.70 0.20 0.70

RQI 0.452 0.70 0.90 0.20 0.20 0.90

OTD 0.267 0.90 0.70 0.20 0.20 0.90

IES 0.160 0.60 0.70 0.40 0.20 0.70

Transactional environment 0.263

ICR 0.501 0.90 0.80 0.60 0.30 0.90

ICS 0.310 0.60 0.40 0.40 0.50 0.60

PCB 0.149 0.80 0.80 0.50 0.50 0.80

KJC 0.040 0.20 0.60 0.90 0.10 0.90

Contextual environment 0.078

SGL 0.374 0.40 0.50 0.30 0.30 0.50

FER 0.571 0.40 0.60 0.40 0.20 0.60

JER 0.055 0.50 0.60 0.50 0.60 0.60

Total Euclidean Distance from the Ideal Benefit (Bm) 0.023 0.016 0.267 0.343 0.000

Risks

Internal environment 0.229

IRD 0.278 0.40 0.90 0.40 0.10 0.10

LPM 0.529 0.20 0.50 0.70 0.60 0.20

RTW 0.149 0.40 0.60 0.60 0.30 0.30

HET 0.044 0.60 0.70 0.40 0.30 0.30

Transactional environment 0.696

JPI 0.066 0.60 0.70 0.70 0.10 0.10

HCB 0.066 0.30 0.30 0.90 0.10 0.10

CCS 0.557 0.30 0.70 0.70 0.70 0.30

LOF 0.311 0.50 0.70 0.50 0.40 0.40

Contextual environment 0.075

BTE 0.134 0.60 0.70 0.70 0.10 0.10

JCD 0.134 0.30 0.30 0.90 0.10 0.10

CJR 0.692 0.70 0.70 0.50 0.70 0.50

PTF 0.040 0.50 0.70 0.50 0.40 0.40

Total Euclidean Distance from the Ideal Risk (Rm) 0.027 0.163 0.159 0.083 0.000



selected as the consensus ranking procedurebecause of its simplicity, flexibility, and generalperformance.

6. MANAGERIAL IMPLICATIONS

A strategic assessment committee is responsiblefor the evaluation of strategic alternatives atSemicon. This committee is comprised of 12members, most of whom are from senior manage-ment. The individual preferences from Euclid canbe combined into group preferences using severalapproaches (Barzilai et al., 1986; Blin and Whin-ston, 1974; Bowman and Calantoni 1973; Cookand Seiford, 1978). Cook and Kress (1985)proposed a network model for deriving theoptimal consensus ranking that minimizes dis-agreement among a group of DMs. Ali et al.(1986) presented an integer programming ap-proach to consensus ranking. While these techni-ques are complex, Beck and Lin (1983) havedeveloped a simple analytical procedure calledmaximize agreement heuristic (MAH) to arrive ata consensus ranking that maximizes agreement

among DMs. At Semicon, MAH was used inconjunction with an intuitive graphical approach.

Each committee member was given three reportsin a graphical format depicting the individual’sown preferences and both a summary and thedetails of the group preferences. The individualfeedback shows how each alternative is positionedwithin the strategic zones based on the DM’sassessment of its benefits and risks. Figure 2 showsthe total Euclidean distance of the four alterna-tives from the ideal benefit and risk for committeemember 1.

The committee members also received tworeports representing the group’s preferences. Thesummarized group feedback in Figure 3 shows thegroup’s overall mean total Euclidean distancefrom the ideal benefit and risk for the fouralternatives. Based on the group’s assessments,alternative A falls in the exploitation zone, B fallsin the challenge zone, C falls in the desperationzone, and alternative D falls in the discretion zone.

The detailed group feedback in Figure 4 showsthe position of each committee member’s prefer-ences compared to the group’s overall mean. Forexample, committee member 1’s preferences place

CHALLENGE ZONE

DESPERATION ZONE

EXPLOITATION ZONE

DISCRETION

ZONE

0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

Total Euclidean Distance from the Ideal Opportunity (Um)

A (0.023,0.027)

MEAN (0.162,0.108)

IDEAL (0.00,0.00)

0.05

0.10

0.15

0.20

To

tal E

ucl

idea

n D

ista

nce

fro

m t

he

Idea

l Th

reat

( T

m)

B (0.016,0.163) C (0.267,0.159)

D (0.343,0.083)

Figure 2. The four strategic alternatives (committee member 1).

M. TAVANA90


alternative A in the exploitation zone, alternativeB in the challenge zone, and both alternatives Cand D in the desperation zone. Figure 4 alsoreveals 11 occurrences of alternative A in theexploitation zone, eight occurrences of alternativeB in the challenge zone, five occurrences ofalternative C in the desperation zone, and nineoccurrences of alternative D in the discretion zone.

In order to develop a consensus ranking of allalternatives with MAH, the overall Euclideandistance of each alternative (Dm) was calculatedfor all 12 committee members. The individualcommittee member rankings along with the con-sensus ranking of the committee are presented inTable VIII. The MAH ranking for the committeeis A>B>C>D.

After receiving the graphical and analyticalfeedback in Figures 2–4 and Table VIII, thecommittee met to finalize the decision. In thismeeting, members were encouraged to sharetheir views with other members of the committee.Given the feedback provided by Euclid, thestrategic assessment committee at Semicon unan-imously selected A>B>C>D as their order ofpreference.

Prior to implementation of Euclid, QSPM wasused to evaluate the strategic alternatives at

Semicon. Although QSPM made significant con-tributions to the strategic evaluation process,increasing environmental complexities forced thecommittee to look for a similar but morecomprehensive approach. Semicon needed anapproach that combined the analytical and in-tuitive aspects, that was simple yet comprehensiveand that was flexible enough to provide structurein a dynamic environment. Euclid was designed toprovide these capabilities while allowing visualcommunication of output to the DMs andpermitting them to perform sensitivity analysis.

A 10-item questionnaire was pre-tested andutilized to measure each committee member’sperception of QSPM. This questionnaire con-tained a definition for each of the following tenattributes: analytical, intuitiveness, simplicity,comprehensiveness, flexibility, structured, visual,sensitivity analysis, decision process quality, anddecision quality. The questionnaire employs aseven-point Likert scale with 1 indicating strongdisagreement, 4 indifference, and 7 strong agree-ment (Lewis and Butler, 1993). The same ques-tionnaire was given to the committee membersafter the implementation of Euclid. Table VIIpresents statistics describing the results of thequestionnaires.

A D

C

OVERALL MEAN (0.085,0.074)

B

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16


To

tal E

ucl

idea

n D

ista

nce

fro

m t

he

Idea

l Ris

k (R

m)

Figure 3. The Four strategic alternatives (average for all committee members).



One-sided Wilcoxon signed ranks test withn=12 subjects and a ¼ 0:05 was used to comparethe committee ratings. As it is shown in Table VII,the ratings of Euclid on the analytical, intuitive-ness, comprehensiveness, visual, decision processquality, and decision quality were significantlygreater than the QSPM ratings. There was nostatistical difference between Euclid and QSPMratings for simplicity, flexibility, structured, andsensitivity analysis at 0.05 significance level. Thecommittee concluded that Euclid had significantlyenhanced the decision process and their confidencein decision.

The common wisdom is that good decisionsresult in good outcomes, and poor decisionsresult in poor outcomes. A questionnaire was usedto assess each committee member’s perception ofthe process and outcome in QSPM and Euclid. Theuse of the questionnaire approach has beencriticized by Eden (1995) and therefore should beapplied carefully. Wright and Rohrbaugh (1999)critique various approaches to the evaluation ofdecision conferencing and suggest that evaluationsshould focus on processes rather than outcomes,

address the group rather than individual roles andbehaviors, and view the group in organizationalcontext rather than in isolation. The competingvalues approach to organizational analysis can beused to enhance the quality outcome of groupdecision making (McCartt and Rohrbaugh, 1989;Reagan and Rohrbaugh, 1990). Ideally, thedecision process and quality of Euclid could becompared with various competing methods. How-ever, realistically, such an experiment requires aconsiderable amount of time and effort by a groupof busy executives attending a 3-day decisionconference. In sum, it should be noted that it ispossible for a most unreasonable method to belinked over time with a windfall, while in anotherinstance, for a most reasonable method to fall farwide of the mark.

7. CONCLUSION

Global competition, data availability, and ad-vances in computer technology have made strategicdecision making more complex than ever. Euclid

D3

A6

A10D8

D7

C10B9

D12

MEAN (0.085,0.074)

A1

A9

C9B12

C7

C1

B7B1

B5

B4

C5

B10

A3

A11

A8

C2

B2

B8

D11

A12

C11

A5

A4C3

D2

D6

C6D5

D10

C4D1

C8

B11

B6

A7

B3

C12

A2

D4

D9

0.000.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

0.24

0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36 0.40


To

tal E

ucl

idea

n D

ista

nce

fro

m t

he

Idea

l Ris

k (R

m)

Figure 4. The four strategic alternatives (committee members 1–12).

M. TAVANA92


uses environmental scanning, AHP, subjectiveprobabilities, and the theory of displaced ideal toreduce these complexities by decomposing thestrategy evaluation process into manageable steps.This decomposition is achieved without overlysimplifying the issues within the strategic decision.Previous strategic evaluation models tend to beeither intuitive or highly analytical. While intuitionis still favored by practicing managers, it may bedangerously unreliable when used to solve complexstrategic problems. On the other hand, highlyanalytical models are often avoided or misusedbecause of their technical difficulties. Euclid hasseveral attractive features that address some of thelimitations inherent in the existing models.

(i) Analytical: The value analysis model utilizedin Euclid is considered a multicriteria deci-sion making approach.

(ii) Intuitive: Euclid captures the intuitivepreferences of DMs by using methodssuch as AHP and subjective probabilityestimation.

(iii) Simple: Euclid has a simple scoring systemthat is easy to understand.

(iv) Comprehensive: Euclid processes a widerange of information concerning the benefitsand risks in the internal, transactional, andcontextual environments.

(v) Flexible: Euclid does not limit the number ofalternatives or environmental factors to beexamined.

(vi) Structured: Euclid decomposes the strategyevaluation process into manageable stepsand then integrates the results from each stepto arrive at an optimal solution.

(vii) Visual: Euclid allows DMs to visuallyexamine the benefits and risks of the

strategic alternatives by comparing themwith the ideal alternative.

(viii) Sensitivity Analysis: Euclid can be used in aninteractive mode to perform ‘‘goal seeking’’and ‘‘what-if’’ analyses.

Using a step-by-step and structured approachlike Euclid does not imply a deterministic ap-proach to strategic decision making. While Euclidenables DMs to crystallize their thoughts andorganize their beliefs, it should be used verycarefully. Managerial judgment is an integralcomponent of Euclid; therefore, the effectivenessof the model relies heavily on the DM’s cognitiveabilities to provide sound judgments. Euclidutilizes intuitive methods such as subjectiveestimation of probabilities and weights becausethey cannot be supported by empirical analysis.While these judgments often mirror a DM’s beliefin the importance of certain events, they should be

Table VII. Means and mean differences for euclid and QSPM

Attribute Euclid mean Euclid std. dev. QSPM mean QSPM std. dev. Mean difference Significant a ¼ 0:05

Analytical 6.17 0.72 2.67 0.78 3.50 Y

Intuitiveness 6.25 0.62 1.75 0.62 4.50 Y

Simplicity 4.08 1.00 4.92 0.90 -0.83 N

Comprehensiveness 6.17 0.83 1.83 0.83 4.33 Y

Flexibility 6.33 0.49 5.67 0.89 0.67 N

Structured 5.83 0.83 5.17 0.83 0.67 N

Visual 6.00 0.85 1.92 0.79 4.08 Y

Sensitivity analysis 5.92 0.90 5.17 1.11 0.75 N

Decision process quality 6.25 0.75 1.92 0.51 4.33 Y

Decision quality 6.58 0.51 1.75 0.45 4.83 Y

Table VIII. Individual and consensus ranking of alter-natives

Committee member Order of preferences

1 A>B>C>D

2 A>B>C>D

3 A>C>B>D

4 A>B>C>D

5 A>B>C>D

6 B>A>C>D

7 A>C>D>B

8 A>B>D>C

9 A>B>D>C

10 A>C>D>B

11 A>B>D>C

12 A>B>D>C

Committee A>B>C>D



used with caution. As with any decision calculusmodel, the researchers and practicing managersmust be aware of the limitations of subjectiveestimates. When empirical analysis is feasible andmakes economic sense, it should be utilized toimprove these estimates (Lodish, 1982). Euclidshould not be used to plug-in numbers and crank-out optimal solutions. Potentially, DMs couldmake poor judgments as they do with anyapproach. Such judgments can generate mislead-ing results and ultimately poor decisions.

Finally, practicing managers should be awarethat the practical application presented in thispaper suffers from the confusion between ‘‘means’’and ‘‘ends’’ objectives (Keeney, 1992). Value-focused thinking suggested by Keeney (1992) canbe used to eliminate this problem in futureapplications. Constructing objectives for decisionproblems is more of an art than a science andvalues play an important role in developingobjectives. When DMs express a preference orjustify an action, they typically refer to theirvalues. Objectives are specific expressions ofvalues. It is useful to distinguish between meansand ends objectives. Ends objectives are the onesthat a DM truly cares about. Ends objectives canbe identified by asking why a DM cares about astated objective. If the answer is ‘‘That is selfevident’’, it is an ends objective. If the answer is‘‘Because achieving this objective contributes toachieving another objective’’, it is a meansobjective. Means and ends objectives can berepresented by a means-ends network (Keeney,1992). The purposes of creating means-ends net-works are to clearly distinguish between meansand ends and to clarify their causal relationships.Keeney (1992) provides a more detailed list ofquestions that are useful to elicit objectives.Hammond et al. (1999) provide additional gui-dance on how to identify objectives. Means andends objectives can be distinguished in Euclid byconsidering the overall internal, transactional, andcontextual concerns as end objectives and thoseobjectives that contribute to each environmentalconcern can be considered as the mean objectives.

ACKNOWLEDGEMENTS

The author is grateful to the editor, theanonymous reviewers, and Professors DennisKennedy and Anne Marie Smith for their insight-ful comments and suggestions.

REFERENCES

Aczel J, Saaty TL. 1983. Procedures for synthesizingratio judgments. Journal of Mathematical Psychology27: 93–102.

Aguilar FJ. 1967. Scanning the Business Environment.MacMillan: New York, NY.

Ali I, Cook WD, Kress M. 1986. Ordinal rankingand intensity of preference: a linear pro-gramming approach. Management Science 32(12):1642–1647.

Ansoff HI. 1980. Strategic issue management. StrategicManagement Journal 1: 131–148.

Barzilai J, Cook WD, Kress M. 1986. A generalizednetwork formulation of the pairwise comparisonconsensus ranking model. Management Science32(8): 1007–1014.

Beck MP, Lin BW. 1983. Some heuristics for theconsensus ranking problem. Computers and Opera-tions Research 10(1): 1–7.

Blin JM, Whinston AB. 1974. A note on majority ruleunder transitivity constraints. Management Science20(11): 1439–1440.

Bouyssou D. 1990. Building criteria: a prerequisite forMCDA. In Readings in MCDA, Bana e Costa CA(ed.). Springer-Verlag: Heidelberg.

Bowman VJ, Calantoni CS. 1973. Majority rule undertransitivity constraints. Management Science 19(9):1029–1041.

Brun WAT, Teigen, KH. 1988. Verbal probabilities:ambiguous, context-dependent, or both? Organiza-tional Behavior and Human decision Processes 41:390–404.

Budescu DV, Wallsten TS. 1985. Consistency in inter-pretation of probabilistic phrases. OrganizationalBehavior and Human Decision Processes 36:391–405.

Camillus JC. 1975. Evaluating the benefits of formalplanning. Long Range Planning 8: 33–40.

Cardoza RN, Wind J. 1985. Risk return approach toproduct portfolio strategy. Long Range Planning 18:77–85.

Cook WD, Kress M. 1985. Ordinal rankings withintensity of preference. Management Science 31(1):26–32.

Cook WD, Seiford LM. 1978. Priority ranking andconsensus formulation. Management Science 24(16):1721–1732.

Daft RL, Sormunen J, Parks D. 1988. Chief executivescanning, environmental characteristics, and companyperformance: an empirical study. Strategic Manage-ment Journal 9: 123–139.

David FR. 1985. Computer-assisted strategic planningin small businesses. Journal of Systems Management36(7): 24–34.

David FR. 1986. The strategic planning matrix}quan-titative approach. Long Range Planning 19(5):102–107.

M. TAVANA94


David FR. 1993. Concepts of Strategic Management,(4th edn). Merrill Publishing Company: Columbus,OH.

De Kluyver CA, Moskowitz H. 1984. Assessing scenarioprobabilities via interactive goal programming. Man-agement Science 30(3): 273–278.

Dyer JS. 1990a. Remarks on the analytic hierarchyprocess. Management Science 36(3): 249–258.

Dyer JS. 1990b. A clarification of ‘Remarks on theanalytic hierarchy process.’ Management Science36(3): 274–275.

Eden C. 1995. On evaluating the performance of ‘‘wide-band’’ GDSSs. European Journal of OperationalResearch 81: 302–311.

Edwards W, von Winterfeldt D, Moody DL. 1988.Simplicity in decision analysis: an example anddiscussion. In Decision Making: Descriptive, Norma-tive, and Prescriptive Interactions, Bell DE,Raiffa H, Tversky A (eds). Cambridge UniversityPress: New York, NY.Expert Choice Software. 1990.Decision Support Software, Inc.: McLean, VA.

Fahey L, King WR, Narayanan VK. 1981. Environ-mental scanning in strategic planning}the state of theart. Long Range Planning 14(1): 32–39.

Festinger L. 1964. Conflict, Decision, and Dissonance.Tavistock Publications, Ltd.: London, Great Britain.

Ginter P, Rucks AC. 1984. Strategic models andsimulations: an emerging decision-making aid. Jour-nal of Systems Management 35: 12–16.

Godet M. 1987. Scenarios and Strategic Management.Butterworth Scientific: London, England.

Grant JH, King WR. 1982a. The Logic of StrategicPlanning. Little, Brown and Company: Boston.

Grant JH, King WR. 1982b. Strategy formulation:analytical and normative models. In Strategic Man-agement: a New View of Business Policy and Planning,Schendel DE, Hofer CW (eds). Little, Brown andCompany: Boston, MA.

Hammond JS, Keeney RL, Raiffa H. 1999. SmartChoices. Harvard Business School Press: Cambridge,MA.

Harker PT, Vargas LG. 1987. The theory of ratio scaleestimation: Saaty’s analytic hierarchy process. Man-agement Science 33: 1383–1403.

Harker PT, Vargas LG. 1990. Reply to ‘Remarks on theanalytic hierarchy process’ by J. S. Dyer.ManagementScience 36(3): 269–273.

Hill CEL, Jones GR. 1989. Strategic Management:an Integrated Approach. Houghton Mifflin: Boston,MA.

Hofer CW, Schendel DG. 1978. Strategy Formulation:Analytical Concepts. West Publishing Company: St.Paul, MN.

Hurd DA. 1979. Vulnerability analysis. Planning Review7: 31–34.

Keeney RL. 1992. Value-focused Thinking: a Path toCreative Decision Making. Harvard University Press:Cambridge, MA.

Keeney RL, Raiffa H. 1976. Decisions with MultipleObjectives: Preference and Value Tradeoffs. Wiley:New York, NY.

Keller LR, Ho JL. 1988. Decision problem structuring:generating options. IEEE Transactions on Systems,Man, and Cybernetics 18: 715–728.

King WR. 1980. Implementing strategic plans throughstrategic program evaluation. OMEGA 8(2):173–181.

Kirkwood CW. 1997. Strategic decision making: multi-objective decision analysis with spreadsheets. Dux-bury Press: Belmont, CA.

Klein HE, Newman W. 1980. How to use SPIRE: asystematic procedure for identifying relevant environ-ment for strategic planning. Journal of Businessstrategy 1(1), 32–45.

Kleindorfer PR, Kunreuther HC, Schoemaker PJH.1993. Decision Sciences: an Integrative Perspective.Cambridge University Press: New York, NY.

Koriat A, Lichtenstein S, Fischoff B. 1980. Reasons forconfidence. Journal of Experimental Psychology 6:107–118.

Lewis HS, Butler TW. 1993. An interactive frameworkfor multi-person, multiobjective decisions. DecisionSciences 24(1): 1–22.

Lodish LM. 1982. Experience with decision calculusmodels and decision support systems. In MarketingDecision Models, Schulz R, Zoltners A (eds). North-Holland: New York, NY.

McCartt AT, Rohrbaugh J. 1989. Evaluating groupdecision support system effectiveness: a performancestudy of decision conferencing. Decision SupportSystems 5: 243–253.

Merkhofer MW. 1987. Quantifying judgmental uncer-tainty: methodology, experiences, and insights. IEEETransactions on Systems, Man, and Cybernetics 17:741–752.

Mitroff II. 1982. Dialectical squared: a fundamentaldifference in perception of the meaning of some keyconcepts in social science. Decision Sciences 13:222–224.

Moore CM. 1994. Group Techniques for Idea Building.Sage Publications: Thousand Oaks, CA.

Moore S. 1995. Making sense of strategic management:towards a constructive guide. Management Decision33: 19–23.

Nees DB. 1983. Simulation: a complementary methodfor research on strategic decision making processes.Strategic Management Journal 4: 175–185.

Naylor ME. 1983. Planning for uncertainty: thescenario-strategy matrix. In The Strategic Manage-ment Handbook, Albert KK (ed.). McGraw-Hill: NewYork, NY.

Naylor TH. 1982. Strategic planning models. Manage-rial Planning 30: 3–11.

Naylor TH, Schauland H. 1976. A survey of users ofcorporate planning models. Management Science 22:927–937.



Radford KJ. 1980. Strategic Planning: an AnalyticalApproach. Reston: Reston, VA.

Reagan P, Rohrbaugh J. 1990. Group decision processeffectiveness: a competing values approach. Group andOrganization Studies 15(1): 20–43.

Rhyne LC. 1985. The relationship of information usagecharacteristics to planning systems sophistication: anempirical examination. Strategic Management Journal6: 319–337.

Roy B. 1990. Decision-aid and decision making.European Journal of Operational Research 45:324–331.

Roy B, Bouyssou D. 1987. Famille de crit"eeres: probl"eemede coh!eerence et de d!eependence. Decument du Lamsade37, Universite Paris-Dauphine.

Saaty TL. 1977. A scaling method for priorities inhierarchical structures. Journal of Mathematical Psy-chology 15: 234–281.

Saaty TL. 1980. The Analytic Hierarchy Process.McGraw-Hill: New York, NY.

Saaty TL. 1989. Decision making, scaling, and numbercrunching. Decision Sciences 20: 404–409.

Saaty TL. 1990a. How to make a decision: the analytichierarchy process. European Journal of OperationalResearch 48: 9–26.

Saaty TL. 1990b. An exposition of the AHP in reply tothe paper ‘Remarks on the analytic hierarchy process.’Management Science 36(3): 259–268.

Schelling TC. 1960. The Strategy of Conflict. HarvardUniversity Press: Cambridge, MA.

Schmid H. 1986. Managing the environment: strategiesfor executives in human service organizations. HumanSystems Management 6: 307–315.

Schoemaker PJH. 1993. Multiple scenario development:its conceptual and behavioral foundation. StrategicManagement Journal 14: 193–213.

Schoemaker PJH, Russo JE. 1993. A pyramid ofdecision approaches. California Management Review36: 9–31.

Schoemaker PJH, Waid CC. 1978. A comparison ofseveral methods for constructing additive representa-tion of multi-attribute preferences. Wharton AppliedResearch Center, University of Pennsylvania.

Schweiger DM, Sandberg WR, Rechner PL. 1989.Experiential effects of dialectical inquiry, devil’sadvocacy, and consensus approaches to strategicdecision making. Academy of Management Journal32: 745–772.

Schweiger DM, Sandberg WR, Ragan JW. 1986. Groupapproaches for improving strategic decision making: acomparative analysis of dialectical inquiry, devil’sadvocacy, and consensus. Academy of ManagementJournal 29: 51–71.

Schwenk CR. 1988. The Essence of Strategic DecisionMaking. Lexington Books: Lexington, MA.

Shim JP. 1989. Bibliographical research on the analytichierarchy process (AHP). Socio-Economic PlanningSciences 23: 161–167.

Spetzler CS, Stael von Holstein CAS. 1975. Probabilityencoding in decision analysis. Management Science22: 340–358.

Steeb R, Johnson SC. 1981. Computer-based interactivesystem for group decision-making. IEEE Transactionson Systems, Man, and Cybernetics 11(8), 544–552.

Stover JG, Gordon TJ. 1978. Cross-impact analysis. InHandbook of Futures Research, Fowles J (eds).Greenwood: Westport, CN.

Tavana M, Banerjee S. 1995. Strategic assessment model(SAM): a multiple criteria decision support system forevaluation of strategic alternatives. Decision Sciences26: 119–143.

Tavana M, Kennedy D, Mohebbi B. 1997. An appliedstudy using the analytic hierarchy process to translatecommon verbal phrases to numerical probabilities.Journal of Behavioral Decision Making 10: 133–150.

Thompson AA, Strickland AJ. 1998. Strategy Formula-tion and Implementation, (10th edn). Irwin: Home-wood, IL.

Triantaphyllou E. 2000. Multi-criteria Decision MakingMethods: a Comparative Study. Kluwer AcademicPublishers: Boston, MA.

Triantaphyllou E, Mann SH. 1995. Using the analytichierarchy process for decision making in engineeringapplications: some challenges. International Journal ofIndustrial Engineering: Applications and Practice 2(1):35–44.

Vickers B. 1992. Using GDSS to examine the futureEuropean automobile industry. Futures 24: 789–812.

Vincke P. 1992. Multicriteria Decision Aid. John Wileyand Sons: New York, NY.

Weigelt K, Macmillan I. 1988. An integrative strategicanalysis framework. Strategic Management Journal 9,27–40.

Wright B, Rohrbaugh J. 1999. Evaluating the strengthsand weaknesses of group decision-making processes: acompeting values approach. Group Facilitation: AResearch and Applications Journal 1: 5–13.

Zahedi F. 1986. The analytic hierarchy process}asurvey of the method and its applications. Interfaces16, 96–108.

Zeleny MA. 1974. Concept of compromise solutionsand the method of the disptaced ideal.Computers and Operations Research 1(Nos. 3–4):479–496.

Zeleny MA. 1982. Multiple criteria decision making.McGraw-Hill: New York, NY.

M. TAVANA96


Documents

Euclid: Strategic AlternativeAssessment Matrixtavana.us/publications/EUCLID.pdf · Euclid: Strategic AlternativeAssessment Matrix MADJIDTAVANA* Management Department, La Salle University,