How Does NGO Partnering Change Over Time? A Longitudinal Examination of Factors That Influence NGO Partner Selection

Abstract

Scholars suggest three partnering strategies that nongovernmental organizations (NGOs) can use to pursue strategic relationships in civil society networks: (a) the development of overlapping ties associated with network closure, (b) adopting an intermediary role between two disconnected organizations associated with brokerage, and (c) complying with the match-making demands of third-party organizations. Collaborative relationships among 489 NGOs were examined over a 14-year period (1990-2004) to determine which of these strategies NGOs use and when. The network demonstrated a strong preference for closure at the beginning of the observation period, after which time partnerships settled into a more stable pattern of intra-sector collaboration after 1996. Brokerage and constrained-choice strategies were not prevalent at any point over the observation period. Results are discussed in terms of network evolution and implications of the observed NGO preferences for closure. The potential benefits of emergent stability are also discussed.

Keywords

NGOs partnering networks interorganizational collaboration

Collaboration among nongovernmental organizations (NGOs) is both normatively encouraged and extensively studied (Gazley & Guo, 2015; Huxham & Vangen, 2013). An under-examined issue within this literature is what strategies NGOs actually employ to select partners (Atouba & Shumate, 2015). Partner-selection research often uses case study or cross-sectional methods that infer, but do not test, how partner selection might have occurred over time (Gazley & Guo, 2015). The purpose of this study is to longitudinally examine partnering choices and to consider the characteristics of potential partners and features of the partnering network in which NGOs are already embedded. We suggest that partner selection is in part determined by the current set of partnerships among NGOs in a field and that partner selection transforms the network, influencing that network in subsequent time periods.

We examine the network of partnerships among NGOs and between NGOs and Intergovernmental Organizations (IGOs) in three major issue sectors over a 14-year period. We investigate three strategies that can explain partnership selection in the network. The key point is that existing partnerships create constraints and opportunities that influence future partner selection. The three network strategies reflect different—if not competing—orientations toward collective action and network building in NGO networks. Each strategy reflects different competitive, collaborative, and institutional interests that all NGOs have to manage.

Literature Review

The mechanisms driving NGO partner selection have recently generated a great deal of interest among organizational scholars (Atouba & Shumate, 2015; Austin & Seitanidi, 2012; Doerfel & Taylor, 2004, 2017; Guo & Acar, 2005; Seitanidi, 2010; Stephens, Fulk, & Monge, 2009). Like for-profit organizations, NGOs strategically form partnerships because they can leverage these relationships to achieve certain goals (Hager, Galaskiewicz, & Larson, 2004; Meyer & Hyde, 2004; Passey & Lyons, 2006; Schneider, 2009). Partnerships increase NGOs’ ability to fulfill their objectives through greater solidarity, information, resource flows, and influence, among other benefits (King, 2004; Rabrenovic & Pierce, 2003; Seitanidi, 2010).

In the interorganizational network literature, two partnering strategies are especially common. First, organizations can build dense communities by developing and maintaining strong, overlapping connections with other organizations. This is referred to as closure. Second, organizations may establish partnerships which position them between disconnected organizations allowing them to act as intermediaries. This is described as brokerage. In a third strategy, organizations may form collaborative ties to meet the demands of powerful third parties. This has been described as a constrained-choice strategy (Stephens et al., 2009). Although this third strategy is mentioned in the literature, it has not been empirically tested. We review existing findings on these strategies as background to the empirical examination of the importance of each strategy to NGOs over time. Hypotheses focus on the prevalence of each strategy and serve as the basis for our larger research question: How do NGO partnering strategies change over time?

Closure

Closure refers to the tendency of organizations sharing ties with the same partner to disproportionately form ties with each other (Monge & Contractor, 2003). Close-knit, densely connected communities characterized by obligation, reciprocity, and trust are observed when many organizations in a network are connected directly and when those connections are reinforced by overlapping connections with third parties.

Baker and Obstfeld (1999) and Obstfeld (2017) describe network closure as a tertius iungens or union orientation. The tertius iungens is “the third who joins” and has “a strategic, behavioral orientation toward connecting people in one’s social network by either introducing disconnected individuals or facilitating new coordination between connected individuals” (Baker & Obstfeld, 1999, p. 102). For-profit organizations often pursue closure as a way to reduce the likelihood of opportunistic behavior by interorganizational partners (Gulati, 1995; Gulati & Gargiulo, 1999).

Civil society organizations also pursue closure. In the United Kingdom, Baldassari and Diani (2007) found what they described as “horizontal solidarity” among civic organizations which developed closure in their community. In his analysis of Mexican NGOs, Fox (1996) emphasized the importance of closure, or what he describes as “societal thickness.” Keyes, Schwartz, Vidal, and Bratt (1996) examined closure in the nonprofit housing sector, arguing that the reciprocity, trust, shared vision, and mutual interest that were developed as organizations partnered with known others repeatedly over time contributed to organizational success (see also Gittell & Vidal, 1998). Atouba and Shumate (2010) found significant evidence of closure among development NGOs. They argued that closure is beneficial for NGOs, since overlapping and reinforcing connections facilitate activity coordination. Drawing upon prior research, we hypothesize that:

Hypothesis 1 (H1): NGOs will form multiple, overlapping relationships with other NGOs, leading to network closure.

Brokerage

Another view, closely associated with the work of Burt (1992, 2001) identifies brokerage as a partnering strategy that keeps other organizations apart, identifying and occupying “those places where people are unconnected in a network” (Monge & Contractor, 2003, p. 143). An organization behaves as a broker by serving as a conduit between two organizations which were not previously or otherwise connected. This creates indirect links between the newly connected other organization and all the other connections of the broker, an outcome that typically favors the broker.

A brokerage position in the network gives an organization access to unique sources of timely information (King, 2004), as well as knowledge about which organizations in the network might find that information valuable (known as the vision advantage). Brokers also can exploit the lack of direct connections between partners, for example, by playing parties off against one another (known as the control advantage). Unlike the tertius iungens role described earlier, organizations that pursue a brokerage strategy act as a tertius gaudens, “the third who benefits” by keeping others apart. Uzzi (1999) found that firms receive lower interest rates when they have a mix of both close ties and brokerage ties with banks. He argued that brokerage allows these firms access to unique information they would not have been able to get on the open market or from close ties.

Research also confirms the use of brokerage by NGOs. Taylor and Doerfel (2003) found that the initial development of the civil society sector in Croatia was dominated by closure. However, 2 years later (Doerfel & Taylor, 2004), they found that the civic organizations had changed their dominant strategy from closure to brokerage. Similarly, Klerx and Leeuwis (2009) show that NGOs benefited by identifying relationship gaps in the Dutch agricultural sector and filling them by acting as “innovation brokers.” Drawing upon prior research, we propose that:

Hypothesis 2 (H2): NGOs will pursue partners through a brokerage strategy by creating direct links to organizations which were themselves previously unconnected.

Constrained Choice

Research on interorganizational relationships emphasizes voluntary choices organizations make regarding whether, with whom, and under what conditions to partner (Gazley & Guo, 2015). Nevertheless, some NGO relationships are instigated by intermediaries, such as foundations or government agencies (Brinkerhoff, 2002; Guo & Acar, 2005; Longoria, 2005; Wood & Gray, 1991). These intermediary organizations often require NGOs to participate in partnerships to yield “the benefits of increased efficiency and innovation, local adaptation, increased flexibility and enhanced community ties” (Graddy & Chen, 2006, p. 2). Take, for instance, the National Response Framework created by Federal Emergency Management Agency (FEMA), in which various nonprofits, including the American Red Cross and members of National Voluntary Organizations Active in Disaster (VOAD), complied with government mandates to collaborate in response to Hurricane Katrina (FEMA, 2008, pp. 7-8).

Stephens et al. (2009) define this type of mandated interorganizational relationship as a constrained-choice alliance, in which “external, ‘cupid’ organizations broker alliances between other target organizations . . . although the cupids do not themselves directly participate in the brokered alliances” (p. 509). These relationships differ from voluntary relationships in two ways. First, partners are not dependent on one another, but are dependent on the powerful third-party that stands to benefit from the partnership. Second, whereas in voluntary collaborations, partners are usually known to one another either through experience or reputation, powerful organizations may arrange partnerships between organizations that have no prior experience with each other, and that do not share other third-party connections that could serve as conduits for reputational information. The constrained-choice perspective suggests that NGOs may be encouraged or obliged to comply with the match-making demands of powerful organizations to collaborate with specific others (Atouba & Shumate, 2015; Stephens et al., 2009).

The constrained-choice collaboration is not motivated by mutual benefit among organizations. Rather, the mandating organization’s role is that of a self-interested party, whose motivation in joining other organizations is to benefit itself. The mandating organization is a broker, using its power and exploiting the lack of connection between target organizations for its own benefit. However, because the mandating organization also creates a direct connection between the targets, it pursues a strategy of union rather than disunion (Baker & Obstfeld, 1999). In other words, the “third who joins,” in this case, is also the “third who benefits.” However, the benefit is from bringing organizations together rather than keeping them apart, suggesting a blended strategy of partnership formation we call tertius disponens, “the third who arranges.”

Among NGOs, constrained-choice arrangements are found when relationships between NGO targets are orchestrated by funders, grantors or IGOs. IGOs and NGOs have a unique relationship, due in part to their mandates. Most IGO charters recognize the sovereignty of states, so IGO authority does not ordinarily supersede national law where these conflict (Stohl & Stohl, 2005; White, 2005). As independent organizations, NGOs are freer than IGOs to act in pursuit of social goals in ways that challenge state sovereignty, or which seek to change the rule of law (Stohl & Stohl, 2005; White, 2005). IGO-NGO relations are perhaps best understood as a policy dance, with IGOs pulling NGOs in as they seek expertise and implementation of IGO projects, and NGOs pushing IGOs as they seek to further their own agendas (Steffek, 2013). In particular, NGOs are dependent upon IGOs to enact policy as well as for funding. Although both NGOs and IGOs are engaged in the policy dance, IGOs take the lead; therefore, many IGO-NGO relationships are defined by the kind of power imbalance that characterizes constrained-choice collaborations. Previous research shows that when NGOs share a relationship with the same IGO (Shumate, Fulk, & Monge, 2005), or are funded by the same IGO (Atouba & Shumate, 2010), they are more likely to have a direct relationship with one another.

If a constrained-choice strategy is operating in this network, we would expect to see more relationships between NGOs that have each had the same IGOs as partners in the past. Exploring the hypothesis that participation in mandated alliances is common in NGO networks, we hypothesize that:

Hypothesis 3 (H3): NGOs will demonstrate constrained choice in their partnering by forming relationships with NGOs who share a preexisting connection to a common IGO.

NGOs operate in dynamic environments (e.g., resource, institutional, and sociopolitical), and these environmental changes are likely to affect their partnering strategies (Doerfel & Taylor, 2004; Margolin et al., 2015; Tushman & Romanelli, 1985). In addition to identifying which partnering strategies NGOs pursue, given the network that exists at any point in time, we also examine how NGOs’ overall relative reliance on these strategies change over 14 years. Thus, we propose the following research question:

Research Question (RQ): How do NGO partnering strategies change over time?

Methods

Sample

Data for this study were drawn from the Yearbook of International Organizations (YIO). The YIO is an NGO clearinghouse for information about international NGOs and government agencies (Union of International Associations [UIA], 2017). Founded in 1907, its mission is to serve as a repository for authoritative information about international organizations. The YIO includes entries on more than 37,000 active international NGOs. Entries include information about organizational mission, founding, funding, and partners.

The YIO surveys organizations via mail, email, and fax and sends out proofs of organizations’ updated entries yearly to ensure that the information is current and correct. The average response rate is 35% (UIA, 2017). The YIO uses a variety of sources to confirm the proofs when the organization does not respond, including reviewing websites, annual reports, and newsletters, and following up with phone calls to the organization. Researchers agree that it is the most comprehensive source of information on international NGOs, public or private (Atouba & Shumate, 2015; Keck & Sikkink, 1998). Keck and Sikkink (1998) found that most organizations were included in the YIO within a few years of their founding. Researchers have successfully used the YIO to map networks of international nongovernmental organizations (INGOs; see Atouba & Shumate, 2015; Keck & Sikkink, 1998; Smith, 1997).

Using the online version of the YIO, a list of international organizations with names, aims, or activities focusing on children’s rights, infectious diseases, or sustainable development was created. We chose these populations because of our familiarity with them and because the inclusion of several populations, as opposed to only one, enhanced the generalizability of the research. After coding 14 volumes of the YIO published between 1990 and 2004, NGOs which had disbanded or for which information could not be confirmed were removed. This yielded 489 unique NGOs.

Measures

Collaboration network

This study takes a social network approach to the study of interorganizational networks, in contrast to the study of coalitions that describe themselves as a network. In social network research, all interorganizational ties within a set of organizations are examined as a network. The relationship examined in this study is collaboration. Collaboration or partnership means that two organizations work together and share resources or services (see Guo & Acar, 2005). NGOs report to the YIO a list of partner organizations. The YIO defines a partner as any organization the NGO “collaborates with,” “links with,” “partners,” “hosts,” or is “hosted by” (Atouba & Shumate, 2015). For example, if the American Thoracic Society (ATS) indicates that they collaborate with the Global Health Council (GHC), a relationship is recorded (i.e., a tie from ATS → GHC). All ties were recorded, and a matrix of relations was created from the various interdependent dyads (Robins, 2015, p. 128). Partnerships for each organization were coded for each year (1990-2004). These connections constitute the 14 NGO collaboration networks under analysis, one for each year of observation.

NGO communities

The network consists of IGOs and three communities of NGOs recorded in the YIO between 1990 and 2004. IGOs were included in the sample if they were named by one or more NGOs as a collaborator. NGOs were included in the sample based on their activities and aims.

The infectious diseases NGO community included organizations whose aims or name described infectious diseases in general or more specific infectious diseases including tropical diseases, tuberculosis, leprosy, malaria, HIV/AIDS, and STDs (n = 137; 28%). NGOs whose aims or name mentioned third world or sustainable development in third-world countries in general or specific development issues including sustainable food systems, agricultural development, economic structures, debt relief, microfinance, water systems, unemployment, developing economic industries, and education were included in the sustainable development NGO community (n = 245; 50%). The children’s rights NGO community consisted of NGOs whose aims or name mentioned children’s rights in general or more specific rights issues including child slavery, exploitation, welfare, and children’s education and health (n = 107; 22%).

Analysis

The prevalence of network structures associated with each of the partnering strategies was estimated using Separable Temporal Exponential Random Graph Modeling (STERGM; Krivitsky & Handcock, 2014). STERGM is an extension of the Exponential Random Graph Model (ERGM; see Lusher, Koskinen, and Robins (2012) for an overview and Atouba and Shumate (2015) for an application of ERGM in the NGO context. These models identify differences between an observed network and randomly generated networks with similar features. The model determines this difference through a blend of mathematical and simulation-based inferences. That is, ERGM simulates thousands of random networks with the same number of organizations and partnership ties as the observed network and then compares the observed network to the average of these thousands of simulated networks. The result is the generation of Maximum Likelihood Estimates and a final ERGM formula, similar to a multiple regression formula. This process reveals whether the specific parameters representing the hypothesized pattern of relationships in the observed network is more or less likely to occur compared to thousands of randomly simulated networks. In other words, the model tests the research hypothesis against the null hypothesis which states that the pattern of relationships in the observed data is only as prevalent as would be expected to occur by chance alone.

STERGM replicates this ERGM process, except that instead of looking at the ties present during one time period, it looks at ties formed between two time periods in the same network. Rather than interpreting whether a tie is likely to be present, as in cross-sectional ERGM, STERGM interpretation is temporal, suggesting whether or not a tie is likely to form between one time period and the next. STERGM allows us to statistically determine whether various factors like closure, brokerage, or constrained-choice connections influence the probability of NGOs forming partnership ties over time. STERGM tests the research hypothesis against the null hypothesis that particular types of ties form only as often as would be expected by chance alone.

Goodness-of-fit (GOF) statistics determine how well the simulated networks created by the final STERGM formula match the observed networks in the data set. A good-fitting model should produce networks that look very similar to the empirical data. There is no hard and fast rule for GOF assessment, but the more parameters are explained by the STERGM equation, the better the model. Parameters are considered explained when the value of a parameter for the simulated networks based on the model is not statistically different than the value for the observed network.

Model construction: A 2-year interval approach

STERGM models were run for seven 2-year intervals (1990-1992, 1992-1994, 1994-1996, 1996-1998, 1998-1900, 2000-2002, and 2002-2004), for two reasons. First, STERGM assumes that all organizations are observed in each wave of analysis. However, as in many longitudinal analyses (see Wasserman & Faust, 1994), this NGO network demonstrated high turnover over the 14 years. An omnibus model from 1990 to 2004 would include, in the sample, only those NGOs that were active across all 14 years (n = 95), biasing the results toward NGOs active across the entire period. Second, examining the 2-year intervals allows us to determine whether specific structural patterns were present in particular years, and whether those patterns persisted over time. Models were constructed using a forward-selection approach recommended by Green and Wasserman (2013). Entered one at a time, parameters were retained when models converged and removed when they caused degeneracy.

Parameter Definition

Closure

To estimate closure, the geometrically weighted edgewise shared partner distribution (GWESP) parameter was estimated (Hunter, 2007). The idea behind GWESP is to capture the tendency of organizations to cluster together and form close ties. For example, if at Time 1, hypothetical NGOs A and B did not collaborate with each other but both collaborated with NGO C, closure would be indicated at Time 2 if A formed a collaborative tie with B while both organizations also retained ties with C. The GWESP term would indicate whether such a closure tendency was more or less likely to occur than by chance.

Brokerage

To measure brokerage, the geometrically weighted dyad-wise shared partner distribution (GWDSP) parameter was estimated. GWDSP captures the tendency of organizations to bridge unconnected organizations. For example, if at Time 1, NGOs A and B collaborated with each other, and in Time 2, A formed a new tie with C, A would be able to act as a broker between B and C. The GWDSP would indicate whether such a tendency is more or less likely to occur than by chance alone.

Constrained choice

To determine the prevalence of constrained-choice partnerships among NGOs, a dyadic covariate parameter was used as a proxy. This parameter measures the likelihood of a collaborative tie forming between two NGOs during Time 2, if they each had an independent relationship with the same IGO during Time 1. The presence of an existing independent relationship with the same IGO is a necessary but not sufficient, condition for constrained choice. We infer that the partnership is mandated or suggested by the IGO. We explore this further in the “Discussion” section.

Figure 1 depicts conceptual visualizations of each of the parameters. Like regression models, STERGM estimates the relative influence of each parameter while holding the others constant, allowing independent interpretation of each partnering mechanism in the context of the others. This also permits that the three strategies may operate in concert.

Figure 1.

Three collaborative formation strategies.

Control parameters

To control for alternative explanations of tie formation, the model included three common network tendencies, homophily, activity, and preferential attachment, which tend to exert a strong influence on network formation generally (Atouba & Shumate, 2015; Robins, Pattison, Kalish, & Lusher, 2007) but are not of primary interest in the analysis presented here. Homophily (McPherson, Smith-Lovin, & Cook, 2001) suggests that NGOs are more likely to form ties if they share similar important attributes or characteristics (Atouba & Shumate, 2015). NGOs were more likely to partner with other NGOs if they belonged to the same issue domain. Activity accounts for whether some NGOs were more likely to form ties in general. Finally, preferential attachment (Merton, 1968), the geometrically weighted degree (GWD) parameter, was included (Hunter, 2007). GWD measures the tendency for popular organizations with more existing connections to gain additional relationships over time (i.e., the “rich get richer” phenomenon). These three parameters are routinely controlled in ERGM-type analyses such as the STERGMs undertaken here, (Robins et al., 2007). One additional parameter, the arc parameter (Lusher et al., 2012), controls for the overall tendency of NGOs to form ties.

Results

Table 1 describes the number of NGOs and partnerships present in each wave of analysis. The intense activity of forming and dissolving collaborative ties in the first four waves between 1990 and 1998 was striking, with the addition of 49 NGOs. In the final three waves from 1998 to 2004, the total number of NGOs increased by only three. There was a lower number of collaborative relationships overall but an increase in repeated ties.

Table 1.

Sample Sizes Across Waves and Populations.

	1990-1992	1992-1994	1994-1996	1996-1998	1998-2000	2000-2002	2002-2004
Infectious disease NGOs	18	24	30	54	53	45	60
Sustainable development NGOs	76	73	75	85	96	77	93
Children’s rights NGOs	83	87	88	87	84	80	83
Total NGOs	177	184	193	226	233	202	236
Unique number of IGOs	43	51	50	58	61	38	68
Number of ties Time 1^a	121	172	157	262	122	129	115
Number of ties Time 2^a	150	154	250	125	112	127	99
Average degree Time 1^a	0.68	0.93	0.81	1.15	0.52	0.64	0.49
Average degree Time 2^a	0.84	0.83	1.29	0.55	0.48	0.63	0.42

Note. NGO = nongovernmental organizations; IGO = intergovernmental organizations.

The number of ties in Time 2 in one wave will not match the number of ties in Time 1 of the subsequent wave, since organizations had to be present in both years of each time pair to be included in the analysis of that year.

Table 2 provides the results of the STERGM models which include the three hypothesized parameters for closure (GWESP), brokerage (GWDSP), and constrained-choice connections (dyadic network covariate), and the four control parameters for homophily, activity, preferential attachment, and arc (the tendency to form ties).

Table 2.

STERGM Results.

	1990-1992 (N = 177)	1992-1994 (N = 184)	1994-1996 (N = 193)	1996-1998 (N = 226)	1998-2000 (N = 233)	2000-2002 (N = 202)	2002-2004 (N = 236)
Arc	−6.26**	−6.89**	−6.13**	−7.94**	−7.83**	−5.90**	−6.30**
Hypotheses
Closure (H1)	0.92**	0.67*	0.79**	0.63*	0.28	0.15	0.29
Brokerage (H2)	0.02	−0.23*	−0.01	−0.05	−0.09	−0.06	−0.09
Constrained choice: IGO (H3)	−1.58	−0.28	−0.32	−0.80	0.10	–	−0.45
Control parameters
Preferential attachment^a	−1.57*	−1.92**	−1.35**	−1.73**	−0.09	−1.20*	−1.50*
Homophily	0.71	0.55	1.05**	1.23*	1.50*	–	1.41*
Sustainable development (activity)	−0.24	−0.10	0.57	0.44	0.25	–	−0.72
Children’s Rights (activity)	0.40	1.56*	0.44	1.21*	0.93	–	−0.36
AIC	314.7	354.8	946.7	425.7	435.8	246.7	194.8
GOF parameters explained	92%	97%	88%	97%	97%	95%	100%

Note. IGO = intergovernmental organizations; AIC = Akaike information criterion; GOF = goodness of fit.

Negative parameter indicates preferential attachment.

p < .05. **p < .01.

H1 predicted that the network would exhibit a tendency toward closure. The positive and significant GWESP estimates across the first four 2-year time periods from 1990 to 1998 support this hypothesis. However, in the last three 2-year waves from 1998 to 2004, closure was not significant. This shows partial support for H1, indicating that NGOs tended to form partnerships that contributed to network closure throughout most of the 1990s but did not continue this trend into the early 2000s.

The models showed no significant support for H2, which argued that the network would exhibit a structural tendency toward brokerage. In one time period (1992-1994), the GWDSP parameter estimate was significant but negative, suggesting that formation of ties was less likely to contribute to the formation of brokerage patterns than expected.

H3 predicted that NGOs would participate in constrained-choice partnerships. The analyses did not support this hypothesis, as there were no significant positive estimates across any of the seven waves.

Five of the seven GOF diagnostics for the models were at 95% or higher showing good fit for the model. A sixth parameter was at 92% indicating acceptable fit for the model. The seventh parameter was 88%, reflecting weak fit. Collectively, the GOF showed that the estimated parameters closely mirror the distribution of observed parameters in the networks in all waves except one.

Discussion

Research has shown that NGOs strategically build partnerships to achieve civil society goals (Hager et al., 2004; King, 2004; Meyer & Hyde, 2004; Passey & Lyons, 2006; Rabrenovic & Pierce, 2003; Schneider, 2009). Less is known about what specific strategies NGOs actually employ to select partners (e.g., Atouba & Shumate, 2015). The focus of the limited research on strategic partner selection to date has been on rich case studies of a few organizations or cross-sectional studies that infer how partner selection might have occurred over time by examining an NGO network at one point at time. The few studies that examine changes over time are focused on single countries or issue sectors or do not differentiate the special role of IGOs. This study was designed to advance this literature by studying partnering choices over time in multiple issue sectors and incorporating features of the partnering network in which each of the potential NGO and IGO partners is already embedded. A key motivation is the view that the current set of partnerships among NGOs influences partner selection and, in turn, partner selection transforms the collaboration network, influencing the partnering network in subsequent time periods.

Three potential strategies that NGOs use to pursue partnerships have been identified, but previous research has not specified which of these strategies NGOs favor at different points of development of the partnering network over time. The results of this STERGM analysis paint a clear picture: controlling for homophily and preferential attachment, early network development was dominated by a process in which organizations continued to add additional ties until they reached a point of network closure. However, this pattern did not persist in later network development. As in previous research, homophily and preferential attachment also influenced the process of tie formation. However, there was no significant evidence of NGOs using the brokerage strategy or selecting partners on the basis of constrained choice.

Why Closure?

Closure was the primary partnering strategy evident in this sample, supporting H1. The results show that closure was important in early years of the analysis (through 1998). In examining the descriptive statistics, 1998 marks a key turning point in the network. The overall number of collaborations decreased between 1996 and 1998 and never recovered, suggesting that these organizations reached the limits of their ability to add new partner links, a phenomenon known as relational carrying capacity (Monge, Heise, & Margolin, 2008).

Extensive network closure is indicated by the significant homophily parameter which emerges and persists starting in the third time period, as well as the reduction in the number of unique partnerships. This suggests that after 1994, organizations began to prefer collaborations with partners in the same issue domain. In other words, NGOs adopted and persisted in a strategy which focused on deepening relationships with homophilous partners over committing resources to the pursuit of collaborators in different issue areas. Another possibility is that after an initial period of experimenting with partnering strategies in the 1990s, NGOs learned both how to partner and how to select good partners, allowing partnerships to stabilize over time. This interpretation is consistent with the significant increase in NGO partnering during the 1990s (Sikkink & Smith, 2002) as well as with the general increase in interconnectedness among organizations in a globalizing world (Castells, 1999, 2004).

Why Not Brokerage?

The lack of evidence for partnering through brokerage suggests that NGOs do not collectively position themselves in partnerships to accrue individual organizational gains (Burt, 1992). One explanation for this finding is that a strategy of seeking advantage by “keeping others apart” may not be consistent with the general goals of NGOs. Unlike firms, whose primary goal is to generate profits, the missions of the NGOs examined here focus on the provision of public goods such as child welfare, human rights, and the eradication of pandemics. Although NGOs do (and must) engage in competition, previous research (O’Brien & Evans, 2017) suggests that NGO leaders experience discomfort with this aspect of their operations. Failure to take advantage of the potential benefits of brokerage may reflect community norms which discourage self-interested behavior. In addition, in close-knit networks like this one, such opportunistic behavior would be more likely to be observed and sanctioned.

Why Not Constrained Choice?

Although there are documented instances of third-party organizations mandating relationships among NGOs (e.g., Heffren, McDonald, Casebeer, & Wallsten, 2003; Rodríguez, Langley, Béland, & Denis, 2007), these results suggest that mandated ties did significantly drive collaboration. IGOs did not seem to play a major role in influencing NGOs’ partner selection, at least as evidenced by prior shared ties to the same IGO. IGOs depend on NGOs to meet the goals of providing services to target populations and energizing social change (Stohl & Stohl, 2005; White, 2005). In some instances, IGOs might be better served by initiating voluntary cooperative relations where possible, reserving their power to mandate relationships among NGOs for occasions where barriers exist to voluntary cooperation among the targets.

An Evolutionary Explanation

Research on organizational evolution sheds light on the patterns identified here. According to Bryant and Monge (2008), longitudinal changes in the forces that influence network ties are not simply random fluctuations. Instead, they actually might mark a shift from one stage of a community’s development to another.

For example, Doerfel and Taylor’s study (2004) of Croatian civil society organizations demonstrated how network metrics evolved during times of intense governmental transition and social mobilization to more maintenance and democratic institutionalization efforts. In particular, they found that the Croatian civil society network moved from heavy cooperation via closure in more uncertain times to more structural holes during entrepreneurial times of early democracy. Likewise, Bryant and Monge (2008) and Margolin et al. (2015) observed that reciprocity would dominate the early stages of a particular nonprofit community, only to decrease as the community began to enter more stable periods of maintenance. This is important because reciprocity is sometimes also used a proxy for closure (Gargiulo & Benassi, 2000).

Astley (1985) theorized about the evolution of closure, writing that closure is useful in early states of community development because it creates a system where organizations can exchange resources among their particular network communities rather than having to deal explicitly with the broader environment. However, the strategy of closure cannot last forever. In particular, the “growth of internal complexity accompanying system closure fosters a stabilization of communities but also sets them up for eventual collapse . . . Community closure progressively eliminates open space and inhibits the potential for new populations to emerge” (p. 236-237). The results presented here align well with Bryant and Monge’s (2008) stage model of community evolution, which suggests that organizational communities move through periods of emergence, maintenance, self-sufficiency, and transformation. The period characterized by partnership closure suggests community emergence, and the shift into a period of stability reflects a maintenance period in which the NGO’s communities settle into relatively stable patterns of ties, particularly with like-minded NGOs (i.e., homophily). In other words, the reduction in closure may signal that nonprofits in these communities are no longer as reliant on previous connections and have moved on to foster relations with other similar nonprofits.

Practically, these results have two implications for NGO leaders. First, as part of strategic planning, NGO leaders might fruitfully consider what phase their population is in and seek partners accordingly. In the emergence phase, cross-sector ties are perhaps more easily formed, but in the maintenance phase, ties with other similar NGOs are more common. Our results also suggest that closure and homophily dominate NGO partner selection. Closure and homophily both mitigate risk but may also reduce the potential for new ideas and innovation to be introduced through partnering (Phelps, 2010). NGO leaders might do well to consider how a network closure strategy might be balanced with greater diversity in network composition (i.e., greater heterophily), as well as to consider discrete partnering choices in the context of the larger network.

Limitations and Future Research

This study makes a significant advance in our understanding of the development of NGO partnership networks but has limitations which should be addressed in future research. First, our sample only includes INGOs across three issue domains. Although a focus on INGOs helps us understand organizing patterns among an important category of organizations in the international system, it does overlook the role of national organizations. However, since the mid-1990s, an approach that has been increasingly adopted by INGOs to improve their legitimacy and the impact of their action has been to establish more and closer collaborative ties with national organizations, especially in the global South (Madon, 1999). An increased focus on building local ties, which are not captured in our study, might also help explain the drop-off in INGO partnerships after 1996. It is also possible that smaller domestic NGOs which have a greater dependence on IGOs are more subject to mandated relationships than the INGOs examined. Future research should investigate broader samples of national and international organizations working in more issue areas.

Second, we examine patterns of tie formation among NGOs over time and infer the partnering strategies pursued by these organizations. Although such an approach is common in the organizational literature, we cannot entirely rule out alternative explanations for the observed patterns. For instance, an alternative interpretation of our constrained-choice analysis is that IGOs might not be mandating a future partnership between NGOs. Rather, the prior link through the IGO might serve as an introduction, generating an opportunity for two NGOs to become acquainted. The fact that we found no significance for this parameter across all seven waves would suggest that NGO partnerships and networks are less impacted by IGO influence than expected. An interesting topic for future research would be to develop a model for predicting when IGOs can promote emergent cooperative strategies among NGOs, and when the IGOs must instead mandate collaboration.

A related limitation is that our analytical approach does not allow us to determine the degree to which multiparty arrangements contribute to the findings observed here. Whereas some of the classic work on multiparty partnerships (i.e., Gray, 1989; Selsky, 1991) assumes collaboration between two organizations, NGOs may instead enter into collaborations that have three or more parties. If such arrangements are prevalent, there would likely be fewer unconnected NGOs, overestimating the network’s tendency toward closure. In such a network, fewer opportunities would exist for brokerage because coalitions would bring many organizations into direct partnership. The dyad-based data examined here do not allow us to empirically investigate this question, but it is a fruitful opportunity for future research.

Our use of a 2-year interval approach was appropriate, given high organizational turnover and allowed us to identify trends across the network over time, but does not allow us to determine whether ties between specific organizations were maintained or terminated between waves. Moreover, it does not allow us to identify other factors (e.g., resources) that may influence turnover. Understanding more about the persistence of ties between NGOs is a valuable arena for future work.

Finally, the extent to which these data conform to the evolutionary network pattern suggested by Bryant and Monge (2008) is persuasive but incomplete. Future research might examine additional data extending from 2004 into the present to determine whether and how the network moves from a period of maintenance into its next phase. In addition, more longitudinal analysis of different populations of NGOs is needed to determine if there is indeed a life cycle or other nonlinear progression of collaboration dynamics.

Conclusion

This study increases our understanding of the ways NGOs collaborate by investigating which partnering strategies characterize the network over time. The results confirm that NGOs tend to pursue closure, a strategy which has been shown to reduce risk (Gulati, 1995; Gulati & Gargiulo, 1999), increase trust (Saxenian, 1994), promote accountability (Burt, 2001), mobilize resources (Obstfeld, 2017), and generate other collective benefits. Furthermore, this analysis shows that closure declined as the NGO network settled into more stable and consistent partnering patterns.

This research makes several contributions to our understanding of NGO partnering. First, it compares, in a single study, the relative influence of three partnering strategies. Although previous research found support for each of the strategies in different samples, this research examines how the three strategies collectively shape NGO organizing across issue sectors over time.

Second, this research contributes to the nascent application of STERGM as a methodological approach for the study of dynamic organizational networks and demonstrates one methodological strategy available to scholars who hope to study NGO networks longitudinally. The use of ERGMs to analyze these networks is important because it controls for the high level of interdependency that occurs among social phenomena and social networks.

Third, this research suggests that NGOs partnering strategies are not static over time, which should be of deep concern to researchers who seek to make generalizable claims about civil society partnering from cross-sectional or limited-horizon panel analyses. This is important because researchers can mistakenly assume that mechanisms in a cross-sectional design hold across time. Clearly, our results reveal interesting differences across waves. To the extent that phase or other nonlinear models of community network evolution apply, we will need to hypothesize and explore dynamic dimensions of NGO networks.

Finally, this research paints a picture of NGO’s strategic partnering behavior as highly collaborative and community-building. Rather than conform to partnership mandates from IGOs or position themselves to generate individual benefits, NGOs partner in ways that build dense network connections and promote collective benefits for the larger networks to which they belong.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors wish to acknowledge the Annenberg School for Communication and Journalism at the University of Southern California for their financial support for this project.

Author Biographies

Nina F. O’Brien (PhD, University of Southern California) is an assistant professor, dually appointed in the Departments of Communication Studies and Management. Her research focuses on the organizational and communication practices and systems which characterize creative industries like film, television, and news media production, as well as the not-for-profit and civil society sectors. Her research also extends to the study of labor conditions, relations, and negotiations as they occur in these unique industries.

Andrew Pilny (PhD, University of Illinois Urbana-Champaign) is an assistant professor and studies organizational communication, teams, and network science. He also studies communication and behavior in virtual worlds, including experimental gaming designs.

Yannick C. Atouba (PhD, University of Illinois at Urbana-Champaign) is an assistant professor of organizational communication in the Department of Communication at the University of Texas-El Paso, where he teaches courses on organizational communication, conflict communication, leadership, and quantitative research methods. His research focuses on (a) the examination interorganizational forms, processes, and outcomes, especially among nonprofits, and (b) the examination of organizational stakeholders’ behaviors and attitudes, and their impact on organizations and organizing. He also specializes in quantitative methods and social network analysis to examine organizational and interorganizational phenomena, and his research has been published in various communication and nonprofit journals.

Michelle Shumate (PhD, University of Southern California) is the founding director of the Network for Nonprofit and Social Impact. In addition, she is an associate professor in Communication Studies and associate faculty at the Institute for Policy Research at Northwestern University. She introduced theory and research inaugurating the bona-fide network perspective, a public communication-centered model of corporation-nonprofit relationships, and a typology of communication networks. Her current work centers on the question: how can nonprofit networks be rewired for maximum social impact?

Janet L. Fulk (PhD, The Ohio State University) is a professor of communication in the Annenberg School for Communication and professor of Management & Organization in the Marshall School of Business at the University of Southern California. Dr. Fulk’s research focuses on the role of communication and information technology for organizations.

Peter R. Monge (PhD, Michigan State University) is a professor of communication at the Annenberg School for Communication & Journalism and professor of Management and Organization at the Marshall School of Business, University of Southern California. He is also the director of the Annenberg Networks Network, a research center focused on communication network theory and research.

References

Astley

W. G.

(1985). The two ecologies: Population and community perspectives on organizational evolution. Administrative Science Quarterly, 30, 224-241.

Atouba

Y. C.

Shumate

(2010). Interorganizational networking patterns among development organizations. Journal of Communication, 60, 293-317. doi:10.1111/j.1460-2466.2010.01483.x

Atouba

Y. C.

Shumate

(2015). International nonprofit collaboration: Examining the role of homophily. Nonprofit and Voluntary Sector Quarterly, 44, 587-608. doi:10.1177/0899764014524991

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, 726-758.

Baker

W. E.

Obstfeld

(1999). Social capital by design: Structures, strategies, and institutional context. In Leenders

R. Th. A. J.

Gabbay

S. M.

(Eds.), Corporate social capital and liability (pp. 88-105). New York, NY: Springer.

Baldassari

Diani

(2007). The integrative power of civic networks. American Journal of Sociology, 113, 735-780. doi:10.1086/521839

Brinkerhoff

J. M.

(2002). Government-nonprofit partnership: A defining framework. Public Administration and Development, 22, 19-30. doi:10.1002/pad.203

Bryant

J. A.

Monge

P. R.

(2008). The evolution of the children’s television community, 1953-2003. International Journal of Communication, 2, 160-192.

Burt

R. S.

(1992). Structural holes: The structure of social capital competition. Cambridge, MA: Harvard University Press.

10.

Burt

R. S.

(2001). Structural holes versus network closure as social capital. In Lin

Cook

K. S.

Burt

R. S.

(Eds.), Social capital: Theory and research (pp. 31-56). New Brunswick, NJ: Transaction.

11.

Castells

(1999). Information technology, globalization and social development (Vol. 114). Geneva, Switzerland: United Nations Research Institute for Social Development.

12.

Castells

(2004). The network society: A cross-cultural perspective. Cheltenham, UK: Edward Elgar.

13.

Doerfel

M. L.

Taylor

(2004). Network dynamics of interorganizational cooperation: The Croatian civil society movement. Communication Monographs, 71, 373-394. doi:10.1080/0363452042000307470

14.

Doerfel

M. L.

Taylor

(2017). The story of collective action: The emergence of ideological leaders, collective action network leaders, and cross-sector network partners in civil society. Journal of Communication, 67, 920-943.

15.

Federal Emergency Management Agency. (2008). National response framework. Retrieved from http://www.fema.gov/pdf/emergency/nrf/nrf-core.pdf

16.

Fox

(1996). How does civil society thicken? The political construction of social capital in rural Mexico. World Development, 24, 1089-1103. doi:10.1016/0305-750X(96)00025-3

17.

Gargiulo

Benassi

(2000). Trapped in your own net? Network cohesion, structural holes, and the adaptation of social capital. Organization Science, 11, 183-196.

18.

Gazley

Guo

(2015). What do we know about nonprofit collaboration? A Comprehensive systematic review of the literature. Academy of Management Proceedings, 2015(1), Article 15409. doi:10.5465/ambpp.2015.303

19.

Gittell

Vidal

(1998). Community organizing: Building social capital as a development strategy. Thousand Oaks, CA: SAGE.

20.

Graddy

E. A.

Chen

(2006). Influences on the size and scope of networks for social service delivery. Journal of Public Administration Research and Theory, 16, 533-552. doi:10.1093/jopart/muj005

21.

Gray

(1989). Collaborating: Finding common ground for multiparty problems. San Francisco, CA: Jossey-Bass.

22.

Green

H. D.

Wasserman

(2013). Network analysis: A definitional guide to important concepts. In Little

T. D.

(Ed.), Oxford handbook of quantitative methods (pp. 154-174). New York, NY: Oxford University Press.

23.

Gulati

(1995). Social structure and alliance formation patterns: A longitudinal analysis. Administrative Science Quarterly, 40, 619-652.

24.

Gulati

Gargiulo

(1999). Where do interorganizational networks come from? American Journal of Sociology, 104, 1439-1493.

25.

Guo

Acar

(2005). Understanding collaboration among nonprofit organizations: Combining resource dependency, institutional, and network perspectives. Nonprofit and Voluntary Sector Quarterly, 34, 340-361. doi:10.1177/0899764005275411

26.

Hager

M. A.

Galaskiewicz

Larson

J. A.

(2004). Structural embeddedness and the liability of newness among nonprofit organizations. Public Management Review, 6, 159-188.

27.

Heffren

McDonald

Casebeer

Wallsten

(2003). Mandated intersectoral, interdisciplinary collaboration: Opportunity or oxymoron? In Hibbert

(Ed.), Co-creating emergent insight (pp. 221-227). Glasgow, Scotland: Graduate School of Business, University of Strathclyde.

28.

Hunter

D. R.

(2007). Curved exponential family models for social networks. Social Networks, 29, 216-230. doi:10.1016/j.socnet.2006.08.005

29.

Huxham

Vangen

(2013). Managing to collaborate: The theory and practice of collaborative advantage. New York, NY: Routledge.

30.

Keck

M. E.

Sikkink

(1998). Activists beyond borders. Ithaca, NY: Cornell University Press.

31.

Keyes

L. C.

Schwartz

Vidal

A. C.

Bratt

R. G.

(1996). Networks and nonprofits: Opportunities and challenges in an era of federal devolution. Housing Policy Debate, 7, 201-229. doi:10.1080/10511482.1996.9521219

32.

King

N. C.

(2004). Social capital and nonprofit leaders. Nonprofit Management & Leadership, 14, 471-486.

33.

Klerx

Leeuwis

(2009). Establishment and embedding of innovation brokers at different innovation system levels: Insights from the Dutch agricultural sector. Technological Forecasting and Social Change, 76, 849-860. doi:10.1016/j.techfore.2008.10.001

34.

Krivitsky

P. N.

Handcock

M. S.

(2014). A separable model for dynamic networks. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 76, 29-46. doi:10.1111/rssb.12014

35.

Longoria

R. A.

(2005). Is inter-organizational collaboration always a good thing? Journal of Sociology & Social Welfare, 32, 123-138.

36.

Lusher

Koskinen

Robins

(Eds.). (2012). Exponential random graph models for social networks: Theory, methods, and applications. Cambridge, UK: Cambridge University Press.

37.

Madon

(1999). International NGOs: Networking, information flows and learning. Journal of Strategic Information Systems, 8, 251-261.

38.

Margolin

D. B.

Shen

Lee

Weber

M. S.

Fulk

Monge

(2015). Normative influences on network structure in the evolution of the children’s rights NGO network, 1977-2004. Communication Research, 42, 30-59.

39.

McPherson

Smith-Lovin

Cook

J. M.

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415-444. doi:10.1146/annurev.soc.27.1.415

40.

Merton

R. K.

(1968). Social theory and social structure. New York, NY: Simon & Schuster.

41.

Meyer

Hyde

(2004). Too much of a “good” thing? Insular neighborhood associations, nonreciprocal civility, and the promotion of civic health. Nonprofit and Voluntary Sector Quarterly, 33, 77S-96S.

42.

Monge

P. R.

Contractor

N. S.

(2003). Theories of communication networks. New York, NY: Oxford University Press.

43.

Monge

P. R.

Heise

B. R.

Margolin

(2008). Communication network evolution in organizational communities. Communication Theory, 18, 449-477.

44.

O’Brien

N. F.

Evans

S. K.

(2017). Civil society partnerships: Power imbalance and mutual dependence in NGO partnerships. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 28, 1399-1421. doi:10.1007/s11266-016-9721-4

45.

Obstfeld

(2017). Getting new things done: Networks, brokerage and the assembly of innovative action. Stanford, CA: Stanford University Press.

46.

Passey

Lyons

(2006). Nonprofits and social capital: Measurement through organizational surveys. Nonprofit Management and Leadership, 16, 481-495.

47.

Phelps

C. C.

(2010). A longitudinal study of the influence of alliance network structure and composition on firm exploratory innovation. Academy of Management Journal, 53, 890-913.

48.

Rabrenovic

Pierce

(2003). Connecting the community: Information as a resource to create social capital and enhance community economic development. In Body-Gendrot

Gittell

(Eds.), Social capital and social citizenship (pp. 65-77). Lanham, MD: Lexington Books.

49.

Robins

(2015). Doing social network research: Network-based research design for social scientists. New York, NY: SAGE.

50.

Robins

Pattison

Kalish

Lusher

(2007). An introduction to exponential random graph (p*) models for social networks. Social Networks, 29, 173-191.

51.

Rodríguez

Langley

Béland

Denis

J. L.

(2007). Governance, power, and mandated collaboration in an interorganizational network. Administration & Society, 39, 150-193. doi:10.1177/0095399706297212

52.

Saxenian

(1994). Lessons from silicon valley. Technology Review, 97, 42-51.

53.

Schneider

J. A.

(2009). Organizational social capital and nonprofits. Nonprofit and Voluntary Sector Quarterly, 38, 643-662.

54.

Seitanidi

M. M.

(2010). The politics of partnerships: A critical examination of nonprofit-business partnerships. New York, NY: Springer.

55.

Selsky

J. W.

(1991). Lessons in community development: An activist approach to stimulating interorganizational collaboration. The Journal of Applied Behavioral Science, 27, 91-115. doi:10.1177/0021886391271005

56.

Shumate

Fulk

Monge

(2005). Predictors of the International HIV–AIDS INGO network over time. Human Communication Research, 31, 482-510. doi:10.1111/j.1468-2958.2005.tb00880.x

57.

Sikkink

Smith

(2002). Infrastructures for change: Transnational organizations, 1953-1993. Minneapolis: University of Minnesota Press.

58.

Smith

(1997). Characteristics of the modern transnational social movement sector. In Smith

Chatfield

Pagnucco

(Eds.), Transnational social movements and global politics: Solidarity beyond the state (pp. 42-58). Syracuse, NY: Syracuse University Press.

59.

Steffek

(2013). Explaining cooperation between IGOs and NGOs—Push factors, pull factors, and the policy cycle. Review of International Studies, 39, 993-1013.

60.

Stephens

K. J.

Fulk

Monge

P. R.

(2009). Constrained choices in alliance formations: Cupids and organizational marriages. Human Relations, 62, 501-536. doi:10.1177/0018726708101982

61.

Stohl

(2005). Human rights, nation states, and NGOs: Structural holes and the emergence of global regimes. Communication Monographs, 72, 442-467. doi:10.1080/03637750500322610

62.

Taylor

Doerfel

M. L.

(2003). Building interorganizational relationships that build nations. Human Communication Research, 29, 153-181. doi:10.1111/j.1468-2958.2003.tb00835.x

63.

Tushman

M. L.

Romanelli

(1985). Organizational evolution: A metamorphosis model of convergence and reorientation. Organizational Behavior, 7, 171-222.

64.

Union of International Associations. (2017). Yearbook of international organizations online: Guide to global civil society networks. Retrieved from http://www.uia.org

65.

Uzzi

(1999). Embeddedness in the making of financial capital: How social relations and networks benefit firms seeking financing. American Sociological Review, 64, 481-505. doi:10.2307/2657252

66.

Wasserman

Faust

(1994). Social network analysis: Methods and applications. Cambridge, UK: Cambridge University Press.

67.

White

B. N. F.

(2005). Shifting positions on child labor: The views and practice of intergovernmental organizations. In Weston

(Ed.), Child labor and human rights: Making children matter (pp. 319-342). Boulder, CO: Lynne Rienner.

68.

Wood

D. J.

Gray

(1991). Toward a comprehensive theory of collaboration. The Journal of Applied Behavioral Science, 27, 139-162. doi:10.1177/0021886391272001