You are on page 1of 44

NetworkAnalysis,Culture, and the

Problem of Agencyl
Mustafa Emirbayer
New Schoolfor Social Research

JeffGoodwin
New York University

Networkanalysisis one ofthemostpromisingcurrentsin sociologi-


cal research,and yetit has neverbeen subjectedto a theoretically
informedassessmentand critique.This articleoutlinesthetheoreti-
cal presuppositionsof networkanalysis. It also distinguishesbe-
tweenthreedifferent (implicit)modelsin the networkliteratureof
the interrelationsof social structure,culture,and human agency.
It concludesthatonlya strategy forhistoricalexplanationthatsyn-
thesizessocial structuraland culturalanalysiscan adequatelyex-
plain theformation, reproduction, and transformation ofnetworks
themselves.The articlesketchesthebroad contoursof such a theo-
reticalsynthesisin the conclusion.

Recentyearshave witnessedthe emergenceof a powerfulnew approach


to the studyof social structure.This mode of inquiry,commonlyknown
as "networkanalysis,"has achieveda highdegreeof technicalsophisti-
cation and has proven extremelyusefulin a strikingly wide range of
substantiveapplications.Since the seminalwork of Barnes (1954) and
Bott(1971), sociologicalstudiesutilizingnetworkanalysishave appeared
withincreasingfrequency;a veritableexplosionof such workhas taken
place overthelast 15 years,particularlywiththefoundingoftwospecial-

' We wouldliketo thankparticipantsin theCROPSO Workshop at HarvardUniver-


sity(esp. Theda Skocpol)fortheirmanyhelpfulcomments, in
as wellas participants
theSociologyStaffSeminarat theNew SchoolforSocial Research(esp. JanetAbu-
Lughod,Diane Davis, Karl-DieterOpp, and ArthurVidich)and the Methodology
Workshop at theNew YorkUniversitySociology
Department (esp.WolfHeydebrand,
Guillermina Jasso,and JamesJasper).We wouldalso liketo thankPeterBearman,
GerardoDel Cerro,KarenGelb,Jeffrey Goldfarb,RogerGould,AnnMische,Calvin
Morrill,JohnPadgett,AlessandroPizzorno,MargaretSomers,CharlesTilly,and the
AJS referees fortheirmanyhelpfulsuggestionsalongtheway. Correspondence may
be addressedto MustafaEmirbayer, Department ofSociology,New SchoolforSocial
Research,65 FifthAvenue,New York,New York 10003.
? 1994byThe University
ofChicago.All rightsreserved.
0002-9602/94/9906-0001$01.50

AJS Volume 99 Number6 (May 1994): 1411-54 1411

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

ized journals,Social Networksand Connections,in thelate 1970s.Today


networkanalysis is one of the most promisingcurrentsin sociological
research.Its practitioners includesome of the mosthighlyrespectedfig-
ures in the profession:Ron Breiger,Ronald Burt, Mark Granovetter,
David Knoke, Peter Marsden, Barry Wellman, and Harrison White.
Many other prominentsociologists,such as Claude Fischer, Edward
Laumann, Doug McAdam, David Snow, and CharlesTilly,draw exten-
sivelyupon networkconcepts.The late GeorgeHomans, in his reflections
upon the last 50 yearsof sociology,justlydescribednetworkanalysisas
one ofthemostencouragingnew developments in thediscipline(Homans
1986, p. xxvi).
Despite its growingprominence,however,networkanalysishas yetto
be subjected to a theoretically informedassessmentand critique.The
secondaryliteratureon this perspectivehas tended to restrictitselfto
outliningbasic concepts,discussingtechnicalprocedures,and summariz-
ing empiricalresearchfindings.There has been an unfortunate lack of
interestin situatingnetworkanalysiswithinthe broadertraditionsof
sociologicaltheory,much less in undertakinga systematicinquiryinto
its underlyingstrengthsand weaknesses. Theoretical"precursors"of
networkanalysishave oftenbeen invokedin passing-especially Durk-
heimand Simmel-but networkanalysis,itselfa constellation of diverse
methodological strategies,has rarelybeen systematically groundedin the
conceptualframeworkstheyelaborated.In addition,therehas been a
notable absence in this literatureof any sustainedconsiderationof the
potential usefulnessof networkanalysis for historicalinvestigation.
Meanwhile,social theorists and historicalsociologists,fortheirown part,
have largelyignoreddevelopmentsin this field;we have yet to see a
sustaineddiscussionof this approach in recentworks of social theory,
even in the writingsof such wide-ranging thinkersas AnthonyGiddens
(1984, 1987) and Jeffrey Alexander(1982, 1987, 1988a) or, alternatively,
in studiesof recentdevelopmentsin comparativeand historicalresearch
(e.g., Skocpol 1984; D. Smith 1991).2 In short,the task of rethinking
networkanalysis,sociologicaltheory,and historicalsociologyin lightof
one anotherhas been sadly neglected.
In thisessay we aspireto accomplishpreciselysuch a task. We begin
withan expositionoftheunderlying theoreticalpresuppositions and con-
ceptual strategiesof networkanalysis,outliningthe characteristic fea-
turesof this approach in relationto broadercurrentsin social theory.
Along the way, we examineseveral exemplarystudiesthat investigate
historicalprocessesof social change usingthe tools and insightsof net-

2 Giddensdoes,however,
criticize
the"structuralism"
ofPeterBlau, whichhas strong
affinities
withnetworkanalysis(Giddens1984,pp. 207-13).

1412

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

work analysis. Upon this basis we thenelaboratea critique-a funda-


mentallysympathetic critique-of the networkperspective,stressingits
inadequateconceptualizations of humanagencyon theone hand, and of
cultureon theother.We carefullydistinguish herebetweenthreedistinct
models in the networkliteratureof the relationshipsamong culture,
agency,and social structure; each ofthesemodels,in ourview, conceptu-
alizes theserelationships in varyingdegreesof theoreticalsophistication.
Throughouttheessay,we attemptto groundour criticalargumentscare-
fullyin detailed and substantiveconsiderationsof actual worksof net-
work analysis-studies that, in our estimation,rank among the most
powerfuland impressiveapplicationsto date ofthenetworkperspective.
Throughoutthis article,our primaryfocus is on analyticalcategories,
not on authors.Our concernwith agencyand culturehas implications
fornetworkanalysisin general,but our focushere remainson studies
thatattemptto explainsocial changeover time-includingthe transfor-
mationof social networksthemselves-and henceon specifically histori-
cal studies.3
Our argumentis thatwhilethisnew mode ofstructuralist inquiry-in
all threeof its versions-offersa morepowerfulway of describingsocial
interactionthan do otherstructuralperspectivesthatfocussolelyon the
categoricalattributesof individualand collectiveactors,it has yet to
provide a fullyadequate explanatorymodel for the actual formation,
reproduction,and transformation of social networksthemselves.Net-
workanalysisall too oftendeniesin practicethecrucialnotionthatsocial
structure,culture,and human agencypresupposeone another;it either
neglectsor inadequatelyconceptualizesthe crucialdimensionof subjec-
tive meaningand motivation-includingthe normativecommitments of
actors-and therebyfailsto show exactlyhow it is thatintentional,cre-
ativehumanactionservesin partto constitute thoseverysocial networks
thatso powerfully constrainactorsin turn.In its less nuanced versions,
in fact,thenetworkapproachemergesas themirrorimageofitsinterpre-
tive and hermeneuticcounterpart(Schutz 1967; Rabinow and Sullivan
1979). Whetherfromthe standpointof objectivesocial structuresor of
subjectiveexperience,both of theseperspectivesin themselvesprovide
no morethana descriptionof social reality;ultimately, bothfailto grasp
in conceptsthe dynamicprocessesthatshape thisrealityover time.We

3 We arenotinterestedhere,itshouldbe pointedout,inthemorepurelymethodologi-
cal and technicalcontributions
thatnetworkanalystshave produced(e.g., Boorman
and White1976;Davis 1967;Freeman1977,1979;Harary,Norman,and Cartwright
1965; Lorrainand White 1971; White,Boorman,and Breiger1976). Nor are we
concernedwithcoveringall ofthemanyimportant empiricalstudiesthathave been
producedin recentyearsusingnetworktechniquesand concepts.See, e.g., Fischer
1982;Laumannand Knoke1987;and Wellman1979.

1413

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

believe,by contrast,thatan adequate approachto historicalexplanation


mustencompassbothsocial structural and culturalperspectiveson social
action. We demonstratehow more sophisticatedversionsof network
analysisdo approximatesuch a strategy,and we sketchout the broad
contoursof our own synthetic explanatoryperspectivein the concluding
pages.

ANALYZING SOCIAL NETWORKS


The Priorityof Relationsover Categories
Networkanalysisis not a formalor unitary"theory"that specifiesdis-
tinctivelaws, propositions,or correlations,but rathera broad strategy
forinvestigating social structure.It is not,thatis, a "deductivesystem"
in which "lower-orderpropositions"follow "as a logical conclusion
from. . . generalpropositionsunder . . . specifiedgiven conditions"
(Homans 1964,p. 812). Rather,like modernization or dependencytheory
in the field of economic development,it is more a "paradigm" or a
"perspective"-"a loose federationof approaches"(Burt 1980a)-than
a predictive"social theory."As a resultof the internaldiversityof net-
work approaches,networkanalyststhemselvesdebate the usefulnessof
alternativemodels of social relationsand methodologicalstrategies.In-
deed, importantdisagreements have arisenovertheverydefinition of its
fundamentalconcepts.Networkanalystsdisputethe mannerin which
ideas such as social structure,networkcentrality,distance,cohesion,
and social networkitself-terms used by othersociologistssimplyas
metaphors-can be operationalizedforpurposesof empiricalresearch.
(We have provideda shortglossaryof termscommonlyused by network
analystsin the appendixbelow.)
Nevertheless,networkanalysisproceedsfromcertainbasic theoretical
presuppositionsand premisesthat are acceptableto most,if not all, of
itspractitioners. It holdsto a setofimplicitassumptionsaboutfundamen-
tal issues in sociologicalanalysissuch as the relationshipbetweenthe
individualand society,the relationshipbetween"micro"and "macro,"
and the structuringof social action by objective, "supra-individual"
patternsofsocialrelationships. The pointofdeparturefornetworkanaly-
sis is what we shall call the anticategoricalimperative.This imperative
rejectsall attemptsto explainhumanbehavioror social processessolely
in termsof the categoricalattributesof actors,whetherindividualor
collective.Networkanalysis,as BarryWellmanputsit, rejectsexplana-
tionsof "social behavioras the resultof individuals'commonpossession
of attributesand normsratherthanas the resultof theirinvolvementin
structured social relations"(Wellman1983, p. 165). In otherwords,one

1414

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

can neversimplyappeal to such attributesas class membershipor class


consciousness,politicalpartyaffiliation, age, gender,social status,reli-
giousbeliefs,ethnicity, sexual orientation,
psychologicalpredispositions,
and so on, in order to explain why people behave the way they do.
"Network theorybuilds its explanationsfrompatternsof relations,"
notes Ronald Burt. "It capturescausal factorsin the social structural
bedrockof society,bypassingthespuriouslysignificant attributesof peo-
ple temporarily occupyingparticularpositionsin social structure"(Burt
1986, p. 106).4In thisrespect,networkanalysispursuesthe Simmelian
goal of a formalisticsociology(Simmel 1971, chap. 3), one that directs
attentionexclusivelyto the overall structureof networkties while sup-
pressingconsiderationof theirsubstantivecontent(see also Bearman
1993, p. 48).5
Given this anticategoricalimperative,networkanalysisemphatically
rejectsall varietiesofculturalism,essentialism,and methodological indi-
vidualism.It stands fundamentally opposed, forexample,to certainof
theassumptionsofstructural-functionalism, whichstressesthenormative
integrationof societies.Despite a commonemphasison the priorityof
structuresover "essences"-an emphasis(derivingultimatelyfromthe
earlyDurkheim)thatlinksbothperspectivesto tendenciesin structural
anthropology, linguistics,and ordinarylanguage philosophy-network
analysts"take seriouslywhat Durkheimsaw but most of his followers
did not: that the organicsolidarityof a social systemrestsnot on the
cognitionof men, but ratheron the interlockand interactionof objec-
tively definablesocial relationships"(Boorman and White 1976, p.
1442).6 It shouldbe pointedout,on theotherhand,thatnetworkanalysis
does notrejectall ofthetenetsofstructural-functionalism. Manynetwork
analysts(e.g., White, Boorman, and Breiger1976; Burt 1982) employ
functionalist notionssuch as "role," "role set," and "status," although
they reconceptualizethese notionsinto relationalor network-analytic

4 Thus one networkanalyst(Brym1988)is able to showin his workon intellectual


radicalism amongthefin-de-siecle in Russiathatsharpdiscrepan-
Jewishintelligentsia
ciesin politicalideologiescan developevenwithinnarrowly circumscribedcategories,
depending upontheactors'different of"structural
patterns rootedness"withinpartic-
ularsocialnetworks.
5 This last formulation appliessomewhatless well,it shouldbe pointedout,to the
thirdoftheseveralnetwork modelsthatwe shallgo on toanalyze,thatofstructuralist
constructionism. This modelsupplements a "static"and formalisticanalysisof net-
workstructure witha more"dynamic"accountof processesof identity transforma-
tion,one thatnecessarily devotesmoreattention to thecontentofnetwork ties.
6 Network analysts'disagreement view of societyis
withthestructural-functionalist
thusreminiscent ofSeligPerlman'sobjectionto thoseMarxisttheoristswhoviewthe
workingclass as "an abstractmass in thegripof an abstractforce"(1979,p. 6).

1415

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

terms.In thisregard,theycan indeed be said to be workingwithinthe


traditionof Mertonand Rossi (1950), Parsons(1951), and Nadel (1957).
As a "structuralist"approach to social analysis, the networkper-
spectivenecessarilyalso questionsthe explanatorypotentialof all those
conceptualstrategiesthatemphasizethe nonrelationalattributesand/or
purposive actions of individuals or collectivities-strategiessuch as
interpretivesociology,phenomenology, symbolicinteractionism, ethno-
methodology,certainvariantsof Marxian analysis,and rationalchoice
theory.Fromthenetworkpointofview,analyticalapproachesthatdirect
attentionto the "intrinsiccharacteristics,""essences," attributes,or
goals ofindividuals,as opposedto theirpatternedand structured interre-
lationships,are all inherentlysuspect(Wellmanand Berkowitz1988, p.
5). Althoughcertainrationalchoiceanalystsdo employthemethodologi-
cal tools and techniquesof networkanalysis(e.g., Coleman 1990), and
while,conversely,certainpractitioners of networkanalysistodaytacitly
rely on instrumentalist conceptionsof action (as we shall see below),
networkanalysis itselfas a broad perspectiveon social structurecan
clearlybe distinguished fromtheradicalindividualismof rationalchoice
theory.
Finally,networkanalysisdepartsin importantways fromthe "main-
stream"sociologicalapproach to empirical"quantitative"research.As
bothJamesColeman and AndrewAbbotthave observed,Americansoci-
ology beforethe "watershed"momentof the 1940s was marked by a
proliferation of qualitativecommunitystudies,influencedin large part
by the Chicago school and, ultimately,by the writingsof RobertPark
(Park, Burgess,and McKenzie 1967),W. I. Thomas (1966), and Georg
Simmel(1955). These theoristsregardedsocial factsas ecologicallyem-
beddedwithinspecificcontextsoftimeand space - thatis to say, within
particularinteractional fieldscomposedof concrete,historically specific
"natural areas" and "natural histories.""All social facts," notes Ab-
bott,wereseen as "locatedin particularphysicalplaces and in particular
social processes.They werealso locatedwithinthetemporallogicof one
or moreprocessesofsuccession,assimilation,conflict, and so on.... The
Chicago visionwas ofa social structure embeddedin time,a structure in
process"(Abbott1992b,p. 14).
Afterthe 1940s,however,Americansociologyshifteditsfocusofatten-
tion away fromcontextualdetermination and onto statisticalsurveyre-
searchand othertypesof workthattookas theirunitof analysisnotthe
communityor the social group,but ratherthe individual."In much of
the work followingthis change," remarksColeman, "the focus shifted
fromsocial processeswithinthecommunity shapingthe system'sbehav-
ior to psychologicalor demographicprocessesshapingindividualbehav-
ior" (Coleman 1986, p. 1315).

1416

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

Therewasnocomparable developmentoftoolsforanalysis
ofthebehavior
ofinteracting
systems orforcapturing
ofindividuals theinterdependencies
actionsas theycombineto producea system-level
ofindividual outcome.
The fargreater
complexity oftoolsforthesepurposes
required constituted
a serious impedimentto theirdevelopment.. . . The end result[was]
extraordinarily
elaborated
methods foranalysisofthebehaviorofa setof
independententities
(mostoftenindividuals),
withlittledevelopmentof
methodsforcharacterizing
systemicactionresulting
fromtheinterdepen-
dentactionsofmembers ofthesystem. [Coleman1986,p. 1316]

Fromthishistoricalvantagepoint,contemporary networkanalysiscan
be viewed as part of a second crucial watershedperiod in American
sociology,one in whichempiricalresearchis now directingits attention
back again to the systemiclevel, thistimeassistedby the development
of quantitativetechniquesand methodsof a highlysophisticatednature.
In thissecondpivotalmoment,analyticconcernsare shifting back once
moreto thosequestionsof interactional fieldsand contextualdetermina-
tionthathad been so centralto sociologistsbeforethe"variablesrevolu-
tion" of the 1940s.
How, then,does networkanalysisproposeto accountforsocial behav-
ior and processes?The answeris implicitin theprecedingremarks.Such
behaviorand processes,it suggests,mustbe explainedwithreference to
networksof social relationsthat link actors or "nodes." These social
relations,significantly, mustbe understoodas independentof the actors'
wills, beliefs,and values; theymustalso be assumed to allocate scarce
resourcesdifferentially (Wellman1983, p. 176). Social structure,in this
view, is "regularities in thepatternsof relationsamongconcreteentities;
it is not a harmonyamongabstractnormsand values or a classification
of concreteentitiesby theirattributes"(Whiteet al. 1976, pp. 733-34;
emphasisin original).A social networkis one of manypossible sets of
social relations of a specificcontent-for example, communicative,
power,affectual,or exchangerelations-thatlinkactorswithina larger
social structure(or networkof networks).The relevantunitof analysis
need not be an individualperson,but can also be a group,an organiza-
tion,or, indeed,an entire"society"(i.e., a territorially
boundednetwork
of social relations);7any entitythat is connectedto a networkof other
such entitieswill do.
Networkanalystsoftenfinditdesirableto carryouttheirinvestigations
at boththeindividualand grouplevels;suchcombinations highlightwhat
some ofthem,in a Simmelian(1955) fashion,referto as the"dualism"of
groupsand actors-the factthatthe natureof groupsis determinedby
the intersectionof the actorswithinthem(i.e., by theties of theirmem-

7One recent writer,Margaret Somers (1993), prefersthe term "relational setting."

1417

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

bersto one anotheras well as to othergroupsand individuals),whilethe


nature of actors is determinedby the intersectionof groups "within"
them (i.e., by theirown various group affiliations; see Breiger 1974).
Individualand groupbehavior,in thisview, cannotbe fullyunderstood
independently of one another.By thus facilitating analysesat both the
individualand grouplevel, networkanalysismakesit possibleto bridge
the "micro-macrogap"-the theoreticalgulfbetweenmicrosociology,
which examines the interactionof individuals, and macrosociology,
whichstudiesthe interaction of groupsor institutions.
Thus theway in whichnetworkanalysisconceptualizessocialstructure
is at once moregeneraland moreconcretethan alternativestructuralist
approaches.It is moregeneralbecause manydifferent kindsof groups,
relations,and institutions thatputativelyorganizeor structure social pro-
cesses can be understoodin, or be "translated"into, networkterms.
And it is moreconcretebecause thesestructures need not be treatedas
"black boxes," but rathercan be disaggregatedinto theirconstituent
elementsof actorsand relations.The significance of thesetwo features
ofnetworkanalysisforempiricalresearchis considerable.Networkana-
lystsare now able to providefarmorepreciseand accuraterepresenta-
tionsofsocial structures and social relationsthanare proponentsofcom-
petingresearchstrategies.
Most networkanalysts,of course, also purportto do far more than
simplydescribethe ways in which actorsare connectedin society.As
Knoke and Kuklinskipointout, "If networkanalysiswere limitedto a
conceptualframeworkforidentifying how a set of actorsis linked to-
gether,it would not have excitedmuchinterestand effortamongsocial
researchers.But networkanalysiscontainsa further explicitpremiseof
greatconsequence:The structure ofrelationsamongactorsand theloca-
tionof individualactorsin thenetworkhave importantbehavioral,per-
ceptual, and attitudinalconsequencesboth forthe individualunitsand
for the systemas a whole" (Knoke and Kuklinski1982, p. 13). "The
hallmarkof networkanalysis," in Edward Laumann's (1979, p. 349)
words,"is to explain,at least in part,the behaviorof networkelements
(i.e., thenodes)and ofthesystemas a wholebyappeal to specificfeatures
of the interconnections amongthe elements."More specifically, the net-
workapproachinvestigatesthe constraining and enablingdimensionsof
patternedrelationshipsamong social actors withina system.It is 'this
emphasis,in fact,thatprovidesthe link betweenits theoreticalinsights
and its key contributions to empiricalresearch.What networkanalysis
provides,in particular,is a way of avoidingthe pitfallsof what Arthur
Stinchcombeterms"epochal interpretations" (or, morememorably,"ep-
ochal garbage"); that is, of causal explanationsthat proceedby "using
theapparentcausal structure createdbynarrativeofa sequenceofevents

1418

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

to create the illusion that epochal theoriesare being substantiated"


(Stinchcombe1978, p. 10; see also Tilly 1992). Networkanalysisallows
historicalsociologists,by contrast,to pinpointthose wide-rangingand
recurrentcausal mechanisms"whose combinationsproduce," as Tilly
expressesit, "the actual unique historieswe observe"(Tilly 1992,p. 11).
How, specifically, does networkanalysisattemptsuchan important task?

RelationalAnalysisas a Way of Representing


Social Structure
Networkanalystsgenerallymake use of one oftwo conceptualstrategies
in order to explain how networksconstrainand enable-and thus to
accountforvarioustypesof social behavior(Burt 1980a, 1987). On the
one hand, many analystsadopt a "relational"or "social cohesion"ap-
proachthatfocuseson thedirectand indirectconnectionsamongactors.
This approach explainscertainbehaviorsor processesthroughthe fact
of social connectivity itself-as well as throughthe density,strength,
symmetry, range,and so on, ofthetiesthatbind. Fromthisperspective,
verystrong,dense, and relativelyisolatedsocial networksfacilitatethe
developmentof uniform"subcultures"and of strongcollectiveidentities;
thisnotion,ofcourse,dates back bothto Durkheim(1984) and to Simmel
(1955; 1971, chap. 18). Relationalanalyses,however,also demonstrate
that"weak" tiesindirectly connectingindividualsor bridgingthe"struc-
tural holes" between isolated social groups may be crucial for many
importantsocial processes,such as locatingemploymentopportunities
(e.g., Granovetter1973; Burt 1992). The mathematicaltools of "graph
theory"have been helpfulin developingthese approaches(see Harary
et al. 1965).
One importantapplicationof networkanalysisthatemploysthe rela-
tionalor social cohesionapproachis a studyby Naomi Rosenthalet al.
(1985) of women'sreformorganizationsin New York State duringthe
latterhalfof thenineteenth century.Rosenthaland herassociatesexam-
ine the organizationalaffiliationsof 202 prominentwomenreformers in
state reformactivityduringthe yearsbetween1840 and 1914. By map-
pingout theseaffiliations, theydevelopa detailedportraitof the "multi-
organizationalfield"of social movementactivitythat obtained during
thatperiod.Measuresof centrality enable themto identifytheparticular
groupsthatweremostimportant to thatnetworkofreform organizations.
Otherrelatedmethodsprovidethemwith carefulmeasurementsof the
boundaries,shape, and textureof variousclustersof such organizations.
Finally, "directionalanalysis" of flowsof individualsacross organiza-
tionsprovidesthemwitha pictureof "threeuniquelyshaped map con-
figurations" of social movementclustersin threedistincthistoricalpe-
riods between 1840 and 1914, each of these manifestinga "different

1419

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

contentto the relationsamongorganizationsand activity"(Rosenthalet


al. 1985, p. 1043).
Anotherimportantline ofnetworkresearchemployingrelationalanal-
ysis concernsitselfwith processesof recruitment to social movements.
Its pointof departureis a studyby Snow, Zurcher,and Ekland-Olson
(1980) thatexaminesmaterialson a wide varietyofmovements.Snow et
al.'s work suggeststhat individualswith preexistingties to movement
membersare morelikelyto be contactedand recruitedto those move-
mentsthan are individualswithoutsuch ties; individualswith few or
weak ties to alternativenetworksare morelikelyto respondfavorably
to theserecruitment efforts thanare individualswithstrongcommitments
to countervailing networks.Snow et al.'s studydoes notemploysophisti-
cated networkanalysistechniques,nor does it examinesystematically
the independenteffectsof networkstructures themselves;it focusesin-
stead upon individualtiesalone. And yetit does providea usefulcorrec-
tive to social psychologicaland culturalistapproachesthatplace undue
explanatoryweightupon such variables as "individualmotivation"to
the exclusionof actors'patternsofembeddednesswithinactual networks
of social ties.
A seriesof studiesby Doug McAdam (1986, 1988; see also Fernandez
and McAdam 1988, 1989) further corroborates and advances thisline of
investigation.McAdam's studiesare concernedwiththe participationof
collegestudentsin the 1964 FreedomSummerprojectin Mississippiand
focus on the patternsof social relationshipsof studentparticipants,as
comparedto thoseofstudentswho appliedfortheproject,wereaccepted,
but failed to participate.McAdam's principalconclusionis that "all of
the applicants-participantsand withdrawalsalike-emerge as highly
committed,articulatesupportersof the goals and values of the summer
campaign" (McAdam 1986, p. 73). What clearlydifferentiates partici-
pants fromwithdrawalsare not theirsocioeconomiccharacteristics nor
attitudinaldifferences that "push" individualsinto participating,but
threestructural"pull" factorsthatfacilitatedand encouragedparticipa-
tion: (1) participantsbelongedto a greaternumberof organizationsand
to moreexplicitly politicalorganizationsthandid withdrawals;(2) partici-
pants had higherlevels of involvementin prior civil rightsactivities
than did withdrawals;and, mostimportant,(3) participantshad more
ties-especially "strong"friendship ties-to otherFreedomSummerap-
plicantsthan did withdrawals.
An articlebyMcAdam and RobertoFernandez(1988; see also Fernan-
dez and McAdam 1989) extendsthe scope of thisanalysisto includethe
effects ofeven morespecifically structuralvariablessuch as the"network
prominence"of applicantsupon theireventualparticipation in Freedom
Summer. McAdam and Fernandez argue that individuals' positions

1420

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

withinmultiorganizational fields-in thiscase, the variousnetworksof


interlockingactivistorganizationsat theirrespectiveuniversities-did
undercertainconditionssignificantly affectthe likelihoodof theirbeing
recruitedto the summerproject.It was not the quantityof individuals'
ties to social movementorganizationsor to otherapplicantsto Freedom
Summerthatdetermined theireventualparticipation, but ratherthepat-
ternof theirinterpersonal connectionsand commonmemberships within
these organizations.To capture the effectsof these patterns,Fernan-
dez and McAdam employa standardcohesionmeasureof centrality-
"prominence"-in theiranalysis;theyuse it to focus attentionon the
densityof applicants'tiesto othermoreor less centrallylocatedindividu-
als withinfieldsof overlappingorganizations.
Despite his increasinginterestin the significanceof "networksas
such" in movementrecruitment, however-as opposed to social ties
"treat[ed]. . . in piecemealfashion"(Gould 1991, p. 717)-McAdam's
researchleaves out of considerationone furtherstructuralvariable:
namely,themultiplexity ofnetworksthemselvesand thecomplexinterac-
tionsamongthem.WhileMcAdam's studiesneglectthepreexisting webs
of relationships withinwhichFreedomSummerapplicantshad been em-
bedded beforejoiningactivistorganizations,RogerGould's (1991, 1992)
studiesof the Paris Communeexploretheinteractiveeffectsof precisely
such networkswith more formalorganizationalstructures.Together,
Gould argues-not separately-theseindigenousstructures (specifically,
the neighborhoodswithinwhich the Parisian insurgentslived) and the
National Guard unitsto whichthe insurgentswere assigneddecisively
affectedtheiroveralllevels of solidarityand resistance.The key to this
interaction processwas theresidentialrecruitment systemoftheNational
Guard. "Membersof each battalionwere tied to each othernot only
throughtheirsharedorganizationalaffiliation, but also by the factthat
theywere neighbors."Gould findsthat

in different
insurgents neighborhoodsinfluencedeach other'sdegreeof
commitment to theinsurrection
through thenetwork oflinkscreatedby
overlapping
enlistments
[inguardunits].Highlevelsofcommitment inone
areaenhancedcommitment elsewhere
whenenlistment patterns
provided
a conduitforcommunicationand interaction.. . . [Thus] neighborhoods
respondedtoeventsinotherareaswheretheirresidents
servedinNational
Guardunits.Forinstance, inthefifth
resistance arrondissementwas posi-
affected
tively bythefactthatmanyofitsresidentsservedinthethirteenth
legion,whosemembers a strong
demonstrated commitment totheinsurgent
effort.
[Gould1991,p. 726]

Gould concludesfromthis that "cross-neighborhood solidarity"was a


significantfeatureof the Parisianinsurrection. "The interdependence of
resistancelevels acrossresidentialareas was . . . intimatelytiednotonly

1421

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

to the quantity,but also to the structureof overlappingenlistments"


(Gould 1991, p. 727). Gould thusshowsthatstructuralanalysisneeds to
take into accountnot onlyindividual-levelvariablessuch as thosethat
are employedin theMcAdam studies,but also thecomplexinfluencesof
multiplexor overlappingnetworksof social ties.

Social Structure
PositionalAnalysisas a Way of Representing
Many networkanalystsemploya different approachto conceptualizing
social structure;their"positional"strategyfocusesupon the natureof
actors'ties not to one another,but to thirdparties.This strategymakes
sense of certainbehaviorsand processesin termsof the patternof rela-
tions that definesan actor's positionrelativeto all otheractorsin the
social system.Positionalanalyses emphasizethe importanceof "struc-
turalequivalence"-that is, thesharingbytwo or moreactorsofequiva-
lentrelationsvis-'a-visa thirdactor-for understanding bothindividual
and collectivebehavior(see Lorrainand White1971). The relevantissue
fromthis pointof view is the specific"position"or "role" that a set of
actorsoccupies withinthe systemas a whole. Any such set is termeda
"block." An algebraic procedure called "blockmodeling"partitions
overall populationsinto sets of structurallyequivalentactors(Whiteet
al. 1976; Boormanand White 1976).
Structuralequivalencemodelsdifferfromrelationalmodelsin at least
two crucialrespects.First,whilethelatterfailto distinguishamongthe
membersof social "cliques" on the basis of those members'different
typesof ties to externalactors,the formerdo concernthemselveswith
the structureof the social systemas a whole. They generatemodels"in
whichan actoris one of manyin a systemofinterconnected actorssuch
that all definedrelationsin which he is involvedmust be considered"
(Burt 1980a, p. 80). Second, structuralequivalencemodelspay no heed
to whetheractorsin a givenpositionhave any directtiesto one another.
A block, or set of structurallyequivalentactors,fromtheirperspective,
may not be a denselyknitsocial clique at all (Whiteet al. 1976).
One interesting exampleof networkanalysisthatmakes use of block-
modelingand positionalanalysisis theworkof PeterBearman(1993) on
local elitesocial structurein England duringthe centurybeforethe En-
glish Civil War. Bearman examinesthe actual patternsof kinshipand
patron-clientties that bound elite actors together(and that simulta-
neouslydrovethemapart)duringthisperiod,and thereby"inducesparti-
tions" or "equivalencyclasses" of these actors that describetheirpat-
ternedinteractionsmore accuratelythan do the standard categorical
classifications.Ratherthan specifythe relevanteliteactorsfromafarin
such categoricaltermsas "middling,""rising,""falling,""court," or

1422

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

"country"gentry,he generatesspecificblockmodelsof theirinterrela-


tionshipsin fourdiscretetimeperiodsbetween1540and 1640. "Individu-
als with distinctivepersonal biographiesact coherently,"he argues,
"with respectto the interestswhich arise fromthe structuralpositions
theyshare in local and nationalnetworks"(Bearman 1993, pp. 11-12).
Changes over timein the structureand distribution of these networks
thus provide the key to understandinglong-termstructuralchanges.
Bearman's largerargumentis that the social transformations that did
occur at the elite level created the "structuralprerequisites"for the
widespreadadoptionof certainabstracttheologicaltenets,whichin turn
would play a crucialrolein inspiringthetumultuouspoliticaleventsthat
were to come. They led, in otherwords, "to the formationof an elite
subworldorganizedon the basis of social relationsthattranscendedthe
[localistand kinship-based]traditionalsocial orderand createda context
in whichabstractreligious[and constitutionalist] rhetoricscould emerge
as thecriticaldeterminants of elitesocial actionin thecenturypreceding
the EnglishCivil War" (Bearman 1993, p. 1).
Anotherhighlysophisticatedand innovativehistoricalcase studyem-
ployingthepositionalapproachto structural analysisis JohnPadgettand
ChristopherAnsell's brilliantwork on the rise of the Medici in early
15th-century Florence. Padgettand Ansell, much like Bearman before
them,employthe conceptof structural equivalenceto identify "thefam-
ily, economic,and patronagenetworksthat constitutedthe Medicean
politicalparty"(Padgettand Ansell 1993,p. 1260),as well as theirmain
rivals in Florentinepolitics,the "oligarchs."They use archivaldata on
marital,economic,political,and personalties to produce an "overall
relationalpictureof Florence'ssocial structure, within[a] 92-familyrul-
ing elite" (Padgettand Ansell 1993, p. 1274). Throughblockmodeling
theyfindthat they are able to predictactual partymembershipsfar
more accuratelythan throughstandardcategoricalanalyses based, for
example, on class and status. Indeed, theiranalyses of marriageand
economicties in particularprove highlyusefulin revealingthe connec-
tionsamong networks,groups,and partymembership.In a pointedre-
minderof the potentialempiricalusefulnessof what we have termed
here the "anticategoricalimperative,"Padgettand Ansell concludethat
"ratherthan partiesbeing generatedby social groups,. . . both parties
and social groupswere inducedconjointlyby underlying networks....
We do not argue . . . thatsocial attributesand groupsare irrelevantto
partyformation;merelythattheirrole needs to be understoodwithina
deeperrelationalcontext.Thereis no simplemappingofgroupsor spatial
dimensionsontoparties;social attributes and groupinterestsare 'merely'
cognitivecategories,which party mobilization,networks,and action
crosscut"(Padgettand Ansell 1993,pp. 1277-78, 1274;emphasisadded).

1423

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

Blockmodelingalso provesusefulin explainingthe superiorcapacity


forcollectiveactionof theMedici partyvis-'a-visits rivalfaction,despite
the bitterdivisionbetweenthosepatricianelitesand "new men" aligned
with the Medici. The Medici party,Padgett and Ansell observe, was
"deeplycleaved on two attributional dimensionssimultaneously-social
class (i.e., prestige)and neighborhood.Not onlydid thevariouscompo-
nentsdespise each other;theydid not run intoeach othermuch either.
Only the Medici familyitselflinkedthe segments"(Padgettand Ansell
1993,p. 1281).In fact,theMedici familyexertedan exceptionaldegreeof
centralizedcontroloveritsfollowerspreciselyby bridgingthe"structural
hole" betweenpatriciansand new men. (On theimportanceof structural
holesmoregenerally,see Burt[1992].)The morecliquisholigarchparty,
by contrast,was constantlybesetwith"cross-pressure[s] on each family
insteadof collectiveconvergence,"due preciselyto itsfarhigherlevelsof
networkmultiplexity and attributionalhomogeneity (Padgettand Ansell
1993, p. 1279).
Padgettand Ansellrevealthroughan analysisof "networkdynamics"
just how these patternsof social relationshipscame about in the first
place. "Elite marriageand economicnetworks,"theyargue, "were re-
configuredby working-classrevoltand by wartimefiscalcrisis,respec-
tively"(Padgettand Ansell 1993, p. 1287). As part of theirstrategyof
reconsolidationin the aftermathof the Ciompi wool workers'revoltof
1378, the temporarily victoriousoligarchsdeliberatelyexcludedthe los-
ers' elite collaborators,includingthe Medici, fromtheirmarital net-
works. "The oligarchicclique and the Medicean . . . networks,"claim
Padgettand Ansell, thus "both emergedin tandem,a singlenetwork,
each reflexively and asymmetrically the other"(Padgettand
structuring
Ansell 1993, p. 1298). In the meantime,foreignwars withMilan (1423-
28) and Lucca (1430-33) had theeffectofseverelyweakeningtheFloren-
tineeconomyand of sendingmanyelitefamilies(withthenotableexcep-
tion of the Medici themselves,who enjoyed special papal ties) into
bankruptcy.The elitefamiliesattemptedto helpthemselvesbydepriving
in turnthe sociallyand politicallymorevulnerablenew mercantileele-
ments-the new men-of officesand tax revenues.After1427, the Me-
dici began to pursueeconomicrelationswiththesenew elementsin their
local residentialneighborhoods. It was preciselyat thatmomentthatthe
Medici partyemergedas a powerfuland self-consciouspoliticalactor
engagedin struggleagainstthe dominantoligarchfaction.

THREE NETWORK MODELS OF HISTORICAL EXPLANATION


Each of thesetwo alternativemodelsof social structure
has its strengths
and weaknesses. Proponentsof the relationalapproach emphasize its

1424

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

suitabilityformappingthe typicalrelationsthat individualshave with


one another.It is more amenable to traditionalsurveyresearchtech-
niques,theyclaim,in contrastto positionalapproachesthatrequiredata
forall the elementsin the system.Proponentsof structuralequivalence,
conversely,stressits moreconsistently structuralistnature,its capacity,
that is, simultaneouslyto take into account all of the relationaldata
pertainingto a given actor, includinghis patternsof externalrelation-
ships-"the relationsin whichhe is involvedas well as the relationsin
which he is not involved" (Burt 1980a, p. 131). "A global [positional]
approach," notesMichael J. Mandel (1983, pp. 376-77), "examinesall
interlocks betweenrolesat thesame time .... [This]simultaneousconsid-
erationofall actorsin an entirepopulation. .. spotlights theinterdepen-
dence of the different rolesfoundin the population.Each role therefore
has builtintoit theelementswhichdifferentiate itfromotherroleswithin
the same overallstructure."Positionalanalystscontendthatthisfeature
lendstheirmodelsa significantly greaterdegreeof predictivepower. For
example,in a reconsideration of a classic "cohesion"studyby Coleman,
Katz, and Menzel (1966), Burt showshow the structuralequivalenceof
physiciansbetteraccountsforthe informalsocial pressuresthat led to
the diffusionof the use of the drug tetracycline than does theirsocial
cohesion(Burt 1987).
Networkpractitioners employbothof theseapproachesto make sense
of an impressivearrayof substantiveissues. Despite the considerable
methodologicalusefulnessof both strategies,however,we contendthat
each can be subjected to a numberof importantcriticisms.The very
assumptionsand premisesthat have made networkanalysisin both of
its methodologicalvariantssuch a powerfultool raise seriousquestions
about itsadequacy as an overallresearchstrategy.We contendthatthere
have been threemodelsimplicitin the literatureon networkanalysis-
modelsoftherelationships amongculture,agency,and social structure-
thathave led to varyingdegreesof difficulty in elaboratingsatisfactory
explanationsof historicalprocesses.The firstof these implicitmodels,
thatof structuralist determinism, neglectsaltogetherthepotentialcausal
role of actors' beliefs,values, and normativecommitments-or,more
generally,of the significance of culturaland politicaldiscoursesin his-
tory.It neglectsas wellthosehistoricalconfigurations ofactionthatshape
and transform pregivensocial structures in thefirstplace. A secondand
more satisfactory-butstill deeply problematic-approach is that of
structuralistinstrumentalism. Studieswithinthisperspectiveaccept the
prominentrole of social actorsin history,but ultimatelyconceptualize
theiractivityin narrowlyutility-maximizing and instrumentalforms.
And finally,themostsophisticated networkperspectiveon social change,
which we termstructuralist constructionism, thematizesprovocatively

1425

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

certainhistoricalprocessesof identityconversionand "robustaction."


It is themostsuccessfulofall oftheseapproachesin adequatelyconceptu-
alizinghuman agencyand the potentiallytransformative impactof cul-
tural idioms and normativecommitments on social action. However,
even thisperspectivefallsshortin understanding thefullcomplexities of
the theoreticalinterconnectionsamongculture,agency,and social struc-
ture. It too pays insufficientattentionto the structuring influencesof
culturaland politicaldiscoursesupon historicalactors.8

Structuralist
Determinism
All threeof these basic approachesare representedby practitioners of
bothrelationaland positionalnetworkanalysis.The aforementioned rela-
tionalstudyby Rosenthalet al. (1985), forexample,standsas an illumi-
natingcase studyin structuralist determinism. Rosenthaland herassoci-
ates delineatethreedistincthistoricalperiodsofwomen'sreformactivity
in New York State betweentheyears 1840 and 1914. The firstof these,
theyclaim, lasted from1840 to the late 1860s; it was dominatedby a
women'srightsmovementand also featuredhighlevels of involvement
in antislaveryand temperanceorganizations.The second periodwas a
transitionalone; betweenthe end of the Civil War and the late 1880s
therewas comparatively littleactivity,manyof theearliergroupsdisap-
peared, and "the possibilityof creatingnew nationalorganizationswas
limited"(Rosenthalet al. 1985, p. 1044). And finally,betweenthe late
1880sand 1914,therewas a resurgence ofintensereformactivity,center-
ing around an increasingnumberof new organizationslinkedprimarily
by thesuffrage issue. Rosenthalet al.'s delineationofthesethreeperiods
of women'sreformactivityby means of relationalanalysissurelyranks
as a significantand worthycontribution.But its limitationsare also
considerable:the studyprovideslittlesystematicexplanationas to pre-
ciselywhythesechangesoccurredfromone historicalperiodto thenext,
settlinginsteadfora successionof static"map configurations" or rela-
tional "snapshots"'of networkpatterns.The individualand social ac-
tionsthatled fromone structuralconfiguration of reformactivityto the
nextare leftunanalyzed,as are thedevelopmentsin social structure and

8
Our termstructuralistconstructionismis coincidentallyreminiscentof Pierre Bour-
dieu's phrase, "constructiviststructuralism"(Bourdieu 1990). But we do not imply
by this any direct connectionbetween the networkanalysts whom we are discussing
and Bourdieu, although it is true that these various thinkersall share an underlying
concern to overcome at both the theoreticaland empirical levels the dichotomybe-
tween "subjectivist" and "objectivist" standpoints. Bourdieu's understanding of
"fields," e.g., does bear strikinganalytical affinitieswith that of "social networks"
(Bourdieu and Wacquant 1992, p. 114).

1426

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

culturaland politicaldiscoursethat underlayand motivatedthem. At


best, Rosenthalet al. treatthesevariousdevelopmentsin theiranalysis
as exogenousvariables.
On theside ofpositionalratherthanrelationalanalysis,theearlywork
ofHarrisonWhiteand hisassociateson blockmodeling techniques(White
et al. 1976; see also Boormanand White 1976) providesanotherillumi-
natingexampleof structuralist determinism.9 White'sresearchincludes
studiesof networkdata describingstructuralchangesover time(White
et al. 1976, pp. 763-68); each of these studies,however(like that of
Rosenthalet al.), ultimately culminatesin a successionofstaticrepresen-
tationsof social structure.Each manages,thatis to say, to generate"no
modelsof processesover time." Whiteet al. recognizethattransforma-
tionsin social structurestillneed to be explained:"Models of structure
are not sufficient untothemselves.Eventuallyone mustbe able to show
how concretesocial processesand individualmanipulationsshape and
are shaped bystructure.... One fundamental problemhereis thatmany
settingsmay admitnot just a singleequilibriumoutcome,but multiple
alternativeequilibria. . . . In turnthe interesting questionsmay bear
on what externalforcesmay cause a social structureto pass fromone
equilibriumconfiguration to another"(Whiteet al. 1976,p. 773; empha-
sis added). Toward the end of anotherimportantarticleon blockmodel-
ing techniquesand role structures, Whiteand his collaboratorsfurther
acknowledgethat "the next analytictask is to provideways to probe
how role structures. . . actuallycome intobeing"(Boormanand White
1976, p. 1442). But again, theyprovideno systematic way of buildinga
concernforhumanagencyand processuality intotheirexplanations.
Anothershortcoming thatbothoftheseexamplesofstructuralist deter-
minismhave in commonis theassumptionthatsocial networkscan best
be conceptualizedas linkingtogether"concrete"entitiessuch as persons
and organizations,ratherthan as also embodyingideals, discursive
frameworks, and "cognitivemaps." The latter,fromtheirperspective,
remain mere "abstractions."'0Structuralistdeterminismrests analyti-
callyon a reificationofsocial relations;it transforms
theimportanttheo-
retical distinctionbetween a structureof social relations,on the one
hand, and culturalformations, on theother,intoan ontologicaldualism.
It thereby"ruthlesslyabstracts"the formalor "objective" dimensions
of social relationsfromtheirculturaland intersubjective contextsso as
to be able to representand analyze such relationswith sophisticated

9 We discussbelowtheratherdifferent thatWhitehas subse-


analyticalperspective
quentlyadopted-e.g., in Identityand Control(1992).
10 For a tellingexample,see Granovetter's
synopsisof CambridgeUniversity
Press's
serieson "StructuralAnalysisin theSocial Sciences"(Knoke1990,frontmatter).

1427

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

technicaltools; in the process,however,it drainssuch relationsof their


active, subjectivedimensionand theirculturalcontentsand meanings.
To compoundtheproblem,thisvariantofnetworkanalysisthentheoreti-
cally privilegesone side of this dualism-namely, that of "social rela-
tions"itself(albeitsocial relationsnow blanchedof boththeiractiveand
symbolicaspects)-over that of culturaland discursiveformations.It
uses social networks("social being,"in Marx's famousformulation) to
explain "social consciousness"and culture,but not (also) the otherway
around. Indeed, in Wellmanand Berkowitz'swords(1988, p. 5), "sym-
bols, meanings,and values . . . are a derivativeand oftenresidual
concern.)"

Instrumentalism
Structuralist
Perhaps because of its theoreticallimitations,structuralist determinism
has givenriseto relatively littleempiricalresearchon historicalprocesses.
Far more commonhas been the perspectiveof structuralist instrumen-
talism,which has also been adopted by practitioners of both relational
and positionalnetworkanalysis.This perspectivecertainlytakesthehis-
toricalrole of social actorsverycarefullyintoaccount.However,it also
draws implicitly upon "residualcategories"fromoutsideits own concep-
tual framework-in particular,a model of homo economicus-for ex-
plainingtheformation and transformation ofsocialnetworksthemselves.
Thereis a striking tendencyamongstructuralist instrumentalists,in fact,
to "smuggle"conceptionsof agency into theirinvestigations,whether
overtlyor covertly,fromthe domainof rationalchoicetheory.Many, if
notmost,such networkanalystsassumeunproblematically thatactors-
individualsand even groups or organizations-are utilitymaximizers
who pursuetheirmaterialinterestsin money,status,and powerin pre-
ciselythe ways predictedby theoristsof rationalchoice. In effect,these
analystsprojecttheirown anticategoricalism on the actorstheystudy,
neglectinghow the latter'sown culturaland moral categorieshelp to
structuretheirbeliefsand behaviors.
A usefulexampleof structuralist instrumentalism on the side of rela-
tional analysis are Roger Gould's aforementioned studies of the 1871
Paris Commune(1991, 1992). Gould beginsby "posit[ing]an influence
processin whicha [Parisian]district'sresistancelevel is a functionof a
set of exogenousvariables and of the resistancelevels of all the other
districts,weightedby the strength of its linkswiththem"(Gould 1991,
p. 721; emphasis in original).He investigatesthree such "exogenous
variables"thathelp to explainthesevaryinglevels of resistance:(1) the
levels of povertyin an arrondisement; (2) the percentageof skilledsala-
riedworkersresidingtherein;and (3) thepercentageofwhite-collar, mid-

1428

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

dle-classemployeeslocatedin any specificneighborhood (Gould 1991,p.


723). "Resistancewas stronger,"concludesGould, "in areas that were
poor and workingclass," althoughhis "expectationthat white-collar
and unskilledworkerswould playless prominent rolesin theinsurrection
is not supported"(Gould 1991, p. 725). Gould departs here fromhis
own anticategoricalapproach by relyingupon "exogenous"gradational
measuresofpovertyand occupationalcomposition-in additionto social
networks-in orderto make senseof themobilizationin each arrondise-
ment(see also Gould 1992, p. 727).
Not onlymustGould movefromrelationalto gradationaldata in order
to identifyaccuratelythe social base of the Paris Commune,but, more
troublingly, he neverprovidesa plausiblecausal accountas to whyPari-
sians would have riskedtheirlives forthe Communein the firstplace.
At timeshe seemssimplyto assume,dubiouslyin our view, thatworking
people (or, at least, working-classmales of 19th-centuryFrance)" would
automaticallyfightand die fora staterepresenting theirclass interests.
He predicts,forexample,thatwhite-collar employees,"who oftencame
frommiddle-classfamilies,"would have been less likelythanartisansor
workersto participatein theinsurrection (Gould 1991,p. 724). As noted
above, this expectationis unsupportedby the data, an outcomethat
Gould himselfnever explains. At othertimes, Gould emphasizesthe
importanceof neighborhood (and cross-neighborhood) loyaltyor solidar-
ity. In this account,he argues that Parisiansjoined the insurgencynot
because of class interestsper se, but ratherbecause of "social pressure"
fromneighbors:"Failure to participatein the insurgenteffortwas con-
struedas a betrayalof loyaltyto the neighborhoodand was sanctioned
accordingly"(Gould 1992, p. 748). This argument,however,fails to
explainwhyor how certaintiesand interactions amongneighbors,which
Gould does not specifyconcretely, generatedsuch a powerfulneighbor-
hood loyaltyin the firstplace; it begs the questionas to how certain
(unspecified)workerscame to believe,and apparentlymanaged to con-
vince others,that "neighborhoodloyalty"requirednothingless than
riskingtheirlives for the Commune. Gould's assumptionsabout the
purelyinstrumental foundationsofpoliticalmobilization-whetherclass
or statusbased-in the presentstage of his researchpreventhim from
analyzingsuch culturaland normativeinfluencesin a fullysatisfactory
manner.Gould impliesthatsimplybelongingto a neighborhoodwitha
certainoccupationalcomposition(and ties to othersuch neighborhoods)
producedpro-Communesolidarities.But whydid such "belongingness"
generatepowerfulsolidarityinsteadof,forexample,interpersonal indif-

" Goulddoesnotsaymuchaboutthecomplexlygendered
character
oftheinsurgency,
simplynotingthatwomenwerenotadmittedintotheNationalGuard.

1429

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

ferenceor even cutthroatcompetition, as amongthe oligarchicclique in


Padgettand Ansell'sstudy?
We recognizethatGould's workhas thusfarappearedonlyin journal
articlesand thatlater,moreexpansiveversionsmightwell exemplifya
different sortof explanatorystrategy.In its presentform,however,we
claim that it does relativelylittleto explorethe specificallynormative
commitments of the social actorsengagedin politicalresistanceduring
the Commune.(The difficulty hereis not that Gould neglects"culture"
per se, but ratherthattheunderlying logicofhis argument-itstheoreti-
cal logic-fails to accord normativecommitments any independentex-
planatorysignificance.) In a morecompletehistoricalexplanation,Gould
would have to directfarmoreattentionthan he does to the specifically
culturalbases of (cross-)neighborhood solidarityand theirinfluenceupon
individuals'projectsof action. Exactlywhat sortsof practicesand ritu-
als, he would need to ask, producedthepowerfulsolidaritiesof Parisian
neighborhoodsand the National Guard? What role did politicalideolo-
giesand culturaldiscoursesplayin sustainingor evenexpandingrelations
of solidarity?(Gould mentionsthe discoursesof "socialism"and of "re-
publican patriotism"in thiscontext,but he neverexaminestheirmean-
ingsto workersand thustheircausal significance in any systematic fash-
ion.) And finally,what do sourcesofinformation such as diaries,letters,
unionrecords,and journalisticaccountsreveal to us about popularcul-
tural practicesin 19th-century Paris and duringthe Commune itself?
(See, e.g., Sewell [1980]and thesourcescitedin Edwards [1973,p. 175].)
Such data sourceswould demonstrate how popularpracticesand popular
culture,farfrombeing"additional"factorsor forcesneedingto be either
"controlledfor"or examinedalongsideof networkstructures,were in
factimportantdimensionsof thosestructures themselves-groundedin,
and sustaining,specific,identifiable social ties.
On the side of positionalanalysis,PeterBearman (1993), too, adopts
the perspectiveof structuralist instrumentalism. Bearman tellsthe story
of changingelite social relationsin England duringthe "long sixteenth
century"by means of a sequence of blockmodelrepresentations of elite
social structure.By examiningtheseblockmodels,he specifies"the slow
and arduousprocessby whichreligiousheterodoxy was embeddedin the
fabricof local elite social and politicallife,"the tangiblemechanisms
"by whichlocal elitescame to perceivethemselvesand othersas actors
whose activitywas of religioussignificance" (Bearman 1993, pp. 171,
132). Bearman's approach is not dissimilarto what Max Weber (1949)
long ago termedthe analysis of "electiveaffinity"-thestudyof how
particulardiscoursesand culturalformationscome to finda "match"
and to "resonate"withspecifichistorically embeddedactors.Wherehis
approach differsfromthatof Weber,however,is in its tacitinstrumen-

1430

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

talism-that is, itstendencyto devotealmostall ofitsanalyticalattention


to uncoveringthe "structuralpreconditions"for this elective affinity,
ratherthanto also exploringtheindependentcausal significance of these
discursiveframeworks themselves.Indeed, Bearman seems to attribute
littlemorethan purelymaterialinterests(in money,status,and power)
to the historicalactors at the centerof his account. "The eclipse of
localism,the decline of kinship,and the emergenceof a national elite
subworld,"he argues,"were aggregateoutcomesof the gentry'spursuit
of local statusand power"(Bearman 1993, p. 3). The searchfor"politi-
cal advantage"was the drivingmechanism,in otherwords,behindthe
crucial historicaltransformationsof the period-and not (also) the be-
liefs, values, and normativecommitments of these elite actors them-
selves. Bearmanmightwell have noted,in theopeningpage ofhis study,
that he is primarilyinvestigatingonly one side of "the interactionbe-
tween structuraland ideationalprocesses. . . the complexrelationship
betweenaction and structure"(Bearman 1993, p. 1).

Structuralist
Constructionism
Severalnetworkanalystshave, in recentyears,developedmoresophisti-
cated approachesthantheseto studyinghistoricalprocesses,approaches
that take into account cultureand agencyas well as social structure.
Again, practitioners of bothrelationaland positionalanalysishave pur-
sued suchinvestigations. One revealingexampleofstructuralist construc-
tionismon theside of relationalanalysisis Doug McAdam's recentwork
on Freedom Summer(1986, 1988). McAdam explicitlyrejectsmany of
the instrumentalist claims of othernetworkanalysts.He qualifiesthe
notion,specifically,that "structuralavailability"for social movement
participationrenders"attitudinalaffinity" to a movementcompletely
irrelevant.Such a notionmighthold true of "low-risk/cost activism,"
but"participation in instancesofhigh-riskactivism[suchas theFreedom
Summerprojectand, we mightadd, theParis Communeand theEnglish
Civil War]would appear to dependon an intenseattitudinaland personal
identificationwiththe movement"(McAdam 1986, p. 73). Surely,high
were requiredforthe FreedomSummervol-
levels of such identification
unteersto aspireto participatein such a demandingand potentially dan-
gerousundertaking; moreover,thevolunteers'accountsoftheirown mo-
tives in open-endedapplicationquestionnairesclearlydemonstratethat
theyfelta deep-seatedidealismand a strongcommitment to theproject's
goals. "The real questionis: Were the volunteers'priorattitudessuffi-
cientin themselvesto accountfortheirparticipation? My answerhereis
a qualifiedno.... Attitudinalaffinity [as well as] biographicalavailabil-

1431

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

itymustbe considerednecessarybutnotsufficient causes ofparticipation


in high-risk/cost activism"(McAdam 1986, pp. 73, 87).
Exactly where, then, did this "attitudinalaffinity" forthe Freedom
Summerproject(or, moregenerally,forany social action)actuallycome
from?One of McAdam's major innovationsin FreedomSummer(1988)
is to elaboratean implicittheoryof identityconversionthattakes seri-
ously the formationof motivationsand identitieswithoutsacrificing at
all the momentof "structurallocation." McAdam argues that the key
to his accountlies in thoseorganizationsthatdrew "the applicantsinto
civil rightsactivitybefore FreedomSummer.... Extremelyrisky,time-
consuminginvolvementssuch as FreedomSummer,"he contends,"are
almostalways precededby a seriesof safer,less demandinginstancesof
activism.In effect,people committhemselvesto movementsin stages,
each activitypreparingtheway forthenext"(McAdam 1988,pp. 50-51;
emphasisin original).Individualsfirstcomeintocontactwithmovement
participantsand engage in discussionsand joint activitieswith them,
confronting firsthand theissuesthatthemovementhas setout to address
and gainingin theprocessa deeperunderstanding of and moralcommit-
mentto its goals. Then, eventually,"at thelevel of identity," theybegin
to " 'play at' and [to] grow morecomfortablewith the role of activists
themselves"(McAdam 1988, p. 51). Unlike many othernetworkana-
lysts,McAdam recognizesthatactorscan undergofar-reaching processes
ofidentityformation in thecourseoftheirinvolvements in extraordinary
affairs.Such an insightis especiallyimportantto bear in mind when
analyzingtheirparticipationin "high-risk/cost" activitiessuch as Free-
dom Summer-or, for that matter,in any major social or political
movement.
In a recentpublication,McAdam concludesthat"networktheoryfails
to offera plausiblemodelofindividualactionand therefore a convincing
mechanismby whichinterpersonal contactsand organizationalmember-
ships draw individualsinto activism"in the firstplace (Friedmanand
McAdam 1992, p. 160). Friedmanand McAdam claim insteadthatnet-
worktheorycan yieldrobustexplanationsof collectiveactiononlywhen
synthesizedwith a modifiedrationalchoice model of individualaction,
one that views collectiveidentities(and not just materialresources)as
potentiallypowerfulincentivesfor action: "One of the most powerful
motivatorsof individualaction," theywrite,"is the desire to confirm
throughbehavior a cherishedidentity.. . . Integrationinto [activist]
networksmakes it morelikelythattheindividualwill value theidentity
of 'activist' and choose to act in accordance with it" (Friedman and
McAdam 1992, pp. 169-70). It is our contentionthat the distinctive
contributionof McAdam's work is, indeed, to expand the conceptof
purposiverationality itselfto the pointof burstingthroughthe seams of

1432

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

standardrationalchoicetheory.It is historicalactors'specifically norma-


tivecommitments, ratherthan(or in additionto) theirpursuitofmaterial
goals, thateffectively drivestheirsocial movementparticipation.
If thereis a weaknessto McAdam's analysisof FreedomSummer,it
lies in his insufficientattentionto preciselythis elementof normative
commitment to cherishedideals. In one of theirarticlesFernandezand
McAdam (1988) notethatrecruitment contexts,"theresidueof a protest
culture,"affected"thenumberand formof interactions amongpotential
recruits,"and thereby"exert[ed]an importantinfluenceon all the pro-
cesses involvedin . . . recruitment" (Fernandezand McAdam 1988, p.
379). McAdam devotesseveralpassages in his book (althoughnot in his
articles)to explainingtheoriginsoftheserecruitment contextson theeve
of FreedomSummer.He discusses,forexample,theroleofdemographic
and economicdevelopmentsin producinga youthcohortdrivenby an
exaggeratedsense of its own culturalimportanceand potencyand the
role of the liberalizationof domesticpoliticsunderthe leadershipof fig-
ures such as JohnF. Kennedyand MartinLutherKing, Jr. (McAdam
1988, pp. 13-24). And yetMcAdam stopsshortof providinga satisfac-
toryaccountofhow establishedcurrents ofAmericancultural(and politi-
cal) discoursewere selectivelydrawn upon by civil rightsleaders and
othersand refashionedinto a new and powerfuldefinition of the situa-
tion.It was thecompellingnatureof thisculturaland politicaldefinition
thatdrewfreshrecruitsto thecivilrightscause in thefirstplace, and that
helpedto createthoseverynetworksthemselvesthatMcAdam regardsas
the startingpointof his analysis.McAdam mighthave provideda more
convincingaccountthanhe does oftheculturaland politicalidiomsofthe
day-not all of which,afterall, weresupportiveof racial equality-and
examinedhow and why certainof themcame to have such a powerful
resonanceforso manypeople,especiallyyoung,relativelyaffluent college
students,preciselyat thatspecifichistoricaljuncture.
On theside ofpositionalanalysisusingblockmodeling techniques,Pad-
gettand Ansell's(1993) workon the FlorentineRenaissanceservesas a
usefulexampleofinnovativeresearchon networks,culture,and agency.
Padgettand Ansell employtwo importanttheoreticalideas to help ex-
plain theriseof theMedici familyin early15th-century Florence:"struc-
tural channelingof learning"and "robust action." All of the major
figuresin theiraccount, theyargue, were highlyactive and dynamic
players.Cosimo de' Medici, forexample,steeredhis familyfroma posi-
tionofabject defeatin Florentinepoliticsto one ofnear-totalascendancy
duringa periodof no morethan 40 years. And yet,claim Padgettand
Ansell, Cosimo did not pursue fromstart to finishsome omniscient
"grandstrategy";rather,he shrewdlyand opportunistically took advan-
tage of the local "openings"that a successionof exogenouseventshad

1433

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

fortuitously broughthis way. Specifically,the reconsolidation strategies


of the oligarchsafterthe Ciompi uprisingof 1378 had leftopen several
important"structuralholes" in theirmarriagenetwork,one around the
Medici themselvesand anotherarounda different segmentoftheoligar-
chic elitewhichwas based in the Santo Spiritoquarterof Florence.The
Medici learnedthatthe latter"breachin the oligarchs'defenses"could
be successfullyexploitedthrougha focusedmarriagestrategy.In addi-
tion, a "second structuralchain reaction"was made possible by the
oligarchs'self-enclosure fromthe nonpatriciannew men. This pattern,
togetherwithneighborhood-based policies oftax extractionduringa pe-
riod of financialcrisiscaused by foreignwars, led to defensivealliances
of the new men firstwithone anotherin theirown neighborhoodsand
theneventuallywithlocal neighborhood patronsamongtheelite-again,
the Medici themselves.Thus, "theMedici partygrewup . . . fromraw
networkmaterialunintentionally channeledto them.... Only verylate
in thegame . . . did [theMedici]adaptivelylearnofthepoliticalpotential
ofthesocial networkmachinethatlay at theirfingertips. In almostHege-
lian fashion,[the]oligarchscreatedthe networksof theirown destruc-
tion" (Padgettand Ansell 1993, p. 1287).
The Medici were able to exploitthe possibilitiesbroughttheirway
throughsuch processesof "networkcascading" by virtueof theirown
structurally anomalous positionwithinthe overall Florentinenetwork.
Involved in several different "games" at once-pursuing higherstatus
throughmarriagestrategiesand simultaneously pursuingmonetarygain
and patronageinfluencethroughneighborhoodeconomiccontacts-the
Medici soughtaftera multiplicity ofnotalwayscompatiblegoals. Having
such genuinelymultivocalinterests,they appeared inscrutablebefore
theirvariousfollowers(thenew menand patricians)and outmaneuvered
their"opponentsinto the forcedclarification of their(but not [the Me-
dici's])tacticallines of action"(Padgettand Ansell 1993, p. 1264). The
key to Cosimo de' Medici's styleof robust action, note Padgett and
Ansell,was "maintainingdiscretionary optionsacross unforeseeablefu-
turesin the face of hostileattemptsby othersto narrowthoseoptions.
. . .Victory meanslockingin others,but notyourself[Cosimo],to goal-
orientedsequences of strategicplay that become predictablethereby"
(Padgett and Ansell 1993, pp. 1263-64). In this respect,Padgett and
Ansellecho Eric Leifer'sconceptionof"local action"as thosemovesthat
allow actorsin "face-to-face competitionwith otherswho have similar
credentials. . . to avoid claiminga (global) role untilthereis evidence
[thatsuch a role] will be conferred"(Leifer1988, pp. 865, 866). They
echo as well Charles Tilly's conceptualizationof "contentiousaction,"
the resourcefulness of individualsand groupswho "performin dramas
[repertoiresof contention] in whichtheyalreadyknowtheirapproximate

1434

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

parts,[but]duringwhichtheynevertheless improviseconstantly"(Tilly
1992, p. 15; 1986).
It is importanthere not to confusePadgett and Ansell's notion of
robustactionwithrationalchoiceconceptualizations of instrumental ac-
tion. It is true, of course, that Padgett and Ansell themselvesspeak
repeatedlyoftheMedici's pursuitofmoney,status,and power. But they
also take special pains to point out that none of these ends of action
makesanysenseat all outsidethetermsoftheculturalcategories,values,
and beliefsprevailingin Florentineelitesocietyat thatparticularjunc-
tureof history.Moreover,theyquite explicitlydistancethemselvesfrom
the Machiavellian presuppositions of game theoryby emphasizingthe
"mutually adaptive learning processes" and "bounded rationality"
characteristicof elite conflicts,especially during "tumultuoustimes"
such as those of the Milan and Lucca wars (Padgettand Ansell 1993,
pp. 1301-2; see also Padgett 1986). During such momentsof "compli-
cated chaos," theysuggest,"the games themselvesare all up forgrabs.
Rational choice requiresa commonmetricof utilityforfooting,but re-
vealed preferences (the basis forinferringtrade-offs across goals/roles)
only exist post hoc. . . . Clear goals of self-interest. . . are not really
featuresofpeople;theyare Florentine(and our)interpretations ofvarying
structures of games" (Padgettand Ansell 1993, pp. 1307-8).
If thereis a major deficiencyto the Padgettand Ansell account, it
lies elsewhere:in theirclosingdiscussionof the Medici party'sultimate
accessionto statepower.Padgettand Ansellfailto explainthereprecisely
why Cosimo de' Medici came in the end to be considerednothingless
than the "pater patriae," the fatherof his country,by so many of his
contemporaries or why he was installedin power with the supportof
"those Florentineswho have remainedon the marginsof our account
thusfar-the politicalneutrals"(Padgettand Ansell 1993,p. 1308). The
key to the problem,suggestthe authors,is containedin "the cognitive
category'oligarch.'. . . When the oligarchswere firmly in control,they
were not labeled 'oligarchs';theywere republican'public citizensof the
state.' Loss of legitimacyand Medici victoryare what got themtheir
pejorativetag. No longerpublic-spirited and selflessin attribution, they
came to epitomizeclass self-interest in Florentineeyes" (Padgett and
Ansell 1993, p. 1308). It was when the rulingfamilies,squeezed by the
fiscalcrisisbroughton byforeignwars,in turnsoughtto repressthenew
men that theyearned the epithet"oligarchs."Clearly,this was a most
significantmomentin thehistoryrecountedby Padgettand Ansell;with-
out it, theMedici, despiteall of theirtacticalmaneuverings, quite possi-
bly would neverhave takenpowerat all. And yet,thiscrucialaspect of
thestory-the evidentsuccessoftheMedici in manipulating towardtheir
own ends the very contentof such key termsas "public-spiritedness"

1435

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

and "self-interest"-isleftcompletelyout of the pictureuntilthe final


pages of the study.Moreover,the authorsneverprovidetheirreaders
with any analysis as to why such issues as public-spiritedness, self-
interest,and corruption mighthave had so muchmeaningforFlorentines
to begin with. Could the discursiveframeworkof civic republicanism
(Baron 1966; Pocock 1975), virtuallyunmentionedin this study,have
profitably been broughtintothe storyat preciselythisjuncture?(Might
still other culturaland politicaldiscoursesalso prevalentduringthat
historicalperiodhave emphasizedsuchthemes?)It was civic republican-
ism,afterall-by itselfor, morelikely,in complexinteraction withother
discursiveformations-thatmade possiblethat special combinationof
deep concernover social statusand hatredof class interest(as opposed
to civic virtue)that togetherallowed the Medici successfully to engage
in robustactionin the firstplace and ultimatelyto seize statepower.

RETHINKING THE PROBLEM OF AGENCY


Social Networksand CulturalNarratives
Clearly,then,none of the threedifferent networkapproachesthat we
have been examiningoffersa completely satisfactory
approachto histori-
cal explanation.Whileeach of thesethreemodelsdoes representa more
nuanced understandingof the complex interrelationships among net-
works, culture,and agency than the one preceding,none completely
succeeds by itselfin addressingall of the difficult
issues at hand. The
modelof structuralist forone, featuresa successionof net-
determinism,
work "snapshots"of social structure,while neglectingaltogetherthe
potentialcausal significance of symbolicand discursiveformationsand
offering fewinsightsintotheconcretehistoricalmechanismsleadingfrom
one such networkconfiguration to another. Structuralistinstrumen-
talism,by contrast,clearlyacknowledgestheexplanatorysignificance of
social action,but, on theotherhand, conceptualizesthedeterminants of
such action in excessivelynarrowterms,oftenrelyingon unwarranted
assumptionsabout the overridingimportanceto historicalactors of
money,status,and power.And finally,structuralist constructionism af-
firmsthepossibilitythatactors'goals and aspirationsmightwell be com-
plex, multivalent,and historicallydetermined;it inquires,forexample,
intosuch intricateprocessesas identityconversion,structural channeling
of learning,and flexibleopportunism.And yet not even this model, at
least as it has been elaboratedby networkanalyststo date, fullyrecog-
nizesthe(potentially) autonomouscausal significance
ofculturalor politi-
cal discoursesin shapingthe complexeventsequencesthatit examines.
In recentyears, by far the mostsignificant theoreticalsteps takenin

1436

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

the alternativedirectionthatwe are proposinghave been the important


writingsby HarrisonWhiteon identity, temporality,and narrative.For-
merlya leading exponentof structuralist determinism, White has now
begun to argue that the interrelationships among networkstructures,
culture,and agencythemselvesneed to be reconceptualized.In Identity
and Control(1992), he suggeststhat "agency is the dynamicface of
networks,"agency understoodnot only as a "by-productof control,"
but also as "ways of . . . upend[ing]institution[s] and . . . initiat[ing]
freshaction" (White 1992, pp. 315, 245). Significantly-andperhaps
surprisingly, givenhis earlieranticulturalist bias-White includesa the-
oryof culturalsymbolsand discoursesin this new versionof his social
theory;he now takes discursive"narratives"and "stories"to be among
the key featuresof social life. "Social networks,"he asserts,"are phe-
nomenologicalrealities,as well as measurementconstructs.Storiesde-
scribethe tiesin networks.. . A social networkis a networkof mean-
ings" (White 1992, pp. 65, 67).
White'sdiverseconceptualizations of "identity"play a prominentrole
in this new social networktheory."Identity,"claims White, signifies
"any source of action not explicablefrombiophysicalregularities,and
to which observerscan attributemeaning"(White 1992, p. 6). Four
distinctsenses or levels of identity"come wound togetherin the same
constructedreality."The firstof these is identityas a "primordialand
continuingurge" for "secure footing"in "an otherwisechaotic social
world." The second is identityas a "social face," the basis "for our
everydayconstructforthe person. . . as an actorin a role [supplying]
preferences that may guide him or her toward goals, and into rational
action." The thirdis "identityfromfrictions and errorsacross different
social settingsand disciplines . . . aris[ing]exactlyfromcontradictions
across disciplines, from mismatchesand social noise." And the fourthis
"identityas more-or-less coherentaccounts,as biography. . . afterthe
fact as presented accountswhichmaybecomewovenintosomeunique
in
narrativestory"(White 1992, pp. 315, 312-14; emphasisadded).
Despite its obviousimportanceforsocial networktheory,what is per-
haps moststrikingabout White'snew approachare two closelyrelated
omissions.First,Whiteneverexplainspreciselywhyactorsor identities
engage in these "contendingcontrolattempts"in the firstplace (as in
his remarkson the firstlevel of identity).It is clearlyinadequate to
explain away thesecontrolprojectssimplyas some ahistorical"primor-
dial and continuingurge" for"securefooting."And second, White ne-
glectsto analyze closelythe role of culturalidiomsand normativecom-
mitmentsin helping to shape the very identitiesand aspirationsof
historicalactors.Indeed, he devotesverylittlespace at all to exploring
theinternalstructure and patterning ofthesesymbolicformations. White

1437

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

begins his book by declaringthat he will be "focus[ing]. . . upon the


purelysocial" (1992, p. 14) as distinguishedfromtheculturallevel. This
is a "perilousundertaking," as he himselfclearlyrecognizes(White1992,
p. 14n.20). White accordinglyrelegates"narratives"and "stories"to a
secondarypositionin his theory,in favorof moresocial structuralpat-
ternssuch as networksof social interaction(his second and thirdlevels
of identity,e.g., are both constitutedby actors' locations withinrole
structuresand disciplines).Ultimately,thesenarrativesand storiesfail
to receive,even in the fourthand finallevel of identity(constitutedby
"accounts . . . afterthe fact"),the sort of carefulanalyticalattention
thatWhiteaccordsto social networksthroughout.
It is our contentionhere, by contrast,that networks,culture,and
agency relate to one anotherin ways that not even White, nor even
the structuralistconstructionists whom he mostcloselyresembles,have
adequatelyconceptualized.Let us not be misunderstood on this crucial
point. White is certainlycorrectin assertingthat cultural(as well as
political) discourses do inform-and are deeply embedded within-
networkpatternsof social relationships.Social networksare, indeed,
"phenomenologicalrealities,"as Whiteputs it-or "networksof mean-
ing." Culture and social relationsempiricallyinterpenetrate with and
mutuallyconditionone anotherso thoroughly thatit is well-nighimpossi-
ble to conceiveof theone withouttheother.This is the respectin which
culturecan, indeed, be said to constitute,in Charles Tilly's felicitous
formulation, nothingless than the very"sinews"of social reality.12
And yet thereis another-and itselfno less critical-sense in which
culturaldiscourses,narratives,and idiomsare also analyticallyautono-
mouswithrespectto networkpatternsofsocial relationships. These sym-
bolicformations have emergentproperties-an internallogicand organi-
zationoftheirown-that requirethattheybe conceptualizedas "cultural
structures"(Rambo and Chan 1990; Barber 1992) analyticallyseparate
fromsocial structure."3 "When theyare interrelated," noteJeffrey Alex-
anderand PhilipSmith,"symbolsprovidea nonmaterialstructure.They

12 An important ofthiscan be foundina recentarticlebyMargaretSomers


illustration
(1993)on thedevelopment of citizenship
practicesin 18th-centuryEngland.Somers
demonstrates therehow different typesofpoliticalculturewereembodiedin distinc-
tive"relationalsettings,"or "patternedmatri[ces]ofinstitutional
relationships
among
cultural,economic,social, and politicalpractices"(p. 595). Politicalculture-and
citizenshippracticesthemselves-hardly stood apart frompatternsof familylife,
socialgeography, and "theinstitutional
dynamics oflaw, governance, and administra-
tive structure"(p. 603). Citizenship,as Somerspointsout (following Karl Polanyi
[1957]),is bestconceivedof as an "institutedprocess."
13 One couldalternatively use theterm"culturalframes"hereas well(see,e.g., Snow
et al. 1986; Snow and Benford1988;and the variousarticleson "collectiveaction
frames"in Morrisand Mueller[1992]).

1438

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

representa level of organizationthatpatternsactionas surelyas struc-


turesof a more visible, materialkind" (Alexanderand Smith 1993, p.
156; emphasisadded). This is an insightthatgoes back at least as faras
the classical Parsoniandistinctionbetweenculturaland social systems
(Parsons and Shils 1951; see also Sorokin1947); indeed,it originatesin
thelaterDurkheim,who in his religioussociologywas thefirstto under-
score the internallogic of systemsof symbolicclassification (Durkheim
1965;Durkheimand Mauss 1963;Douglas 1966;Alexander1988b,1988c;
Kane 1991; P. Smith 1991; Alexander and Smith 1993; Emirbayer
1993).14 Those who, like Whitein his morerecentwork,stressthe utter
inseparability of cultureand social structurecome close to denyingthis
all-important theoreticalinsight.They veer offin the direction,in fact,
of what MargaretArcherhas termedthe fallacyof "centralconflation":
the assumptionthat because cultureand social structureare mutually
constitutive,"thereis no way of 'untying'the constitutive elements.The
intimacyof theirinterconnection denies even relativeautonomyto the
componentsinvolved. . . . In the absence of any degreeof autonomyit
becomes impossibleto examine theirinterplay"(Archer1988, p. 80;
emphasisin original).15Examplesofculturalor discursivestructures that
need to be analyzedinternally (as well as in theirinterplaywithnetwork
structures)includethe civic republicanismof Padgettand Ansell'sFlor-
ence, the civil rightsdiscourseof FreedomSummer,and the socialism
and republicanpatriotismof the Paris Commune.
Recognitionof theanalyticalautonomyof culturalstructures certainly
does not necessitatea returnto "values-based"sociology.We are not
speakinghereof mere"norms"and "values," as Parsonsand his follow-
ersdid, but ratherofmuchlargersymbolicformations such as discursive
frameworks and culturalidioms.(For a discussionofhow thisperspective
pointswell beyondthe limitationsof functionalist "value analysis,"see

14 Another approach,thatof "narrativeanalysis,"also inquiresintotheinnerlogic


ofculturalstructures. See Steinmetz(1992)and Somers(1992).As Steinmetzsumma-
rizes it, thisapproachholds that"one . . . shouldpay special attentionto [such
elements as] thecentralsubjectand actors,theformoftheplotand itsrelationto the
story,therulesforexcluding eventsfromthenarration,theturningpoints,repetitions,
and 'fillingin.' One shouldask whether a givenhistory
assumestheformofa complete
historical narrative, ofannals,orofchronicles.Finally,oneshouldidentifythenarra-
tor,the actorsin the story,and theexplicitor implicitaudience"(Steinmetz1992,
p. 501). Such a programfornarrativeanalysisis notnecessarily incompatiblewith
late-Durkheimian approachesto culturalinquiry.
15
White'sown workcan be said to vacillatebetweencentralconflation and what
Archer(1988) terms"upwardconflation," which,like its counterpart,
"downward
she definesas a varietyof epiphenomenalism.
conflation," The formerdenotesfor
Archeran analyticalprivileging
ofthesocialstructural
realm,whilethelatterentails
a reductionto theculturaldomain.

1439

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

Alexanderand Smith[1993].) Nor do theseinsightsnecessarilylead in


the directionof a "reification"of culture,a renderingof culturalframe-
works as "concretesocial entities"with theirown appropriatesets of
institutions,rules, and resources.There is a significant differencebe-
tweenconceivingof culturalnarratives,idioms,and discoursesas sym-
bolic patternspossessingtheirown autonomousinnerlogic, on the one
hand, and thinkingof themas substantively distinct"domains"of social
life,on theother-surelya "fallacyofmisplacedconcreteness," in Alfred
NorthWhitehead'smemorablephrase(see Goodwin 1994). And finally,
in speakingofculturalstructures we are notsimplyaddinganother"vari-
able" to our complexexplanatoryequation,as ifculturewereitselfnoth-
ing morethana residualcategoryto be broughtin afterthefactin order
to completea task that social structuralanalysis alone had failed to
accomplish.(For two compellingcritiquesof the "variables" approach
in general,see Blumer1969, chap. 6; Abbott1992c.) The pointwe wish
to make is perhapsarticulatedbest by Theda Skocpol,who affirms that
"it [does] makea difference whichidiomor mixtureofidiomsis available
to be drawnupon bygivengroups.Indeed,theverydefinitions ofgroups,
theirinterests,and theirrelationsto one anotherwill be influencedby
culturalidioms"(Skocpol 1985, p. 91; emphasisadded).
Whyis it so important, then,to thinkofsymbolicformations as ifthey
wereanalyticallyautonomousculturalstructures? We proposethatthese
culturalformations are significantbecausetheybothconstrainand enable
historicalactors,in muchthe same way as do networkstructures them-
selves. Culturalstructuresconstrainactors,to begin with,by blocking
out certainpossibilitiesforaction,as, forexample,by rendering it incon-
ceivablefortheoligarchsof 15th-century Florenceto have pursuedmar-
riage ties with nonpatriciannew men, even when it mighthave been
materiallyadvantageousforthemto do so. Eviatar Zerubavel(1991) has
referredto these cognitivecategorizationsphenomenologically as "is-
lands of meaning"-an interesting analogue,certainly,to Ronald Burt's
(1992; and Padgettand Ansell'sown) conceptof "structural holes." Cul-
tural structuresalso constrainactors by preventingcertainarguments
frombeingarticulatedin publicdiscourseor,oncearticulated,frombeing
favorablyinterpreted by othersor even understood(Swidler1987). It is
oftenunder theiraegis, moreover,that contendingsocial groupswage
their cultural and political battles. Each of these contendinggroups
"sketch[es]out a different blueprintfroma commonset of [ideological]
principles,. . . out of the same terminology and the same essentialset
of concepts"(Sewell 1985, pp. 74, 76; see also Sewell 1980; Skocpol
1985, p. 89). Alexandertermsthis a contextof culturalrefraction: "In
this situationdifferent interestshave been refractedthroughthe same
culturallens" (Alexander1988a, p. 155).

1440

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

But culturalformations also enablehistoricalactorsin diverseways-


forexample,by orderingtheirunderstandings of thesocial worldand of
themselves,by constructing theiridentities,goals, and aspirations,and
by renderingcertainissuessignificant or salientand othersnot. (Indeed,
by constrainingactors' possibilities,these culturalformationsalready
"enable" them as well, since, as Niklas Luhmann [1982, 1990] has
pointedout, the "reductionof complexity"servespreciselyto enhance
the range of alternativesopen to individualand/orcollectiveactors.)'6
Symbolicpolaritiescrystallizewithinculturalstructures, dividingsocial
and metaphysicalrealityinto such antitheticalcategoriesas "pure" and
''polluted,'""just" and "unjust,'" and "sacred" and "profane.'"Such
categoriesprovidethe groundworkfornormativeevaluationsas well as
forguidelinesforaction,as in the case of RenaissanceFlorence,where,
as we have seen, the decisive momentoccurredonlywhen the Medici
had succeededin becomingidentified in the mindsof politicalneutrals
as "pure," "just," and "public-spirited," while the oligarchshad come
to be seen as exemplarsof political"impurity," "corruption," and "self-
interest."Under such circumstances,certainidentities,interests,and
coursesof actioncome to be morevalued thanothers,to thepointwhere
individualsand groupsoftenpreferto sacrificetheirown materialinter-
ests out of a deep-seatedcommitment towardthem,as in the cases not
only of the Medici's ascensionto power, but also of the English Civil
War, the Paris Commune,and the civil rightsstrugglein the United
States (see also Calhoun 1991). Strugglesto redefinethe culturaland
symbolicdefinition of such situations,and in so doingto identify certain
actors (and typesof action) with purityand sacralityand otherswith
impurityand pollution,constitute one of themostimportantdimensions
of social conflict(see Emirbayer1992a; 1992b). PierreBourdieu'stren-
chant insightsinto the dynamicsof classificationstrugglesare helpful
here in bringingthese conflictsinto sharperfocus(Bourdieu 1984, pp.
466-84; Bourdieuand Wacquant 1992; see also Emirbayer1993).
Culturalstructures are, then,bothconstraining and enablingof social
action. They are also, in conclusion,multipleand interpenetrating. As
Skocpol puts it, "Multiple culturalidioms coexist, and theyarise, de-
cline,and intermingle in temposthatneed to be exploredby intellectual
and socioculturalhistorians"(Skocpol 1985, p. 91; emphasis added).
There has probablyneverbeen but one overarchingculturalidiom,nar-
rative,or discourseoperativein anygivenhistoricalcontext.A fullexplo-
rationof a case such as that of the rise of the Medici would thus have
to featurenot onlya carefulconsiderationof the internallogic of civic

16 We are grateful
to AnnMischeforthisobservation.

1441

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

republicanism, but also an inquiryintootherpopularidiomsoftheday-


and of how these possiblyinteractedwith (or stood in tensiontoward)
civic republicanismitself.Social actors nearlyalways findthemselves
withina cultural"environment" (more on this specificconceptbelow)
markedby a richpluralityof culturalformations; while theremightin
forarguingthatone or anotherof
certaincases be historicaljustification
theseis primary,the questionitselfcan onlybe resolvedthroughcareful
empiricalinvestigation.

Human Agencyand Social Action


This last setofconsiderations allowsus now to shiftourfocusofattention
away fromthe analyticalautonomyof culturalformations-themain
topicoftheprecedingsection-onto thequestionoftheirconcreteinterre-
lationshipwithother(network)structures and withthesocial actorsthat
reproduceand transform them. This lattertopic is surelyeverybit as
significant as the former.For as Ann Swidlerhas pointedout, "It is ...
the concretesituationsin which . . . culturalmodels are enacted that
determinewhich take root and thrive,and which wither and die"
(Swidler1986,p. 280). Unfortunately, many-if notmost-cultural soci-
ologiststodayfailto addresssuchissues,or even engagethemin thefirst
place, at least in a systematicfashion.All too often,theytake cultural
formations as unproblematically reflectingthebeliefsand assumptionsof
certaincategoriesof individuals,includingentiresocial classes, nations,
and even genders,withoutrecognizingthattheseformations themselves
have a relationalcharacterand are groundedin specificconcretesettings.
There moreoverremainsthe questionas to how structures of all types
-cultural as well as societal-interrelatewith social action itselfand
withthe verypotentialforhumanagency.These questionsrequirethat
we consider,in turn,the influencethatculturaland societalformations
have upon social actorsand thetransformative impactthatsocial actors,
fortheirown part,have upon culturaland societalstructures.
We beginbyaffirming thatWhiteand thestructuralist constructionists
we have examinedare surelyon therighttrackin theircommonemphasis
upon the role of human agencyin history.Networkanalystssuch as
McAdam, Padgett,and Ansellquitecorrectly stressthevolitionalaspects
of social life, the capacity of social actors to transformas well as to
reproducelong-standing structures,frameworks, and networksof inter-
action.In thisessay,we ourselvesholdto sucha view. In ourunderstand-
ing, human agencysignifiesthatmomentof freedom-or of "effort," as
Talcott Parsonstermedit (1937)-that existsas an analyticaldimension
of all actual empiricalinstancesof social action. Human agency,as we
conceptualizeit, entailsthe capacityof sociallyembeddedactorsto ap-

1442

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

propriate,reproduce,and, potentially, to innovateuponreceivedcultural


categoriesand conditionsof action in accordance with theirpersonal
and collectiveideals,interests, and commitments (Emirbayerand Mische
1994). We hope to have shownabove whya recognition of thiscapacity
forhumanagencyis criticalto any adequate attemptat historicalexpla-
nation.
On the otherhand, it is also our implicitunderstanding in thisarticle
thatsuch a capacityforhumanagencydoes not-and shouldnot-mean
preciselythe same thingas "social action"per se. While the momentof
"effort" or agencyis presentin all empiricalunitsof action,thisdimen-
sion mustbe understoodas an analyticalmomentonly.The "identifica-
tionof actorand agency,"writesJeffrey Alexander,rendersone "guilty
of [thefallacyof]misplacedconcreteness.Ratherthanreplacingor rein-
terpreting the familiardichotomybetweenactors and structures,[this]
identification. . . actuallyreproducesit in anotherform.. . . Actorsper
se are much more than, and [simultaneously] much less than, 'agents'
[alone]"(Alexander1992, pp. 1-2; see also Alexander1988a). Empirical
action,then,is multiplydetermined.It is not drivenexclusivelyby hu-
man agency, but ratheris deeply structuredas well by several other
"environments" of action (to use Alexander'sterminology), such as the
societal (network)and culturalenvironments.17 Each of these environ-
mentsinterpenetrates withand givesshape and directionto themoment
of human agencyitself.Any empiricalinstanceof action is structured
simultaneously by thedynamicsof societalas well as culturalstructures,
even though-in principle,at least-it is nevercompletelydetermined
and structuredby them. Long beforeAlexander,Parsons had come to
thisverysame conclusionas well,bymeansofhis structural-functionalist
model and his celebrated"AGIL" schema (Parsons and Smelser 1956;
Parsons 1961). More recently, JurgenHabermas, too, has reliedheavily
uponthisanalyticaldistinction in his majorwork,The TheoryofCommu-
nicativeAction(1984-87).
Our aspirationhere is not to endorse any one of these particular
theoriesof social action unambiguously-certainlynot the structural-
functionalism of Parsons, for example, with its powerfulidealist and
consensualisttendencies.The fundamentalpointthatwe wish to make,
rather,is thatnetworkanalysiserrsseriouslyin ignoringthe conceptual
insightssharedby all of thesevarioustheories,in particularthe notion
thatagencyand structure interpenetrate withone anotherin all individ-

17 environment-
Like Parsonsand Sorokin,Alexanderalso speaksofa "personality"
an analyticalmove thatwe certainlyendorse,even thoughwe have not had the
opportunityto discusspersonality
structuresin thisparticularcontext(see Smelser
1968;Chodorow1989;and Goodwin1992).

1443

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

ual units(as well as complexes)ofempiricalaction,and thatall historical


processesare structuredat least in part by culturaland politicaldis-
courses, as well as by networksof social interaction.Earlier we had
quoted fromMargaretArcher,who called foran analysisof thecomplex
interaction(or "interplay")betweenculturaland social structuralforma-
tions. Here we suggestthat it is preciselythroughempiricalsocial ac-
tion-multiplydetermined, and undertakenbyconcretely situatedhistor-
ical actors-that these various analyticalenvironmentsrelate to one
another.The severalenvironments of actionoughtneverto be reifiedas
separate,concreteentities,muchlesshierarchized as ifone ofthem(social
networks)were always morecausallysignificant than the others.
Two crucial implicationsfollowimmediatelyand directlyfromthese
remarks.One is the notionthathistoricalactors'veryidentities,goals,
and aspirationsare themselvesfundamentally constructedphenomena.
This is a pointperhapsmostforcefully articulatedin the classical socio-
logical literatureby GeorgeHerbertMead (1962), who argued that the
mind and the social self-no less than societyitself-arise and are sus-
tainedthroughinteraction, and thatthehumansubjectmustbe regarded
as an ongoingdevelopmentalprocess. More recently,Craig Calhoun,
too, has suggestedthat identity"is not a static, preexistingcondition
that can be seen as exertinga causal influenceon collectiveaction; at
bothpersonaland collectivelevels,itis a changeableproductofcollective
action." In particular,identitycannotbe "adequatelycapturedby the
notionofinterest.Identityis a no morethanrelativelystableconstruction
in an ongoingprocess of social activity"(Calhoun 1991, pp. 59, 52).
Structuralist such as McAdam, Padgett,and Ansellde-
constructionists
velop thesepointsquite forcefully. The insightthat"riskyand unusual
collectiveactionplaces one's identityon thelinein an especiallypowerful
way" (Calhoun 1991, p. 61) receivesstrongcorroboration, forexample,
in McAdam's argumentsabout processesof identityconversionin social
movements.And Padgettand Ansell'sperceptiveremarkconcerningthe
"varying structuresof games"-and thereforeof "goals of self-
interest"-duringtimesof "complicatedchaos" illustratesthis point as
well. Thereis simplyno suchthingas a prestructured individualidentity;
both individualsand societiesare the productsand the contents-but
notthestartingpoints-of interaction."What is primaryis theintersub-
jective process"(Calhoun 1991, p. 59).
It is importantto notehere,incidentally, thatindividualautonomyis
itselfa constructedphenomenon;that is, individualautonomyis only
made possibleby the sheermultiplicity of structures-societalas well as
cultural-withinwhichsocial actorsare situatedat any givenmoment.
Not only is autonomylinkedto locationwithinoverlappingand inter-
sectingnetworksof social ties, as Simmelpointedout long ago (1955,

1444

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

1971, chap.18; see also Coser 1975; Burt 1980b), but it is also made
possible by actors'locationamong a multiplicity of culturalstructures,
such as idioms,discourses,and narratives.18
Of course,if culturaland societal(network)structuresshape actors,
thenit is equallytruethatactorsshape thesestructures in turn.Cultural
and social structuresdo not,in otherwords,bythemselvesbringaboutor
somehow"cause" historicalchange.Rather,it is theactionsof historical
subjectsthatactually"reconfigure" (givenhistorically conducivecircum-
stances)existing,long-term structuresofaction,bothculturaland societal
(Sewell 1992b, p. 46). Hence the second implicationof our earlierre-
marks: namely,that for a more comprehensiveunderstandingof pro-
cesses of change,it is necessaryto devotemoreattentionnotonlyto the
structural levelsofcausation,butalso to thosemoreephemeraldynamics
of historical"events,"that"relativelyrare subclass of happenings"that
transform such structuresin "significant"ways (Sewell 1992b,p. 31).19
While structuralist determinists such as the earlyWhite(and Rosenthal
et al. 1985) neglectaltogetherthisdimensionof historicalevents,contin-
gencies,and processesof social change in favorof static,side-by-side
comparisonsof networkconfigurations, structuralistinstrumentalistsare
considerablymoresensitiveto the causal significance of human agency.
It is the structuralistconstructionists,however,such as McAdam and
his associatesand, especially,Padgettand Ansell,who mostsuccessfully
incorporatethislevel of analysisintotheirexplicitframeworks of expla-
nation. Indeed, theirnotionsof "robustaction," "identityconversion,"
and "channelednetworkcascading" capturefar betterthan any other
networkconceptsthe inherentprocessualityand temporality-the"se-
quentialconnectednessand unfolding"(Griffin 1993,p. 1097)-of social
action (Ricoeur 1985; Abbott 1992b, 1992c; Sewell 1992c; Griffin1992;
Aminzade 1992). Togethertheyrevealmoreclearlythanever beforethe
extentto which, as Sewell has put it, "structures[findthemselves]at
risk,at least to some extent,in all of the social encounterstheyshape"
(Sewell 1992a, p. 20).
The mostfruitful directionin whichnetworkanalysiscan now proceed
is thustowarda deeperexplorationofjust such analyticalthemes.Net-
work analystswould do well, in our view, to thematizemoreexplicitly
thantheyhave theinherently constructed natureofindividualand collec-
tive identities.They would do well also to thematizethe complexways

18 Here, too, thevariousstructuresof the"personality"


environment-andthe ten-
sionsand contradictionsamongthem-mustbe noted,althoughonceagainit would
takeus too farafieldto discussthesestructures
at greaterlength.
19 For a provocative
and historically
groundeddiscussionof"theinteraction
ofsystem
and event,"see Sahlins(1981, 1991).

1445

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

in which actors' identitiesare culturallyand normatively,as well as


societally,determined-theempiricalinterpenetration, in otherwords,
of those culturaland social structuresthatwe have been arguingmust
be carefullydistinguishedfromone anotheron an analyticalplane. And
finally,networkanalystswould do well to distinguishtheoretically the
differentordersoftemporality and causationthatappear alreadyin their
historicalexplanations-theorders,thatis, oflong-term structuraldeter-
mination,on the one hand, and of the dynamiceffortsand projectsof
historicalactors,on theother.It is onlyby takingthesevarioustheoreti-
cal notionsintoaccountthatnetworkanalystswill realizemostfullythe
considerableresearchpotentialthatalreadyinheresin theirtechniques,
theirmethodologies, and theirhighlydistinctiveview ofthesocial world.

CONCLUSION
Despite thequantityand qualityofthescholarshipthatnetworkanalysis
has produced over the past 20 years, historicalsociologistsand social
theoristshave failedto examineand systematically to criticizeits funda-
mentaltheoreticalpresuppositions. The abstruseterminology and state-
of-the-artmathematicalsophisticationof this unique approach to the
studyof social structureseem to have preventedmanyof these"outsid-
ers" fromventuringanywherenear it. The resulthas been an unfortu-
nate lack of dialogue amongnetworkanalysts,social theorists,and his-
toricalsociologists,and a consequentimpoverishment of theirrespective
domains of social inquiry.In networkterms,all threecamps have re-
mained isolatedcliques separatedfromone anotherby structuralholes,
withunbridgeablesubculturalstylesand mutuallyincomprehensible dis-
courses. By examiningand criticizingthe theoreticalpresuppositions
of networkanalysis,we have triedto provide some sort of a link-a
"weak" tie, so to speak-between thesevariouscamps, and therebyto
renderthis approach more accessibleto sociologistsin a wide range of
fields.
We have shown how networkanalysis, despite its forbiddingself-
presentation, actuallyproceedsfroma fewspecific,simple,and elegant
theoreticalpresuppositions. Its principalachievement,we have argued,
has been to transform a merelymetaphoricalunderstanding of the em-
beddednessof actorsin networksof social relationships intoa morepre-
cise and usable toolforsocial analysis.We have also suggested,however,
that despiteits powerfulconceptualizationof social structure,network
analysisas it has been developedto date has inadequatelytheorizedthe
causal role of ideals, beliefs,and values, and of the actorsthatstriveto
realize them;as a result,it has neglectedthe culturaland symbolicmo-
mentin the verydetermination of social action. Networkanalysisgains

1446

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

its purchaseupon social structureonlyat the considerablecost of losing


its conceptualgrasp upon culture,agency,and process. It providesa
useful set of tools forinvestigating
the patternedrelationshipsamong
historicalactors.These tools,however,by themselvesfail ultimatelyto
make sense of the mechanismsthroughwhich these relationshipsare
reproducedor reconfigured over time. Our own positionis that a truly
syntheticaccountof social processesand transformations thattakes into
considerationnot onlystructuralbut also culturaland discursivefactors
will necessarilyentaila fullerconceptionof social actionthan has been
providedthusfarby networkanalysts.

APPENDIX
A ShortGlossaryof NetworkAnalysisTerms
The followingdefinitions are meantto orientthe generalreaderto the
basic termsof networkanalysisas well as to conveythe way in which
certainmore generalsociologicalconceptshave been "translated"into
theseterms.
Actor.-A person,group,organization, thing,event,and so on, linked
to othersin a network.This is sometimesreferred to as a "node."
Asymmetric tie.-A relationwhose form,content,or bothis different
forthe linkedactors.See also symmetric tie.
Block.-A setofstructurally equivalentactorsin a multiplexnetwork.
See also multiplexnetworkand structuralequivalence.
Blockmodeling.-A technique for findingor "partitioning"(and
graphicallyrepresenting) equivalentactors(or blocks) in a
structurally
network.
Boundaryproblem.-The problemofdefining thepopulationof actors
to be studiedthroughnetworkanalysis a way whichdoes not depend
in
on a prioricategories;in otherwords,theproblemofdelimiting thestudy
of social networkswhichin realitymay have no limits.
Catnet(fromcategoryand network). -A sociallycohesiveset of struc-
turallyequivalentactorshypothesizedas moreable and likelyto share
ideas or a commoncultureand to engagein collectiveactionthan other
sortsof real or latentgroups.See also social cohesion.
Centrality.-The numberof an actor'stiesto others,weightedby the
numberof the latter'sties to others.
Clique.-A group of actors in which each is directlyand strongly
linkedto all of the others.Comparesocial circle.
Content.-The specificnatureor typeof relationlinkingactorsin a
network(e.g., exchange,kinship,communicative, affective,instrumen-
tal, or powerrelations).

1447

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

Density.-The ratio of actual relationsor ties among a set of actors


in a networkand the maximumpossiblenumberof ties.
Distance.-A functionof the numberand strength of social tiessepa-
ratingtwo actors.
Dualism.-The idea that the natureof groupsis determinedby the
intersection of the actorswithinthem(i.e., by the actionsof theirmem-
bers),and thatthe natureof actorsis determinedby the intersection of
groups"within"them(i.e., by theirgroupaffiliations).
Egocentricnetwork.-An actor (sometimescalled the "anchorage"),
the actors with which it has relation,and the relationsamong those
actors. This is sometimesreferred to as a "personalnetwork."
Form.-The formalpropertiesof the relationsamongactorsin a net-
work(e.g., strength or weakness,density,symmetry or asymmetry).
Multiplexnetwork. -A networkwith two or moretypesof relations
linkingactors(e.g., exchangeand communication in a marketor commu-
nicativeand affectivetiesin a clique).
Network. -The set of social relationsor social ties among a set of
actors(and the actorsthemselvesthuslinked).
Networkstructure.-The patterningof relationsand "holes" among
actorsin a network.See also structuralholes.
Position.-A set of structurally equivalent actors or nodes (e.g., a
block).
Positional approach.-An analysisthat focuseson the patterningof
structurally equivalent relationsamong actors. See also relationalap-
proach.
Range.-The numberof an actor'sties to others.See also centrality.
Relational approach.-An analysisthat focuseson patterningof so-
ciallycohesiverelationsamongactors.See also positionalapproach.
Role.-The patternof relationsof a set of structurally equivalent
actors(i.e., a block) to otherblocks.
Simultaneity,the principleof. The idea that all positionsand roles
(see above) are determinedrelativeto one anotherand thus cannot be
assumed or alteredindependently of one another.
Social circle.-A groupin which each actor is directlyand strongly
linkedto most(e.g., 80%), but not necessarilyall, others.This is some-
timescalled a "social cluster."Compare clique.
Social cohesion.-The presenceof a dense networkof strongties
among a set of actors.See also clique and social circle.
Social structure.-The persistingpatternof ties among actors;more
specifically,a network(microstructure) or thenetworkofnetworks(mac-
rostructure).
Strengthof ties.-The relativefrequency,duration,emotionalinten-

1448

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

sity,reciprocalexchange,and so on whichcharacterizea giventie or set


of ties.
Structuralequivalence.-The sharing,by a set of two or moreactors
who are not necessarilylinkedthemselves,of equivalentrelationsto a
thirdactor.
Structuralhole.-The absenceofa relationamongactorsin a network
(a crucialelementof networkstructure).
Symmetric tie.-A relationwhoseformand/orcontentis thesame for
the linkedactors.See also asymmetrictie.

REFERENCES
Aminzade,Ronald. 1992. "HistoricalSociologyand Time." SociologicalMethods
and Research 20, no. 4 (May 1992): 456-80.
Abbott,Andrew.1992a. "FromCauses to Events:Noteson NarrativePositivism."
SociologicalResearchand Methods20:428-55.
. 1992b."Of Time and Space: The Contemporary Relevanceof the Chicago
School." SorokinLecture,deliveredto the SouthernSociologicalSociety,New
Orleans.
. 1992c. "What Do Cases Do?" Pp. 53-82 in WhatIs a Case? editedby
CharlesRaginand HowardBecker.New York:CambridgeUniversity Press.
Alexander,JeffreyC. 1982.Theoretical
LogicinSociology.Vol. 1:Positivism,
Presup-
positions,and Current Controversies.
BerkeleyandLos Angeles:UniversityofCali-
forniaPress.
. 1987.Twenty Lectures:SociologicalTheorysinceWorldWarII. New York:
ColumbiaUniversity Press.
.1988a. Actionand Its Environments.New York:ColumbiaUniversity Press.
,ed. 1988b.Durkheimian Sociology:CulturalStudies.New York:Cambridge
UniversityPress.
. 1988c. "The New TheoreticalMovement."Pp. 77-101 in Handbookof
Sociology,editedby Neil J. Smelser.BeverlyHills,Calif.:Sage.
. 1992."SomeRemarkson 'Agency'in RecentSociologicalTheory."Perspec-
tives 15:1-4.
Alexander,Jeffrey C., and PhilipSmith.1993. "The Discourseof AmericanCivil
Society:A New ProposalforCulturalStudies."Theoryand Society22:151-207.
Archer, Margaret.1988.CultureandAgency.Cambridge: Cambridge UniversityPress.
Barber,Bernard.1992."Neofunctionalismand theTheoryof theSocialSystem."Pp.
36-55 in TheDynamicsofSocial Systems,editedbyPaul Colony.London:Sage.
Barnes,JohnA. 1954. "Class and Committees in a NorwegianIsland Parish."Hu-
manRelations7:39-58.
Baron, Hans. 1966. The Crisisof theEarly Italian Renaissance.Princeton, N.J.:
Princeton UniversityPress.
Bearman,Peter.1993.RelationsintoRhetorics:Local Elite Social Structurein Nor-
folk,England,1540-1640.New Brunswick, N.J.: RutgersUniversity Press.
Blumer,Herbert.1969.SymbolicInteractionism. Berkeley:University of California
Press.
Boorman,ScottA., and HarrisonC. White. 1976. "Social StructurefromMul-
tiple Networks:II. Role Structures."
AmericanJournalof Sociology81:1384-
1446.

1449

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

Bott,Elizabeth.1971.Familyand Social Network, 2d ed. New York:Free Press.


Bourdieu,Pierre.1984. Distinction:A Social Critiqueof theJudgement of Taste.
Cambridge,Mass.: HarvardUniversity Press.
. 1990.In OtherWords.Stanford, Calif.:Stanford UniversityPress.
Bourdieu,Pierre,and LoYcJ. D. Wacquant.1992.An InvitationtoReflexive Sociol-
ogy.Chicago:University ofChicagoPress.
Breiger,Ronald L. 1974. "The Dualityof Personsand Groups."Social Forces53:
181-90.
Brint,Steven.1992."HiddenMeanings:CulturalContentand Contextin Harrison
White'sStructural Sociology."SociologicalTheory10:194-208.
Brym,RobertJ. 1988."Structural LocationandIdeological Divergence:JewishMarxist
Intellectualsin Turn-of-the-Century
Russia."Chap. 13 in Social Structures, edited
byBarryWellmanand S. D. Berkowitz. New York:Cambridge UniversityPress.
Burt,RonaldS. 1980a."ModelsofNetworkStructure." AnnualReviewofSociology
6:79-141.
. 1980b."Autonomy in a Social Typology."AmericanJournalof Sociology
85:892-925.
1982. Towarda Structural TheoryofAction.New York:AcademicPress.
1986. "Comment."Pp. 105-7 in Approachesto Social Theory,editedby
SiegwartLindenberg, JamesS. Coleman,and StefanNowak. New York: Russell
Sage.
. 1987."SocialContagionand Innovation: CohesionversusStructural Equiva-
lence."AmericanJournalofSociology92:1287-1335.
. 1992. StructuralHoles: The Social Structureof Competition. Cambridge,
Mass.: HarvardUniversity Press.
Calhoun,Craig. 1991."The ProblemofIdentity in CollectiveAction."Pp. 51-75 in
Macro-MicroLinkagesin Sociology,editedbyJoanHuber. BeverlyHills, Calif.:
Sage.
Chodorow,Nancy.1989.FeminismandPsychoanalytic Theory.New Haven,Conn.:
Yale University Press.
Coleman,James.1986. "Social Theory,Social Research,and a Theoryof Action."
American JournalofSociology91:1309-35.
. 1990.FoundationsofSocial Theory.Cambridge, Mass.: HarvardUniversity
Press.
Coleman,James,Elihu Katz, and HerbertMenzel. 1966.Medical Innovation.Indi-
anapolis:Bobbs-Merrill.
Coser,Ruth. 1975."The Complexity ofRolesas SeedbedofIndividualAutonomy."
Pp. 237-63 in The Idea ofSocial Structure: Essays in HonorofRobertMerton.
New York:HarcourtBraceJovanovich.
Davis, JamesA. 1967. "Clustering and Balance Theoryin Graphs."HumanRela-
tions20:181-87.
Douglas,Mary. 1966.Purityand Danger.London:PenguinBooks.
Durkheim,Emile. 1984.TheDivisionofLabor in Society.New York:Free Press.
Durkheim,Emile,and MarcelMauss. 1963.PrimitiveClassification. Chicago:Uni-
versity ofChicagoPress.
Edwards,Stewart,ed. 1973.TheCommunards ofParis, 1871.Ithaca,N.Y.: Cornell
University Press.
Emirbayer, Mustafa.1992a."The Shapingofa VirtuousCitizenry: EducationalRe-
formin Massachusetts,1830-1860."Studiesin AmericanPoliticalDevelopment
6:391-419.
. 1992b."BeyondStructuralism and Voluntarism: The Politicsand Discourse
of Progressive SchoolReform,1890-1930."Theoryand Society21:621-64.
1993."UsefulDurkheim. " CSSC Working PapersSeriesno.159, New School
forSocial Research.

1450

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

Emirbayer, Mustafa,and AnnMische. 1994."WhatIs Agency?"Manuscript.New


SchoolforSocial Research.
Fernandez,RobertoM., and Doug McAdam. 1988. "Social Networksand Social
Movements:Multiorganizational Fields and Recruitmentto MississippiFreedom
Summer."SociologicalForum3 (3): 357-82.
. 1989. "MultiorganizationalFields and Recruitment
to Social Movements."
In OrganizingforSocial Change:Social Movement OrganizationsacrossCultures,
editedby BertKlandermans. Greenwich, Conn.:JAIPress.
Fischer,Claude S. 1982. To Dwell amongFriends:PersonalNetworks in Townand
City.Chicago:University ofChicagoPress.
Freeman,LintonC. 1977."A Set ofMeasuresofCentrality Based on Betweenness."
Sociometry40:35-41.
in SocialNetworks:ConceptualClarification."
. 1979."Centrality Social Net-
works 1:215-39.
Friedman,Debra, and Doug McAdam. 1992. "CollectiveIdentityand Activism:
Networks,Choices,and theLifeofa Social Movement."Pp. 156-73 in Frontiers
in Social MovementTheory,editedby Aldon D. Morrisand Carol McClurg
Mueller.New Haven, Conn.:Yale University Press.
Giddens,Anthony.1984. The Constitution of Society.Berkeleyand Los Angeles:
UniversityofCaliforniaPress.
. 1987.Social Theoryand ModernSociology.Stanford, Calif.:Stanford
Uni-
versityPress.
Goodwin,Jeff.1992."The LibidinalConstitution ofSocialMovements: The Case of
theHuk Rebellion."Paper deliveredat themeetingof theAmericanSociological
Association,Pittsburgh.
.1994. "Towarda New SociologyofRevolutions." TheoryandSociety,inpress.
Gould,RogerV. 1991."MultipleNetworksand Mobilization in theParisCommune,
1871."AmericanSociologicalReview56:716-29.
. 1992."TradeCohesion,Class Unity,and UrbanInsurrection: ArtisanalAc-
tivismin theParisCommune."American JournalofSociology98:721-54.
Granovetter,Mark. 1973."The StrengthofWeakTies." American JournalofSociol-
ogy 78:1360-80.
Larry. 1992. "Temporality,
Griffin, Events,and Explanationin HistoricalSociol-
ogy."SociologicalMethodsand Research20, no. 4 (May 1992):403-27.
1993. "Narrative,Event-Structure in
Analysis,and Causal Interpretation
HistoricalSociology."AmericanJournalofSociology98:1094-1133.
Habermas,Jiurgen. 1984-87.The TheoryofCommunicative Action.Boston:Beacon
Press.
Harary,Frank,RobertZ. Norman,andDorwinCartwright. 1965.StructuralModels:
An Introduction to theTheoryofDirectedGraphs.New York:Wiley.
Homans,GeorgeC. 1964. "Bringing Men Back In." AmericanSociologicalReview
29:809-18.
. 1986. "FiftyYears of Sociology."Annual Review of Sociology12:xii-
xxx.
Kane, Anne. 1991. "CulturalAnalysisin HistoricalSociology."SociologicalTheory
9:53-69.
Knoke, David. 1990. Political Networks:The StructuralPerspective.Cambridge:
CambridgeUniversity Press.
Knoke, David, and JamesH. Kuklinski.1982. NetworkAnalysis.BeverlyHills,
Calif.:Sage.
Laumann,Edward0. 1979."Network Analysisin LargeSocialSystems:SomeTheo-
Problems."Chap. 18 in Perspectives
reticaland Methodological on Social Network
Research,editedbyPaul W. Hollandand SamuelLeinhardt.New York:Academic
Press.

1451

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

Laumann,EdwardD., and David Knoke.1987.TheOrganizational State:A Perspec-


tiveon NationalEnergyand HealthDomains.Madison:University of Wisconsin
Press.
Leifer,Eric M. 1988. "Interaction
Preludesto Role Setting."AmericanSociological
Review 53:865-78.
Lorrain,Francois,and HarrisonC. White.1971."StructuralEquivalenceofIndivid-
uals in Social Networks."JournalofMathematicalSociology1:49-80.
Luhmann,Niklas. 1982.TheDifferentiationofSociety.New York:ColumbiaUniver-
sityPress.
. 1990.Essays on SelfReference.
New York:ColumbiaUniversity Press.
Mandel,MichaelJ. 1983."Local Rolesand SocialNetworks." AmericanSociological
Review 48:376-86.
McAdam,Doug. 1986. "Recruitment to High-RiskActivism:The Case of Freedom
Summer."AmericanJournalofSociology92:64-90.
. 1988.FreedomSummer.New York:OxfordUniversity Press.
Mead, GeorgeHerbert.1962.Mind,Self,andSociety.Chicago:University ofChicago
Press.
Merton,RobertK., and A. S. Rossi. 1950."Contributionsto theTheoryofReference
GroupBehavior."Pp. 40-105 in Continuities in Social Research,editedby Rob-
ertK. Mertonand Paul F. Lazarsfeld.New York:Free Press.
Morris,AldonD., and CarolMcClurgMueller,eds. 1992.Frontiers in Social Move-
mentTheory.New Haven, Conn.:Yale University Press.
Nadel, S. F. 1957.The TheoryofSocial Structure.Glencoe,Ill.: Free Press.
Padgett,JohnF. 1986. "RationallyInaccessibleRationality." Contemporary Sociol-
ogy 15:26-28.
Padgett,JohnF., and Christopher K. Ansell.1993."RobustActionand theRise of
theMedici,1400-1434."American JournalofSociology98:1259-1319.
Park, Robert,ErnestBurgess,and RoderickMcKenzie. 1967. The City. Chicago:
University ofChicagoPress.
Parsons,Talcott.1937.TheStructure ofSocial Action,2 vols.New York:FreePress.
1951.The Social System.New York:Free Press.
ed. 1961.TheoriesofSociety.New York:Free Press.
Parsons,Talcott,and EdwardShils.1951.Towarda GeneralTheoryofAction.Cam-
bridge,Mass.: HarvardUniversity Press.
Parsons,Talcott,and Neil Smelser.1956.Economyand Society.New York: Free
Press.
Perlman,Selig. 1979. A Theoryof the Labor Movement.Philadelphia:Porcupine
Press.
Pocock,J. G. A. 1975.TheMachiavellianMoment.Princeton, N.J.: Princeton
Uni-
versityPress.
Polanyi,Karl. 1957."The Economyas Instituted Process."Pp. 243-70 in Tradeand
Marketin theEarlyEmpires,editedbyKarlPolanyi,ConradArensberg, andHarry
Pearson.Glencoe,Ill.: FreePress.
Rabinow,Paul, and WilliamSullivan,eds. 1979. Interpretive Social Science: A
Reader.Berkeley:University ofCaliforniaPress.
Rambo, Eric, and Elaine Chan. 1990. "Text, Structure, and Actionin Cultural
Sociology."Theoryand Society19:635-48.
Ricoeur,Paul. 1985.Timeand Narrative.Chicago:University ofChicagoPress.
Rosenthal,Naomi,MerylFingrutd, MicheleEthier,RobertaKarant,and David Mc-
Donald. 1985."Social Movementsand NetworkAnalysis:A Case Studyof Nine-
teenth-CenturyWomen'sReform in New YorkState."American JournalofSociol-
ogy 90:1022-54.
Sahlins,Marshall.1981.HistoricalMetaphorsand MythicalRealities.AnnArbor:
ofMichiganPress.
University

1452

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
NetworkAnalysis

. 1991. "The Return of the Event, Again." Pp. 37-100 in Clio in Oceania,
edited by A. Biersack. Washington,D.C.: SmithsonianInstitutionPress.
Schutz, Alfred. 1967. The Phenomenologyof the Social World. Evanston, Ill.: North-
westernUniversityPress.
Sewell,WilliamH., Jr.1980.WorkandRevolutionin France:TheLanguageofLabor
from the Old Regime to 1848. Cambridge: Cambridge UniversityPress.
. 1985. "Ideologies and Social Revolutions." Journal of Modern History 57:
57-85.
. 1992a. "A Theory of Structure: Duality, Agency, and Transformation."
American
JournalofSociology98:1-29.
. 1992b. "Three Temporalities: Toward an Eventful Sociology." Manuscript.
Universityof Chicago.
. 1992c. "Introduction:Narratives and Social Identities." Social Science His-
tory16:479-88.
Simmel, Georg. 1950. The Sociology of Georg Simmel. Translated and edited by
Kurt H. Wolff.New York: Free Press.
. 1955. Conflictand the Web of GroupAffiliations.
Translatedby KurtH.
Wolffand Reinhard Bendix. New York: Free Press.
. 1971.GeorgSimmel:On Individuality
and Social Forms,editedby Donald
Levine. Chicago: Universityof Chicago Press.
Skocpol, Theda, ed. 1984. Vision and Method in Historical Sociology. Cambridge:
Cambridge UniversityPress.
. 1985. "Cultural Idioms and Political Ideologies in the RevolutionaryRecon-
structionof State Power: A Rejoinder to Sewell." Journal of Modern History 57:
86-96.
Smelser, Neil J. 1968. "Social and PsychologicalDimensions of Collective Behavior."
Pp. 92-121 in Essays in Sociological Explanation. Englewood Cliffs,N.J.: Prentice
Hall.
Smith, Dennis. 1991. The Rise ofHistorical Sociology. Philadelphia: Temple Univer-
sityPress.
Smith, Philip. 1991. "Codes and Conflict." Theory and Society 20:103-38.
Snow, David A., and Robert D. Benford. 1988. "Ideology, Frame Resonance,
and Participant Mobilization." International Social Movement Research 1:197-
217.
Snow, David A., E. Burke Rochford,Jr.,Steven K. Worden, and RobertD. Benford.
1986. "Frame AlignmentProcesses, Micromobilization,and Movement Participa-
tion."AmericanSociologicalReview51:464-81.
Snow, David A., Louis A. Zurcher, Jr., and Sheldon Ekland-Olson. 1980. "Social
Networks and Social Movements: A MicrostructuralApproach to DifferentialRe-
cruitment."American Sociological Review 45:787-801.
Somers, Margaret. 1992. "Narrativity,Narrative Identity,and Social Action." Social
Science History 16:591-630.
. 1993. "Citizenship and the Place of the Public Sphere." American Sociologi-
cal Review 58:587-620.
Sorokin,Pitirim.1947.Society,Culture,and Personality:
TheirStructure
and Dy-
namics. New York: Harper.
Steinmetz,George. 1992. "Reflectionson the Role of Social Narratives in Working-
Class Formation." Social Science History 16:489-516.
Stinchcombe,Arthur. 1978. TheoreticalMethods in Social History. New York: Aca-
demic Press.
Swidler, Ann. 1986. "Culture in Action: Symbols and Strategies."AmericanSociolog-
ical Review 51:273-86.
. 1987. "The Uses of Culture in Historical Explanation." Paper presentedat
the 82d annual meetingof the American Sociological Association, Chicago.

1453

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
AmericanJournalof Sociology

Thomas,W. I. 1966.On Social Organization and Social Personality,editedbyMorris


Janowitz.Chicago:University ofChicagoPress.
Tilly,Charles.1986.TheContentious French.Cambridge, Mass.: HarvardUniversity
Press.
. 1992. "Historyand SociologicalImagining."Manuscript.New School for
Social Research.
Weber,Max. 1949. FromMax Weber,editedby Hans Gerthand C. WrightMills.
New York: OxfordUniversity Press.
Wellman,Barry.1979. "The Community Question:The IntimateNetworksof East
Yorkers."AmericanJournalofSociology84:1201-31.
. 1983."NetworkAnalysis:SomeBasic Principles." Pp. 155-200in Sociologi-
cal Theory1983,editedby R. Collins.San Francisco:Jossey-Bass.
Wellman,Barry,and S. D. Berkowitz,eds. 1988. Social Structures.New York:
CambridgeUniversity Press.
White,HarrisonC. N.d. "Notes on the Constituents of Social Structure."
Manu-
script.HarvardUniversity, Department ofSociology.
1992.Identityand Control.Princeton,
N.J.: Princeton University Press.
1993. "NarrativesintoTimes."Manuscript.ColumbiaUniversity, Depart-
mentof Sociology.
White,HarrisonC., ScottA. Boorman,and RonaldL. Brieger.1976."Social Struc-
turefromMultipleNetworks:I. Blockmodelsof Roles and Positions."American
JournalofSociology81:730-80.
Zerubavel,Eviatar. 1991.The Fine Line. New York:Free Press.

1454

This content downloaded from 141.218.001.105 on August 09, 2016 14:08:44 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

You might also like