emir bayer 2782580
TRANSCRIPT
-
8/10/2019 Emir Bayer 2782580
1/45
Network Analysis, Culture, and the Problem of AgencyAuthor(s): Mustafa Emirbayer and Jeff GoodwinSource: American Journal of Sociology, Vol. 99, No. 6 (May, 1994), pp. 1411-1454Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/2782580.
Accessed: 12/12/2014 20:29
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.
http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
The University of Chicago Pressis collaborating with JSTOR to digitize, preserve and extend access to
American Journal of Sociology.
http://www.jstor.org
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=ucpresshttp://www.jstor.org/stable/2782580?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2782580?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ucpress -
8/10/2019 Emir Bayer 2782580
2/45
Network
Analysis,
Culture,
and the
Problem of Agencyl
Mustafa Emirbayer
New Schoolfor Social Research
Jeff
oodwin
New York University
Network nalysis s one ofthemostpromisingurrentsnsociologi-
cal research, nd yet
t
has neverbeensubjected o
a
theoretically
informedssessment nd critique. his article utlines hetheoreti-
cal
presuppositions
f network
nalysis.
t also
distinguishes
e-
tweenthreedifferentimplicit)models
n
the network iterature f
the interrelationsf social structure, ulture, nd humanagency.
It concludes hatonly strategy orhistorical xplanation hat yn-
thesizes ocial structural nd cultural nalysiscan adequately
ex-
plain the formation,eproduction,nd transformationf networks
themselves. he article ketches he broad contours f
such a theo-
retical ynthesisn the conclusion.
Recentyearshave witnessed he emergence f a powerful ew approach
to
the study f social structure. his mode of nquiry, ommonly nown
as
network nalysis, has achieved a highdegree
of
technical ophisti-
cation and has
proven extremely
seful
n
a
strikingly
ide
range
of
substantive pplications.Since the seminalwork of Barnes (1954)
and
Bott 1971), sociological tudies tilizing etwork nalysishave appeared
with ncreasing requency; veritable xplosion fsuchworkhas taken
place overthe ast 15 years,particularly ith hefounding ftwo special-
'
We would ike o thank
articipants
n
the
CROPSO
Workshop
t HarvardUniver-
sity esp.ThedaSkocpol) or heirmany elpful omments,s well s participants
n
the
Sociology
taff
eminar
t theNew
School
for
ocial
Research
esp. Janet
Abu-
Lughod,
Diane
Davis, Karl-Dieter pp, and Arthur idich) nd
the
Methodology
Workshop
t
the
New York
Universityociology epartmentesp.
Wolf
Heydebrand,
Guillermina
asso,
nd
JamesJasper).
We would
lso liketo thank eter
Bearman,
Gerardo el
Cerro,
Karen
Gelb,Jeffreyoldfarb, ogerGould,
Ann
Mische,
Calvin
Morrill, ohn adgett, lessandroizzorno,Margaret omers, harles illy, nd the
AJS referees
or
heirmanyhelpful uggestionslong
he
way. Correspondence ay
be
addressed o Mustafa
mirbayer,epartment
f
Sociology,
ew
School
or
ocial
Research, 5
Fifth
Avenue,New York,New
York 10003.
?
1994
by
The
University
f
Chicago.
All
rights
eserved.
0002-9602/94/9906-0001$01.50
AJS Volume 99 Number
6
(May 1994):
1411-54 1411
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
3/45
AmericanJournal
f
Sociology
ized ournals,Social Networks nd Connections,
n
the ate 1970s.Today
network nalysis
s
one of the most promising urrents
n
sociological
research. ts practitionersnclude ome of the mosthighly espected ig-
ures
in
theprofession: on Breiger,Ronald Burt,Mark Granovetter,
David Knoke, PeterMarsden, Barry Wellman,
and Harrison
White.
Many
other
prominent ociologists,
uch as
Claude
Fischer,
Edward
Laumann, Doug McAdam,
David
Snow,
and
CharlesTilly,
draw exten-
sively pon
network
oncepts.
he
late
GeorgeHomans,
n
his reflections
upon
the ast 50
years
of
sociology,ustly
describednetwork
nalysis
s
one ofthemost
ncouraging
ew
developments
n
the
discipline Homans
1986, p. xxvi).
Despite tsgrowing rominence, owever,networknalysishasyet o
be
subjected
to a
theoretically
nformed ssessment nd
critique.
The
secondary iterature
n this
perspective
as
tended
to restricttself o
outlining
asic
concepts, iscussing
echnical
rocedures,
nd summariz-
ing empirical esearch indings. here has been
an
unfortunateack of
interest
n
situating
network
nalysis
within he broader traditions f
sociological heory,
much ess
in
undertaking systematicnquiry nto
its
underlying trengths nd weaknesses. Theoretical precursors f
network
nalysis
have
often een invoked
n
passing-especially Durk-
heim nd Simmel-but networknalysis, tself constellationfdiverse
methodologicaltrategies,
as
rarely
een
systematicallyrounded
n the
conceptual
frameworks
hey
elaborated.
n
addition,
here
has
been
a
notableabsence
in
this iterature f any sustained onsideration f the
potentialusefulnessof networkanalysis for historical nvestigation.
Meanwhile,
ocial theoristsnd historical
ociologists,
or
heir wn
part,
have
largely gnoreddevelopments
n
this
field;we have yet
to
see a
sustaineddiscussion f this
approach
n
recentworks
of
social theory,
even
in
thewritingsfsuchwide-ranginghinkers s AnthonyGiddens
(1984, 1987)
and
Jeffreylexander 1982, 1987, 1988a) or, alternatively,
in
studies
f recent
evelopments
n
comparative
nd
historical esearch
(e.g., Skocpol 1984;
D.
Smith
1991).2
In
short,
he
task
of
rethinking
network nalysis, ociological heory, nd historical ociology
n
light f
one another
has
been
sadly neglected.
In
this
essay we aspire to accomplish recisely uch a task. We begin
with n
exposition
f the
underlyingheoreticalresuppositionsnd con-
ceptual strategies
f network
nalysis,outlining
he characteristic
ea-
turesof thisapproach n relation o broadercurrentsn social theory.
Along
the
way,
we examine
everal
exemplary
tudies
that nvestigate
historical
rocesses
f social
change using
the
tools
and
insights
f
net-
1412
2
Giddens
oes,however, riticizehe structuralismf
Peter lau, which as strong
affinitiesithnetworknalysis Giddens 984,pp.
207-13).
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
4/45
Network
Analysis
work
analysis. Upon this basis we then elaboratea critique-a funda-
mentally ympatheticritique-of the network erspective,
tressingts
inadequate conceptualizationsf humanagency n the onehand, and of
culture n the other.We carefully istinguish erebetween hree istinct
models
in the
network iterature f the relationshipsmong culture,
agency, nd social structure;ach ofthesemodels, n our view,
conceptu-
alizes
theserelationships
n
varying egrees f theoreticalophistication.
Throughout he essay,we attempt o ground ur critical rguments are-
fully
n
detailed and
substantive onsiderations
f
actual works
of
net-
work
analysis-studies that, n our estimation, ank
among the most
powerful nd impressive pplications o date of thenetwork
erspective.
Throughout his article,our primary ocus s on analytical ategories,
not on authors.Our concernwith agency nd culturehas
implications
for
network nalysis n general,but our focus here remainson
studies
that
ttempt o explain ocial changeover time-including the
transfor-
mation
f
social networks hemselves-and henceon specificallyistori-
cal studies.3
Our
argument
s
thatwhilethisnew mode of structuralist
nquiry-in
all three
f ts versions-offers more
powerfulway
of
describing
ocial
interactionhan do other
tructural erspectives hatfocus
olely
on the
categorical ttributes f individual and collective ctors, t has yetto
provide a fully dequate explanatorymodel for the actual
formation,
reproduction, nd transformation
f
social networks
hemselves.Net-
work
nalysis
ll
too often enies
n
practice
hecrucialnotion hat ocial
structure, ulture,
nd human
agencypresuppose
ne
another;
t either
neglects r inadequately onceptualizes he crucialdimension
f
subjec-
tive
meaning
nd
motivation-including
he normative ommitments
f
actors-and
thereby
ails o show
exactly
how it is
that
ntentional,
re-
ativehuman
ction erves
n
part
o constitutehose
very
ocial networks
that o powerfullyonstrain ctors n turn. n its ess nuancedversions,
in
fact,
he
network
pproach merges s
themirror
mage
of ts
nterpre-
tive
and hermeneutic
ounterpartSchutz 1967; Rabinow
and
Sullivan
1979).
Whether rom he
standpoint
f
objective
ocial structures r of
subjective xperience,
oth
of
these
perspectives
n
themselves
rovide
no
more han a description
f
social reality; ltimately, othfailto
grasp
in
concepts
he
dynamic rocesses hat shape thisreality ver time. We
3We are not nterestedere,t hould epointed ut, n themore urelymethodologi-
cal and technical ontributionshatnetworknalysts ave producede.g., Boorman
and White 976;Davis 1967;Freeman 977,1979;Harary,Norman, nd Cartwright
1965; Lorrain nd White 1971; White,Boorman, nd Breiger 976). Nor are we
concerned ith overing ll of themany mportantmpiricaltudies hathave been
produced n recent earsusingnetwork echniquesnd concepts. ee, e.g., Fischer
1982;Laumann nd Knoke 1987; nd Wellman 979.
1413
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
5/45
AmericanJournal f Sociology
believe,by contrast, hat
n adequate approach o historical xplanation
must ncompassboth ocial
structuralnd cultural erspectives
n social
action. We demonstrate ow more sophisticated ersionsof network
analysis do approximate
uch a strategy, nd we sketchout the broad
contours f
our own
synthetic
xplanatory erspective
n the concluding
pages.
ANALYZING SOCIAL NETWORKS
The Priority f Relations
ver Categories
Network nalysis s not a formal r unitary theory hatspecifies is-
tinctiveaws, propositions,
r
correlations, ut
rather
broad strategy
for
nvestigating
ocial structure.
t is
not,
that
s,
a deductive
ystem
in
which lower-order
ropositions ollow as
a logical conclusion
from . . general propositions
nder . . specified iven conditions
(Homans 1964,p. 812).Rather, ike modernization
r dependency heory
in
the field of economic
development, t is more a
paradigm or a
perspective - a
loose federation f approaches Burt 1980a)-than
a
predictive
social
theory.
As a result
f
the nternal iversity f
net-
workapproaches,network nalysts hemselves ebatethe usefulness f
alternative
models
of
social
relations
nd
methodological
trategies.
n-
deed, important isagreements
ave arisenover
the
verydefinitionf
ts
fundamental
oncepts.
Network
nalystsdispute
the manner
n
which
ideas such as social structure,
etwork entrality, istance,
cohesion,
and
social
network tself-terms used
by
other sociologists imply s
metaphors-can
be
operationalized
or
purposes
of
empirical esearch.
(We
have
provided
short
lossary
f terms
ommonly
sed
by
network
analysts
n the
appendixbelow.)
Nevertheless,
etwork
nalysisproceeds
rom ertain asic theoretical
presuppositions
nd
premises
hat
are
acceptable
to
most,
f
not
all,
of
its
practitioners.
t
holds o a setof mplicit ssumptionsboutfundamen-
tal issues
in
sociological
nalysis
such
as
the
relationship
etween
the
individualand
society,
he relationship etween micro
nd macro,
and the
structuring
f social action
by objective,
supra-individual
patterns
f
ocial
relationships.
he
point
f
departure
ornetwork
naly-
sis
is
what we
shall
call
the
anticategoricalmperative.
his
imperative
rejects ll attempts o explainhuman behavior r social processes olely
in
termsof the categorical ttributes
f
actors, whether
ndividual or
collective.Network
nalysis,
s
Barry
Wellman
puts
t, rejects xplana-
tions
of
social behavior
s the
result
f
ndividuals' ommon
ossession
of
attributes
nd
norms
ather han
as the
result f their nvolvement
n
structuredocial relations
Wellman1983, p. 165).
In
other
words,
one
1414
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
6/45
Network
Analysis
can never imply ppeal to such attributess class membership r class
consciousness, oliticalparty ffiliation,ge, gender, ocial status,reli-
giousbeliefs, thnicity,exual orientation, sychological redispositions,
and so on,
in
order to explain why people behave the way they do.
Network
theory
builds its
explanations
from
patterns
f
relations,
notes Ronald Burt. It capturescausal factors n the social structural
bedrock
f
society, ypassing hespuriously ignificantttributes f peo-
ple temporarilyccupying articular ositions n social structure Burt
1986, p. 106).4
n
thisrespect,network nalysispursues he Simmelian
goal
of a
formalistic ociology Simmel 1971, chap. 3), one
that
directs
attention
xclusively o the overall structure f network
ies while
sup-
pressing onsideration f their ubstantive ontent see also Bearman
1993, p.
48).5
Given this
anticategoricalmperative, etwork nalysisemphatically
rejects ll varieties f culturalism,ssentialism,nd methodologicalndi-
vidualism.
t stands
fundamentallypposed, forexample,to certain
f
the
ssumptions f tructural-functionalism,hich tresses henormative
integration f societies.Despite a common mphasison the priority f
structures ver
essences
an
emphasis deriving ltimately
rom he
earlyDurkheim)
hat inks both
perspectives
o tendencies
n
structural
anthropology,inguistics,nd ordinary anguage philosophy-network
analysts take seriouslywhat Durkheim aw but
most
of
his
followers
did not: that the
organic olidarity
f a social
system
estsnot on the
cognition
f
men, but
rather
n
the
interlock
nd
interaction
f
objec-
tively definable social relationships Boorman
and
White 1976, p.
1442).6 It shouldbe pointed ut, on theother and, thatnetwork nalysis
does not
reject
ll ofthe enets
f
tructural-functionalism.
any
network
analysts e.g., White, Boorman,
and
Breiger1976;
Burt
1982) employ
functionalist otions uch as
role,
role
set,
and
status, although
they reconceptualize hese notions nto relationalor network-analytic
4 Thus one network nalyst Brym 988) s able
to show n his work n intellectual
radicalismmong he in-de-siecleewishntelligentsia
n Russia hat harp iscrepan-
cies n politicaldeologies an develop venwithin
arrowlyircumscribed
ategories,
depending pon
he
ctors' ifferentatterns
f
structuralootedness ithin artic-
ular
ocialnetworks.
5 This last formulationpplies omewhatess
well, t shouldbe pointed ut, to the
third f he everal etwork odels hatweshallgo ontoanalyze,hat f tructuralist
constructionism.his model upplements static nd formalisticnalysis
f
net-
work tructure ith more dynamic
ccount f
processes
f
dentity
ransforma-
tion, ne thatnecessarilyevotesmore ttention
o thecontent fnetworkies.
6
Network nalysts' isagreement ith he tructural-functionalistiew of society
s
thus eminiscentf SeligPerlman's bjection o
thoseMarxist heorists ho viewthe
working lass as
an abstractmass
n
the
grip
f
an abstract orce
1979,p. 6).
1415
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
7/45
-
8/10/2019 Emir Bayer 2782580
8/45
NetworkAnalysis
Therewas
no
comparableevelopment
f
ools or nalysisf
he
ehavior
of
nteractingystemsf ndividuals
rfor
apturinghe nterdependencies
of ndividualctions s
they
ombine
o
produce system-level
utcome.
The far reateromplexityequiredf ools or hese urposesonstituted
a serious mpediment o theirdevelopment.
.
. The end result was]
extraordinarilylaborated ethodsor nalysisf hebehaviorf setof
independentntitiesmost ftenndividuals), ith ittle evelopmentf
methodsor haracterizingystemicction esultingrom he nterdepen-
dent ctions
f
membersfthe
ystem.Coleman 986, . 1316]
Fromthishistorical antagepoint, ontemporaryetwork nalysis an
be
viewed as part of a second crucial watershedperiod
in
American
sociology,
ne
in
which
empirical esearch
s now
directing
ts
attention
back again to thesystemicevel,thistimeassistedbythedevelopment
of
quantitative echniques nd
methods
f
a
highly ophisticated ature.
In
thissecond pivotalmoment, nalytic oncerns re shifting ack once
more o
thosequestions f nteractional ields nd contextual etermina-
tionthat
had been so central o sociologists efore he variablesrevolu-
tion
of
the 1940s.
How, then,does network nalysis ropose o account or ocial behav-
ior andprocesses? he answer s implicit
n
thepreceding emarks. uch
behavior nd processes,tsuggests,mustbe explainedwithreferenceo
networks f
social relations
hat
link
actors
or
nodes.
These social
relations, ignificantly, ustbe understoods independentf theactors'
wills, beliefs,
nd
values; they
must also be assumed to allocate scarce
resources ifferentiallyWellman 1983, p. 176). Social structure,
n
this
view,
s
regularities
n
the
patterns
f relations
mong
concrete
ntities;
it
is
not a
harmony mong
abstractnorms nd
values
or a classification
of concrete ntities
ytheir
ttributes
White
et al.
1976, pp. 733-34;
emphasis
n
original).
A
social
network s one of
manypossible
sets of
social relations of a specificcontent-for example, communicative,
power, affectual,
r
exchange
elations-that ink actors
within
larger
social structure
or
network f
networks).
he
relevant
nit
of analysis
need not be
an
individualperson,but can also be a group, n organiza-
tion,or, ndeed,
n
entire
society i.e.,
a
territorially
ounded
network
of social
relations);7 ny entity
hat
s
connected o
a
network
f
other
such
entitieswill
do.
Network
nalysts
ften ind
t
desirable o
carry
ut their
nvestigations
at both he ndividual
nd
group evels;
uch
combinations
ighlight
hat
some ofthem, na Simmelian1955) fashion, efer oas the dualism of
groups
nd
actors-the
factthat the natureof
groups
s determined
y
the ntersectionf the actors
within
hem
i.e., by
the ties
of
their
mem-
7One
recent writer,
Margaret Somers
(1993), prefers he term relational setting.
1417
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
9/45
AmericanJournal f
Sociology
bersto
one another
s
well as to other
roups
nd
individuals),
whilethe
natureof actors s
determined y the
intersection f groups within
them i.e., by theirown variousgroupaffiliations;ee Breiger1974).
Individual and group
behavior, n thisview,
cannotbe fully nderstood
independentlyf one
another.By thusfacilitatingnalyses at both
the
individual nd group evel,network nalysis
makes t possible o
bridge
the
micro-macrogap -the theoretical
ulf
between
microsociology,
which examines the
interaction
f
individuals,
and
macrosociology,
which
tudies
he
nteraction
f
groups
r
institutions.
Thus theway nwhich
network nalysis
onceptualizes ocialstructure
is at
once moregeneral nd moreconcrete han
alternativetructuralist
approaches. t is moregeneralbecausemanydifferentinds ofgroups,
relations, nd institutionshatputatively rganize r
structureocial
pro-
cesses can be understood
n,
or be
translated nto,network
erms.
And it
is more concrete ecause these tructures eed not
be treated
s
black
boxes,
but
rather
an be
disaggregated nto
their
constituent
elements f actors
and
relations.
The
significance
f these
two
features
of
networknalysis or mpirical esearch s
considerable.Network na-
lysts re now able to
providefar moreprecise
nd accuraterepresenta-
tions f social structures
nd
social
relations han re
proponents
f com-
peting esearch trategies.
Most
network nalysts, f course,also
purport o
do far
more than
simplydescribethe
ways
in
which
actors
are connected
n
society.
As
Knoke and Kuklinski
point
out,
If
network
nalysis
were
imited
o a
conceptual
framework or
dentifying
ow a
set
of
actors s linked to-
gether,
t would
not
have excited
much
nterest nd effort
mong
ocial
researchers. ut network
nalysis ontains furtherxplicit
remise f
great
onsequence:The
structuref relationsmong ctors nd the oca-
tionof
ndividual ctors nthe network ave
important ehavioral,
er-
ceptual,
and attitudinal
onsequences
oth for he
ndividualunits
and
for the
system
s
a whole
(Knoke
and
Kuklinski
1982, p. 13).
The
hallmark
of
network
nalysis,
n
Edward Laumann's (1979, p.
349)
words,
is to
explain,
t
least
n
part,
the
behavior f network lements
(i.e., the
nodes)
nd ofthe
ystem s
a
wholeby
appeal
to
specific eatures
of the
nterconnections
mong
the
elements. More
specifically,
he net-
work
approach nvestigates
he
constraining
nd
enabling
imensions
f
patterned elationships
mong
social actors
within
system.
t is this
emphasis, n fact, hatprovides he inkbetween tstheoreticalnsights
and its
key
contributionso
empirical
esearch.What network
nalysis
provides,
n
particular,
s a
way
of
avoiding
he
pitfalls
f
what Arthur
Stinchcombe
erms
epochal
nterpretations
or,
more
memorably,ep-
ochal
garbage );
that
s,
of causal
explanations
hat
proceed by using
the
apparent
ausal
structurereated
y
narrative
f
sequence
of
events
1418
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
10/45
Network
Analysis
to
create the
illusion that
epochal
theories re
being substantiated
(Stinchcombe 978, p. 10;
see
also
Tilly 1992).
Network
nalysis llows
historical ociologists, y contrast, o pinpoint hosewide-rangingnd
recurrent ausal mechanisms whose combinations roduce,
as
Tilly
expresses t,
the actual
unique
histories e observe
Tilly 1992, p. 11).
How, specifically,
oes network
nalysis ttempt
uchan
important
ask?
RelationalAnalysis s a Way ofRepresentingocial Structure
Network
nalysts enerally
make use
of one of two
conceptual trategies
in order to explain how networks
onstrain nd enable-and thus to
accountforvarious types f socialbehavior Burt 1980a, 1987). On the
one
hand, many analysts dopt a
relational r social cohesion ap-
proachthatfocuses n the direct nd indirect onnections mong ctors.
This approach explainscertainbehaviors r processes hrough he fact
of social
connectivitytself-as well as through hedensity, trength,
symmetry,ange, nd so on, of thetiesthatbind. Fromthis
perspective,
very trong, ense,
and
relatively
solated ocial networks acilitate he
development f uniform subcultures nd
of
strong ollectivedentities;
this
notion,
f
course,
dates
back
both
o
Durkheim
1984) and
to
Simmel
(1955; 1971, chap. 18). Relationalanalyses,however, lso demonstrate
that weak ties
ndirectly
onnectingndividuals r bridging he struc-
tural holes
between solated social
groups may be crucial for many
important ocial processes, uch as
locating mployment pportunities
(e.g., Granovetter 973;Burt 1992).
The
mathematical ools
of
graph
theory
ave
been helpful
n
developing hese approaches see Harary
et al.
1965).
One
important pplication
f
network
nalysis
hat
employs
he
rela-
tional
or
social cohesion pproach s a
studyby Naomi
Rosenthal t al.
(1985)
of
women's reformrganizations
n
New York State during he
latter
halfof the nineteenthentury. osenthal nd her associates xam-
ine the
organizational ffiliationsf
202
prominent
omenreformers
n
state reform
ctivity uring
he
yearsbetween1840 and
1914. By map-
pingout these ffiliations,hey
evelop
a detailed
portrait
f the
multi-
organizational
ield
of
social movement
ctivity
hat obtained
during
that
period.Measures of centralitynable
them
oidentifyheparticular
groups
hatweremost
mportant
o thatnetwork f
reform
rganizations.
Otherrelatedmethodsprovidethemwithcarefulmeasurements f the
boundaries, hape,
and
texture
f
variousclusters
f
such
organizations.
Finally,
directional
nalysis
of
flows
of
individuals cross
organiza-
tions
provides
hemwith a
picture
f three
uniquely haped map
con-
figurations
f social movement
lusters
n
threedistincthistorical
e-
riods between 1840 and 1914, each of these manifesting different
1419
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
11/45
AmericanJournal f Sociology
content o the relations
mongorganizations nd activity Rosenthal t
al. 1985, p. 1043).
Another mportantine of network esearch mploying elational nal-
ysis
concerns
tselfwith processes f recruitmento social movements.
Its point of departure s a studyby Snow, Zurcher, nd Ekland-Olson
(1980) thatexaminesmaterials n a wide variety f movements. now et
al.'s work
suggests hat
ndividualswith
preexistingies to movement
members re more
ikely o be contacted nd recruited o those move-
mentsthan are individualswithout uch ties; individualswith few or
weak
ties
to
alternative etworks re more ikely o respondfavorably
to these
recruitmentffortshan
re
ndividualswith trong ommitments
tocountervailingetworks. now etal.'s study oesnotemploy ophisti-
cated network nalysis techniques,nor does it examine systematically
the
independent ffects
f
network tructureshemselves; t focuses
n-
stead
upon
ndividual ies alone. And
yet
t does
provide
useful
orrec-
tive
to
social
psychological
nd
culturalist pproaches
hat
place undue
explanatoryweightupon
such variables as
individualmotivation o
the exclusion f actors'
patterns
f embeddedness
ithin ctual
networks
of social
ties.
A
seriesof
studiesby Doug McAdam (1986, 1988;
see
also Fernandez
and McAdam 1988, 1989)furtherorroboratesnd advances this ineof
investigation.
cAdam's
studies re concernedwith he participation
f
college tudents
n
the
1964
FreedomSummer
roject
n
Mississippi
nd
focus
on the
patterns
f social
relationships
f student
participants,
s
compared
o
those
f
tudentswho
applied
for he
project,
were
ccepted,
but failed
to
participate.McAdam's principal
onclusion s that all of
the
applicants-participants
nd withdrawals
like-emerge
as
highly
committed,
rticulate
upporters
f the
goals
and values
of the summer
campaign McAdam 1986, p. 73). What clearlydifferentiatesartici-
pants
fromwithdrawals
re
not their ocioeconomic haracteristics or
attitudinal ifferenceshat
push
individuals
nto
participating, ut
three tructural
pull
factors
hat
facilitated
nd
encouraged articipa-
tion:
1) participants elonged
o a
greater
umber f
organizations
nd
to more
xplicitly olitical rganizations
handid
withdrawals;2) partici-
pants
had
higher
evels
of involvement
n
prior
civil
rights
ctivities
than did
withdrawals; nd,
most
mportant,3) participants
ad more
ties-especially strong riendship
ies-to other reedomSummer
p-
plicants han did withdrawals.
An
article
yMcAdam
and Roberto ernandez
1988;
see also Fernan-
dez
and McAdam
1989)
extends
he
scope
of
this
nalysis
o include he
effects f even more
pecifically
tructural ariables
uch
as
the network
prominence
f
applicantsupon
their
ventual
participation
n
Freedom
Summer. McAdam and Fernandez
argue
that individuals'
positions
1420
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
12/45
Network
Analysis
within
multiorganizationalields-in this case, the various networks
f
interlockingctivistorganizations t theirrespective niversities-did
undercertain onditionsignificantlyffect he ikelihood f theirbeing
recruited o the summer roject. t was not the quantity f ndividuals'
ties to social movement
rganizations
r to other
pplicants
o
Freedom
Summer hatdetermined
heir ventualparticipation, ut rather hepat-
tern
f
their
nterpersonal
onnectionsnd commonmemberships ithin
these organizations. o capture the effects f these patterns, ernan-
dez and McAdam
employ standard ohesionmeasureof centrality-
prominence -in their nalysis; theyuse it to focus attention n the
density f applicants' ies
to othermore r ess centrallyocated ndividu-
als within ields foverlapping rganizations.
Despite
his
increasing nterest
n
the significance f networks
s
such
in
movement ecruitment, owever-as opposed to social ties
treat[ed] . . in
piecemealfashion Gould 1991, p. 717)-McAdam's
research eaves out of
consideration ne further tructural ariable:
namely, hemultiplexity
fnetworks hemselvesnd thecomplex nterac-
tions mongthem.While
McAdam's studiesneglect hepreexisting ebs
of
relationships ithinwhichFreedom
ummer pplicants
had been
em-
bedded before
oining ctivist
rganizations, oger
Gould's
1991, 1992)
studies f the ParisCommune xplore he nteractive ffectsfprecisely
such networkswith more
formalorganizational tructures. ogether,
Gould
argues-not
separately-these ndigenous
tructures
specifically,
the
neighborhoods ithinwhich the Parisian
nsurgentsived)
and
the
National Guard unitsto which the
nsurgents
ere
assigneddecisively
affected heir verall
evels
of
solidarity
nd resistance.
he
key
to
this
interaction rocesswas theresidential ecruitmentystem ftheNational
Guard. Members
of
each battalionwere tied to each other
not
only
throughheir haredorganizationalffiliation,utalso bythe factthat
they
were
neighbors.
Gould finds hat
insurgents
n
differenteighborhoodsnfluencedach other's egree f
commitment
o
the
nsurrectionhroughhenetworkf inks reated y
overlappingnlistmentsinguard nits]. igh evels f ommitment
n
one
area
enhanced ommitmentlsewhere hen nlistment
atternsrovided
a
conduitfor
communicationnd interaction.
. .
[Thus] neighborhoods
respondedoevents
n
other reaswhere heir esidentserved
n
National
Guard nits.
or
nstance,
esistance
n
the ifth
rrondissement
as
posi-
tively
ffected
y
he act hat
many
f
ts esidentserved
n
the hirteenth
legion, hosemembersemonstratedstrongommitmento hensurgent
effort.
Gould 991, .
726]
Gould concludes
from his
that
cross-neighborhoodolidarity
was a
significant
eature f the Parisian nsurrection.The
interdependence
f
resistanceevels across residential reas was .
.
. intimatelyied notonly
1421
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
13/45
AmericanJournal
f Sociology
to the quantity,
but also to the structure
f overlapping nlistments
(Gould 1991,
p. 727). Gouldthus hows
that
tructural
nalysisneeds to
take into accountnot only ndividual-level ariables suchas thosethat
are employed
n
theMcAdam
studies,but
also the complexnfluences f
multiplex roverlapping
etworks f social
ties.
PositionalAnalysis s a Way
of Representingocial Structure
Many network
nalysts mploy
differentpproachto conceptualizing
social structure;
heir
positional trategy
ocusesupon
thenatureof
actors'
ies
not to one another,
ut to third arties.This strategy
makes
senseofcertainbehaviors nd processesn terms fthepattern f rela-
tions that defines
n actor'spositionrelative
o all other
ctors n the
social system.Positional nalyses
emphasizethe importance
f
struc-
tural
quivalence -that
is,
the
sharing y
two
or
more
ctors f
equiva-
lent relations is-'a-vis
third
ctor-for understandingoth
ndividual
and collective ehavior
see
Lorrain nd White
1971).
The relevant ssue
from his point
of
view
is the specific position
r role that a set
of
actors
occupies
within he system s a whole.
Any
such set
is
termed
block.
An
algebraic procedure
called blockmodeling partitions
overallpopulations ntosets ofstructurallyquivalent ctors Whiteet
al.
1976;
Boorman and White 1976).
Structural
quivalence
modelsdiffer
rom elational
models
n
at least
two crucialrespects. irst,
whilethe atter
ail todistinguishmong
the
membersof social cliques
on the basis of those members'
different
types
of
ties
to
external ctors,
he former
o
concern
hemselveswith
the structure
f the
social
system
s a whole.
They generate
models
in
which n
actor s one of
many
n
a
system
f
nterconnected
ctors
uch
thatall defined elationsn whichhe is involvedmustbe considered
(Burt 1980a,
p. 80). Second,structuralquivalence
models
pay no heed
to whether ctors
n
a givenposition
ave any direct
iesto one another.
A
block, or
set of structurallyquivalent
ctors,from heir
erspective,
may
not be a
densely
knit ocial
clique
at all
(White
t
al.
1976).
One
interestingxample
of network
nalysis
hatmakes use of block-
modeling
nd
positional
nalysis
s
the work
of
Peter
Bearman
1993)
on
local elite
social
structure
n
England
during
he
century
efore he
En-
glish Civil
War. Bearman
examines heactual patterns f
kinship nd
patron-clienties that bound elite actorstogether and that simulta-
neously
rove
hem
part)
during
his
period,
nd
thereby
inducesparti-
tions
or
equivalency
classes
of
these
actors that
describetheir
pat-
terned nteractionsmore
accurately
han
do
the
standard
categorical
classifications.
ather
han
specify
he relevant
liteactors
from far
n
such
categorical
terms
as
middling,
rising, falling,
court,
or
1422
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
14/45
Network
Analysis
country
entry, e generates pecific
lockmodels f their nterrela-
tionships
n
four
iscrete imeperiods etween 540and 1640. Individu-
als with distinctive ersonal biographies ct coherently, e argues,
with
respect o the interestswhich arise from
he structural ositions
they hare n
local and nationalnetworks
Bearman 1993, pp. 11-12).
Changesover time
n
the
structure nd
distribution
f
thesenetworks
thus provide
the
key to understandingong-term tructural hanges.
Bearman's largerargument s that the social
transformationshat did
occur at the elite level created the structural
rerequisites or the
widespread doption f
certain bstract
heological enets,which n turn
would play a crucial
role
n
inspiring
he
tumultuous olitical vents hat
were to come. They led, in otherwords, to the formationf an elite
subworld rganized
n
the
basis
of
socialrelations
hat
transcended
he
[localist nd
kinship-based]raditionalocial
order nd created context
in
which abstract eligious and
constitutionalist]hetorics ould emerge
as
the
critical
eterminantsf elite ocial action
n
the
century receding
the English Civil
War (Bearman 1993, p. 1).
Another ighly ophisticated nd innovative
istorical ase study m-
ploying he
positional pproach o structural
nalysis s John adgett nd
Christopher
nsell's brilliant
work on
the rise of the Medici
in
early
15th-centurylorence. Padgett nd Ansell,much ike Bearman before
them, mploy
he
concept
f structural
quivalence
o
identify
thefam-
ily, economic,
nd patronagenetworks hatconstituted he Medicean
politicalparty
Padgett
nd Ansell
1993, p. 1260),
as well as theirmain
rivals
n
Florentine olitics, he oligarchs. They use archivaldata on
marital,
economic,political, and personal ties to producean overall
relational icture f
Florence's ocial structure, ithin a] 92-family ul-
ing elite Padgettand Ansell 1993, p. 1274).
Throughblockmodeling
they
find that
they
are able to
predict
ctual
party memberships
ar
moreaccurately
han through tandardcategorical
nalyses based,
for
example,
on
class and
status. Indeed,
their
nalyses
of
marriage
nd
economic ies
n
particular rove highly seful
n
revealing
he connec-
tions mong
networks, roups,
nd
party
membership.
n a
pointed
re-
minderof the
potential mpiricalusefulness
f what we have termed
here the
anticategoricalmperative, adgett nd
Ansell
concludethat
rather
han parties
being generated y socialgroups, . . both parties
and
social
groups
were nduced
conjointly yunderlying
etworks.
..
We do notargue . . that social attributes nd groups re irrelevant o
party ormation;
merely
hat their
ole needs
to
be understood
within
deeper
relational ontext. here
s
no
simple
mapping fgroups
r
spatial
dimensions nto
parties;
ocial
attributesnd
group
nterests
re
merely'
cognitive
ategories,
which
party mobilization,
networks,
nd action
crosscut
Padgett
nd Ansell
1993,pp.
12
7-78, 1274; emphasis dded).
1423
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
15/45
AmericanJournal f Sociology
Blockmodeling lso proves useful
n explaining he superior apacity
for ollective ctionoftheMedici
party is-'a-vists rivalfaction, espite
thebitter ivisionbetween hosepatrician lites nd new men aligned
with the Medici. The
Medici party,Padgett and Ansell observe, was
deeplycleaved on two attributional
imensions imultaneously-social
class
(i.e., prestige) nd
neighborhood. ot onlydid the variouscompo-
nentsdespise each other; heydid not run nto each othermuch either.
Only
the Medici
family tself
inked
the segments Padgett nd Ansell
1993,p. 1281). n fact, heMedicifamily xerted n exceptional egree f
centralized ontrol ver ts followers
recisely y bridging he structural
hole betweenpatricians nd new men.
On
the
mportance
f
structural
holes moregenerally, ee Burt 1992].) The more liquish ligarch arty,
by contrast,was constantly
eset
with
cross-pressure[s]
n each
family
instead f collective onvergence, ue precisely o tsfarhigher evelsof
networkmultiplexitynd attributional
omogeneityPadgett
nd Ansell
1993, p. 1279).
Padgett nd Ansellreveal through
n analysis f network ynamics
just
how
these patterns f social
relationships
ame about
in
the first
place. Elite marriage nd
economicnetworks, hey rgue, were re-
configured y working-classevolt nd
by
wartime iscal
risis,respec-
tively Padgett nd Ansell 1993, p. 1287).As partof their trategyf
reconsolidation
n
the aftermath f the
Ciompi wool workers' evoltof
1378, the temporarilyictorious
ligarchs eliberately xcludedthe
os-
ers'
elite
collaborators, ncluding
he
Medici,
from heir
marital
net-
works. The
oligarchic lique
and
the Medicean . . .
networks,
laim
Padgett
and
Ansell,
thus
both
emerged
n
tandem,
a
single network,
each reflexively
nd
asymmetrically
tructuring
he other
Padgett
nd
Ansell 1993, p. 1298). n the
meantime, oreignwars withMilan (1423-
28) and Lucca (1430-33) had theeffect fseverelyweakening heFloren-
tine
conomy
nd of
sendingmany
lite
families
with
he notable
xcep-
tion of the Medici
themselves,
who
enjoyed special papal ties)
into
bankruptcy.
he elite
families
ttempted
o
help
themselves
y depriving
in
turn he socially
nd
politicallymore vulnerablenew mercantile le-
ments-the new men-of
offices nd tax revenues.
After
427,
the
Me-
dici
began
to
pursue
conomic elations
with
hese
new elements n their
local residential eighborhoods.t was preciselyt thatmoment hat he
Medici
partyemerged
s a
powerful
nd
self-consciousoliticalactor
engaged n struggle gainstthe dominant ligarch action.
THREE
NETWORK MODELS OF HISTORICALEXPLANATION
Each of thesetwoalternativemodelsof social
structure
as
its strengths
and
weaknesses.
Proponents
f
the relational
pproach emphasize
ts
1424
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
16/45
Network
Analysis
suitability ormappingthe typicalrelations hat ndividuals
have with
one another. t is more
amenable to traditional urveyresearch ech-
niques, they laim, ncontrast o positional pproaches hatrequire ata
for
ll the elements
n
the
system. roponents f structuralquivalence,
conversely, tress ts moreconsistentlytructuralistature, ts
capacity,
that is, simultaneously
o take into account all of the
relationaldata
pertaining o
a
givenactor, ncluding
his
patterns fexternal
elation-
ships- the relations n
which
he is
involved s well as therelations
n
which
he
is
not involved
Burt 1980a, p. 131).
A
global [positional]
approach,
notes Michael
J.
Mandel
(1983, pp. 376-77), examines ll
interlocks etween oles t
the ame time ... [This]
imultaneous onsid-
eration f all actors n an entire opulation .. spotlightshe nterdepen-
dence of the differentolesfound
n
thepopulation.
Each role
therefore
has built nto t the
elementswhichdifferentiatet from ther oleswithin
the same overall tructure. ositional
nalysts
ontend hat
hisfeature
lends theirmodels significantlyreater egree f predictive
ower. For
example,
n
a reconsiderationf a
classic cohesion tudyby
Coleman,
Katz,
and
Menzel (1966),Burt shows how the structural quivalenceof
physicians etter ccounts
forthe informal ocial pressures hat ed to
the diffusion
f
the use
of
the
drug tetracycline
han
does their ocial
cohesionBurt 1987).
Network
practitioners
mploy
oth
of
these
pproaches
o
make sense
of
an
impressive rray
of
substantive ssues.
Despite
the
considerable
methodological sefulness f
both strategies, owever,we contend hat
each can be
subjected
to a number
of
important
riticisms. he
very
assumptions nd premises hat have
made network nalysis
n
both
of
its
methodological
ariants uch a powerful ool raise seriousquestions
about ts
adequacy
as an
overall
research
trategy.We contend hat
here
have been three
models mplicit
n the
iterature n network
nalysis-
models
f
the
relationshipsmong ulture, gency, nd
social
structure-
that
have
led to
varying
egrees
f
difficulty
n
elaborating atisfactory
explanations
f
historical
rocesses.
The
first
f
these
mplicit
models,
thatof structuralist
eterminism,eglects ltogether
he
potential
ausal
role of
actors' beliefs,values, and normative
ommitments-or,more
generally,
f the
significance
f
cultural
nd
politicaldiscourses
n
his-
tory.
t
neglects
s well thosehistorical
onfigurations
f
ction hat
hape
and
transform
regiven
ocial
structures
n
the first
lace.
A
second and
more satisfactory-butstill deeply problematic-approach is that of
structuralist
nstrumentalism.tudieswithin his
perspective
ccept
the
prominent ole of social
actors
n
history, ut ultimatelyonceptualize
their
activity
n
narrowly tility-maximizing
nd
instrumental orms.
And
finally,hemost ophisticatedetwork
erspective
n
social
change,
which
we term tructuralist
onstructionism,
hematizes
rovocatively
1425
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
17/45
AmericanJournal f
Sociology
certainhistorical rocesses f
identity onversion
nd robust
ction.
It is themost uccessful f ll ofthese pproachesnadequately onceptu-
alizing human agency nd thepotentiallyransformativempactof cul-
tural idioms
and
normative ommitmentsn
social
action.
However,
even thisperspective alls hort
n
understandinghe full omplexitiesf
the theoretical
nterconnectionsmongculture, gency, nd
social
struc-
ture. It too pays insufficientttention
o
the
structuring
nfluences f
cultural nd political
discourses pon historical ctors.8
Structuralisteterminism
All threeof these basic approaches are represented y practitionersf
bothrelationalndpositional
etwork nalysis.
The
aforementionedela-
tional
tudyby
Rosenthal t al.
(1985),
for
xample,
tands s
an
illumi-
nating ase study n
structuralisteterminism. osenthal nd herassoci-
ates
delineate
hree
istinct istorical
eriods
f women'sreform
ctivity
in New York State between he years 1840 and 1914.The first fthese,
theyclaim,
lasted from1840
to the
late
1860s; t was
dominated
by
a
women's
rights
movement nd also featured
igh
evels of nvolvement
in
antislavery nd temperance rganizations. he second period was a
transitional ne; betweentheend of the CivilWar and the late 1880s
there
was
comparatively
ittle
ctivity,many
of the
earlier
roupsdisap-
peared,and the
possibility
f
creating
ew national
organizations
was
limited
Rosenthal t al. 1985, p. 1044). And finally, etween he ate
1880s nd 1914,therewas a
resurgence
f
ntense
eform
ctivity, enter-
ing around an increasing
umber f new organizationsinkedprimarily
by
the
suffragessue. Rosenthal t
al.'s
delineation
f
these hree eriods
of
women's
reform
ctivity y
means
of
relational
nalysis urely
anks
as a significantnd worthy ontribution. ut its limitations re also
considerable: hestudyprovides ittle
ystematic xplanation
s
to
pre-
ciselywhy
hese
hanges
ccurred rom
ne
historical
eriod
o the
next,
settling
nstead for
a succession f static
map configurations
r rela-
tional
snapshots '
f network
atterns.
The individualand social ac-
tions
that
ed from
ne structural
onfiguration
f
reform
ctivity
o the
next re
leftunanalyzed,
s
are the
developments
n
social structure
nd
8
Our term tructuralist onstructionisms coincidentally eminiscent f Pierre Bour-
dieu's phrase, constructivist tructuralism Bourdieu 1990). But we do not imply
by
this
any direct connectionbetween the network nalysts whom we are discussing
and Bourdieu, although t is true that these various thinkers ll share an underlying
concern
to overcome at both
the
theoretical
nd
empirical evels
the
dichotomy
be-
tween subjectivist and objectivist standpoints. Bourdieu's understanding of
fields, e.g., does bear striking nalytical affinitieswith that of social networks
(Bourdieu and Wacquant 1992, p. 114).
1426
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
18/45
Network
Analysis
cultural nd politicaldiscourse hat underlay nd
motivated hem. At
best, Rosenthal t
al.
treat hese
various
developments
n
their nalysis
as exogenous ariables.
On
the ide of
positional
ather hanrelational
nalysis,
he
early
work
ofHarrisonWhite nd his associates n blockmodeling
echniquesWhite
et
al. 1976; see also Boorman and White 1976) provides nother llumi-
nating xample of structuralisteterminism.9 hite's
research ncludes
studies
of network
ata
describing tructural hangesover time
White
et al.
1976, pp. 763-68); each of these studies,however like that
of
Rosenthal t al.), ultimatelyulminatesn a succession f static epresen-
tations f social structure.
ach manages, hat
s
to
say,
to
generate
no
modelsof processes ver time. Whiteet al. recognize hat transforma-
tions
n
social structuretillneed to
be
explained: Models of structure
are not
sufficientntothemselves. ventually ne mustbe able to show
how concrete ocial processes nd individual
manipulations hape and
are
shaped bystructure.... One fundamentalroblem ere
s
that
many
settingsmay
admit not
ust
a
single quilibrium utcome,
but
multiple
alternative quilibria. .
. In
turntheinteresting
uestionsmay
bear
on
what external
orcesmay
cause a social structure o
pass
from ne
equilibrium onfiguration
o
another
White
t al.
1976, p. 773; empha-
sis added). Toward the end of anothermportantrticle n blockmodel-
ing techniques nd role structures,White and his collaborators urther
acknowledge
hat the next
analytic
ask is
to provide
ways
to
probe
how
role structures
. .
actually ome ntobeing
Boorman nd White
1976, p. 1442). But again, theyprovideno systematic
ay
of
building
concern
or
human
agency
nd
processualityntotheir xplanations.
Another hortcominghatboth fthese xamples f
structuralisteter-
minism ave
in
common
s
theassumptionhat ocial networks an best
be
conceptualized
s
linking ogether
concrete ntities uch as
persons
and
organizations,
ather
than
as
also
embodyingdeals,
discursive
frameworks,
nd
cognitivemaps. The latter, rom heir
perspective,
remain mere abstractions. '0 tructuralist eterminism
ests analyti-
cally
on
a
reification f
social
relations;
t
transformshe
mportant
heo-
retical distinction etween a structure f social relations, n the one
hand,
and
cultural ormations,n theother, nto n
ontological ualism.
It thereby ruthlessly bstracts he formal r
objective
dimensions
of
social relations rom heir
ultural
nd
intersubjective
ontexts o as
to be able to representnd analyze such relationswithsophisticated
9
We
discussbelowtherather ifferent
nalytical erspective
hat
White
has
subse-
quently
dopted-e.g.,
in
Identity
nd
Control1992).
10
For a
telling xample, ee
Granovetter'synopsis f Cambridge niversityress's
series n
Structural nalysis
n
the
Social Sciences Knoke 1990,frontmatter).
142
7
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
19/45
AmericanJournal f Sociology
technical ools;
n
the process,however, t drains uch relations
f
their
active, subjectivedimension nd their ultural ontents nd meanings.
To compound heproblem, hisvariant fnetwork nalysis hen heoreti-
cally privileges ne side of this dualism-namely, that of social rela-
tions tselfalbeit ocial relations ow blanched f boththeir ctive and
symbolic spects)-over that of cultural nd discursive ormations.
t
uses social networks social being,
n Marx's
famousformulation)o
explain social consciousness
nd
culture,
ut not
also)
the
otherway
around.
ndeed,
in
Wellman
nd
Berkowitz'swords
1988, p. 5), sym-
bols, meanings, nd values . .
.
are a derivative nd oftenresidual
concern.)
Structuralist
nstrumentalism
Perhaps because of its theoreticalimitations,tructuralisteterminism
has givenrise o relativelyittle mpirical esearch n historical rocesses.
Far more commonhas been
the
perspective f structuralistnstrumen-
talism,
which has also
been adopted by practitionersf both relational
and positional etwork nalysis.
This
perspective ertainlyakesthe his-
toricalrole
of
social actors
very arefully
ntoaccount.
However,
t also
draws mplicitly pon residual ategories rom utside ts ownconcep-
tual framework-in particular, model
of homo economicus-for ex-
plaining heformationnd transformationf social networks hemselves.
There s a strikingendency mong tructuralistnstrumentalists,n fact,
to
smuggle conceptions
f
agency
nto
their
nvestigations,
hether
overtly
r
covertly,
rom he domainof rational hoice
theory.Many,
f
not most,
uch
network nalysts ssume unproblematically
hat ctors-
individuals and even groups or organizations-are utility
maximizers
who
pursue
theirmaterial
nterestsn
money, tatus,
nd
power
n
pre-
cisely he ways predicted y
theorists f
rational
hoice.
n
effect,
hese
analystsprojecttheirown anticategoricalism
n the
actorsthey tudy,
neglecting ow the latter'sown cultural nd moral categorieshelp
to
structureheir
beliefs nd behaviors.
A
useful xample
of structuralistnstrumentalism
n
the
side
of
rela-
tional
analysis
are
Roger
Gould's aforementionedtudies
of the 1871
Paris Commune 1991, 1992). Gould begins by posit[ing]n influence
process
n
which a [Parisian]district's esistance evel
is a
function f
a
set of exogenousvariables and ofthe resistance evelsofall theother
districts,weighted y
the
strength
f
ts inkswiththem
Gould 1991,
p. 721; emphasis n original).He investigates hree such exogenous
variables that
help
to
explain
these
varying
evels
of
resistance:
1)
the
levels
of
poverty
n
an arrondisement;2)
the
percentage
f
skilled
ala-
riedworkers esiding herein; nd (3) thepercentage
f
white-collar,
id-
1428
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
20/45
Network
Analysis
dle-class mployees ocated nany specific eighborhoodGould 1991, p.
723). Resistancewas stronger,
oncludesGould, in areas that were
poor and working lass, althoughhis expectation hat white-collar
and unskilledworkerswouldplay ess prominentoles n the nsurrection
is not supported Gould 1991,
p. 725). Gould departs here fromhis
own
anticategorical pproach byrelying pon exogenous gradational
measures f poverty nd occupational omposition-in addition o social
networks-in order o make
sense of the mobilizationn each arrondise-
ment see also Gould 1992, p.
727).
Not
only
mustGould move from
elational o
gradational
ata
inorder
to identifyccurately he social base of the Paris Commune,but, more
troublingly,e neverprovides plausiblecausal account s to whyPari-
sians would have risked heir
ives for he Commune n the first lace.
At timeshe seems imply o
assume,dubiously n our view, thatworking
people or, at least, working-class
ales of 19th-centuryrance) would
automatically ight nd die for state representingheir lass interests.
He predicts, or xample, hat
white-collarmployees, who often ame
frommiddle-class amilies,would have been ess ikely hanartisans r
workers o
participate
n
the nsurrectionGould 1991,p. 724). As noted
above, this expectation s
unsupported y the data, an outcome that
Gould himselfneverexplains.At othertimes, Gould emphasizes the
importance
f
neighborhoodand cross-neighborhood)oyalty r solidar-
ity.
n
this account, he argues
that Parisians oined the nsurgency ot
because of
class interests er se, but rather ecause of social
pressure
fromneighbors: Failure to
participate n the nsurgent ffort as con-
strued
s a
betrayal f oyalty o the
neighborhood
nd was sanctioned
accordingly Gould 1992, p.
748).
This
argument, owever,
fails
to
explainwhy
or
how certain ies nd interactions
mongneighbors,
hich
Gould does
not
specify oncretely, enerated
uch
a
powerful eighbor-
hood loyalty n the first lace; it begs the questionas to how certain
(unspecified)
orkers
ame
to
believe,
and
apparentlymanaged
to con-
vince
others,
that
neighborhoodoyalty required nothing
ess than
risking
heir ives for
the Commune. Gould's
assumptions
bout
the
purely
nstrumental
oundations
f
political
mobilization-whether lass
or
status based-in the
present
tage
of his
research
revent
him
from
analyzing
uch cultural nd normative nfluences
n
a
fully atisfactory
manner.
Gould implies hat
simply elonging
o a
neighborhood
ith
a
certain ccupational ompositionand tiesto other uchneighborhoods)
producedpro-Commune
olidarities. ut why did such belongingness
generate owerful olidarity
nstead
of,
for
xample, nterpersonal
ndif-
Goulddoesnot aymuch bout he omplexlyenderedharacterf
he nsurgency,
simply oting hatwomenwerenot dmittednto heNationalGuard.
1429
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
21/45
AmericanJournal f Sociology
ference r even
cutthroat ompetition,s amongthe oligarchic lique
in
Padgett nd Ansell's study?
We recognize hatGould'sworkhas thusfar ppeared only n ournal
articles nd that ater,more
expansiveversionsmightwell exemplify
differentortof explanatory trategy.
n
its present orm, owever,we
claim
that it does relatively ittle o
explore
the
specifically
ormative
commitments
f
the social actorsengaged
n
politicalresistance uring
the Commune.
The difficultyere
s not
that Gould
neglects culture
per se, but rather hat heunderlyingogicof hisargument-its theoreti-
cal
logic-fails to accord normative
ommitmentsny independent x-
planatory ignificance.)n a more omplete istorical xplanation,Gould
would have to direct ar moreattention han he does to thespecifically
cultural ases
of
cross-)neighborhoodolidaritynd their nfluencepon
individuals'projects f action.
Exactlywhat sorts f practices nd ritu-
als,
he
would need to ask, produced hepowerfulolidarities f Parisian
neighborhoodsnd the National Guard?Whatrole did political deolo-
giesand cultural iscourses lay n
sustaining r even expanding elations
of
solidarity?Gould mentions
he discourses
f socialism nd
of re-
publican patriotism
n
thiscontext, ut
he
neverexamines heirmean-
ingsto
workers
nd thus heir ausal
significance
n
any systematic ash-
ion.) And finally, hat do sources f nformationuch as diaries, etters,
union
records,
nd
journalistic ccounts
reveal
to us about
popular
cul-
tural
practices
n
19th-century
aris and
during he Commune tself?
(See, e.g.,
Sewell
1980]
nd the sources
ited
n
Edwards
1973, p. 175].)
Such
data
sources
would demonstrate ow
popularpractices
nd
popular
culture,
arfrom
eing
additional actors r forces
eeding o be either
controlled or or examined
alongside
of
network tructures, ere
in
fact
mportant
imensions f
those tructures
hemselves-grounded n,
and sustaining, pecific,dentifiableocialties.
On the side
of
positional nalysis,
PeterBearman
1993), too, adopts
the
perspective
f
structuralist
nstrumentalism.earman
tells
the
story
of
changing
lite
social relations
n
England during
he
long
sixteenth
century y
means of
a
sequence
of
blockmodel
epresentations
f
elite
social
structure.
y examining
hese
blockmodels,
e
specifies
the
slow
and arduous
processby
which
religious
eterodoxy
as embedded
n
the
fabricof
local elite social
and
political ife,
the
tangible
mechanisms
by
which ocal elitescame
to perceive hemselves nd others s actors
whose activitywas of religious ignificance Bearman 1993, pp. 171,
132). Bearman's approach s not
dissimilar
o
what Max Weber 1949)
long ago
termed he
analysis
of
electiveaffinity -the tudy
of
how
particular
discourses nd
culturalformations
ome to find a
match
and to
resonate with
specific istorically
mbedded ctors. Where
his
approach
differs
rom
hat
of
Weber,
however,
s in its
tacit nstrumen-
1430
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
22/45
NetworkAnalysis
talism-that is, tstendency o devote lmost ll
of ts nalytical ttention
to uncovering he structural reconditions or this elective affinity,
rather han to also exploring he ndependentausal significancef these
discursive rameworkshemselves. ndeed, Bearman
seems to attribute
littlemorethan purelymaterial nterestsinmoney, tatus, nd power)
to the historical ctors at the centerof his account. The eclipse of
localism, the decline of kinship, nd the emergence f a national elite
subworld, he argues, were aggregate utcomes
f the gentry's ursuit
of ocal
status
nd
power Bearman 1993, p.
3). The searchfor politi-
cal
advantage
was the
drivingmechanism,
n
otherwords,behindthe
crucial historical ransformationsf the
period-and
not (also) the be-
liefs, values, and normative ommitmentsf these elite actors them-
selves. Bearmanmightwellhave noted,
n
theopening age of
his
study,
that he
is primarilynvestigatingnly one side
of
the
interaction e-
tween
structural nd ideationalprocesses .
. the complexrelationship
between ction and structure
Bearman 1993,p. 1).
Structuralist
onstructionism
Several networknalystshave, inrecent ears,developedmore ophisti-
cated
approaches han these o studying istorical rocesses, pproaches
that
take
into account
culture
nd
agency
as well as social structure.
Again, practitionersf both relational nd positional
nalysishave pur-
sued such
nvestigations. ne revealing xample
f tructuralistonstruc-
tionism n the
side of
relational
nalysis
s
Doug
McAdam's recent
work
on Freedom
Summer
1986, 1988).
McAdam
explicitly
ejects
many
of
the
instrumentalistlaims
of
othernetwork nalysts.
He
qualifies
the
notion, specifically,
hat
structural vailability
for social
movement
participation enders attitudinal ffinity o a movement ompletely
irrelevant. uch a notionmighthold true of
low-risk/costctivism,
but
participation
n
nstances f
high-risk
ctivism
such
s
theFreedom
Summer
roject nd,
we
might dd,
theParis Commune nd the
English
Civil
War]
would
appear
to
depend
on an intense ttitudinal
nd
personal
identification ith
the
movement McAdam
1986, p. 73). Surely,high
levels of such
dentification ere required
or he FreedomSummer
ol-
unteers o
aspire
to
participate
n such a
demanding
nd
potentially
an-
gerousundertaking; oreover,hevolunteers' ccounts ftheir wn mo-
tives n
open-ended pplication uestionnaires
learly
demonstrate hat
they
elt
deep-seated
dealism nd a
strong
ommitmento the
project's
goals.
The
real
question
s:
Were
the volunteers'
rior
ttitudes
uffi-
cient
n
themselves o accountfor heir
articipation?
y
answer
here s
a
qualified
no.... Attitudinal
ffinityas
well
as] biographical
vailabil-
1431
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
23/45
AmericanJournal f Sociology
itymustbe considered ecessary ut not ufficientauses of participation
in high-risk/costctivism McAdam 1986, pp. 73, 87).
Exactly where, then, did this attitudinal ffinity or the Freedom
Summer roject or, moregenerally, or ny social action) ctually
ome
from?One of McAdam's major nnovations n Freedom ummer 1988)
is to elaborate n implicit heory f identity onversion hat takes seri-
ously the formation f motivations nd identitieswithout acrificingt
all the moment f structuralocation. McAdam argues that the key
to his account ies
in
those
organizations
hat drew the
applicants
nto
civilrights ctivity eforereedomSummer.... Extremely isky, ime-
consuming
nvolvements
uch as
Freedom
Summer,
he
contends,
are
almost lways precededbya seriesofsafer, essdemandingnstances f
activism. n effect, eople commit hemselves o movementsn stages,
each
activity reparing
he
way
for
he
next
McAdam 1988, pp. 50-51;
emphasis
n
original).
ndividuals
irst
ome
nto
ontact
withmovement
participants
nd
engage
in
discussions
nd
joint
activitieswith
them,
confronting
irsthandhe
ssuesthat hemovement
as
set
out to address
and
gaining
n the
process deeperunderstanding
f
and moral
ommit-
ment o its goals. Then, eventually, at the
evel of
dentity, heybegin
to
'play
at'
and
[to] grow
more comfortable
ith the role of activists
themselves McAdam 1988, p. 51). Unlike many othernetwork na-
lysts,McAdam recognizes hat ctors an undergo ar-reachingrocesses
of
dentity ormation
n
the course ftheir nvolvements
n
extraordinary
affairs. uch an
insight
s
especially mportant
o bear
in mind when
analyzing
heir
participation
n
high-risk/cost
ctivities
uch as Free-
dom
Summer-or, for that matter,
n
any major social
or
political
movement.
In a recent ublication,McAdam concludes hat network heory ails
to offer
plausible
modelof ndividual ction nd therefore
convincing
mechanism ywhich nterpersonalontacts ndorganizationalmember-
ships
draw
individuals nto
activism
n
the first
lace (Friedman
and
McAdam 1992, p. 160). Friedman nd
McAdam claim
nstead hatnet-
work
theory
an
yield
robust
xplanations
f
collective ction
only
when
synthesized
ith
a
modified
ational
hoice
model of individual
ction,
one
that views collective dentities
and
not
ust
material
esources)
s
potentially owerfulncentives or action: One of themostpowerful
motivators f individual
action, theywrite,
is
the
desire
to
confirm
through ehaviora cherisheddentity.
.
. Integrationnto [activist]
networks
makes
t more
ikely
hat he
ndividualwill value the
dentity
of
'activist'
and choose
to
act
in
accordance with it
(Friedman
and
McAdam
1992, pp. 169-70).
It is our contention hat the distinctive
contribution f McAdam's work
is, indeed,
to
expand
the
concept
of
purposive ationality
tself o the
point
of
bursting hrough
he
seams of
1432
This content downloaded from 148.206.159.132 on Fri, 12 Dec 2014 20:29:50 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp -
8/10/2019 Emir Bayer 2782580
24/45
Network
Analysis
standard
ational
hoice heory.t
is
historical ctors' pecifically orma-
tivecommitments,ather han or
n addition o) their ursuit fmaterial
goals,thateffectivelyrives heir ocial movement articipation.
If
there
s
a weaknesstoMcAdam's analysisof Freedom
Summer,
t
lies in his insufficientttention
o precisely his element f normative
commitmento cherisheddeals. In
one
of
their rticlesFernandez
and
McAdam
(1988)
notethatrecruitment
ontexts,
the residue f
a
protest
culture, ffected the number
nd
form