421. Characteristics of Organizational Environments and (1)

Eobert B, Duncan

Characteristics of Organizational Environments andPerceived Environmental Uncertainty

Twenty-two decision groups in three manufacturing and three research anddevelopment organizations are studied to identify the clmracteristics of the en-vironment that contribute to decision unit members experiencing uncertainty indecision making.^

Two dimensions of the environment are identified. The simple-complex dimen-sion is defined as the number of factors taken into consideration in decision mak-ing. The static-dynamic dimension is viewed as the degree to which these factorsin the decision unit's environment remain basically the same over time or are ina continual process of clmnge. Results indicate tlmt individuals in decision unitstvith dynamic-complex environments experience the greatest amount of uncer-tainty in decision making. The data also indicate that the static-dynamic dimen-sion of the environment is a more important contributor to uncertainty than thesimple-complex dimension.

Organizational theorists emphasize thatorganizations must adapt to their environ-ment if they are to remain viable.^ One ofthe central issues in this process is copingwith uncertainty (Crozier, 1964; Thompson,1967). The concept of the environment, withits components and relevant dimensions,however, has not been well specified in theliterature (Dill, 1958; Emery and Trist, 1965;Lawrence and Lorsch, 1967; Perrow, 1967;Thompson, 1967). If a theory of organization-environment interaction is to be developedto facilitate empirical research, it is necessar)^that the components and dimensions of theenvironment be more clearly defined. Thisis the object of the research presented here,

The components of the environment aswell as its specific dimensions, are identified,This identification, in turn, then facilitatesthe identification of types of environments—

^ The research on which this article is based wassupported by a National Science Foundation Grant,-\o. GS-3054. T_his assistance is gratefully acknow-ieagea. The author aJso wishes to acknowledge theht lpful comments of Clayton Alderfer, Chris Argyris,Ucwid Baron, Douglas T. Hall, and Lawrence Zahn.

•^ The author is also currently developing a modelof how organizations learn to adapt to their environ-ments (Duncan, 1971b).

that contribute to different degrees of un-certainty as perceived by individuals in-volved in decision making.^

Twenty-two decision units are studied inthree manufacturing organizations (ten de-cision units) and in three research and de-velopment organizations (twelve decisionunits). An organizational decision unit isdefined as a formally specified work groupwithin the organization under a superiorcharged with a formally defined set of re-sponsibilities directed toward the attainmentof the goals of the organization. Decisionmaking per se may be centered in the formalleader and/or distributed to various mem-bers of the specific unit. Decision makingfor this analysis is more broadly definedthan in most decision models to includethe gathering and processing of informationcarried out by groups of individuals, whichprecedes the actual choice process.

It should be emphasized that environ-mental uncertainty and the dimensions of

environment are defined here in terms

3 The author has also identified the types of struc-t^âl modifications decision units implement in mak-ing decisions under uncertainty and the relationshipbetween these adaptation processes and organiza-tional effectiveness (Duncan 1971c).

313

314 ADMINISTRATIVE SCIENCE QUARTERLY

of the perception of organization members.Research has indicated that there are dif-ferences among individuals in their percep-tions and tolerance for ambiguity or uncer-tainty (Adorno et al, 1950; Berlyne, 1968).Given the role individual difiEerences playin one's reaction to events, considerationwill be given in this research to the differ-ences among individuals in their perceptionof uncertaintv and the environmental dimen-sions before aggregate measures are con-structed.

THE ENVIRONMENT

One of the shortcomings of much of thetheoretical and empirical research on orga-nizational environments has been the failureclearly to conceptualize organization en\d-ronment or the elements comprising it (Law-rence and Lorsch, 1967; Thompson, 1967;Terreberr}^ 1968). Pugh et al. (1969) havestudied organizational contexts, that is, ori-gin and history, ownership-control, size,location, and so on—the settings withinwhich organization structure is developed.As they have specified, however, this is nota model of an organization in an environ-ment. Lawrence and Lorsch (1967), for ex-ample, have studied how organizations seg-ment their environment into related sectorsbut have not clearly conceptualized the en-vironment or its makeup. They have alsoconceptualized the environment as a totalentity but have looked onlv at the environ-ment from the organization outward (Law-rence and Lorsch, 1967: 4). Dill's (1958)concept of the task environment again fo-cused only on those parts of the organi-zation's external environment which wererelevant or potentially relevant to the orga-nization's goal setting and goal attainment.

In the present analysis environment isthought of as the totality of physical andsocial factors that are taken directly intoconsideration in the decision-making behav-ior of individuals in the organization.

If the environment is defined in this way,there are then factors within the boundariesof the organization or specific decision mak-ing units that must be considered as partof the environment. A differentiation ismade, therefore, between the system's in-ternal and external environment.

The internal environment consists of thoserelevant physical and social factors witninthe boundaries of the organization or spe-cific decision unit that are taken direcrlvinto consideration in the decision-makingbehavior of individuals in that system.

The external environment consists of thoserelevant physical and social factors outsidethe boundaries of the organization or spe-cific decision unit that are taken directhinto consideration.

This distinction between internal and ex-ternal environments is more comprehensivethan Rice's (1963) definition of the internalenvironment as including the interpersonalrelations of members and their interactionswith each other and the external environ-ment as including other individuals, groups,and institutions.

In an effort to identify environmental com-ponents, research was carried out in a large,industrial, manufacturing organization (Dun-can, 1968). Nineteen individuals in variousdecision units in several functional areasand organizational levels were interviewed.A semistructured interview focused on thenature of the decision unit's en\dronmentand the decision-making process. From thisresearch a list of environmental componentswas constructed comprising a decision unit'sinternal and external environment (Table 1).No decision unit is expected to identify allthe components in its particular internal andexternal environment.

The list of environmental components pre-sented in Table 1 may be partictilarly rele-vant to industrial organizations and may var}for other types of organizations. These find-ings go beyond existing research by moreclearly conceptualizing the environment andfactors comprising it. The empirical analysisof organization-environment interaction is.therefore, facilitated.

THE SIMPLE-COMPLEXDIMENSION

The next step was the identification ofthe environment's dimensions in order tomake predictions about the kinds of envi-ronments in which different levels of pti'ceived uncertainty are expected to exist. Inthe work of organizational theorists such as

and Trist (1965), Thompson (1967),

Duncan: ORGANIZATIONAL ENVIRONMENTS 315

TABLE 1. FACTORS AND COMPONENTS and Terreberr)' (1968), two environmentalCOMPRISING THE ORGANIZATION'S dimensions can be inferred: the simple-

iNTERNAL AND EXTERNAL complex dimension and the static-dynamicENVIRONMENT dimension. These dimensions were then con-

ceptualized for this research in the followingInternal environment manner.

TlTorganizational personnel component The simple part of the simple-complex(A) Educational and technological background dimension deals with the degree to which

and skills the factors in the decision unit's environment(B) Previous technological and managerial skill ^ j . ^ f ^ | j ^ number and are similar to one(G) Individual member's involvement and ^jêr in that thev are located in a few

commitment to attaining systems goals -- , , • j - 4.(D) Interpersonal behavior ttyles components. The complex phase indicates(E) Availability of manpower for utilization that the factors in the decision units envi-

within the system ronment are large in number.(2) Organizational functional and staff units com- An example of a decision unit with a sim-

ponent pie environment would be a lower-level pro-(A) Technological characteristics of organiza- ûction unit whose decisions are affected(B) Interdependence of organizational units in only by the parts and materials department

carrying out their objectives upon which it IS dependent tor supplies, and(G) Intra-unit conflict among organizational the marketing department upon which it is

functional and staff units dependent for determining its output vol-(D) Inter-unit conflict among organizational ^ ^ ^ ^ j ^ . ^ environment is thus homogeneous

functional and stait units . , , i . • ^ J _». t-, , , m that the parts and materials department

(3) Orgamzational level component ^^^ ^^ marketing department are both un-(A) Organizational obiectives and goals , , • . i 4. 4--U(B) Integrative process integrating individuals der the same environmental component, the

and groups into contributing maximally to organizational function and staff units (seeattaining organizational goals Table 1).

(G) Nature of the organization's product ser- ^ ^ example of a decision unit with a com-^Jf^ plex environment would be a decision unit

External environment in a programing and planning department.(4) Gustomer component ' These decision unit membei-s, when making

(A) Distributors of product or service a decision, may consider a wide variety ot(B) Actual users of product or service factors in botJi the internal and external

(5) Suppliers component environment. They may, for example, focus(A) New materials suppliers on the internal environmental factors of the(B) Equipment suppliers marketing and materials departments of the(G) Product parts suppliers organizational functional and staff unit com-(D) Labor supply ^^^^^ ^^ ^ ^ ^^^^ ^^^ ^^^^ ^^^ f ^ ^

(6) Gompetitor component ^^ ^^^ external environmental factors of(A) Competitors for suppliers customer demand of the customer compo-(B) Competitors for customers vûsLumci yxK. , . i j

, , nent, the availability of raw matenals and(7) Socio-political component ^^ ^^^ ^ j . ^ ^ component,

^ ^ S d T t ^ " ^ " ^ " and government regulatoi^ control over the(B) Public political attitude towards industry industr}^ of the socio-political component

and its particular product in formulating programs and plans for or-(G) Relationship with trade unions with juris- ganizational action. This decision unit ap-

diction in the organization preaches the complex end of the simple-8) Technological component complex dimension as there are a larger

(A) Meeting new technological requirements ^ f f ^^^j . N = 6 identified that areof own industry and related industries m " " " ^ ^ ' 'production of pVoduct or service dissimilar or heterogeneous to one another

(B) Improving and developing new products This is due to the fact that they are locatedby implementing new technological ad- jn. several different environmental compo-vances in the industry nents, N = 4 (see Table 1).


A specific simple-complex environmentalindex is then developed by multiplying thenumber of decision factors (F) identifiedby decision unit members that they consid-ered in decision making by the number ofcomponents (C)2. This product expressesthe contribution of both the number of fac-tors and the degree to which they are sim-ilar (found in one component) or are dis-similar (found in several components). Thus,the simple-complex environmental index =

) ( )Squaring the number of components is

an indicator of similarity-dissimilarity in thatthe more components the factors are in, themore dissimilar they are. This is expressedin (C)^ A decision unit with three factorsin one component, for example, would havea simple-complex environmental index of 3(3 X12 = 3). A unit with three factors inthree different components would have anindex of 27 (3 X 3^ = 27), indicating theincreased complexit}^ of their environmentas a function of the dissimilar nattire of thefactors. The rationale for squaring (C) isthat the amount of variance between com-ponents is greater than the amount of vari-ance between factors and, thus, should beweighted in the development of the index.

The simple-complex environmental indexfor the lower-level production unit would be2 X 12 = 2 (that is, the parts and materialsdepartment and the marketing department,which represents two factors and, since bothof these factors are in the organizationalfunctional and staff component, there is onecomponent represented).

The simple-complex environmental indexfor the programing and planning departmentdescribed above as having a complex envi-ronment would be 6 X 42 = 96 (that is,marketing department, materials department,customer demand, availability of raw mate-rials, availabilit)' of product parts, and gov-ernment regulatory control over the industry;this represents six factors and these factorsare in the organizational functional and staff,customer, supplier, and socio-political com-ponents which represent four components).

THE STATIC DYNAMIC DIMENSION

The static-dynamic dimension indicatesthe degree to which the factors of the deci-

sion unit's internal and external environmeniremain basically the same over time or artin a continual process of change. It is composed of two subdimensions. The first focuses on the degree to which the factorsidentified by decision unit members in theunit's internal and/or external environmentare stable, that is, remain the same overtime, or are in a process of change. For ex-ample, the environmental factors in a pro-duction decision unit which are always takeninto consideration in decision making maybe the marketing department and the mate-rials department. These factors would char-acterize a static environment if the market-ing department's requests for productionoutput remained stable and if the materialsdepartment was able to supply a steady rateof inputs to the production decision unit. Onthe other hand, if the marketing departmentwas continually changing its request for dif-ferent production outputs and the materialsdepartment was variable in its ability to sup-ply parts, these factors would characterize amore dynamic environment for the decisionunit. This particular subdimension of thestatic-dynamic dimension is measured byasking respondents how often each of thefactors that they identified as being impor-tant in decision making in their internaland/or external environment change. The re-sponse categories vary along a five-pointscale of: (1) never, (2) ahnost never,(3) sometimes, (4) frequently, and (5) ver}'often.

The decision units are then given a scoreon this subdimension from the average re-sponse of all the entities as changed. Individ-ual responses are averaged to form the unit'soverall score on this subdimension. Themembers of the production decision unit,for example, may indicate as a group thatthe marketing department is frequentlychanging its request for different products(score = 4) and that the materials depart-ment almost never changes in its input ofparts and supplies (score = 2). The totalscore on this first subdimension for the de-cision unit would be 2 + 4/2 = 3.

The second subdimension of the static-dynamic dimension focuses on the frequencywith which decision unit members take intoconsideration new and different internal


and/or external factors in the decision-naking process. An example of the secondsubdimension might be a production unitthat always takes the marketing departmentand the niaterials department into considera-tion in its decision making. This would con-tribute to a static environment as the envi-lonment was not changing; the same tvvofactors, marketing and materials, were con-tinually considered in decision making. Onthe other hand, a programing and planningdecision unit's environment would be moredynamic if the members of this unit indi-cated that they focused on a variety of dif-ferent factors over time. In developing pro-grams for one type of product, for example,they might focus on the customer demandand the production and marketing depart-ments. In planning and developing programsfor a different type of product, the relevantenvironment to be considered in decisionmaking mav have changed to include, inaddition to^ the marketing and productiondepartments, a different group of customers,possible government rlgulator^^ agencieswith jurisdiction over this t^rve of product.and the implications for this" new producton labor-management relations. Thus, therelevant environment tor this decision unitis changing. This particular subdimensionIS measured by asking respondents ot a giveDdecision unit how often thev consider newand different factors m decision maknig.Again the response categories varv along thesame hve-pomt scale.

The decision unit as a group then receivesthe raw score as indicated bv its members(individual decision unit members responsesare averaged to form the overall unit s score)In the programing and plannmg departmentthe members as a group may mdicate thatthe factors that they take into considerationm decision making change very often (score

The scores obtained on these two subdi-mensions of the static-dynamic dimensionof the environment are then added togetherto obtain the decision unit's static-dynamicindex. Units are then rank ordered accordingto index scores and split at the median. The•ligh scoring half of the distribution is de-fined as decision units having a dynamic

environment and the low scoring half asunits having a statie environment.

PERCEIVED UNCERTAINTYOne of the primar}^ tasks of this research

was to place environmental uncertainty ina logical framework so that it could be op-erationalized more effectively in the future.The concept of uncertainty' has been defined'« a variet\^ of ways in the literature. In-formation theorists such as Attneave (1959)and Camer (1962) have defined the concept''n a narrow fashion. Carner's (1962: 19)definition is representative in statmg that"the uncertainty of an event is the logarithmo the number of possible outcomes theevent can have . . . .

Decision theorists such as Knight (1921)Luce and Raiffa (1957) defined uncer-

^f those situations where the probabil-the outcome of events is unknown as

to risk situations where each out-J known probablllt>^ At the wider

of analysis, Lawrence and Lorschf^ 2 / ; state that unc'ertamt>. consists

three components: (1) the lack of claritv

feedback, and (3) the general un-certamtv or causal relationships.

In survevms; these diirerent concepts, itwas concluded that the wide dehnitions weretoo broad m scope and did not facihtate theoverall obiective or trying to dehne the con-cepts m the model more specihcallv so thatthe\ could be operationalized. Lawrence andLorsch (196 /) were vague with their dehni-tion or lack or information and general un-certamtv of causal relationships. This lackof ciaritv m definition then mlubits the de-

^p^^^g^ operational measuresaintv, which is an immediate objec-

^ ^^.^ research.narrower definitions of Carner (1962),

, and Luce and Raiffa (1957)tend to focus on the more mathematical as-p^^^^ f uncertainty such as the individual'sability or inability to assign probabilities toevents. It was concluded that, although thismay be an important component of uncer-tainty, there may be other components thatshould also be included. Another considera-tion in not adopting the narrower definitionof uncertainty was that the definition of un-


certaint}^ implemented here had to provide going to the moon or becoming a millionairean operationalization of the concept to which but he is less confident about the probabilityactual organizational members could respond, estimates than under conditions of risk. TheGiven the more specific mathematic defini- real question becomes one of how confidenttion of uncertainty by the information theo- the individual is in his estimate. In uncertainrists (Gamer, 1962; Attneave, 1959), it was situations there is less predictability withbelieved that these definitions were too ab- respect to the outcome of events than understract for managers to respond to. conditions of risk. The present definition of

Environmental uncertainty was defined on uncertainty thus takes a broader perspectivethe basis of the preliminar)^ research dis- and is concerned with the individual's abil-cussed above (Duncan, 1968). Although ity to assign probabilities. It builds on boththere was difficulty in the preliminary re- the wide and narrow formulations to providesearch in getting respondents to verbalize a definition that is both comprehensive andtheir views of uncertainty, there was a re- yet specific, so that it can be used in futuremarkable degree of similarity' in the way in research.which the concept was ultimately defined. TTxrrp^TiTATiVTY lV/fFAtiTTBTrThree components of uncertainty were men- ™^ UNCERTAINTY MEASUREtioned by some or all of the eighteen indi- Dimensions 1 and 2 of perceived environ-viduals who gave a definition: (1) the lack mental uncertainty are measured by scaleof information regarding the environmental items similar to those in the Likert systemfactors associated with a given decision- (Duncan, 1971a). The first dimension—lackmaking situation, (2) not knowing the out- of information regarding the environmentalcome of a specific decision in terms of how factors associated with a given decision-mak-much the organization would lose if the deci- ing situation—contains six scale items ofsion were incorrect, and (3) inability to which the following is an example: how oftenassign probabilities with any degree of con- do you believe that the information you havefidence with regard to how environmental about this factor is adequate for decisionfactors are going to affect the success or making?failure of the decision unit in performing its The second dimension—not knowing thefunction. outcome of a specified decision in terms of

The first two components focus on the how much the organization would lose ifgeneral lack of information that is involved the decision were incorrect—is composed ofin decision making. This is similar to Law- six scale items. An example is: how often dorence and Lorsch's (1967) broad formula- you feel you are unable to predict how thistion. The third component in the present factor is going to react to or be affected b)study is similar to narrower mathematical decisions made in this group?definitions in its focus on assigned prob- The response categories varied along aabilities, but it does differ in a fundamental five-point scale of: (1) never, (2) seldom,way. Decision theorists (Knight, 1921; Luce (3) occasionally, (4) fairly often and (5) al-and Raiffa, 1957) have normally defined un- ways. The factors taken into considerationcertainty as a situation where the individual in decision making were identified by ancannot assign probabilities to the outcome of interview prior to administering the ques-events. The third component of uncertainty tionnaire. The environmental components indefined above seems to indicate that this Table 1 were used as a guide in this inter-definition is too restricted. The individuals view. In the interview decision unit membersin this preliminary research indicate that might indicate that they took Factor A (ex-they could assign probabilities to the out- production department), Factor B (exmar-come of events but that in uncertain situa- keting department). Factor C (excustomertions the question becomes one of how sure demand) into consideration in decisionor confident the respondent is in his prob- making. Each decision unit member wasability assessment. Thus, in uncertain situa- then asked to answer each question in thetions the individual can still assign proba- scales for the first two dimensions of per-bilities to the outcome of events, for instance, ceived environmental uncertainty for each


I the factors taken into consideration in de- the individual's degree of ability to assigncn making. Individuals then received an probabilities as to the effect of a given factor

ivei age score on each of the questions on on the success or failure of the unit in per-ie scales for the first and second dimensions forming its function. For each factor he indi-V means of the following formula: cates that he takes into consideration in deci-

total score on a given sum of answers for

number of factors taken into consideration

The third dimension of perceived environ-mental uncertaint)^ deals with the respon-dent's ability or inability to assign probabil-ities as to the effect of a given factor on thesuccess or failure of a decision unit in per-oiming its function. There are two com-ponents to the question that measures thisdimension. First, the respondent was askedto indicate on a scale how sure he was ofhow each of these factors was going to af-fect the success or failure of his work groupin canning out its function. The scale was asoUows:

ompletely unsure completely sure0.10.2^0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

This is the probability assigning task thatthe strict definition of uncertaint}^ in theliterature (Knight, 1921; Luce and Raiffa,1957) indicates that the individual cannotBo. The assumption was, however, that even|n uncertain situations, the individual may

l be able to assign some probability esti-mates. The real question then becomes oneM how confident he is in this estimate.^

The second part of the question for eachrespondent, after he assessed his certainty,has what range of numbers he was consider-ing between 0 and 1.0 in indicating this cer-laintv, that is, how confident was he in hisestimate. For example, if a person answered(that he was 0.3 sure about a factor, what wasHie range he was considering in giving this|answer? Was it between 0.2 and 0.4 or 0.1and 0.7, or 0 and 1.0, and so forth? A wideSpread in this range would indicate lack ofConfidence in his probability assessment.

This question is then scored to deteiinine

^ Some direct support of this procedure is foundRaiffa (1968: 161-168); Professor Gerrit Wolf

f Yale University was also helpful in the develop-nt of this measure.

sion making, he receives a score measuringhis degree of abilit)^ to assign probabilitiesas to the effect of that factor on the decisionunit's performance. This score is derived byweighting his certainty about the effects ofa given factor (part one of the question) bythe range between 0 and 1.0 he considers inmaking this assessment (part two of thequestion). The specific formula is as followswith larger scores indicating greater abilityto assign probabilities:

degree of ability to assign probabilities =(certainty of effects of factor) X

(1-range of certainty estimate)

If a person, for example, responds by indi-cating he is 0.3 sure about the effects ofFactor A on the performance of his workgroup and the range he is considering ingiving this answer is between 0 and 0.5, hisdegree of ability' to assign probabilities scorefor this factor would be 0.3 X (1-0.5) = 0.15.

The respondent's total score for this ques-tion is then ayeraged for the number offactors taken into consideration in decisionmaking which may yar)^ from indiyidual toindiyidual:

sum of degree of ability to assignprobabilities scores for all factors

identifiednumber of factors identified

The scores of the three components of un-certainty are added to form a total uncer-tainty score. The rationale is that these threecomponents are conceptually related in thesense of representing a general lack of infor-mation about the enyironment.

HYPOTHESESBoth the simple-complex and static-dy-

namic dimensions are important in deter-


mining the state of the decision unit'senvironment. By considering the interactionof these two dimensions, different states ofthe decision unit's environment can be identi-fied. Once these are identified, predictionscan then be made as to the degree of per-ceived environmental uncertainty expected toexist in these different tvpes of environments.

-ansion so that the outcome of the decisionbe assessed.

Hypothesis two. Decision units with com-plex-dynamic environments (Cell 4) willexperience the greatest perceived environ-mental uncertainty.

In environments characterized by complex-dynamic dimensions where a large number

Table 2 represents a conceptualization of of changing factors differ from one another.

TABLE 2. ENVIRONMENTAL STATE DIMENSIONS AND PREDICTED PERCEIVED UNCERTAINTYEXPERIENCED BY INDIVIDUALS IN DECISION UTSTITS

Simple

Gelll:low perceived uncertainty

Static

Dynamic

(1) Small number of factors and compo-nents in the environment

(2) Factors and components are somewhatsimilar to one another

(3) Factors and components remain basi-cally the same and are not changing

Gell3:moderately high perceived uncertainty

(1) Small number of factors and compo-nents in the environment

(2) Factors and components are somewhatsimilar to one another

(3) Factors and components of the en-vironment are in continual process ofchange

Gomplex

Gell2:moderately low perceived uncertainty

(1) Large number of factors and compo-nents in the environment

(2) Factors and components are not similarto one another

(3) Factors and components remain basi-cally the same

Gell4:high perceived uncertainty

(1) Large number of factors and compo-nents in the environment

(2) Factors and components are not similarto one another

(3) Factors and components of environmentare in a continual process of change

these two dimensions for the combined in-ternal and external environment. Thus, Table2 is a simplified presentation in that there isno distinction being made between the in-ternal and external environment.^From this simplified typolog)' the followinghypotheses can be derived.

Hypothesis one. Decision units withsimple-static environments (Cell 1) will ex-perience the least perceived environmentaluncertainty. In environments that are char-acterized by simple-static dimensions wherethere is a smaller number of relatively similar,unchanging factors considered in decisionmaking, little uncertainty is expected to exist.Here, decision unit members are predictedto be able to have the relevant informationregarding the factors associated with a deci-

o The sample size of this research did not allow forthis more differentiated analysis (Duncan 1971a).

uncertaint\' is predicted to be high. BotliThompson (1967), in his theoretical analysisof organizational adaptation, and Ud\(1959), in his comparative research, indi-cated that environmental uncertainty in-creased when organizational environmentswere changing and heterogeneous. Here, itis predicted that decision unit members willnot have the relevant information availablefor the factors associated with a decision, sothe outcome of the decision cannot be as-sessed.

Hypothesis three. Decision units withsimple-dynamic environments (Cell 3) ^ illexperience greater perceived environmentaluncertainty than individuals in decision unitswith complex-static environments (Cell 2)Decision units having simple-dynamic en-vironments (Cell 3) are predicted to have amore difficult time obtaining the relevantinformation for decision making than units

Duncan: ORCANIZATIONAL ENVIRONMENTS 321

having complex-static environments (Cell 2)because of the continually changing natureof factors in their environment. As a result,they were predicted to experience greaterperceived environmental uncertainty thandecision units in Cell 2 type of environments.

The implication of the third hypothesis isthat the static-dynamic dimension is a moreimportant contributor to uncertainty thanthe simple-complex dimension. In a dynamicenvironment, where the factors taken intoconsideration in decision making are con-tinually changing, it is going to be difficultto have available the relevant informationfor the decision-making situation. When theenvironment is changing, the system mustcontinually learn to readapt. The systemcannot rely on past procedures and practices;rather, it is faced with a new situation inwhich its members will have to learn newmethods. In a dynamic environment the sys-tem is faced with many possible outcomeswhereas with an unchanging, static environ-ment, there is onlv a finite number of out-comes to events. This will exist regardless ofwhether there are many or few factors takeninto consideration in decision making, thatis, whether the decision unit's environmentis simple or complex. The result is that mod-erate to high levels of perceived environ-mental uncertainty were predicted to existfor decision unit members. This predictionis consistent with both the work of Env^^vand Trist (1965) and Terreberr\^ (1968).They found that organizations in dxnamic-turbulent environments often exceeded theircapabilities for prediction and control withthe result that the outcome of events becameless certain.

METHODOLOGY

Unit of Analysis

Since the unit of analysis is the organiza-tional decision unit, responses obtained fromdecision unit members on all the items ona variable are pooled to reflect the degree ofthe given variable experienced bv the unitas a whole. This is accomplished in threesteps. First, a mean score on each of theitems of a variable is computed for eachtype of social role in the decision unit.Second, the decision unit's score for a given

item is then determined by computing theaverage of all social role means in the uniton the given item. Third, the decision unit'stotal score on a variable is then computedby adding the scores on the items makingup the variable. For a more complete dis-cussion of this pooling, see Lazarsfeld andMenzel (1960) and Hage and Aiken (1967).

Pooling Responses

One of the initial problems in poolingperceptual measures is to determine the de-gree of variance among individuals makingup the group. Are individuals responding tothe variable under measurement in the sameway or are individuals in the group respond-ing differently to the same variable? If thereare large variances in the way individuals areresponding, it is difficult to pool individualresponses to represent the group as a whole.

A five-point change scale was used to in-vestigate these questions: (1) never, (2) al-most never, (3) sometimes, (4) frequently,and (5) very often. If in a group composedof individuals Ii, I2, Ia, h responds with 1 indi-cating very low change, 12 responds with 3indicating medium change, and I3 respondswith 5 indicating a very high level of changeexperienced tbe averasje pooled score for the

1 + 3 + 5group would be = 3 = median

level of cbange experienced in this group.This pooled group score would be misleadinggiven the wide variance in the wav individ-uals are exneriencing change in their en-vironment. Thus, in assessing perceptualmeasures, consideration must be given todetermining how individual members re-spond to the same phenomenon. Schneiderand Bartlett (1970) support this view in theirresearch on perceptual measures of organiza-tional climate. Their research indicated alack of congruence between the manager'sand agent's view of climate in life insuranceagencies. Forehand and von Haller Gilmer(1964) in their analysis of environmentalvariation in organizational behavior havealso indicated that for individual scores tobe pooled to represent a group score, theremust be some evidence that the dimensionunder consideration is perceived similarlyby all those in the group.


To assess the homogeneity of group mem- 4 on the simple-complex and static-dynamicbers' perception of a particular variable, one- dimensions. The results of this analysis con-way analysis of variance was computed across firm that in this sample, research and devel-individuals in a given decision unit to dis- opment organizations have more complexcover any significant differences among indi- (t = 4.388, p < 0.001) and dynamic (t =vidual perceptions. This was done before 3.453, p < 0.01) environments than manu-individual scores were pooled to get total facturing organizations.decision unit scores on the simple-complex Civen the wide differences in environ-and static-dynamic emdronmental dimensions ments between these two types of organiza-and on the perceived environmental uncer- tion, it is important to consider whether ittainty variable. The data indicate no signifi- is the nature of the environment or type ofcant differences across individuals in groups organization that is most important in a deci-for the simple-complex environmental dimension unit experiencing uncertainty. It hassion and perceived uncertaint}^ On the static- been indicated that the initial statistical con-dvnamic environmental dimension, in one trol for organization type had to be elimi-of the twenty-two groups in the sample, nated because of zero entries in some cellsindividuals exhibit a significant difference in Table 3.in their perception of this dimension of their A somewhat rougher analysis is performedenvironment (F = 7.630 p < 0.01). Civen to indicate the amount of variance in uncer-the general homogeneity of group member tainty that is explained by environmentalperceptions on the variables, however, indi- t)'pe and organizational type. A T-test be-vidual responses are summed to form group tween the mean amount of uncertainty ex-scores, perienced by manufacturing and research

and development organizations is computed.iit!iaui-.ic> 'pjjg amount of variance explained for un-

Before specific hypotheses derived from certainty by organizational t}'pe is then iden-Table 2 are tested, a constraint of the re- tified by computing co-, omega squaredsearch sample must be considered. Type of (Hays, 1963: 324-332).organization was initially controlled for in To indicate the amount of variance ex-the data collection. Examination of the distri- plained by environmental type, a one-waybution of the sample on the simple-complex analysis of variance (Table 6) is performedand static-dynamic environmental dimensions across the four different t}^pes of environmentin Table 3/however, indicates that it is im- in Table 3 and then co (omega squared) ispossible to control for organizational type in computed to indicate the amount of variancethe statistical analysis because of the zero explained for uncertainty by environmentalentrs' of research and development organiza- type (Hays, 1963: 381-384).tions in Cell 1 and the zero entry of manu- The data presented in Tables 5 and 6 indi-facturing organizations in Cell 4. cate that organizational type explains ap-

In this particular sample, there were no proximately 30 percent of the variance inresearch and development organizations with uncertainty, while environmental type ex-simple-static environments and no manufac- plains 70 percent of the variance. The con-turing organizations with complex-dynamic elusion, then, is that it is the nature of theenvironments. This confirms the idea that organization's environment rather than thedifferent organizations operate in different kind of organization that is most importantenvironments and seems to suggest that in explaining the degree of uncertainty ex-manufacturing organizations tend to have perienced in decision making,more simple and static environments, while In testing the hypotheses derived fromresearch and development organizations tend Table 3, a 2 X 2 (simple-complex X static-to have more complex and dynamic environ- dynamic) analysis of variance is performed,ments. T-tests between the means of the The results of that analysis are presented inmanufacturing and research and develop- Table 7. The multiple comparisons of thement organizations are performed in Table four cell means based on the a priori pre-


TABLE 3. DISTRIBUTION OF DECISION UNITS SAMPLED ON SIMPLE-COMPLEX ANDSTATIC-DYNAMIC ENVIRONMENTAL STATE DIMENSIONS

Static

Simple

Celll:

manufacturingorganizations

# 1# 2# 3# 4# 5# 7

research anddevelopmentorganizations

Complex

Cell 2:


#6


#20#22#23

#10N = 7 N = 4

Cell 3: Cell 4:


# 9#11


#13#14



#12#15#16#17#18#19#21

Dynamic

N = 4 N = 7

TABLE 4. MEAN DIFFERENCES BETWÊEN MANUFACTURING AND RESEARCH ANDDEVELOPMENT ORGANIZATIONS ON THE SIMPLE-COMPLEX AND STATIC-DYNAMIC

ENVIRONMENTAL DIMENSIONS

Number of Standard Student's Degree ofobservations Means de\iation T-test freedom

Bartlett'stest results

Research anddevelopmentorganizations

Manufacturingorganizations



12

10

12

10

Simple-complex variable

600.25 117.738 4.388*

296.00 203.277

Static-dynamic variable

6.774 1.092 3.453f

5.014 1.301

20 Chi Sq. = 2.775(1 D.F.) varianceassumed equal;

pooled variance =26218.912

20 Chi Sq. = 0.2925(D.F.) variance as-sumed equal;

pooled variance ==1.417

* p < 0.001t p < 0.01


TABLE 5. MEAN DIFFERENCES BETWEEN MANUFACTURING AND RESEARCH AND

DEYELOPMENT ORGANIZATIONS IN PERCERVED UNCERTAINTY



* p < 0.01co2 = 0.299

Number ofobservations

12

10

Means

43.967

29.610

Standarddeviations

12.345

7.354

Student'sT-test

3.224*

Degree offreedom

20

Bartlett'stest results

Chi Sq. = .235 (1D.F.) variances assumed equal;pooled variance ~108.150

TABLE 6. ANALYSIS OF VARIANCE

ENVIRONMENTAL TYPE AND

PERCEIVED UNCERTAINTY

Source

BetweenWithin

C02 =

0.0010.707

Meansquare

829.89644.316

Degree offreedom

318

F-test

18.727*

dieted differences in different levels of un-certaint}^ to be experienced in the differenttypes of enyironments is presented in Table 8.

Tables 7 and 8 confirm the first hypothesisin that decision units with static-simple en-^dronments experience the least amount ofperceiyed uncertainty^ (25.960). This is sig-nificantly lower than the groups in Gells 3and 4, while not significantly lower than thegroups in Cell 2.

Tables 7 and 8 confirm the second hypo-

thesis in that decision units with dynamic-complex enyironments experience, on theayerage, the greatest degree of perceiyedenyironmental uncertainty (51.729), whichis significantly different from all the otherthree types of enyironment (Cells, 1, 2, and3).

Tables 7 and 8 proyide some support forthe third hypothesis. Decision units withsimple-dynamic environments (Cell 3) ex-perience, on the ayerage, a greater leyel ofperceiyed uncertaint)^ (38.337) than groupswith complex-static enyirorunents (Cell 2)(31.635). Table 8, howeyer, indicates thatthe difference between the means for Cells 2and 3 is not significant. Other data do pro-yide support for the general implication ofthe hypothesis that the static-dynamic di-mension of the enyironment is a more im-portant contributor to the perception ofuncertainty. The results of the two-wa\analysis of yariance presented in Table 7

TABLE 7. ANALYSIS OF yARiANCE, SIMPLE-COMPLEX AND STATIC-DYNAMIC

EN\ÎRONMENTAL STATE DIMENSIONS UPON PERCEIVED ENyiRONMENTAL

UNCERTAINTY

Source ofvariance

A = simple-complexB = static-dynamicA X BTotal between cellsTotal within cells

* p < 0.01Rank order of cell means predictedlow to high uncertainty:

Cell I: static-simpleCell II: static-complexCell III: dynamic-simpleCell IV: dynamic-complex

Sum ofsquares

462.6561,341.926

75.7751,880.357

797.693

Degree offreedom

1113.00

18.00

Meansquare

462.6561,341.926

75.775

44.316

ObservedSimple

Static 125.960

Dynamic III 38.337

F ratio

10.438*30.504*

1.710

cell means:ComplexII 31.735

IV 51.729


TABLE 8. MULTIPLE COMPARISONS OF CELL

MEANS OF UNCERTAINTY, SIMPLE-COMPLEX

AND STATIC-DYNAMIC ENVIRONMENTAL

DIMENSIONS ANALYSIS

Gellscompared

I & IVI & IIII & IIII & IVII & IIIIII & IV

Differencebet^veen cells

25.76912.3775.675

10.0946.702

13.392

Griticaldifferencefor t, 18

degrees offreedom

p < 0.01p < 0.01

NSp < 0.01

NSp < 0.01

indicate that the static-dynamic main effectis much higher (F =" 30.504) than thesimple-complex main effect (F = 10.438).The potency of the static-dynamic dimensionin contributing to perceived uncertainty isfurther enhanced by the insignificant inter-action effect between the two environmentalstate dimensions.

Inspection of the multiple comparisons inTable 8 provides additional information asto the importance of the static-dynamic di-mension. The comparison between static-simple (Cell 1) and static-complex (Cell 2)environments indicates no significant dif-ference in the amount of uncertainty expe-rienced by decision units with these typesof environments. Comparison between d\' -namic-simple (Cell 3) and dynamic-complex(Cell 4) environments, however, indicatesa significant difference in the amount of un-certainty experienced. Thus, the differencein perceived uncertainty between static anddynamic environments is always significantregardless of whether the environment issimple or complex. Difference in perceiveduncertainty between simple and complexenvironments is contingent upon the environ-ment being dynamic. Thus, it appears thatthe complexity' of the decision unit's environ-ment does not have much impact on uncer-tainty until those factors considered in deci-sion making begin to change, that is, becomedynamic.

SUMMARY

Two dimensions of the environment areidentified. The simple-complex dimension is

defined as the number of factors taken intoconsideration in decision making. The static-dynamic dimension is defined as the degreeto which these factors in the decision unit'senvironment remain basically the same overtime or are in a continual process of change.Results indicate that individuals in decisionunits experiencing d}'namic-complex environ-ments experience the greatest amount of un-certainty in decision making. The data alsoindicate that the static-dynamic dimensionof the environment is a more important con-tributor to uncertainty than the simple-com-plex dimension. Decision units with dynamicenvironments always experience significantlymore uncertainty' in decision making regard-less of whether their environment is simpleor complex. The difference in perceived un-certaint)^ between decision units with simpleand complex environments is not significant,unless the decision unit's environment is alsodynamic.

This finding is somewhat consistent withthe theoretical work of Thompson (1967),Terreberry (1968), and Emery and Trist(1965) but, in addition, provides the firstsystematic conceptualization and empiricalanalysis of the dimensions of the environ-ment that lead to different degrees of per-ceived uncertaint).

It is also emphasized that uncertaint}' andthe degree of the complexity and dynamicsof the environment should not be consideredas constant features in an organization.Rather, they are dependent on the percep-tions of organization members and thus canvary in their incidence to the extent thatindividuals differ in their perceptions. Someindividuals may have a very high tolerancefor ambiguit}^ and uncertainty so they mayperceive situations as less uncertain thanothers with lower tolerances.

Future research should thus focus on theinterface between individual differences andorganizational properties. If the view thatan organization has no properties aside fromthe way people perceive it (Hunt, 1968) isgiven some credence, we need to begin toidentify more clearly how individual dif-ferences affect perceptions of organizationalproperties. For example, the research re-ported here should now be expanded tolook at the impact of individual differences


on the perception of uncertainty and thecomplexity and dynamics of the organiza-tion's environment. This research would helpdevelop a more comprehensive contingencytheory of organizations. Most contingencytheories now tend to be one sided (Bumsand Stalker, 1961; Duncan, 1971c; Lawrenceand Lorsch, 1967) in that they focus on thecharacteristics of the environment or tasksituation while ignoring an equally importantcontingent factor of individual differencesamong organizational members. It is onlyby beginning to focus on these individualdifferences that we can begin to develop ourcontingency theory more fully.

Robert B. Duncan is assistant professor oforganization behavior at the Graduate Schoolof Management, Northwestern University.

REFERENCES

Adorno, T., Else Frenkel-Brunswick, DavidLevinson, and Richard Sanford

1950 The Authoritarian Personality. NewYork: Harper & Row.

Attneave, Fred1959 Applications of Information Theory to

Psychology. New York: Holt, Rine-hart & Winston.

Berlyne, Donald E.1968 "The motivational significance of col-

lative variables and conflict." In R.Abelson, E. Aronson, W. McCuire, T.Newcomb, M. Rosenberg, and P. Ten-nenbaum (eds.), Theories of Cogni-tive Consistency: A Sourcebook: 257-267. Chicago: Rand McNally.

Bums, Thomas, and Cerald Stalker1961 The Management of Innovation. Lon-

don: Tavistock.Crozier, Michel

1964 The Bureaucratic Phenomenon. Chi-cago: University of Chicago Press.

Dill, William1958 "Environment as an influence on man-

agerial autonomy." AdministrativeScience Quarterly, 2: 409-443.

Duncan, Robert B.1968 "Preliminary report on the pretesting

of: the effects of internal and externalperceived environmental uncertaintyon the flexibility of organizational de-cision unit structures." Unpublishedpaper, Yale University, Departmentof Administrative Sciences.

1971a "The effects of perceived environ-mental uncertainty on organizationaldecision unit structure: a cyberneticmodel." Ph.D. dissertation, Yale Uni-versity.

1971b "Toward an operant model oforganization learning: adaptation toenvironmental uncertainty." WorkingPaper #40—71, Northwestern Univer-sity, Graduate School of Management.

1971c "Multiple decision making structuresin adapting to environmental un-certainty: the impact on organiza-tional effectiveness." Working Paper#54—71, Northwestern University,Graduate School of Management.

Emery, F.E., and E. Trist1965 "The causal texture of organizational

environments." Human Relations, 18:21-31.

Forehand, C , and B. von Haller Cilmer1964 "Environmental variation in studies

of organizational behavior." Psycho-logical Bulletin, 62: 381-382.

Camer, William1962 Uncertainty and Structure as Psy-

chological Concepts. New York:John Wiley.

Hage, Jerald, and Michael Aiken1967 "Relationships of centralization to

other structural properties." Admini-strative Science Quarterly, 12: 72-93.

Hays, William1963 Statistics for Psychologists. New York:

Holt, Rinehart & Winston.Hunt, Raymond

1968 Review of E. J. Miller and A. K. Rice,Systems of Organization. Administra-tive Science Quarterly, 13: 360-362.

Knight, F.1921 Risk, Uncertainty and Profit. New

York: Harper & Row.Lawrence, Paul and Jay Lorsch

1967 Organization and Environment: Man-aging Differentiation and Integration.Boston: Harvard University, Gradu-ate School of Business Administration,Division of Research.

Lazarsfeld, Paul and P. Menzel1960 "On the relation between individual

and collective properties." In A.Etzioni (ed.). Complex Organiza-tions: A Sociological Reader: 422—441. New York: Holt, Rinehart &Winston.

Luce, Robert Duncan, and Howard Raiffa1957 Cames and Decisions. New York:

John Wiley.


perrow, Charles1967 "A framework for the comparative

analysis of organizations." AmericanSociological Review, 32: 194-208.

Pugh, D. S., D. Hickson, C. Hinings, and C.Turner

1969 "The context of organization struc-tures." Administrative Science Quar-terly, 14: 91-115.

Raiffa, Howard1968 Decision Analysis: Introductory Lec-

tures on Choices under Uncertainty.Reading, Mass.: Addison-Wesley.

Rice, A. Kenneth1963 The Enterprise and Its Environment.

London: Tavistock.Schneider, B., and C. Bartlett

1970 "Individual differences and organiza-

tional climate II: measurement oforganizational climate by the multi-trait-multi-rater matrix." PersonnelPsychology, 23: 493-513.

Terreberry, Shirley1968 "The evolution of organizational

environments." Administrative ScienceQuarterly, 12: 377-396.

Thompson, James D.1967 Organization in Action. New York:

McGraw-Hill.Udy, Stanley

1959 Organization of Work: A Compara-tive Analysis of Production amongNonindustrial People. New Haven;Yale University. Human RelationsArea Files.

Documents

421. Characteristics of Organizational Environments and (1)