Uploaded by zhangyin114

2. Meindl etal 1985

advertisement
Sage Publications, Inc.
Johnson Graduate School of Management, Cornell University
The Romance of Leadership
Author(s): James R. Meindl, Sanford B. Ehrlich and Janet M. Dukerich
Source: Administrative Science Quarterly, Vol. 30, No. 1 (Mar., 1985), pp. 78-102
Published by: Sage Publications, Inc. on behalf of the Johnson Graduate School of Management,
Cornell University
Stable URL: http://www.jstor.org/stable/2392813
Accessed: 28-10-2015 13:43 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected]
Sage Publications, Inc. and Johnson Graduate School of Management, Cornell University are collaborating with JSTOR
to digitize, preserve and extend access to Administrative Science Quarterly.
http://www.jstor.org
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
The Romanceof
Leadership
James R. Meindl,
Sanford B. Ehrlich,and
Janet M. Dukerich
? 1985 by Cornell University.
0001 -8392/85/3001-0078/$1 .00.
We are grateful to three anonymous ASO
reviewers for their helpful comments.
This research is an attempt to explore and understand the
prominence of the concept of leadership in our collective
consciousness. Inthree archival studies, we examined the
attention and interest in leadership as reflected in a variety
of publications, in conjunction with national, industry, and
firm variations in performance. In a series of experimental
studies, we examined the effects of performance outcome
levels on the strength of leadership attributions. The results of these studies supported an attributionalperspective in which leadership is construed as an explanatory
concept used to understand organizations as causal
systems; results were interpreted in terms of a romanticized conception of leadership.*
The sheer volume of theoryand researchdevoted to the study
of leadershipover the decades is testimony to its prominence
in ourcollectiveefforts to understandand improveorganizations. However,it has become apparentthat, afteryears of
trying,we have been unableto generate an understandingof
leadershipthat is both intellectuallycompellingand emotionally satisfying.The concept of leadershipremainslargelyelusive
and enigmatic.Criticshave made us aware of a rangeof
scientificdeficiencies that have plaguedrelevanttheories and
research,citingpoormethodology,conceptualproblems,definitionalambiguities,inappropriate
focus, lackof coherence,
and so on (e.g., Bennis, 1959; Stogdill,1974; Miner,1975;
Greene, 1976; Karmel,1978; McCalland Lombardo,1978).
Others have told us that leadershipis best construedas a mere
substitute for and,thus, is functionallyequivalentto other,
more mundaneorganizationalarrangementsand processes
(e.g., KerrandJermier,1978). Stillothers confrontus with
disturbingevidence that ourassumptions aboutthe direct
instrumentalpotency of leadershipon organizationaloutcomes
have vastlyoutstrippedreality(e.g., Liebersonand O'Connor,
1972; Salancikand Pfeffer, 1977). Finally,there are persuasive
argumentsthat cause one to suspect that the greaterrelevance of leadershipas a concept for organizationalscience is
that it is a phenomenologicallyimportantaspect of how observers and participantsunderstand,interpret,and otherwise
give meaningto organizational
activitiesand outcomes (Calder,
1977; Pfeffer, 1977; Pfefferand Salancik,1978). Despite
these assaults on traditionalviews, it appearsthat the concept
of leadershipis a permanentlyentrenchedpartof the socially
constructedrealitythatwe bringto bearin ouranalysisof
organizations.Andthere is every sign that the obsessions with
and celebrationsof it willpersist. The purposeof this analysis
is to shed some lighton this collectivecommitmentto
leadership.
Inourview, the social constructionof organizationalrealities
has elevated the concept of leadershipto a loftystatus and
level of significance.Such realitiesemphasize leadership,and
the concept has therebygaineda brilliancethat exceeds the
limitsof normalscientificinquiry.The imageryand mythology
typicallyassociated with the concept is evidence of the mystery and nearmysticismwith which it has been imbued.A
sample listingof some articleson leadershipthat were found
in recent volumes of the Indexof Business Publicationsre78/AdministrativeScience Quarterly,30 (1985): 78-102
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
flects this imagery:"Leadershipand MagicalThinking";
"BlackArtof Leadership";"IThinkContinuallyof Those Who
Are Great";"ProteanManagerialLeadership";and "Casting
Demons: An Exorcisein Leadership."
Out Organizational
Itappearsthat as observers of and as participantsin organizations, we may have developed highlyromanticized,heroic
views of leadership- what leadersdo, what they are able to
accomplish,and the generaleffects they have on our lives.
One of the principalelements in this romanticizedconception
is the view that leadershipis a centralorganizationalprocess
and the premierforce in the scheme of organizationalevents
and activities.Itamountsto what mightbe considereda faith
inthe potentialif not inthe actualefficacyof those individuals
who occupy the elite positionsof formalorganizationalauthorof leadershipis hintedat in the
ity.The romanticization
observationsmade by a numberof social and organizational
analystswho have noted the esteem, prestige, charisma,and
heroismattachedto variousconceptions and forms of leadership(e.g., Weber, 1946; Klapp,1964; House, 1977; Burns,
1978; Goode, 1978; McCalland Lombardo,1978; Staw and
Ross, 1980; March,1981).We suspect that the romanticized
conceptionof leadersand leadershipis generalizedand prevalent.The argumentbeing advancedhere is that the concept of
leadershipis a perceptionthat playsa partinthe way people
relevantpheattemptto make sense out of organizationally
nomena. Moreover,inthis sense-makingprocess, leadership
role.
has assumed a romanticized,larger-than-life
An importantpartof the sense-makingprocess involvesan
attemptto generate causal attributionsfor organizational
events and occurrences (Thompsonand Tuden,1959; Weick,
1979). The possibilityof takingan attributionalperspectiveon
leadershipwas firstraisedby Calder(1977)and by Pfeffer
(1977).Since then, there has been a growingbody of research
andtheorydevoted to attributional
analyses of leadership(see
McElroy,1982; Lordand Smith, 1983, for recent reviews).
However,that literature,with butfew exceptions (e.g., Phillips
and Lord,1981), has not dealt in a directway with the basic
issue raisedby Calderand by Pfeffer,which we are addressing
here: namely,leadershipis perhapsbest construedas an
explanatorycategorythat can be used to explainand account
for organizational
activitiesand outcomes. Staw (1975)
reacheda similarconclusion,but in a more generalcontext, by
arguingthat the self-reportedopinionsand beliefs of organizationalactors and observers regardingcausalitymay infact
constitute attributional
inferences ratherthanactualcausal
determinantsof events and occurrences. Unfortunately,most
researchershave respondedby focusing narrowlyon the
methodologicalramificationsof this view (e.g., DeNisiand
Pritchard,1978; Downey, Chacko,and McElroy,1979; Binning
and Lord,1980; McElroyand Downey, 1982), forthe most part
ignoringthe wider, underlyingimplicationthat manyorganizationalbehaviorconcepts can be used by individualsto form
coherent explanationsof events and occurrences.This is precisely the premise fromwhich the present analysisproceeds.
The significanceplacedon leadershipis a response to the
ill-structured
problemof comprehendingthe causal structure
of complex, organizedsystems. Imaginefora moment the
problemfaced by an observerwho must comprehenda large
79/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
and complex system: there are manycausalforces to consider
and they occurtogether in highlyintricateand overlapping
networks,complete with multipleinputsand outcomes,
numerousfeedback loops, and allexisting in some dynamic
state of flux.Totalcomprehensionof the system will easily be
beyondthe power of the observer. Insuch a task, the particular
understandingthat is gainedwilldepend at least as much on
the characteristicsof the observeras it does on the system
itself. Ourinformal,implicitlyheld models and perhapsour
more formaltheories, as well, are limitedresponses to the task
of comprehendingthe causal complexitiesthat characterizeall
organizations.Of course, the potentialways inwhich an understandingcan be achieved are many,and it would be difficultto
choose among them on a purelyrational,logicalbasis. However, what is attended to andwhat causalfactors emerge as the
"figure"againstthe backgroundof allother possibilities,even
if arbitrary
with respect to the system, is not randombut is a
process guided by the psychologyand sociology of the observer. Ineffect, the results representa systematic bias about how
a system is understood,how relevantevents and outcomes
are defined and explained,and to what factorsthey are attributed. The term "bias"is used here inthe way Schlenker
(1982: 205) defined it: "Abias inthe interpretationand explanationof events is a subjectivetendency to preferone
interpretationover another;such an interpretationmay or may
not be an erroraccordingto some 'objective'criterionfor
assessing the event." Such preferences occur, in part,because of the ambiguityof relevantinformationand the perceived importanceof events. The romanticizedconceptionof
leadershipresultsfroma biased preferenceto understand
importantbutcausallyindeterminantand ambiguousorganizationalevents and occurrencesinterms of leadership.Accordingly,inthe absence of direct,unambiguousinformationthat
would allow one rationallyto inferthe locus of causality,the
romanticizedconceptionof leadershippermitsus to be more
comfortablein associating leaders- by ascribingto them
controland responsibility-with events and outcomes to
which they can be plausiblylinked.'
1
This collective, idealized representation of
leadership bears a certain similarityto
what has been examined under the general rubricof "implicit leadership theories"
(e.g., Rush, Thomas, and Lord, 1977; Staw
and Ross, 1980); however, we presumed
the conception to be much more generic,
powerful, and less well articulated.
The Present Research
The researchreportedhere examinedthe hypothesis that the
relativeprominenceof the use of leadershipin understanding
complex, organizedsystems variesto a significantdegree with
the performancelevels of such systems. Generallyspeaking,
the need to understandand make sense shouldcorrespondto
the occurrenceof salientevents (Anderson,1983). Itis possible that observers are generallyproneto overestimatethe
impactof leadershipintheirexplanationsof events; however,
it seems likelythatvariationsin events would be importantfor
uncoveringany bias towardunderstandingevents interms of
view of
leadership.One implicationof a heroic,larger-than-life
leadershipis that its effects on an organizationare not trivial.
Thatis, associations between leadershipand events will be
consistent with the romanticizedconceptionandwilltherefore
be most appealingwhen those events are in some way
defined as extraordinary
(i.e., largecause, largeeffect). We
reasonedthat the romanticizedconceptionwill have greatest
sway in extreme cases - eithervery good or very poor
performance- causingobservers to understandthese events
80/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
in terms of leadership.A strongeremphasis on leadership
shouldoccur underconditionsinwhich high-magnitudeoutcomes obtain,andweaker preferences should be foundwhen
low-magnitudeoutcomes obtain.We know frompast research
that leadersare often held responsibleand are "scapegoated"
for poororganizationalperformance(e.g., Gamsonand Scotch,
1964). Otherevidence suggests that informationabout performanceis sometimes used to inferthe good and badquality
of leadershipthat must have existed (e.g., Staw, 1975). Thus,
we expected that a bias towardleadershipcould be systematicallyrelatedto performancelevels in a positive or negative
way. These ideas were tested in a series of archivaland
experimentalstudies.
ARCHIVAL
STUDIES
Inthe archivalstudies, we attempted to findevidence of the
bias representedinthe romanticizedconceptionof leadership
by explaininghow, if at all,an interest in leadershipis associated with the performanceof firms,industries,and the national
economy. Inorderto do so, we examinedpublishedsources
and dissertationsforthe appearanceof leadershipas a topic of
interestand attention.The workingassumptionwas that an
analysisof the correspondencebetween attentionto leadershipand performancescould providean indirectand very
broadindicationof the extent to which outcomes are collectivelyunderstoodinterms of and attributedto leadership.In
Study 1, we examinedthe relativeemphasis on corporate
leadersand leadershipinthe popularpress. InStudies 2 and 3,
we focused on the correspondencebetween variationsin
nationaleconomic performanceand the generalemphasis
placedon leadershipby young scholarsand by the business
communityin general.Allthree studies were designed to test
the hypothesisthat the amountof interest in and attention
devoted to leadershipinthe publicationsstudied would vary
directlyor inverselywith generalperformance.
STUDY1: THEPOPULARPRESS
Method
Forthis study, we examinedtitles of articlespublishedin the
WallStreet Journal,from 1972 through1982, on a sample of
34 business firmsdrawnfromthe Fortune500 list of largeU.S.
corporations.We measuredthe amountof attentionand publicitythis well-knownpublisherof business news devoted to
the topic of corporateleadershipand determinedwhether or
not that attentionboreany relationshipto performancelevels
- defined here in generalterms of the sales or profitgrowthof
the firmsand industriesinvolved.The WallStreet Journalwas
chosen because it has an impeccablereputationas a highly
crediblesource of business news, it has an extraordinary
readership,and it is perhapsone of the most powerful,leading
publicationsinthe world(Neilsonand Neilson, 1973). For
performancedataforthe same period,we reliedon the Value
LineInvestmentSurvey.
Ourselection of firmswas guided by several considerations.
First,we attempted to sample a rangeof differentindustries,
with several sample firmsin each industry.We also triedto
choose firmsthat showed a rangeof differentperformance
81/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
curves over the time period.Finally,we wanted to select firms
that we felt would have receivedample press coverage during
those years. We hadoriginallysampled 35 firms; however, we
realizedlaterthatwe had unwittinglyselected one firmthat
was a whollyowned subsidiary,and it was thereforedropped.
The finalsample of firmsis listed inTable1.
Table 1
Sample of Firms
Abbott Labs
Allied Chemical
American Airlines
American Cyanamid
American Home Products
American Motors
Armco
Bethlehem Steel
Boeing
Bristol Myers
Burroughs
Chrysler
Continental Airlines
Data General
Delta Airlines
DuPont
Eastern Airlines
Ford Motor Company
General Electric
General Mills
General Motors
Hewlett-Packard
IBM
Lilly
Lockheed
McDonnell-Douglas
McGraw-Edison
NCR
Pan Am
RCA
Republic Steel
Texas Instruments
U.S. Steel
Westinghouse
Emphasis on leadership. To get an estimate of the extent to
which corporateleadershipwas emphasizedfor a given firmin
a given year,we used the annualindexof the WallStreet
Journal,which containsa listingby title, undereach large
corporation,of every articleon the corporationappearinginthe
WSJinthat year. Ourprocedurewas to readthe titles under
the headingsfor each firminthe sample and for every year.An
.articlewas classifiedas leadershiporiented(LA)if its title
includeda keywordor phrasethat appearedin a "dictionary,"
developed specificallyforthis research,containinga short,
selected list of items. The items includedreferences to names
of corporateofficers, references to senior executive positions,
and phrases such as "top management,""seniorexecutive,"
"top brass,"and otherdescriptorscommonlyused to referto
corporateleadership.Insome cases, whether or not an item
was a keyworddepended on its use in the context of the title.
Forexample, the word "management"was includedif it
referredto the administratorsof the firm,but it was excluded
when the referentwas a process, as in "the managementof
innovation."Ifthe title did not includea listed item, then the
articlewas assigned to an "other"(OA)category.Two coders
hadthe task of scanningthe titles and coding articles.Each
coder was assigned responsibilityfortabulatingthe frequencies in each categoryon a firm-by-firm,
yearlybasis, for halfthe
sample. The two coders underwentseveral preliminaryexercises inwhich they were asked to search and classifythe
articlesfroma numberof pages of the Index,using the
dictionaryto guide theirdecisions. These exercises led to
some modificationsinthe dictionary.Insubsequent trialruns,
each coder independentlysearched and classifiedthe articles
fromtwo pages selected at randomfromthe Index.The extent
of theiragreement was scored, revealingan errorrateof less
82/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
than3 percent.The coders' tabulationssummarizingthe number of articlesclassifiedas LArelativeto the numberof articles
classifiedas OAwas taken as a roughindicationof the degree
to which leadershipwas being emphasized.Thisemphasis
was capturedina "LeadershipQuotient"(LQ),equivalentto
the ratio,(LA/OA)x 100.
Of course, there are some obvious, inherentlimitationsin
using titles to classify articles.Ifnothingelse, measurement
errorwould be increasedthroughany misclassifications.
However,there are several considerationsthatjustifiedits use
and providedus with at least some reassuranceabout its
suitabilityfor this research.First,this method allowed us to
scan and code a very largenumberof references in a reasonable periodof time, thus enablingus to expandthe numberof
data pointsfarbeyondthat possible throughan analysisof the
actualcontents of articles.Second, titles are usuallyintended
to highlightthe mainthemes of an article,which suggests that
there is a reasonablecorrespondencebetween title and content. Third,even if the correspondencebetween titles and
content is a loose one, references to leaders inthe title head of
an articlesymbolicallyemphasize the concept of leadership,
increasingits prominencerelativeto other concepts and thereby producingan implicitassociationbetween top management
and whatever informationthen appearsinthe articleor elsewhere aboutthe firm.Fourth,our method is conceptually
consistent with the systems used by a numberof well-known
and popularlibrarydata bases. Forexample, the SocialSciences CitationIndex(SSCI)and the EducationalResources
InformationCenter(ERIC)data bases both make use of titlekeywordsystems to classify publicationsintoa varietyof
content areas.2Althoughthese considerationscannotgive us
perfect confidence,when takentogether they allowed us to
feel reasonablycomfortablewith ourtitle scan and classification procedure.
2
Unlike the SSCI data base, which relies
exclusively on a title-keyword system, the
ERICdata base uses both a title-keyword
system and a controlled vocabulary system. Under the controlled vocabulary system, assignment to a particularcontent
area ("access code") is made by a coder
who reads publications for their content
themes. We conducted separate searches
of the ERICdata base using the keyword
descriptor method and the controlled vocabulary method to retrieve the number of
entries on a yearly basis, for every year the
data base was available (1966-1983),
under the access code "leadership." The
number of entries retrieved under each
system were highly correlated,
r(l6 df) =.94, p<.O01.
Results and Discussion
A total of 33,248 articlesaboutthe firmsin oursample
appearedin the WSJover the periodexamined.Of those,
2,832 hadtitles that were coded as emphasizingthings other
than leadership(OA).The average numberof articlesfor any
given firm,in any given year,was 88.90. Of those, an average
of 7.57 were coded as LA,rangingfroma minimumof zero to a
maximumof 59. The average numberof OAarticleswas
a minimumof 6 to a maximumof 995. The
81.33, ranging-from
averageyearlyLQwas 14.48 percent, rangingfroma minimum
of 0 percentto a maximumof 70 percent.The comparable
figurefor mean annualsales growthwas 13.33 percent, ranging froma low of - 45.57 percentto a highof 131.03 percent.
Year-wiseanalysis.To findevidence that a generalemphasis
on leadershipis associated with performance,we examined
the yearlyLQ'sforthe entiresample of firmsin conjunction
with theiryearlyperformances.The results indicatedthat LQ
scores were positivelyrelatedto performance,measured here
interms of yearlyannualsales growth,r(9df)= .53, p<.05.
Thissuggests thatyears inwhich companies are on average
doingwell are also the years inwhich leadershipon average
tends to be more highlyemphasized.
83/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Analysisby industry.To examinethe relativeemphasis on
leadershipwith respect to differentindustryperformances,we
classifiedthe 34 firmsin oursample intoten majorindustrial
groupingson the basis of the firsttwo digitsof theirSIC
designations.The mean annualincrease in sales performance
across these industrieswas 10.37 percent,with a standard
deviationof 2.70 percent, rangingfroma highof 15.80 percent,
to a low of 6.92 percent.The numberof firmsin oursample in
the same industrialgroupwas quite small: usuallythree or
four,but in one case two, and inanothercase six. Evenso, a
one-way ANOVArevealedthat the mean LQvariedsubstantially across the industrygroups, F(9,34)= 2.28, p<.05. Moreover,
the variancein LQappearsto be systematicallyrelatedto
industryperformance:a plannedcomparisonrevealedthat
firmsassociated with the five highest performanceindustries
had, on average, significantlyhigherLQ'sthanthose firms
associated with the lowest performanceindustries,
F(1,33) = 8.99, p<.01. Thatfindingwas corroboratedby a
significantcorrelationbetween averagefirmLQand industry
performance,r(8df)= .64, p<.05.
Company-by-company
analysis.To examine how the emphasis on leadershipmayvaryin relationshipto a firm'sown
performancesover time and how that relationshipmayvary
across firms,we conductedseparate analyses for each firm,
correlatingLQwith annualperformance.Since we have data
availablefor only 11 years, the potentialdegrees of freedom
availableforthese analyses are quite small (df= 9). However,
we felt that the 34 replicationscould provideus with a reasonablygood estimate of the pervasiveness of the expected
effect. Giventhe inherentdifficultiesof specifyinga prioriwhat
definitionof performancewill be used to make inferences
aboutand associations to the leaders of any given firm,and
since performanceof a single firmis often judgedinterms of
how well it is doing relativeto others in its own industry,we
expandedthe generalperformanceoutcomes forthis analysis
to includegrowth in profitsand sales relativeto the comparable industry-widefigures.The results of these analyses indicated that for 25 of the 34 firms(74 percent),LQwas significantly(p<.09 or greater)associated with at least some of our
definitionsof performance.If50 percent is used as an extremelyconservativeexpected value, then a simple onedegree-of-freedomchi-squaretest suggests this is a nonrandom pattern;x2(1 df)= 6.89, p<.01 9. Also, of the 25 firms
showing an association between LQand performance,16 (64
percent)were positive,and the remaining9 (36 percent)were
negative.
Within-yearanalysis.Ourfinalanalysisfocused on the covariation of LQand performanceacross companies in each of the
11 years. The data summarizedinTable2 show that, in every
year examined, LQwas correlated(p<.08 or greater)with
performanceoutcomes. Ineight of those years, the significant
correlationswere negative, indicatingthat in each of those
years, the poorerthe performance,the more leadershipwas
emphasized. Inthe remainingthree years, the significantcorrelationswere positive,such that the better the performances,
the greaterthe emphasis on leadership.
84/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
Table 2
Within-year Analysis: Direction of Significant Correlations between Leadership Quotient and Company
Performance Measures
Year
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
neg*
neg
neg
neg
neg
neg*
Pos
Pos
Pos
neg
neg
*p<.08; all others are p<.05 or greater.
The foursets of analyses gave us an opportunityto gain
somewhat differentperspectives on the dataand providedus
with differentfocal pointsfor examiningthe tendency to
associate leadershipwith performance.Ineach of the analyses, the weight of evidence supportedour expectationsthat
the emphasis on a firm'stop managementwillvarysignificantly with performancelevels. The industryand the year-wise
analyses revealedthat an emphasis on leadershipincreases
with increasinglypositive performance.The within-company
andwithin-yearanalyses introducedadditionalevidence that,
on some occasions and for some firms,leadershipis more
likelyto be emphasizedwhen performancesare poor.These
two majorpatternsof results,when takentogether, provided
us with initialsupportfor the propositionthat the perceived
causal priorityof and attributionsto leadershipin understandevents and occurrencesare likelyto occur
ing organizational
when performancesare eithervery good or very bad.
TOPICS
STUDY2: DISSERTATION
Withthe evidence obtainedfromStudy 1, we turnedour
attentionto exploringthe societal aspect of ourtheory,which
suggests that the level of collective interestand significance
invested inthe concept of leadershipis responsiveto fluctuations inthe generaleconomic performanceof the entire nation.
Inorderto test that notion,we chose, inthis study, to trackthe
level of interest in leadershipthroughthe dissertationtopics
young scholarschose. We assumed that the commitmentand
devotion representedby a dissertationtopic would provideus
with a glimpse of the collective investment in the concept of
leadership.
Method
We counted the numberof doctoraldissertationsdevoted to
the topic of leadershipand relatedit to the generaleconomic
source of
conditionsover the years 1929-1983. Our-primary
informationwas DissertationAbstractsInternational(DAI),an
recognizedreferencetool that summarizesand
internationally
indexes virtuallyallthe currentdissertationsaccepted inthe
U.S. and Canada(DAIUser's Guide,1983). We used the
subject index,which lists and groupsdissertationsintoover
200 specializedsubject headings,one of which is "leadership."The numberof dissertationsappearingevery year
underthat heading("LD")formedthe basis of ouranalysis.
However,because DAIdid not give comparabledatathat
would allow us to estimate easily the total numberof dissertations in allthe social sciences, we obtainedan estimate from
anothersource, AmericanDoctoralDissertations.We used
85/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
theirannualfiguresto findthe total numberof social science
dissertationsaccepted each year ("TD").Inorderto estimate
generaleconomic conditions,we reliedon figures publishedby
the EconomicStatistics Bureauof Washington,DC,in their
Handbookof Basic EconomicStatistics,to compute year-toyear percentage changes in the GNP(deltaGNP).Thismeasure was chosen because it is a very broadandfamiliar
indicatorof swings inthe nation'seconomy.
Results and Discussion
From1929 to 1983 there was a dramaticincrease in the
numberof dissertationsawarded.In 1929, there were under
2,000; in 1979, there were over 35,000. This historicaltrend
showed up in ourpreliminary
analysisas a very strong correlation between years andTD,r= .91, p<.001, and LD,r= .81,
p<.001. Thus, in lightof this strong historicaltrend,we controlledforyears, throughpartialcorrelations,to examine the
relationshipbetween economic conditionsand interest in leadership.We also formeda ratio,LD/TD,which yielded a leadership quotient (LQdissertations)
conceptually similar to that used
in Study 1. Italso seemed reasonableto expect a lag of several
years between economic conditionsand completed dissertations, althoughwe could not specify exactlywhat that lag
would be. On the basis of these considerations,ouranalyses
focused on the lagged (0 to + 6 years) partialcorrelations
between delta GNP and LMdissertations,
controlling for linear,
historicaltrends.Table3 shows that the relationshipbetween
delta GNP and LQdissertations
was negative, indicating that downTable 3
Lagged Partial Correlations between
General Economy (Delta GNP)
Uldissertations
and Changes in the
Lag
0
+1
+2
+3
+4
+5
+6
-15
-22
[email protected]
[email protected]
-3900
-17
-02
8p<.05; *[email protected]<.01.
turnsin the growthof the economy were subsequentlyfollowed by a greaterinterest in leadership,relativeto allother
topics and aftercontrollingfor historicaltrends. This relationshipbecomes reliableaftera two-yearlag, reachingits
highest level inthe plus-fourthyear, andthen dropsoff. These
results, then, suggest that there is an association between
good or badeconomic times andthe interest in leadership,at
least among scholarschoosing dissertationtopics.
STUDY3: GENERALBUSINESSPERIODICALS
Thisstudy was conceptuallysimilarto Study2. Thistime,
however, ourstrategywas to focus more specificallyon the
business community.Accordingly,we deliberatelychose an
availabledata base thatwas muchwider in scope thanthat
used in Study2 and capturedto a greaterdegree the interests
of the generalbusiness communityin the topic of leadership.
Giventhe resultsof Study2, we expected that the negative
86/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
relationshipbetween the state of the generaleconomy and
interest in leadershipwould be replicatedhere. Thisstrategy
allowed us to observe whether or not the generalbusiness
community'scollective interest in leadershipis also responsive
to fluctuationsinthe nationaleconomy. Ifso, then we could
of ourguiding
have added confidence inthe generalizability
hypothesis. Inaddition,the study affordedus an opportunityto
determineif interest is such that it tends to emphasize leadershipto a greaterdegree duringgood or duringbad economic
times.
Method
We examinedthe annualvolumes from 1958 to 1983 of the
Business PeriodicalIndex(BPI),publishedby the H.W. Wilson
Company,which consists of subject entries for a wide rangeof
business-orientedpublications.In1981 alone, the contents of
over 250 differentperiodicalswere indexedand groupedinto
hundredsof subject headings,one of which is "leadership."
Althoughscholarlyjournalssuch as ASQ andAMJare indexed,
the majorityof the publicationsindexedare nonacademicand
Forexample, this indexincludespopular
practitioner-oriented.
periodicals,such as Barrons,Business Week, Forbes,and
Fortune,as well as more specialized,often industry-specific
publications,such as ChemicalWeek, ElectronicsWorld,and
Pipelineand GasJournal.Because of these characteristics,
this databasewas chosen over others, such as ERIC,or SSCI.
Inaddition,the indexwas publishedfrom 1958 to the present
-the longest runningperiodwe couldfind.
As in Study2, obtaininga yearlyestimate of the interest in
leadershipentaileda simple count of the numberof titles listed
underthe subject heading,"leadership"(LABpI).
However, no
publisheddatawere availableon the total numberof articles
indexed (TABpI),
andthis hadto be estimated. Fortunately,
because the format,page size, andtype size have remained
the same across volumes andyears, we were able to obtain
the average numberof entries per page by drawinga sample
of 50 pages (two pages for every year)and then countingthe
numberof entries on each (M= 65.24; SD = 7.15). We then
multipliedthe numberof pages in each volume by the average
numberof entries per page to arriveat a yearlyestimate of
TABPI.
Results and Discussion
Therehas been a stronggrowth inthe numberof business
periodicalspublishedover the years, and this historicaltrend
was reflected in ourdata by the zero-ordercorrelationsber=.83,
tween years and TABp1,
r=.88, p<.001, and LABPI,
p<.001. Inan analysisparallelto that inthe previousstudy, we
and delta
examinedthe relationshipbetween LQBPI(LA/TA)
GNP,controllingfor that linearhistoricaltrend.The partial
correlationswere lagged inthe same manneras in Study2 (0
to + 6 years),althoughlittledelaywas anticipated,given that
the intentof the majorityof the periodicalsis to stay current.
The results, summarizedinTable4, show that, as in Study2,
there appearedto be some association between economic
performanceand interest in leadership,aftercontrollingfor
historicaltrends. However,unlikeinthe previousstudy, the
relationshipis predominantlypositive,suggesting that the
interest in leadershipinthe generalbusiness community,at
87/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
least interms of publications,seems to be at its greatest levels
when there are upswings inthe nation'seconomic growth.
Apparently,the relationshipis more immediateintime than
that foundwith the dissertationdata,which is not surprising,
given the differences inthe natureand goals of those publications. However, why LQdissertations
in Study 2 varied inversely
andwhy in this study LQBPIvarieddirectlywith delta GNPis
intriguing.Perhapsthose patternsrepresentsome underlying
differences between academicand practitioner-oriented
views. Whateverthe case, the relationshipsare not likelyto
representrandomassociations and are both generallyconsistent with ourexpectations, if not.indirection,at least interms of
degree.
Table4
Lagged PartialCorrelationsbetween LQpBP
and Changes in the General
Economy (Delta GNP)
Lag
0
.67000
+1
+2
+3
+4
+5
+6
.4800
.52--
.02
-.33
-.35
-.02
Op<.05; *-0p<.01; 000p<.001.
EXPERIMENTAL
STUDIES
The precedingarchivalstudies providedreasonablyclearevidence of a general relationshipbetween performanceoutcomes and degree of emphasis on leadership.The following
series of experimentswas designed to test more preciselythe
notionthatthe use of leadershipas an explanatoryconcept in the formof causal attributions- varieswith performance.In
particular,given the theoreticalargumentsand the patternof
positive and negative relationshipsuncoveredin the archival
studies, we sought to determineif, undercontrolledexperimentalconditions,leadershipattributionswould indeed be
more likelyto occur- andtherebycreate a strongerassociation- when performanceis eithervery good orvery bad. In
the three experimentsreportedhere, business school students were presented with minimalinformationandwere
asked to accountfor instances of performancethatvariedin
terms of the magnitudeof outcomes. Ineach case, they were
asked to considera leaderas a possible reason for the outcome event. Forcomparisonpurposes, individuals'attributions
to alternativedeterminantsof performanceotherthanto the
leaderwere also obtained.Study4 provideda partialtest of the
hypothesis by examiningattributional
patternswhen observers
were presented with informationthat variedthe magnitudeof
positive performanceoutcomes. Study5 provideda more
complete test of the hypothesis by replicatingand extending
Study4 and includedconditionsthat variedthe magnitudeof
negative as well as positive performanceoutcomes. Studies 4
and 5 laidthe groundworkfor Study6, which attempted to
replicatethe patternof results undermore refinedconditions
and began to explorethe roleof expectations on leadership
attributions.AlthoughStudies 4 and 5 were preliminary,because they were instrumentalto the development of Study6,
we will brieflydescribe them here.
88/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
STUDY4
Method
Fifty-nineundergraduatesenrolledinan introductoryorganizationalbehaviorcourse at a largenortheasternuniversityparticipated inthis study. Theirmean age was 21.90 years, and on
the averagethey reportedhavingthe equivalentof 2.56 years
of workexperience.
Subjectswere randomlyassigned to readone of three different versions of an extremelybrieforganizationalperformancerelatedvignette. Eachversioncontainedthe same summary
descriptionof an organizationalunitand its members, including
the leader.The vignettes differedonly interms of the information they providedon performanceoutcomes, which were
defined interms of sales increase. Low, moderate,and high
magnitudeeffects were conveyed to subjects by providing
them with informationthat the unithadexperiencedeithera
slight (2 percent),moderate(10 percent),or large(25 percent)
increase in sales performance.The vignettes readas follows:
John Smithis the Directorof Sales fora majornortheasternappliance
firm.John assumed this positionfive years ago followinghis attainment of an MBAdegree. Priorto his MBA,John hadcompleted a
bachelor'sdegree in Marketing.Inthis positionhe has gainedthe
respect of both his subordinatesand superiors.On his last evaluation
John was ratedas a capableworkerand his subordinateshave
indicatedthatthey enjoyworkingfor him.John currentlyis in charge
of five subordinates.Allof the subordinatesworkinginJohn's department have a good workingknowledgeof marketingprinciplesas
demonstratedby theirpriorand currentworkexperience.At the end
of the fourthquarter,new customeraccounts hadshown a slight/
moderate/largeincrease(2%/10%/25%)duringthe year,over last
year's performance.
Immediatelyafter readingthe vignettes, subjects were asked
to rate(on a 7-pointscale) the extent to which they considered
the leaderto be an importantcausaldeterminantof the performanceoutcome.3And,in orderto insurethat subjects were
aware of and at least consideredother, perhapscompeting
explanationsforthe outcome, parallelquestions asked them to
express the extent to which alternative,plausiblefactors may
have contributedto the outcome, includingother actors (subordinates),environmentalfactors (generaleconomy), and anythingelse they felt shouldbe considered (other).Responses to
these last questions were consideredtogether as "alternative" attributionsand were thereforeaggregated,for conceptualand analyticpurposes.
3
We considered using open-ended response measures; however, recent evidence suggests that structured, scale
measures similar to those used here are
generally preferable to other methods for
assessing attributions (Eligand Frieze,
1979).
Results and Discussion
Attributions("leader"vs. "alternatives")were examinedconjointlywith outcome effects (low, moderate,and highmagnitudes) in a 3 x 2 ANOVA.The data inTable5 revealthat the
general level of attributionmakingdid not differacross the
three magnitudeconditions(overalllow M= 5.08; overallmoderate M=4.93; overalllargeM= 5.03), F(1,56)= .21; ns. The
analysisalso revealedthat ingeneral,attributionsto leader
(overallM= 5.27) were preferredto attributionsto alternatives
(overallM=4.75), F(1,56)= 11.59, p<.001. Most importantly,
however, and as expected, the maineffects were qualifiedby
an interactionbetween type of attributionand magnitudeof
outcome, F(2,56),p<.06, indicatingthat largermagnitudeout891ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Table 5
Mean Attributions in Each Magnitude Condition: Study 4
Attributions*
Leader
Alternatives
Low
(N= 19)
5.10
5.00
Magnitude of Increase
Moderate
(N= 20)
5.20
4.67
High
(N= 20)
5.50
4.55
*7-point scales; higher scores indicate stronger attributions.
comes caused greater use of the leader as an explanation and
less use of alternative explanations. A planned comparison
between the leader and alternative attributions in the large
magnitude condition (M = 5.50 and M = 4.55, respectively) was
significant, F(1,56)= 13.43, p<.001. These two types of
attributions were not reliablydifferent in the other low and
moderate magnitude conditions.
Thus, the pattern of results provided initialsupport for the
hypothesis that the preference to use leaders in understanding
organizational outcomes increases with increasingly large
magnitudes of positive effects. Although, by itself, the increase in attributions to the leader was not great, the trend
upward is compelling when compared with the "baseline"
provided by the use of alternative explanations. Such comparisons reveal that the increase in leadership attributions occurred despite the fact that attributions to alternatives
decreased.
STUDY 5
The support found in Study 4, although suggestive, is limited
by the fact that only positive performance conditions were
examined. In its general form, the hypothesis is indifferent to
the direction of performance changes -the effect should
occur in the negative as well as in the positive cases. Accordingly, the goal in Study 5 was to further verify the hypothesis
by examining attributionalresponses for negative performance
cases, as well, especially in light of the negative associations
uncovered in our archival studies.
Method
One hundred and sixteen undergraduates enrolled in introductory organizational behavior and human resource courses participated in this study. Their mean age was 22.32 years, and on
average they reported having the equivalent of 2.24 years of
work experience.
The vignettes used in Study 4 were modified to accommodate
the inclusion of negative as well as positive outcomes of
varying degrees. That resulted in six different versions: large
negative (25 percent decrease), moderate negative (10 percent
decrease), small negative (2 percent decrease), small positive
(2 percent increase), moderate positive (10 percent increase),
and large positive (25 percent increase) effects. Thus, the
descriptions subjects received ranged from very poor sales
performance at the high-magnitude, negative end, to very
good sales performance at the high-magnitude, positive end.
On the basis of feedback obtained from initial pre-testing, the
90/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
briefdescriptionof the leaderwas made consistent with the
generalpositiveor negative directionof performancechange,
in orderto insurethat he remainedan equallyplausibleexplanationfor alloutcome effects. Thus, inthree cases of
increasedperformance,a somewhat positive impressionwas
conveyed; inthe decreased performancecases, a somewhat
negative impressionwas conveyed. Of course, the description
of the leaderwithineach type (increaseversus decrease in
performance)was held constant.
Immediatelyafter readingthe vignettes, subjects were asked
to ratethe performanceof the uniton a 7-pointscale, from
"extremelypoor"to "extremelygood." As in Study4, subjects
were then asked to attributeperformanceoutcomes to the
leaderand to alternativecauses.
Results and Discussion
The performanceand attributiondataare summarizedinTable
6. Ourpredictionwas that the greatest level of leaderattribuTable 6
Mean Perceived Performance and Attributions in Each Performance Outcome Condition: Study 5
Dependent
Variable
Large
Decrease
(N= 19)
Moderate
Decrease
(N= 19)
Performance Condition
Slight
Slight
Decrease
Increase
(N= 20)
(N= 18)
Moderate
Increase
(N= 20)
Large
Increase
(N= 20)
Attributions*
Leader
Alternatives
Perceivedperformancet
5.00
3.82
2.47
4.26
3.96
3.11
4.55
4.07
3.55
5.05
3.91
5.55
5.10
3.80
6.35
4.94
4.09
5.17
*7-point scales; higher scores indicate stronger attributions.
t7-point scales; higher scores indicate better perceived performance.
4
A 1 x 6 ANOVAon the performance ratings revealed a significant main effect,
F(5,1 10) = 77.22, p<.001, attesting to the
efficacy of the manipulation. The attribution measures were examined in a 2 x 6
ANOVA (one within factor: two types of
attributions; one between factor: six performance levels). This analysis revealed a
significant main effect for type of attribution, such that leader attributions (overall
M = 4.82) were generally stronger than
attributions to alternative causes (overall
M=3.94), F(1,1 10) = 55.38, p<.001. The
main effect for performance level was not
significant, F(5, 1 0) = .99, p = .42; ns, indicating that the level of attributions to all
sources did not vary across conditions. The
interaction between attribution and performance was marginal, F(5, 1 0) = 2.01,
p<.08.
tions would occurat both extremes of the performancecontinuum- i.e., where positiveand negative magnitudesare
greatest - implyinga curvilinearrelationshipbetween performancelevel and leaderattributions.The means across the
six conditionssuggest such a patterndid indeed occur.4
However,the most sensitive and useful test compares subjects' own perceptionsof performanceoutcomes with the
strengthof theirattributionsto the leader.Accordingly,an
orthogonalpolynomialregressionanalysiswas conducted, using subjects' own perceptionsof performanceto predictthe
extent to which leadershipwas used as an explanatoryconcept. The hypothesis, inthis case, is a quadratic(20 polynomial)
model. Linear,quadratic,and cubic models were examined,
and, as predicted,the onlycoefficientto reachsignificance
was associated with the quadraticcomponent, B= 3.09(1.02);
t= 3.02, p<.001. Inaddition,goodness-of-fittests forthe
polynomialmodel at each degree were conducted.These tests
estimated the lackof fit of models at each degree, relativeto
the residualMS fromfittingpolynomialsof higherdegrees.
Thus, a highF ratiois an indicationof a poorfit. These tests
revealedthat the linearmodel produceda significantlypoorfit,
F(2,112)= 4.87, p = .009, while the quadraticmodel provided
the best fit, F(1,1 12)= .62, p= .43. A scatter plot of the data
91/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
confirmedthe U-shapeddistributionof scores. Figure1 shows
the mean attributionsfor subjects at each perceivedperformance level. Similaranalyses conducted on the alternative
attributionsindicatedthat such curvilineartrends did not occur,
lendingadded supportto the hypotheses.
6.00
Predicted
.
Observed
5.25 at
4.50
STRENGTHOF
ATTRIBUTIONS
TO LEADER
_ _
3.75
3.00
2.25
1.2
1.8
2.4
3.0
3.6
4.2
4.8
5.4
6.0
6.6
PERFORMANCE
PERCEIVED
Figure 1. Predictedvalues and mean attributions to leader at each level of
perceived performance: Study 5.
As an even bettertest of the hypothesis,the perceivedperformanceswere used to predictsubjects' preferences for
using the leader as a causal explanation relative to their
tendency to make alternativeattributions.Accordingly,a parallelregressionanalysiswas conducted using the difference
between the leaderand alternativeattributionsas the dependent variable.Again,the quadraticcomponent was significant,
B = 4.60(1.22); t= 3.76, p<.001. Also as expected, the subsequent fit test revealeda poorlinearfit, F(2,112)= 7.34, p<.001,
but a good fittingquadraticmodel, F(1,1 12)= .57, p= .45.
These results,then, paralleledthose of the previousregression
analyses.
When takentogether, the results providedgood supportfor
the hypothesisthat largeroutcomes - whether they are
positiveor negative- are most likelyto lead observers to
make the inferencethat a leaderwas an importantcause.
Nevertheless, several issues remained,and those became the
focus of Study6.
STUDY6
Thisfinalstudy hadtwo generalobjectives. One was to
replicatethe patternof resultsfound in the previousexperiments undersomewhat more refinedconditions.The vignettes used inthose experiments raisedsome issues that
could be relevantto the observed effects. One issue con92/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
cerned the salience and general prominenceof the leaderin
the vignettes. Upon reflection,we had not paidmuch attention
to the positioningand lengthof the leader'sdescriptionin the
vignettes. Itwas possible thatwe had unwittinglyand artificially inflatedthe extent to which the leaderwas subsequently
consideredas an importantcausal determinantof performance. Thatis a concern, because other researchindicatesthat
attributionsin generalare highlysensitive to the contextual
propertiesof causal information,such as saliencyand primacy
(e.g., Jones et al., 1968; Taylorand Fiske, 1978; McArthur,
1981). Infact, it is preciselyfor those reasons that the "main
effects" for type of attribution(leaderversus alternatives)
observed in Studies 4 and 5 must be treatedwith caution.A
less likely,but nevertheless present possibilityis that extreme
outcomes may somehow have been attributedto the leaderin
response to such artificial,externallyinducedprominence,
ratherthan being the results of internalprocesses (e.g., Phillips
and Lord,1981). A relatedissue was the descriptionof the
leader:in Study5, inorderto insurethatthe leaderremained
an equallyplausible,potentialcausal determinantacross the
entirespan of positiveand negative performanceconditions,
he was portrayedsomewhat positivelyinthe three positive,
increasedperformanceconditionsand somewhat negativelyin
the three negative,decreased performanceconditions.
Althoughthis is not necessarilya problem,Study6 allowed us
to make use of an alternativestrategy inwhich alldescriptive
informationon the leaderwas deleted fromthe variousvignettes, andwe were able to clear up any ambiguitiesthat
mighthave been associated with the previousstrategy.
A second generalpurposewas to examine the roleof performance expectations in makingleaderattributions.Otherliterature (e.g., Jones and Davis, 1965; Jones et al., 1971; Pyszczynskiand Greenberg,1981; Hastie, 1984; Weiner, 1985)
suggests that spontaneous attributionmakingis exacerbated
by, among otherthings, the degree to which events depart
fromobservers'generaland normativeexpectations. Surprising, extraordinary
events increase the need to search for
plausiblecausal determinants.Inthe present context, expectations may be stronglyrelatedto the magnitudeof the performance outcome and to the subsequent tendency to make
attributionsto the leaderand perhapsto alternativecauses, as
well. One reasonablehypothesis is that the more extreme
performancesdeviatedfromobservers' less extreme expectations of what performancechanges are typicalfor an organization. Ifthat is true,then perhapsit is the size of deviationfrom
expectations- not simplythe magnitudeof performance
that is responsiblefor the observed patternof leadership
attributions.Althoughthat reasoningis not inconsistentwith
the generalperspectivetaken inthis analysis, it does suggest
that some attempt should be made to take intoaccount
observers' expectations,alongwith performanceoutcomes, in
orderto examine theireffects on attributions.
Method
Seventy-two undergraduatebusiness majorsin sections of an
evening introductoryorganizationalbehaviorcourse participated in this experiment.Theirmean age was 25.0 years, and
on averagethey reportedhavingnearlyfive (4.87)years of
full-timeworkexperience.
93/ASO, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
The vignettes used inthe previousstudies were modified,in
orderto decrease the prominenceof the leaderrelativeto
other potentialcausal determinants,by weaving intothe text
the mentionof the leader,alongwith a numberof other causal
determinants.The finalvignettes readas follows:
The GeminiCorporation
is a largevolume manufacturerof household
appliances.Theyhave been in business fora numberof years and
have several plantslocatedthroughoutthe country.The appliance
industryis characterizedby an environmentwhose marketand economicfactors have been changingover the past few years. Sales is
one of the functionaldepartmentswithinthis corporationand is
headed by a Director,John Smith,who assumed this positionat the
beginningof the last fiscalyear.At aboutthe same time, a new group
of sales representativeswere hiredand reporteddirectlyto Mr.
Smith.At the end of the fiscalyear,gross sales hadshown a
slight/moderate/largeincrease/decrease(2%/10%/25%)over last
year's performance.
As a furtherprecaution,subjects were asked on the rating
form itself to considerallof the potentialcauses before they
made theirattributionalratingsof the impactof any single
causal determinant.Before makingthese ratings,however,
subjects were firstasked to ratethe performanceof the unit,
as inthe previousstudies. And, in orderto assess the extent to
which the level of performancedeviatedfromtheirown implicit and generalexpectations, after ratingthe unit'sperformance
they were also asked to rate(on a 7-pointscale) how surprising
they foundthe increaseor decrease in performance.
Results and Discussion
Initialanalyses. A series of analyses of variancewas conducted on the six performanceconditions,examiningattributions, expectations, and perceivedperformance.These data
are summarizedin Figure2. A one-way ANOVAof the perceived performanceattested to the efficacy of the manipulation, F(5,66)= 9.65, p<.001. A similaranalysisof the expecta-
6.0
5.5
Leader
5.0_/5.0~~~~~~~~~~~~~~~~~~
STRENGTHOF
4.5
ATTRIBUTIONS
4.0
0 Alternatives
-
3.5
3.0
Large
Decrease
Moderate
Decrease
Slight
Decrease
Slight
Increase
Moderate
Increase
Large
Increase
CONDITION
PERFORMANCE
Figure 2. Graphic representation of attributions to leader and alternatives
in each performance outcome condition: Study 6.
94/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
tion measure revealeda significanteffect of performance
conditionon expectations, F(5,66)= 3.14, p= .013. Attributions
were examined ina two-way analysisof variancewith one
between factor(performance)and one withinfactor(leader
versus alternativeattributions).Thisanalysisrevealeda significant maineffect for performanceoutcome, F(5,66)= 2.55,
p = .036, and for type of attribution, F(1,66) = 7.49, p = .008.
Moreover,both maineffects were qualifiedby a significant
interaction, F(5,66) = 2.42, p= .045.
Expectations and performance. First,in orderto establish the
relationshipbetween performanceand expectations, a polynomialregressionanalysiswas conducted inwhich ratedperformancewas used to predictthe reporteddeviationsfrom
expectations.Thisanalysisrevealedthat the coefficient on the
quadraticcomponentwas significant,B = 2.989(1.26);
t(69)= 2.38, p<.025. Moreover,the goodness-of-fittests indicated thatthe linearmodel provideda significantpoorfit,
F(2.89),p= .062, while the U-shapedmodel (20 polynomial)
providedthe best fit, F(1.68)= .19, p = .67. Thus,extremely
good and extremelypoorperformancewere judgedto be more
surprisingandthereforerepresentedlargerdeviationsfrom
subjects' generalexpectations.
Leader attributions.The next task was to incorporateexpectations intothe model specified by the originalhypothesis.
However,the polynomialregressiontechnique used in the
previousexperimentto test the predictedcurvilinearrelationshipbetween perceivedperformanceand leaderattributions did not lend itself to the inclusionof more thana single
predictorvariableand, therefore,could not be used to control
and test forthe additionaleffect of expectations. Consequently, a more traditionalmultiple-regressionprocedurewas employedas a reasonableapproximationof the model. First,the
relationshipbetween leaderattributionsand perceivedperformancewas estimated by includingthe performancevariable
and its squaredterm as predictorsof leaderattributions.Ifthe
hypothesized"quadratic"relationshipwere true, then a significant, but negative coefficientshould be obtainedon the performancevariable,in combinationwith a significantbut positive coefficient on the squaredterm. Withonlythese two
predictorterms, the overallequationwas significant,R2= .347;
F(2.69)= 18.30, p<.001. Moreimportantly,however, and as
expected, the coefficientassociated with the performance
term was significantand negative, B= -2.41(.438), p<.001;
and the coefficientassociated with the squaredterm was
significantand positive,B= .322(.545),p<.001. Ineffect, then,
these results replicatedthose of the previousstudy and, in
lightof the changes made inthe vignettes, providedus with
of the
more confidence inthe validityand generalizability
effect.
The overallequationremainedsignificantwhen the expectation ratingswere added intothe above model as a predictor,
R2= .344; F(3,68)= 13.41, p<.001. However,the coefficient
associated with the expectationratingwas not significant,
B = .1829(. 110),p = .103. The negative coefficient on the performance term remained significant, B = - 2.24(.444), p<.001,
as didthe positivecoefficienton the squaredterm,
B= .297(.559),p<.001. Thusone must concludethat although
extreme performancesdeviatedfromexpectations andwere
95/ASO, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
generallyviewed as more surprisingthan lower magnitude
performances,inthis case, such deviationsprobablydid not,
by themselves, have a largeindependenteffect on the
strengthof leaderattributions.Moreover,when controllingfor
expectations,the observed relationshipbetween the magnitude of outcomes andthe tendency to understandperformance in terms of leadershippersisted.
A plausibleargumentis thatwhen faced with explaininga large
magnitudeoutcome, individualsmake more attributionsto all
relevantsources. Accordingto that lineof reasoning,the level
of leadershipattributionsmay simplybe an artifactof a more
generaltrend.Consequently,we performedone finalset of
analyses that attempted to controlfor individuals'general
tendency to make attributionsto allsources. Inorderto do
that, the same set of variableswas used to predicta ratioin
which the strengthof leadershipattributionswas dividedby
the strengthof attributionsto allother sources. Thisratio
roughlyparallelsthe LQmeasure used inthe previousarchival
studies. Accordingto the hypothesis, extreme performances
should be associated with higherratios.The predictionmodel
was also significantfor this dependent variable,R2= .283,
p<.001. And,as with the previousdependent variable,the
coefficient associated with the expectationterm was not
significantB= .238(.458),p= .604. However,as predicted,the
coefficient on the performanceterm was significantand negative, B= - .811(.184),p<.001; and the coefficient associated
with the squaredterm was significantand positive,
B= .109(.232),p<.001. Thislast analysis,then, indicatedthat
extreme performancesdid indeed lead to a proportionalincrease in the preferenceto use the leaderas a causal explanation and provideda strong confirmationof ourexpectations.
GENERALDISCUSSION
The romanticizedconceptionof leadershipsuggests that leaders do or should have the abilityto controland influencethe
fates of the organizationsintheircharge.Thisassumptionof
controland the responsibilityit engenders is a double-edged
sword: not onlydoes it implygivingcreditfor positive outcomes, but it also entails layingblamefor negative ones
(Salancikand Meindl,1984). Ourexperimentalstudies revealed that pattern.However,the results of ourarchivalstudies suggest that one or the othertendency, forwhatever
reasons, may predominateinany given case. The negative and
positiveassociations in Studies 2 and 3, respectively,between
an interest in leadershipand the state of the nationaleconomy
are particularly
intriguing.The positiveassociation uncovered
in Study3 suggests that the popularpress that serves the
general business communitycontributesto the credit-giving
aspect of the romanticizedview. Of course, the popularpress
is in parta reflectionof the communityit serves, andfirms, by
theirown activities,can promptan interest in and association
to leadershipfactors.Thus,the findingthat leadershipis accentuated duringtimes of economic prosperityis, in retrospect, not so difficultto understand.Bythe same token, the
scholarlycommunitymay have less reason to favorgiving
creditover layingblame. Infact, the negative association
between an interest inthe topic of leadershipand economic
96/ASO, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
prosperitymay reflectthe problem-orientedresponse of young
scholarsto hardtimes.
Others(e.g., Pfeffer, 1977; Pfefferand Salancik,1978) have
suggested that the tendency to ascribe highlevels of control
and influenceto leadersarises fromprivateneeds to find
causes among humanactors. Accordingly,the exacerbationof
those needs would tend to foster the development of a
romanticconception inwhich leadershipwas indeed believed
to be highlysignificant.Infact, a subsequent analysisof our
experimentalstudies revealedthat attributionsto different
personalcausal agents (inthis case, leaderand subordinates)
tended to be positivelycorrelated(r= .20, .22, and .37 in
Studies 4, 5, and 6, respectively),as this general lineof
reasoningwould suggest. A romanticizedview of leadershipis
probablyalso an outgrowthof a generalfaithin humanorganizationsas potentiallyeffective and efficientvalue-producing
systems that fulfillthe variousinterests of theirparticipants
and perhaps,also, society at large.The potency and promiseof
humanorganizationsand allthe values they representcome to
be symbolizedinthe formalhierarchyof authorityand the
officialswho occupy the elite positions of power and status
(Milgram,1974). Giventhis, a faithinthe significanceof
leadershipmay be one manifestationof internalizedvalues
aboutthe validityof organizationsand therefore,by implication, the roles occupied by people who are chargedwith the
responsibilityto maintainand controlthem.
Because observers are proneto overestimate the amountof
when the event or outcontrolthat leaders exert, particularly
come in question is especiallysignificant,a subscriptionto a
romanticizedview could be dysfunctionalto the goals of an
"objective"or rationalassessment of importantbut causally
indeterminantevents. At the same time, however, it seems
possible that an excessive belief inthe potency of leadership
couldalso be functionalforthose who willoccupy positionsof
formalauthorityand status. Ifwe assume that on some
occasions leadershipcan, in reality,make a difference- but
that we cannot be sure when - then it may be importantfor
organizationsto have leaderswho operate, at some level, on
the assumptionthat they do make a differenceand thatthey
are in control.Withoutthe benefits of a workingassumption
that conveys a sense of efficacy and control,the initiationof
and persistence in potentiallyrelevantactivitywould be considerablymore difficult.The end resultmay be somewhat
depressed functioningand a sense of helplessness in situations inwhich controlis infact possible.
The present researchmay begin to provideus with some new
insightsaboutthe reason for changingleadershipor decisions
to extend an incumbent'stenure in response to perceived
variationsin an organization'sfortunes. Forexample, there is a
small, but somewhat paradoxicalliteraturethat attempts to
understandthe causes and consequences of managerialsuccession. Severaltheoreticalperspectives have been offered
(e.g., Grusky,1963; Gamsonand Scotch, 1964; Gordonand
Rosen, 1981), allof them based on more or less implicit
assumptions aboutthe attributionsof relevantand powerful
others to leadershipfactors in response to poororganizational
performance.Infact, the theories and the empiricalstudies
make a convincingcase that poorperformanceincreases the
97/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
probabilityand rateof successions (Grusky,1963; Helmichand
Brown,1972; Liebersonand O'Connor,1972; Helmich,1974,
1977; Osbornet al., 1981). Thereis less theoreticalagreement
aboutthe effect of succession on subsequent organizational
performance.Conventionalwisdom impliesthatthe effect on
performanceof changingleadersshould be positive,since
such events are ostensibly guided by the positive intentions
and expectations of those in a positionto inducethem. Some
(e.g., Grusky,1963) arguethat successions are disruptiveto so
many importantprocesses that subsequent performanceswill
decline. Stillothers (e.g., Gamsonand Scotch, 1964) emphasize the symbolicaspects of successions and considerthem
exercises in "ritualisticscapegoating"that involveprocesses
that are only incidentallyor tangentiallyrelevantfor subsequent performance.The availableempiricalevidence tends to
runcontraryto the conventionalwisdom, suggesting that
althoughpoorperformancesmay often precipitatesuccessions, such events have littleor negative effects on subsequent performance(e.g., Allen,Panian,and Lotz,1979; Brown,
1982). The paradoxis that, despite the absence of clearly
instrumentaleffects, successions are nevertheless a popular
response to poorperformance.At least a partof that paradox
can be understoodas reflectingan inclinationto construe
events and outcomes in terms of leadership.Pfeffer(1977)
arguedthat the limitedimpactthat manyleadershipsuccessions have on performanceoutcomes is due in largepartto the
lackof variabilityinthe pool of individualsfromwhich both the
incumbentand successor have been drawn.One interesting
and testable hypothesis precipitatedby the present analysisis
that interested partiesare very likelyto overperceivethe
degree of relevantvariationinthat pool, seeing more heterogeneity than reallyexists between the old andthe new leader
and among the potentialsuccessors. Giventhe romanticization
of leadership,it is less difficultto understandthe optimisticfaithinthe effectiveness of successions -the shiftingof
commitmentfromthe old to the new leaderand the maintenance of positiveexpectations for outcomes, even inthe face
of contraryevidence.
Needless to say, organizationshave always been influencedby
theirenvironments,yet it is only recently-within the last ten
years - that organizational
dependencies have been fully
appreciatedin ourtheoreticalperspectives (e.g., Aldrich,1979;
Hall,1982; Pfefferand Salancik,1978). InStudy 1, the average
companysales growthperformanceover the 11-yearperiod
and the correspondingfiguresfor relevantindustrieswere
stronglycorrelated,R(9df)= .80, p<.01. Thatis not surprising,
given the numberof industriessampled and the size of the
firms in oursample that were chosen to representthese
industries.However, it does providea roughindicationthat a
given firm'sfate, interms of performance,is closely tied to
externalfactorsaffectingwhole industries,as opposed to
being underthe direct,uniquecontrolof its top management.
As expected, however, we were able to findevidence that
there is nevertheless a tendency to linkleadershipnot only
with variationsin companyperformance,but also with the
performanceof entire industries,which are undoubtedly
affected by factorswell beyondthe controlof any single firmor
management.Otherresearchershave also found systematic
98/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
evidence indicatingthat traditionalviews have overestimated
the amountof variancein performanceoutcomes that is
logicallyand empiricallyattributableto leadership(e.g., Lieberson and O'Connor,1972; Salancikand Pfeffer, 1977). Such
evidence shifts attentionand the locus of controlaway from
top-levelleadersand the positionsthey occupy to other causal
entities and forces not directlytied to the qualitiesand activities of leadership.The implicationis that perhapsleadershipis
not as importantas we normallythink- at least not inthe
traditionalsense (Pfeffer,1978, 1981).Thatimplicationis
provocative,because it contradictsthe romanticizedconception of leadership,and some resistance to it is predictable.To
the extent that observers are psychologicallyinvested ina
romanticizedview of leadership,then, we mightexpect selective perceptions,confirmatorybiases, and other processes
(Ross, 1977) to be present that cause the observerto avoidor
resist informationand evidence that diminishesthe significance of leadershipto organizational
functioning.Considerthe
reactionof Burke(1979: 121) to Pfeffer(1978):
Pfefferindeedwent out on a limbby proclaimingthat leadersdo not
matterthat much. Manyvariablesotherthanthe leaderper se
accountfor organizational
outcomes. Moreover,"leadershipis the
outcome of an attributionprocess inwhidhobservers- in orderto
achieve a feeling of controlover theirenvironment- tend to attribute
outcomes to persons ratherthanto context, and the identificationof
individualswith leadershippositionsfacilitatesthis attributionprocess" (p.31). An interestingbelief, interpretation,hypothesis,or
whatever, but methinksPfefferbrokethe limband fell off. Inother
words, inan apparentattemptto be provocative,Pfefferseems to
have leanedtoo heavilytowardiconoclasm.
Itis possible to take the positionthat leadershipmay infact
contributeto a largeportionof the variancethat is controllable
and thus warrantsintense attention.However,the results of
ouranalysissuggest that the faithin leadershipis likelyto
exceed the realityof controlandwill be used to accountfor
variancethat is infact uncontrollable.Thisis a convenientstate
of affairsfor managements motivatedto do just that. Salancik
and Meindl(1984)presented evidence documentingthe
attempt by top managements to create an illusionof control
throughthe manipulationof causal reasoningaroundperformance issues. Such motivationsappearto be strongest among
managements whose firmshave displayedthe sorts of erratic
performancehistoriesthatwould implylittlerealcontrol.Our
analysissuggests thatwhat otherwise mightbe considered
patentlyobviousattempts to create the illusionsof control
where none exists is likelyto be complemented by a high
degree of receptivityamong observers.
When consideringthe "symbolicrole"of management(Pfeffer and Salancik,1978; Pfeffer, 1981), the greatersignificance
of leadershiplies not inthe directimpacton substantive
matters but inthe abilityto exert controlover the meanings
and interpretationsimportantconstituencies give to whatever
events and occurrencesare considered relevantfor the organization'sfunctioning(Pondy,1978; Daftand Weick, 1984).
The manipulationof languageand other organizationally
relevant symbols allows leadersto manage the politicaland social
processes that maintainorganizedactivityinthe face of potentiallydisruptiveforces (Pondyet al., 1982). One plausible
hypothesis is that the development of a romanticizedconcep99/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
tion of leadershipcauses participantsmore readilyto imbue
the symbolicgestures of leaderswith meaningand significance. Accordingly,the psychologicalreadinessto comprehend things interms of leadership,whatever dysfunctionsit
represents, may playan importantrole in determiningthe
ultimateeffectiveness of symbolismas a politicaltool, benefiting most those leaderswho are adept at its manipulation.
CONCLUSION
Therehas been in recentyears some question concerningthe
viabilityof leadership,both as a concept and as an areaof
inquiry.Indeed,there is ample reasonto modifyourtraditional
assumptions aboutthe instrumentalpotency of leadership
factors inthe largerscheme of things. Giventhe present
analysis,however, it appearsthat the obsession with the
concept will not easily be curtailed.Whilethere are some
obvious limitationsto the studies reportedhere, together they
providereasonablycoherent and compellingevidence forthe
premise that a romanticizedconceptionof leadershipis an
importantpartof the social realitythat is broughtto bear in our
informalanalysisof organizations- and perhapsin our more
though, a heroicvision of
formaltheories as well. Ironically,
what leadersand leadershipare allaboutvirtuallyguarantees
that a satisfyingunderstandingwill remainbeyondthe grasp of
since the thrustof scienourbest scientificefforts, particularly
tific inquiryis to do away with mysteries. The majoreffect is to
objectify,quantify,and in some cases trivializethe unique
importof leadership.Inthat sense, the productof such efforts
is contraryand antitheticalto the romanticizedconception.
And, if ouranalysisis correct,the continuinginfatuationwith
leadership,forwhatever truthsit yields aboutthe qualitiesand
behaviorof our leaders,can also be used to learnsomething
aboutthe motivationsof followers. Itmay be that the romance
and the mysterysurroundingleadershipconcepts are critical
for sustainingfollower-shipandthat they contributesignificantlyto the responsiveness of individualsto the needs and goals
of the collectiveorganization.
REFERENCES
Aldrich, Howard E.
1979 Organizations and Environments. Englewood Cliffs, NJ:
Prentice-Hall.
Allen, Michael P., Sharon K. Panian, and Roy E. Lotz
1979 "Managerial succession and
organizational performance: A
recalcitrant problem revisited."
Administrative Science
Quarterly, 24: 167-180.
Anderson, Craig A.
1983 "The causal structure of situations: The generation of plausible causal attributions as a
function of type of event situation." Journal of Experimental
Social Psychology, 19: 185203.
Bennis, Warren
1959 "Leadershiptheoryand administrative behavior: The
problem of authority." Administrative Science Quarterly,4:
259-301.
Binning, John F., and Robert G.
Lord
1980 "Boundary conditions for performance cue effects on group
process rating: Familiarityversus type of feedback." Organizational Behavior and Human
Performance, 26: 115-130.
Brown, M. Craig
1982 "Administrative succession
and organizational performance: The succession effect."
Administrative Science
Quarterly, 27: 1-16.
Burke, W. Warner
1979 Review of Morgan W. McCall
and Michael M. Lombardo
(eds.), Leadership: Where Else
Can We Go? Journal of Applied
Behavioral Science, 15: 121122.
Burns, John M.
1978 Leadership. New York: Harper
& Row.
Calder, Bobby J.
1977 "An attributiontheory of leadership." In BarryM. Staw and
Gerald R. Salancik (eds.), New
Direction in Organizational Behavior: 179-204. Chicago: St.
Clair.
100/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Romance of Leadership
Daft,RichardL.,and KarlE.Weick
Helmich, Donald L.
growthand
1984 "Towarda modelof organiza- 1974 "Organizational
systions as interpretation
succession patterns."
Academyof Management
tems." Academyof ManageJournal,17: 771-775.
ment Review,9: 284-295.
1977 "Executivesuccession inthe
DeNisi, Angelo S., and RobertD.
corporateorganization:A curPritchard
rentintegration."Academyof
1978 "Implicittheoriesof performManagementReview,2: 252ance as artifactsin surveyre266.
search:A replicationand
Be- Helmich, Donald L.,and W. B.
extension."Organizational
haviorand HumanPerformBrown
1972 "Successortypeandorganizaance, 21: 358-366.
tionalchange inthe corporate
Downey, H. Kirk,Thomas I.Chacko,
enterprise."Administrative
and James C. McElroy
Science Quarterly,17: 371of the 'causes' of
1979 "Attribution
381.
performance:A constructive,
replicationof House, RobertJ.
quasi-longitudinal
the Staw (1975)study."Orga- 1977 "A1976 theoryof charismatic
nizationalBehaviorand Human
leadership."InJ. G. Huntand
L.L.Larson(eds.), Leadership:
Performance,24: 287-299.
The CuttingEdge: 189-207.
Elig,TimothyW., and IreneHanson
Carbondale,IL:SouthernIlliFrieze
nois UniversityPress.
1979 "Measuringcausalattributions
forsuccess andfailure."Jour- Jones, EdwardE., and KeithE.
Davis
nalof PersonalityandSocial
1965 "Fromacts to dispositions:
Psychology,37: 621-634.
Theattributionprocess in perGamson, WilliamA., and Norman
son perception."InL.BerkoA. Scotch
witz (ed.),Advancesin Ex1964 "Scapegoatingin baseball."
perimentalSocialPsychology,
AmericanJournalof Sociology,
2: 219-266. New York:
70: 69-72.
AcademicPress.
Goode, WilliamJ.
Jones, EdwardE., Leslie Rock,Kel1978 The Celebrationof Heroes.
ly 0. Shaver, George Goethals, and
Berkeley,CA:Universityof
L.M. Ward
Press.
California
1968 "Patternof performanceand
An unexabilityattribution:
Gordon,Gil E., and Ned Rosen
factorsin leadership
pected primacyeffect." Jour1981 "Critical
nalof Personalityand Social
succession." Organizational
Psychology,10: 317-340.
Behaviorand HumanPerformance, 27: 227-254.
Jones, EdwardE., Steven Worchel,
George R. Goethals, and Judy
Greene, Charles N.
Grumet
1976 "Disenchantmentwith leadershipresearch:Some causes, 1971 "Priorexpectancyand behavior
extremityas determinantsof
recommendations,andalternaJournalof
attitudeattribution."
tive directions."InJ. G. Hunt
SocialPsycholoExperimental
and L.L.Larson(eds.), Leadgy, 7: 59-80.
ership:The CuttingEdge:5767. Carbondale,IL:Southern
Karmel,Barbara
IllinoisUniversityPress.
1978 "Leadership:A challengeto
traditionalresearchmethods
Grusky,Oscar
andassumptions."Academy
succession and
1963 "Managerial
of ManagementReview,3:
effectiveness."
organizational
475-482.
AmericanJournalof Sociology,
69: 21-31.
Kerr,Steven, and John M. Jermier
1978 "Substitutesfor leadership:
Hall,RichardH.
Theirmeaningand measureStructuresand
1982 Organizations:
Bement."Organizational
Process, 3d ed. Englewood
haviorand HumanPerformCliffs,NJ: Prentice-Hall.
ance, 22: 375-403.
Hastie, Reid
1984 "Causesandeffects of causal
Klapp,OrrinE.
1964 SymbolicLeaders.Chicago:
Journalof Perattributions."
Aldine.
sonalityandSocialPsychology,
46: 44-56.
Lieberson, Stanley, and James F.
O'Connor
1972 "Leadershipandorganizational
performance:A studyof large
corporations."American
SociologicalReview,37: 117130.
Lord,RobertG., and Jonathan E.
Smith
1983 "Theoreticalinformationprocessing andsituationalfactors
affectingattribution
theory
models of organizational
behavior."Academyof Management Review,8: 50-60.
March,James G.
1981 "Howwetalkand howwe act:
Administrative
theoryand
administrativelife."Unpublishedmanuscript,Stanford
University.
McArthur,Leslie Z.
1981 "Whatgrabsyou?The roleof
attentionin impressionformationandcausalattribution."In
E.T. Higgins,C. P. Herman,
and M. P. Zanna(eds.), Social
Cognition:The OntarioSymposium, 1: 201-246. Hillsdale,
NJ: Erlbaum.
McCall,MorganW., Jr., and
Michael M. Lombardo(eds.)
1978 Leadership:Where Else Can
We Go?Durham,NC:Duke
UniversityPress.
McElroy,James C.
1982 "Atypologyof attributionleadershipresearch."Academyof
ManagementReview,7: 413417.
McElroy,James C., and H. Kirk
Downey
1982 "Observationinorganizational
research:Panaceato the performance-attribution
effect?"
Academyof Management
Journal,25: 822-835.
Milgram,Stanley
1974 Obedienceto Authority.New
York:Harper&Row.
Miner,John B.
1975 "Theuncertainfutureof the
leadershipconcept: Anoverview." InJ. G. Huntand L.L.
Larson(eds.), LeadershipFrontiers: 197-208. Kent,OH:Kent
State UniversityPress.
Neilson, Winthrop,and Frances
Neilson
1973 What'sNews - Dow Jones.
Radnor,PA:Chilton.
1O1/ASQ,March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Osborn, RichardN., LawrenceR.
Jauch, Thomas N. Martin,and William F. Glueck
1981 "Theevent of CEOsuccession, performanceandenvironmentalconditions."Academy
of ManagementJournal,24:
183-191.
Pyszczynski,T. A., and Gerald
Greenberg
1981 "Roleof disconfirmedexpectancies inthe instigationof
attributional
processing."Journalof PersonalityandSocial
Psychology,40: 31-38.
Ross, Lee
1977 "Theintuitivepsychologistand
Pfeffer,Jeffrey
his shortcomings:Distortions
1977 "Theambiguityof leadership."
inthe attributionprocess." In
Academyof ManagementReLeonardBerkowitz(ed.),Adview, 2: 104-112.
Social
vances in Experimental
1978 "Theambiguityof leadership."
Psychology,10: 174-220. New
InMorganW. McCalland
York:AcademicPress.
MichaelM. Lombardo(eds.),
Leadership:WhereElse Can
Rush, MichaelC.,Jay C.Thomas,
We Go?: 13-34. Durham,NC: and RobertL.Lord
DukeUniversityPress.
1977 "Implicitleadershiptheory:A
1981 "Managementas symbolicacpotentialthreatto the internal
tion:The creationand maintevalidityof leaderbehavior
parananceof organizational
questionnaires."Organizationdigms." InL.L.Cummingsand
al Behaviorand HumanPerin
B. M. Staw (eds.), Research
formance,20: 92-110.
Behavior,3: 1Organizational
GeraldR., and James R.
Salancik,
52. Greenwich,CT:JAIPress.
Meindl
Pfeffer,Jeffrey, and GeraldR.
1984 "Corporateattributionsas
Salancik
strategicillusionsof manage1978 The ExternalControlof Orgament control."Administrative
nizations:A ResourceDepenScience Quarterly,29: 238dence Perspective.New York:
254.
Harper& Row.
Salancik,GeraldR., and Jeffrey
Phillips,James S., and RobertG.
Pfeffer
Lord
1977 "Constraintson administrative
1981 "Causalattributionsand
discretion:The limitedinfluperceptionsof leadership."
ence of mayorson city
Behaviorand
Organizational
budgets."UrbanAffairs
HumanPerformance,28: 143Quarterly,12: 475-498.
163.
Schlenker, BarryR.
Pondy, Louis R.
1982 "Anidentity-analytic
approach
1978 "Leadershipis a language
to the explanationof social
game." InM.W. McCall,Jr.,
conduct."InLeonardBerkoand M. M. Lombardo(eds.),
witz (ed.),Advancesin ExLeadership:Where Else Can
perimentalSocialPsychology.
We Go?:87-99. Durham,NC:
15: 194-248. New York:
DukeUniversityPress.
AcademicPress.
Pondy, Louis R., Peter Frost,Gareth Staw, BarryM.
Morgan,and Thomas Dandridge
of the 'causes' of
1975 "Attribution
(eds.)
performance:A general
1982 Organizational
Symbolism.
of
alternativeinterpretation
Greenwich,CT:JAIPress.
cross-sectionalresearchon
organizations."Organizational
Behaviorand HumanPerformance, 13: 414-432:
Staw, BarryM., and Jerry Ross
1980 "Commitmentinan experimentingsociety: A study
of the attributionof leadership
scenarfromadministrative
ios." Journalof Applied
Psychology,65: 249-260.
Stogdill, RalphM.
1974 Handbookof Leadership.New
York:FreePress.
Taylor,Shelly E., and Susan G.
Fiske
1978 "Salience,attention,and
Topof the head
attribution:
phenomena."InL.Berkowitz
(ed.),Advancesin ExperimentalSocialPsychology,
11: 250-288. New York:
AcademicPress.
Thompson, James D., and A.
Tuden
1959 "Strategies,structures,and
deprocesses of organizational
cision."InJ. D.Thompson,P.
B. Hammond,R.W. Hawkes,
B. H. Lunker,andA. Tuden
(eds.), ComparativeStudies in
195-216. PittsAdministration:
burgh:PittsburghUniversity
Press.
Weber, Max
1946 "Thesociologyof charismatic
authority."InH. H.Gerthand
C.W. Mills(eds.), FromMax
Weber: Essays in Sociology:
245-252. New York:Oxford
UniversityPress.
Weick, KarlE.
1979 The SocialPsychologyof Organizing,rev.ed. Reading,MA:
Addison-Wesley.
Weiner, BernardE.
1985 "Spontaneouscausalthinking."PsychologicalBulletin,
97: 74-84.
102/ASQ, March 1985
This content downloaded from 193.60.79.168 on Wed, 28 Oct 2015 13:43:46 UTC
All use subject to JSTOR Terms and Conditions
Download
Related flashcards
Cultural studies

21 Cards

Swords

20 Cards

Legal documents

27 Cards

Create flashcards