Paradigm Development and Communication in Scientific Settings: A Contingency Analysis

Author(s): Joseph L. C. Cheng

Source: The Academy of Management Journal, Vol. 27, No. 4 (Dec., 1984), pp. 870-877
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/255884 .
Accessed: 15/06/2014 09:25

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 support@jstor.org.

Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy
of Management Journal.

http://www.jstor.org

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
?Academy of Management Journal
1984, Vol. 27, No. 4, 870-877.

Research Notes
PARADIGM DEVELOPMENT AND COMMUNICATION
IN SCIENTIFIC SETTINGS:
A CONTINGENCY ANALYSIS'

JOSEPH L. C CHENG
Virginia Polytechnic Institute and State University

More than ten years ago, Lodahl and Gordon (1972) published an investi-
gation in whichthey appliedthe conceptof paradigm(Kuhn,1962)to the
studyof taskuncertaintyin scientificsettings.Theirdatashowedthat scien-
tists in fields with developedparadigms(physicsand chemistry)exhibited
greaterconsensuson theory,methodology,and trainingthan did those in
fields with undevelopedparadigms(sociology and political science).This
consensusprovideda morecertainor predictableworkenvironment,which
in turnfacilitatedteachingand researchactivities.Lodahland Gordoncon-
cludedthat these findingsprovideencouragingsupportfor conceptualiz-
ing paradigmdevelopmentin terms of task uncertainty,and they recom-
mended furtherresearchinto its organizationalimplicationsin scientific
settings.
In an attemptto explorefurtherthe usefulnessof the paradigmconcept,
researchers haveinvestigatedthe effectsof paradigmdevelopmenton various
organizationalprocesses,includinginfluencepatternsin universities(Beyer
& Lodahl, 1976), grant allocations(Pfeffer,Salancik,& Leblebici,1976),
editorialpracticesin journal organizations(Beyer,1978), academicturn-
over (Salancik,Staw,& Pondy, 1980),and universitybudgeting(Pfeffer&
Moore, 1980),amongothers.Withoutexception,their findingsareconsis-
tent with the originalconceptualizationof Lodahland Gordon(1972)and,
together,providea strongempiricalbase for the use of paradigmdevelop-
ment as a measureof task uncertaintyin scientific settings.
Basedon this foundation,Chengand McKinley(1983)recentlyexamined
the implicationsof paradigmdevelopmentfor bureaucratic controland per-
formancein scientificsettings.Theypresenteddata suggestingthat the ef-
fect of bureaucraticcontrolon scientificperformanceis contingenton the
paradigmdevelopmentof the scientific field, In fields with highly devel-
oped paradigms,bureaucratic controlwaspositivelyrelatedto performance
As paradigmdevelopmentdecreased,however,this relationshipdeclined
significantlyand becamenegativein fields with less developedparadigms.
at the 1983NationalAcademyof Management
'An earlierversionof thispaperwaspresented Meeting,
Dallas.The authorexpresseshis appreciationto Bill McKinleyand two anonymousreviewersfor their
helpful comments.

870

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
1984 Cheng 871

Thesefindingsare potentiallysignificantin that they point to the need for

a contingencyapproachto the study of scientificperformance.They also
raise questionsabout the generalityof some earlierfindingson scientific
performancethat did not takeaccountof differencesin paradigmdevelop-
ment (Pelz & Andrews,1966).
The purposeof this studyis to investigatefurtherthe paradigmconcept
as a contingencyfactorin relationto scientificperformance.The focus of
the studyis on the contingencyeffect of communicationon scientificper-
formance,and the organizationsstudiedare researchunits in universities.
The investigationincludesa theoreticalanalysisand an empiricaltest of
the propositionderivedfromthat analysis.The theoreticalperspectivethat
guidesthis studyis the information-processingmodel of organization(Gal-
braith, 1972).
Analysis
andTheoretical
Background

Severalpremisesprovidethe theoreticalunderpinningsfor this study.An

initialpremiseis thatthe greaterthe taskuncertainty,the greaterthe amount
of informationan organizationhas to processin orderto accomplishits
task (Galbraith,1972).If the task is well understood,much of the activity
can be preplanned,and the informationprocessingrequirementsfor the
organizationare relativelysmall. If the task is not well understood,new
informationmustbe processedduringtask execution,and the information
processingrequirementsfor the organizationbecomesubstantial(Thshman
& Nadler,1978).In scientificsettings,the importanceof informationpro-
cessingvarieswiththe field'slevelof paradigmdevelopment.The lowerthe
paradigmdevelopment,the greaterthe taskuncertainty,andthus the greater
the importanceof informationprocessing.
A second premiseis that in scientificsettings,oral communicationis a
particularlyefficientmediumthroughwhichinformationis gathered,trans-
ferred,and processed.Unlikewrittencommunication,oralcommunication
permitsrapid feedback,decoding,and synthesisof complexinformation
(Porter& Roberts,1976;Tushman,1978). These qualitiesare essentialto
scientificcommunicationbecausemost researchideas are ambiguousand
difficultto articulate,particularlyat the formulationstage.Thus, oralcom-
municationrepresentsa primarymediumfor informationexchangein scien-
tific settings.It will be used in this study as a measureof a researchunit's
informationprocessingactivity.
A thirdpremiseis that oral communicationis effectiveonly to the extent
that the actorssharea commonlanguagescheme(Thshman& Katz, 1980).
Accordingto these authors,lack of linguisticcommonalitybetweenactors
can lead to errorsin the interpretationof messagesanalogous to ''noise"
in physicalsystemsthat causeerrorin messagereception.In scientificset-
tings, researchersspend almost as much time in task-relatedcommunica-
tion with colleaguesoutsidetheir own unit (but withinthe same organiza-
tion) as with colleagueswithintheirown unit (Katz& Tushman,1979;Pelz

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
872 Academy of Management Journal December

& Andrews,1966). To the extent that differentorganizationalunits have

differentlanguageschemes(Porter& Roberts,1976),interunitcommunica-
tion could resultin misinformationbecauseof an incorrector incomplete
understandingof the message'sinformationcontent. This misinformation
stemmingfrom interunitcommunicationmay havenegativeconsequences
for researchunits as far as their scientific performanceis concerned.
Tikentogether,thesethreepremisesarguefor an interactioneffectof par-
adigmdevelopmentandcommunicationon researchunit performance.The
lowerthe paradigmdevelopment,the higherthe information-processing re-
quirements,and thus the greaterthe effect of communicationon research
unit performance.Becauseof the limitationsdiscussedin the thirdpremise,
thiseffectwill be positivefor intraunitcommumncationbut negativefor inter-
unit communication.Theseideas can be summarizedinto the followinghy-
pothesis:
Hypothesis:Thelowertheparadigmdevelopment,the morepositive
the relationshipbetween intraunitcommunicationand researchunit
performance,but the morenegativethe relationshipbetweeninterunit
communicationand researchunit performance.

Method

Sample.The studysampleconsistedof 350academicresearchunitslocated

in six countries:Austria,Belgium,Finland,Hungary,Poland,and Sweden.
Included were 88 academic institutions and 3 scientific fields-biology,
chemistry,and physics.The use of these three fields, which vary in para-
digm developmentbut are all "hardscience"disciplines,permitsa more
stringenttest of the hypothesis.A typicalresearchun}itconsistedof a unit
headand a staff of about nine people, includingthe scientists,technicians,
and clericalpersonnel.
Data Collection.The datawerecollectedusinga combinationof personal
interviews,organizationalrecords,and on-siteadministrationof standard-
izedquestionnalres. A detaileddescriptionof the data-collectioninstruments
and samplingprocedureis providedby Andrews(1979).
Measures.Ieo single-itemscales wereused for measuringintraunitand
interunitcommunication.Intraunitcommunicationwas measuredby ask-
ing scientistsin the researchunit to indicate(on a 5-point scale) how fre-
quently they met as a group to discuss scientific/technicalmatters.They
also wereaskedto indicate(on a 6-pointscale)how frequentlytheydiscussed
scientific/technicalmatterswith colleaguesin other researchunits within
the same organization.The latter informationwas used as a measureof
interunitcommumcation.Forboth measures,the homogeneityof unit mem-
bers'responseswas testedfor the appropriatenessof poolingusinga 2-step
proceduredevelopedby Tushman(1978). Resultsfrom the first step indi-
cated that pooling was appropriatefor each of the two communication vari-
ables(p <.05). The secondstepeliminated(p < .01) threeand five cases for
the intraunitand interunitcommunicationvariables,respectively.

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
1984 Cheng 873

Researchunit performancewasassessedin termsof threetypes of scien-

tific outputthe unit had producedduringthe past threeyears:(1) number
of books published,(2) numberof journal articlespublishedoutside the
country,and (3) numberof journal articlespublishedwithin the country.
Thesedata werecollectedfromprojectrecordsprovidedby the unit heads.
Six units did not have completedata on all three output items and were
deletedfromthe study sample.Becausethe originaloutput scoresshowed
skeweddistributions,they all weretransformedto lognormalscoresusing
the Pelzand Andrews(1966)procedure.To adjustfor the unevenscientific
importanceof the threetypes of output, a weightof 3 was assignedto the
numberof books published,2 to the numberof journalarticlespublished
outsidethe country,and 1 to the numberof articlespublishedwithin the
country(Cheng& McKinley,1983).A compositeindexwas constructedby
summingthe weighted,transformedscores across items.
Paradigmdevelopmentwasassessedin termsof the unit'sscientificfield.
Units performingresearchin physicsand chemistrywereassignedto a de-
velopedparadigmgroup(N= 202), and units performingresearchin biol-
ogy wereassignedto a less developedparadigmgroup(N= 134).This clas-
sificationwas based on (1) Lodahland Gordon's(1972)rankingsof para-
digm developmentin sevenscientificfields and (2) the combinedrankings
of 20 academicdisciplinesreportedby Salanciket al. (1980) and Pfeffer
and Moore(1980).In the LodahlandGordonstudy,paradigmdevelopment
wasmeasuredon the basis of scientists'evaluationsof how developedthey
perceivevariousfieldsto be Theirdatashowedthat physicshad an average
rankscoreof 1.38, chemistryhad a scoreof 1.75, and biology 2.93. In the
Salanciket al. (1980)and Pfeffer and Moore(1980) studies,paradigmde-
velopmentwas measuredin termsof (1) coursesequencingin departments
and (2) the lengthof dissertationabstracts.TheSalanciket al. studyranked
phsyics2, chemistry5, andgeneralbiology 15. ThePfefferand Moorestudy
rankedboth physicsand chemistry5.5 and physiology(a specializeddisci-
pline in biology) 4. Becausephysicsand chemistryhad similarrankingsin
these studies,they werecombinedinto one paradigmgrouphere.Averag-
ing the rankingsfrom the Salanciket al. and Pfeffer and Moore studies,
physicsand chemistryas a group were ranked4.5, and biology 9.5.
Sevenadditionalvariableswereincludedas controls.The first five were
dummyvariablesfor the six countriesstudied.The remainingtwo wereunit
size-measured in termsof the numberof staff membersin the unit (in-
cludingscientists,technicians,and clericalpersonnel)and the numberof
researchunitswithinthe organizationthatwereactivein the sameor similar
field(s). Data for this variablewere providedby the unit heads.
Results
Threeseparatemoderatedregressionswereperformedto test the hypoth-
esis. The firsttestedfor a positiveinteractioneffect of intraunitcommuni-
cation and paradigmdevelopment(coded 1 for low and 0 for high) on

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
874 Academy of Management Journal December

Table 1
Moderated Regression Resultsa
Research Unit Productivity
Predictor Analysis 1 Analysis 2 Analysis 3

Hungaryb -.10 -.07 -.10

Austria -.13 -.11 -.12
Belgium -.01 -.00 .01
Sweden -.20t -.20t -.19t
Poland -.25tt -.24tt -.25tt
Unit size .29tt .29tt .30tt
Number of units inside organization .12t .12t .1it
Intraunit communication .00 -.00
Interunit communication .02 .02
Paradigm developmentC -.21 .24 -.01
Intraunit communication x paradigm development .28** .32***
Interunit communicationxparadigm development -.22* -.27**
R2= 18% 18% 20%
aCoefficients are the standardized regression coefficients ($3's).
bFinland is the base comparative country.
CReversescale (1= low, 0=high).
*p <.10; **p <.05; ***p <.025 (one-tailed for hypothesized relationships)
tp<.05; ttp<.01 (two-tailed for control variables)

researchunit performance.The secondtestedfor a negativeinteractionef-

fect of interunitcommunicationand paradigmdevelopmenton researchunit
performance.Finally,a more stringenttest was performedto assess both
interactioneffectssimultaneously usinga singleregressionequation.Because
the directionof the interactioneffectshad alreadybeen specified,one-tailed
tests wereemployed.
As hypothesized,the first two sets of analyses(see columns 1 and 2 in
Table 1) show a positive interactioneffect for intraunitcommunication
(/3=.28; p < .05), and a negativeinteractioneffect for interunitcommunica-
tion (fi = -.22; p <.10). These effects remainedsignificanteven when they
were assessed simultaneouslyusing a single regression equation (see
Figure 1
Mean Performance of Research Units by
Level of Unit Communication for High and Low Paradigm Groups
15- 15-_
14 - 14-
Mean Unit 13 , Mean Unit 13
Performance Performance
12 - 12 -

11- 11-
I 1 1- 1
Low High Low High
Intraunit Communication Interunit Communication
_______= High paradigm group
?------=Low paradigm group

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
1984 Cheng 875

column 3). The regressioncoefficientsfor the two interactionterms were

3=.32 (p< .025)for intraunitcommunication and : =-.27 (p < .05) for inter-
unit communication.These relationshipswere confirmed by the graphs
shown in Figure 1. For both communicationvariables,the slopes of the
graphswerein the predicteddirection(positivefor intraunitand negative
for interunitcommunication).Further,the slopes weresteeperin the low
paradigmgroupthan in the high paradigmgroup.Takentogether,these re-
sults providefull supportfor the hypothesizedrelationships,

Discussion

The purposeof this study has been to explorefurtherthe explanatory

powerof the paradigmconceptas a contingencyfactorin relationto scien-
tific performance.Data from three "hardscience"disciplinessupportthe
hypothesizedinteractiverelationshipamong paradigmdevelopment,com-
munication,andresearchunitperformance. Thelowerthe paradigmdevelop-
ment, the morepositivethe relationship between intraunitcommunication
and performance,but the morenegativethe relationshipbetweeninterunit
communicationand performance.Thesefindingsthus reinforcethe impor-
tanceof includingparadigmdevelopmentas a contingencyvariablein study-
ing scientificperformance(Cheng& McKinley,1983). They also provide
new and additionalsupportfor conceptualizingparadigmdevelopmentin
terms of task uncertaintyas originallyproposedby Lodahl and Gordon
(1972).
The presentfindings are similarto those of Katz and Thshman(1979)
but contradictthose of Pelz and Andrews(1966). Based on data from a
subsampleof 13 researchprojectsin a largeindustriallaboratory,Katzand
TIhshman reporteda significantpositive relationshipbetweenintraproject
communicationand researchproject performance,and a nonsignificant
negativerelationshipbetweenextraproject(but withinthe samelaboratory)
communicationand researchprojectperformance.PelzandAndrews,based
on data from 1,131 scientistsin 11 academic,industrial,and governmental
laboratories,foundsignificantpositiverelationshipsbetweenboth intraunit
and interunitcommunicationand scientists'technicalperformances.One
possibleexplanationfor this discrepancymight be that Pelz and Andrews'
sampleconsistedprimarilyof "applied"scientistsdoing developmentalre-
search.To the extentthat developmentalresearchis more standardizedin
its terminologythan basic research,interunitcommunicationcould be as
effectiveas intraunitcommunicationas faras informationprocessingis con-
cerned.Takentogether,these findings suggest a two-dimensionalcontin-
gencyframeworkfor the studyof scientificcommunication:(1) levelof par-
adigmdevelopmentof a scientificfield, and (2) the basic-appliedorienta-
tion of the research.Futurestudiesmight explorethe predictiveutility of
this two-dimensionalframework,thus leadingto a more completeunder-
standingof the role of communicationin scientific settings.

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
876 Academy of Management Journal December

The presentfindingsalso haveimplicationsfor the currenteffort on in-

tegratingorganizationalresearchand practice(Beyer,1982). Thomasand
Tyrmon (1982) recentlyhave recommendeda strategythat calls for active
involvementof practitionersin feedbackand reviewprocesses.Although
this strategyis intuitivelyappealing,it might have unanticipatednegative
consequencesbecauseof differentframesof referenceor languageschemes
betweenresearchersand practitioners.A more fruitfulapproachmight be
first to encourageeach group to developa frameworkas to what its con-
stituentsconsideras utilizableresearch,followedby two-wayinteractions
betweenrepresentatives fromeachgroupwho arewell versedin both orga-
nizationresearchand practice.This strategyis similarto the two-steppro-
cess suggestedby Tushmanand Katz (1980) for effectivecommunication
betweenengineersworkingon "locally-oriented"developmentalresearch
and externalsourcesof information.
Finally,froma morepragmaticstandpoint,this studyreinforcesthe im-
portanceof managingcommunicationactivitiesin researchorganizations.
The presentdata show that the more productiveunits had more intraunit
but less interunitcommunicationthan the less productiveunits. In terms
of managerialactions, these resultssuggestthat administratorsmay have
to be selectivein theirattemptsto promotecommunicationactivitiesin re-
searchorganizations.Becauseof the problemsinvolvedin interunitcom-
munication,as discussedearlier,frequentcontactsbetweenscientistsfrom
differentresearchunitsmayneedto be de-emphasized. Intraunitcommunica-
tion, on the other hand, can be encouragedand facilitated.This can be
achievedby (1) de-emphasizingformalityand status differencesin the re-
searchunit, thus reducingthe psychologicalbarriersto communication,
(2) promotingsupportiveand trustingrelationshipsamongscientistsin the
unit, (3) arrangingregularmeetingsfor progressreportsand idea exchange,
and (4) abolishing physical barriersto communicationand placing re-
searchersin close proximityto one another.
References

Andrews, F. M. (Ed.). Scientific productivity: The effectiveness of researchgroups in six countries. Cam-
bridge,England:CambridgeUniversityPress, 1979.
Beyer,J. M. Editorialpoliciesandpracticesamongleadingjournalsin fourscientificfields.Sociological
Quarterly, 1978, 19, 68-88.
Beyer, J. M. Introduction. Administrative Science Quarterly, 1982, 27, 588-590.
studyof patternsof influencein UnitedStatesandEnglish
Beyer,J. M., & Lodahl,T.M. A comparative
universities. Administrative Science Quarterly, 1976, 27, 104-129.
Cheng,J. L. C., & McKinley,W. Towardan integrationof organizationresearchand practice:A con-
tingencystudyof bureaucratic Science
controland performancein scientificsettings.Administrative
Quarterly,1983, 28, 85-100.
Galbraith,J. R. Organizationdesign:An informationprocessingview.In J. W. Lorsch& P. Lawrence
(Eds.), Organization planning. Homewood, Ill.: Irwin-Dorsey, 1972, 49-74.
Katz,R., & Tbshman,M. Communicationpatterns,projectperformance,and taskcharacteristics: An
empirical evaluation and integration in an R&D setting. Organizational Behavior and Human Per-
formance, 1979, 23, 139-162.

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions
1984 Cheng 877

Kuhn,T. S. Thestructureof scientificrevolutions.Chicago:Universityof ChicagoPress, 1962.

Lodahl,J. B., & Gordon,G. The structureof scientificfieldsandthe functioningof universitygraduate
departments.AmericanSociologicalReview, 1972, 37, 57-62.
Pelz, D. C., & Andrews,F. M. Scientistsin organizations.New York:Wiley,1966.
Pfeffer,J.,& Moore,W. L. Powerin universitybudgeting:A replicationand extension.Administrative
ScienceQuarterly,1980, 25, 387-406.
Pfeffer,J., Salancik,G. R., & Leblebici,H. The effect of uncertaintyon the use of social influence
in organizationaldecisionmaking.AdministrativeScienceQuarterly,1976, 21, 227-245.
Porter,L., & Roberts,K. Communicationsin organizations.In M. Dunnette(Ed.),Handbookof in-
dustrialand organizational psychology. Chicago:RandMcNally,1976, 1553-1589.
Salancik, R., Staw, M., Pondy,L. R. Administrative
G. B. & turnoveras a responseto unmanagedor-
ganizationalinterdependence. Academyof ManagementJournal, 1980, 23, 422-437.
Thomas,K. W., & Tyrmon, W. G., Jr.Necessarypropertiesof relevantresearch:Lessonsfrom recent
criticismsof the organizationalsciences.Academyof ManagementReview, 1982, 7, 345-352.
Thshman,M. Technicalcommunicationin R&Dlaboratories:The impactof projectworkcharacteris-
tics. Academyof ManagementJournal, 1978, 21, 624-645.
Thshman,M., & Katz,R. Externalcommunicationand projectperformance: An investigationinto the
role of gatekeepers.ManagementScience, 1980, 26, 1071-1085.
Thshman,M., &Nadler,D. A. Informationprocessingas an integrating conceptin organizational design.
Academyof ManagementReview, 1978, 3, 613-624.

Be-
JosephL. C ChengisAssociateProfessorof Organizational
haviorand a memberof the TechnologicalInnovationStudy
Groupin theDepartmentof Management,VirginiaPolytechnic
Instituteand State University.

?Academy of Management Journal

1984, Vol. 27, No. 4, 877-885.

THE INFLUENCE OF ORGANIZATIONAL STRUCTURE

ON INTRINSIC VERSUS EXTRINSIC MOTIVATION

J. DANIEL SHERMAN
Universityof Alabamain Huntsville
HOWARDL. SMITH
MedicalCollege of Virginia

Researchon intrinsicmotivationhas demonstratedthat extrinsicrewards

can havea negativeeffect on intrinsicmotivationundercertainconditions
(Calder& Staw,1975;Daniel & Esser, 1980;deCharms,1968;Deci, 1971,
1972, 1975a,1975b;Greene& Lepper,1974;Pinder,1976;Ross, 1975;Staw,
Calder,Hess, & Sandelands,1980). Accordingto Deci's cognitiveevalua-
tion theory,thereare two processesby which extrinsicrewardscan affect
intrinsicmotivation.The first occursif there is a changein the perceived
locus of causality.Hence,whena personis intrinsicallymotivated,the locus
of causalityis withinthemselves(deCharms,1968). However,when exter-
nal rewardsare made contingenton behaviorand the individualbeginsto
perceivethat he or she is engagingin the activity for these rewards,then

This content downloaded from 91.229.229.210 on Sun, 15 Jun 2014 09:25:59 AM

All use subject to JSTOR Terms and Conditions