t^ 1The Ambiguity of Leadership

JEEFREYPFEFEERUniversity of California, Berkeley

Problems with the concept of leadership are addressed: (a) the am-biguity of its definition and measurement, (b) the issue of whetherleadership affects organizational performance, and (c) the process ofselecting leaders, which frequently emphasizes organizationally-irrel-evant criteria. Leadership is a process of attributing causation to indi-vidual social actors. Study of leaders as symbols and of the process ofattributing leadership might be productive.

Leadership has for some time been a majortopic in social and organizational psychology.Underlying much of this research has been theassumption that leadership is causally related toorganizational performance. Through an analysisof leadership styles, behaviors, or characteristics(depending on the theoretical perspective cho-sen), the argument has been made that more ef-fective leaders can be selected or trained or, al-ternatively, the situation can be configured toprovide for enhanced leader and organizationaleffectiveness.

Three problems with emphasis on leadership

Jeffrey Pfeffer (Ph.D. — Stanford University) is Associate Pro-fessor in the School of Business Administration and AssociateResearch Sociologist in the Institute of Industrial Relations atthe University of California, Berkeley.

Received 72/77/75, Accepted 2/27/76; Revised 4/20/76.

as a concept can be posed: (a) ambiguity in defi-nition and measurement of the concept itself; (b)the question of whether leadership has discerni-ble effects on organizational outcomes; and (c)the selection process in succession to leadershippositions, which frequently uses organizationallyirrelevant criteria and which has implications fornormative theories of leadership. The argumenthere is that leadership is of interest primarily as aphenomenological construct. Leaders serve assymbols for representing personal causation ofsocial events. How and why are such attributionsof personal effects made? Instead of focusing onleadership and its effects, how do people make

1 An earlier version of this paper was presented at the con-ference. Leadership: Where Else Can We Go?, Center forCreative Leadership, Greensboro, North Carolina, June 30 -July 1,1975.

104

Academy of Management Review-January 1977 105

inferences about and react to phenomena la-belled as leadership (5)?

The Ambiguity of the Concept

While there have been many studies of lead-ership, the dimensions and definition of the con-cept remain unclear. To treat leadership as a sep-arate concept, it must be distinguished fromother social influence phenomena. Hollanderand Julian (24) and Bavelas (2) did not draw dis-tinctions between leadership and other proc-esses of social influence. A major point of theHollander and Julian review was that leadershipresearch might develop more rapidly if moregeneral theories of social influence were incor-porated. Calder (5) also argued that there is nounique content to the construct of leadershipthat is not subsumed under other, more generalmodels of behavior.

Kochan, Schmidt, and DeGotiis (33) at-tempted to distinguish leadership from relatedconcepts of authority and social power. In lead-ership, influence rights are voluntarily conferred.Power does not require goal compatability —merely dependence — but leadership impliessome congruence between the objectives of theleader and the led. These distinctions depend onthe ability to distinguish voluntary from involun-tary compliance and to assess goal compatibility.Goal statements may be retrospective inferencesfrom action (46, 53) and problems of distinguish-ing voluntary from involuntary compliance alsoexist (32). Apparently there are few meaningfuldistinctions between leadership and other con-cepts of social influence. Thus, an understandingof the phenomena subsumed under the rubric ofleadership may not require the construct of lead-ership (5).

While there is some agreement that leader-ship is related to social influence, more disagree-ment concerns the basic dimensions of leaderbehavior. Some have argued that there are twotasks to be accomplished in groups — mainte-nance of the group and performance of sometask or activity — and thus leader behavior might

be described along these two dimensions (1, 6, 8,25). The dimensions emerging from the OhioState leadership studies — consideration and in-itiating structure — may be seen as similar to thetwo components of group maintenance and taskaccomplishment (18).

Other dimensions of leadership behaviorhave also been proposed (4). Day and Hamblin(10) analyzed leadership in terms of the closenessand punitiveness of the supervision. Several au-thors have conceptualized leadership behaviorin terms of the authority and discretion subordi-nates are permitted (23, 36, 51). Fiedler (14) anal-yzed leadership in terms of the least-preferred-co-worker scale (LPG), but the meaning and be-havioral attributes of this dimension of leader-ship behavior remain controversial.

The proliferation of dimensions is partly afunction of research strategies frequently em-ployed. Factor analysis on a large number ofitems describing behavior has frequently beenused. This procedure tends to produce as manyfactors as the analyst decides to find, and permitsthe development of a large number of possiblefactor structures. The resultant factors must benamed and further imprecision is introduced.Deciding on a summative concept to represent afactor is inevitably a partly subjective process.

Literature assessing the effects of leadershiptends to be equivocal. Sales (45) summarizedleadership literature employing the authoritar-ian-democratic typology and concluded that ef-fects on performance were small and inconsist-ent. Reviewing the literature on considerationand initiating structure dimensions, Korman (34)reported relatively small and inconsistent results,and Kerr and Schriesheim (30) reported moreconsistent effects of the two dimensions. Betterresults apparently emerge when moderating fac-tors are taken into account, including subordi-nate personalities (50), and situational character-istics (23, 51). Kerr, et al. (31) list many moderat-ing effects grouped under the headings of sub-ordinate considerations, supervisor considera-tions, and task considerations. Even if each set ofconsiderations consisted of only one factor

106 The Ambiguity of Leadership

(which it does not), an attempt to account for theeffects of leader behavior would necessitate con-sidering four-way interactions. While social real-ity is complex and contingent, it seems desirableto attempt to find more parsimonious explana-tions for the phenomena under study.

The Effects of Leaders

Hall asked a basic question about leadership:is there any evidence on the magnitude of theeffects of leadership (17, p. 248)? Surprisingly, hecould find little evidence. Given the resourcesthat have been spent studying, selecting, andtraining leaders, one might expect that the ques-tion of whether or not leaders matter would havebeen addressed earlier (12).

There are at least three reasons why it mightbe argued that the observed effects of leaders onorganizational outcomes would be small. First,those obtaining leadership positions are selected,and perhaps only certain, limited styles of be-havior may be chosen. Second, once in the lead-ership position, the discretion and behavior ofthe leader are constrained. And third, leaderscan typically affect only a few of the variables thatmay impact organizational performance.

Homogeneity of Leaders

Persons are selected to leadership positions.As a consequence of this selection process, therange of behaviors or characteristics exhibited byleaders is reduced, making it more problematicto empirically discover an effect of leadership.There are many types of constraints on the selec-tion process. The attraction literature suggeststhat there is a tendency for persons to like thosethey perceive as similar (3). In critical decisionssuch as the selections of persons for leadershippositions, compatible styles of behavior probablywill be chosen.

Selection of persons is also constrained bythe internal system of influence in the organiza-tion. As ZaId (56) noted, succession is a criticaldecision, affected by political influence and byenvironmental contingencies faced by the or-

ganization. As Thompson (49) noted, leaders maybe selected for their capacity to deal with variousorganizational contingencies. In a study of char-acteristics of hospital administrators, Pfeffer andSalancik (42) found a relationship between thehospital's context and the characteristics and ten-ure of the administrators. To the extent that thecontingencies and power distribution within theorganization remain stable, the abilities and be-haviors of those selected into leadership posi-tions will also remain stable.

Finally, the selection of persons to leader-ship positions is affected by a self-selection proc-ess. Organizations and roles have images, pro-viding information about their character. Personsare likely to select themselves into organizationsand roles based upon their preferences for thedimensions of the organizational and role char-acteristics as perceived through these images.The self-selection of persons would tend to workalong with organizational selection to limit therange of abilities and behaviors in a given organ-izational role.

Such selection processes would tend to in-crease homogeneity more within a single organ-ization than across organizations. Yet many stud-ies of leadership effect at the work group levelhave compared groups within a single organiza-tion. If there comes to be a widely shared, social-ly constructed definition of leadership behaviorsor characteristics which guides the selectionprocess, then leadership activity may come to bedefined similarly in various organizations, lead-ing to the selection of only those who match theconstructed image of a leader.

Constraints on Leader Behavior

Analyses of leadership have frequently pre-sumed that leadership style or leader behaviorwas an independent variable that could be se-lected or trained at will to conform to what re-search would find to be optimal. Even theoristswho took a more contingent view of appropriateleadership behavior generally assunned that withproper training, appropriate behavior could beproduced (51). Fiedler (13), noting how hard it

Academy of Management Review-January 1977 107

was to change behavior, suggested changing thesituational characteristics rather than the person,but this was an unusual suggestion in the con-text of prevailing literature which suggested thatleadership style was something to be strategicallyselected according to the variables of the partic-ular leadership theory.

But the leader is embedded in a social sys-tem, which constrains behavior. The leader has arole set (27), in which members have expecta-tions for appropriate behavior and persons makeefforts to modify the leader's behavior. Pressuresto conform to the expectations of peers, subor-dinates, and superiors are all relevant in deter-mining actual behavior.

Leaders, even in high-level positions, haveunilateral control over fewer resources and few-er policies than might be expected. Investmentdecisions may require approval of others, whilehiring and promotion decisions may be accom-plished by committees. Leader behavior is con-strained by both the demands of others in therole set and by organizationally prescribed limi-tations on the sphere of activity and influence.

External Factors

Many factors that may affect organizationalperformance are outside a leader's control, evenif he or she were to have complete discretionover major areas of organizational decisions. Forexample, consider the executive in a construc-tion firm. Costs are largely determined by opera-tion of commodities and labor markets; and de-mand is largely affected by interest rates, avail-ability of mortgage money, and economic condi-tions which are affected by governmental poli-cies over which the executive has little control.School superintendents have little control overbirth rates and community economic develop-ment, both of which profoundly affect schoolsystem budgets. While the leader may react tocontingencies as they arise, or may be a better orworse forecaster, in accounting for variation inorganizational outcomes, he or she may accountfor relatively little compared to external factors.

Second, the leader's success or failure may

be partly due to circumstances unique to the or-ganization but still outside his or her control.Leader positions in organizations vary in terms ofthe strength and position of the organization.The choice of a new executive does not funda-mentally alter a market and financial positionthat has developed over years and affects theleader's ability to make strategic changes and thelikelihood that the organization will do well orpoorly. Organizations have relatively enduringstrengths and weaknesses. The choice of a partic-ular leader for a particular position has limitedimpact on these capabilities.

Empirical Evidence

Two studies have assessed the effects ofleadership changes in major positions in organi-zations. Lieberson and O'Connor (35) examined167 business firms in 13 industries over a 20 yearperiod, allocating variance in sales, profits, andprofit margins to one of four sources: year (gen-eral economic conditions), industry, companyeffects, and effects of changes in the top execu-tive position. They concluded that compared toother factors, administration had a limited effecton organizational outcomes.

Using a similar analytical procedure, Salancikand Pfeffer (44) examined the effects of mayorson city budgets for 30 U.S. cities. Data on ex-penditures by budget category were collectedfor 1951-1968. Variance in amount and propor-tion of expenditures was apportioned to theyear, the city, or the mayor. The mayoral effectwas relatively small, with the city accounting formost of the variance, although the mayor effectwas larger for expenditure categories that werenot as directly connected to important interestgroups. Salancik and Pfeffer argued that the ef-fects of the mayor were limited both by absenceof power to control many of the expendituresand tax sources, and by construction of policiesin response to demands from interests in the en-vironment.

If leadership is defined as a strictly interper-sonal phenomenon, the relevance of these twostudies for the issue of leadership effects be-

108 The Ambiguity of Leadership

comes problematic. But such a conceptualizationseems unduly restrictive, and is certainly incon-sistent with Selznick's (47) conceptualization ofleadership as strategic management and decisionmaking. If one cannot observe differences whenleaders change, then what does it matter who oc-cupies the positions or how they behave?

Pfeffer and Salancik (41) investigated theextent to which behaviors selected by first-linesupervisors were constrained by expectations ofothers in their role set. Variance in task and socialbehaviors could be accounted for by role-set ex-pectations, with adherence to various demandsmade by role-set participants a function of sim-ilarity and relative power. Lowin and Craig (37)experimentally demonstrated that leader behav-ior was determined by the subordinate's own be-havior. Both studies illustrate that leader behav-iors are responses to the demands of the socialcontext.

The effect of leadership may vary dependingupon level in the organizational hierarchy, whilethe appropriate activities and behaviors may alsovary with organizational level (26, 40). For themost part, empirical studies of leadership havedealt with first line supervisors or leaders withrelatively low organizational status (17). If leader-ship has any impact, it should be more evident athigher organizational levels or where there ismore discretion in decisions and activities.

The Process of Selecting Leaders

Along with the suggestion that leadershipmay not account for much variance in organiza-tional outcomes, it can be argued that merit orability may not account for much variation in hir-ing and advancement of organizational person-nel. These two ideas are related. If competence ishard to judge, or if leadership competence doesnot greatly affect organizational outcomes, thenother, person-dependent criteria may be suffi-cient. Effective leadership styles may not predictcareer success when other variables such as so-cial background are controlled.

Belief in the importance of leadership is fre-

quently accompanied by belief that persons oc-cupying leadership positions are selected andtrained according to how well they can enhancethe organization's performance. Belief in a lead-ership effect leads to development of a set of ac-tivities oriented toward enhancing leadership ef-fectiveness. Simultaneously, persons managingtheir own careers are likely to place emphasis onactivities and developing behaviors that will en-hance their own leadership skills, assuming thatsuch a strategy will facilitate advancement.

Research on the bases for hiring and promo-tion has been concentrated in examination ofacademic positions (e.g., 7, 19, 20). This is possi-bly the result of availability of relatively preciseand unambiguous measures of performance,such as number of publications or citations. Evi-dence on criteria used in selecting and advancingpersonnel in industry is more indirect.

Studies have attempted to predict either thecompensation or the attainment of general man-agement positions of MBA students, using per-sonality and other background information (21,22, 54). There is some evidence that managerialsuccess can be predicted by indicators of abilityand motivation such as test scores and grades,but the amount of variance explained is typicallyquite small.

A second line of research has investigatedcharacteristics and backgrounds of persons at-taining leadership positions in major organiza-tions in society. Domhoff (11), Mills (38), andWarner and Abbeglin (52) found a strong pre-ponderance of persons with upper-class back-grounds occupying leadership positions. The im-plication of these findings is that studies of grad-uate success, including the success of MBA's,would explain more variance if the family back-ground of the person were included.

A third line of inquiry uses a tracking mod-el. The dynamic model developed is one inwhich access to elite universities is affected bysocial status (28) and, in turn, social status and at-tendance at elite universities affect later careeroutcomes (9,43,48, 55).

Unless one is willing to make the argument

Academy of Management Review-January 1977 109

that attendance at elite universities or comingfrom an upper class background is perfectly cor-related with merit, the evidence suggests thatsuccession to leadership positions is not strictlybased on meritocratic criteria. Such a conclusionis consistent with the inability of studies attempt-ing to predict the success of MBA graduates toaccount for much variance, even when a varietyof personality and ability factors are used.

Beliefs about the bases for social mobility areimportant for social stability. As long as personsbelieve that positions are allocated on merito-cratic grounds, they are more likely to be satis-fied with the social order and with their positionin it. This satisfaction derives from the belief thatoccupational position results from application offair and reasonable criteria, and that the oppor-tunity exists for mobility if the person improvesskills and performance.

If succession to leadership positions is deter-mined by person-based criteria such as social ori-gins or social connections (16), then efforts toenhance managerial effectiveness with the ex-pectation that this will lead to career success di-vert attention from the processes of stratificationactually operating within organizations. Leader-ship literature has been implicitly aimed at twoaudiences. Organizations were told how to be-come more effective, and persons were told whatbehaviors to acquire in order to become effec-tive, and hence, advance in their careers. Thepossibility that neither organizational outcomesnor career success are related to leadership be-haviors leaves leadership research facing issuesof relevance and importance.

The Attribution of Leadership

Kelley conceptualized the layman as:an applied scientist, that is, as a person con-cerned about applying his knowledge of caus-al relationships in order to exercise control ofhis world (29, p. 2).

Reviewing a series of studies dealing with the at-tributional process, he concluded that personswere not only interested in understanding theirworld correctly, but also in controlling it.

The view here proposed is that attributionprocesses are to be understood not only as ameans of providing the individual with a verid-ical view of his world, but as a means of en-couraging and maintaining his effective exer-cise of control in that world (29, p. 22).

Controllable factors will have high salience ascandidates for causal explanation, while a biastoward the more important causes may shift theattributional emphasis toward causes that arenot controllable (29, p. 23). The study of attribu-tion is a study of naive psychology — an exami-nation of how persons make sense out of theevents taking place around them.

If Kelley is correct that individuals will tendto develop attributions that give them a feelingof control, then emphasis on leadership may de-rive partially from a desire to believe in the effec-tiveness and importance of individual action,since individual action is more controllable thancontextual variables. Lieberson and O'Connor(35) made essentially the same point in introduc-ing their paper on the effects of top manage-ment changes on organizational performance.Given the desire for control and a feeling of per-sonal effectiveness, organizational outcomes aremore likely to be attributed to individual actions,regardless of their actual causes.

Leadership is attributed by observers. Socialaction has meaning only through a phenomeno-logical process (46). The identification of certainorganizational roles as leadership positionsguides the construction of meaning in the direc-tion of attributing effects to the actions of thosepositions. While Bavelas (2) argued that the func-tions of leadership, such as task accomplishmentand group maintenance, are shared throughoutthe group, this fact provides no simple and po-tentially controllable focus for attributing causal-ity. Rather, the identification of leadership posi-tions provides a simpler and more readilychangeable model of reality. When causality islodged in one or a few persons rather than be-ing a function of a complex set of interactionsamong all group members, changes can bemade by replacing or influencing the occupantof the leadership position. Causes of organiza-

110 The Ambiguity of Leadership

tional actions are readily identified in this simplecausal structure.

Even if, empirically, leadership has little ef-fect, and even if succession to leadership posi-tions is not predicated on ability or performance,the belief in leadership effects and meritocraticsuccession provides a simple causal frameworkand a justification for the structure of the socialcollectivity. More importantly, the beliefs inter-pret social actions in terms that indicate poten-tial for effective individual intervention or con-trol. The personification of social causality servestoo many uses to be easily overcome. Whetheror not leader behavior actually influences per-formance or effectiveness, it is important be-cause people believe it does.

One consequence of the attribution of cau-sality to leaders and leadership is that leaderscome to be symbols. Mintzberg (39), in his dis-cussion of the roles of managers, wrote of thesymbolic role, but more in terms of attendanceat formal events and formally representing theorganization. The symbolic role of leadership ismore important than implied in such a descrip-tion. The leader as a symbol provides a target foraction when difficulties occur, serving as ascapegoat when things go wrong. Gamson andScotch (15) noted that in baseball, the firing ofthe manager served a scapegoating purpose.One cannot fire the whole team, yet when per-formance is poor, something must be done. Thefiring of the manager conveys to the world and tothe actors involved that success is the result ofpersonal actions, and that steps can and will betaken to enhance organizational performance.

The attribution of causality to leadershipmay be reinforced by organizational actions,such as the inauguration process, the choiceprocess, and providing the leader with symbolsand ceremony. If leaders are chosen by using arandom number table, persons are less likely tobelieve in their effects than if there is an elabor-ate search or selection process followed by anelaborate ceremony signifying the changing ofcontrol, and if the leader then has a variety ofperquisites and symbols that distinguish him or

her from the rest of the organization. Construc-tion of the importance of leadership in a givensocial context is the outcome of various socialprocesses, which can be empirically examined.

Since belief in the leadership effect pro-vides a feeling of personal control, one mightargue that efforts to increase the attribution ofcausality to leaders would occur more when it ismore necessary and more problematic to attri-bute causality to controllable factors. Such anargument would lead to the hypothesis that themore the context actually effects organizationaloutcomes, the more efforts will be made to en-sure attribution to leadership. When leadersreally do have effects, it is less necessary to en-gage in rituals indicating their effects. Such ritualsare more likely when there is uncertainty andunpredictability associated with the organiza-tion's operations. This results both from the de-sire to feel control in uncertain situations andfrom the fact that in ambiguous contexts, it iseasier to attribute consequences to leadershipwithout facing possible disconfirmation.

The leader is, in part, an actor. Throughstatements and actions, the leader attempts toreinforce the operation of an attribution processwhich tends to vest causality in that position inthe social structure. Successful leaders, as per-ceived by members of the social system, arethose who can separate themselves from organ-izational failures and associate themselves withorganizational successes. Since the meaning ofaction is socially constructed, this involves mani-pulation of symbols to reinforce the desiredprocess of attribution. For instance, if a managerknows that business in his or her division is aboutto improve because of the economic cycle, theleader may, nevertheless, write recommenda-tions and undertake actions and changes thatare highly visible and that will tend to identifyhis or her behavior closely with the division. Amanager who perceives impending failure willattempt to associate the division and its policiesand decisions with others, particularly personsin higher organizational positions, and to disas-sociate himself or herself from the division's

Academy of Management Review-January 1977 111

performance, occasionally even transferring ormoving to another organization.

Conclusion

The theme of this article has been that anal-ysis of leadership and leadership processes mustbe contingent on the intent of the researcher. Ifthe interest is in understanding the causality ofsocial phenomena as reliably and accurately aspossible, then the concept of leadership may bea poor place to begin. The issue of the effects ofleadership is open to question. But examinationof situational variables that accompany more orless leadership effect is a worthwhile task.

The more phenomenological analysis of

leadership directs attention to the process bywhich social causality is attributed, and focuseson the distinction between causality as perceivedby group members and causality as assessed by anoutside observer. Leadership is associated with aset of myths reinforcing a social construction ofmeaning which legitimates leadership role occu-pants, provides belief in potential mobility forthose not in leadership roles, and attributes so-cial causality to leadership roles, thereby provid-ing a belief in the effectiveness of individualcontrol. In analyzing leadership, this mythologyand the process by which such mythology is cre-ated and supported should be separated fromanalysis of leadership as a social influence proc-ess, operating within constraints.

REEERENCES

1. Bales, R.F. Interaction Process Analysis: A Method forthe Study of Small Groups (Reading, Mass.: Addison-Wesley, 1950).

2. Bavelas, Alex. "Leadership: Man and Function," Admin-istrative Science Quarterly, Vol. 4 (1960), 491-498.

3. Berscheid, Ellen, and Elaine Walster. Interpersonal At-traction (Reading, Mass.: Addison-Wesley, 1969).

4. Bowers, David G., and Stanley E. Seashore. "PredictingOrganizational Effectiveness with a Four-Factor Theoryof Leadership," Administrative Science Quarterly, Vol. 11(1966), 236-263.

5. Calder, Bobby J. "An Attribution Theory of Leadership,"in B. Staw and G. Salancik (Eds), New Directions in Or-ganizational Behavior (Chicago: St. Clair Press, 1976), inpress.

6. Cartwright, Dorwin C , and Alvin Zander. Group Dynam-ics: Research and Theory, 3rd ed. (Evanston, III.: Row,Peterson, 1960).

7. Cole, Jonathan R., and Stephen Cole. Social Stratifica-tion in Science (Chicago: University of Chicago Press,1973).

8. Collins, Barry E., and Harold Guetzkow. A Social Psychol-ogy of Group Processes for Decision-Making (New York:Wiley, 1964).

9. Collins, Randall. "Functional and Conflict Theories ofStratification," American Sociological Review, Vol. 36(1971), 1002-1019.

10. Day, R. C , and R. L. Hamblin. "Some Effects of Close andPunitive Styles of Supervision," American Journal of So-ciology, Vol. 69 (1964), 499-510.

11. Domhoff, G. William. Who Rules America? (EnglewoodCliffs, N.J.: Prentice-Hall, 1967).

12. Dubin, Robert. "Supervision and Productivity: EmpiricalFindings and Theoretical Considerations," in R. Dubin,G. C. Homans, F. C. Mann, and D. C. Miller (Eds.), Lead-ership and Productivity (San Francisco: Chandler Pub-lishing Co., 1965), pp. 1-50.

13. Fiedler, Fred E. "Engineering the Job to Fit the Manager,"Harvard Business Review, Vol. 43 (1965), 115-122.

14 Fiedler, Fred E. A Theory of Leadership Effectiveness(New York: McGraw-Hill, 1967).

15. Gamson, William A., and Norman A. Scotch. "Scapegoat-ing in Baseball," American journal of Sociology, Vol. 70(1964), 69-72.

16. Granovetter, Mark. Getting a lob (Cambridge, Mass.:Harvard University Press, 1974).

17. Hall, Richard H. Organizations. Structure and Process(Englewood Cliffs, N.).: Prentice-Hall, 1972).

18. Halpin, A. W., and J. Winer. "A Factorial Study of theLeader Behavior Description Questionnaire," in R M.Stogdill and A. E. Coons (Eds), Leader Behavior: Its De-scription and Measurement (Columbus, Ohio: Bureau ofBusiness Research, Ohio State University, 1957), pp. 39-51.

19. Hargens, L. L. "Patterns of Mobility of New Ph.D.'sAmong American Academic Institutions," Sociology ofEducation, Vol. 42 (1969), 18-37.

20. Hargens, L. L., and W. O. Hagstrom. "Sponsored andContest Mobility of American Academic Scientists," So-ciology of Education, Vol. 40 (1967), 24-38.

21. Harrell, Thomas W. "High Earning MBA's," PersonnelPsychology, Vol. 25 (1972), 523-530.

112 The Ambiguity of Leadership

22. Harrell, Thomas W., and Margaret S. Harrell. "Predictorsof Management Success." Stanford University GraduateSchool of Business, Technical Report No. 3 to the Officeof Naval Research.

23. Heller, Frank, and Gary Yukl. "Participation, ManagerialDecision-Making, and Situational Variables," Organiza-tional Behavior and Human Performance, Vol. 4 (1969),227-241.

24. Hollander, Edwin P., and James W. Julian. "Contempo-rary Trends in the Analysis of Leadership Processes,"Psycho/ogical Bulletin, Vol. 71 (1969), 387-397.

25. House, Robert J. "A Path Goal Theory of Leader Effec-tiveness," Administrative Science Quarterly, Vol. 16(1971), 321-338.

26. Hunt, J. G. "Leadership-Style Effects at Two ManagerialLevels in a Simulated Organization," Administrative Sci-ence Quarterly, Vol. 16 (1971), 476-485.

27. Kahn, R. L., D. M. Wolfe, R. P. Quinn, and J. D. Snoek.Organizational Stress: Studies in Role Conflict and Am-biguity (New York: Wiley, 1964).

28. Karabel, J., and A. W. Astin. "Social Class, Academic Abil-ity, and College 'Quality'," Social Forces, Vol. 53 (1975),381-398.

29. Kelley, Harold H. Attribution in Social Interaction (Mor-ristown, N.J.: General Learning Press, 1971).

30. Kerr, Steven, and Chester Schriesheim. "Consideration,Initiating Structure and Organizational Criteria—An Up-date of Korman's 1966 Review," Personnel Psychology,Vol. 27 (1974), 555-568.

31. Kerr, S., C. Schriesheim, C. J. Murphy, and R. M. Stogdill,"Toward A Contingency Theory of Leadership BasedUpon the Consideration and Initiating Structure Litera-ture," Organizational Behavior and Human Perform-ance, Vol. 12 (1974), 62-82.

32. Kiesler, C , and S. Kiesler. Conformity (Reading, Mass.:Addison-Wesley, 1969).

33. Kochan, T. A., S. M. Schmidt, and T. A. DeCotiis. "Supe-rior-Subordinate Relations: Leadership and Headship,"Human Relations, Vol. 28 (1975), 279-294.

34. Korman, A. K. "Consideration, Initiating Structure, andOrganizational Criteria—A Review," Personnel Psychol-ogy,Wo\. ^9 0966), 349-362.

35. Lieberson, Stanley, and James F. O'Connor. "Leadershipand Organizational Performance: A Study of Large Cor-porations," /American Sociological Review, Vol. 37 (1972),117-130.

36. Lippitt, Ronald. "An Experimental Study of the Effect ofDemocratic and Authoritarian Group Atmospheres,"University of Iowa Studies in Child Welfare, Vol. 16 (1940),43-195.

37. Lowin, A., and J. R. Craig. "The Influence of Level ofPerformance on Managerial Style: An Experimental Ob-ject-Lesson in the Ambiguity of Correlational Data," Or-ganizational Behavior and Human Performance, Vol. 3(1968), 440-458.

38. Mills, C. Wright. "The American Business Elite: A Collec-tive Portrait," in C. W. Mills, Power, Politics, and People(New York: Oxford University Press, 1963), pp. 110-139.

39. Mintzberg, Henry. The Nature of Managerial Work (NewYork: Harper and Row, 1973).

40. Nealey, Stanley M., and Milton R. Blood. "LeadershipPerformance of Nursing Supervisors at Two Organiza-tional Levels," Journal of Applied Psychology, Vol. 52(1968), 414-442.

41. Pfeffer, Jeffrey, and Gerald R. Salancik. "Determinants ofSupervisory Behavior: A Role Set Analysis," Human Re-lations, Vol. 28 (1975), 139-154.

42 Pfeffer, Jeffrey, and Gerald R. Salancik. "OrganizationalContext and the Characteristics and Tenure of HospitalAdministrators," Academy of Management Journal, Vol.20 (1977), in press.

43. Reed, R. H., and H. P. Miller. "Some Determinants of theVariation in Earnings for College Men," Journal of Hu-man Resources, Vol. 5 (1970), 117-190.

44. Salancik, Gerald R., and Jeffrey Pfeffer. "Constraints onAdministrator Discretion: The Limited Influence of May-ors on City Budgets," Urban Affairs Quarterly, in press.

45. Sales, Stephen M. "Supervisory Style and Productivity:Review and Theory," Personnel Psychology, Vol. 19(1966), 275-286.

46. Schutz Alfred. 7he Phenomenology of the Social World(Evanston, III.: Northwestern University Press, 1967)

47. Selznick, P. Leadership in Administration (Evanston, III.:Row, Peterson, 1957).

48. Spaeth, J. L. and A. M. Greeley. Recent Alumni andHigher Education (New York: McGraw-Hill, 1970).

49. Thompson, James D. Organizations in Action (New York:McGraw-Hill, 1967).

50. Vroom, Victor H. "Some Personality Determinants of theEffects of Participation," journal of Abnormal and SocialPsychology, Vol. 59 (1959), 322-327.

51 Vroom, Victor H., and Phillip W. Yetton. Leadership andDecision-Making (Pittsburgh: University of PittsburghPress, 1973).

52. Warner, W. L., and J. C. Abbeglin. Big Business Leadersin America (New York: Harper and Brothers, 1955).

53 Weick, Karl E. The Social Psychology of Organizing(Reading, Mass.: Addison-Wesley, 1969).

54. Weinstein, Alan G., and V. Srinivasan. "Predicting Man-agerial Success of Master of Business Administration(MBA) Graduates," Journal of Applied Psychology, Vol.59 (1974), 207-212.

55. Wolfle, Dael. The Uses of Talent (Princeton: PrincetonUniversity Press, 1971).

56. ZaId, Mayer N. "Who Shall Rule? A Political Analysis ofSuccession in a Large Welfare Organization," Paci//c So-ciological Review, Vol. 8 (1965), 52-60.