Sie sind auf Seite 1von 4

Network Analysis: A Reappraisal

Author(s): Jeremy Boissevain


Source: Current Anthropology, Vol. 20, No. 2 (Jun., 1979), pp. 392-394
Published by: The University of Chicago Press on behalf of Wenner-Gren Foundation for Anthropological
Research
Stable URL: http://www.jstor.org/stable/2741937 .
Accessed: 06/11/2013 13:48

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.

The University of Chicago Press and Wenner-Gren Foundation for Anthropological Research are collaborating
with JSTOR to digitize, preserve and extend access to Current Anthropology.

http://www.jstor.org

This content downloaded from 128.227.214.253 on Wed, 6 Nov 2013 13:48:55 PM


All use subject to JSTOR Terms and Conditions
NetworkAnalysis: A Reappraisal' social field(Wellman1976,Shulman1976). In short,network
analysishas promisedto providea releasefromsome of the
constraintsof structural-functional analysis. It has conse-
byJEREMYBOISSEVAIN quently appealed to differentsocial scientistsfor varying
Department ofEuropean-Mediterrantean Studies,Universityof reasons.
Amsterdam, Sarphatistraat106A, Amsterdam,The Nether- Networkanalysis,whilenot a theory,has theoreticalimpli-
lands.3 iv 78 cations.It is an analyticalinstrument whichviews circlesof
relativesand friends,coalitions,groupsand businesshouses,
Since the networkrevivalin anthropologyin the late 1960s
industrialcomplexes,and even nation-statesas scatteringsof
(Barnes 1968,1969; Boissevain1968; Mitchell1969), therehas
pointsconnectedby lines that formnetworks.The pointsare
been ever increasinginterestin the field.There have been at
of coursethe unitsof analysis,the lines social relations.Net-
least a dozen conferencesand symposia,a floodof articlesand
workanalysisasks questionsabout who is linkedto whom,the
discussionpapersby anthropologists, and political
sociologists, contentof the linkages,the patternthey form,the relation
scientists,a computerizedbibliographywithalmost 1,000 en-
betweenthe patternand behaviour,and the relationbetween
tries(Freeman1975), the collectionand consolidationof com-
thepatternand othersocietalfactors.This has theoreticalim-
puterprogrammes, and, to crownthisinterdisciplinary
activity, plicationsin that it formspart of a paradigmaticshiftaway
theestablishment of theInternationalNetworkforSocial Net-
fromstructural-functionalism. The failureto recognizethese
work Analysisand the journal Social Networks.How is the
theoretical implicationsand to providea consistenttheoretical
enthusiasmfornetworkanalysisto be explained?Barnes(1954)
framework withinwhichnetworkanalysiscan be used has re-
and Bott (1957) planted the conceptsin the mid-1950s,but
sultedin a sterileoverelaboration of classificationand defini-
theyonlysproutedinto substantialgrowth15 yearslaterand
tion,in short,a methodological involution(cf.Kapferer1973:
now threatento becomean impenetrable jungle.
167). By linkingnetworkanalysisto theoreticalassumptions,
bothKapfererand Boissevainhave attemptedto movebeyond
NETWORK AND THEORY the butterfly-collector'spreoccupationwith classification and
technique,as represented, at leastin anthropology, by Barnes's
The enthusiasmfornetworkanalysisis related-toand part of recentwork(1968, 1969, 1972), into the realmof ideas (Kap-
the theoreticalshiftin thesocial sciencesaway fromthestruc- ferer1969, 1972, 1973; Boissevain 1974; forfurther discussion
tural-functional analytical frameworkwhich dominatedan- of the relationof networkanalysisto theory,see Whittenand
thropology, sociology,and politicalsciencein Britainand the Wolfe 1974,Mitchell 1974). The most fruitful theoreticalas-
United States forthe past 30 years.This is obviouslynot the sumptionsat presentappear to be derivedfromexchangeand
place to explorethereasonsand dimensions ofthismethodologi- (trans)actiontheory.Even withoutexplicitconsiderationof
cal and paradigmaticshift(cf. Boissevain 1974, 1975,among basic theoreticalassumptions,however,networkanalysisis a
others).I can only discuss the appeal that networkanalysis powerfultoolforsocial scientistsseekingto further theirunder-
has had foranthropology, althoughI suspectthatsimilarcon- standingof social behaviourand processes.
siderationshave also influencedsociologistsand political
scientists.
Networkanalysisopened a door to permitthe entryof in- WHAT NETWORK ANALYSIS CAN AND CANNOT Do
teractingpeopleengagedin actionsthatcouldalterand manip-
ulate the institutionsin whichtheyparticipated.This intro- As an adjunct or complementto other researchtechniques,
networkanalysishas at least ten importantvirtues:
duceda new dimensionintotheself-regulating structural-func-
tionaledificeofformalgroups,systems,and moralorderwhich 1. Networkanalysisfocusessystematicattentionon inter-
was seen as impinging linkagesbetweenunits of analysis. These interlinkages may
upon people,socializingthem,moulding
theircharacter,and determining be outwardlinksbetweenindividualsand betweengroups;they
theirbehaviour.In anthro-
pology the workof Firth (1951), Leach (1954), and some of mayalso be inwardlinks,settingout theinterrelations between
Gluckman'sstudents(Turner1957,Van Velsen 1964) had led membersof a groupor otherunitof analysis.
to a growingconcernwithpeople and theirrelationsto the 2. By focusingsystematically on therelationsbetweenunits
institutions of analysis,networkanalysishighlights theirinterdependency.
whichweresupposedto dominatethem.Network
In fact,this interdependency and its consequencesforsocial
analysisprovidedan analyticalframework fordata at a lower
levelofabstractionthantheinstitutional action are assumptionsunderlying the networkapproach.The
complex.It was more
configurations of interlinked,and thereforeinterdependent,
down-to-earth. Moreover,it also providedapparently"hard"
data whichcould be plottedand even computerized. personsand groupsare thus taken into account in tryingto
This last
has particularlyappealed to sociologists,who, more than an- predictbehaviour.By systematically tracingall interlinkages
betweenunits of analysis,one eliminatesprior assumptions
thropologists, seem to revelin data thatcan be quantifiedand
fittedintoelaborateformulae,therebyseemingto supporttheir and therefore biases in favourof particulartypesof relations.
claim to being considereda hard science. Networkanalysis Kinsmen, neighbours,and friendsare not singled out and
has also appealed to thosewhohave soughtto plot and analyze viewedin isolationfromotherrelations.
the manipulationof powerbrokers,leaders,and coalitionsas 3. The focuson interlinkage and interdependency provides
a framework withinwhichit is verydifficult to separatemicro-
theyseek to further theirinterestsand in so doingbringabout
or block developmentof the groups,institutions, frommacro-analytical levelsand partfromwhole.Amongother
and society
of whichtheyformpart. Finally,and morerecently,network things,thenetwork approachdevelopstheviewofa socialfieldor
ofa societyas a network ofnetworks. Whilethisis metaphorical-
analysishas providedsocial scientistsworkingin citieswitha
toolwhichenablesthemto deal withtherelationbetweenface- for a city or nation-stateis obviously more than simplya
to-faceinteractionand institutionsin an extremelycomplex networkof networks-networkanalysis does force upon the
social investigator pathwaysthat lead away frommicro-units
1 This paper was presentedto the conference"Mathematical Ap- of analysis.These last are therefore placed in a widerfieldof
proachesin Social NetworkAnalysis,"held at the Werner-Reimers- social relations.It is only throughfocusingon such outward
Stiftung,Bad Homburg,Federal RepublicofGermany,March 17-19, links that Wolf (1956), for example,developedthe concepts
1977 (Hummeland Ziegler1977). Earlier versionswerepresentedto necessaryto understandthe relationbetweendifferent levelsof
seminarsat the universitiesof Toronto,York, and Amsterdam.I am integration in the
gratefulto them for their hospitalityand discussion,to Hannie same society,thus breaking down the arti-
Hoekstra for convertingword into print,and to Rod Aya, Norm ficialboundariesbetween part and whole that had hitherto
Shulman,and MarilouDreightonforcommenting on thefinalversion. impededsocial analysis in complexsocieties.
392 CURRENT ANTHROPOLOGY

This content downloaded from 128.227.214.253 on Wed, 6 Nov 2013 13:48:55 PM


All use subject to JSTOR Terms and Conditions
4. Networkanalysis focusesnot only on interlinkage, but importanceof kinshipin complexsocieties,the ways in which
also on the contentof the relations.In otherwords,the first leadersrecruitand manipulatesupport,and the way in which
plot of a networkof relationsprovidesa systematicblueprint gossip is circulated.It has been used to combat organized
forfurther investigation into theircontent. crime,to delineatetheoverlappingpositionsfromwhichpower
5. Networkanalysis,by also focusingupon content,sensi- is exercisedthroughinterlocking directorships,and to examine
tizestheinvestigator to theinherenttensionin social relations many otherproblems.Networkanalysis can also be used to
betweenpersonswhohave differential access to resourceswhich learnmoreabout class and interclassrelations,interethnic re-
affectpowerchances.The way in whichnetworkanalysisac- lations, the ramification of multinationals,and the way in
cents thisinherenttensionand asymmetry in social relations whichsocialmilieuaffectsmentalhealth.Yet it has made little
is an antidoteto the structural-functional preoccupationwith contribution to thesefields.
consensus,order,balancedopposition,and harmony. Networkanalysishas not realizedits potentialfora number
6. Networkanalysis,thus,by providinga systematicframe- of reasons.Amongthese are an overelaborationof technique
workforanalyzingtensionand asymmetry in social relations, and data and an accumulationof trivialresults.Basically,net-
sensitizesthe investigatorto the inherentdynamicsin such workanalysisis very simple: it asks questionsabout who is
relations.Since such relationsare part of groups as well as linkedto whom,thenatureofthatlinkage,and how thenature
institutionalcomplexes,the social investigatoris alerted to of the linkageaffectsbehaviour.These are relativelystraight-
the dynamicnatureof societyand to the humandimensionof forwardquestions,the resolutionof whichis fairlysimple.For
such dynamism.Changes are thus perceivedas inherentin variousreasons,theyhave givenriseto an arsenalof concepts,
personalrelationsand hencein society.This again is an anti- terms,and mathematicalmanipulationsthat terrifies potential
dote to thestructural-functional assumptionofequilibrium. users.Anthropologists, sociologists,and politicalscientistshave
7. Networkanalysisalso gets away fromthe piecemealor borrowedheavily-fartoo heavily,in myopinion-frommathe-
institutionalapproach. By charting,for example,a person's maticalgraphtheory.As a result,theyare in veryreal danger
networkof intimatesor the networkactivated by an action ofsuffocation bythejargon,theory, and techniquesdevelopedto
set or that of a politicianmobilizingvotes, networkanalysis resolvequitedifferent problemsin anotherdiscipline.To present
movesbeyondthe traditionof limitinganalysisto discretein- theanthropologist interestedin politicalmobilizationwiththis
stitutionalspheressuch as economics,politics,or, especially arsenal is like givinga do-it-yourself programmefornetwork
foranthropologists, kinship.Networkanalysiscuts across the analysisand a computerterminalto a fisherman who merely
conceptualbarriersof an institutional approach. wishesto explainto his son how to unravelhis tanglednet.
8. By its focus on interrelation, interdependency, and in- The batteryof techniqueswithwhichsocial scientistshave
teraction,networkanalysisalso makes it possibleto deal with equipped themselvesto answerthe limitedquestionsthat net-
formsof social organizationthatemergefrominteraction, such work analysis can resolveproducesoverkill.Flies are killed
as patron-client chains,leader-follower coalitions,cliques,fac- withdynamite.Certainly,the help of statisticaland computer
tions,cartels,and othertemporaryalliances at various social specialistsis neededifthenumbersofinformants and variables
levels. These formsof social organizationin the recentpast make hand computationproblematic.Most calculations,how-
weregenerallyignoredor relegatedto interstitial, peripheral, or ever,have to do withsimplenosecountingand cross-tabulation.
residualcategoriesof social analysis(Boissevain 1968). It will Neitherthe questionsasked northe typeand reliabilityof the
be obviousthat thereare formsof social organizationthe un- data normallywarrantthe use of the techniquesand concepts
derstandingof which is essential to the comprehensionof whichhave reachedus fromgraphtheory.As enthusiasticnet-
manylargeand smalleventsin thelivesofpersonsand groups. workpractitioners strivetowardsevergreaterrigour,network
9. Networkanalysisprovidesa way of relatingformal,ab- analysisrisksbecomingfurtherremovedfromhumanlifeand
stractsociologicalanalysisto everydayexperience,forit links boggeddown ever deeperin the swampof methodological in-
interpersonal relationsto institutions.It thushumanizessocial volution(Hannerz 1975:27; Leeds 1972:5; Sanjek 1974:596;
analysisby reintroducing "people," as opposed to "roles,"and Ottenberg1971:948; Kapferer1973:167).
theirchoicesand actions into the streamof events that con- The seconddangerfacingnetworkanalysisis thatthosewho
stituteshistory. have chosen to use this methodof researchtend to trivialize
10. Finally,networkanalysisbringsinto sharp sociological its results.As Sanjek has remarked,"One does not studynet-
focusthedifficult analyticalcategoryoffriends-of-friends, those works;one uses networkmethodsto answeranthropological
personswho lie just beyondthe researcher'shorizonbecause questions"(1974:589). Far too muchoftheresearchnowbeing
theyare not in directcontactwithhis informants. done on networkslacks any clearformulation of theproblems
These, then,are some of the thingsthat networkanalysis it seeks to resolve. Networksare comparedwith regard to
can do. There are also thingsthatit cannotdo. density,size,and even composition, muchin thewaybutterfly-
Whilenetworkanalysiscan help plot the directionand con- collectorscompare the colouring,wingspread,and number
centrationof immigrants and the locationof industry,forex- of spotsof theirfavouritespecies.Trivialbut extremely costly
ample,used alone it cannotdeal withthe social processesthat resultsbased on samplesof thousandsare put forwardwith
bringabout immigration and industrialization. In otherwords, greatsolemnityby sociologists.Thus we learn that if you ask
it cannotdeal withthe social forcesunderlying long-term pro- severalhundredpersonsto name a fewpersonsoutside their
cesses. Nor can it deal adequatelywith the impactof educa- householdwithwhomtheyhave close relationships, theseturn
tional reform, land distribution, morerightsforwomen,etc., out typicallyto be kin and friends.Other studieshave dis-
or with culture,cognition,or the social forcesderivingfrom coveredthataffective relationschangeovertime.Is thisnews?
economicactivity.These dimensionsare essentialfora com- What is the social or theoreticalsignificance of these "scien-
plete understandingof social behaviour and developments. tific"discoveries?We are left,too often,to drawour own con-
Networkanalysisalone cannotprovidethem.Used alongside clusions.My conclusionis thatmanyof the studiespresented
otherresearchmethodsand formsof conceptualization, how- by enthusiasticnetworkanalystsseem merelyto confirmthe
ever,it can provideimportantadditionaldimensions. popularlyheld view that sociologyis the disciplinewhichsets
out theobviousat greatcost in an unintelligible language.The
concernwith method,classification, and networks-as-things-
THlE FUTURE OF NETWORK ANALYSIS
in-themselves, ratherthan with the ideas and problemsthat
Networkanalysishas an importantfuture.Researchershave thepractitioners are attempting to solve,characterizes notonly
alreadydemonstratedthatit is usefulforgaininginsightinto theresultsbut also, alas, theway in whichthoseresultsare re-
urban-ruralcontrasts,male-femalerelationships,the relative viewed.For example,Barnes's review(1974) of Boissevainand
Vol. 20 * No. 2 * June 1979 393

This content downloaded from 128.227.214.253 on Wed, 6 Nov 2013 13:48:55 PM


All use subject to JSTOR Terms and Conditions
Mitchell(1973) is exclusivelyconcernedwithterminology and BOTT, ELIZABETHI. 1957. Familyand socialnetwork. London: Tavi-
technique,while Sanjek's (1974) also examinesthe problems stock.
FIRTH, RAYMOND.1951. Elementsof social organization.London:
withwhichthe analystsattemptedto deal. Watts.
It is becomingincreasingly obvious that if anthropologists FREEMAN, LINTONC. 1975. A bibliography ofsocial networks.
Depart-
and sociologistscontinueto view networkanalysisas a special mentof Social Relations,Lehigh University.
fieldof inquiry,and if thosewho use it continueto encourage HANNERZ, ULF. 1975. Thinkingwith networks.MS, Departmentof
Social Anthropology, Universityof Stockholm.
thisview,it willrapidlybecomeoverlytechnicaland its results HUMMEL,HANS J.,and ROLF ZIEGLER.Editors. 1977. Anwendung
progressivelytrivial (Sanjek 1974:596). Networkanalysis is mathematischer Verfahren zur A nalysesocialerNetzwerke. Duisberg:
a researchinstrument whichcan help resolvecertainsocial and Sozialwissenschaftlichen Kooperative.
theoreticalproblems.It must not become an esotericend in KAPFERER,BRUCE.1969. "Norms and the manipulationof relation-
ships in an Africanfactory,"in Social relationsin urbansituations.
itselfwhose practitionerscan communicateonly with each Edited by J. Clyde Mitchell,pp. 181-244.Manchester:Manchester
otherabout scientificpuzzles of interestonlyto themselves.If UniversityPress.
those who have used networkanalysis considerthat it can -. 1972. Strategyand transactionin an Africanfactory.Man-
providevaluable insights,let themdemonstratethis to their chester:ManchesterUniversityPress.
. 1973. "Social networkand conjugal role in urban Zambia:
scepticalcriticsby makingtheirresultsand methodsrelevant Toward a reformulation of the Bott hypothesis,"in Networkanal-
and understandable.Conferencesof network"specialists,"a ysis. Edited by J. Boissevain and J. Clyde Mitchell,pp. 269-80.
journal,and a special societyto cater to theirneeds are dis- The Hague: Mouton.
turbingsignsof an involutionwhichwill ultimatelyresultin LEACH, E. R. 1954. Politicalsystems ofhighlandBurma.London: Lon-
don School of Economics.
networkanalysis's joining the dodo, Neanderthalman, and LEEDS, ANTHONY. 1972. Urban anthropologyand urban studies.
sociometry as an extinctspecies. UrbanAnthropology Newsletter 1:4-5.
MITCHELL, J. CLYDE. 1969.Social networks in urbansituations.Man-
chester:ManchesterUniversityPress.
-. 1974. Social networks.Annual Reviewof Anthropology 3:
279-99.
References
Cited OTTENBERG, SIMON. 1971. Review of: Social networks in urbansitua-
tions,by J. Clyde Mitchell (Manchester: ManchesterUniversity
BARNES, J. A. 1954. Class and committeesin a Norwegian island Press, 1969). AmericanAnthropologist 73:946-48.
parish.Human Relations7:39-58. SANJEK, ROGER. 1974. What is networkanalysisand what is it good
. 1968. "Networksand politicalprocess,"in Local levelpolitics. for?Reviewsin Anthropology 1:588-97.
Edited by Marc Swartz,pp. 107-30. Chicago: Aldine. SHULMAN, NORMAN. 1976. Networkanalysis: A new addition to an
--. 1969. Graph theoryand social networks:A technicalcom- old bag of tricks.Acta Sociologica19(4).
menton connectednessand connectivity.Sociology3:215-32. TURNER, U. W. 1957. Schism and continuity in an Africansociety.
Manchester:ManchesterUniversityPress.
1972. Social networks.
Reading, Mass.: Addison-Wesley. VAN VELSEN, J. 1964. Thepoliticsofkinship.Manchester:Manchester
1974. Review of: Networkanalysis,edited bv J. Boissevain UniversityPress.
and J. Clyde Mitchell (The Hague: Mouton, 1973). Man 9: WELLMAN, BARRY. 1976. Urban connections. Centre for Urban and
497-99. CommunityStudies and Departmentof Sociology,Universityof
BOISSEVAIN, JEREMY. 1968. The place of non-groupsin the social Toronto,Research Paper 84.
sciences.Man 3:542-56. WHITTEN, NORMAN E., JR., and ALVIN W. WOLFE. 1974. "Network
1974. Friendsoffriends:Networks,manipulatorsand coali- analysis," in Handbookof social and culturalanthropology. Edited
tions.Oxford:Basil Blackwell. by J. Honigmann.Chicago: Rand McNally.
BOISSEVAIN, JEREMY, and J. CLYDE MITCHELL. Editors. 1973. Net- WOLF, ERIC R. 1956.Aspectsofgrouprelationsin a complexsociety:
workanalysis:Studiesin humaninteraction. The Hague: Mouton. Mexico. AmericanAnthropologist 58:1065-78.

Continuities
andChangein Tropical na nationstoday. Human occupationof the savannas reaches
back manymillenniain Africa,Asia, Australia,and theAmeri-
SavannaEnvironments cas, duringwhichtime diverseways of life evolved in these
environments; but todaypatternsof resourceuse are changing
rapidlyas establishedsystemsof pastoralismand cultivation
byDAVID R. HARRIS are beingmodifiedand replacedby commercialranchingand
Department of Geography, University CollegeLondon,Gower by large-scaleprojectsof agriculturaland industrialdevelop-
St., LondonWC1E 6BT, England.14 x 78 ment.Many of thesechangesare takingplace in nation-states
Between the equatorial rain forestsand the perenniallydry thatareexperiencing rapidpopulationgrowthand urbanization
desertsof the subtropicalhigh-pressure belts lie the world's while enduringpersistenthazards of droughtand endeKiic
"savanna lands." They occupyabout one-fourth of theworld's disease.Analysisofthecapacityof savannaecosystemsto sup-
land surface,supportvaried plant and animal communities, port morepeople and to sustainnew modes of land use is a
and encompassthe greatestnumberand diversityof human prerequisiteforsuccessfuldevelopment.
societieswithinthe tropics.Perhapsbecauseof theirgreateco- The conferencebroughttogethera groupwhoseexperience
logical and social diversity,they are seldom perceivedas a spannedthe tropicalcontinentsand whoseexpertisefellwithin
geographicalentity,and theyhave attractedless scholarlyat- thefieldsofarchaeology, anthropology,botany,economics,epi-
tentionthan eitherthe desertsor the rain forests.It was to demiology, geography, nutrition,
physiology,and zoology.The
help redressthis imbalancethat a Wenner-GrenFoundation firstquestionaddressed-and quicklydisposedof-by partici-
conferencewas held at Burg WartensteinAugust4-13, 1978, pantswas thedefinition ofsavannaenvironments. A necessarily
on the theme"Human Ecologyin Savanna Environments." arbitraryclimaticdefinition proposedby the organizerin his
The purposeof the conference was to examinein worldwide preconference paper was acceptedas settingbroadlimitswithin
comparativeperspectivethe ways in whichpast and present whichgradientsof environmental variationwere recognized.
humanpopulationshave adapted to and made use of tropical Thus theIntermediate Tropicalor SavannaZone can be defined
savanna environments.Comparative examination of this as thatpart of the tropicalworldthat experiencesa dryseason
themewould,it was hoped,yield improvedunderstanding of of2.5 to 7.5 months'duration.Its commonclimaticdenomina-
theecologicaland socioeconomic changestakingplace in savan- tor is the occurrenceof a winterdryseason that checksplant
394 CURRENT ANTHROPOLOGY

This content downloaded from 128.227.214.253 on Wed, 6 Nov 2013 13:48:55 PM


All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen