Content uploaded by Rong Wang
Author content
All content in this area was uploaded by Rong Wang on Dec 17, 2018
Content may be subject to copyright.
1"
Is Collective Impact the Destination? A Typology of Interorganizational Collaboration
Rong Wang, University of Kentucky, rong.wang@uky.edu
Katherine R. Cooper, DePaul University, kcooperm@depaul.edu
Anne-Marie Boyer, Northwestern University, annemarie@u.northwestern.edu
Shaun M. Dougherty, Vanderbilt University, shaun.dougherty@vanderbilt.edu
Michelle Shumate, Northwestern University, shumate@northwestern.edu
Paper draft prepared for ARNOVA 2018. Please do not cite without authors’ permission.
2"
Abstract
“Collective impact” has gained prominence as a particular means for organizations to respond to
social problems in their community, though there is some concern that the term is over-used or
improperly applied. In this study, we draw from research on collective impact, collaborative
initiatives and network governance to suggest that what constitutes “collective impact” varies
widely by community. We introduce a 2 by 2 matrix to describe a variety ways of organizing
partners along two dimensions: a). the degree to which program planning and implementation are
enacted by centralized leadership, and b). the degree to which cross-sector partners engage in
collaboration. With interview and archival data collected from 28 communities across the United
States, this study suggests that these networks may be classified according to one of four
approaches: holistic coalitions, low-overhead coalitions, community-led coalitions, and multi-
stakeholder coalitions.
Keywords: Collective impact, coalitions, network governance, cross-sector, education reform
3"
Is collective impact the destination? A typology of interorganizational collaboration
Interorganizational collaboration is a long-used approach for responding to complex
social problems within communities, and though various models and approaches to collaboration
exist, recent academic and practitioner conversation has focused on one model in particular.
Since its 2011 introduction in Stanford Social Innovation Review, collective impact has been
touted as a model for improving educational outcomes and health disparities, among other
concerns, and has been adopted in hundreds of communities across the United States and the
world (Kania, Hanleybrown, & Splansky Juster, 2014; Kania & Kramer, 2011). Despite the
popularity of this model, it remains somewhat controversial with respect to its effectiveness
(Walzer, Weaver, & McGuire, 2016; Wolff, 2016) and distinction from other collaborative
approaches (Christens & Inzeo, 2015).
In particular, we question whether the term “collective impact” is overutilized, and
whether communities would benefit from the availability of other terms or approaches to
describe their efforts. Advocates for collective impact have argued for reserving this term for a
specific type of collaborative initiative (see Edmondson & Hecht, 2014; Kania et al., 2014), and
previous research suggests that communities that use this term may adopt it differently, with
different implications for organizational participation (Cooper, 2017). In this paper, we further
explore the nuances among self-described collective impact initiatives and other models of
collaboration in the hopes of better understanding why communities adopt different approaches
and how they structure their initiatives as a result.
This paper is structured as follows. First, we review the definition of collective impact
and a growing body of related academic literature. Second, we situate our understanding of
4"
collective impact in the broader discourse concerning cross-sector collaboration and community-
based coalitions. Third, we review network literature, and in particular network governance. We
then introduce our methodology, a grounded approach that relies on interview data and archival
documents from 28 communities across the United States.
This paper makes two contributions. First, we introduce what we believe to be one of the
largest and most diverse samples of community-based, interorganizational collaboration. Second,
we introduce a typology that draws upon elements previously discussed in nonprofit and
collaboration literature, such as network governance (Provan & Kenis, 2008), and cross-sector
collaboration (Bryson, Crosby, & Stone, 2006). Ultimately, our analysis suggests that
communities may be organized according to two factors: the degree to which partners from
different sectors engage in collaboration, and the degree to which program planning and
implementation are enacted by centralized leadership. We then suggest a 2 by 2 matrix that
encompasses four different approaches to local networks: holistic coalitions, low-overhead
coalitions, community-led coalitions, and multi-stakeholder coalitions. Examples for each
collaboration type explain why a community is located in that quadrant and its key
characteristics. Implications of this typology for both researchers and practitioners are discussed.
Literature Review
Collaboration among nonprofits as well as between sectors are commonly discussed in
nonprofit literature (see Gazley & Guo, 2015). Although related concepts surface in both the
collaboration and network literatures, each area tends to rely on its own terminology and
theories. Some models favored by practitioners, such as collective impact, come with their own
sets of terms. In this section, we provide a brief overview of collective impact, placing it in the
context of cross-sector collaboration and network governance literatures.
5"
Collective impact
Collective impact was introduced by consultants in a Stanford Social Innovation Review
article as a way for nonprofits to avoid isolated impacts by leveraging their efforts with other
cross-sector partners in a measured, managed way (Kania & Kramer, 2011). Specifically, Kania
and Kramer (2011) describe collective impact as “the commitment of a group of important actors
from different sectors to a common agenda for solving a specific social problem” (p. 36).
Kania and Kramer - and many additional consultants or practitioners that utilize the term
- have been clear that for an initiative to actually be considered collective impact, it must meet
several conditions. Collective impact networks rely upon a “backbone” organization to manage
the partnership and requires that partners agree upon a common agenda; partners pursue their
shared goal by engaging in shared measurement, mutually reinforcing activities to meet that
goal, and continuous communication. In the time since the term first appeared, hundreds of
collective impact initiatives have been adopted by communities (Kania, Hanleybrown, &
Splansky Juster, 2014) and by such varied groups as school districts, local governments, state
governments, philanthropies and foundations, and the White House (Christens & Inzeo, 2015).
Initially put forth by consultants, collective impact was met with some skepticism from
researchers, who have long focused on related issues such as collaboration and networks (see
Christens & Inzeo, 2015). But in recent years, researchers have begun to study collective impact
initiatives in addressing social issues such as health disparity, and education reform (Cooper,
2017; Lawlor & Neal, 2016). Their approaches vary. In a reflection of philanthropic interest in
collective impact, many of the existing large-scale studies of collective impact are reports
commissioned by foundations (e.g., Henig, Riehl, Houston, Rebell, & Wolff, 2016). Some
researchers have focused on particular stakeholders in collective impact, such as the role of the
6"
facilitators (Gillam, Counts, & Garstka, 2016), or the presence and participation of nonprofit
stakeholders in collective impact initiatives (Cooper, 2017). Others have focused on the internal
or external contexts of collective impact, finding the internal dynamics a unique context for
exploring issues of dialogue and power (see Page & Stone, 2017) or focusing on the community
characteristics that lead to the creation of a collective impact initiative (Boyer, Cooper,
Dougherty, Wang, & Shumate, 2018).
Many have been critical of collective impact, with much of the criticism focusing on two
concerns. The first is that from researchers and practitioners is that collective impact relies too
heavily on the involvement of formalized power, in particular organizations, the business sector,
or leaders (Wolff, 2016; Wolff, Minkler, Wolfe, et al., 2017) and may omit local or grassroots
involvement (Christens & Inzeo, 2015); Cooper’s (2017) study suggested that collective impact’s
reliance on data and the financial resources required to maintain these initiatives present
challenges for nonprofit participation. The second commonly discussed criticism of collective
impact is that advocates for this approach do not necessarily place collective impact in a broader
context of collaborative and community-based models (Christens & Inzeo, 2015). In the next
section, we revisit the broader literature on cross-sector collaboration.
Cross-sector collaboration
The definition of collective impact specifies that actors from various sectors are needed
to solve complex social problems (Kania & Kramer, 2011). However, the concept of different
sectors working together predates the collective impact movement. Cross-sector collaboration
has been described as “the linking or sharing of information, resources, activities, and
capabilities by organizations in two or more sectors to achieve jointly an outcome that could not
be achieved by organizations in one sector separately” (Bryson, Crosby, & Stone, 2006: 44).
7"
These partnerships may include alliances among government, business, nonprofits and
philanthropies, communities, and the public. Bryson et al. (2006) cite several propositions that
sound similar to those concepts later stressed by Kania and Kramer (2011) in their description of
collective impact. For example, Bryson et al. also emphasize the importance of sponsors, a
similar understanding of the problem, continuous trust-building activities, and the gathering and
interpretation of data.
Additionally, many other scholars have focused on cross-sector partnerships in varying
forms, such as the relationship between nonprofits and business (Austin, 2000; Austin &
Seitanidi, 2012; Rondinelli & London, 2003) and nonprofits and government (Gazley &
Brudney, 2007), or across all three sectors (Selsky & Parker, 2005); many of these researchers
include typologies to describe the ways these sectors interact with one another and to what end.
Other researchers have focused on typologies that classify interorganizational partnerships
depending on activity (Snavely & Tracy, 2000) or the degree to which the partners are integrated
(Kagan, 1991; Keast, Brown, & Mandell, 2007) or the formality of the partnership (Guo & Acar,
2005).
Despite the existence of so many typologies, however, there is little research to depict the
extent to which different sectors are involved, and whether the presence or absence of various
sectors influence the way partnerships are structured. Although Cooper (2017) suggested that
self-identified collective impact initiatives may vary in terms of the extent to which sectors are
represented or play a role within the initiative, previous research has been limited to small
samples and has not explored cross-sector engagement across a large sample of networks. Thus,
we ask:
8"
RQ1: How do different sectors engage in collaborative initiatives in response to
community social issues?
Though we suspect that the presence or absence of various organizational partners and
sectors is important in shaping a network, the structure of networks likely also matters. In
particular, we explore the role of network governance in the next section.
Network governance mechanism
Education collaboratives, such as collective impact, are often depicted as networks.
Following Provan and colleagues, we defined these networks as “whole networks” which consist
of “multiple organizations linked through multilateral ties” to facilitate achievement of a
common goal (Provan, Fish, & Sydow, 2007, p. 482). Examining the whole network helps to
uncover how collective impact initiatives evolve and how they are governed within their specific
community contexts. Studying the whole network also provides implications for how to better
structure multilateral collaboration to accomplish collective goals.
Networks are thought to be more flexible and adaptive compared to other forms of
collaboration (Powell, 1990; Powell, White, Koput, & Owen‐Smith, 2005). However, as pointed
out in Provan et al. (2007: 482) whole networks "are often formally established and governed,
and goal-directed, rather than occurring serendipitously" and that "relationships among network
members are primarily non-hierarchical and participants often have substantial operating
autonomy". This line of research leads to the conceptualization of the network governance
model, which focuses on analyzing the structure and process of whole networks (Provan &
Kenis, 2008).
Network governance framework is built upon the assumption that network coordination
could lead to positive network outcomes such as enhanced learning, more efficient use of
9"
resources, and increased capacity in addressing complex problems, and understanding the
functioning of networks helps to explain why networks produce certain outcomes (Provan &
Kenis, 2008). The network governance approach views networks as the unit of analysis, which
moves beyond the sum of organizational actors and their links to investigate the variation of
networks regarding their structural characteristics.
To solve collective problems, the right amount of integration across and among network
members is critical, as too little coordination and collaboration may limit the value of adopting a
network form; yet too much integration may be too costly and not necessarily beneficial (Provan
& Lemaire, 2012). Locating the right amount of integration is related to different roles a network
facilitator could play. As laid out in Provan and Kenis (2008), network governance mechanisms
could range from self-organizing, to lead-organization governed, and to network administrative
organization governed. In self-organizing model, there is no existence of a distinct or formal
administrative entity and the network acts collectively. Organizational members are responsible
for managing both internal and external management relations with stakeholder groups. In lead-
organization governed networks, one organizational member functions as the coordinating body
for the network and facilitates the activities of other partners to achieve the collective goal. The
lead organization may play the role of fiscal agency to seek and control access to external
funding. In network administrative organization governed models, a separate entity is set up to
manage and sustain partnership activities. It differs from the lead-organization governance model
that the administrative body is not a member of the network providing its own services. The
network is externally governed.
Another related characteristic is the degree to which an entity has central control over the
funding within the network. When a network is more centrally governed, funding management
10"
takes on a top-down approach where a government body defines how the funding will be
allocated and monitored (Provan & Huang, 2012). In a less centrally governed network, no
organizational members dominate in the flow of funding and members have flexibility in how to
seek and manage their funding sources. The degree of centralized funding captures the extent to
which funds are viewed as a property of a network, thus affording a broad pattern of interaction
among members (Provan & Huang, 2012).
The other characteristic of network governance is the problem or agenda that brings
together an interorganizational network. In more centrally governed networks, there is a focused
agenda all the members collectively tackle. It is possible that there is considerable variance
across network members regarding their expertise areas and focal issue areas. Centralized
governance helps in this situation to narrow down the issue area and strategically coordinate
operational decisions. It ensures all members are in agreement with central issues and are
committed to align all their effort (Lawlor & Neal, 2016). In less centrally governed networks,
members are given enough flexibility to work on multiple agenda.
Given the implication of network conveners for questions of a common agenda and
program planning, we examine how these networks are organized:
RQ2: How do collaborative initiatives self-organize in response to community social
issues?
The focus on the two elements in organizing partners to solve social issues helps us to
uncover the variations that might emerge from diverse communities. Therefore, we examine the
following research question to address an overarching inquiry:
RQ3: How do cross-sector engagement and network governance shape the dynamics of
collaborative initiatives?
11"
Methods
Participants & procedure
This study draws from archival data and interview data from 28 collaborative initiatives
aiming to improve educational outcomes. See Table 1 for the list of communities sampled. These
initiatives were sampled through the following procedures. First, we compiled a database of
collective impact networks by identifying member affiliations with national networks such as
StriveTogether, Collective Impact Forum, Ready by 21, and the Campaign for Grade-Level
Reading. Second, we received participation consent from 14 communities that were on our
recruitment list who also self-identifies as collective impact sites. Third, we used the following
matching criteria to locate another 14 communities to be added in our sample: geographic (e.g.,
population density, coverage area, and number of school districts or municipalities),
demographic (e.g., race, and poverty rate), and labor market (e.g., unemployment rate, and
median income). The second half of our sample tend to be less structured though some of them
self-identified as early-stage collective impact sites.
Interviews were conducted with all the 28 communities. Questions included their history,
mission statement, funding sources, strategies used to align partners, community engagement
activities, and data collected. In addition, we also requested for archival data from all these
communities, including their founding documents such as Memorandum of Understanding
(MOU), meeting notes, and partner roster.
Coding procedures
Because previous research has hinted at the role of the convener and the representation of
the different sectors (Cooper, 2017), and both cross-sector engagement e.g., Austin, 2000;
Gazley & Brudney 2007; Selsky & Parker, 2005) and network governance (e.g., Provan &
12"
Kenis, 2008) are well covered in collaboration research, we opted to take a deductive approach
to coding that relied largely upon procedural methods, or more “standardized” ways of coding
data (Saldaña, 2013, p. 67). As a result, we drew upon these previous studies to develop
provisional codes that fit our study’s “conceptual framework, paradigm, or research” (Saldaña,
2013: 65), and created codes pertaining to the presence and engagement of different sectors and
different forms of network governance. We then relied upon magnitude coding, and pilot-tested
our coding choices as recommended by Saldana (2013).
Magnitude coding typically builds upon existing codes by adding a category to indicate
its presence or intensity (Saldana, 2013). For each community, we coded two dimensions:
network governance and cross-sector engagement, on a scale ranging from -2 to 2. Along the
dimension of network governance, a community is coded as 2 if it has at least 3 of the following
characteristics: a strong presence of a lead agency, a clear top down structure of how the lead
agency works with the partner organizations, a structured way of how funding will be
distributed, and a clearly defined common agenda. It is considered being most centrally
governed. If a community has 1 or 2 of the above mentioned characteristics, it is coded as 1
indicating the network is to some degree centrally governed. If a community has at least 3 of the
following characteristics, it is coded as -2 indicating it is most decentrally governed: lack of
coordination in project planning or implementation among partners, tackling multiple agendas,
no strong presence of a lead agency, and loose structure of how funding will be distributed. If a
community has 1 or 2 of the above mentioned characteristics, it is coded as -1 indicating it is to
some degree decentrally governed.
Along the dimension of cross-sector engagement, a community is coded as 2 if it has the
following characteristics: a strong presence of cross-sector partnerships (eg: presence of
13"
nonprofits working closely with government or the private sector), open and sustained
communication among partners. If a community has some presence of cross-sector partnerships
(eg: presence of nonprofits working closely with government or the private sector), but
communication among partners is not well sustained (eg: less frequently meeting), it is coded as
1. If a community has a single sector dominating, but still maintains some level of
communication among partners, it is coded as -1. If a community has a strong presence of a
single sector dominating and no sustained communication among partners, it is coded as -2.
Results
Our magnitude coding led us to create a continuum along two dimensions: the degree of
cross-sector collaboration, and the extent to which the network is centrally governed. Each of
these is further described below.
Mode of Partnership Engagement
To answer RQ1, we found that private, public and nonprofit sectors decide how to engage
in collaborative initiatives in response to community social issues by considering the following
factors: the availability of shared resources, continuous communication among partners, and the
existence of a dominating sector in leading the effort. To further explain the different patterns of
stakeholder collaboration, we propose the model of partnership engagement which moves along
the continuum of high to low cross-sector engagement.
In high cross-sector engagement, there is a strong presence of cross-sector partnerships.
For example, nonprofits work with businesses, or government agencies form partnerships with
nonprofits. In some cases more than two sectors are engaged in the collaboration. Partners tend
to have open and continuous communication which is conducive to the mutually reinforcing
activities emphasized by proponents of collective impact.
14"
At the other end of the continuum is low cross-sector engagement, where there is a single
sector dominating the implementation. The collaboration network could be led by a nonprofit or
a government agency. Such limited cross-sector engagement may pertain to the lack of interest,
involvement, or resources from cross-sector partners, or may reflect that the size of the
community simply does not enable cross-sector engagement. Consequently, there is no sustained
communication among partners.
Mode of Network Governance
To answer RQ2, we propose another model to describe the degree of centralized
leadership in planning and implementing programs, ranging from centralized to decentralized
governance. In centralized governance, there is a strong presence of a backbone organization, a
structured way of distributing funding, and a commonly shared agenda - characteristics
originally described in the collective impact model. The role of a backbone organization is to
ensure all partners work toward a collective goal through “ongoing facilitation, technology and
communications support, data collection and reporting, and handling the myriad logistical and
administrative details needed for the initiative to function smoothly” (Kania & Kramer, 2011:
40).
In decentralized governance, partners tend to be self-organizing and lack coordination in
program planning or implementation. Partner organizations may work together without
completely agreeing on the problem to be solved, and thus they tend to work on multiple agendas
under distributed leadership. In addition, there is no or no strong presence of a backbone
organization and generally the initiative has a loose structure in how to distribute the funding to
different programs. As is the case with partner engagement, the tendency to be less organized
15"
may be a result of resources available in the community or a conscious choice to work together
in this fashion.
Based on the coding results of 28 communities, we propose a Community System
Solutions framework to capture how organizations collaborate to solve social issues without
assuming whether a particular form of collaboration is likely to be more effective than others.
The framework is depicted in Figure 1, which is composed of a 2X2 matrix and illustrates four
quadrants of community system solutions derived from the intersection of the axes. These
quadrants help to explain a diverse set of ways that move beyond the collective impact model.
Mapping the Community System Solutions
Four quadrants are derived from the intersection of the axes to depict different ways of
organizing organizational partners for solving community-based problems. In this section, we
draw examples from our sampled communities to showcase the key characteristics of each
quadrant; in response to RQ3, we found that depending on community issues at hand, history of
partnership in the community, and the level of sufficient resources, the dynamics of collaborative
initiatives can be depicted in the following 4 models of coalition. See the mapping of all the
sampled 28 communities in Figure 2.
Holistic Coalition
This quadrant captures networks that are highly centralized in program design and
implementation, and also are characterized by high cross-sector engagement. Communities
located in this quadrant are often collective impact initiatives that have been operating for a
while. Our coding identified 14 of the sampled communities within Quadrant 1. Though these
communities share some common characteristics, they also differ in their collaboration patterns.
16"
In this section we draw from several communities to explain why they chose to work in this
particular structured way.
Within the Quadrant 1, the following eight communities were coded as both highest in
network governance and diverse partnerships: Summit Education Initiative (SEI) in Akron Ohio,
Higher Expectations of Racine County in Wisconsin, ROC the Future in Rochester NY, Achieve
the Brown County in Wisconsin, Learn to Earn Dayton in Ohio, Family Success Alliance in
Orange County North Carolina, Impact Tulsa in Oklahoma, Flint and Genesee Literacy Network
in Michigan. All of these initiatives are affiliated with national networks such as StriveTogether
and Collective Impact Forum, and they also identified themselves as Collective Impact. In each
community, the backbone organization plays a significant role in aligning partners’ effort toward
a collective agenda, and functions as the fiscal agency. All the partners in these networks meet
regularly and have a strong focus on data collection and data use. To explain these characters in
more detail, we will discuss two communities specifically. The first one is SEI which was
established in 1994 to improve reading scores in the Akron Public Schools. With sufficient
funding to support programming, SEI has a strong backbone organization that helps manage over
300 partners, including school districts, higher education institutes, investors, and community-
based organizations. All partners collaborate on the principle of “acting on education data.” SEI
maintains a solid database. Another example to highlight is ROC the Future, which was founded
in 2011 in a community of a rich history of cross-sector collaboration. Its strong backbone
agency organizes the coalition around different task forces that are defined by academic
outcomes such as school readiness network, attendance network, and college access network.
The diverse partnerships are built among local nonprofits, research institutes, local foundations,
businesses, and government agencies.
17"
Another set of examples are communities that are slightly high on centralized governance
and highest on cross-sector engagement, indicating that the backbone organizations tend to play
a coordinating role and provide some degree of autonomy for its diverse network members.
These include: Coalition for New Britain’s Youth in Connecticut, Westbrook Children's Project
in Maine, Youth Thrive in Wake County North Carolina, and Hartford Partnership for Student
Success in Connecticut. In these communities, the backbone agency facilitates efforts from
diverse sectors but sometimes the initiative as a whole tends to tackle more than one collective
agenda. All of these communities are Collective Impact sites except Hartford Partnership for
Student Success led by a United Way in the community.
The other two communities that are also located in Quadrant 1 but coded as lower on
centralized governance and lower on cross-sector engagement: Berkshire United Way in
Massachusetts, and Communities that Care in Franklin Massachusetts. These are both Collective
Impact sites too. The Berkshire initiative has multiple collective impact initiatives which are set
to tackle different social issues such as reducing teen pregnancy, improving early childhood
academic performance, building a drug-free community, positive youth development, and parent
engagement. Though there is a strong presence of a backbone organization, the network itself
advocates the model that heavily relies on volunteers from local communities to implement
programs. Communities that Care in Franklin County as a network brings youth, parents,
schools, community agencies and local governments to promote the health and well-being of
youth people, and thus education is only a part of their bigger picture. The backbone
organization views itself as a facilitator of the partners.
Holistic coalition requires sufficient financial and human resources to sustain. The
advantage of this model is that partnerships’ effort toward a common agenda can be align
18"
efficiently. However, the downside is that the network lacks direct community engagement.
Most of the communities located in the Quadrant 1 from our sample heavily rely on their local
partners to gain access to local communities to raise the awareness of education reform.
Low-Overhead Coalition
This quadrant is characterized by networks that feature low-cross-sector engagement and
emphasize centralized governance. As suggested in the network governance literature, networks
that are centrally governed typically have a strong lead agency that takes the initiative to
coordinate and manage other partners; however, networks that fall in this category typically have
fewer partners to manage and, given the low cross-sector engagement, little representation from
business and government sectors. This approach is commonly seen among newer initiatives,
managed by an interested or funding party with limited resources that typically rely on existing
partnerships in the community.
Examples of low-overhead coalitions within our sample include the Impact Committee
for Education in Davidson County North Carolina, and the United Way of Saginaw Michigan.
These networks founded in 2015 and 2014, respectively, were, at the time of data collection,
represent fairly young initiatives that are still building key partnerships and amassing resources.
Both networks are centrally governed, in this case by local United Way agencies that also act as
funders.
Within Davidson County’s Impact Committee for Education, the United Way acts as the
lead agency by coordinating meetings and managing communication among partners. That said,
their partnerships at time of data collection were few and relied typically among local nonprofits,
(e.g.,the Boys and Girls Club of Davidson County, YMCA, and other local nonprofits that offer
education and community outreach programs). The partnership includes the local school districts
19"
but only one local business; and at the time of data collection, leadership was seeking to recruit
local superintendents and policymakers. Likewise, in Saginaw, the United Way coordinates
activities among partners, which are similarly limited to local nonprofits and school districts with
minimal involvement from business or local government.
Two more communities also fall in Quadrant 2: Campaign for Grade Level Reading
Marshalltown Iowa, United Way of Miami-Dade Homestead City Florida. We refer to coalitions
within this quadrant as “low-overhead” because the convening agency typically has limited
human and financial resources. This may be a good model for those collaborative initiatives with
limited finances and a limited number of partners to coordinate. With that said, however,
coalitions are typically small and rely on participation from the nonprofit sector as opposed to a
diverse representation of sectors. Although many early-stage education initiatives we spoke to
operated as Low-Overhead Coalition, several of them expressed that there limited in what they
could accomplish through this format and ultimately aspired to be more Holistic Coalition.
Community-Led Coalition
Community-led coalitions in the third quadrant are characterized by decentralized
network governance and low cross-sector engagement. The networks are typically dominated by
community-based organizations from a single sector (nonprofits or civil society organizations),
have informal or irregular communication among partners with no particular organization in the
lead, may pursue multiple agendas or have fluid and evolving network goals, and may even have
a continually evolving roster of partner organizations. From our sample of 28, the networks that
fall within this quadrant are United Way of York County in Biddeford, Maine; My Brother’s
Keeper in Mt. Vernon, New York, Campaign for Grade Level Reading in Delray Beach, Florida;
and the Howard Local Management Board in Howard County, Maryland.
20"
Situated in rural Maine, United Way of York County in Biddeford is a larger coalition of
58 loosely-connected community-led and nonprofit organizations (for e.g. YMCAs and Boy
Scouts). The partners in the coalition are voluntarily associated and do not require formal
membership. Lead by a group of six staff members at the local United Way, the coalition faced
severe funding and capacity issues during their initiation in 2007 with an early childhood
coalition, and later in 2014 with a youth-focused coalition. These funding challenges restricted
their work to an event-based approach with communication limited to sporadic meetings like
annual breakfast events. A lack of buy-in from the local government has added to its list of
capacity challenges. Due to these external challenges, the coalition relies heavily on alliances
with local nonprofits to assist with events.
In certain cases, community-led coalitions are born from the coordinated efforts of a few
individual community champions. My Brother’s Keeper in Mt. Vernon, New York, is one such
example. Led by two community leaders, the coalition was created in 2016 by an advocate who
reached out the school district, mayor’s office, and other community individuals to rally support.
Funding from school district grants, and an additional grant from New York City, supported the
resulting coalition of 18 organizations. Despite the presence of a lead agency in the coalition,
fiscal powers are controlled by the school district leading to decentralization of authority in the
network. The coalition has multiple agendas around education ranging from reading to mental
health, and coordinating among various partners on these agendas has been a challenge
Additionally, similar to United Way of York County, My Brother’s Keeper lacks support from
local government agencies.
However, we identified two networks in the Quadrant 3 that are led by government
agencies: the Campaign for Grade Level Reading (CGLR) in Delray Beach, Florida, and the
21"
Howard Local Management Board in Howard County, Maryland. In contrast with previously
discussed networks in this quadrant, these two initiatives follow a structured collective impact
model which intends to promote centralized governance and cross-sector collaborations. In
CGLR Delray Beach, the main players are the city and the school district, but their main agendas
are less defined as to include multiple broader goals to achieve. The City of Delray Beach
functions as a backbone organization yet leadership and succession in the network are a
challenge which impacts goal alignment among partners. That said, the network works hard to
maintain accountability to its community by releasing regular reports on grade-level proficiency.
Its efforts to improve education in the community have earned it an All-America City Award.
Howard County, lead by the Howard Local Management Board, has structured itself as a
centralized governance network in writing; but in practice the lead agency is struggling with
gaining buy-in from all its partners. Founded in 2017, the initiative has 29 members, with most
of them from the government sector like juvenile services, police, social services, and family and
children services. The network’s goals comprise ensuring physical and mental well-being for
children and young adults, and equitable opportunities to succeed for all youth in the county. The
lead agency works mainly as a coordinator and communicator, yet communication appears to be
scattered and irregular in the network. That said, they engage closely with the community and
focus on racial equity in their outreach efforts.
The communities that fall in the community-led coalition quadrant have struggles with
partner alignment and goal direction of the network. Funding and capacity issues compound
these challenges which is why the sustenance of the network falls upon a few key actors in the
community. The term “community-led” is meant to highlight the efforts of these individual and
22"
organizational actors in ensuring that education initiatives and motivations are supported in the
community even in the face of capacity and partnership alignment struggles.
Multi-Stakeholder Coalition
The fourth quadrant is Multi-Stakeholder Coalition as it captures the effort of aligning
diverse partners to work toward multiple agenda without a backbone organization. Partners in
these initiatives self-organize themselves into program design and implementation. Our coding
identified six communities that are located in Quadrant 4: Anne Arundel Local Management
Board in Maryland, Blue Ribbon Commission on the Prevention of Youth Violence in North
Carolina, Building Our Future Kenosha, Success by 6/Smart Start (United Way of Southwest
Oklahoma Lawton City), Sparks! La Crosse in Wisconsin, Campaign for Grade Level Reading
Grinnell Iowa. Among these communities, the Anne Arundel network, Building Our Future
Kenosha and Sparks! La Crosse are Collective Impact. However they are located in Quadrant 4
as the backbone organization does not organize regular meetings for partners and there is no
structured collaboration among partners. It is often due to the fact that the initiative is still at its
early stage. For example, Sparks! La Crosse and Building our Future are both less than two years
ago as of when the data were collected. The Anne Arundel network is managed by a local
government agency which focuses on multiple social issues and thus lacks a clear agenda in what
needs to achieve at the education level. Partners of the initiative do not engage in continuous
conversation.
Blue Ribbon Commission aims to tackle multi-generational poverty in a local community
by increasing self-sufficiency, social cohesion, collective efficacy and economic stability. With
the coordination of the backbone organization, the initiative did significant amount of
community engagement to reduce youth violence by working with diverse local paters. The
23"
backbone organization views itself as a connector and emphasises the role its partners play in
getting the work done.
The initiative in Grinnell is led by a local community college and a community
foundation. The shared leadership allows the partners to collaborate in their own way. It serves a
small community of just over 9000 population, and the partnerships are relatively diverse to
attract multiple sectors. The initiative in Lawton city is led by the local United Way, yet the
partnerships are loosely organized. There is no attendance requirement for involvement, so
attendance and engagement from partners varies. The initiative focuses on multiple education
programs which rely on informal connections across partners.
Multi-stakeholder coalition is unique in that partners tend to self-organize their effort.
Even with sufficient resources to sustain a backbone organization, the lead agency often plays a
connector or convener’s role to leave the partners enough autonomy in how to engage in
collaboration. This leads to the existence of multiple agendas partners focus on, which could
create challenges to how efficiently utilize and coordinate resources. The advantage of this
model is that partners engage in direct community engagement.
Discussion and Conclusion
In this study, we aimed to move beyond communities’ trend towards self-identified
collective impact models and to further explore nuances in interorganizational partnerships. With
the three research questions proposed, we focus on understanding who the partners are, how they
engage in program planning and implementation, and how the degree of cross-sector engagement
and centralized governance affect the dynamics of collaborative initiatives. With qualitative data
collected from 28 communities across the U.S., we conducted content coding and proposed a
community system solution framework to benchmark where each community is located and why.
24"
With the coding results, we found that collective impact as defined by Kania and Kramer
(2011) and discussed in recent research is not necessarily the intended final destination for all
communities. In some cases, we see communities label themselves as collective impact to meet
funder expectations and to connect to a broader community of advocates of the model across the
country. However, they tweak the model for their circumstances due to capacity concerns or
different roles the backbone organization plays.
The community system solution framework in this study captures the variations of
collaboration models and showcases that collective impact is only one of the pathways to
generate system change at the community level. We acknowledge that communities vary by the
community size, tenure, population served, problems at hand, goals to achieve, existing social
capital in the community, and resources available for mobilization. By no means these
communities discussed here represent all education reform initiatives in the U.S. But they
provide a set of cases to capture diverse needs and solutions at the community level.
The framework could also serve as a guide for communities to configure what is the best
way to align partners and implement programs given their specific community context. In the
result section, we highlighted the advantages of some models over others and reasons why
certain communities might adopt a model other than collective impact. In addition, the
framework provides an alternative language for the communities to describe the varieties of
ways that they could use to organize their partners for social impact, other than the collective
impact model.
We suggest that some communities may see changes over the time - for some
communities that aspire to collective impact, they may indeed begin in one quadrant and
ultimately move to others. However, we suggest that such a progression may not be the goal for
25"
all communities, especially those with limited cross-sector engagement, who do not attract
significant funding, or those communities that have found success with less centralized, more
grassroots approach to organizing. Regardless, we hope that communities will see collective
impact as one approach that is available to them and that an awareness of other approaches might
lead them to the best path forward for their community. Similarly, we hope that funders may also
be open to other approaches of community system solutions as they seek to fund initiatives with
the best chance of success in their respective communities.
" "
26"
References
Austin, J. E. (2000). Strategic collaboration between nonprofits and businesses. Nonprofit and
Voluntary Sector Quarterly, 29(1), 69–97.
Austin, J. E., & Seitanidi, M. M. (2012). Collaborative value creation: A review of partnering
between nonprofits and businesses: Part I. Value creation spectrum and collaboration
stages. Nonprofit and Voluntary Sector Quarterly, 41(5), 726–758.
https://doi.org/10.1177/0899764012450777
Boyer, A.-M., Cooper, K. R., Dougherty, S. M., Wang, R., & Shumate, M. (2018). Predicting
community adoption of collective impact in the United States: A national scan. Presented
at the Cross-Sector Social Interactions, Copenhagen, Denmark.
Bryson, J. M., Crosby, B. C., & Stone, M. M. (2006). The design and implementation of cross-
sector collaborations: Propositions from the literature. Public Administration Review, 66,
44–55. https://doi.org/10.1111/j.1540-6210.2006.00665.x
Christens, B. D., & Inzeo, P. T. (2015). Widening the view: Situating collective impact among
frameworks for community-led change. Community Development, 46(4), 420–435.
https://doi.org/10.1080/15575330.2015.1061680
Cooper, K. R. (2017). Nonprofit participation in collective impact: A comparative case.
Community Development, 48(4), 499–514.
https://doi.org/10.1080/15575330.2017.1332654
Edmondson, J., & Hecht, B. (2014). Defining quality collective impact. Stanford Social
Innovation Review, (Fall 2014). Retrieved from
https://ssir.org/articles/entry/defining_quality_collective_impact
27"
Gazley, B., & Brudney, J. L. (2007). The purpose (and perils) of government-nonprofit
partnership. Nonprofit and Voluntary Sector Quarterly, 36(3), 389–415.
https://doi.org/10.1177/0899764006295997
Gazley, B., & Guo, C. (2015). What do we know about nonprofit collaboration? A
comprehensive systematic review of the literature. Academy of Management
Proceedings, 2015(1), 15409. https://doi.org/10.5465/AMBPP.2015.303
Gillam, R. J., Counts, J. M., & Garstka, T. A. (2016). Collective impact facilitators: How
contextual and procedural factors influence collaboration. Community Development,
47(2), 209–224. https://doi.org/10.1080/15575330.2015.1133684
Guo, C., & Acar, M. (2005). Understanding collaboration among nonprofit organizations:
Combining resource dependency, institutional, and network perspectives. Nonprofit and
Voluntary Sector Quarterly, 34(3), 340–361. https://doi.org/10.1177/0899764005275411
Henig, J. R., Riehl, C. J., Houston, D., Rebell, M., & Wolff, J. (2016). Collective impact and the
new generation of cross-sector collaborations for education. Teachers College, Columbia
University. Retrieved from https://www.tc.columbia.edu/education-policy-and-social-
analysis/department-news/cross-sector-collaboration/CI-corrected-digital-version-3-11-
16.pdf
Kagan, S. L. (1991). United we stand: Collaboration for child care and early education services.
New York: Teachers College Press.
Kania, J., Hanleybrown, F., & Splansky Juster, J. (2014). Essential mindset shifts for collective
impact. Stanford Social Innovation Review.(Collective Insights on Collective Impact).
Retrieved from
http://www.ssireview.org/articles/entry/essential_mindset_shifts_for_collective_impact
28"
Kania, J., & Kramer, M. (2011). Collective impact. Stanford Social Innovation Review, 1, 36–41.
Keast, R., Brown, K., & Mandell, M. (2007). Getting the right mix: Unpacking integration
meanings and strategies. International Public Management Journal, 10(1), 9–33.
https://doi.org/10.1080/10967490601185716
Lawlor, J. A., & Neal, Z. P. (2016). Networked community change: Understanding community
systems change through the lens of social network analysis. American Journal of
Community Psychology, 57(3–4), 426–436. https://doi.org/10.1002/ajcp.12052
Page, S., & Stone, M. M. (2017). Power, conflict, and collaborative decisions in partnerships to
improve public education. Presented at the ARNOVA, Grand Rapids, MI.
Powell, W. (1990). Neither market nor hierarchy: Network forms of organization. Research in
Organizational Behaviour, 12, 295–336.
Powell, W. W., White, D. R., Koput, K. W., & Owen‐Smith, J. (2005). Network dynamics and
field evolution: The growth of interorganizational collaboration in the life sciences.
American Journal of Sociology, 110(4), 1132–1205. https://doi.org/10.1086/421508
Provan, K. G., & Kenis, P. (2008). Modes of network governance: Structure, management, and
effectiveness. Journal of Public Administration Research and Theory, 18(2), 229–252.
https://doi.org/10.1093/jopart/mum015
Provan, Keith G., Fish, A., & Sydow, J. (2007). Interorganizational networks at the network
level: A review of the empirical literature on whole networks. Journal of Management,
33(3), 479–516. https://doi.org/10.1177/0149206307302554
Provan, Keith G., & Huang, K. (2012). Resource tangibility and the evolution of a publicly
funded health and human services network. Public Administration Review, 72(3), 366–
375.
29"
Provan, Keith G., & Lemaire, R. H. (2012). Core concepts and key ideas for understanding
public sector organizational networks: Using research to inform scholarship and practice.
Public Administration Review, 72(5), 638–648.
Rondinelli, D. A., & London, T. (2003). How corporations and environmental groups cooperate:
Assessing cross-sector alliances and collaborations. The Academy of Management
Executive (1993-2005), 17(1), 61–76.
Saldaña, J. (2013). The coding manual for qualitative researchers (Second edition). Thousand
Oaks, CA: SAGE Publications Ltd.
Selsky, J. W., & Parker, B. (2005). Cross-sector partnerships to address social issues: Challenges
to theory and practice. Journal of Management, 31(6), 849–873.
https://doi.org/10.1177/0149206305279601
Snavely, K., & Tracy, M. B. (2000). Collaboration among rural nonprofit organizations.
Nonprofit Management and Leadership, 11(2), 145–165.
https://doi.org/10.1002/nml.11202
Walzer, N., Weaver, L., & McGuire, C. (2016). Collective impact approaches and community
development issues. Community Development, 47(2), 156–166.
https://doi.org/10.1080/15575330.2015.1133686
Wolff, T. (2016). Ten places where collective impact gets it wrong. Global Journal of
Community Psychology Practice, 7(1). Retrieved from
https://www.gjcpp.org/en/resource.php?issue=21&resource=200
30"
Table 1. A list of sampled communities
State
Network Name
ID
Scope of work
Collective impact
Or Not
CT
Coalition for New Britain's Youth
CI1
City level
Collective impact
NC
Family Success Alliance (Orange County)
CI2
County level
Collective impact
ME
Westbrook Children's Project (Portland)
CI3
City level
Collective impact
NC
Youth Thrive
CI4
Couty level
Collective impact
CT
Hartford Partnership for Student Success
CT
EN1
City level
No
NC
United Way of Davidson County
EN2
Couty level
No
ME
United Way of York County
EN3
City level
No
NC
Blue Ribbon Commission
EN4
City level
No
MA
Communities That Care Coalition
(Franklin county)
CI6
Couty level
Collective impact
MA
Berkshire United Way
EN6
Couty level
Collective impact
NY
ROC the future (Rochester)
CI8
City level
Collective impact
NY
My Brother's keeper alliance Mt Vernon
EN8
City level
No
OH
Summit Education Initiative (Akron)
CI5
Couty level
Collective impact
OH
Learn to Earn Dayton (Dayton)
EN5
Couty level
Collective impact
FL
Delray Beach, Campaign for Grade Level
Reading
CI9
City level
Collective impact
FL
United Way of Miami-Dade (Homestead
City)
EN9
City level
Collective impact
WI
Higher expectation Racine
CI7
County level
Collective impact
WI
Building our future Kenosha
EN7
County level
Collective impact
OK
Impact Tulsa
CI10
City level
Collective impact
OK
Lawton City
EN10
City level
No
MI
Flint & Genesee Literacy Network
CI11
County level
Collective impact
MI
Saginaw
EN11
County level
Collective impact
MD
Howard
EN12
County level
Collective impact
MD
Anne Arundel County
CI12
County level
Collective impact
WI
Sparks! La Crosse
EN13
City level
Collective impact
WI
Achieve Brown County
CI13
City level
Collective impact
IA
CGLR Marshalltown
CI14
City level
Collective impact
IA
CGLR Grinnell
EN14
City level
Collective impact
31"
Figure 1: Community System Solutions
32"
Figure 2. Mapping of the sampled communities
