Robinson Evolution of EM Networks

8/6/2019 Robinson Evolution of EM Networks

1/38

The Core and Periphery of EmergencyManagement Networks: A Multi-ModalAssessment of Two Evacuation Hosting

Networks from 2000-2009

Scott E. Robinson - Texas A&M University

Warren S. Eller - University of North Carolina - PembrokeMelanie Gall - Louisiana State University

Brian J. Gerber - University of Colorado - Denver

April 4, 2012

Please direct communication to [email protected]

1


2/38

Abstract

Emergency management is a eld in which collaborative activitiesare inescapable. Emergency planning and response increasingly in-volve close interactions between a diverse array of actors across elds(emergency management,public health, law enforcement, etc.), sector(government, nonprot,and for-prot), and level of government (local,state, and federal). The necessity of collaboration is built into the logicof escalation in the Stafford Act and the nature of emergency eventsas boundary spanning threats. While the necessity of collaboration isclear, the dynamics of this collaboration are less well understood.

This article assesses the temporal dynamics of emergency manage-ment networks in two moderately sized communities that have served

as large-scale disaster evacuation hosting sites in the past decade. Thepaper uses two strategies for tracking the evolution of these networksacross time. First, we develop an network roster using newspaper andnewswire data sources across a decade. Second, we develop a viewof the evolution of the networks by analyzing emergency operationsplans for each community.

Analysis of data from these two strategies reveal a contrast be-tween a core set of consistent (mostly governmental) actors and aperipheral set of rapidly turning (mostly non-governmental) over ac-tors though the account depends on the mode of data on which onefocuses. The article concludes with a discussion of the advantage pre-

sented by having a two-tier network for evacuation hosting that mixescore and periphery across multiple sectors.

2


3/38

1 IntroductionPublic administration and political science are currently focusing quite a bitof attention on issues related to political and administrative networks. ThePublic Administration Review devoted an entire special issue to the subjectin 2005 and the American Political Science Association created a specializedorganized section on the subject of political networks.

While attention to collaborative public management and policy networksis high right now, this is by no means a new subject. The classic argument of the dominance of iron triangles or policy whirlpools is a network argument -albeit of a small network (Redford 1969). The counterargument that policytends to involve broad and uid participation in issue networks is also rather

obviously a network construct (Heclo 1978). More recent integrations of theliterature positing changing levels of participation over time and across policyareas suggest that these networks can evolve over time as characteristics of individual policy domains change (McCool 1998, Sabatier & Jenkins-Smith1993).

While attention to issues of administrative and policy networks has been acomponent of the policy literature for decades, the dynamics of the networksacross time has proven to be a difficult subject to study especially in thearea of emergency management. Given the inherently collaborative natureof emergency management networks, as we discuss below, understandingchange in this domain is especially important. Due to extraordinary demandson data and the necessity of novel inferential techniques for data involvingnetworks, very little work has engaged issues related to network change.

This article represents an initial step toward assessing the evolution of emergency management networks across time - in this case over a decadeinvolving two major events. The next section (Section 2)will discuss someof the existing literature on issues related to the incorporation of new actorsinto a policy network and the evolution of network characteristics over time.The result will be a series of propositions about the nature of emergencymanagement network change with particular attention to the differences involatility between sectors (government, nonprot, private). Section 3 will in-

troduce two approaches to measuring membership1

and relationships within1 Here membership only means co-participation in a community effort. It does not imply

formal membership of acknowledgment. Given our interest in organizations that may actin the periphery of the network, a restriction to formal or self-acknowledged membershipwould exclude potentially interesting actors.

1


4/38

policy networks. Sections 4 and 5 report on the roster of emergency man-

agement networks in two communities for each of the two measurementsstrategies discussed in 3. Finally, Sections 6 and 7 discuss the implicationsof these results for the management of evacuation networks and the admin-istrative / policy networks generally. Section 8 concludes with a summaryand discussion potential avenues for future research.

2 The Evolution of Emergency ManagementNetworks

A central question of research into policy and administrative networks is thescope and uidity of participation. The key characteristic of policy subsystemapproaches to networks was the emphasis on closed and limited participationby predictable actors (e.g. congressional subcommittees, interest groups, andagencies). The principal critique of the argument was that participation inactual policy domains tends to be much more broad and uid. It was arguedthat a large variety of actors may participate within any policy networkincluding those envisioned by subsystem theories as well as representativesof other levels of government and even public interest groups. Furthermore,the participation level of various actors is thought to change over time withsome actors dropping out of active participation while new actors emerge

at different time. An issue network represents an extreme version of thisopen and uid network (Heclo 1978). A key question, then, is the scopeand uidity of participation in actual policy networks. The next subsectionwill discuss the issue of participation in emergency management networksspecically. Following that discussion, we present specic propositions fromthe literature regarding the evolution of emergency management networks.

2.1 Networks and Emergency ManagementOver the past two decades, the importance of collaborative networks has be-come clear to scholars specializing in emergency management(Comfort 2006,

Kapucu 2009). Emergency management represents a classic wicked problem(OToole Jr 1997). Emergencies tend to cross jurisdictional boundaries due tothe geographic scope and the broad range of consequences they present. Forexample, Hurricane Katrina devastated communities across multiple statesand mobilizing reactions from a variety of government agencies (emergency

2


5/38

management, law enforcement, transportation, public health, housing and

welfare, etc.) and nongovernmental agencies (the American Red Cross, Wal-mart, local religious institutions, etc.) (Simo & Bies 2007).Among the most prominent voices for research into collaboration and

networking activities in emergency management has been Louise Comfort.Comfort has argued that emergency management networks are best under-stood as self-organizing systems (Comfort 1994). The emphasis in her ac-count is uidity of participation and the inability to predict mobilizations exante. Rather than following documented plans or stable expectations, mo-bilization tend to involve an unpredictable set of actors that vary greatly interms of prior disaster experience, organizational sector, and other charac-teristics. This can be seen as a radical extension of Heclos characterizationof policy issue networks as involving uid participation.

Concluding that mobilization is unpredictable is unsatisfying in a numberof ways. First, it suggests that efforts to plan mobilizations are doomed tofailure. If one can not predict who will be involved - at least in terms of somecore players - then one can not know whom to involve in emergency planning.Second, to the extent that exercises and other simulations are key preparatory(and possibly even evaluative) elements of emergency management, if onecan not predict who will mobilize following an emergency event then onewill not know who to include in an exercise. The limited composition of exercises preceding Hurricane Katrina has been identied as a key cause of

the eventual problems in evacuating residents of New Orleans with limitedaccess to transportation (Kiefer & Montjoy 2006). Some level of stability isessential for the inclusion of actors in pre-event exercises (Kapucu 2008).

However, the difficulty in predicting which organizations will mobilize inan emergency may have been overstated. In a study of the mobilizationof evacuation hosting activities in the Dallas/Fort Worth, TX area follow-ing Hurricane Katrina, Robinson, Berrett & Stone (2006) found that themobilization of many organizations was predictable given a series of priorrelationships. Relationships that sometimes had little to do with emergencymanagement and evacuation hosting activities served as a basis for the emer-gence of a series of response networks. While there was also evidence of

spontaneous mobilization of organization with no prior membership in emer-gency management networks, a good part of the network - particularly thenetwork leadership - involved prior relationships that could easily escape theattention of emergency management scholars. The case studies collected inthis article provided some hope that relationships can be managed and that

3


6/38

mobilizations can be predicted (within some bounds).

The complexity of the mobilization and management process of emer-gency management networks has raised important questions about the man-agement and leadership of these networks. Waugh & Streib (2006) arguedthat coordination is difficult within emergency management networks despiterecent attempts to provide structure to the networks through such devicesas the National Incident Management System (NIMS) and the IntegratedCommand System (ICS). The difficulties in leadership are accentuated giventhe diversity of these networks. Recent studies of emergency managementfrom Florida and Louisiana have served to illustrate the scope of this com-plexity including diversity across private, public, and nonprot sectors(Kapucu 2006, Kapucu 2007, Kapucu, Augustin & Garayev 2009) Actorsfrom diverse sectors and policy areas bring with them a variety of assump-tions about the nature of emergencies and appropriate forms of coordinationand communication(Comfort 2007).

2.2 Propositions for Network EvolutionGiven the importance of network collaboration to issues of emergency man-agement, research into the development and evolution of these networks isessential to the improvement of management of emergency preparedness andresponse networks. This places our research in the tradition of whole net-work analysis (Provan, Fish & Sydow 2007). The topic of interest is thecharacter of the entire network rather than the actions of specic individu-als (or pairs/dyads) within it. While there is valuable research at the dyadiclevel investigating issues like the relationship between network membershipand performance (Meier & OToole 2001) we seek to investigate questionsat the level of the network itself.

This article will focus on expectations surrounding the key characteris-tics that distinguish subsystem models from issue network models of policynetworks: volatility in network participation (Heclo 1978). Volatility cantake on a number of meanings. In its simplest form, volatility can involvethe change in the size of a network (that is, the number of members it has).

Assessing the aggregate size of networks would yield literature information,though. If one only knew network size, one could not tell whether a constantsize was the result of the lack of change or additions and subtractions of members that offset. For this reason, we are interested in volatility as mea-sured by the number actors at time ( t ) that are not present in the network

4


7/38

at time ( t + 1).

Given the strength of prior research suggesting that there is volatility inemergency management networks, we start with the proposition that:

Proposition 1 Emergency management network membership will experi-ence signicant change from t to t + 1 .

This proposition contrasts with the expectations of iron triangle theoriesthat suggest that policy systems involve only a small number of unchangingparticipants. In its strongest form, it could suggest essential unpredictabil-ity of network members along the lines of Comforts arguments regardingspontaneous or emergent networks.

However, the size of a network may be less important than the resourcesand knowledge contained within that network. As a result, involving differentorganizations may not be as important as increasing representation from di-verse types of organizations. Diversity within a network engages a key trade-off in network studies between brokerage and closure (Burt 2005). Linkagesbetween a particular clique and the rest of the network (like an emergencymanagement network embedded within its broader political community) pro-vide informational advantages to members of the network. Connections tolocal law enforcement, for example, could provide support during emergencysituations. Similarly, a connection to a local nonprot food kitchen may pro-vide an important resource during an emergency event. These are advantagesto brokerage connections across cliques or clusters of organizations.

There are advantages also in having a tightly knit group. A tightly knitgroup expands the opportunities for repeated interaction and trust building.Having the members of a network consistent participants in, say, emergencyexercises provides for community building within the emergency managementclique. In Burts terms, these are advantages to closure (Burt 2005). Theart of effective building of collaborative groups is to balance the advantagesof broad brokerage with the advantages of focused closure.

We will focus on representation of policy sectors (e.g. emergency man-agement, law enforcement, education, public health, etc.) and nonprot /

private sector organizations (Kapucu 2007). A network will experience bro-kerage to the extent it represents a broad range of policy issues sectors andorganizational sectors. A network will experience closure to the extent thatthe network experiences stability in membership.

5


8/38

We propose that these two strategies are not entirely inconsistent. Net-

works can develop a core of stable actors supplemented by a brokered ex-tended group of actors whose presence in the network is transient. In an ap-proach similar to the hierarchal networks advocated by Moynihan Moynihan(2008), emergency management networks may experience closure among itscore while taking advantage of access to resources present within a transientperiphery of organizations where members change based on the particularnature of the event.

The question remains as to why members are in the core or the periph-ery. Traditionally, emergency management research has focused almost ex-clusively on the role of government organizations. Given the central rolegiven to governmental organizations 2 , we anticipate that the stable core of the networks will be dominated by government organizations with privateand nonprot organizations most often found in the periphery. The resultare two more specic propositions.

Proposition 2 There will be an identiable core of government organiza-tions in emergency management networks.

Proposition 3 There will be an identiable periphery of non-governmental organizations in emergency management networks.

3 DataDisaster evacuations offer an opportunity to understand these networks be-cause they typically represent community-scale and often inter-jurisdictionalactivations of collaborative management effort. In each of the communitieswe studied, evacuation activities lasted past the basic planning assumption of 48-72 hours of hosting. These activities then illustrate to whom (if anyone)emergency managers turn when forced to support activities beyond routines.Evacuation hosting is also an activity that can call upon diverse supportfrom the community. The activities needed to support evacuation hostingare built into the host communities themselves ranging from nutrition andfeeding to entertainment to basic retail sales. In a test of the variabilityof government and nongovernmental actors within emergency management

2 But see Robinson (2007) and Kapucu (2007).

6


9/38

networks, this specic activity that engages a wide variety of actors is an

appropriate starting point.Disaster evacuation represent one type of disaster response activity rather than the totality of disaster response. Evacuation hosting provides auseful domain of investigation in that it has historically existed at the inter-section of nonprot and government organizational activities. This specicdomain of disaster response allowed us to focus attention on practical ac-tivities within a narrowly dened domain, analyze a specic set of relevantplanning documents (core and mass care plans), while providing a measureof comparability across the communities under study.

Testing these propositions requires data with a particular set of character-istics. The dataset must record participation within emergency managementnetworks. Furthermore, the observations must be ordered so that a time pathis clear. Ideally, the time path should record participation over a number of years preferably at least a decade (Sabatier 1993). Hypothetically, onecould conduct interviews or surveys annually over a decade but such effortsare incredibly expensive and, predictably, rare. The twin needs of compara-ble measurement and available data across time are best (or at least, mostfeasibly) served by documentary analysis. This paper will focus on two typesof documentary sources: media reports of evacuation related activity andformal emergency operations plans related to evacuation.

To collect these data, we located six comparable communities. We were

interested in evacuation hosting activities so we identied communities thathad recent (within the last decade) experiences with evacuation hosting activ-ities. These communities need to be similar in terms of population size. Forthis reason we chose moderately sized communities rather than the largestcities where the variances in size are large in absolute terms. The six commu-nities have populations between 200,000 and 1 million residents (when nothosting evacuees) and include four communities from Gulf Coast states andtwo from non-Gulf Coast states. In the interest of space, this paper will focuson two Gulf Coast communities: Brazos County, Texas and Caddo/BossierParishes, Louisiana.

3.1 Media Reported NetworksFor our rst data source, we have elected to use newspaper searches to gen-erate a database of media documents. Our goal was to create a single systemfor collecting journalistic coverage of emergency management networks that

7


10/38

could be used for a variety of communities across time. For purposes of this

study, we are focusing attention on evacuation related activities (within theentire realm of emergency management). We elected to search within theWestlaw database using the substantive search term evac! . This will cap-ture all words that begin with the letters evac including evacuation, evacuate,evacuee, and the like. The Westlaw database allowed us to search all newspa-pers and news wires - to ensure we captured local as well as national mediasources. We added geographic limiters to the substantive search term in-cluding the major cities and the county in which the community resides. Forexample, we looked for articles that included a term starting with evac andalso included either College Station , Bryan , or Brazos County . Given thevarying roles that county and city official play in emergency management,we felt it was essential to search based on city and county. This paper re-ports the results of document searches related to Brazos County, TX (whichincludes the cities of College Station and Bryan about 1.5 hours north-west of Houston) and Caddo and Bossier Parishes, LA (including the city of Shreveport). The use of media reports to reconstruct networks was inspiredby Comforts (2006) study of response networks to Hurricanes Katrina andRita.

These searches of the Westlaw database resulted in hundreds of hits foreach year of our sample (2000-2009). For the purpose of this analysis, weaggregated the media reports into two periods: 2000-2005 (up to and in-

cluding Hurricane Katrina) and 2006-2009 (post-Hurricane Katrina throughHurricanes Gustav and Ike). Each of these articles were then read individ-ually to ensure that the article was germane to issues of evacuation. Thiseliminated many articles. Some articles included references to entertainmentor sporting events in the target community (such as Texas A&M Universitysports teams) and coverage of something having to do with an evacuation ina different community. We only selected articles that involved an evacuationor evacuation hosting activity within the target community. We then readeach of the selected articles to identify all organizations mentioned.

3.2 Plan Based NetworksTo complement the media reports, we have also collected formal emergencyoperations plans from each of these communities. The emergency operationsplan serves as a primary coordinating documents for a variety of actors inemergency management within each community. The document lays out the

8


11/38

structure of authority and responsibility as well as providing an assessment

of locally prominent hazards and specic annex documents for a variety of detailed functions.We have coded each formal emergency operation plan to create a roster

of formally included members of the emergency management network withineach community. While the media reported network is a permissive samplethat includes a wide variety of actors, the formal plans tend to have muchsmaller rosters and focus on organizational with legally dened responsibili-ties within emergency management. Where possible, we have collected his-torical plans as well as the currently operations emergency operations plan.Interestingly, few of the emergency management offices within our communi-ties kept historical plans. When a new plan was formally approved, the oldone were often discarded without a copy being kept in the office. Where wehave found historical plans, it has been through the use of the web archiveservice The Wayback Machine at web.archive.org .

The collection of these two sources of data is not novel (e.g. Comfort,Oh & Ertan 2009). However, many studies aggregate the two data sourcesto create an inclusive list (cf. Kapucu & Demiroz 2011). We will maintainthe separation of the data sources to better contrast their strengths andweaknesses 3 .

4 Brazos County, TX ResultsBrazos County lies approximately 1.5 hours (by car) Northwest of the Hous-ton metroplex and has been involved in two major evacuation efforts: onein 2005 (including Hurricanes Katrina and Rita) and one in 2008 (includingHurricane Ike). Additionally, there was a notable local evacuation in 2009stemming from a release from a chemical plant. This local evacuation waslimited in duration but resulted in an evacuation order for most of the city

3 Even with the combination of these two sources, measurement error is possible. Themedia and plan roles were conrmed individually through the context of each documentto ensure that actors had a specic role in emergency management. This process is notinfallible, though. It is possible that an actor has escaped coverage by the media orinclusion in the plans. The justication for combining modes of measurement is to addressthis measurement error through the adoption of contrasting methods where the likelysources of error also contrast. The combination of the methods, then, should provide astrong basis for inference if, still, not perfect. Any roster method (surveys, interviews,eld research, etc.) encounters the same issue.

9


12/38

of Bryan.

4.1 Media Reported NetworksThe media reported networks for Brazos County are diverse and extensive with the active involvement of units of the Texas A&M System. We will lookat each of these rosters in turn. For each roster, the members in red uniquelyappear in that particular time period. Actors who persist across both timeperiods are in bold. For example, the EPA was mentioned in media accountsof evacuation efforts in Brazos County in the 2000-2005 but not in the laterwindow. By comparing the organizations listed in bold to those listed in red,you can see the types of organizations that persist through the period andthose that do not.

4.1.1 2000-2005

The media reported network roster for 2000-2005 is reported in Figures 1 and2. Here, as in the later rosters, the members appearing only on the currentlist are in a red font while members appearing on the roster for each timeperiod are in bold .

[[Insert Figures 1 and 2 About Here]]

The diversity of policy domains and organization types is remarkable.Within the government organization roster, public health, transportation,environmental quality, and housing oriented organizations are all part of themedia-reported networks. Among the actors who persist in the network,there are many educational and health care organizations (e.g. Texas A&MUniversity and the Texas A&M School of Rural Public Health) along withtransportation (e.g. Texas Department of Transportation (DOT)) and tradi-tional emergency management organizations (e.g. FEMA). Among the actorswho do not persist, there is a wider array of policy areas including law en-forcement (e.g. FBI and the Department of Justice), environmental (e.g.Texas Natural Resources Conservation Committee and the EPA), and hous-ing (e.g. HUD), among other policy areas. This roster represents a broadrange of participants with signicant change over time in support of propo-sition 1. While there is a set of actors present in both time period (whichrepresent candidates for core actors in the network), there is a great deal of

10


13/38

change across time even among government actors, challenging proposition

2. The roster of non-governmental organization experiences even more changeover time. While there is a diverse set of nongovernmental actors, all butone of these actors is missing on the later roster. While there was signicantturnover among governmental organizations, there was almost total turnoveramong the non-governmental organizations with the notable exception of theAmerican Red Cross. The difference in the turnover rate between governmentand nongovernmental actors supports proposition 3.

It is also interesting that emergency management organizations, whilepresent, are not particularly prominent on the list. Curiously missing arethe county and city emergency management organizations. State and federalemergency management organizations are present, but not the local offices.It is hasty to conclude from this exclusion that the county and city emergencymanagement offices were not active and important. However, their operationswere missing from media accounts of the activities.

4.1.2 2006-2009

The second time period covers the post-Katrina/Rita period that includedanother major hurricane evacuation (related to Hurricane Ike in the EastTexas area) and a local evacuation related to a chemical release (at the ElDorado Chemical Plant). The media reported network roster is presented inFigures 3 and 4.


In the government organization roster, university units are again promi-nent and persistent along with a handful of other organizations. In this laterperiod, local re and emergency management organizations make it into themedia reported network. The diversity of policy domains is still remarkable- though, often, the specic representatives changes. For example, the EPAdrops out during this time period but the TCEQ (the state equivalent tothe EPA) enters. This raises an interesting question for studies of policy

networks. How important is persistence of specic organizations within anetwork to ensure that a specic perspective (like, say, the importance of environmental issues) to be present within that network?

The specic members of the non-governmental organization roster changedconsiderably in this period. We already saw that only one non-governmental

11


14/38

actor is on both rosters. The new non-governmental actors include a number

of private sector companies specic to the context of the chemical releaseevent mirroring Comforts argument for spontaneous networks. The ag-gregation of the media networks to 5-6 year increments masks a great dealof annual variation. Within this time period, the number of organizationsvaried from 3 (in 2000) to dozens (in 2005 and 2008). This represents clearvolatility in network size on an annual basis. In reference to our propositions,the media reported networks provide evidence for volatility in network sizeand diversity. In the media reported networks, various sectors and policyelds are consistently represented but by different organizations.

4.2 Plan Based NetworksWe also collected rosters for the emergency management and mass care /sheltering from official emergency plans. This data source, while also docu-mentary, provides a quite different view of the relevant policy networks. Theemergency plans include those who are formally responsible for emergencymanagement and a variety of tasks related to evacuee sheltering. For eachplan, we identied all of the actors named as responsible parties in the gen-eral emergency management network within the core emergency operationsplan.

For the Brazos County community, we were able to locate two general(core) emergency operations plans that covered the sample time period. Therst plan is from 2004 while the second was approved in 2009. These plansprovide a pre-Katrina view and an update following the lessons of the varioushurricanes. As in the media reported networks, we have highlighted elementsin red that are unique to that plan (that is, not present in the other plan).

4.2.1 2004

Figures 5 and 6 presents the network roster from the perspective of theemergency plan.


The emergency operations plans provide a starkly different view of thelocal emergency management networks. Most obviously, the roster of the net-work is much shorter than the media reported network rosters. The rosters

12


15/38

are focused almost entirely on governmental organizations (with the notable

exception of the Red Cross) and the actors are largely persistent across theplans. This view of the emergency management network supports proposi-tions 2 and 3.

4.2.2 2009

The roster looks largely similar in the more recent emergency plans asrepresented in Figures 7 and 8.


Here again the roster is stable and focused almost exclusively on govern-mental organizations.From the perspective of the emergency operations plans, the network

includes little volatility over time. There was some consolidation betweenthe two time periods, but most of the actors involved through the decadewere in both plans. The actors stay largely the same contradicting 1. Afew actors drop out (e.g. Radiological Officer) and a couple of actors emerge(notably, the Salvation Army). A diversity of policy domains are representedbut the range is not as large or the membership as diverse as in the mediareported networks. There is some turnover (in support of proposition 1)but not as much as seen in the media data particularly among the non-governmental actors. Instead there seems to be a core of persistent actorswithin the plan consisting of government (notably, emergency managementorganizations) with the persistent addition of the American Red Cross andutility companies. If one had looked at only this data source, one couldreach a different conclusion regarding our propositions related to networkevolution certainly about the strength of proposition 2 and little supportfor proposition 3.

5 Caddo-Bossier Parishes, LA Results

The second community in our study is Caddo-Bossier Parishes in NorthwestLouisiana. These twin parishes include Shreveport and were a major evacu-ation site during Hurricane Katrina (and for months afterward). The key isto compare the networks for these two communities.

13


16/38

5.1 Media Reported Networks

Caddo and Bossier Parishes experienced a tremendous inux of evacuees including tens of thousands of residents for their shelters. These parishes arelocated far enough away from the coast to miss most of the extreme elementsof incoming hurricanes. It is a natural location for intra-state evacuee hostingfor Louisiana.

5.1.1 2000-2005

Figures 9 and 10 present the rosters of the media reported network for Caddo-Bossier Parishes from 2000-2005.

[[Insert Figure 9 and 10 About Here]]

As in the case of Brazos County, there is tremendous instability in theparticipants in the media reported network. In the government organiza-tion roster, the persistent components include local re and law enforcementorganizations along with state and federal emergency management organiza-tions. About half of the governmental actors are persistent. This representsevidence for proposition 1 and mixed evidence regarding proposition 2.

Some nongovernmental are persistent while most organizations intermit-tent participants from nonprot and religious organizations. The American

Red Cross is again a persistent actor. This is not a surprise given its role insheltering activities in most communities. However, it is one of a small num-ber of persistent non-governmental organizations. Here we see more supportfor proposition 3.

As to policy domain representation, we see a similar array in this commu-nity as in the last. Transportation and health care organizations are presentand persistent. There are other policy areas present (e.g. nutrition support,parks and recreation, and agriculture) but in a more intermittent way.

5.1.2 2006-2009

Figures 11 and 12 presents the media reported network for Caddo-BossierParishes from 2006-2009.


14


17/38

As in the previous community, there is diversity in terms of policy domain

and organization sector. Some of the instability is an artifact of name changesthrough the period (the state shifted to the title Office for Homeland Secu-rity and Emergency Preparedness). However, representation by local andstate emergency preparedness offices seems light compared to what one mightexpect. Propositions 1 and 3 nd strong evidence in the media-reported net-works for the Caddo-Bossier Parishes. There is inconclusive evidence relatedto proposition 2.

5.2 Plan Based NetworkAgain we will contrast the media reported network with the official emergencyplan for the jurisdiction. In the case of the Caddo-Bossier Parishes, we werenot able to collect a historical plan. We only have access to the currentplan from 2009. In seeking older versions of the plan, we were assured thatthe previous versions of the plan operative from 2000 were largely the sameas the current plan. When we explained that we were most interested inthe actors present and the relationships between actors within the plan, onerepresentative of the emergency management office said that there wouldbe very few changes. Given the few changes in the Brazos County plansdiscussed previously, this seems credible though we were unable to conrm itwith an old version of the document. For this reason, the color comparisonschange for the Caddo-Bossier plan rosters. Here bold indicates a matchbetween the rosters of Caddo-Bossier and the persistent actors in BrazosCounty. Red indicates an actor present only in the Caddo-Bossier plan and not in the Brazos County plan.

5.2.1 2009

Figures 13 and 14 presents the roster for the Caddo-Bossier Parishes emer-gency plan.


The place rosters for this community is larger than the plan-based rosterfor Brazos County and includes a broader range of organizations. This planincludes a broader range of government actors (including code enforcementand agriculture extension offices). Even with this diversity, though, the diver-sity of the plan network is still quite limited compared to the media-reported

15


18/38

networks. We can not directly evaluate the propositions in the absence of a

formal plan from the earlier time period. However, given the similarities be-tween the Caddo-Bossier plan and the Brazos County plan, we are condentthat the plans experienced similar changes over time. The Caddo-Bossierplan represents the plan after the dramatic evacuation hosting events asso-ciated with Hurricane Katrina. It still represents only a small number of sectors and policy domains.

Again, the conclusion one would draw in relation to our network evolutionpropositions would be quite different if you looked only at the formal plan.From the perspective of the formal plans, there is little evidence of variabilityor diversity. A small number of closely nit actors dominate the network fromthe perspective of the formal plan. In this regard, the network looks morelike a narrow policy subsystem than a broad and volatile issue network. Thegovernment actors are similar to those in the persistent core in the BrazosCounty plans but the few nongovernmental actors are also the persistentactors in the Brazos County plan. The transient actors seen in the mediadata are absent from the plan-based data.

Based on the Caddo-Bossier plan, we have support for proposition 2 butmixed evidence for proposition 3. We are unable to directly assess proposition1 given the absence of an earlier plan.

6 Lessons for Evacuation Hosting and Emer-gency Management

These data conrm that evacuation hosting is a complex and evolving enter-prise. The media networks identify a larger and ever changing networks of participants representing a broad range of policy domains and sectors. Theformal plans do not indicate the same level of diversity. There is not con-vergence on a single image of the respective networks. Instead, the differentdata sources tell somewhat different stories.

One could look at the data in an effort to declare one method morereliable or otherwise better than the other. A pessimistic reading of thediscrepancy between the results, for example, is that the plans are ignorantof, or unrelated to, the volume and nature of activities within the community.We are not convinced that this is the case. Instead, we see the plans asserving a quite specic service. The plans provide for a framework of formal

16


19/38

responsibilities. They do not prevent the emergence of new participants. The

media reports are consistent with this interpretations. The limited networksrepresented in the formal plans coexisted with large and diverse networkswithin the community.

What is not clear is whether the plans would be better off with a moreinclusive vision of their communities. Including a broad range of actors runsthe risk of diluting and confusing the core plan with references to participantswho may come and go over time. The broadly inclusive media networks alsoincluded a great deal of volatility with partners dropping out in the sec-ond time period. Some of the actors who dropped out may have stayed if included in the core plan; but some may have dropped out anyway leavingvestigial references to clot up the formal plan. The inclusion of the brokeragecomponents of the periphery directly completes with the closure of the corenetwork. While further research is needed on the balance of inclusiveness andfocus in formal plans, we encourage great care in creating formalized rela-tionships between non-traditional actors within the core emergency plan. Itmay be ideal to balance the vision advantage of brokerage with the reliabilityadvantage of closure by bifurcating the network into core actors (identiedwithin the formal plans) and peripheral actors (identied within the mediacoverage but absent from the formal plans). The results may be a hierar-chical network (Moynihan 2008). Engagement of non-traditional actors canwork without including those actors in the formal plans themselves.

One theme that emerges clearly from the data is that evacuation hostingis a community event that calls upon the resources of a broad range of actors.Emergency management professionals (including the consistent participantslike the American Red Cross) do not manage this process by themselves.Communities, not just agencies, host evacuees. Whether the formal plansinclude the broad range of actors engaged in evacuation hosting activities,emergency management professionals must be aware of the range of organi-zations that one may potentially recruit to assist in hosting activities.

7 Lessons for Network DynamicsThe lessons for network dynamics, more generally, depend entirely on whichapproach to network measurement one prefers. The formal plans providean image of the networks as limited in scope and stable over time. Thesmall number of participants in these networks tend to stay in the network.

17


20/38

The network provides little evidence of permeability to changes even in

what may be seen as a remarkable challenge to the network. These werenetworks that operated below the radar of public media attention. Despitethe high salience, there is little evidence of change in the participants asviewed through the formal plans.

The media reported networks tell quite a different story. Within thesetwo communities, we see dynamic networks that change size and membership.Each network includes participants from diverse policy domains and sectors.This view suggests participation by the broad civil society in issues of evac-uation hosting. Religious organizations provide a broad range of services aswell as serving as evacuee hosts. Similarly, private sector organizations areinvolved in services ranging from providing supplies and meals (sometimesat a remarkable discount). Government organizations from a broad range of policy domain, including most obviously, public health and transportation;but also social services, education, and environmental quality.

This equivocation within the data do not help to resolve the fundamentalquestion related to policy networks. Whether networks are diverse (as issuenetworks) or limited (as iron triangles) depends to a great extent on the dataone uses to view the network. The formal plans provide an image of limitedparticipation that is larger than the most strict iron triangle models, but stilllimited largely to a steady group of government organizations.

Our conclusion is that neither data source is accurate and both are

accurate. Each data source views the networks in different ways and fromdifferent perspectives. The formal plans focus on participants with speciclegal authority and responsibility. This is a narrow range of participation but an important one. The number of actors with legal authority and respon-sibilities related specically to emergency management activities is small andthere is little variation in this network. At the same time, there is a largernetworks of actors without specic authority and responsibilities who may(or may not) participate in evacuation hosting. These informal participantsare important but they play a different role. The informal actors are morelikely to disappear from the network (or appear only in the later time pe-riod). They are the component of the network that is volatile and diverse.

The data suggest that we need to move away from looking for a single net-work even within a single policy areas. Policy networks involve overlappingmemberships in various networks including core and periphery actors, thosewith and those without formal authority, and those whose presence is con-sistent and those who are not. Narrow and broad versions are both correct

18


21/38

images of networks; though of different networks within a policy area.

In the end, the networks are both diverse and homogeneous both steadyand volatile. There is an inner-core of actors whose membership is steadyover time. Interestingly, this group operates largely beneath the notice of the media. There is an outer-periphery of actors who enter and leave thenetwork quickly, represent a broad range of interests, and draw the attentionof the media. This combination may prove to be a strength of emergencymanagement networks. This may create the potential for both reliable op-erations from a formalized core and rapid reorganization of the supportinginformal ring (Moynihan 2008).

From the standpoint of the measurement of networks, this leaves an im-portant question. The pooling of these two methods of actor identicationpresume that they contribute to a vision of a single network. It may be moretelling to avoid such pooling and instead talk about the subnetworks sepa-rately. This is especially the case with the nongovernmental actors and themedia networks. In volume, the media network members outnumber thoseactors only present in the formal plans. If one pooled the rosters and thenperformed statistical tests or summaries on the aggregate pool, the dynamicsof the media pool would drive the results. Separate analysis of the formalrosters might be appropriate if ones theoretical interest involved the core setof actors. Based on the lack of a convergent image of the networks in thisstudy, we recommend choosing a pooling strategy carefully depending on the

theoretical interest of ones own study.

8 ConclusionNetwork analysis is an emerging and potentially powerful perspective forthe study of policy implementation generally and emergency managementnetworks specically. The results of studies of two emergency and sheltermanagement networks here provide some insight into the dynamics of thesenetworks over time but raise substantial questions about the nature of datacollection for network analysis.

Two different methods of data collection provide contrasting portraits of the emergency and shelter management networks. The media reported net-works provide a vision of a broad issue network including an ever-changingroster representative of various policy domains and organizational sectors.The plan-based networks are much smaller and focus almost exclusively on

19


22/38

government partners and a handful of longstanding nonprot partners. The

resulting image of the networks is of an inner-core of players surrounded bya volatile periphery of actors. This raises important questions for the studyof emergency management networks. Is the division of actors into such tiersefficient? Does it effectively focus attention on the most important actorswithout distractingly cluttering the formal plan? Does the informal roleplayed by so many actors create a disincentive for those actors to continueto participate or to contribute as much as they could to the hosting effort?How many domains within emergency management or in other policy ar-eas deliberately create and formalized such core and periphery distinctions?There is clearly a great deal still to learn about the structure and dynamicsof emergency management networks.

20


23/38

ReferencesBurt, Ronald S. 2005. Brokerage and Closure: An Introduction to Social

Capital . New York, NY: Oxford University Press.

Comfort, L.K. 1994. Self-organization in Complex Systems. Journal of Public Administration Research and Theory 4(3):393410.

Comfort, L.K. 2007. Crisis Management in Hindsight: Cognition, Com-munication, Coordination, and Control. Public Administration Review 67(s1):189197.

Comfort, L.K., N. Oh & G. Ertan. 2009. The Dynamics of Disaster Recov-

ery: Resilience and Entropy in Hurricane Response Systems 20052008.Public Organization Review 9(4):309323.

Comfort, Louise K. 2006. The Dynamics of Policy Learning: CatastrophicEvents in Real Time.. Presented at the 2006 Annual Meeting of theAmerican Political Science Association.

Heclo, Hugh. 1978. Issue Networks and the Executive Establishment. InThe New American Political Establishment , ed. Anthony King. Ameri-can Enterprise Institute pp. 87124.

Kapucu, N. 2006. Public-nonprot partnerships for collective action in dy-namic contexts of emergencies. Public Administration 84(1):205220.

Kapucu, N. 2007. Non-prot Response to Catastrophic Disasters. Disaster Prevention and Management 16(4):551561.

Kapucu, N. 2008. Collaborative Emergency Management: Better Commu-nity Organising, Better Public Preparedness and Response. Disasters32(2):239262.

Kapucu, N. & F. Demiroz. 2011. Measuring Performance for Collabora-tive Public Management Using Network Analysis Methods and Tools.Public Performance & Management Review 34(4):549579.

Kapucu, N., M.E. Augustin & V. Garayev. 2009. Interstate Partnerships inEmergency Management: Emergency Management Assistance Compactin Response to Catastrophic Disasters. Public Administration Review 69(2):297313.

21


24/38

Kapucu, Naim. 2009. Performance Under Stress: Managing Emergencies

and Disasters. Public Performance and Management Review 32(3):339344.

Kiefer, J.J. & R.S. Montjoy. 2006. Incrementalism Before the Storm: Net-work Performance for the Evacuation of New Orleans. Public Admin-istration Review 66:122.

McCool, D. 1998. The Subsystem Family of Concepts: a Critique and aProposal. Political Research Quarterly 51(2):551.

Meier, Kenneth J. & Laurence J. OToole. 2001. Managerial Strategiesand Behavior in Networks: A Model with Evidence from U.S. Pub-lic Education. Journal of Public Administration Research and Theory 11:271294.

Moynihan, D.P. 2008. Learning under Uncertainty: Networks in CrisisManagement. Public Administration Review 68(2):350365.

OToole Jr, L.J. 1997. Treating Networks Seriously: Practical and Research-Based Agendas in Public Administration. Public Administration Re-view 57(1):4542.

Provan, K.G., A. Fish & J. Sydow. 2007. Interorganizational Networks at

the Network Level: A Review of the Empirical Literature on WholeNetworks. Journal of Management 33(3):479516.

Redford, E.S. 1969. Democracy in the administrative state . Oxford UniversityPress, USA.

Robinson, S.E. 2007. A Seat at the Table for Nondisaster Organizations.The Public Manager 5:4.

Robinson, S.E., B. Berrett & K. Stone. 2006. The Development of Collab-oration of Response to Hurricane Katrina in the Dallas Area. PublicWorks Management & Policy 10(4):315.

Sabatier, P.A. & H.C. Jenkins-Smith. 1993. Policy Change and Learning:An Advocacy Coalition Approach . Westview Pr.

22


25/38

Sabatier, Paul. 1993. Policy Change Over a Decade or More. In Pol-

icy Change and Learning: An Advocacy Coalition Approach , ed. PaulSabatier & Hank Jenkins-Smith. Westview Press pp. 1340.

Simo, G. & A.L. Bies. 2007. The Role of Nonprots in Disaster Response:An Expanded Model of Cross-Sector Collaboration. Public Adminis-tration Review 67:125142.

Waugh, W.L. & G. Streib. 2006. Collaboration and leadership for effectiveemergency management. Public Administration Review 66(s1):131140.

23


26/38

9 Appendix: Figures


27/38

Figure 1: Brazos County 2000-2005 Media Report Government Membership unique elements are in red and persistent elements are in bold .

TAMU

Army AF Instr.School

TX Agg. YouthCamp

TX Eng ExtServ

FEMA

TAM HRRC

TX DOT

TAM SRPH

EPA

TX Nat. Res.Cons. Comm.

TX Department of Health

NOAA

Ntnl Hur Cntr

FBI

TX DPS

TX General Land

Office Texas Parks and

Wildlife

Nat. Coun. onRad. Prot.

TTI

Em. Prep. Inst.

DHS

DoJ

TAMU Coop.Inst. for Appl.Mtrlgcl Stud.

Offshore Tech.Research Center

TX NationalGuard

Ntnl Svr Strm Lab

Ntnl Wthr Srvc

Bryan High

School Tulane Univer-

sity

Real Estate Cen-ter at TAMU

TAM Hlth SciCntr

SBA

HUD

UNO

Aggie Guide Dogsand Service Dogs

TX Board of Health

TAM Galveston


28/38

Figure 2: Brazos County 2000-2005 Media Report Non-governmental Mem-bership unique elements are in red and persistent elements are in bold .

Lake JacksonCivic Center

American RedCross

Exxon MobilsFire School

Lakeview MthdstCnf Cntr

La Quinta

S. TX Cncl BoyScouts

Habitat for Hu-manity

Texas City PrairiePreserve

Houston Chroni-cle

Southeastern Li-brary Network

TX Oil and GasAssociation

TX Tank CarriersAssociation

TX Petro. Mar-keters and Con-ven. Stores

Exxon Mobil

Association of Former Students


29/38

Figure 3: Brazos County 2006-2009 Media Report Government Membership unique elements are in red and persistent elements are in bold .

TAMU

TX EngineeringExtension Ser-vice

National Hurri-cane Center

Tulane Univer-sity

TAMU Galve-ston

Center for Retail-ing Studies

College StationFire Department

University PoliceDepartment

Env. Health andSafety Depart-ment

TAMU PhysicalPlant

CommunityEOC

Hazard Reduc-tion and Recov-

ery Center TAM HSC Of-

ce of HS

TAM Center forHomeland Secu-rity

Texas Home-land Security

FEMA TX DOT

TTI

TX NationalGuard

NationalWeather Ser-vice

TAM HSC

US Public HealthServices

Teaching, Learn-ing, and Cul-ture Department- TAMU

Brazos CountyCouncil of Gov-ernment

Texas Task Force1

Bryan Fire De-partment

Roberston CountyEMS

Bryan Police De-partment

TCEQ


30/38

Figure 4: Brazos County 2006-2009 Media Report Non-governmental Mem-bership unique elements are in red and persistent elements are in bold .

Atmos Energy

American RedCross

A&M UnitedMethodist Church

Plaza Hotel

Christ UnitedMethodist Church

Blood Center of Brazos Valley

St. Joseph Reg.Health Center

CS Medical Cen-ter

Brazos ValleyPhysicians Orga-nization

Carter Blood Care

BCS Conv. Cen-ter

Brazos CountyFood Bank

Texas Hotel andLodging Associa-tion

Texas PublicPower Association

Bryan HEB

CS Wal-Mart

El Dorado Chemi-cal Company

Grace BibleChurch

Bryan Police De-partment


31/38

Figure 5: Brazos County 2004 Emergency Operations Plan GovernmentMembership unique elements are in red and persistent elements are inbold.

County Judge/CM

Asst. County Judge/CM

EMC

Law Enforcement

Fire Services

Public Works

Health Services

Human Resources

Community Development

Human Services

Tax Assessors

Transportation/ISD

City/County Attorney

Radiological Officer

Public Information Officer

Parks and Recreation

Treasurer

Justice of the Peace

Figure 6: Brazos County 2004 Emergency Operations Plan Non-governmental Membership unique elements are in red and persistent el-ements are in bold.

Utilities Red Cross


32/38

Figure 7: Brazos County 2009 Emergency Operations Plan Membership unique elements are in red and persistent elements are in bold.

County Judge/CM

Asst. County Judge/CM

EMC

Law Enforcement

Fire Services

Public Works

Health Services

Human Resources

Community Development

Human Services

Tax Assessors

Transportation/ISD

City/County Attorney

Search and Rescue

Figure 8: Brazos County 2009 Emergency Operations Plan Membership unique elements are in red and persistent elements are in bold.

Utilities

Salvation Army

Red Cross


33/38

Figure 9: Caddo-Bossier Parishes 2000-2005 Media Report GovernmentalMembership unique elements are in red and persistent elements are inbold.

S Police Dept. C Sheriff Dept.

National Guard

DHHS

LA DOT

FEMA

US PH Service

S Fire Dept.

NWS

Barksdale AFB

US DHS

Dept VA

LSU

LA Rec. Auth.

FBI

LSU Hospital

LSU HSC

LA DHH US Army

LA Dept. of Ag.

LA Wild. andFish.

LA Dept of PH

US Coast Guard

US Nat. Hurr.Cntr.

VA Hospital

Sthn. University

Calcasieu ParishOHSEP

England AFB

US Forest Service

LA Do Labor


34/38

Figure 10: Caddo-Bossier Parishes 2000-2005 Media Report Non-governmental Membership unique elements are in red and persistent el-ements are in bold.

American RedCross

St. Mary Place

LA HS Athl.Ass.

NO Bus. Council

Shreve Human So-ciety

LA Dis. Rec.Found.

Hab. for Human. S-B Com. Re-

newal

LA IA Found.

Fuller Center

S Fabricators

Coca-Cola

LA Oo Tourism

LA Hospital As-soc.

Goodwill

ORU

St. Lukes Meth.Ch.

Unity in Prayer

Salvation Army

Grace UnitedMeth. Ch.

Natnl. Low Inc.Hous. Coal.

Prof. CLN Assoc.

Brammer Engi-neering

Boffier Civic Cen-ter

Fam. Rec. Corps

US Humane Soc.

Hibernia Corp.

NPR

S Expo Hall

Hirsch Mem.Col.

Indep. Stadium Cham. of Comm.

NAACP

SW Assoc. of Epis. Sch.

Food Bank of S

S Charity Hospital


35/38

Figure 11: Caddo-Bossier Parishes 2006-2009 Media Report GovernmentMembership unique elements are in red and persistent elements are inbold.

Shreveport Po-lice

FEMA

LSU

LA Recovery

LSU Health Sci-ence Center

LA Tech

LA Dept. of Envi-ronmental Quality

LA Public ServiceCommission

LA GOHSEP

LA Poison Con-trol Center

Texas SouthernUniversity

LA Education De-partment

LA DHH

Caddo ParishSheriff Depart-ment

LSU

LSU Hospital

Desoto ParishSheriff Depart-ment

Southwood HighSchool

LA NationalGuard

LA DOT

Shreveport FireDepartment

Department of So-cial Services

LA Animal Re-sponse Team

Caddo OHSEP


36/38

Figure 12: Caddo-Bossier Parishes 2006-2009 Media Report Non-

governmental Membership unique elements are in red and persistent el-ements are in bold.

LA HightSchool AthleticAssociation

LA Association of Business and In-dustry

Fuller Center Assoc. of Comm.

Orgs. for ReformNow

Hal Sutton Foun-dation

NAACP

LA Nursing HomeAssociation

American Red

Cross United Methodist

Committee on Re-lief

Salvation Army

Sams Club inShreveport, LA

Earl K. LongMedical Center

LA Vet Med. As-sociation

Hirsch Memo-rial Coliseum


37/38

Figure 13: Caddo-Bossier Parish 2009 Emergency Operations Plan Govern-ment Membership unique elements are in red and persistent elements arein bold.

Caddo-Bossier OHSEP

Law Enforcement

Fire Service Emergency Medical Ser-

vices

Caddo and Bossier HealthUnits

Coroners Office

Public Works Departments

Water and Sewer Depart-ment

Parks and Recreation Depart-ments

Caddo and Bossier SchoolSystems

Shreveport Airport Authority

SPORTRAN

City and Parish Legal Depart-ment

City and Parish PlanningDepartment

Building Inspection/Code En-forcement

City and Parish FinanceDepartment

City and Parish Fleet Ser-vices

County Agents, Ag ExtensionServices, and Soil ConservationServices

Caddo and Bossier Officesof Family Support, Coun-cils on Aging, and Commu-nity Action Agencies

Military Units


38/38

Figure 14: Caddo-Bossier Parish 2009 Emergency Operations Plan Non-governmental Membership unique elements are in red and persistent ele-ments are in bold.

Hospital and Medical Cen-ters

Private utility Companies

(Natural Gas, Electric, and

Telephone)

American Red Cross

Salvation Army present only

in the later Brazos County plan

Robinson Evolution of EM Networks

Documents

Transcript of Robinson Evolution of EM Networks