STRATEGIC DECISION-MAKING PROCESSES:

THE ROLE OF MANAGEMENT AND CONTEXT 1

Vassilis M. Papadakis * Spyros Lioukas David Chambers !!!!

* Post Doctoral Fellow,London Business School, Sussex Place, Regent's Park, London, NW1 4SA, UKTel. (+44) (0) 171 2625-050. Direct Fax (+44) (0) 171 4028-979.

Professor of Business Strategy and Systems,Athens University of Economics and Business, 76 Patission St., 104-34 Athens, Greece.

Tel. (00301) 8656-459. Direct Fax (00301) 8676-265.

! Professor of Comparative Management,London Business School, Sussex Place, Regent's Park, London, NW1 4SA, UK

Tel. (+44) (0) 171 2625-050. Direct Fax (+44) (0) 171 4028-979.

Key Words: Strategic Decision-Making, Rationality, Politics, Strategic Decisions, StrategyProcess, Top Management

1 The authors gratefully acknowledge that this paper has benefited from an EC Post Doctoralfellowship. They would like to thank three anonymous reviewers of SMJ for their comments andsuggestions which helped improve this paper substantially

1 Paper published in Strategic Management Journal, 1998, 19, 115-147.

Please note that format of this paper, as well as other minor details may bedifferent from the final published paper.

2STRATEGIC DECISION-MAKING PROCESSES:

THE ROLE OF MANAGEMENT AND CONTEXT

A B S T R A C T

This paper investigates the relationship between the process of Strategic Decision (SD)

making and management and contextual factors.

First, drawing on a sample of SDs, it analyzes the process through which they are taken, into

seven dimensions: comprehensiveness/rationality, financial reporting, rule formalization,

hierarchical decentralization, lateral communication, politicization, problem-solving dissension.

Second, these process dimensions are related to (i) decision-specific characteristics, both

perceived characteristics and objective typologies of strategic decisions, (ii) top management

characteristics and (iii) contextual factors referring to external corporate environment and internal

firm characteristics.

Overall, the results support the view that SD processes are shaped by a multiplicity of factors,

in all these categories. But the most striking finding is that decision-specific characteristics appear

to have the most important influence on the strategic decision-making process, as decisions with

different decision-specific characteristics are handled through different processes. The evident

dominance of decision-specific characteristics over management and contextual factors enriches

the traditional external control vs strategic choice debate in the area of strategic management.

An interpretation of results is attempted and policy implications are derived.

1STRATEGIC DECISION-MAKING PROCESSES: THE ROLEOF MANAGEMENT AND CONTEXT

INTRODUCTIONStrategic decision-making has emerged as

one of the most active areas of currentmanagement research. The area has greatlybenefited from such research traditions asbehavioral decision theory and transactioncost economics and has recently gained itsown momentum (Schwenk, 1995). However,despite a substantial body of literature, it isstill widely recognized that our knowledgeof strategic decision-making processes islimited and is mostly based on normative ordescriptive studies and on assumptions mostof which remain untested (e.g. Bateman andZeithaml, 1989; Langley 1990; Pettigrew,1990; Rajagopalan et al. 1993; 1997;Schneider and DeMeyer, 1991). AsEisenhardt and Zbaracki put it, despite thecrucial role of strategic decisions, thestrategy process research has not departedsignificantly from a stage of being based on"mature paradigms and incompleteassumptions" (Eisenhardt and Zbaracki,1992, pp. 17).

In particular, the need has beenrecognized for integrative research whichexplicitly considers the impact of context onstrategic processes (Bateman and Zeithaml,1989; Bryson and Bromiley, 1993;Rajagopalan et al. 1993; 1997; Schneiderand DeMeyer 1991; Schwenk, 1995). Forinstance, Pettigrew (1990) asked whetherthe nature of the decision problem shapesthe process more than does theorganizational context through which theprocess proceeds. In the same vein,Rajagopalan et al. (1997) suggest as one ofthe priorities of future research in strategicdecision making the examination of theextent to which variations in strategicdecision-making processes (DMPs) areexplained by variations in organizational,environmental, decision-specific andmanagerial factors.

The wider literature on environmentaldeterminism and the role of managementchoice is relevant here. Strategic Decisions(SDs) are among the main means throughwhich management choice is actuallyeffected. But empirical research has not beenextended to rigorous investigation of the role

of management factors, contextual factorsand decision-specific characteristics on theactual strategic decision-making processes.

The present paper attempts to contributeto this area drawing upon an in-depthempirical investigation of a number strategicdecisions. Specifically, it focuses onstrategic decisions of an investment nature.These are decisions leading to significantcommitment of resources, with significantimpact on the firm as a whole and on itslong-term performance (Marsh et al. 1988).

First, the paper analyzes the processthrough which organizations arrive at a SD.Using prior research and empirical evidenceit identifies and measures significant genericdimensions of the process. The dimensionsextracted refer tocomprehensiveness/rationality,formalization, configuration of the processand politicization. Second, these dimensionsare related to a number of factors belongingto the following categories: decision-specificcharacteristics, top managementcharacteristics, contextual factors i.e.external corporate environment, and internalfirm characteristics (such as systems,performance, size, ownership).

The structure of the paper is as follows.First, we review the theoretical backgroundand propose an integrated researchframework for studying the effect ofmanagement and context on SD processes.Then, comes our research methodology, theconsideration of the dimensions of theprocess of arriving at SDs and theexplanation of the selection andoperationalization of management and othercontextual variables. Next, we present theanalysis of the data as well as the mainresults of the study. Finally, we discuss ourresults, summarize the main conclusions andderive theory and policy implications.

THEORETICAL FRAMEWORKDimensions of SD Processes

Various dimensions/aspects of SDmaking processes have been emphasized inthe literature. Many studies in the field ofstrategic decision-making describe theprocess as a sequence of steps, phases or

2routes (e.g. Fredrickson, 1984; Mintzberg etal. 1976). Others focus on processdimensions instead (e.g. Bourgeois andEisenhardt 1988; Hickson et al. 1986; Lyles,1987; Miller, 1987; Stein, 1980). Severaldimensions of SD-processes can be derivedfrom the literature. These include thefollowing: Comprehensiveness/rationality

dimension (Dean & Sharfman, 1993a;b;Lyles & Mitroff, 1980; Miller, 1987).Elements of rationality can also be tracedin studies addressing such dimensions ascomplexity of methodology (Langley,1990), degree of inquiry (Lyles, 1987),and scrutiny (Cray et al. 1988).

Centralization (Cray et al. 1988; Lyles,1987; Miller 1987),

Formalization/standardization of theprocess (e.g. Stein, 1980),

Political/problem solving dissensiondimension. This includes among otherspoliticality (Lyles 1987; Hickson et al.1986; Dean and Sharfman, 1993b; Pfefferand Salancik, 1974), andnegotiation/bargaining (Cray et al. 1988;Hickson et al. 1986; Pettigrew, 1973),

Other factors have also been suggestedsuch as dynamic factors (Cray et al. 1988;Mintzberg et al. 1976), forcing (Brysonand Bromiley, 1993), and duration(Hickson et al. 1986; Wally and Baum,1994).Characterization of the decision making

process (DMP) on these dimensions allowsthe researcher to examine possibleinterrelationships with contextual and otherfactors.

The Role of Broader Context in StrategicDecision-making

Many researchers have referred to aspectsof contextual influence on strategic decision-making processes (e.g. Beach and Mitchell,1978; Billings et al. 1980; Bryson andBromiley, 1993; Dutton et al. 1983; Hitt andTyler 1991; Rajagopalan et al. 1993).Schneider and DeMeyer (1991), in anattempt to provide an integrative model,proposed the following categorization offactors which are expected to influencestrategic processes: (i) managers' individualcharacteristics and group dynamics, (ii)

internal organizational context and (iii)environmental factors. Pettigrew, (1990),suggests that in addition to context, researchshould consider the role and significance ofthe nature of the decision problem inshaping the process.

An integration of these contextualdomains into a wider framework looks apromising avenue for research. Such aframework must combine at least thefollowing basic perspectives: an "individualdecision perspective", "strategic ormanagement choice", "environmentaldeterminism", and a "firm characteristicsand resource availability perspective". Thefollowing paragraphs briefly discuss thetheoretical underpinnings of eachperspective, as well as the most importantrelevant research efforts under eachperspective.

i. The Decision PerspectiveThe nature of the decision itself, or the

SD project, may be important. Research intodecision-making cognition and labelingsuggests that the same internal or externalstimulus may be interpreted quite differentlyby managers in different organizations oreven within the same organization (e.g. Deanand Sharfman, 1993a; Dutton, 1993; Haleyand Stumph, 1989). It has been argued thatthe way managers categorize and label adecision in the early stages of the decision-making process (DMP), strongly influencesthe organizations subsequent responses(Dutton, 1993; Fredrickson, 1985; Mintzberget al. 1976). For example, there is evidencethat if a decision is perceived as a crisisdifferent actions will be taken than if thedecision is perceived as an opportunity(Jackson and Dutton 1988; Milburn et al.1983). Fredrickson (1985) found that whendecisions were interpreted as threats asopposed to opportunities, the DMP followedwas characterized by greatercomprehensiveness.

Our understanding, however, of theimpact of decision-specific characteristics onorganizational decision-making processes isstill quite limited (Papadakis and Lioukas,1996; Rajagopalan et al. 1993). Most of theempirical work focuses on: (i) singledecision-specific characteristics (e.g.

3opportunity or crisis) and their influence onaspects of the DMP, or (ii) the early stages ofissue identification and diagnosis (e.g.Billings et al. 1980; Dutton 1986; Jacksonand Dutton, 1988). The authors are notaware of any empirical work that empiricallyexamines a range of decision specificcharacteristics in relationship to a range ofprocess dimensions. With few exceptions(e.g. Dean and Sharfman, 1993a; Dutton,1986; Dutton et al. 1989; Dutton et al. 1983;Fredrickson, 1985), existing research has notyet shown in any detail how decision-specific characteristics shape the DMP as awhole.

ii. The Strategic or Management ChoicePerspective

This perspective emphasizes the role ofdecision makers. It stresses that strategicchoices have an endogenous behavioralcomponent, and partly reflect theidiosyncrasies of decision makers (Child,1972; Cyert and March, 1963). A number ofstudies extend this argument further,contending that the role of "upper echelons"or "top managers" or "strategic leadership" isimportant enough to determine strategycontent and process (Child, 1972; Hambrickand Mason, 1984; Miller and Toulouse,1986).

Research has mainly focused on theinfluence of top management (i.e. CEOand/or top management team) on corporatestrategies (Miller and Toulouse 1986;Finkelstein and Hambrick 1990), onperformance (Haleblian and Finkelstein1993; Smith et al. 1994) and on planningformality (Bantel 1993). There has beenlittle empirical work on the link between topmanagement and the process of making SDs(Bantel 1993; Huff and Reger 1987; Lewinand Stephens 1994; Smith et al. 1994). AsRajagopalan et al. (1993, pp 364) stress in arecent review: "research relatingorganizational factors such as .... topmanagement team (TMT) characteristics tostrategic decision processes is limited".

Moreover, the few studies which havebeen done on the links between topmanagement characteristics and strategicDMPs have produced mixed results.Recently Hitt and Tyler (1991) found that

the demographic characteristics of CEOs(i.e. type of academic education) influencedthe modes of strategic decision-makingfollowed. It is interesting to note thatcounter-arguments have also been advanced.Stein, in studying the strategic DMP, went sofar as to conclude that "leadership does notconstitute a meaningful contextual domaininfluencing strategic procedures" (Stein1980: 332). The same view has also beensupported by Lieberson and O'Connor(1972), and Hannan and Freeman (1977).From another perspective, Lyles and Mitroff(1980, pp. 117), note that managementcharacteristics may not influence theorganizational problem-formulation process.

This is a significant issue that needs to beresolved empirically. The influence of topmanagement on SDs remains unclear. Toadvance our knowledge of the role of theCEO and the TMT we need a betterunderstanding of their impact (if any) onstrategic DMPs and/or the underlyingcharacteristics which are important(Rajagopalan et al. 1997; Smith et al. 1994).

iii. The Environmental DeterminismPerspective

According to environmentaldeterminism, strategic decisions andprocesses are expressing adaptation toopportunities, threats, constraints and othercharacteristics of the environment. The roleof top managers is minimized to afacilitation of this adaptation. Hannan andFreeman (1977), and Aldrich (1979) go evenfurther to propose a process of naturalselection of species for organizations: theenvironment determines who will survive,while top managers are passive agents withminimal impact on corporate development.This view is in line with economic theoriesin which decision outputs rather thaninternal Decision-Making Processes (DMPs)are relevant for the explanation of a firm'sbehavior in a competitive environment.

In the context of SDs the environmentaldeterminism perspective mainly addressesthe question of how environmental factors(e.g. dynamism, hostility) influence strategicDMPs. Few empirical studies can be foundhere (e.g. Fredrickson, 1984; Eisenhardt,1989; Judge and Miller, 1991) and those

4available seem to have producedcontradictory results (Rajagopalan et al.1997). For example, Fredrickson andIaquinto (1989) contend that companiesoperating in stable environments followrational-comprehensive strategic DMPs. Inthe same vein, Stein (1980) argues thatcompanies operating in highly dynamicenvironments may tend to employ both lessextensive search and less explicit analysis ofalternatives. Yet, Bourgeois and Eisenhardt(1988) concluded that in high velocityenvironments effective firms follow morerational decision making processes.Sharfman and Dean (1991) argued for a linkbetween environmental heterogeneity andstandardization in the making of SDs. In asimilar vein, Priem et al. (1995) have foundthat comprehensive processes led to betterperformance in rapidly changingenvironments.

In sum, the results of this body do nothelp us in making any meaningfulgeneralizations (Sharfman and Dean, 1991).Rajagopalan et al. (1993 pp. 354) and Dessand Rasheed (1991) note that the smallnumber of studies adds to the uncertainty asto the effects of each environmental aspecton the process of making SDs. Anothercriticism is that most research seems to focusmainly on one important environmentalcharacteristic (i.e. environmentaluncertainty). Other important characteristicssuch us environmental munificence-hostilityseem to have received somewhat lessattention (Rajagopalan et al. 1993).

iv. The Firm Characteristics and ResourceAvailability Perspective

This perspective emphasizes internalfactors such as: internal systems, companyperformance, size, corporate control (i.e.ownership). At the level of theory, it can belinked to the "inertial" perspectiveproposed by Romanelli and Tushman(1986), according to which existingorganizational arrangements, structures,systems, processes and resources, thoughinitially determined by management andenvironmental forces, in turn constrainfuture strategic decision-making. It is alsorelated to resource availability such asprofitability and slack resources. More

specifically:I n t e r n a l S y s t e m s : The systems of

an organization (especially formal planningsystems-FPSs), might be expected not onlyto exert significant influence on the flow ofinformation between the layers of hierarchy,but also to determine the nature and contextof human interactions, and to influence SDprocesses (Armstrong, 1982; Miller, 1987).The literature is replete with studies arguingthat FPSs are essential tools for managers,since they are designed to improvemanagerial decision making (e.g. Duncan,1990; Langley, 1988). But there is anopposite line of argument, which discountstheir contribution to SDs. It has beenconvincingly argued that much of the actualdecision making may take place outsideFPSs (e.g. King 1983; Sinha, 1990). It seemstherefore a fruitful research question, toexplore empirically this link between formalplanning systems and strategic DMPs.Performance: Since virtually all strategicinitiatives require resources, a "resourceperspective" may be added to thedeterminants of strategic decision processes(Bourgeois, 1981; Pfeffer and Salancik,1978). Research relating past performance tostrategic DMPs is limited (Rajagopalan et al.1993). Much research investigatesperformance in relation to the content ofstrategy, planning and strategy formulationprocesses, rather than SDs. An exception isthe study conducted by Fredrickson (1985)who found that past performance had anegative effect on the comprehensiveness ofstrategic DMPs.

More than three decades ago, Cyert andMarch (1963) reached the same conclusion,i.e. that superior performance is expected tolower the intensity with which organizationswill 'search' for and analyze information. Inthe same vein Bourgeois (1981), March andSimon (1958), suggested that slack resourcesoffer organizations the 'luxury' of'satisficing', and suboptimal decision-making. The above arguments lead us tohypothesize that performance may benegatively related to rational decision-making.

Against this, others have found empiricalsupport for a positive relationship. Forexample Smith et al. (1988) found that, for

5both small and larger firms, comprehensiveout-performed less comprehensive decision-making, and Jones et al. (1992) reportedconsistently positive relationships betweenorganizational effectiveness andcomprehensiveness in decision-making.Taken together, empirical results in this areaare conflicting. This may be due to themoderating effect of other omitted variables(e.g. environment) or to modelunderspecification which characterizes muchof the research (Rajagopalan et al. 1993).

F i r m S i z e : Company size is usuallyconsidered to be of importance in the contextof SDs. Again, the evidence is far from clearor generalizable. Fredrickson and Iaquinto(1989) reported that larger size is associatedwith comprehensiveness in strategicdecision-making. Child (1972) alsosuggested that size affects the framework oforganizational decision-making. However, itis worth mentioning that Dean and Sharfman(1993a) as well as the Bradford studies (e.g.Hickson et al. 1986) found no differences instrategic DMPs which could be attributed tosize.

Corporate Control: Several studieshave provided evidence on the importantimplications of corporate control in strategicDMPs (e.g. Lioukas et al. 1993; Mintzberg,1973). The type of ownership or control typeis a variable which has attracted muchattention, especially lately in the literature onmarkets for corporate control andprivatization. If it is hypothesized thatnationally-owned enterprises display anational style of management and national'culture' in decision-making, whilesubsidiaries of multinationals may representan implanted (probably more 'sophisticated')decision-making style, then it will be ofinterest to test whether important differencescan be detected. The Bradford group ofresearchers have provided evidence of theexistence of different decision-makingpatterns between British and multinationalcompanies operating in Britain (Mallory etal. 1983). Moreover, as suggested by severalauthors, public vs. private ownership maydecisively affect decision-making practicesand processes (e.g. Lioukas et al. 1993).

Towards an Integrated ResearchFramework

It is evident from the above brief reviewthat: (i) there has been little research on theinfluence of broader context on SDs, (ii)most of the studies focus on a limitednumber of antecedents while ignoring otherimportant sources of influence on strategicdecision making processes (modelunderspecification), (iii) most of the studiesfocus on just one characteristic of theprocess (i.e. comphrehensiveness, politics,decentralization), despite the fact thatstrategic DMPs are multidimensional innature, (iv) in addition, much of the evidenceproduced is contradictory and far fromestablishing a coherent theory.

Therefore, we are not able to answer thequestion what are the key influences onthe process of making SDs?. Is it theexternal environment as the populationecologists would argue, or is it the topmanagement (CEO and TMT) as theproponents of management choice theorieswould contend? Do internal enterprisecharacteristics affect the process? Dodifferent decision-specific characteristics, asperceived by management, lead to differenttreatment of the decision? Does pastperformance play any role in influencing themaking of SDs? What is the role andsignificance of formal planning systems?

Since these questions remain largelyunanswered (Bateman and Zeithaml, 1989;Bryson and Bromiley, 1993; Pettigrew,1990; Rajagopalan et al. 1993; 1997;Schneider and DeMeyer, 1991), it seems thatwhat is needed is an exploratory approachwhich views the process of making SDs assubject to multiple influences, and examinesthe effects of factors in three contextualdomains: decision-specific characteristics,top management, and context.

The present paper addresses these issuesby formulating an integrative model ofcontextual influence on strategic DMPs. Thedimensions of the strategic DMP are shownon the right-hand-side of figure 1. Thedecision-specific characteristics are depictedon the left-hand-side. The top block of thediagram indicates management factors whilethe lowest block indicates broader contextualfactors (corporate environment and internal

6firm characteristics).This study operationalizes these

dimensions/factors, and tests their effect onthe DMP. The components of the modeltogether with operationalization andmeasurement issues, follow the discussion ofour methodology. The exploratory nature ofthe paper should again be stressed. Given theprevious studies have reached widelyconflicting conclusions the paper aims toprovide evidence as to which domainspertaining to SD process are more important,and which factors within each domainactually influence various dimensions of theprocess. Further research will be needed toadvance and test particular hypotheses.

Insert Figure 1 about here

RESEARCH METHODOLOGYData Collection and sampling issues

To achieve these objectives an ambitiousstudy was designed and executed, whichtook more than 14 months of intensivefieldwork. This can be characterized as amulti-method, in-depth field research study(Snow and Thomas, 1994). The data sourcesinclude: (i) initial CEO interview, (ii) semi-structured interviews with key participants,(iii) completion of two differentquestionnaires: one general for the CEO andone decision-specific for the keyparticipant(s), (iv) supplementary data fromarchival sources (e.g. internal documents,reports, minutes of meetings).

The research covers 70 SDs in 38manufacturing firms in Greece. A sequenceof steps was followed in order to secure thereliability of data based on participant recall.The process is described in Appendix 1. Thesampling frame comprised all manufacturingenterprises in Greece with more than 300employees, drawn from three industrialsectors (food, chemicals and textiles), a totalpopulation of 89 companies of which 38participated in the research. The average sizeof the companies in the sample is 730 fulltime employees. In most cases two SDs werestudied in each firm, resulting in a sample of70 SDs.

The response rate achieved(approximately 43%) is very highconsidering the intrusive nature of the

research and the fact that top managementwas asked to devote several hours of its time.Comparison between respondent and non-respondent firms on the basis of threeobjective measures (number of employees,total assets, and return on assets), verifiedthe representativeness of the final sample.

Reliability and Validity ConsiderationsA study based on participant recall,

though the dominant method of studyingdecision-making processes, may haveinherent limitations (Bouchard, 1976; Huberand Power, 1985; Kumar et al. 1993). Anumber of procedures have been suggestedto help reduce their impact, including the useof multiple informants (Kumar et al. 1993).Even these methodologies do not guaranteeobjectivity. The nature of the presentresearch (in-depth study of one or two SDsin each company, a separate CEO interview,use of archival data), the specific features ofthe sample (i.e. medium-sized enterprises,existence of few key-informants in each SD),as well as the effort required to findinformants to discuss in depth often delicatematters, relating to a SD, made it difficult touse multiple informants per SD and toaggregate their responses.

Several tactics were followed in anattempt to alleviate possible biases(Bourgeois and Eisenhardt, 1988; Huber andPower, 1985; Kumar et al. 1993), First,archival records documenting the processand its characteristics were collected prior toeach main interview. Second, all thediscussions were recorded. This tacticenabled the researcher to have direct accessto the original discussion and pay attentionto any part of it, at later stages. Interviewnotes, impressions, noteworthy points werewritten down during the first 24 hours afterthe completion of the interview (Bourgeoisand Eisenhardt, 1988). These improvedknowledge of the process.

Third, particular caution was exercised tominimize distortion and memory failureproblems notably by selecting recently takendecisions (Mintzberg et al. 1976), byinterviewing only major participants havingan intimate knowledge of the process(Kumar et al. 1993), by adopting a "funnelsequence" method in conducting interviews

7(Bouchard, 1976), by cross-checkinginterview-derived information against othermanagers recollections (e.g. CEOs), byusing additional informants in cases ofincomplete information, and by cross-checking interview data with other companysources available (e.g. documents, reports,minutes of meetings). In addition, a small number of key processvariables were measured independently (e.g.planning formality, internal reportingactivities) based on archival data. Statisticaltests showed that managers recollectionswere significantly correlated with theselected archival data. In addition, bothsubjective and objective data on corporateperformance were obtained. The twomethods provided similar results, reinforcingbelief in the validity of the data. However,managers recollections were used inmeasuring most of the variables in thisstudy.

Another major consideration was theminimization of common method bias. Tocorrect for such effects the followingprecautions were taken: First, a number ofvariables (e.g. size, performance) arearchival, obviating any danger of commonmethod bias with them. Second, twodifferent questionnaires (general anddecision-specific) were used and they wereanswered by different managers (i.e.dependent and independent variables wereanswered by different persons). Third, theitems used in the analysis were distributedthroughout a lengthy interview. Fourth, scaleanchors were reversed in several places toreduce and compensate for the developmentof response patterns. These precautionsafford some confidence that commonmethod bias was not a problem. Finally, thewillingness and sincerity with which topmanagers participated in the research and theinterest they showed during the interviewingprocess, provide a further reason to believein the face validity of their responses.

OPERATIONALIZATION ANDMEASUREMENT OF SD PROCESSES

The dimensions along which SDprocesses were measured in this study are:comprehensiveness/rationality, extent offinancial reporting, existence of a set of

formalized rules guiding the process,hierarchical decentralization, lateralcommunication, politicization, and problem-solving dissension.These are indicated inappendix 2 along with their measurementdetails, reliability levels and the sourcesfrom which they were derived.

As outlined in the theoretical framework,ample theoretical support can be found forthe above dimensions. For example, theframework adopted is similar to that of Crayet al. (1988). Indeed, the scrutiny dimensionis captured by the comprehensiveness andfinancial reporting dimensions, theinteraction dimension is captured by thepoliticization and problem-solvingdissension dimensions, the centralitydimension is similar to our hierarchicaldecentralization and lateral communicationdimensions.

Moreover, their reliability levels are verysatisfactory. Especially for thecomprehensiveness/ rationality construct,they are higher than those reported by otherresearchers (e.g. Dean and Sharfman, 1993b;Fredrickson, 1984; Smith et al. 1988). It isnoteworthy that despite the fact that theresulting variables tap dimensions of thesame phenomenon (i.e. the strategic DMP)they do not have very high intercorrelationcoefficients (see table 1). All however, are inthe expected direction. For example theformalization construct is positively andsignificantly related to the notion ofrationality, an association argued by severalresearchers (e.g. Langley, 1989).

Insert Table 1 about here

SELECTION ANDOPERATIONALIZATION OFEXPLANATORY VARIABLESMeasuring Decision-specificCharacteristics

To derive generic dimensions, the presentresearch specified and measured sixteendecision-specific characteristics, whichwould apply across the diverse SDs in oursample, and which were based on theliterature review. These issue characteristicsare shown in table 2.

These initial variables were factoranalyzed, using varimax rotation method and

8six factors were derived. Table 2 presents theresults of the factor analysis investigation. Itis worth noting that all factors reflectdistinct, internally consistent patternssuggesting generic characterizations of SDs.A specific name is assigned to each factorbased on the variables loading. The names ofthese factors are: SD's magnitude of impact,uncertainty, amount of pressure anticipatedby the participants, frequency/familiarity,extent to which the SD was perceived as acrisis situation, and finally extent to whichthe SD emerged through the formal planningsystem (planned vs. ad hoc). Appendix 3presents details on variable measurement,sources in the literature from which thesewere drawn, and their reliability levels.

Insert Table 2 about here

But these generic characteristics may notcover the true nature of a project. So furtherobjective decision-specific constants wereadded describing any idiosyncratic aspects ofSDs not accounted by the characteristicsincluded (see Hickson et al. 1986; Shirley,1982). For the purposes of the present paper,a fourfold classification of SDs is identified:new business investment decisions (e.g.acquisitions, mergers, joint ventures, newcompany establishment), investments incapital equipment (e.g. expansion ofproduction equipment, storing facilities,modernization of production equipment),investment in the marketing domain (e.g.new product introduction, marketingchannels), and finally internalreorganization investments (e.g.investments in information systems, internalreorganization etc.). All these are measuredusing dummy (0/1) variables.

Measuring Top ManagementCharacteristics

Both personality and demographicvariables are used to measure characteristicsof the CEO and the top management team(TMT). This would help find out whether itis the CEO, or the top management team, orboth, that play an important role in themaking of strategic decisions.

i. CEO's Personality and Demographic

Characteristics:Two CEO personality characteristics are

incorporated in the present work: need forachievement, and risk attitude. Need forachievement is, according to several writers,one of the basic characteristics positivelyassociated with entrepreneurial success(Gough, 1976). In the present study Steersand Braunstein's (1976) scale is used (seeappendix 3). Attitude towards risk (riskpropensity) is a psychological disposition ofindividuals to show varying degrees of risktaking or risk avoidance behavior. It isamong the major personality dimensionswhich was found to be associated withvarious strategic configurations. Theparticular construct used is derived fromJackson (1976), and Eysenck and Wilson(1975). Appendix 3 describes how thesedimensions were operationalized andmeasured. The resulting reliabilitycoefficients are satisfactory, providingreliability levels similar to those reported inother studies (e.g. Budner, 1962; Steers andBraunstein, 1976).

As regards CEOs demographiccharacteristics several variables have beenused for describing characteristics of'managerial elites'. The present research usestwo variables (i) CEO's length of service inthe company (number of years with thecompany), and (ii) CEO's level ofeducation. Both variables have beenreported to have a profound influence onorganizational processes and outcomes (e.g.Finkelstein and Hambrick, 1990; Hitt andTyler, 1991).

ii. TMT CharacteristicsTwo measures were used. The first,

measures the degree of aggressiveness ofwhat Hage and Dewar (1973) call the'behavioral elite group' (i.e. the CEO and allthose participating in major decisions). Itdraws from Khandwalla (1977) and Stein(1980), and is measured by three itemsexpressing dimensions of the TMTs attitudetowards risk and achievement. The first itemmeasures the degree of 'beat-the-competition attitude, the second TMTs riskpropensity (i.e. attitude towards riskyprojects), and the third the top team's attitudeto innovation. The combination of thesethree items is explained as TMTs

9aggressiveness towards competitors,innovation, and risky projects.

The second variable attempts to capturethe level of education of what Hage andDewar (1973) name as formal elite. It is anobjective variable measuring the percentageof managers, down to the level ofdepartmental heads, who are universitygraduates.

Broader Contexti. Environmental Context

Three environmental dimensions aremeasured using perceptions of top managers:(i) environmental heterogeneity, (ii)environmental dynamism, and (iii)environmental hostility (opposite tomunificence). Appendix 3 describes howthese dimensions were operationalized andmeasured. Cronbach Alpha reliabilitycoefficients are satisfactory, providingreliability levels similar to those reported inother studies which used the same measures.

ii. Internal ContextInternal Systems: For the purposes of the

present work, planning systems are used aspotentially very relevant to SDs (e.g. Sinha,1990). In particular the variableformalization of the planning effort isused. It has been suggested by variousresearchers (e.g. Grinyer et al. 1986) thatformalization is one of the most prominentcharacteristics of planning systems. Thespecific construct used is adapted from thework of Wood and Laforge (1981). Onlyseven of the initial eighteen dimensionsproposed by Wood and Laforge (1981) havebeen selected. These dimensions werepreferred because they refer to the long-termplanning conducted rather than to short termbudgeting practices (see appendix 3). Whenfactor analyzed the seven dimensionsproduce only one factor, further verifying theappropriateness of the modified scale used.

Corporate Performance: Two objectivemeasures of performance are used. First,return on assets (ROA), which is viewed asan operational measure of the efficiency of afirm with regard to the profitable use of itstotal asset base (Bourgeois, 1980). Second,growth in profits, indicating the trend inprofitability improvement. This paper treats

performance as an independent variableinfluencing the strategic DMP. To assurethis, performance measures were calculatedgoing five years prior to the decision studied.This adds confidence in testing whether pastperformance was serious consideration whenmaking the SD.

Firm Size: To measure size this paperuses the log of full time employees (e.g.Fredrickson, 1984).

Corporate Control: Finally, to capture theeffect of type of ownership-control ondecision-making practices two dummy (0/1)variables are used, distinguishing StateOwned Enterprises (SOEs) and privateGreek companies from subsidiaries ofmultinationals.

DATA ANALYSIS CONSIDERATIONSGiven the number of variables involved,

both dependent and independent, separateregression models were applied for each SDdimension. The results reported here presentthe "best" regression equation, i.e. theequation which provides the maximumnumber of significant variables. These wouldgive a first indication of the relativeinfluence of the explanatory variables oneach independent variable.

A consideration in presenting the modelswas whether a full equation model should bepresented along with the best equationmodel, for each of the dependent variables.The size of the sample (i.e. 70 SDs)theoretically would impose limits on thenumber of variables to be introducedsimultaneously. The degrees of freedomwould not be adequate to assure reliable andvalid results in a full variables version.Moreover, the research is exploratory and inmost cases there are not prior strong reasonsto expect relationships with all variables.Versions with full equations can be obtainedby the first author upon request. These donot change the pattern emerging.

Each model was derived by bothbackward elimination and stepwiseregression methods in corroboration. In mostcases the results were identical. In the veryfew cases where the two methods provideddifferent equations, further tests wereattempted by entering and removingvariables from the equation, and finally the

10

model with the best explanatory power wasselected.

Insert Table 3 about here

A second consideration refers to possiblemulticollinearity effects. Table 3 presents theintercorrelations between the independentvariables. Only 2 out of over 300 singlecorrelation coefficients are above 0.50, andthen only slightly, indicating thatintercorrelations are not unduly high. Tosafeguard for multicollinearity effects, theprocedures outlined by Belsley et al. (1980)were followed. According to this, conditionindexes were calculated for each of theregression models of table 4. Thesecondition indices were far below thesuggested warning level of 10.0 for mildcollinearity. Thus, no serious problems areexpected (Belsley et al. 1980).

Several other warning signals were alsoinspected, in order to detect possiblemulticollinearity problems. In none of theequations is there a substantial R2

accompanied by statistically insignificantcoefficients, to raise suspicion about possiblemulticollinearity problems. The stability ofregression coefficients was also tested. Here,several runs were conducted by dropping oradding independent variables in theequation. None of these trial runs hasindicated any extraordinary change inregression coefficients. Most of theregression coefficients appear to producealgebraic signs according to theoreticalexpectations and the coefficients for eachregression show a high consistency withsingle correlations.

Insert Table 4 about here

RESULTSTable 4 summarizes the results of the

paper. All models afford good to satisfactorypredictions of the extent to which each SD-process dimension is determined by thedecision-specific, management andcontextual characteristics. The explanatorypower of the models ranges from .39 to .63and on average exceeds .50. Considering thecross-sectional nature of the research, andcomparing the results to similar efforts this

is seen as very satisfactory (e.g. Dean andSharfman, 1993a; Stein, 1980).

Comprehensiveness/RationalityOverall, results show that

comprehensiveness/rationality is affected bydecision-specific characteristics andinternal context. Environment andmanagement factors are insignificant. Thespecific coefficients of model 1, table 4suggest that SD's magnitude of impact andtype of SD are the most importantdecision-specific characteristics whileplanning formality, corporateperformance, firm size, andownership/control type are the dimensionsof the internal context which significantlyinfluence comprehensiveness in the SD-making process.

Results are in line with Dean andSharfman (1993a) and Stein (1980) whosuggest that the perceived magnitude ofimpact of a decision is among the strongestexplanatory variables of decision-makingbehavior, as decision makers act morecomprehensively/ rationally when decisionsimply important consequences. It is alsonoteworthy that SDs for new businessinvestment and marketing type seem to besubject to less comprehensive/rationalanalysis than SDs on capital investment andinternal reorganization. This follows fromthe negative coefficients of the respectivedummies.

As regards internal context, alldimensions appear to be significant. Resultssupport the normative view that FPSscontribute to more rational decision making(Armstrong, 1982; Duncan, 1990; Langley,1988). Contrary to the line of reasoningwhich understates the contribution of formalplanning systems (FPSs) to decision-making(e.g. King 1983; Sinha, 1990), here planningformality appears to be an importantcontributor to comprehensiveness in SDmaking.

A positive relationship between corporateperformance and comprehensiveness/rationality, is obtained with return onassets. It appears that high levels ofperformance may produce enough resourcesto help in making better, more rationaldecisions; that may mean that high

11

performers are 'offered the luxury' to investin more analysis while poor performers maylack these slack resources. The opposite mayalso obtain, high rationality may lead tobetter performance thus reinforcing apositive relationship. In the same vein,others have argued that more rationaldecisions may themselves lead to betterperformance (Grinyer and Norburn 1977/78;Smith et al. 1988). In our case thisexplanation is less likely, given that ROAfigures used concern previous years. Theresults seem to contradict the oppositeschool of thought, which suggests thatsuperior performance may lower the extentwith which organizations engage in rationaldecision-making (Bourgeois, 1981; Cyertand March, 1963; Fredrickson, 1985).

Size constitutes a significant explanator ofcomprehensiveness. This is in line with thesuggestions of previous work (e.g.Fredrickson, 1984; Fredrickson and Iaquinto1989; Mintzberg, 1973), which state that sizeentails more comprehensive/rationaldecision behavior.

Finally, both dummy variables measuringcontrol/ownership appear to be significantlyrelated to decisioncomprehensiveness/rationality. Morespecifically, using subsidiaries ofmultinationals as benchmarks, statecontrolled enterprises seem to follow morerational processes, while enterprises ofprivate Greek ownership appear to be lessrational. This is an interesting result since itverifies the popular view that Greek privateenterprises follow lesscomprehensive/rational DMPs when makingdecisions of a strategic nature, in comparisonto subsidiaries of multinationals. That stateowned enterprises are closer tomultinationals in this respect is unexpected.Maybe they have qualified staff and engagein more analysis, taking all the necessarytime and effort to collect all necessaryinformation and explore alternatives, inorder to justify their final choices in theirstrong stakeholders, both inside and outsidethe organization.

The insignificant coefficients also provideuseful suggestions. A striking result is thelack of significance of top managementcharacteristics, since no variable loads

significantly in model 1. Only TMTsaggressive philosophy is marginallysignificant (at a 10% level). This seems atodds with current theory, which stresses thevital role of the CEO and/or the TMT oncomprehensive/rational decision-making(Hambrick and Mason, 1984).

A similar observation can be made asregards environmental variables indetermining comprehensiveness, which runsagainst propositions of environmentaldeterminism. We would expectenvironmental variables to influencerationality/comprehensiveness. Indeed,several researchers have argued for thesignificance of environmental heterogeneityin determining strategic processes (Lindsayand Rue, 1980; Smart and Vertinsky, 1984).These authors argued that managers whoperceive their corporate environment ascomplex tend to employ morecomprehensive strategies. Bourgeois andEisenhardt (1988), in their study of strategicdecision processes in high velocityenvironments concluded that effective firmsfollow the rational model in decisionmaking.

Others have argued that companiesoperating in stable environments followrational-comprehensive processes in makingand integrating strategic decisions (e.g. Cyertand March, 1963; Fredrickson, 1984;Fredrickson and Iaquinto, 1989; Hart, 1992).The argument behind this contention is thatstrategists usually find it difficult to rely onformal financial analysis, in-depth study, andrational processes when having to deal withunstable, high velocity environmentscharacterized by information scarcity andrapid change. Instead, they are obliged totake quick, bold decisions in many instances,relying on the available amount ofinformation. In the same vein, companiesoperating in stable environments rarely facesignificant opportunities and thus whenhaving to deal with such a situation theyemploy more rational processes.

Our results do not support either line ofthought. On the contrary they are in line withSmith et al. (1988) who reported a lack ofany statistical significant relationshipbetween environmental dimensions andrationality of strategic decision-making.

12

However, one should consider the results ofthe other dependent variables, before rushingto conclusions.

Financial ReportingFinancial reporting, a dimension of

rationality typically applied in SD especiallyof an investment nature, is significantlyaffected by decision-specific characteristics,and some characteristics of the internalcontext.

More specifically, from a quickinspection of model 2 in table 4 it appearsthat two decision specific characteristicsnotably magnitude of impact andemergence of the SD through planning,CEOs level of education and return onassets are to be positively associated withfinancial reporting, while private Greekownership has a negative association. Ofnote are the marginally significantassociations (10% level of significance)provided by two other generic decision-specific characteristics.

First, results indicate that situationsperceived as crises are actually associatedwith more financial reporting activities. Thisis in line with previous theoreticalargumentation (Dutton, 1986). In general,one might argue that when adversity loomseveryone might want to interpret and explainthe situation in terms of financial analysisand reporting. Or, taking another view, thecompany may seek to exercise control andsupport the meaningfulness of its actions inthe eyes of both internal and externalstakeholders by relying on deeper financialreporting and analysis, since crises usuallyinvolve risks of a significant financial loss.

Second, financial reporting is negativelyrelated to frequency. This result supports theview that frequent/familiar issues are dealtwith by standard rules and analogies frommemories. They therefore are associatedwith less analysis and comprehensivereporting of data (Marmaras et al. 1992).Also the coefficients of the dummiesinvestment in capital equipment andinvestment in marketing are marginallysignificant showing a higher level offinancial reporting for these types ofdecisions, as against investments in internalreorganization.

As regards top management, onlyCEOs level of education is positivelyassociated with financial reporting.Education level has been found to be relatedto the extent of peoples information searchand analysis (Dollinger, 1984). A highlyeducated CEO is thus likely to demand moredetailed information, leading to morefinancial reporting (Bantel, 1993). All othertop management characteristics (e.g. CEOsneed for achievement, risk propensity ortenure) appear to be insignificant.

From internal firm characteristics, againperformance in terms of ROA is positivelyassociated with financial reporting. Animportant finding, is also the negativeassociation between private Greekownership and financial reporting. It impliesthat Greek private firms may rely less onformal financial reporting activities whenmaking strategic decisions thanmultinationals. This strengthens ourargumentation when discussing the results ofthe comprehensiveness/rationality model.SOEs are not different from multinationals.

Of note is the general lack of significanceof external corporate environment. This is inline with the results obtained using therationality/comprehensiveness construct.Interestingly, size does not seem to besignificantly associated with financialreporting. These findings will be furtherdiscussed in later sections of the paper.

Rule formalizationRule formalization in the SD process

(model 3 table 4) is influenced by decision-specific characteristics (decision uncertaintyand emergence through formal planning),top management characteristics (CEO riskpropensity) and corporate control type(Greek-ownership).

From decision-specific characteristics,decision uncertainty and emergence arenegatively associated with rule formalizationof SD process. Uncertainty, as used hererefers to specific decisions, as opposed to theuncertainty caused by the organizationalenvironment. Results are in line withThompson (1967 pp. 134) who contends thatin cases of high uncertainty, managers act inan 'inspirational' manner, by makingobsolete any formal procedures and rules

13

usually followed. One can contend that highuncertainty about the decision may, contraryto the received (common) expectations,result in more intuitive processes (Daft andLengel, 1986; Dean and Sharfman, 1993a)together with use of less formalized rules.Again, as expected, SDs emerging from thediscipline of a formal planning system arefound to follow more formalized paths.

As regards management we observe anegative association between CEO's riskpropensity and rule formalization in takingSDs. Again, such a result is intuitivelyexpected, since risk takers usually break thebounds of organizational systems andformalities and influence the SD processtowards more informal paths. TMTsaggressive philosophy is also related to moreformalized rules.

Finally, private Greek companies seem tobe lagging in rule formalization as thenegative coefficient suggests. SOEs do notdiffer from multinationals.

Overall, however, we cannot argue thatthere exists a 'balanced contribution' of alldomains (SD itself, management andcontext) in explaining SD formalization. Ofinterest is the lack of significant impact ofdecision frequency, corporate environment,planning formality, past performance andsize. This suggests that SD process formalityis independent from the formal planningmachinery of the firm and external corporateenvironment and the other internal companycharacteristics. It is more a matter ofdecision-specific characteristics and topmanagement choice.

Hierarchical decentralization and lateralcommunication

The extent to which SD process isdecentralized and allows participation oflower level managers depends on thedecision-specific characteristics, on CEOtenure and corporate profitability. Moreover,the extent of lateral communication isdetermined by decision-specificcharacteristics, top management teamsaggressiveness, and internal firm planningformality.

More specifically, several conclusionscan be drawn from the results of models 4and 5, table 4. From the decision-specific

characteristics, particularly strong is theeffect of magnitude of impact, followed byperceived pressure and threat/crisis. Resultsimply that SDs with important impact attractthe collective attention of more layers in thehierarchy and more departments as revealedin models 4 and 5. This corroborates Duttonet al. (1989) who argues that issues withgreat magnitude of impact imply highinterconnectedness with other relevantissues. Therefore, such issues attract morecollective attention and thus result in higherhierarchical decentralization and lateralcommunication.

Pressure has a negative coefficient,suggesting that when SDs are taken underpressure there may not be enough time toinvolve more levels and departments.

An interesting effect obtains forthreat/crisis. This suggests that threateningsituations result in more hierarchicaldecentralization. At first this iscounterintuitive. A number of authors (e.g.Dutton, 1986; Herman, 1963) argue thatcentralization of authority is the expectedoutcome of crises, since two opposite forcesclash. First, managerial elites undertake theresponsibility of the whole effort to divertthe crisis. Second, middle managers feelingthat the issue might be 'too heavy' for themto deal with, pass it to top management.

Milburn et al. (1983), provide anexplanation for this counterintuitive result:their findings suggest that althoughcentralization was the immediate outcome ofcrises, the actual intermediate response wasdecentralization of authority. This may beexplained by the distributed character ofinformation and expertise; if we admit thatthe source of vital information is middlemanagement, centralization deprives topmanagement of extremely useful data.Herman (1963) offered another explanation.He argued that the relationship betweencrisis and decentralization of authority iscurvilinear. Thus, under situations which arecharacterized as "mild crises" one mayobserve decentralization. By contrast, whencrises become acute, authority centralizationis found. Further investigating thedescriptive statistics of the variablemeasuring the extent of perceived crisis, wemay see that the variable is measured on a

14

five-point scale and has a mean of 2.51which implies rather 'mild crises', on theaverage. This suggests that our sample ofSDs may not include intense crises whichare assumed to lead to hierarchicalcentralization and less lateralcommunication.

As regards the effect of type ofinvestments on hierarchical decentralization,all dummy coefficients are significantshowing that only internal reorganizationinvestments are relatively more centralized.On the contrary none of the dummyvariables is significantly associated withlateral communication.

Top management has two significantcoefficients. In model 4 the variablemeasuring CEO's tenure is positivelyrelated to hierarchical decentralizationpatterns. This may be explained by the factthat CEO's tenure may influenceparticipation patterns by developing greaterlevels of social integration, and possibly byincluding in the dominant coalition moremanagers from various layers (Wiersemaand Bantel, 1992).

Similarly, TMT's aggressive philosophyappears significantly positively associatedwith lateral communication during SDmaking. This is in line with the view that inorder to follow an aggressive strategy topmanagement may need more informationand cross-departmental involvement. If weconsider that individual departments possessspecialized information and their alliance iscrucial to implementation of an aggressivephilosophy, it is plausible that the level oflateral communication may increase.

It is somewhat surprising, however, tofind again that corporate environment hasinsignificant coefficients. None of theenvironmental variables seems to have anyeffect on hierarchical decentralization andlateral communication patterns during themaking of SDs. This is more intriguing if weconsider the results of various studies whichhave argued for close links betweenenvironmental characteristics anddecentralization and communication. Forexample, Lindsay and Rue (1980) argue thatenvironmental heterogeneity is associatedwith more hierarchical decentralization, andmore lower-levels involvement. Miller and

Friesen (1983), also suggested that anincrease in perceived environmentalheterogeneity is expected to complicate theadministrative task, thus resulting insubsequent changes in structure (e.g. morelateral communication). In the same vein,Grinyer et al. (1986) contend, thatenvironmental stability favors the delegationof authority to lower levels in the hierarchy,during the planning process. Interestingly,none of these relationships are supported byour data.

Performance, expressed by ROA, ispositively related to hierarchicaldecentralization. This is in line with theresults presented by Bourgeois andEisenhardt (1988) and others. They havereported that the more the power to makestrategic decisions is delegated to thefunctional and divisional executives, thehigher the performance of the firm. This lineof thought assumes that greater participation(especially by middle managers) will have apositive impact on organizationalperformance by triggering two parallel,positive phenomena. On one hand, theinvolvement of more people in strategicDMPs increases the level of consensusamong managers, produces a commonunderstanding of the joint task, creates aclimate of shared effort, and facilitatessmooth implementation of strategicdecisions. Smooth implementationcontributes in turn to higher performance(Wooldridge and Floyd, 1990). On the otherhand, lack of involvement of employeesother than 'strategic elites' in the process hasbeen found to create implementationproblems, including sabotage (Guth andMacMillan, 1986).

Secondly, middle managers act asinformation monitors and are usually thefirst to sense potential threats andopportunities in their own particular domains(Pascale, 1984). Increasingly, topmanagement's ability to sense the emergenceand meaning of various challengesencountered, is seen as a critical strategiccapability. Due to information overload, topmanagers may be less and less able to fullyunderstand the world around them.According to this view, strategic DMPs insuccessful firms are, more a product of a

15

shared effort, than deliberation by oneperson.

Two other results seem also important.First, planning formality as expected, has apositive coefficient in model 5, indicatingthat it is associated with higher lateralcommunication. Second, corporate controldoes not appear to influence decentralizationand communication patterns. So there maybe no significant differences between thevarious types of enterprises -state, private,foreign- in decentralization andcommunication, when taking SDs.

Politicization and Problem-solvingDissension

Politicization and problem solvingdissension (models 6 and 7, table 4) aremainly influenced by decision specificcharacteristics (uncertainty, and pressure),one external environmental characteristic(heterogeneity) and certain internal contextcharacteristics (e.g. planning formality,performance and corporate control).

Both politicization and problem solvingdissension seem to be influenced more bySD uncertainty and less by othercharacteristics such as impact, threat/crisisor pressure. This is in line with Lyles (1981),who based on case evidence, argued thatuncertainty about certain aspects of an issue(i.e. definition) may raise politicality in theproblem formulation process. Indeed, whenuncertainty exists (for example about theactions to be taken and/or the information tobe collected), one may expect to find both adivergence of opinions during the initialstages of problem formulation, and a surgeof political activities during the issueresolution process.

Interestingly pressure situations are foundto have positive effect on problem solvingdissension among participants. It seems thatpressure situations may intensify dissensionas many times they call into question theefficacy with which the dominant coalitionhas acted in the past to preserveorganizational interests, and threatens thepower-base of managerial elites (Dutton,1986). It is not at all unexpected to witnesswhat Herman (1963) has called 'factionalism'i.e. various units or departments favoropposite views about the proper reaction to

the problem. It is also interesting to note thatall dummies for type of decision havenegative coefficients. This suggests that allinvestments cause less dissension thaninternal organizational SDs.

Environmental factors also seem tohave no significant effect on politicization,while environmental heterogeneity appearsto negatively influence problem-solvingdissension. This latter result is contrary toexpectations. For example, Dess and Origer(1987) argue for an inverse relationshipbetween environmental heterogeneity andconsensus on goals, since complexity givesrise to more possible points of conflict amongmanagers and makes consensus more difficultto achieve. Others, (e.g. Lyles and Mitroff1980) argued along similar lines.

A positive relationship obtains betweenpoliticization and planning systemsformality. This suggests that planningformality has a positive influence onpoliticization. Results corroborate theprevailing view that FPSs encouragepolitical behavior (Langley, 1988; Rhyne,1986) since managers may perceive planningsystems as a means through which personalviews are communicated, politicalaspirations take effect, and political activitymay develop.

Of note is also that both politicization andproblem solving dissension are found to bepositively related to growth in profits. Wewill draw attention to this finding in ourdiscussion section.

The positive statistical significantassociation between politicization and SOEsmay be attributed to the multiplicity ofinternal and external interests in the contextof SOEs. The fact that numerous parties mayintervene and try to skew the output of thedecision process in their preferred directionmay raise politicization levels in SOEsrelative to other categories of enterprises.

Of note is the general absence of topmanagement characteristics in influencingboth political activities and problem solvingdissension. Results support the view that theemergence of internal politics and dissensiondepends more on the decision-specificcharacteristics and certain characteristics ofthe external and internal context(environmental heterogeneity, planning

16

formality, performance), rather than thecharacteristics of the decision makersthemselves. This appears to be at odds withDean and Sharfman (1993b), who havereported that political behavior (amongothers) stems from the characteristics of thedecision making group (interpersonal trust).

DISCUSSIONOverall, the results of the paper support

the view that for understanding strategicDMPs in depth, an integrative model whichincludes decision-specific, management,environmental and organizational factors isneeded. Results suggest that strategic DMPsare shaped by the interplay of these factors.Neither the external control model(environment), nor the strategic choicemodel (decision makers), nor the corporateinertial model (size), or the resourceavailability (performance), adequatelyexplain actual strategic decision-makingbehavior.

The most striking finding was thedominant role of decision-specificcharacteristics in determining decisionprocesses. To the best of the authorsknowledge this is the first time that thedominant role of decision-specificcharacteristics (generic attributes andobjective categorization) is verified in thecontext of SDs. The specific results showthat the generic characteristics of a SD suchas its perceived magnitude of impact,frequency/familiarity, its uncertainty, itsthreat/crisis component and whether itemerges through discipline of the planningsystem of the firm, significantly influencedimensions of the DMP, more than otherenvironmental, organizational andmanagerial factors.

Somewhat similar results were obtainedby Meyer and Goes (1988), in their study ofinnovation assimilation. In assessing thecomparative influence of various contextualdomains on innovation assimilation theyfound that environment, organization andleadership, taken together, were relativelypoor predictors of innovation. In contrastvery good predictors proved to be the innateattributes of innovations.

The generic decision-specificcharacteristics tested in the present study

may not be exhaustive. This is obvious fromthe statistically significant dummiesexpressing types of decisions. This meansthat the objective type of SD variablesintroduced offer a unique additionalexplanatory power, not captured by thegeneric decision-specific characteristicsincluded. It also calls for a search foradditional generic SD characteristics whichmay, when included, minimize any remnanteffect of dummies. This however wouldrequire further research.

New business investments andinvestments in marketing exhibit lessassociation with rationality in comparison tocapital equipment investments and internalreorganizations. Investments in capitalequipment and marketing exhibit morefinancial reporting in comparison to otherinvestments. Investments in internalreorganization seem to be hierarchicallymore centralized in comparison to otherinvestments. Also internal reorganization isaccompanied by more problem-solvingdissension in comparison to all other types ofSDs.

The study has also established the relativeimportance of top managementcharacteristics. Certain CEO characteristicsentered significantly into the regressionmodels and influenced financial reporting,formalization and hierarchicaldecentralization. This implies that thecontribution of CEO characteristics havealso their own effect when controlling forother contextual factors. These results are inline with research reporting that CEOpersonality and demographic characteristicsare related to aspects of strategic DMPs (e.g.Miller and Toulouse, 1986).

In addition to CEO characteristics, TMTaggressive philosophy appear to influencerule formalization and lateralcommunication and only weaklycomprehensiveness/rationality. It is alsoworth noting that both the CEO and theTMT appear to be insignificant ininfluencing such dimensions as politicalactivities and problem solving dissension.This may be explained by the fact thatexecutives do not always have completelatitude of action (Hannan and Freeman,1977). There exist conditions of restricted

17

discretion where TM becomes less importantand other factors as corporate control, andfirm characteristics or decision-specificcharacteristics become more significant ininfluencing politicality in SD making.

Overall, regression results support theview that both CEO's and TMT'scharacteristics have their effects in SDprocesses and are in line with Kets De Vriesand Miller (1986) and Nahavandi andMalekzadeh (1993). This may lead us to lendcredence to the upper echelons view oforganizations, and put into dispute theallegations of population ecologists (e.g.Hannan and Freeman, 1977) who considerTMT to be but a passive agent.

As regards external corporateenvironment, we found only one significantcoefficient: namely, that environmentaldynamism has a negative effect on problemsolving dissension. All other environmentalvariables we found to be largelyinsignificant. Overall, our results contradictresearchers who argued that environmentalfactors as opposed to internal organizationalfactors are the primary sources of influenceon SDs (e.g. Hannan and Freeman 1977;Jemison, 1981).

No support was found for the role ofenvironmental heterogeneity/complexity onstrategic DMP. Rajagopalan et al. (1993)argued that the degree of environmentalcomplexity in a firms operatingenvironment directly impacts the amountand nature of information that has to beprocessed by decision makers. Research oncognitive processes also suggests thatenvironmental heterogeneity affects strategicdecision process characteristics such ascomprehensiveness, and leads to greater useof cognitive simplification processes(Schwenk, 1988). None of these aresupported by our results. Again,environmental hostility was found toinfluence none of the characteristics of SDprocess. The argument that organizations inhostile environments follow more rationaldecision processes (Dess and Beard, 1984;Rajagopalan et al. 1993) receives noempirical support.

In an attempt to explain this lack ofdominance of the external control model,several speculative assumptions may be

advanced regarding the particular countryfrom which results are drawn. Schneider andDeMeyer (1991), reported that LatinEuropean managers in contrast to otherEuropeans, may be characterized by anattitude of having limited control over theexternal environment. Thus, they may directtheir efforts towards controlling theimmediate, internal environment, and adjustdecision making processes accordingly.Although speculative this conjectureprovides a fruitful avenue for research incomparative decision-making practices.

Internal firm characteristics show moresignificant effects on DMPs. Firstly, formalplanning systems appear to have a positiveinfluence on three aspects of the DMP:comprehensiveness, lateral communication,and politicization. This is in line withtheoretical and normative speculationsarguing that planning systems lead to morerational decision-making (Armstrong, 1982;Duncan, 1990; Langley, 1988). Again,results corroborate the prevailing view thatformal planning systems encourage bothlateral communication (Langley,1988;Tregoe and Tobia, 1991) and politicalbehavior (Langley, 1988; Rhyne, 1986).

Concerning formal planning systems weshould note the difference with anotherdimension which in the course of thisresearch was treated as a projectcharacteristic rather than as an internal firmcharacteristic, i.e. the emergence of a SDthrough the formal planning system. Thisrefers to the decision whether to handle a SDin the formal planning machinery or treat itad hoc outside the planning system, and isrelated to the initial stages of projectformulation. These two variables areconceptually distinct and their inter-correlation is not unduly high (r=0.47). Theresults suggest that both have a significantinfluence on strategic decision-makingprocesses.

Another set of interesting relationshipsrevolves around the significance of corporateperformance in determining decisionprocesses. Return on assets, providessignificant positive associations withrationality, financial reporting andhierarchical decentralization. Profit growthin turn are highly related to politicization

18

and dissension. So different performanceaspects appear to influence differentdimensions of the process. Past profitabilitymay lead to more rationality anddecentralization of DM processes. Growth inprofits may lead to internal politics anddissension. The latter is an interestingfinding which deserves attention. It may bethat high growth of profits over time raisesinternal conflicts as to where these resourcesshould be invested. Particularly telling wasthe example of a fast-growing company inthe food sector which considering investinga sizable part of its retained earnings. Themarketing department wanted majorinvestments in new products and channels ofdistribution, while the productiondepartment supported a single investment instate of the art computerized storingfacilities. The process of deciding where toinvest these surplus resources was a highlypolitical one.

Size is found to be largely insignificant inalmost all regression models except forcomprehensiveness/rationality whichappears to increase with size. A similarpattern of associations was reported by Deanand Sharfman (1993a). This result seems tobe at odds with the conventional wisdomthat as companies grow they tend to movetowards more procedural and formalizeddecision-making (e.g. Mintzberg, 1973).

Finally, it appears that control type has asignificant impact on several aspects ofstrategic DMPs. Indeed, enterprises understate ownership (SOEs) seem to apply morecomprehensiveness/rationality, and morepoliticization, when making decisions of astrategic nature. This is in line with the viewthat in the context of SOEs political andeconomic considerations coexist (e.g.Lioukas et al. 1993). As strategic decisionsin SOEs are usually subject to scrutiny orinfluence by strong stakeholders thedecision-takers may take pains todemonstrate that they act rationally and tojustify their major decisions both inside andoutside the company (Dean and Sharfman,1993a; Romanelli and Tushman, 1986). Thisexplains the positive relationship betweencomprehensiveness and SOEs. The positivestatistical significant association betweenpoliticization and SOEs may be attributed to

the fact that numerous parties, not only frominside but also from the outside of thecompany, may intervene and try to skew theoutput of the decision process in theirpreferred direction. Such an activity mayraise politicization.

By contrast in private Greek companiesless rational processes, less financialreporting, and less rule formalization appearto prevail. Results seem to comply with theparticular view, that private Greekcompanies lag in analytical resources andinternal systems to supportcomprehensive/rational decision-making,extensive financial reporting and formalizedprocedures, or may be less interested inanalysis.

In a study of decision-making practicesbetween British and American-owned firmsin Britain Mallory et al. (1983) reported thatin general American ownership had little orno influence on the way decisions weremade. Among the few alleged differencesthey found more reliance in Britishcompanies on certain aspects offormalization. The results of this study seemto reveal an important gap between privateGreek enterprises and subsidiaries ofmultinationals which tend to haveinternational characteristics. This impliesthat Greek private management may beassociated with less comprehensive/rationalprocesses, less financial reporting and lessrule formalization. This opens up interestingquestions on comparative decision makingpractices across countries or types of firms,which may be a useful avenue for futureresearch.

IMPLICATIONS AND POSSIBLEEXTENSIONS

Some interesting practical implicationsfollow from the significance of decision-specific characteristics, identified in thisresearch, in comparison to management,corporate environment and internal firmcharacteristics. As certain of these decision-specific characteristics are amenable tomanagerial interpretation and perception itmay be in the interest of management toactively manipulate the meaning orcategorization of strategic issues, andthrough them to influence organizational

19

responses. This is in line with much of thethinking in the area. For example, topmanagement deliberately make certaindecision-specific characteristics salient. E.g.it may choose to manipulate theinformation provided from external orinternal systems, such as EnvironmentalScanning or 'Strategic Issue Management 'or 'Boundary Spanning' systems, to serve itsown goals. What may be communicated as athreat by a specific system may becharacterized as an opportunity by another.Filtering information and manipulatingdecision-specific characterizations, mayenable management to subsequently controlrationality, formalization, lateralcommunication, hierarchicaldecentralization and even the extent ofinternal political activity.

The specific results indicate that certainmanagement and internal firm characteristicsbear on the strategic DMP, while others suchas the environment variables appear to beinsignificant. So contrary to allegations onthe significance of environmentaldeterminants, management seems to play animportant role. More specifically, CEOsrisk propensity, education and tenure as wellas the TMTs aggressiveness seem to beimportant determinants of certain processdimensions (e.g. rule formalization, financialreporting, hierarchical decentralization, andlateral communication). These resultsdirectly question the findings of a number ofresearch works arguing for the secondaryrole of managers.

Another implication of the results is thatthey verify the important role of theformality of planning systems in influencingthe making of SDs. Results indicate thatformal planning influences the way in whichstrategic decisions are taken, and thus to anextent, strategy itself. Indeed, by influencingcomprehensiveness, lateral communication,and political activities, formal planningsystems seem to act as a powerful input tothe process of strategy making.

Corporate control is also important,suggesting, an important lagging of privateGreek firms in rationality. SOEs insteadseem to be closer to international standards,as benchmark against the subsidiaries ofmultinationals included in the sample.

The present research effort has touchedon only a few of the research questions in thefield of strategic decision-making. Severalextensions, both methodological andsubstantive need to be made, and a numberof points concerning overall researchrecommendations in the area be highlighted.

First, the research has established thedominance of decision-specificcharacteristics over other management andcontext factors in making strategic decisions.More work needs to be done to test thegeneralizability of the present results in othersettings and sample designs. Another usefulline of research would be to examine thesame hypotheses in more narrowly definedsamples, e.g. controlling for types ofenterprises or SDs and for other contextvariables, so that consistent research findingsbe accumulated and a more focusedcontingency theory on the impact of contexton strategic DMPs be developed.

Second, despite the fact that theregression models tested here appear to havea very good explanatory power over theadopted process dimensions, there stillremains an unexplained percentage ofvariance. Further research can incorporateadditional variables not considered in thecourse of this study (e.g. contextual elementssuch as reward systems, organizationalstructure or further SD attributes), and mayadopt different and possibly moreappropriate operationalizations of theconstructs already used.

Third, theory is needed that moreaccurately reflects the strategic DMP in itscontext and the relative weight of itsdeterminants. Progress in this area couldsignificantly improve both our understandingand eventually the quality of strategic DMPs.A refinement of the formulation may benecessary before a more substantialexplanation or prediction capability can beachieved.

Fourth, it may be useful to includeintermediate outcome variables of SDs (e.g.innovation, learning, decision quality,satisfaction, commitment, and overallcompany performance/ effectiveness). Futureresearch may seriously consider theseaspects, which were outside the scope of thepresent work.

20

Fifth, the present work has established themultidimensional aspects of SD processesand the multiple relationships with the mainvariables of the study. This enhances theneed for producing a more integrated imageof decision-making reality through thesimultaneous study of a large number ofqualities, and use of more sophisticatedmultivariate analysis of contextual influenceon strategic DMPs (Rajagopalan et al. 1993).Simultaneous equation techniques would beuseful to further examine determinants of theactual sets of relationships in practice.

It will also be interesting to investigatehow closely these results, obtained in onesouthern E.U. country, apply to SDs in thecountries where the vast majority of researchin this field has taken place (USA, UK,Canada).

Some of the empirical results from thisstudy concur with the body of researchevidence in the area. This would support the'culture free argument which maintains thatcultural differences may not affectrelationships among structural characteristics(e.g. Negandhi, 1975). There are however,certain findings which may be interpreted asculture specific. For instance, the enhancedrole of decision-specific characteristics asagainst corporate environment and topmanagement may be specific to theparticular context. It is possible that becauseGreek private firms have less formal rulesand less comprehensive decision processesthan their U.S. or British counterparts, theywould be more likely to treat each strategicdecision as unique and thus react in a moreemotional manner. In U.S. or British firmswith more formal rules and procedures forSDs, decision-specific characteristics may beless important and all or most SDs may behandled using similar processes. This doesnot mean decision-specific characteristicsare of no significance in such a context: thework of Dutton et al. (1989) as well as manyothers provide clear evidence to the contrary.It rather implies that the role of thesecharacteristics may be not as dominant asthey appear to be in the Greek context. Thisopens up a very promising avenue for futureresearch on comparative decision-makingpractices, across different country and/orcultural domains. As Charles Schwenk

argues in a recent literature review, it maybe that many of the conclusions aboutstrategic decision making developed in theU.S. context will have to be modified inorder to be applicable across cultures(Schwenk, 1995 pp. 484).

21

REFERENCESAchrol, R.S., and L.W. Stern, "Environmental

Determinants of Decision-Making Uncertainty inMarketing Channels", Journal of MarketingResearch, (1988), 25,36-50.

Aldrich, H.E., Organizations and Environments,Prentice-Hall, New Jersey, 1979.

Armstrong, S.J., "The Value of Formal Planning forStrategic Decisions: Review of EmpiricalResearch", Strategic Management Journal,(1982),3,197-211.

Bantel, K.A., "Top Team, Environment andPerformance Effects on Strategic PlanningFormality", Group and OrganizationManagement, (1993), 18, 4, 436-458.

Bateman, T.S., and C.P. Zeithaml, "ThePsychological Context of Strategic Decisions: AModel and Convergent Experimental Findings",Strategic Management Journal, (1989), 10, 59-74.

Beach, L.R., and T.R. Mitchell, "A ContingencyModel for the Selection of Decision Strategies",Academy of Management Review, (1978),3,439-449.

Belsley, D.A., E. Kuh, and R.E. Welsch,Regression Diagnostics: Identifying InfluentialData and Sources of Collinearity, John Wileyand Sons, New York: 1980.

Billings, R.S., T.W. Milburn, and M.L.Schaalman,"A Model of Crisis Perception: A Theoreticaland Empirical Analysis", Administrative ScienceQuarterly, (June 1980), 25,300-316.

Bouchard, T.J., JR., (1976), "Field ResearchMethods: Interviewing, Questionnaires,Participant Observation, SystematicObservation, Unobtrusive Measures", InDunnette, M.D., (Ed.), Handbook of Industrialand Organizational Psychology, Chicago, and,McNally: 363-413.

Bourgeois, L.J., "On the Measurement ofOrganization Slack", Academy of ManagementReview, (1981),6,29-39.

Bourgeois, L.J. III, "Performance and Consensus",Strategic Management Journal, (1980), 227-248.

Bourgeois, L.J. III, and K.M. Eisenhardt "StrategicDecision Processes in High VelocityEnvironments: Four Cases in theMicrocomputer Industry", Management Science(1988), 34,7,816-835.

Bryson, J.M. and P. Bromiley, Critical FactorsAffecting the Planning and Implementation ofMajor Projects, Strategic Management Journal,(1993),14, 319-337.

Budner, S., "Intolerance of Ambiguity as aPersonality Variable", Journal of Personality,(1962), 30,29-50.

Butler, R., L. Davies, Pike, and J. Sharp, StrategicInvestment Decision-making: Complexities,Politics and Processes, Journal of ManagementStudies,(1991), pp. 395-415.

Child, J., "Organizational Structure, Environmentand Performance: The Role of StrategicChoice", Sociology, (1972), 6,1-22.

Cray,D.,G.R. Mallory, R.J. Butler, D.J. Hickson andD.C. Wilson "Sporadic, Fluid and ConstrictedProcesses: Three types of Strategic DecisionMaking in Organizations", Journal ofManagement Studies (1988),25,1,13-39.

Cyert, R.M., and J.G. March, A Behavioral Theoryof the Firm, Prentice-Hall, 1963.

Daft R.R. and Lengel, R.H., 1986. OrganizationalInformation Requirements, Media Richness andStructural Design. Management Science,32,(5):554-571.

Dean, J.W. Jr., and M.P. Sharfman, "ProceduralRationality in the Strategic Decision MakingProcess", Journal of Management Studies,(1993a),30,3,587-610.

Dean, J.W. Jr., and M.P. Sharfman, "TheRelationship Between Procedural Rationalityand Political Behaviour in Strategic DecisionMaking", Decision Sciences, (1993b),24,6,1069- 1083.

Dess, G.G. and D.W.Beard, Dimensions ofOrganizational Task Environments,Administrative Science Quarterly, (1984),pp.52-73.

Dess, G.G. and A.M.A. Rasheed, "Conceptualizingand Measuring Organizational Environments: ACritique and Suggestions", Journal ofManagement, (1991),17,4,701-710.

Dess, G.G., and N.K. Origer, "Environment,Structure and Consensus in StrategyFormulation: A Conceptual Integration",Academy of Management Review,(1987),12,2,313-330.

Dollinger, M., "Environmental Boundary Spanningand Information Processing Effects on

22

Organizational Performance", Academy ofManagement Journal, (1984),27,351-368.

Duncan, H., "Strategic Planning Theory Today",Optimum, (1990),20,4,63-74.

Dutton, J.E., "Interpretations on Automatic: ADifferent View of Strategic Issue Diagnosis",Journal of Management Studies,(1993),30,3,339-357.

Dutton, J.E., "The Processing of Crisis and Non-Crisis Strategic Issues", Journal of ManagementStudies, (1986), 25, 5, 501-517.

Dutton, J.E., E.J. Walton, and E. Abrahamson,"Important Dimensions of Strategic Issues:Separating the Wheat from the Chaff", Journalof Management Studies, (1989), 26,4,379- 396.

Dutton, J.E., L. Fahey and V.K. Narayanan,"Toward Understanding Strategic IssueDiagnosis", Strategic Management Journal,(1983),4,307-323.

Eisenhardt, K.M., "Making Fast Strategic Decisionsin High Velocity Environments", Academy ofManagement Journal, (1989), 32,3,543-576.

Eisenhardt, K.M. and M.J. Zbaracki, StrategicDecision-making, Strategic ManagementJournal, (1992), pp.17-37.

Eisenhardt, K.M., and L.J. Bourgeois, "Politics ofStrategic Decision Making in High-VelocityEnvironments: Toward a Midrange Theory",Strategic Management Journal, (1988),31,4,737-770.

Eysenck, H.J., and G. Wilson, Know Your OwnPersonality, PenguinLtd,Harmondsworth,Middlesex,England, 1975.

Finkelstein, S., and D. Hambrick, "TopManagement Team Tenure and OrganizationalOutcomes: The Moderating Role of ManagerialDiscretion", Administrative Science Quarterly(1990), 35, 484-503.

Fredrickson, J.W., "Effects of Decision Motive andOrganizational Performance Level on StrategicDecision Processes", Academy of ManagementJournal (1985),28,4, 821-843.

Fredrickson, J.W., "The Comprehensiveness ofStrategic Decision Processes: Extension,Observations, Future Directions", Academy ofManagement Journal, (1984), 27,3, 445-466.

Fredrickson, J.W., and A.L., Iaquinto, "Inertia andCreeping Rationality in Strategic Decision

Processes", Academy of Management Journal,(Sept, 1989), 32,4,516-542.

Guth, W.D. and I.C. MacMillan, "StrategyImplementation vs Middle Management Self-Interest", Strategic Management Journal,(1986), 7,313-327.

Gough, H., "Personality and PersonalityAssessment", In Dunnette, M.D., (Ed.),Handbook of Industrial and OrganizationalPsychology, Rand McNally, Chicago,(1976),571-607.

Grinyer, P., S. Al-Bazzaz, and M. Yasai-Ardekani,"Towards a contingency theory of corporateplanning Findings in 48