Managing Networks: Propositions on What Managers Do and Why They... Author(s): Michael McGuire Source:

Managing Networks: Propositions on What Managers Do and Why They Do It
Author(s): Michael McGuire
Source: Public Administration Review, Vol. 62, No. 5 (Sep. - Oct., 2002), pp. 599-609
Published by: Blackwell Publishing on behalf of the American Society for Public
Administration
Stable URL: http://www.jstor.org/stable/3110019
Accessed: 13/03/2009 22:25
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=black.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].
Blackwell Publishing and American Society for Public Administration are collaborating with JSTOR to
digitize, preserve and extend access to Public Administration Review.
http://www.jstor.org
MichaelMcGuire
of NorthTexas
University
Managing
Networks:
Managers
Do
and
Propositionson
Why
They
Do
What
It
Measuringmanagementin networksis difficultbecause the allocationof managerialresourcesin
networkstructures
is fluid-that is, theutilizationof managementbehaviorsvariesacrosstimeand
space withina given programor project.As a means of focusingthe networkmanagementresearch agenda, propositionsbased in contingencylogic are suggested to test ideas regarding
when,why,and how networkmanagersundertakethesebehaviors.Thepropositionsare intended
to identifythe vastinventoryof networkmanagementbehaviorsand, mostimportantly,
determine
how the manager strategicallymatchesbehaviorswith the governingcontext.Suggestionsare
also offeredto help us understandhow and why managerialresourcesare reallocatedover time
and space. Theproposed researchagenda is offered as a guide to help us determinewhich
choicesare mostlikelyto be effective.
"The study of governanceneeds its bestseller with
snappyaphorismsand vivid stories."
-RA.W. Rhodes
In the midst of remarkableincreasesin the quality and
quantityof researchin public network management,the
story of the public managerin such contexts has yet to be
adequatelytold. Where is the "day in the life of a public
networkmanager"fable to teach us the basic how-tos of
networkmanagement?Where are the textbookswith fullcolor pictorials, biographical accounts of highly skilled
networkmanagers?Wherecan we find documentationof a
managermuddlingthrough,gropingalong,or fallingasleep
at the wheel in networksettings?As moreattentionis given
to governing structuresthat are multiorganizationaland
multisectoral,this aspect of public managementcould use
some simple,richdescription.In the passagequotedabove,
ProfessorRhodes(1997, xv) writesconcisely,andonly partially in jest, that "all too often academicsmake maps of
complexity,insisting that complex problemsrequirecomplex solutions"that stand in contrastto the "snappy'ten
commandments'of the latestmanagementbestseller."However,it is difficultto discussclearlyandconciselythatwhich
is complex and multifaceted.The practice of managing
across governmentsand organizationsoutpacesempirical
descriptionandtheoreticalexplanation.We do not yet have
the vocabularyand imagery to tell meaningful stories of
managementsuccess in networksettings.
This articlemay fail the "tencommandments"test, but
its intentis to simplifyandfocus researchon networkmanagement.It discussesthe challengesof measuringthepublic
manager'scriticalactions andbehaviorsin models of program effectiveness in networkpolicy making and administration.The basic premise is that a research agenda in
networkmanagementmust include threecomponents:(1)
a descriptionof the behaviorschosen by the networkmanager; (2) an explanation of why managers make such
choices; and (3) an evaluationof these choices. Making
the importantcase thatnetworkmanagementhas become
a criticalactivity in public administration,recentresearch
has emphasizedthe thirdcomponent,theorizingaboutthe
meaning of effectiveness in networksand testing models
of effectiveness with empiricaldata. In such valuable but
incompletemodels, networkmanagementhas become the
ultimateindependentvariablein causal modeling of programeffectiveness, but the specific behaviorsof the public manager are not properlymeasuredin these models.
Although advanceshave resultedfrom the more sophisticated studies (Meier and O'Toole 2001; Provan and
MichaelMcGuireis an associateprofessorof publicadministration
at the
of NorthTexas.Hisresearchon interorganizational
networks,colUniversity
laborativepublicmanagement,and economicdevelopmenthas been publishedin manyjournals,includingPublicAdministration
Review,Journalof
PublicAdministration
Researchand Theory,State and LocalGovernment
Review,Publius:The Journalof Federalism,and EconomicDevelopment
Quarterly.Email:[email protected].
on WhatManagersDoandWhyTheyDo It 599
ManagingNetworks:Propositions
Milward1995),empiricalresearchheretoforehasneglected
to describe and explain how particularmanagementresources are employed in network structures.In researchdesign terminology,a premiumhas been placed on establishing internal validity, without a similar concern for
measurementvalidity.The discussion in this article is intendedto provideone roadmap for identifyingandunderstandingthe behaviorsof the networkmanager.
Assumptionsand Definitions
Several assumptionsand definitions must be stated at
the outset. First, the term "network"is used in this article
to describepublicpolicy makingand administrativestructuresinvolvingmultiplenodes (agenciesandorganizations)
with multiple linkages. I use the term not just to describe
social networksor informalpatternsof interaction,but to
connote structuresthrough which public goods and services areplanned,designed, produced,anddelivered(and
any or all of the activities). In keeping with recent public
administrationresearch,I do not definenetworksas purely
sociological phenomena.Instead, I use the network as a
metaphorto describejoint situationsin which more than
one organizationis dependenton anotherto performa task.
My intentis not to apply the concept of the social network
to public administration,butto modify ourmodels of public administrationand managementto apply to networklike settings.
Second, I assume that governmentis properlyviewed
as a steererof policy making and execution,reflectingthe
desires and demands of its citizens. My use of the term
"networkmanagement"still conveys the steeringfunction
of government,but in a way that is wholly differentfrom
command-and-controlprocesses within a bureaucracy.In
practice,networksmay rely on variousleaders at various
times performingvarious roles, all of which may be necessaryfor networkeffectiveness.However,my hypotheses
assume the typical situations in which governmentultimately is held accountablefor the satisfactorydelivery of
public goods and services. Therefore,the network manager in the situations assumed here is the government
managementcharged with completing such a task, who
must do so with and throughnetworkedsettings. Public
managerscan't commandaction in networks,but they are
still responsiblefor their outputs.
Third,unlike many studies of networksand collaboration, my focus is on managingthe networkas a network.
The first-stage, operations-level management behaviors
examined in this article are conducted to achieve a particularpurpose;the hypothesesraisedherearenot intended
to directlyaddressthe networkpurpose.In this regard,the
behaviors can be viewed as analogous to-but certainly
not the same as-behaviors and tasks thatconstituteorgaReview* September/October
Administration
600 Public
2002,Vol.62, No.5
nizational behavior theory. For example, in order for an
organizationto produceits goods andservices,it musthave
the most skilled staff.The staffmustbe motivatedto maintainor increaseproduction,the organizationmustbe structuredin such a way that productionis efficient and effective, and the development of interpersonalrelationships
must be fostered. The manager'sjob is to make this happen. Similarly, any government manager charged with
achievinga goal throughnetworksettingsunderstandsthe
most criticalactivitiesinvolve operations-identifying and
working with the properplayers and resources, keeping
the players committed, defining the roles of the players,
and facilitatingeffective interactionamong the players.A
networkmanagercertainlycannotcommandthese to happen, as a managerin a single organizationcan do, butthey
still must happenfor networksto be effective. Thus, I delimit my analysis to exploringwhat managersdo to operate in effective networks.
TheResearchQuestion
I submitthat the most importantnetworkmanagement
researchagenda involves answeringa simple but elusive
empiricalquestion:Do the actions of a managercontribute to the effectiveness of multiorganizationalarrangements, and if so, how? Governingnetworksdo not simply
emerge spontaneouslyas self-sufficient, automatedentities. There is both an operationaland a strategiccharacter
to networksthat depends largely on the actions of a manager.Recently adoptedmeasuresof networkmanagement
typicallyinvolve countingcontactsandinteractionswithin
the network (Agranoff and McGuire 1998; Meier and
O'Toole 2001; Provanand Milward 1991, 1995). The use
of these preliminary measures in network studies has
opened the floodgates, so to speak, and provided a welcome stream of knowledge regardingthe scope and impact of multiorganizationalstructures,but they fall short
of accuratelydepicting a networkmanager'sbehavior.If
scholarshipultimatelyis intendedto informaction,to provide guidanceto managersoperatingin a complex andrapidly changingenvironment,this empiricaloversightmust
be corrected and next-generation measures of network
managementmust be conceived.
Measuringmanagementin networksis difficultbecause
the allocation of managerialresources in network structuresis fluid, thatis, the utilizationof managementbehaviors varies across time and space within a given program
or project.A managermay believe it is necessaryto stabilize processes within a network-say, by fostering agreement on the rules of interactionamongthe playersor gaining consensuson a commonlanguageto use duringnetwork
operations.But emphasizing stabilizationmeans that, in
that particulartime and place, less attentioncan be given
to exploiting opportunitiessuch as expandingthe size of
the networkor engaging stakeholdersto supportthe aims
of the network.The distributionof managementresources
expendedat a given point in time will vary,makingobservation and identificationproblematic.
As a means of focusing the network managementresearch agenda, I suggest hypotheses to test ideas regarding when, why, and how network managersbehave. The
hypotheses are drawnfrom interviewswith local network
managersand the rapidlyexpandingliteratureon network
management.
TheExample
An example of a very common scenario illustratesthe
complexity and difficulty of measuringnetworkmanagement. One local agency in North Texas is charged with
providingservicesto personswithmentalillness andmental
retardationas partof the Texas mental health and mental
retardation(MHMR) system. The overridinggoal of this
particularagency is to respondto the needs of adults and
children with mental illness, mental retardation,autism,
pervasivedevelopmentdisorder,or substanceabuse,as well
as the needs of their families, by providing quality services. The centerhas contractswith a vast numberof providers thatdeliver services to its consumersandhas established working relationships with no fewer than 31
governmentaland nongovernmentalorganizations.Funding for the center'sactivitiescomes frommultiplesources:
home and community-basedservices contracts,grantsfor
operating expenses from the state MHMR agency, new
generationMedicationfunding,Medicaid, child and adolescent mental health block grants, in-home and family
services, Medicare,and funding from three local governments. Managersfrom the MHMRservice arrayshave recently experiencedthe opening of two new clinics, an additional office, a merger,and a majorrestructuringof the
statefundingsystem. Since the summerof 2000, the home
and community-basedservices programfor persons with
mental retardationhas experiencedfunding reductionsof
approximately25 percentwhile realizing a 39 percentincreasein the numberof people served.Demandfor MHMR
services has far surpassed the center's capacity, while
centerwide funding from the state of Texas has been reduced every year.Uncertainfundingfrom the federaland
state government,a continual shortage of Medicaid providers, increasingcosts of utilities and housing, and other
significantshocks to the agency have resultedin many incrementalandseveralfundamentalchangesto the network
structureandoperations.Even in suchenvironmentalcomplexity, however,evaluationsindicatethe vast majorityof
consumers, family members,communityvolunteers,and
representativesof variousgroupsare satisfiedwith the per-
formance of center activities, rating the effectiveness of
the networkas high or very high.
In this scenario,how do we measure"networkmanagement"? Considerthe multitudeof activitiesperformedby
managers.To meet the demandfor expandedmentalretardationservices,the managerexpandedthe networkof providers and reconfiguredthe roles of existing providers.
After the agency initiated person-directed planning-a
process for developing service plans from the personal
outcomesdesiredandidentifiedby the individuals-managers worked with providersto assist them in realigning
their service approachwith the agency's new philosophy.
When a committeeof volunteersdrafteda plan calling for
improvedinternaland external communication,network
managers enhanced informationexchanges and utilized
informationmore effectively as a means of improvingthe
quality of interactionwithin the network.As the state reduced the center's funding, managers solicited financial
resources from previously inactive community representatives and mobilized supportfrom previouslyuncommitted players in the community.
The only thing extraordinaryaboutthis example is that
it is not extraordinary;environmentalshocks such as those
experiencedat the agency are common in many policy areas. As in many networks,over a shortperiod of time, the
allocation of managementresourcesvaried considerably.
Some behaviors necessarily preceded other behaviors,
while otherswereundertakenin responseto abruptchanges
in the programenvironment.How do we accountfor these
numerousactions andbehaviorsin our models of network
management?At what point should we capturethis activity and assign a value to it? For pedagogical and theoretical reasons,it is not desirable,even if possible, to consider
managementin this regardas a single set of efforts or a
sum of managerialbehaviors.A simple count of the number of contacts with a predeterminedlist of possible networkmemberswill not-indeed, cannot-adequately capturethe extent to which strategicand competentaction by
the networkmanager(s)contributesto the success of the
network. It is even conceivable that such a count would
severely underestimatethe scope of managementactivity,
since the quality of interactionmay have changed more
significantlythan the quantityof interaction.The limitations of applying our currentmeasures of network managementeven to a typical networkare obvious.
TheFluidityof NetworkManagement
Management Behavior
Network managementis an elusive target to properly
measure. The allocation and utilization of management
resourcesexpendedis fluid-it variesacrosstime andspace
within a given programor project. Managingin network
Networks:
onWhatManagers
DoandWhyTheyDoIt 601
Propositions
Managing
structuresinvolvesa complex sequenceof moves andcountermoves,adjustmentsandreadjustments,actionsandnonactions. Some moves are more consequentialthan others.
Some moves merelyestablishthe contextfor makingother
moves. Othersserve as a breachbetween failed andinconsequentialmoves and the promise of eventualsuccess.
Ourunderstandingof networkmanagementis derived
mainly from theoreticallyexamining, ratherthanempirically cataloging, behavior. During a time when observers first became aware of the emerging intergovernmental and interorganizational forms of governing, Hanf
identified how managers intervene in existing interrelationships, promote interactions, and mobilize coordination (Hanf, Hjer, and Porter 1978). Since that time, researchershave documented that officials from all levels
of government perceive managing across governments
and organizationsas involving a number of discrete but
related activities: mobilizing forces within and outside
the community to build support;acquiringthe necessary
financing, expertise, and other resources while setting a
course of action; learning about the externalgovernment
opportunitiesand constraints;reading the ever-changing
signals of program managers and funding agents; and
successfully operating and cooperating within the system (Agranoff 1986; Howitt 1984; Stone 1989).
Anotherway to move towardgreaterunderstandingof
the activities of networkmanagementis to model the impact of public managementin general on governmental
performanceandthen isolate factorsthatarenetwork-specific. O'Toole and Meier (1999) develop a parsimonious
yet robustframeworkfor modeling managementthat discerns not only the impact of these functions, but also the
managerialresources used to performthese functions in
particularstructuralcontexts. O'Toole and Meier's model
is not network-specific;rather,it capturesthe resources
thatgenerallycomprisepublicmanagement,which include
stabilizingthe internaloperationsof a system, exploiting
shocks in the environmentof the system, andbufferingthe
system to minimize the impact of environmentalshocks.
The model is groundedin structuralvariationsthat exist
withinparticularprogramcontextsandthe way suchvariations determinethe allocation of managementresources.
Thus, network managementis a particularallocation of
resourcesin which environmentalmanagement-leveraging externalopportunitiesand bufferingthe system from
unwantedshocks-supplements or opposes more hierarchical functions.
Meier and O'Toole (2001) test the formal model with
performancedata from Texas public school districts.The
variableconceptualizedas the environmentalcomponent
in the model is measuredby the level of interactionbeand
tween the primarynetworkmanagers(superintendents)
the
school districts'organizational
five sets of actorsfrom
Review* September/October
Administration
602 Public
2002,Vol.62, No.5
environment.Using both ordinaryleast squaresregression
and substantivelyweighted analytical techniques (Meier
and Gill 2000), the authorsfind the frequencyof interaction is positively related to school district performance;
the greaterthe numberof actorsandinteractionwith whom
the superintendents"networked,"the higher the performance.In a field withempiricalresearchthathas proceeded
at a snail's pace, the evidence demonstratingthe impactof
network managementon programperformanceprovides
an enormouscontributionto the field. Meier and O'Toole
conclude, "the results of this study offer encouragement
for those who are convincedthatpublic managementmatters, thatnetworkmanagementitself can be importantfor
performance,andthatcomplex models of public management are worthyof serious attention"(291).
Even so, the authorsacknowledgethattheirmeasureof
managementin networked settings is simplified and ignoresthe realessence of management.Indeed,the strength
of the formal model-conceptual sophistication, parsimony, and generalizability-is also its main weakness
when applied to managing in networks. Like other network management studies (Provan and Milward 1995;
Agranoffand McGuire 1998), the frequencyand regularity of network managementare accounted for and measured, but the multiple operationalbehaviors undertaken
by the managerare not.
In orderto look at how managersmanage a networkas
a network,it is necessaryto distinguishbehaviorsin terms
of their operationaldifferences. I examine four different,
thoughcertainlynot all-inclusive,categoriesdiscussedelsewherein the literature(AgranoffandMcGuire2001). These
behaviors have parallels in single-organizationmanagement, but, as arguedthroughoutand as managers in the
field are quick to suggest, they are quite different from
theirorganizational-behavior
counterparts.The operational
is
similar
to
work of Kickert,Klijn, and
the
categorization
who
distinguishthe multitudeof manaKoppenjan(1997a),
gerial tasks in terms of purposes: network management
activities aimed at the ideas and perceptions of network
members,and those aimed at the interactionof members.
The formerinclude preventingor introducingnew ideas,
bargaining,and inducing reflection within the network,
whereasthe latterinvolve arranging,structuring,and mediating interactionamong networkparticipants.
One class of behaviorsundertakenby networkmanagers is referredto here as activation,which managersin the
field suggest may be the most importantactivity of managing networks.I use the term"activation"to referto a set
of behaviors employed for identifying and incorporating
the personsandresources(such as funding,expertise,and
legal authority) needed to achieve program goals. The
single-organizationparallelto activationwould be personnel issues of staffing.Activatinginvolves identifying par-
ticipants for the network and including key stakeholders
in the process (Gray 1989; Lipnack and Stamps 1994;
TermeerandKoppenjan1997). The skills, knowledge,and
resourcesof these potentialparticipantsmust be assessed
andtappedinto (AgranoffandMcGuire1999;Klijn 1996).
Accordingto Scharpf(1978), selective activationis based
on correctlyidentifyingthe participantsandotherresources
needed for the network.Activationis a criticalcomponent
of networkmanagementbecauseresourcessuch as money,
information,andexpertisecan be integratingmechanisms
of networks. Even after a network is operationalbut not
performingas desired, "deactivation"may be needed to
remove some networkparticipants.An example of activation is a directorof economic developmentwho is charged
with developing an incentive plan for recruitingand retaining businesses and submittingthe proposalto the city
council. Applying her knowledge of the community and
her experience in the field, the directorassembles a network that includes key governmentaland nongovernmental representativessuch as the county economic development director,chamberof commerce leadership,director
of local developmentcorporation,and business owners.
Othernetworkmanagementbehaviorsare employed to
help frame the structureand the norms and values of the
network as a whole. Framing is defined as the behaviors
used to arrangeandintegratea networkstructureby facilitatingagreementon participants'roles, operatingrules,and
networkvalues. Like activation,framingis used both during the formationof the networkand when networkeffectiveness diminishes or is suboptimal.Network managers
must arrange,stabilize,nurture,and integratethe network
structure(O'Toole 1997). Framinginvolvesfacilitatingthe
internalstructureandposition of the participants(Kickert,
Klijn, and Koppenjan1997b), as well as influencing the
operating rules and norms of the network (Klijn 1996;
Mandell 1990; Termeerand Koppenjan1997). Managers
cannotdrawup an organizationalchartin a network,as is
done in single-organizationstructures,but they do try to
influence the roles that each participantmay play at any
given time and the perceptionsone has aboutthe common
purposeof the network(Benson 1975;Gray 1989;Lipnack
and Stamps 1994). Managersdo this by facilitatingagreement on leadershiproles; helping to establish an identity
and culturefor the network,even if it is temporary;assisting in developinga workingstructurefor the network(that
is, committeesor network"assignments");andalteringthe
perceptionsof participantsto understandthe uniquecharacteristics of working with persons in contexts without
organizationalmechanismsbased in authorityrelations.
As organizationalbehaviortheoryinstructsus, motivated
peoplewho "consistently,creatively,andenergeticallywork
towardthe attainmentof organizationalgoals are the key
to organizational success" (Denhardt, Denhardt, and
Aristigueta 2002). Network managers also must induce
individuals to make and keep a commitment to the network. Mobilizing behaviorsare used to develop commitmentandsupportfor networkprocessesfromnetworkparticipants and external stakeholders.Mobilization in this
regardis a commonandsometimesongoingtaskfor achieving network effectiveness. Managers build support by
mobilizing organizationsand coalitions and by forging an
agreementon the role and scope of network operations
(Kickert and Koppenjan 1997; Mandell 1990). An economic development directorin city government,for example, shouldhave the cooperationof all networkparticipants as well as the city council and the chief executive as
a means to more confidently engage in networking and
achieve the strategic purpose at hand (Agranoff and
McGuire 1999).
Managersalso must employ synthesizingbehaviorsto
create an environmentand enhance the conditions for favorable,productiveinteractionamongnetworkparticipants.
One critical behaviorof the networkmanageris to build
relationshipsand interactionsthat result in achieving the
networkpurpose.The strategiesof each networkparticipant and the outcomes of those strategiesare influenced
by the patternsof relations and interactionsthat have developed in the network.Network managementbehaviors
include facilitatingand furtheringinteractionamong participants (Agranoff and McGuire 1999; Mossbergerand
Hale 1999; O'Toole 1988), reducing complexity and uncertaintyby promotinginformationexchange (Gray 1989;
LipnackandStamps 1994;TermeerandKoppenjan1997),
and facilitating linkages among participants (O'Toole
1997). Weiner (1990) suggests that organizationalmanagementtechniquessuch as teambuildingandgroupproblem solving are applicableto networks,but networkmanagementis based on informationratherthanauthority.The
networkmanagerseeks to achieve cooperationamong actors while minimizing and removing informationaland
interactionalblockages to the cooperation.This steering
of networkprocessesis like gamemanagement,in the sense
that the result of the network process "derives from the
interactionbetween the strategiesof all actors involved"
(Klijn and Teisman 1997, 99).
The categorizationof networkmanagementbehaviors
offered here is less importantthan the point that it is necessaryto catalog suchbehaviorsandto determinehow and
why to use such behaviors.However one categorizes behaviors, the differentmanagementresourcesallocated by
managersare nearly seamless in their applicability.Multiple behaviors are utilized in network settings. For example, the addition or removal of a critical network resource-human or otherwise-can have myriadeffects on
the network.While roles are rearranged,the managerwill
try to facilitate an environmentin which greaterinterac-
DoandWhyTheyDoIt 603
onWhatManagers
Networks:
Propositions
Managing
tion can takeplace. This mightbe done by improvingcommunicationor by changingthe incentivesof participation.
Similarly,synthesizingis often accompaniedby reframing
or by a conscious attemptto change the networkparticipants' perceptionof the interactionchanges. Duringor after deactivationor reframing,managersoften must mobilize supportfor the changes,reestablishthe purposeof the
network,and make sure all participantsare "on board."
Network Environment
Thereis a fundamentalneed to discover some explanation or rationale-the set of managerialdecision rulesfor why managerschoose one activity over anotherand
how they allocate their time and effort on these activities.
What does a networkmanagerdo when the networkmust
undergoa transformationof vision andphilosophy?What
should she do? Whatdoes a networkmanagerdo when the
goals of a programor project are elusive or conflicting?
Whatshouldhe do? Broadly,what arethe criticalenvironmental variables affecting network operations,and what
are the preferredresponses to these variables?
Contingencytheorysuggeststhereis an identifiableand
predictablelogic to network managementthat can assist
researchersin explaininghow and why managersallocate
resourcesin a given context.Applying this logic can lay a
foundationfor determiningwhetheractivities and behaviors chosen by managersalways, sometimes, or never operate systematically.Even as researchreveals the vast inventory of network management behaviors, it is also
importantto understandhow the managermatches these
behaviorswith the policy context. One assumptionof this
logic is thatthe network'sfulfillmentof purposesandgoals
is contingenton its ability to function in, and its capacity
to adaptto, the environmentin which it operates.According to Simon (1981), goal attainmentis a functionof adapting the many and varied inner environmentsin a system
(thatis, the network)to the outerenvironment.He argues
that "in very many cases whethera particularsystem will
achieve a particulargoal or adaptationdepends on only a
few characteristicsof the outerenvironmentand not at all
on the detail of that environment"(11). If networkmanagementis indeedcontingent,it is necessaryto isolatethose
"few characteristics"of the outer environmentthat affect
the allocationof managerialactivities.
Mandell (1990) was one of the first researchersto suggest how managerialconstraintsvarysystematicallywithin
and across networksettings, and thus can affect what she
refersto as "strategicnetworkmanagementstyle."Thebasis
of her strategic-contingencylogic is similar to Simon's
framework:Programperformanceis contingenton the relationship of managementand its problem context. Network settings offer a unique problemcontext with sets of
constraintsthataffect, in some way, the allocationof manAdministration
Review* September/October
604 Public
2002,Vol.62, No.5
agement resources, and thus the choice of particularbehaviors. Some of the relevant environmentalcontingencies are not exclusive to networks,but others are, and all
are importantchallenges for managers.
The hypothesesofferedhere extend her analysis by examining not just the roles played by network managers,
butthe activitiesof the manager.Before we can adequately
describeand informthe practiceof networkmanagement,
it is necessary to examine actual behaviors performedin
networksandhow managementresourcesmay be allocated
differentlywithin networksat differentpoints in time and
across networks. I discuss a necessarily brief list of key
environmentalvariablesas a means of illustratingthe fluid
natureof networkmanagement;a more exhaustivediscussion of such variablesultimatelyis needed to properlydevelop theories of networkmanagement.
Matching Environment and Behavior
Goal Consensus. The degree to which program or
projectgoals are unambiguousis a critical environmental
factor affecting the allocation of managementresources
by the networkmanager(Thompson1967). Local government executives involved in network settings suggest the
final productis by far the most importantcriterionof effectiveness.This instrumentalperspectiveof networkmanagementis commonamongadministrators(Kickert,Klijn,
Koppenjan1997b), who lear early on thatone's ability to
steer government processes is closely related to goal
achievement.Some programshave very clear objectives
and readily determinedperformanceindicators,which allow managers to expend more energy on collecting the
properresourcesandcoordinatingactivities,just as a managermay do in a bureaucraticorganization.Otherprograms
areconceived in vague terms,andthe networkmust establish its objectivesjointly. Consequently,it is reasonableto
predictthe following hypothesis:
Hypothesis 1: Effectivemanagersin networkswith
relativelyclear programobjectiveswill allocate the
greatestshareof managerialresourcesto activating
(identifying participants,tapping the skills and resources of these persons, ensuring all interests are
included)and synthesizing(promotinginformation
exchange, developingproceduresof interaction).
A program infused with many conflicting goals requires a different type and level of managerial energy.
Mandell (1990) argues that "the problem is how to meet
the overriding interorganizationalgoal (or set of goals)
while at the same time allowing each organizationin the
networkto also meet its own...." The presence of unclear
programgoals means that one of the first tasks the manager must attend to is facilitating goal consensus-indeed, some networks cannot (should not) proceed without some level of agreement on the ultimate objective.
Faced with this constraint,the goals and actions of network memberswill be mutuallycreatedand adjustedover
time (Agranoff 1986). Because goals often are embedded in other goals, disentanglingthe primaryor immediate programobjectiveis a criticalmanagementtask.Thus,
one can posit the following:
Hypothesis 2: Effectivemanagersin networkswith
relativelyunclearor multipleprogramgoals will allocate the greatestshareof managerialresourcesto
mobilizing (motivatingnetworkparticipants,gaining supportfrom stakeholders,fosteringtheir commitment) and framing (influencing the operating
structureof the network,influencing its prevailing
values and norms, creatinga sharedpurpose or vision).
Resource Distribution.The structureof organizational
interdependenciesin a networkis defined by the distribution of resourcesacross networkmembers.The positional
resourcesof policy making-legal authority,funding, organization,materials,needed supportservices, expertise,
information,and experience-are at the disposal of many
differentorganizations(Franz 1991). Each party literally
possesses some level of resourcesthat will make a policy
work. Just as organizationscan adapt their structuresto
the environment(Pfeffer and Salancik 1978), so, too, will
network managers shape the structureand membership
based on resourcedependencies.
Resourcesmay be distributedwidely acrossmanyplayers, resultingin a necessarily large network,or they may
be relatively concentratedin a few key players. Persons
who contributemay come from many different sources,
and their hierarchicalposition in their "home"organization is less importantthan the resourcesthey bring to the
network.The types andquantityof resourcesheld by each
playermay varyas well. One networkmight includemany
members with a relatively equal distribution(control) of
policy-makingresources,andanothernetworkof the same
size might includeone playerwho possesses an influential
amountof a criticalresource,such as fundingor legal authority (Mandell 1990). The network manager needs to
identify and include in the network the needed expertise
and other resourcesto move a project forward.Effective
managersknow who has such resources and will be successful in bringingthem into the networkstructure.Thus,
we can predictthe following:
to manageexternallyis relatedto the internalconditionof
the manager'sprimaryorganization.City administrators
reportthattheircapacityto functioneffectivelyin networks
is directlyrelatedto the scope of cooperationwith the city
council throughparticipatingin strategysessions and representingthe city in public forums. Supportalso involves
a willingness on the part of governmentofficials to conduct an open and continuous dialogue with existing and
potentialpartners,thus sharinginformationfrom government to network,and vice versa. The repertoireof a networkmanagercertainlymust include these support-building activities.This suggests the following:
Hypothesis 4: Effectivemanagersin programswith
a relativelylow level of supportfromkey stakeholders will allocate the greatestshareof managerialresources to mobilizing.
Relationships.Networkscomprisingmemberswho have
worked togetherpreviously may requireless attentionto
developing a common purpose or establishing trust than
will networkswith relativenewcomers. Trustis a critical
resourceas public and nonpublicorganizationsattemptto
redefinetheirusual legal-based(hierarchical,contractual)
relationships(Nohria 1992). It is commonly acceptedthat
people join, remain,and work togetherbecause some element of trust exists, and trust is developed most easily
throughexperienceand familiarity.Managerialbehaviors
such as sharingand discussing information,finding similarityin processingandtechniques,andjustifying transaction costs may be less criticalwhen trustis presentin networks.Similarly,managerscan allocatefewermanagement
resourcesfor formulatingproductiverelationshipsamong
the networksmembers when a sharedbelief or common
purpose-a programrationale(Mandell1988)-is already
present.Thus, one could predictthe following:
Hypothesis 5: Effectivemanagersin networkscomposed of some memberswho have workedtogether
previouslywill allocatethe lowest shareof managerial resourcesto synthesizing.
Policy Orientation.The content of the policy practices
endemic to a particularnetworkis anotherenvironmental
contingency on which the fluid nature of network management may rest. What Benson (1982) calls a "policy
paradigm,"which refers to a commitmentwithin the sector to a particularset of policy options, consists of the
network's choice of policy instrumentsor tools (Elmore
Hypothesis 3: Effective managersin programenvi1987; Salamon 1981). Some programareasmightrely prironments with a wide distribution of needed remarily on grants-in-aidfor resolving problems, whereas
sources will allocate the greatest share of manageothersdependon informationtransferand technicalassisrial resourcesto activationand mobilizing.
tance. The dominantpolicy focus constrains,to some deSupport.As in any public program,political and social gree, the extent to which a network managercan exploit
supportis an importantenvironmentalcontingencyor con- the environmentfor needed resources;when grant fundstraintfor single organizationsand networks.The ability ing is needed,fundingmustbe sought,andactivatingmemonWhatManagers
DoandWhyTheyDoIt 605
Networks:
Propositions
Managing
bers with resourcesotherthanfundingmay have to be deferred. Some programareas depend on directly provided
instrumentsor regulationandauthorityinstruments,resulting in mandatedaction. Otherpolicy areas, such as economic development,rely increasinglyon instrumentsdesigned to tap into the special qualities of a community,
thus requiringmore diversity in policy formulationand
implementation.
McGuire (2000) demonstratesthat the process of designing and administeringlocal economic development
policy can vary accordingto the types of policies and institutionsadoptedin the jurisdiction.Similarly,O'Toole's
examinationof replacingfederalgrantsupportfor municipal wastewatertreatmentwith state revolving loan funds
concludes that "decisions to shift programsto the states,
deregulate, privatize, and employ market-basedmechanisms have consequencesfor interorganizationalarrangements andprogramsin practice"(1996, 239). This discussion suggests the following:
Hypothesis 6: Effectivemanagersin programareas
with a reliance on policy instrumentsusing local
resourceswill allocate a greatershareof managerial
resourcesto activatingand mobilizing.
Because the implementationof programsusing intergovernmentalinstrumentssuch as grants-in-aidand mandated regulationsrequiresa great numberof actors with
disparategoals, sometimesconflictingobjectives,andlittle
incentive to cooperatein operationalactivities, one might
predictthat:
Hypothesis 7: Effectivemanagersin programareas
with a reliance on intergovernmentalsubsidies and
regulationswill allocate a greatershareof managerial resourcesto framingand synthesizing.
Strategic Orientation. A complicated environmental
factor concerns the multiple and sometimes conflicting
purposesfor which networksexist. In its most fundamental form, managementis the mechanism for achieving a
programobjective, such as educatingchildren,improving
the local economy, or providing critical health care services to those in need. Even if the objective is unclearin
the relevantlegislation or becomes compromisedduring
the process, managers seek to deliver that which policy
makersintend and citizens demand.All tasks undertaken
by the managerderivefromachievingthatproject.In many
fields, neither the principals (lawmakers)nor the clients
(servicerecipients)areas concernedaboutprocessas much
as they are goal achievement.
Because managerial resources are finite, managers
sometimes must choose between facilitatingnetwork action to achieveprogramobjectivesandcateringto key networkplayers.Programobjectivesmay be clear-thus proReview* September/October
606 Public
Administration
2002,Vol.62, No.5
viding a clear measureof effectiveness for networkmembers, in theory-but the stability of the network may be
fleeting if key stakeholdersandnetworkmembersperceive
little benefit from membershipand threatento bail from
the network by demandingmore agency-level resources.
In this scenario, the managermust brokera solution that
balances network effectiveness with the efficient allocation of resourceswithin the network.If the networkmember withdrawsfrom participation,effectiveness is jeopardized; if resources are reallocated, efficiency may be
sacrificed.
Provan and Milward (2001) argue that networks may
be expected to fulfill criteriafor effectiveness that are different from simple goal achievement. Satisfying groups
thatrepresentdiverse communityinterests, such as advocacy organizations,is a crucial criterionof effectiveness.
As the authorssuggest,in termsof stakeholders,"networks
must satisfy the needs and expectation of those groups
within a community that have both a direct and indirect
interestin seeing thatclient needs are met"(417). The primaryprogramobjectivemay be to deliver and/orenhance
client services, but the community at large may demand
success in achievingvisible aggregateoutcomes.Although
a social services networkmay seek to enhanceits arrayof
services to reach needy populations,the population as a
whole may determinethe community-levelcosts of such a
programare too high, either because of competition for
limited resourcesor for philosophicalreasons. Effectiveness may be viewed by externalgroups as "dependingon
whatspecific serviceproviderseitherdo or do not do, rather
thanhow well services areprovidedas a resultof network
activities,"thusresultingin community-andnetwork-level
decisions made "at the expense of networkparticipants"
(422). Thus, one could posit that:
Hypothesis 8: The allocation of managerial resourcesto be expendedin a networkmay varyto the
extent thatclient needs may not be compatiblewith
community-or organizational-levelneeds.
It follows then that:
Hypothesis 9: Managersin networks where competing assessments of effectiveness exist will allocate the greatest share of managerialresources to
framingand mobilizing.
The complete range of environmentalcontingencies is
not specifiedhere.The foregoinghypothesesillustratehow
to proceedwithnetworkmanagementresearch,butthe level
of generalityis variable.Forexample,one can reconfigure
thehypothesesto be moregeneral,perhapsphrasedin terms
of stability,exploiting the environment,and bufferingthe
environment(O'Toole andMeier 1999). In this regard,the
first hypothesiscould be restated:
with
innetworks
Hypothesisla: Effectivemanagers
relativelyclearprogramobjectiveswill allocatea
resourcesto exploiting
greatershareof managerial
thanto stabilizingthesystheexternalenvironment
tem.
Alternatively,one could test a hypothesisby examining
a particularbehaviorratherthan a broad category of behaviors;therefore,the first hypothesis could be restated:
innetworks
with
Hypothesislb: Effectivemanagers
relativelyclearprogramobjectiveswill relyprimarily on identifyingand securingneededresources
neededpolicy resources,andpromotingeffective
communication
amongparticipants.
The hypotheses offered here suggest the need for simplified,even elementary,researchon networkmanagement.
Regardlessof the level of generality,researchis needed to
test these and otherhypotheses.
TheFluidityof the Match
servationsof managersthan on the networkmanagement
literature,it is reasonableto inferfromthe manager'slogic
in use thata single packagerepresentingthe allocationand
reallocationof tasks exists and can be observed.There is
an implicit understandingthat certain behaviors work in
certainsituationswithin certaincontexts, so a playbookof
sorts can be constructedfrom the manager'sexperience.
When asked to describe the generalprocess of managing a network, network managers operating in a largely
technicalenvironment-involving clearobjectives,broadbased support,and ampleresources-assert the use of ostensibly linear strategies. An example of a network in
which linearstrategiesmay prevailis a temporary,projectbased networkor a networkformed as a result of a catastrophicevent. The quality (measuredby the high level of
goal consensus, lack of contentiousness among network
members,supportof key stakeholders,presence of a program rationale,etc.) of coordinatingrelief efforts across
governments and organizations after the attacks on the
WorldTrade Center and the Pentagon illustrates the essence of a technical environment.Such an environment
addressesproblems that have been referredto as "tame"
(Rittel and Webber 1973; O'Toole 1997). In these contexts, managersactivatea network,mobilize commitment,
frame the operating structuresand rules of the network,
and then synthesize interactioninto a productivewhole.
Although certainlyoversimplified,networkmanagement
may operate like this in specific contexts where the vast
majority of resources are dedicated to one set of tasks,
then the next, then the next, and then the next, with only
occasional changesmade. Sharedprogramgoals andrelatively clear operatingobjectivesmake for relativelystable
systems, so one would expect linearnetworkmanagement
strategiesto result in stability even while leveraging externalopportunities(table 1, la).
Othernetworkmanagementstrategiesmight be identifiable in different contexts. Network managers lacking
broad-basedsupportfor theiractivitiesdescribehow often
the process of mobilizing must be revisited over time. In
these contexts, the allocation of resources alternatesbetween activationand mobilization:Persons become part
The first step in properlymodeling network management is to identify specific behaviorsundertakenby managers in various situations.The hypotheses offered above
are meant as guides for such research.However, observing the way networkmanagementresourcesare allocated
within a networkis necessary,but not sufficient;because
managementin networksis fluid, we also mustunderstand
how and why managerialresources are reallocated over
time and space. Even in networksin which a certaintype
of resourceis most salient,many activities are performed,
andwe mustidentifywhy andhow managersrotateamong
and reallocateresources"on the fly."
The concept of strategyoften is used ambiguously to
referto these many activities,butthe presentanalysisdoes
not view strategyin this manner.The term "strategy"is
best used to describewhatO'Toole andMeier (1999) refer
to as the "whole set of behaviors"relatedto management.
In keeping with contingency logic, the totality of behaviors and the fluid allocationof such behaviorsis itself the
strategy.Choosing to expandthe size of the network-an
activationbehavior-is not the same as formulatinga management strategy.Expandingthe network is a single be- Table1 SpeculativeNetworkManagementStrategies
havioramongmanybehaviorsthattypically areemployed
la. LinearStrategy(technicalenvironment,clear programobjectives,
to effect a single purpose. Thus, a networkmanagement
project-based)
- Framing- Synthesizing
strategyis defined not only by the numberand types of
Activation- Mobilization
instrumentsthat are used to solve a policy problem, but
(Withminimalreallocation)
also by the ways in whichtaskdeploymentvariesovertime.
1b. RecursiveStrategy(lackof support)
- Activation
- Activation- Mobilization
Activation- Mobilization
(Withminimalreallocationand less need forframingand synthesizing)
Audacious Speculation
Interviews with local government executives suggest
some common,genericstrategiesinvolvingmultiplemanagerial behaviors.Althoughthey are based more on the ob-
1c. RecursiveStrategy(institutional
environment)
- Framing
- Framing- Synthesizing
Activation- Framing- Motivation
(Withemphasison identifyingsharednormsand valuesof the network)
on WhatManagersDoandWhyTheyDo It 607
ManagingNetworks:Propositions
of the network,supportis solicited, more personsare activated and some are deactivated,commitmentis solicited,
and so on. As hypothesis4 suggests, effective managersin
these contexts will allocate the greatestshareof managerial resourcesto mobilizing, but managersin the field describe the process as recurring.The other management
behaviorsof framingand synthesizingareundertaken,but
the main approachresultsin whatcan be called a recursive
strategy(table 1, lb).
Similarly, it is plausible to hypothesize that network
managersemploy recursivestrategiesin institutionalenvironments(Scott and Meyer 1991), where immediategoal
achievementis impossibledue to a lack of goal consensus,
and perhapseven undesirableif the network must build
long-termlinkageswithin a communityandnetwork-level
"soulsearching"definesmuchof the network'soperations.
In suchenvironments,a greatdeal of attentionwill be given
to framingandreframingthe network.As rules andnorms
are established,some networkmembersleave, others are
added;supportis mobilized internallyand externally;and
interactionis enhanced, but framing tasks dominate the
activities of the networkmanager(table 1, lc).
Conclusion
The focus of this article has been the actions of the
many governmentaland nongovernmentalofficials operating in network settings on a daily basis. To them, management may not seem predictable,and it certainly may
be presumptuousto call it systematic. The contingency
logic suggested here does not imply rationality.It passes
no judgment on the long-term public value of network
management,nor does it advocate specific management
strategies. Particularmanagementstrategies are not preferredover others. Instead, a simple contingency logic is
used as a tool for magnifying a process that seems too
complicated to comprehend.
Why does managerB allocate managerialresourcesin
one way, and managerC allocates resources differently?
Why are some networkedprogramssuccessful, but others
arenot?The answersto these questionscan be foundsomewhere throughthe kind of observationthatprovidesa rich
description of network managementand the type of inquiry that assumes public managersdo and must manage
networkas networks.Observationof managerialbehavior
is one mechanismfor more accuratelydocumentinghow
managersmatchbehaviorwith environmentalconstraints,
as is interviewingmanagers,eitherthroughin-depthinterviewing or with a survey.The preliminaryclassificationof
behaviorsofferedhere shouldbe tested andrefined-even
rejected-if sound data are collected from managersfor
this purpose.Bottom line, it is necessary to hear from the
managersthemselves.Researchersmay assistpractitioners
with a vocabularyandpreliminaryframeworkwithinwhich
their actions can be described,but the managersmust be
the ones who reveal their actions.
Both good andbadmanagementchoices exist; anypublic manager would assert that. The proposed research
agenda is offered as a guide to help us determinewhich
choices aremost likely to be effective. If nothingelse, such
researchmay providea vivid story and one or two snappy
aphorisms.
References
Agranoff, Robert. 1986. IntergovernmentalManagement:Human Services Problem-Solving in Six MetropolitanAreas.
Albany,NY: State Universityof New YorkPress.
Agranoff, Robert, and Michael McGuire. 1998. Multinetwork
Management:Collaborationand the Hollow State in Local
EconomicPolicy.Journalof Public AdministrationResearch
and Theory8(1): 67-91.
. 1999. Managing in Network Settings. Policy Studies
Review 16(1): 18-41.
. 2001. Big Questions in Public Network Management
Research. Journal of Public AdministrationResearch and
Theory 11(3): 295-326.
Benson, J. Kenneth. 1975. The InterorganizationalNetwork as
Political Economy.AdministrativeScience Quarterly20(2):
229-49.
. 1982. A Framework for Policy Analysis. In
Interorganizational Coordination: Theory, Research, and
Implementation,edited by David L. Rogers and David A.
Whetten, 137-76. Ames, IA: Iowa State UniversityPress.
Review* September/October
Administration
608 Public
2002,Vol.62, No.5
Denhardt Robert B., Janet Vinzant Denhardt, and Maria P.
Aristigueta.2002. ManagingHumanBehavior in Public and
NonprofitOrganizations.ThousandOaks, CA: Sage Publications.
Elmore, Richard F. 1987. Instrumentsand Strategy in Public
Policy. Policy StudiesReview 7(1): 174-86.
Franz,Hans-Jurgen.1991. Interorganizational
Policy Coordination: Arrangementsof Shared Government.In The Public
Sector-Challenges for Coordinationand Learning, edited
by Franz-XaverKaufmann,469-99. Berlin:Walterde Gruyter.
Gray,Barbara.1989. Collaborating:Finding CommonGround
for MultipartyProblems.San Francisco,CA: Jossey-Bass.
Hanf, Kenneth,Benny Hjern,and David D. Porter.1978. Local
Networks of ManpowerTrainingin the FederalRepublic of
GermanyandSweden. In InterorganizationalPolicy Making,
edited by KennethHanf and FritzW. Scharpf,303-41. London: Sage Publications.
Howitt, Arold. 1984. Managing Federalism:Studies in Intergovernmental Relations. Washington, DC: Congressional
QuarterlyPress.
Kickert,WalterJ.M., and Joop EM. Koppenjan.1997. Public
Managementand Network Management:An Overview. In
ManagingComplexNetworks,editedby WalterJ.M.Kickert,
Erik-HansKlijn, and Joop F.M. Koppenjan,35-61. London:
Sage Publications.
Kickert,WalterJ.M.,Erik-HansKlijn,andJoopF.M.Koppenjan.
1997a. Introduction:A ManagementPerspective on Policy
Networks.InManagingComplexNetworks,editedby Walter
J.M. Kickert,Erik-HansKlijn, and Joop EM. Koppenjan,113. London:Sage Publications.
. 1997b. Managing Networks in the Public Sector. In
ManagingComplexNetworks,editedby WalterJ.M.Kickert,
Erik-HansKlijn, and Joop EM. Koppenjan, 166-91. London: Sage Publications.
Klijn, Erik-Hans. 1996. Analyzing and Managing Policy Processes in Complex Networks. Administrationand Society
28(1): 90-119.
Klijn,Erik-Hans,andG.R.Teisman.1997. StrategiesandGames
in Networks. In Managing Complex Networks, edited by
Walter J.M. Kickert, Erik-Hans Klijn, and Joop F.M.
Koppenjan,98-118. London:Sage Publications.
Lipnack,Jessica, and JeffreyStamps. 1994. TheAge of the Network. New York:JohnWiley and Sons.
Mandell, Myrna P. 1988. IntergovernmentalManagement in
InterorganizationalNetworks:A Revised Perspective.InternationalJournal of Public Administration11(4): 393-416.
.1990. NetworkManagement:StrategicBehaviorin the
Public Sector.In Strategiesfor ManagingIntergovernmental
Policies and Networks,editedby RobertW. Gage andMyrna
P. Mandell, 29-53. New York:Praeger.
McGuire,Michael. 2000. CollaborativePolicy MakingandAdministration:The OperationalDemands of Local Economic
Development.EconomicDevelopmentQuarterly14(3): 27691.
Meier, KennethJ., and Jeff Gill. 2000. WhatWorks:A New Approach to Program and Policy Analysis. Boulder, CO:
Westview.
Meier, KennethJ., and LaurenceJ. O'Toole. 2001. Managerial
StrategiesandBehaviorin Networks:A Model with Evidence
from U.S. Public Education.Journal of Public Administration Researchand Theory 11(3): 271-93.
Mossberger,Karen,and KathleenHale. 1999. InformationDiffusion in an IntergovernmentalNetwork:The Implementation of School-to-Work Programs. Paper presented at the
Annual Meeting of the AmericanPolitical Science Association, September2-5, Atlanta,Georgia.
Nohria, Nitin. 1992. Is a Network Perspectivea Useful Way of
Studying Organizations?In Networks and Organizations:
Structure,FormandAction, editedby Nitin NohriaandRobert G. Eccles, 1-22. Boston: HarvardBusiness School Press.
O'Toole, Laurence J. 1988. Strategies for Intergovernmental
Management:ImplementingProgramsin Intergovernmental
Management.InternationalJournalof Public Administration
11(4): 417-41.
. 1996. Hollowing the Infrastructure:Revolving Loan
Programsand Network Dynamics in the American States.
Journal of Public AdministrationResearchand Theory6(2):
225-42.
. 1997. TreatingNetworks Seriously: Practicaland Research-BasedAgendas in Public Administration.Public AdministrationReview 57(1): 45-52.
O'Toole, LaurenceJ., and Kenneth J. Meier. 1999. Modeling
the Impactof Public Management:Implicationsof Structural
Context. Journal of Public Administration Research and
Theory9(4): 505-26.
Pfeffer, Jeffrey, and Gerald R. Salancik. 1978. The External
Controlof Organizations.New York:Harperand Row.
Provan,Keith G., and H. BrintonMilward. 1991. InstitutionalLevel Norms and OrganizationalInvolvementin a ServiceImplementationNetwork.Journal of Public Administration
Researchand Theory 1(4): 391-417.
. 1995. A PreliminaryTheory of InterorganizationalEffectiveness:A ComparativeStudy of FourCommunityMental Health Systems. AdministrativeScience Quarterly40(1):
1-33.
.2001. Do Networks Really Work?A Frameworkfor
Evaluating Public-Sector OrganizationalNetworks. Public
AdministrationReview 61(4): 414-23.
Rhodes, R.A.W. 1997. Foreword.In Managing ComplexNetworks, edited by WalterJ.M. Kickert,Erik-HansKlijn, and
Joop EM. Koppenjan,xi-xv. London:Sage Publications.
Rittel, Horst,and Melvin Webber.1973. Dilemmas in a General
Theory of Planning.Policy Sciences 4(1): 155-69.
Salamon, Lester M. 1981. Rethinking Public Management:
Third-PartyGovernmentand the Changing Forms of GovernmentAction. Public Policy 29(3): 255-75.
Scharpf,Fritz W. 1978. InterorganizationalPolicy Studies: Issues, Concepts,andPerspectives.InInterorganizational
Policy
Making,edited by KennethHanf and FritzW. Scharpf,34570. London:Sage Publications.
Scott, W. Richard,and JohnW. Meyer. 1991. The Organization
of Societal Sectors:Propositionsand EarlyEvidence. In The
New Institutionalismin OrganizationalAnalysis, edited by
WalterW. Powell and Paul J. DiMaggio, 108-40. Chicago:
Universityof Chicago Press.
Simon, HerbertA. 1981. The Sciences of the Artificial. 2nd ed.
Cambridge,MA: MIT Press.
Stone, ClarenceN. 1989. RegimePolitics. Lawrence,KS: University Press of Kansas.
Termeer,C.J.A.M.,and Joop EM. Koppenjan.1997. Managing
Perceptionsin Networks. In Managing ComplexNetworks,
editedby WalterJ.M.Kickert,Erik-HansKlijn,andJoopEM.
Koppenjan,79-97. London:Sage Publications.
Thompson,JamesD. 1967. Organizationsin Action. New York:
McGraw-Hill.
Weiner,MyronE. 1990. HumanServices Management.2nd ed.
Belmont, CA: Wadsworth.
DoandWhyTheyDoIt 609
onWhatManagers
Networks:
Propositions
Managing