Role Conflict and Ambiguity in Complex Organizationscyb.ox.or.kr/lms_board/bbs_upload/%C1%F7%B9%AB%BD%BA%C6%AE%B7%B9...Complex Organizations The literature indicates that dysfunctional

John R. Rizzo, Robert J. House, and Sidney I. Lirtzman

Role Conflict and Ambiguity in

Complex Organizations

The literature indicates that dysfunctional individual and organizational conse-quences result from the existence of role conflict and role ambiguity in complexorganizations. Yet, systematic measurement and empirical testing of these roleconstructs is lacking. This study describes the development and testing of ques-tionnaire measures of role conflict and ambiguity. Analyses of responses of mana-gers show these two constructs to be factorially identifiable and independent.Derived measures of role conflict and ambiguity tend to correlate in two samplesin expected directions with measures of organizational and managerial practicesand leader behavior, and toith member satisfaction, anxiety, and propensity toleave the organization.

The purpose of this paper is to report onthe development of a questionnaire consist-ing of factorially independent scales de-signed to measure role confiict and ambiguityin complex organizations.

The questionnaire was developed for useas part of a broader survey to identify man-agement development needs and barriers forthe effective implementation of a plannedmanagement-development program in a largemanufacturing company (House, 1967: ch. 4and appendix I; House, 1968). Question-naires were developed and administered totwo samples of employees to measure em-ployee need satisfaction, job-induced anxiety,leader behavior, organizational and manage-ment practices, and role confiict and ambi-guity. This paper reports a construct valida-tion of scales designed to measure roleconfiict and ambiguity against demographicdata and other variables measured as part ofthe survey.

PREVIOUS LITERATURE ON ROLECONFLICT AND AMBIGUITY

Theory

In classical organization theory the prin-ciple of chain of command and the principleof unity of command and direction have im-plications for role conffict in complex organi-zations. Even though these principles are not

new to modern organizations, they have beensubjected to empirical testing in only a smallnumber of situations.

Aecording to the chain-of-command prin-ciple, organizations set up on the basis ofhierarchical relationships with a clear andsingle How of authority from the top to thebottom should be more satisfying to mem-bers and should result in more effective eco-nomic performance and goal achievementthan organizations set up without such anauthority fiow. Theoretically, such a singlechain of command not only provides topmanagement with more effective control andcoordination, but is also desirable becauseit is consistent with the principle of unity ofcommand.

The principle of unity of command statesthat for any action an employee should re-ceive orders from one superior only, and thatthere should be only one leader and one planfor a group of activities having the same ob-jective. The essence of this prineiple is thatthe structure of an organization should keepa member from being caught in the crossfireof incompatible orders or incompatible ex-pectations from more than one superior. Acorollary principle advanced by a more re-cent classical theorist (Davis, 1951) is theprinciple of single accountabiHty, whichstates that a person should be accountablefor the successful execution of his tasks to

150

Rizzo et al: ROLE CONFLICT AND AMBIGUITY 151

one and only one superior. Sueh an arrange-ment is intended to ensure systematie andeonsistent reporting, evaluation, and eontrolof the work of the subordinate. This preventsthe alloeating of time and effort aeeordingto individual preferenees, rather than aeeord-ing to tbe demands of the task, or the diree-tions of superiors. Subordinates eannot playone superior against another, and thus pre-vent aeeurate evaluation of performanee byambiguous or misleading reporting of per-formanee and alloeation of time.

Role theory states that, when the behaviorsexpected of an individual are inconsistent—one kind of role conflict—he will experieneestress, become dissatisfied, and perform lesseffectively than if the expectations imposedon him did not conflict. Role conflict cantherefore be seen as resulting from violationof the two classical principles and causingdecreased individual satisfaction and de-creased organizational effectiveness.

Both classical organization theory and roletheory deal with role ambiguity. Accordingto classical theory, every position in a formalorganizational structure should have a speci-fied set of tasks or position responsibilities.Such specification of duties, or formal defini-tion of role requirements, is intended toallow management to hold subordinates ac-countable for specific performance and toprovide guidance and direction for subor-dinates. If an employee does not know whathe has the authority to decide, what he isexpected to accomplish, and how he will bejudged, he will hesitate to make decisionsand will have to rely on a trial and errorapproach in meeting the expeetations of bissuperior.

Role theory likewise states (Kahn et al.,1964) that role ambiguity—lack of the nec-essary information available to a given or-ganizational position—will result in copingbehavior by the role incumbent, which maytake the form of attempts to solve the prob-lem to avoid the sources of stress, or to usedefense mechanisms which distort the realityof the situation. Thus, according to roletheory, ambiguity should increase the prob-ability that a person will be dissatisfied withhis role, will experience anxiety, will distortreality, and will thus perform less effectively.

Previous Research

Relation of unity of command to membersatisfaction and performance. Are organiza-tions with a clear and single flow of author-ity from the top to the bottom less charac-terized by role conflict among their membersand more satisfying for their members; anddo such organizations enjoy more effeetiveand economic performance and goal achieve-ment than organizations without a singleclear flow of authority? Altliough not muchresearch has been conducted to compare therelative effeetiveness of organizations thatviolate the chain-of-command principle andthose that observe it, one can draw infer-ences from the experience of organizationsthat have more than one flow of authority.

Professional organizations frequently ex-hibit violations of the chain-of-commandprinciple. As Blau and Scott (1962) pointedout, two sources of authority exist when or-ganizational discipline is based not only onposition power—supported by formal sanc-tions, and derived from the legal contractgoverning employment of the organizationalmember and the formal sanctions vested inthe superior's position—but also on profes-sional expertise which is enforced by colle-gial authority. Several studies have shownthat (J) multiple authority disrupts the in-dividual's orientation to his organization orto his profession by requiring him to choosebetween the two (Kaplan, 1959; Etzioni,1959; La Porte, 1965; Evans, 1962; Reiss-man, 1949; Gouldner, 1958a; 1958b); (2)individuals oriented primarily toward theirprofessional norms are more critical of theorganization and more likely to ignore ad-ministrative details (Blau and Scott, 1962);and (3) professionals in sucb organizationsfrequently experience stress as result of be-ing caught in the middle (Kaplan, 1959; LaPorte, 1965; Evans, 1962).

These findings are reinforced by researchconducted in a completely diflerent kind oforganization setting, namely hospitals. Con-cerning hospital hierarchies, Perrow (1965:957) stated:

The differences between hospitals and most or-ganizations are obvious, and a basic one thatresearchers have focused upon is a system ofmultiple authority or multiple subordination.

152 ADMINISTRATIVE SCIENCE QUARTERLY

Speeifically, the reference is to the administra-tive and medical hierarchies, with nurses as theprime example of groups who are caught be-tween the two lines of authority. Another theme,again concerning the anomalous position of thedoctor, is the way the doctor appears to be botha staff member (providing a specialized skillthat is required outside the organization) and amember of the "line" (doing the productive ef-fort). Finally, there is the anomalous positionof the doetor as a "guest" in the house, bringingin his own patients and using the faeilities butnot being a bona fide member of the organiza-tion in the usual sense of being reimbursed byit and subjeet to its authority; yet also being amember in that he exercises control, serves ina bureaucratie role, and to some degree has hisown interests tied to the interest of the hospital.Sueh deviations from the Weberian (classical)model of bureaucracy . . . . have fascinatedsoeial scientists.

Zawacki (1963) found that role conflict re-sults from the dual hierarehy of hospitalsand that those afleeted respond with hostilityto physicians and passive resistance to formalrules. Ben-David (1958) reported that phy-sicians who became members of a govern-ment public health insuranee system feltdissatisfied, exploited, and experienced a lossof professional independenee beeause of con-flict with the organization.

Thus, the evidence indieates that multiplelines of authority are aecompanied by roleconflict and dissatisfaction for the membersand loss of organizational efiBeiency and ef-fectiveness. However, these dysfunctions ap-pear to be neeessary coneommitants andcosts of providing professional control overthe technical aspects of the organization'sactivities.

Do violations of the prineiple of unity ofcommand result more in role conflict anddeereased satisfaction for recipients of com-mands and in decreased organizational ef-fectiveness? There are a few studies that,examine applications or violations of theprinciple of unity of command. On the basisof case analysis, both Frank (1958) andDitz (1959) found that the Communist fac-tory manager was subject to performanceexpectations not only from the rationalclaims of his functional role—the rule of hissuperiors and the implicit rule of the market—but also from the political hierarchy of the

party organization and from workers' com-mittees. Frank's study indicated that themanager responded by (I) providing forsafety factors, (2) simulating the meetingof standards, and (3) using personal influ-ence to obtain favors and concessions.

Phelen (1960) used Frank's hypothesisabout ambiguity of goals and conflictingstandards in a historical analysis of theSpanish colonial bureaucracy. He concludedthat his flndings and Frank's suggested thatin the presence of multiple and conflictingstandards whose relative importance is un-defined, subordinates must determine thepriority of accomplishment. They select thosestandards most in accord with their own in-centives and those most likely to be used bysuperiors in evaluating them. Thus, multiplestandards allowed a wide range of discretionto subordinates. Kahn et al. (1964) foundthat persons reporting role conflict statedthat their trust in the persons who imposedthe pressure was reduced, they liked themless personally, they held them in lower es-teem, they communicated less with them,and that their own eflectiveness was de-creased.

Woodward's (1965) findings were not en-tirely consistent with the findings reportedby others. She found consistent adherenceto the principle of unity of command onlyin large batch and mass production firms;adherenee to the principle was not widelypracticed in firms with greater or less tech-nological complexity. Suecessful multiplecommand was found only in technically ad-vanced process firms where the design ormechanism for the eoordination of work isintrinsie in the proeess itself. Furthermoreshe found successful multiple-command re-lationships in a ease analysis of two organi-zations. In one, 30 supervisors—all of whomhad had experience working under a singlesuperior at some time in their careers—re-ported to and received direetion from 5 ex-ecutives; yet 28 of them said tliat closeassociation with the 5 supervisors meant thatthey knew everything that was g'oing on inthe firm and any development contemplated.In the second firm, two-thirds of those inter-viewed said that on the whole they thoughtthe organization worked well. First-hand ob-servation by Woodward's researehers led to

Rizzo et al: ROLE CONFLICT AND AMBIGUITY 153

the same conclusion. No evidence was foundin either firm of incompatible or eonflictinginstructions being given or of attempts beingmade to play one superior against another(Woodward, 1965:121-122):

It was clear that, in these two firms, joint ac-countability was not only the basis of organi-zation, but was associated both with businesssuccess and good interpersonal relationships.Moreover, these were not the only two firms inwhich joint accountability was a workable con-cept. At a higher level there were two firms withjoint managing directors. At an informal levelof organization, too, there were staff managersin line-staff organizations with so much powerthat in practice their advice had the weight ofexecutive orders. The two extreme cases werethe firms with departmentalized organization inwhich a move toward divisionalization was notunder discussion.Although the organizational charts drew a dis-tinction between line roles and staff roles, thestatus of the staff managers, and the way inwhich they were used, implied that they werejointly responsible with the line managers forend results. This showed itself in the way inter-departmental memoranda were routed and ad-dressed, and in the composition of the differentmanagement and policy committees. One ofthese firms was a process production firm. Itseems that in some circumstances people canwork quite happily under more than one boss.

She also reported that in two firms the prin-ciple of unity of command was applied eon-scientiously in assigning responsibility anddefining tlie role of the personnel manager.Yet (Woodward, 1965:116):

It soon became obvious . . . that the patternof informal relationships emerging from day-to-day operations in these firms was not very dif-ferent from what it would have been if thepersonnel manager had been made formally re-sponsible to more than one superior. He re-garded other senior executives as his superiorsand their requests as executive instructions. Asso frequently happens, informal organizationhad created the very situation which formalorganization had been devised to avoid.

The above studies support the positionthat violations of the principles of unity ofcommand and single accountability fre-quently result in consequences predietedfrom classical theory: increased role eon-fliet, less satisfied employees, less eflective

performance, and less adequate use of time.Yet Woodward's study suggested that it mayapply only in certain situations. Otlier factorspresent in the organization can shield orga-nizational members from harmful eflects ofconflicting demands, and prevent them fromplaying one superior against another. Physi-cal technology may serve to control the flowof directions and Information in sueh a man-ner that application of these principles re-sults in undesirable encumbrances.

Relation of role conflict to member satis-faction and. performance. Is role conflictassociated with decreased satisfaction anddecreased organizational efleetiveness? Anexamination of research findings concernedspeeifically with role conflict will furtlierspecify the efleets of role conflict on individ-uals and the conditions under which suchrole conflict is likely to have negative eflects.

Many role-conflict situations are tempo-rary but certain positions constantly visitconflicting role pressures upon their incum-bents. The best known of sueh positions isthat of the foreman, who is often eaught inthe middle (Roethlisberger, 1965) betweenconflicting demands from superiors and sub-ordinates. And as Bai-nard (1938:277) noted,there are many organizational conditionsthat impose on the executive funetion "amoral complexity' and a moral conflict pre-sumably not soluble." Both Frank (1958)and Ditz (1959) fotmd that conflicting di-rections made it necessary for managers touse multiple eriteria in judging personnelperformanee; however, the necessary eriteriacould not be speeified, even theoretically,and thus posed a dilemma. In a case studyof the resolution by managers of formal andinformal role systems, Dalton (1955) elassi-fied executives as strong or weak accordingto the behavior they displayed in reconcilingrole confiict. Strong executives had a hightolerance for eonflict and, unlike the weakexecutives, carried home little of tlie eflectsof job discord. In addition to the above easestudies, several surveys doeumented tlie ef-fect of role conflict in organizations. Seeman(1953) found that for sehool exeeutives po-tential sourees of role conflict resulted insignificant decision-making difficulty.

Gullahorn (1956) found that unresolvedrole conflicts among union members was as-

154 ADMINISTRATIVE SCIENCE QUARTERLY

sociated with an increasing tendency to viewproblems unreaHstically. Getzels and Guba(1954) in a study of role conflict in armyschools found that: (J) schools with militarysubject matter had the least conflict amongstaff, because there was little difference be-tween officer and instructor roles; (2) re-spondents who did not volunteer for teaeh-ing duty, but were not given the opportunityto refuse teaehing assignments, had higheonfliet scores; (3) ineffective teachers ex-perieneed the most conflict.

Gross et al. (1958) found a significant neg-ative eorrelation between perceived role con-flict and three of four measures of jobsatisfaction. Gharters (1952) found tbat eon-flicting expectations were associated with thefeeling of being caught in the middle onlyif foremen: (1) interacted with the workersand management personnel who entertainedthe conflicting expectations; (2) accuratelyperceived the contradictory nature of the ex-pectations; and (3) were motivated to con-form to the expectations of both workers andmanagement. Many of the foremen who felt"in the middle" aeeepted the situation as alegitimate aspect of their jobs and wereneither insecure nor discontent. Thus fromthe above studies it seems clear that roleconflict is associated with decreased satis-faction, coping behavior that would be dys-functional for the organization, and experi-ences of stress and anxiety.

Relation of role ambiguity to member sat-isfaction and performance. Is role ambiguityassociated with decreased satisfaction anddecreased organizational effectiveness? Kahnet al. (1964) asserted that role ambiguity re-sults from organizational size and complexitywhich exeeed the individual's span of com-prehension, rapid organizational growthwhich is usually accompanied by frequentreorganizations, frequent changes in tech-nology which in tum require associatedchanges in social structure, frequent ehangesin personnel which disturb interdependen-cies, changes in the environment of theorganization whieh impose new demands onits members, and managerial philosophieswhich foster restriction on information flowthroughout the organization. The studyshowed that 35 percent of a national sampleof employees were disturbed by lack of aclear idea of the scope and responsibilities

of their jobs. The interviews indicated tliathigh degrees of role ambiguity were associ-ated with inereased tension, anxiety, fear andhostility, decreased job satisfaction, and lossof self-confidence, often with lower produc-tivity.

Gohen (1959) found that ambiguous defi-nition of a task and ineonsistent directionfrom a superior resulted in an increase inanxiety, a less favorable attitude toward thesuperior, and a decrease in productivity.Wispe and Thayer (1957) interviewingthree levels of management in a life in-surance company, found that managerswhose roles were largely ambiguous werethe most anxious. Mandell (1956) surveyed695 persons in trades, clerical work, and en-gineering and accounting occupations. Re-spondents agreed that one of the most im-portant characteristics of a good supei"visorwas the ability to issue clear instructions.Supervisors who were rated low in perfor-mance were also rated low in the extent towhich they gave clear information to tlieirsubordinates. Supervisory willingness to in-form subordinates of their performance wasfound to be a factor in morale.

Smith (1957), in an experimental studywith 140 college students systematicallyvaried the amount of role ambiguity andmeasured the effects on problem solving.Results showed that: (J) when groups wereasked to solve problems without clarificationof the role each member was to perform,their effieiency was significantly less thanwhen the roles were made clear; (2) roleambiguity markedly reduced gioup satis-faction with the experience; and (3) thehostility level was significantly higher forgroups under conditions of role ambiguityas compared to control groups. Subsequentrole clarification significantly reduced thehostility, but did not completely counteractthe effects of the original ambiguity. Thusprevious survey and e.xperimental evidencesuggests that role ambiguity, hke role con-flict, results ill undesirable consequences forboth organizational members and for organ-izational performance.

THE PRESENT STUDY

It is clear that both role conflict and am-biguity are important intervening variablesthat mediate the effects of various organiza-

Rizzo et al: ROLE CONFLICT AND AMBIGUITY 155

tional practices on individual and organiza-tional outcomes. However, it is also clearthat little is known about the relationshipbetween these concepts or among theirtheoretical components, and that the re-search cited used varied measures andmethods and very often did not systemati-cally relate these concepts to other variablesin a complex organizational setting. Tliisstudy is addressed to the development andvalidation of these two role constructs againstmeasures of organizational and managementpractices, leadership behavior, satisfaction,anxiety, propensity to leave, and demo-graphic variables.

DEFINITIONS AND QUESTIONNAIREDEVELOPMENT

The questionnaire developed consisted of30 items (Table 1), 15 of which dealt withrole ambiguity (even numbers) and 15 withrole conflict (odd numbers).

A role is most typically defined as a set ofexpectations about behavior for a position ina social structure. Expectations define behav-ioral requirements or limits ascribed to therole by the focal person filling that position,or by others who relate to the role or simplyhave notions about it. The expectations areconditioned by general experience andknowledge, values, perceptions, and specificexperience with focal person(s). They serveas standards for evaluating the worth orappropriateness of behavior, and they tendto condition or determine such behavior.Role eoncepts such as the following arefound in the literature: (1) role sets orsystems—set(s) of roles in a social structure;(2) role sector(s) or sent role(s)—subset(s)of expectations from counter position(s) inthe role structure; (3) role ambiguity orclarity; and (4) role confiict.

Role ConflictRole confiict is defined in terms of the

dimensions of congruency-incongruency orcompatibility-incompatibility in the reqiiire-ments of the role, where congrueney orcompatibility is judged relative to a set ofstandards or conditions which impinge uponrole performance. Kahn et al. (1964) usedsuch concepts as person-role confiict, inter-role confiict, intersender eonfiict, and intra-

sender confiict. Gross et al (1958) usedintrarole and interrole confiict with em-phasis on exposure to incompatible expecta-tions and on the perceiver of the incom-patibility, i.e., focal person or observer.Definitions of these role confiict components,as used in the present scale are given below.

Incompatibility or incongruency may re-sult in various kinds of confiict;

1. Confiict between the focal person's in-ternal standards or values and the definedrole behavior (items 3, 5, 27, 29). This is aperson-role confiict or intrarole confiict ofthe focal person as he fills a single positionor role.

2. Confiict between the time, resources,or capabilities of the focal person and de-fined role behavior (items 1, 11, 15, 17, 25).Where one other person in a related rolegenerates the ineompatibility, this may beviewed as intrasender eonfiict. It may alsobe organizationally generated. From thepoint of view of the focal person, there isintrarole confiict or person-role confiict, e.g.,insufficient capability.

3. Confiict between several roles for thesame person which require different or in-compatible behaviors, or changes in behav-ior as a function of the situation (items 7and 19); i.e. role overload. This is interroleconfiict for the focal person as he fills morethan one position in the role system.

4. Confiicting expectations and organiza-tional demands in the form of incompatiblepolicies (items 9 and 13), confiicting re-quests from others (item 21), and incom-patible standards of evaluation (item 23).

These role confiict items suggest sanctionsattached to role behavior and are thereforerelated to the role ambiguity componentswhich involve the prediction of the outcomeof one's behavior. Also, the items may refiectmore than the category cited. Finally, thesources of confiict cited in the literatin-e—multiple authority, professional values—arefor the most part not specifically included inthe items.

Role Ambiguity

Role ambiguity is not elaborately definedin the literature. The definition used here isin terms of ( i ) the predictability of the out-come or responses to one's behavior (items8, 16, 24, 30), and (2) the existence or clarity

156 ADMINISTRATIVE SCIENCE QUARTERLY

TABLE 1. QUESTIONNAIRE ITEMS AND FACTOR LOADINGS

Itemnumber Statement

1. I have enough time to complete my work.2. I feel certain about how much authority I have.3. I perform tasks tliat are too easy or boring.4. Clear, planned goals and objectives for my job.5. I have to do things that should be done differently.6. Lack of policies and guidelines to help me.7. I am able to act the same regardless of the group I am with.8. I am corrected or rewarded when I really don't expect it.9. I work under incompatible policies and guidelines.

10. I know that I have divided my time properly.11. I receive an assignment without tlie manpower to complete it.12. I know what my responsibilities are.13. I have to buck a rule or policy in order to carry out an assignment.14. I have to "feel my way" in performing my duties.15. I receive assignments that are within my training and capability.16. I feel certain how I will be evaluated for a raise or promotion.17. I have just the right amount of work to do.18. I know that I have divided my time properly.19. I work with two or more groups who operate quite differently.20. I know exactly what is expected of me.21. I receive incompatible requests from two or more people.22. I am uncertain as to how my job is linked.23. I do things that are apt to be accepted by one person and not

accepted by others.24. I am told how well I am doing my job.25. I receive an assignment witliout adequate resources and materials

to execute it.26. Explanation is clear of what has to be done.27. I work on unnecessary things.28. I have to work under vague directives or orders.29. I perform work that suits my values.30. I do not know if my work will be acceptable to my boss.

Items 10 and 18 on tliis administration were identical, owing to a clerical error.

Factorloadings

.30

Role Roleconflict ambiguity

.60

.43

.60

.56

.54

.36

.43

.56

.41

.52

.52

.59

.51

.42

.31

.62

.61

-.35

.34

.32

.59

.61

.35

.39

of behavioral requirements, often in termsof inputs from the environment, whichwould serve to guide behavior and provideknowledge that the behavior is appropriate(the remaining even-numbered items).Therefore, the items reflect certainty aboutduties, authority, allocation of time, andrelationships with others; the clarity orexistence of guides, directives, policies; andthe ability to predict sanctions as outcomesof behavior.

METHOD

Role Questionnaire

Subjects were requested to respond toeach role item, indicating the degree to

which the condition existed for him, on aseven-point scale ranging from very false tovery true.

Other Variables

For the purpose of relating the derivedrole measures to other measures, tlie rolemeasures were correlated with 45 variablesincluded in the study. A complete list ofthese variables, including the number ofitems in each, means, standard deviations,reliabilities, and correlations appears inTable 2. They fall into the following cate-gories. The satisfaction, leadership, organi-zation, and anxiety variables were derived inpart from separate factor analyses of eachset.

Rizzo et al: ROLE CONFLICT AND AMBICUITY 157

Satisfaction: tlie degree to which the satis-faction condition described in the item ispereeived to exist or be fnlfiUed (7-pointscale).

Leadership: tlie freqnency with which therespondent perceives his boss engaging ingiven behaviors (Ohio State UniversityLeader Behavior Description Questionnaire,5-point scale).

Organization: the degree of which certainorganizational or management practices areperceived to exist (7-point scale).

Anxiety: trne-false description of existenceof physical symptoms, feelings of pressnre orstress, both general and job induced. (Partof the items were drawn from the TaylorManifest Anxiety Scale).

Demographic: age, education, tenure,organization level.

Propensity to leave: the plans for and con-ditions under which the respondent wouldleave the organization, each ranging fromlow to high propensity to leave.

Sample

The questionnaire was administered to a35 percent random sample of the centraloffices and main plant of the firm and to a100 percent sample of the research andengineering division. All respondents weresalaried managerial and technical employees,excluding salesmen, first level foremen, andclerical personnel. The total number of re-spondents were treated as two samples.Sample A (IV = 199) represents a 35 per-cent sample of central offiee and main plantpersonnel plus 35 percent of the respondents,the total universe, of the research and en-gineering division. Thus sample A consistsof a 35 percent random sample of salariedemployees of the total organization with ex-ceptions previously noted. Sample B (JV =91) represents the remaining 65 percent ofthe research and engineering personnel whocompleted the questionnaire but were notrandomly placed in sample A. The sampleswere divided in the above manner to permita representative sampling of the entire firmand to permit maximum sampling of the re-search and engineering division. Approx-imately 70 percent of both samples were be-tween the ages of 25-50 years; 93 percentwere married; and 65-70 percent were with

the company less tlian 10 years. The samplesdiffered in two respects. Although 60 to 65percent of both samples were 3 to 5 levelsfrom the top of the organization, sample Aincluded 30 corporate staff employees nearerthe top, while sample B contained a slightlyhigher percentage of respondents at 6 to 7levels from tlie top. Also members of sampleB were slightly better educated, with 80 per-cent having bachelor's and master's degreescompared to 67 percent in sample A.

Administration

The variables analyzed were all derivedfrom a single combined instrument of ap-proximately 350 items. The instiiiment wasadministered to gi'oups ranging from 10 to50. Anonymity was assured and participationwas voluntary; only 5 percent of the drawnsample did not complete the questionnaire.Administration time took between one anda half and two hours.

RESULTS

Factor Analysis

The responses to the role questionnaireitems were factor analyzed using an imagecovariance method and rotated using avarimax criterion (Kaiser, 1958a, 1958b) inorder to test relationships and structural re-lationships of the role confiict and ambiguitydefinitions. Table 3 presents item means,standard deviations, and the unrotated androtated factor sti'uctures. Two factors wereextracted which account for 56 percent ofthe common variance of the 30-item set.

Factor I was named role confiict becauseit primarily refiected items drawn from therole confiict definition. Items loading .30, orgreater are shown in Table 1. Of the 15 roleconfiict items, 9 are represented with load-ings greater than or equal to .30. Among theremaining 6 items, 3 items (7, 15, 29) loadedprimarily on factor II, role ambiguity, 2items (1, 17) were complex lower-loadingitems, and one item (3) loaded in the ex-pected direction but with a low magnitude.Five of these 6 items were from that part ofthe role confiict definition representingperson-role or intrasender confiict. Four roleambiguity items loaded on this factor (6, 14,28, 30). They are drawn from both parts of

158

g

Ia

o

a

1WK

oq

m

So

X!

oO

ffl

I-i

ADMINISTRATIVE SCIENCE QUARTERLY

O I CO lO CD C^ • ^ lO lO CO (N O C<J (N Cl CM O C^ d i—I i—I O CO lO " ^ C<1 O

I f I i' I \ i I I • • I I r \ \ I " • • i' \ \ \ '

. . . . . . . . . . . . . ^.-H.co. cq. CO CO. cj) . u^ . Tj; cq.

O O It » f t # flOO

I I I I I I I I I I I I I I I

in i-Hcq^incqcqin i-Hoo-HOrHinTf-^'**

r I i \ I I \ \ I I " " " [ • " ' ^ - " [ •

4 O 0'OCOt>005Cs lCOCOI I O ' — < O C ^ O c S I O O

I I

I I I I I

o

.82

CD

00

80

o00

so

CO

67

ocq00

S3

CO CD

O3 CD

00 t~

CO cqin 0000 t^

00inOO

n n c o tCD0OCT>CD0O

1 O t^ 00 CD O>: in cq CO t^ oCD in CO _ -. . _

OOOOt^CDt^OOt^t^CD

t~in COCDOCO^t-CO C<lt~-HCO( ,rHi-H *—i *—1.^ '—^ i -H i—* C O C D C D t ^ ^ C D C O C D l

l t c q c D H c O O i O Oq !-; m co q -^ oq cpi-H'i-irHi-HrH'i-iO»-i

i-Hoo t- '^ in '^ '^cocDcqo i-Hi-Hcqcoi-Hi-HO

tcDcoinH^cDcqo5cDqqcDincDCDcocDO O O O O O C J O O

cooinI-H cq cq

o i n c D O i H f H O J tcq cq cq cq q • * q cq

CDCO f -H t^ t^ t ^OOt^O COCDl~~-' CO t^ CO t^ l>.00 o o o ^ O5 cq f-H ^ m in o co •^ in oo oo co cDqcqcpCTicocqCTiin

CO CO • * cq CO CO CO co' • * • * ' •^ ' T T r j i co' co' •^ ' CO co'

cqin.. q .

COCO-*cq'cOCOCOCOCO

CO CO CO cq i n

cq CO •* in CD i:~ 00 OiO'-HfNco'^w^'^r^—icaccio fsicqc ifMrM

Rizzo et al: ROLE CONFLICT AND AMBIGUITY 159

(U 3

• . a

CS

O

o

-2 c

iS >5 5 ^

cq o cq o S oI I I I I I I I I

cqcDcqcqoocoo5cooot~ocO"^cqcqTfOco"^c<ico

q ql' ' l' f

11> cq O5 00 cq inI o cq d o c<i o

I I I I I I I I

I l l l l I I I

I l l l l I I I

cqcoocoocqicDn<

i n * t o o o t 0 5 0 5 c q * T rin in o •-; o cq CO oq cq o 'H

cqoq_ _OOO

O J O ^ G S C D c q t C O ^ O O ^i n p 1-; t - C O •-; 00 i n OS 1-H oTji lri Ti< co' -*' •*' CO CO CO •'^' -^

in ^ cqoq CO cq

cqi>oc*D!t—in_ T-H COi n ' CO - ^ CO

'-Hoo-H'-Mt> " ^ c o i nM; in_ t-; 00 05 d co cq_

' • ^ co' CO CO CO i-H ^ rH

oXI

T3

c <u c

£ fi °>- o c

o•OrS

£ c

HHill

oCJ.213

CH 0

.9 3c r

tu a)

00 O5(Ncq

CO " ^ i n <cococoi > t~ 00) CO CO

l l

o o o oC» 00 00 O5cq ^H cq I-H

l l

I I I

in_ a; q CJI-H' d I—i ^

^ cq OS COCO -H i n CO

00 CO ^ CD-H cq 1 - ; ^CO •^' cq' in'

cq 00 cq .-;CO CO co' i n

IIo l

O O5cq cq

I COI O

in O500 CD

^ cqO5 05

d d

o inl ~ CO

ai in

cq cq'

Qvi

1 ^I E

.2 ? " -2ao'•o u .2 2 .

S S

0) =bOTJ

< WH O

"TS

(So

in

qV

§-§

160 ADMINISTRATIVE SCIENCE QUARTERLY

TABLE 3. FACTOR LOADINGS FOR ROLE CONFLICT AND AMBIGUITY ITEMS, FOR SAMPLE A

Item

123456789

101112131415161718192021222324252627282930

Items

Mean

3.854.004.003.954.194.124.462.873.604.164.505.053.664.335.904.053.013.964.704.203.883.014.353.664.243.923.663.764.523.32

10 and 18 were

Standarddeviation

1.811.801.881.701.801.801.721.611.931.482.041.601.981.921.141.881.631.682.061.672.041.881.891.761.821.581.881.771.581.69

identical; only item

Factor

Unrotated

Roleconflict

- . 3 3- .49

.08- . 3 0

.54

.49- . 1 9

.13

.53- .27

.44- . 4 9

.43

.50- . 1 9- . 1 7- . 3 5- . 2 4

.33- . 5 3

.50

.33

.44- . 2 3

.45- . 3 8

.53

.61- . 2 5

.38

10 was scored.

Roleamhiguity

.08

.27

.18

.30

.29

.07

.24

.17

.29

.57

.34

.39

.33- . 0 5

.18

.29

.13

.56

.28

.36

.28

.03

.11

.14

.26

.15

.17

.19

.29

.00

loadings

Rotated

Roleconflict

- . 2 1- . 2 2

.17- . 0 5

.60"

.43

.00

.21

.60

.14

.56*- . 1 6

.54*

.36- . 0 4

.05- . 2 0

.15

.48*- . 2 0

.56"

.28

.41"- . 0 9

.52*- . 2 1

.52"

.59- . 0 2

.30

Roleamhiguity

.26

.51*

.09

.42"- . 1 0- . 2 4

.31

.06- . 1 0

.62*

.00

.61"

.00- . 3 5

.26

.34

.32

.59

.02

.61*- . 0 8- . 1 7- . 1 8

.25- . 0 7

.35*- . 1 9- . 2 2

.39- . 2 3

* Items used in scoring factors.

the ambiguity definition (predictability ofsanctions and external inputs). This factoraccounts for 32 percent of the eommonvariance.

Factor II was named role ambiguity be-cause it reflected items drawn primarily fromthe same definition. Most items were in thedirection of role clarity. The sign of suchitems are reflected in scoring. Items loadinggreater than or equal to .30 are shown inTable 1. Of the 15 role ambiguity items, 9are represented on this factor. Among theremaining 6 items, 4 (6, 22, 24, 30) loadedprimarily on factor I, but might also beviewed as complex items with lower load-ings (opposite sign) on factor I (item 14was also eomplex): one item (8) loadedlow, but primarily on factor I; and one item

(24) loaded in the expected manner, butwith a low magnitude. Factor II accountsfor 26.3 percent of the common variance ofthe set.

The factor analysis revealed that the twofactors extracted strongly parallel the twotheoretical concepts of role conflict and roleambiguity; therefore, the unexamined yetoften presumed separation of the two con-structs seems warranted by this study. How-ever, the specific definitional parts of thesetwo concepts, as exti-acted from the liter-ature and defined by the authors, did notemerge as separate independent factors.

Scale Development

For purposes of developing scales, itemswere selected for scoring on each factor

Rizzo et al: ROLE CONFLICT AND AMBICUITY 161

using several criteria. First, only items load-ing greater than or equal to .30 were con-sidered. Second, complex items—those withrelatively high loadings on both factors—were excluded in order to achieve greaterindependence of scores. Third, items werethen subjected to reliability analysis (Kuder-Richardson internal consistency reliabilitieswith Spearman-Brown corrections), using aninterative technique which selected itemscontributing to the reliability of the finalsets for each scale. Items seleeted for scor-ing—marked with an asterisk in Table 3—were summed for each respondent anddivided by the number of items in the set.Factor II items were reflected before scor-ing (response of 7 changed to 1, 6 to 2, etc.).Reliabilities are reported in Table 2.

Correlations with Other Variables

Table 2 lists the scales, the number ofitems for each scale, means, standard de-viations, and reliabilities computed for allmultiple item scales in the interative fashiondescribed above. As with the role scales,means reflect the location on the 7-point, 5-point, and true-false response scales. Table2 indicates the response made for eaeh setof variables and lists the product-momentcorrelations of the 45 variables with therole-conflict and role-ambiguity measures.The intercorrelations between the role mea-sures was .25 (sample A) and .01 (sampleB), indicating relative independence. Table2 also shows that the sign of the eorrelationsbetween both role measures and other vari-ables is generally in the same direction, withthe exception of variable 34.

Among the satisfaction variables, there isa clear tendency for higher correlations be-tween the role ambiguity measure and othervariables. The overall negative correlationsindicate lowered degrees of need fulfillmentwith increased role conflict and role am-biguity. Need areas associated with workitself, the reward system, and the pleasant-ness of the social environment appear sim-ilarly afl ected. Job security (variable 6),generally high in this organization, (X =4.97) showed the lowest relationship to therole measures.

In the leadership variables the behavior of

superiors that might be expected to have aless direet influence on subordinates, i.e.,representation of the group to outsiders, up-ward aspiration, and persuasion, show theweakest relationships with the role variables.On the other hand, behaviors indicative ofmore direct superior-subordinate relationsshow stronger relationships. In these eases,role conflict and role ambiguity tend to belower under conditions in which superiorsare described as more frequently engagingin emphasizing production under conditionsof uncertainty, providing structure andstandards, facilitating teamwork, toleratingfreedom, and exerting upward influence.Tolerance of freedom by the superior didnot increase role conflict or role ambiguity,nor was there a very strong correlation forrole abdication.

Both the leadership and the organizationpractices tend to show higher correlationsin sample A than sample B, with no signifi-cant reversals in sign across samples. Thespecific organization practices which tendto be associated with high role conflict androle ambiguity are goal confliet and in-consistency, delay in deeisions, distortionand suppression of information, and viola-tions of the chain of command. The practiceswhich tend to be associated with lower roleconflict and role ambiguity are emphasis onpersonal development, formalization, ade-quacy of communication, planning, horizon-tal communication, top management recep-tiveness to ideas, coordination of work flow,adaptability to change, and adequacy ofauthority.

These patterns of relationships, taken to-gether with those variables not highly re-lated to the role variables, tend generally tofit with what one would predict from roletheory and previous research. However, var-iable 34, in which increases in requests forinformation from superiors are associatedwith increased role conflict but decreasedrole ambiguity, indieates that respondentsmight have viewed such requests as clarify-ing expectations, but possibly conflict induc-ing. Furthermore, violations in ehain ofcommand (variable 35) which would ex-clude the immediate supervisor as an inputto the respondent is associated with in-ereased role eonflict and role ambiguity.

162 ADMINISTRATIVE SCIENCE QUARTERLY

There is only a slight tendency for therole variables to correlate with demographic,anxiety, and propensity-to-leave variables;and among the latter two variables, correla-tions tend to be positive.

DISCUSSION AND CONCLUSIONS

The results of factor and item analysestended to show (J) that the two conceptsof role conflict and role ambiguity emergedas separate dimensions, (2) that scales de-rived on the basis of the factor analysis ofsample A were relatively independent forboth samples A and B, and (3) that theoret-ical components of these concepts did notemerge as distinct factorial entities.

When the scales developed to measurethese concepts were correlated with othervariables, the scales tended to correlate (1)negatively with measures of need fulfill-ment, (2) more strongly with leader behav-iors indicative of direct as opposed to in-direct interactions with subordinates, (3)most cases with leadership and varied or-ganizational practices in a pattern one wouldpredict from classical theory, role theory, andprevious research, and (4) weakly, butpositively, with anxiety and propensity toleave the organization.

There is no adequate explanation of whyrole ambiguity correlated more highly thanrole conflict with the satisfaction variables.Role theory and research often suggest thatrole conflict might be the more dysfunc-tional. This possibility is generally not borneout for any set of variables in this research.Furthermore, one might have expectedstronger relationships between role variablesand measures of anxiety or propensity toleave. It may be that role stress does notgenerate anxiety, when leaving is possible, ora strong desire to leave the system; or thatfor certain levels of role stress, adaptive be-havior reduces stress, as some literatureindicates. It may also be that for certainindividuals, inputs which might be viewedas stressful are rewarding. For example,violations of the chain of command may beviewed by some as stressful and others as anopportunity to gain visibility and recogni-tion. These questions are beyond the scopeof the present paper, but are the subjeet of

a larger research project eurrently beingconducted by the authors.

The correlational analysis of this studysuggests that a more precise experimentalinvestigation of the relation between causeand effect is warranted. Further, more needsto be known about various levels of roleconflict and role ambiguity. The presentsample had relatively high levels of dysfunc-tional organizational practices, yet norma-tive data on the scales used are not availableat present to test this.

In conclusion, this paper represents anattempt to deal with role conflict and roleambiguity in a more empirical and opera-tional manner than usual. In the past, vari-ous procedures have been reported leadingto fairly consistent eonelusions, but lackingrigor in measurement or explanation. Fur-ther research of the present kind and likethat of Kahn et al (1964) offer hope ofmaking role theory a more iiseful tool in or-ganizational analysis and tlieory.

John R. Rizzo is an associate professor inthe department of management at WesternMichigan University. Roljert J. House andSidney I. Lirtzman are associate professorsin the Rernard M. Raruch College of theCity University of New York.

REFERENCES

Barnard, Chester I.1938 The Functions of the Executive. Cam-

bridge: Harvard University Press.Ben-David, Joseph

1958 "The professional role of the physicianin bureaucratized medicine: a studyin role conflict." Human Relations, 2:255-257.

Blau, Peter M., and W. Richard Scott1962 FoiTnal Organizations. San Francisco:

Chandler.Charters, W. W., Jr.

1952 A Study of Role Conflict among Fore-men in a Heavy Industry. Doctoral dis-sertation. University of Michigan.

Cohen, Arthur R.1959 "Situational structure, self esteem, and

threat-oriented reactions to power."In D. Cartwright (ed.). Studies in So-cial Power:35-52. Ann Arhor: Insti-tute for Social Research, University ofMichigan.

Rizzo et al: ROLE CONFLICT AND AMBIGUITY 163

Dalton, Melville1955 "Managing the managers." Human Or-

ganization, 14:4—10.Davis, Ralph C.

1951 Fundamentals of Top Management.New York: Harper.

Ditz, Gerhard W.1959 "Industrial administration in commu-

nist East Europe." Administrative Sci-ence Quarterly, 4:82-96.

Etzioni, Amitai1959 "Authority structure and organizational

effectiveness." Administrative ScienceQuarterly, 4:43-67.

Evans, William M.1962 "Role strain and the norm of recipro-

city in research organization." Ameri-can Joumal of Sociology, 68:346-354.

Frank, Andrew G.1958 "Goal ambiguity and conflicting stan-

dards: approach to the study of orga-nizations." Human Organization, 17:8-13.

Getzels, Jacob W., and Egon G. Guba1954 "Role, role conflict and effectiveness:

an empirical study." American Socio-logical Review, 19:164-175.

Gouldner, Alvin W.1958a "Cosmopolitans and locals: toward an

analysis of latent social roles—I." Ad-ministrative Science Quarterly, 2:281-306.

1958b "Cosmopolitans and locals: toward ananalysis of latent social roles—IL" Ad-ministrative Science Quarterly, 2:444-480.

Gross, Neal, Ward S. Mason, and Alexander W.McEachern

1958 Explorations in Role Analysis. NewYork: Wiley.

Gullahorn, John T.1956 "Measuring role conflict." American

Joumal of Sociology, 61:299-303.House, Robert J.

1967 Management Development: Design,Evaluation and Implementation. AnnArbor: Bureau of Industrial Relations,University of Michigan.

1968 "Leadership training: some dysfunc-tional consequences." AdministrativeScience Quarterly, 12:556-571.

Kahn, Robert L., Donald M. Wolfe, Robert P.Quinn, J. Diedrick Snoek, and RobertA. Rosenthal.

1964 Organizational Stress. New York:Wiley.

Kaiser, Henry F.1958a Best Approximation of a Common Va-

riance Faetor Space. Working paper.Bureau of Educational Research, Uni-versity of Illinois, Research Report No.25.

1958b "The varimax criterion for analytic ro-tation in factor analysis." Psycho-metrika, 23:187-200.

Kaplan, Norman1959 "The role of tlie research administra-

tor." Administrative Science Quarterly,4:20-41.

La Porte, Todd R.1965 "Conditions of strain and accomoda-

tion in industrial research organiza-tions." Administrative Science Quar-terly, 10:21-28.

Mandell, Milton M.1956 "Supervisory characteristics and rat-

ings: a summary of recent research."Personnel, 32:435-440.

Perrow, Charles1965 "Hospitals: technology, structure, and

goals." In James G. March (ed.).Handbook of Organizations:910-971.Ghicago: Rand McNally.

Phelen, John L.1960 "Authority and flexibility in Spanish

imperial bureaucracy." AdministrativeScience Quarterly, 5:47-65.

Reissman, Leonard1949 "A study of role conceptions in bu-

reaucracy." Social Forces, 27:305—310.Roethlisberger, Fritz J.

1965 "The foreman: master and victim ofdouble talk." Harvard Business Re-view, 45: 22fl:\

Seeman, Melvin1953 "Role conflict and ambivalence in lead-

ership." American Sociological Review,18:373-380.

Smith, Evvart E.1957 "The effects of clear and unclear role

expectations on group productivityand defensiveness." Joumal of Abnor-mal and Social Psychology, 55:213-217.

Wispe, Lauren G., and Paul W. Tliayer1957 "Role ambiguity and anxiety in an

occupational group." Journal of SocialPsychology, 46:41-48.

Woodward, Joan1965 Industrial Organization. New York:

Oxford University Press.Zawacki, J. S.

1963 A System of Unofficial Rules of a Bu-reaucracy: A Study of Hospitals. Doc-toral dissertation. University of Pitts-burg.