Module 7: Group Discussion

©The Author 2017. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: [email protected].

1

Journal of Public Administration Research And Theory, 2017, 1–18 doi:10.1093/jopart/mux026

Article

Article

Picking the Team: A Preliminary Experimental Study of the Activation of Collaborative Network Members Chris Silvia

Brigham Young University

Address correspondence to the author at [email protected].

Abstract

Among a collaborative leader’s most important duties is selecting a collaborative partner. Numerous perspectives, including resource dependency theory, institutional theory, transaction cost theory, and personality typologies, have been used to help explain this decision. Clearly, a collaborative leader would desire to work with an individual who has access to needed resources, has a personality that fits the network, and is familiar. However, such a perfect partner does not often, if ever, exist. Therefore, a collaborative leader must make trade-offs between the issues of resource access, personality, and familiarity. Using an experimental design, this study explores how collaborative leaders make these trade-offs when considering potential collaborative part- ners. The findings suggest that while prospective partner personality may be the most significant driver of the partnership decision, it is actually the combination of factors, especially personality and resource access that interact to determine partner desirability.

Introduction

Over the last few decades, government leaders have increasingly utilized collaborative approaches to prob- lem solving and service delivery (O’Toole 1997). As McGuire and Silvia (2010) commented, “the public manager of the current era must regularly and skillfully navigate a multitude of actors and programs in the intergovernmental system,” and “there is little doubt that interdependence and interconnectedness charac- terize the intergovernmental environment of today’s public organizations” (279). However, despite the ubiquity of collaborative arrangements, the practice of collaboration is not easy. Collaborative leaders and the networks they lead face many challenges, such as knowledge gaps, legal barriers, resource scarcity, exer- cise of agency negative power, and political opposition (Agranoff 2012). Central to a collaborative leader’s role is the selection of the people and resources that the network needs to achieve its objectives (O’Leary

and Vij 2012). A collaborative leader must determine the discrepancy between what the network needs in terms of people, perspectives, and resources and what it currently has. Given that the rationale for collabo- rating is often the realization that a problem “cannot be solved—or solved easily—by a single organization” (McGuire 2006, 34), the decision regarding with whom to collaborate is not trivial. Graddy and Chen (2009) suggest “three factors that [they] believe explain the choice of partners in collaborative arrangements: pro- grammatic needs that promote resource exchange; organization legitimacy goals; and efforts to reduce the transaction costs associated with partnership for- mation and management” (55). By identifying and recruiting partners, a collaborative leader can amass the resources, garner the information, and assem- ble the people that individual organizations lack, yet require, to address the issue that brings them together. This accumulation and retention of “resources like

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx2

money, information, and expertise can be the integrat- ing mechanism of networks” (Agranoff and McGuire 2001, 298) that maintains a sense of interdependence among the collaborative parties and engenders stabil- ity within the network membership by incentivizing participation and fostering the network’s ability to achieve its mission and goals (Ansell and Gash 2008).

Many scholars have also argued that the incorpora- tion of the right partners is a key to success (Graddy and Chen 2009). Central to this idea is that who is asked to participate in the collaborative is as impor- tant as what the potential partners can contribute in terms of resources. This is because, while collabora- tive partners each represent their own home agencies or interest groups, it is the people, not the organiza- tions that actually collaborate. The convener of the collaborative must carefully consider the personalities of each partner candidate. Hackman (2011, 155) pos- ited that “60 percent of the difference in how well a team eventually performs is determined by the quality of prework,” including determining the group com- position. Identifying the “right” people is important because of the shared leadership arrangement inherent in many collaboratives. Perhaps more so than in a hier- archical structure, “every man in a group is to some extent a leader in so far as every man has some effect upon the syntality of a group” (Cattell 1951, 25). In other words, the collaborative’s personality is a func- tion of the personalities of the individual members of the collaborative and the results of the collaborative effort “depend on the team members’ characteristics” (Vincentini and Boccardelli 2014, 40). Each partner’s contribution to the overall success of the collaborative is not only a function of the resources that they con- tribute, such as money, information, reputation, etc., but also each individual’s attributes.

Collaborative arrangements are not without their skeptics. A  common caution from practitioners and scholars is to collaborate only when necessary because collaboration is a “seriously resource-consuming activ- ity” (Huxham and Vangen 2005, 13)  that requires a great deal of time and energy to make it successful. In their examination of the impact of collaborative lead- ership behaviors on network effectiveness, McGuire and Silvia (2009) found that engaging in framing behaviors, including those leadership behaviors that focus on establishing agreement regarding individu- als’ roles and the operating rules of the network, were negatively related to network effectiveness. They argue that if the leader must continually spend time getting everyone on the same page, then the members lack the unity necessary for success. The lack of agreement means that the network must keep its focus on deter- mining how to work together instead of actually work- ing together toward accomplishing their mission and

goals. “Organizational performance is an outcome of how well . . . interdependent elements are aligned . . . [and] are working in concert to attain specific goals that ultimately help the organization fulfill its mission” (Daft 2008, 216). One way that managers attempt to decrease these transaction costs is to work with those with whom they are familiar. “Members who are familiar with one another and with their work con- text are able to settle in and focus on the work rather than waste time and energy getting oriented to new coworkers or circumstances” (Hackman 2011, 62). Thus, working with familiar individuals is not only easier, since they are a known quantity, but also leads to greater efficiency and effectiveness.

Recently, there has been an increase in scholarly interest regarding collaborative partner selection (e.g., Berardo and Scholz 2010; Calanni et al. 2015; Feiock, Lee, and Park 2012; Graddy and Chen 2009; Ryu 2014). Each of these studies has added much to the conversation regarding this important management topic. This article seeks to build on these earlier studies by examining the trade-offs collaborative leaders make when selecting pro- spective partners. As discussed above, partner selection is influenced by the potential partner’s access to resources, potential to increase the collaborative’s legitimacy, famil- iarity among the current members of the collaborative, and his or her personality. Certainly, a collaborative leader would like to include a person who represents an organization that has needed resources, has a personality that fits the network, and is familiar. However, since the perfect collaborative partner likely does not exist, trade- offs must be made. This article strives to discover how public managers make these trade-offs when considering a potential collaborative partner. Thus, the research ques- tion explored in this article seeks to determine which mix of benefits (i.e., resource exchange, associational legiti- mization, transaction cost reduction, and personality type) is seen as providing the best value when confront- ing the inevitable trade-offs that must be addressed when selecting partners.

Literature Review

This study brings together a number of different lit- eratures as a foundation for the examination of how collaborative partners are selected. In so doing, there are a number of terms that will be used in this article that either may not be consistently defined within or across the fields from which they are taken. For exam- ple, a term such as “collaboration” likely means differ- ent things to different people based upon the field of scholarship. Therefore, it is essential that these terms be defined for the purpose of this study.

Consistent with definitions used by Agranoff and McGuire (2003) and Huxham and Vangen (2005), the

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 3

term “collaboration” will be used in this study to refer to the process by which two or more entities work across organizational boundaries to achieve something that they could not accomplish alone. Networks, on the other hand, are structures. O’Toole defined networks as “structures of interdependence involving multiple organizations or parts thereof, where one unit is not merely the formal subordinate of the others in some larger hierarchical arrangement” (1997, 45). Thus, the term “network,” which will be used synonymously in this study with terms like “a collaborative” or “a col- laborative group,” refers to the organizational struc- ture within which the process of collaboration takes place.

The literature differentiates between types of net- works based upon their function. For example, knowl- edge networks (e.g., Eglene, Dawes, and Schneider 2007) form as a forum within which to share infor- mation, governance networks (e.g., Klijn and Skelcher 2007) form to make and implement policy, and service delivery networks (e.g., Romzek et al. 2014) form to facilitate the rendering of public services. However, regardless of the function of the network, they must all find and incorporate new members to ensure that the network has the right people and resources to execute its mission.

The literature also differentiates between types of networks based upon their structure. Perhaps the most notable example is Provan and Kenis’ (2008) typol- ogy consisting of three forms: network administrative organization, lead agent-governed, and participant- governed. In these structures, the network adminis- trator, lead agent, and individual network members, respectively, are all tasked with identifying and recruiting new members as the circumstances require. Therefore, in the context of this current study, the dif- ferences between network functions and structures are not critical.

Others differentiate networks based upon their institutional constraints. “Collaboration can be either formal (mandated by the state) or informal, involve many organizations or few, it can be vertical and/or horizontal, and it can be intra- and interorganiza- tional” (Smith 2009, 1). Constraints such as these play a significant role in the partnership decision. For exam- ple, there are many situations where collaboration is not only mandated, but the collaborative participants are also mandated. Further, all collaborative arrange- ments are different and thus will operate by different rules. While these are important factors in collabora- tion, institutional factors that constrain partner choice are not explicitly included in this study. Instead, a basic framework for partnership selection based upon col- laborative style, resources, and familiarity is proposed and tested. By removing these contextual variables,

the underlying partner selection logic of the decision maker can be identified more clearly. This study is a first step and can serve as a scaffold upon which future studies can build to understand how the findings pre- sented here can be applied to specific institutional arrangements.

For purposes of this article, an individual within the network that is considering incorporating a new mem- ber will be referred to using the term “collaborative leader” and the new member will be referred to using the term “potential partner.” The term “collaborative leader” does not indicate that this individual exerts complete control over who participates in the network. As Feyerherm (1994) concludes in her study, in a col- laborative setting, leadership is often exhibited by both the acknowledged leader as well as by the other mem- bers of the collaborative. Instead of a single collabora- tive leader, the leadership role is shared by multiple, and often all, members of the network (e.g., Ansell and Gash 2008; Crosby and Bryson 2005). This diffusion of leadership functions means that all members can, and likely do, play the role of a leader and that those who are recruited as partners are both future members and future leaders of the group. The potential partner is an individual that represents his or her home organi- zation in the network.

There are numerous perspectives that can speak to collaborative partner selection. Those considered in this study are resource dependency theory, institutional theory, transaction cost theory, and syntality/person- ality. Each perspective offers a different rationale for the inclusion of a potential partner. Importantly, these perspectives are not mutually exclusive, since each pro- spective partner will have a unique mix of access to resources, legitimating power, familiarity among the current collaborative partners, and personality type. Thus, the aim of this study is not to test these four per- spectives against each other, but rather to identify the trade-offs among them that dictate partner selection.

Resource Dependency Theory Arguably the most common perspective (Hu, Khosa, and Kapucu 2015) used in the context of collabora- tive partner selection is resource dependency (Pfeffer and Salancik 2003). A  single organization’s inability to address an issue on its own often results from a lack of necessary resources. Thus, when a network lacks the resources it needs to accomplish its goals and mis- sion, it is often incumbent upon a collaborative leader to identify and recruit partners that have access to those resources. These partners work for organiza- tions that can provide the network with the desired resources. A collaborative leader, therefore, may pur- sue a collaborative partnership as a way to manage uncertainties in their network’s resource environment

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx4

(Gray and Wood 1991). Based upon her case study work, Cigler (2001) identified a set of nine precon- ditions, including capacity building, that help explain the decision to pursue a collaborative venture. Cigler’s work “could not uncover even one organization that had emerged without some type of capacity-building assistance external to the organization” (78–9). As some Pentagon planners have said, “a vision without resources is a hallucination” (Friedman 2009, 205). Thus, resource dependency theory would suggest that the desirability of a collaborative partner is based on their access to the resources needed to successfully meet the network’s objectives that are not currently possessed by either the collaborative leader’s home organization or by the other members of the network (Bardach 1998; Gray and Wood 1991; Malatesta and Smith 2014).

Institutional Theory Another theory that is frequently used to help explain collaboration is institutional theory, which suggests that the selection of a collaborative partner is based on the network’s strategic decision to improve its reputa- tion, image, or prestige among its own membership and stakeholders (DiMaggio and Powell 1983;Graddy and Chen 2009). “Collaborative projects often need broad public support and/or public legitimacy because they normally represent the facilitated sharing of power and resources from many agencies and affected persons” (Agranoff 2012, 12). However, collaborations are not automatically granted legitimacy (Bryson, Crosby, and Stone 2006). Instead, the collaborative must build and maintain legitimacy, an objective that can potentially be achieved through the efforts and reputation of col- laborative partners.

The reputation of collaborative partners has a num- ber of distinct advantages for the network at various phases of the network’s life cycle. Human and Provan (2000) identified three distinct aspects of network legitimacy: as a form, as an entity, and as an interac- tion. Since the collaborative partners are the face of the collaborative—both within the network and to exter- nal stakeholders—prestigious partners can help the network achieve legitimacy in all three areas.

In terms of building legitimacy of the network as a form, having a highly visible individual involved can help establish the network as an acceptable form of organizing for the purposes of solving an intractable problem. This is particularly the case in the beginning stages of the collaborative’s existence because it helps with recruiting collaborative partners (Linden 2010). In addition to helping establish the credibility of the network as an organizational form in the eyes of poten- tial collaborative partners, networks must also estab- lish the credibility of the network as a form in the eyes

of external groups and stakeholders, including poten- tial funders (Human and Provan 2000). This point was illustrated in Crosby and Bryson’s (2005) discussion of the 2002 Vital Aging Summit in Minnesota, in which “members of the summit planning committee used their personal connections and credibility with foun- dation directors to win donations” (303).

Once the legitimacy of the network as a form of organ- izing is established, the network must establish its own legitimacy as an entity. Of his seven key collaborative factors, Linden (2010) identified the inclusion of “some- one with credibility and clout . . . [as potentially] the most important” (49). Linden’s argument for this assertion is essentially that a champion can help the network gain legitimacy as an entity both among the network mem- bers and their stakeholders, thereby gaining their sup- port. Demonstrating the legitimacy of the network and actively mobilizing the support of both members and stakeholders of the network have been found to be inte- gral to the success of the network (McGuire and Silvia 2009). This was illustrated during the establishment of the Hennepin County African American Men Project, where the collaborative leaders sought out “African Americans who had rich connections and credibility in their own ethnic community,” resulting in a collabora- tive “that had prestige in the eyes of many stakeholders” (Crosby and Bryson 2005, 66). As Crosby and Bryson remark, “solutions gain prestige through attachment to respected people, institutions, or processes” (248).

The final dimension of legitimacy that is important for collaborative leaders to consider is that of the network as an interaction. “The interaction process itself [has] to be legitimized so network members [will] be willing to work together to build and maintain the levels of involve- ment and norms of cooperation that would be critical for sustaining the network (Powell 1990)” (Human and Provan 2000, 340). This is akin to McGuire and Silvia’s (2009) findings regarding mobilizing and synthesizing, both of which were found to be positively associated with network effectiveness. Mobilization behaviors are those that “develop support for network processes from network participants” and involve “establishing and maintaining [the network’s] legitimacy” (39–40), and synthesizing behaviors are those that “create and main- tain trust among network participants as a means to build relationships and interactions that result in achiev- ing the network’s purpose” (40). Partner trustworthiness has been cited as being especially important in situations where there is increased commitment risk (Feiock, Lee, and Park 2012) or where there is low social trust and a lack of potential collaborative partners with whom one is familiar or with whom one has an existing personal relationship. In such contexts, collaborating with a repu- table party “could provide a ‘guarantee’ of reliability” (Koljatic, Silva, and Valenzuela 2006, 65).

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 5

Transaction Cost Theory As mentioned previously, the selection of a collabora- tive partner is frequently driven by the identification of a need that is not being met by the current compo- sition of the network. Often it is familiar individuals who come to mind when deciding whom to recruit to fill that deficiency. From a transaction cost perspective, this makes sense: recruiting a known entity decreases the search and information cost of identifying others with the personalities and capabilities of the familiar individuals. As also mentioned earlier, uncertainty is a concern for a collaborative leader, both in terms of whether to and with whom to collaborate. One way to reduce this uncertainty is to collaborate with some- one the leader knows and has worked with in the past (Hinds et al. 2000). Hinds et al. (2000) identified two key mechanisms by which this decision occurs. First, past experience is the best predictor of future perfor- mance. Second, less time is required to socialize and bring a familiar partner up to speed (Cohen, Ledford, and Spreitzer 1996). There is also evidence that work- ing with familiar individuals increases productivity and decision-making effectiveness (Sawyer 2007). Sawyer (2007) identified familiarity as being particu- larly helpful for tasks requiring “problem solving crea- tivity” (52) and as one of the ten conditions that foster group flow, or “a peak experience, a group performing at its top level of ability” (43). Familiarity often arises from a history of successful past collaboration, which has been found to engender trust because partners are better able to predict each other’s actions (Gulati and Sytch 2008). Familiarity decreases the uncertainty of a partnership and helps collaborators to establish a “common language and a common set of unspo- ken understandings” (Sawyer 2007, 51) and “mutual understanding, internal legitimacy, and shared com- mitment” (Emerson, Nabatchi, and Balogh 2012, 14).

Personality The final major factor that has been used to explain col- laborative partner selection is personality. Among the resources that a collaborative has at its disposal is the individual characteristics of its members. While little research has been done regarding the impact of person- ality traits in the area of collaboration, there is a rich literature on this subject relating to group dynamics and group effectiveness (e.g. Bell 2007; Kramer, Bhave, and Johnson 2014; LePine et al. 2011). The major premise behind this stream of research is that the “personality factors of team members are characteristic patterns of thinking, feeling, and acting that should affect team per- formance through a variety of processes ranging from how team members approach task completion to how team members interact with one another” (Bell 2007, 597). Personality and individual characteristics are

thought to be particularly important in group dynam- ics because they “are not only directly relevant to the task-focused contributions individual members make to team outcomes, but they also influence the manner in which members react and relate to each other in the course of performing work together as ongoing units of work” (LePine et al. 2011, 312). Thus, there is likely an interaction between the partners’ personalities that not only impacts the effectiveness and efficiency of the col- laborative’s work but also the group dynamic in terms of trust and mutual understanding.

Among the myriad personality typologies are Parker’s (2008) team player styles. Parker’s typology was developed using data from approximately 3,400 individuals working in team, group, and collaborative arrangements. The styles include contributor (task- oriented), collaborator (goal-directed), communica- tor (process-oriented), and challenger (questioning). Parker (205–6) provides the following descriptions:

Contributor The contributor is a task-oriented team mem- ber who enjoys providing the team with good technical information and data, does his or her homework, and pushes the team to set high per- formance standards and to use their resources wisely. Most people see a contributor as depend- able, although they believe that at times a con- tributor may become too bogged down in the details and data or fail to see the big picture or the need for a positive team climate. Collaborator The collaborator is a goal-directed member who sees the vision, mission, or goal of the team as paramount but is flexible and open to new ideas, willing to pitch in and work outside his or her defined role, and able to share the limelight with other team members. Most people see a collab- orator as a big-picture person, but they believe that at times a collaborator may fail to periodi- cally revisit the mission, give enough attention to the basic team tasks, or consider the individual needs of other team members. Communicator The communicator is a process-oriented mem- ber who is an effective listener and facilitator of involvement, conflict resolution, consensus build- ing, feedback, and the building of an informal, relaxed climate. Most people see a communica- tor as a positive “people person,” but they find that at times a communicator may see the process as an end in itself and that a communicator may not confront other team members or give enough emphasis to completing task assignments and making progress toward team goals.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx6

Challenger The challenger is a member who questions the goals, methods, and even the ethics of the team; is willing to disagree with the leader or higher authority; and encourages the team to take well- conceived risks. Most people appreciate the value of a challenger’s candor and openness, but they think that at times a challenger may not know when to back off an issue or may become self- righteous and try to push the team too far.

There are a number of ways that personality could be theorized as affecting the partnership decision. If Wernerfelt’s (1984) broad definition of a resource as being anything that could be considered as a strength for an organization is applied to collaborative groups, it is easy to see how these collaborative role person- alities described above could themselves be seen as resources. Each role type has unique strengths that balance out the weaknesses of the others. Achieving a balance between the different roles could be important for the proper functioning of a group. Therefore, see- ing personality as a resource that can be incorporated into the network, a leader may purposefully collabo- rate with those unlike themselves or the other members of the collaborative in order to harness the comple- mentarity of the roles. Conversely, a leader may see a potential partner’s personality as impacting the trans- action costs of working together. Certain personalities are just more difficult to work with than others. Hence, leaders may purposefully avoid potential partners who are challengers, a personality type that has a reputa- tion as being difficult to work with (Clay-Williams and Braithwaite 2015). Finally, homophily would predict that collaborative leaders will choose partners that are like themselves because similarity not only increases the likelihood of collaboration (Feiock, Lee, and Park 2012) but also may spawn stronger relationships (McPherson, Smith-Lovin, and Cook 2001).

Data and Methods

Although experimental research designs are becom- ing increasingly more common across the social sci- ences, such designs remain relatively uncommon in public management research (Christensen et al. 2013; Margetts 2011). While novel research methods are interesting, the decision regarding which method to use must be based on the research question being asked. Since the objective of this study was to inves- tigate the trade-offs that a collaborative leader makes when selecting a collaborative partner, this article relies on a mixed-experimental design to uncover the causal relationships underlying this decision. The fol- lowing discussion highlights how this design meets the three criteria necessary to make a claim of a causal

relationship between the intervention and group differ- ences: an intervention, the measurement of outcomes, and the random group assignment (Margetts 2011).

Intervention and Measurement of Outcomes The mixed-experimental design used in this study is similar to that used in Christensen et  al. (2013) and Whiting et al. (2008). Namely, this study also incorpo- rates a within-subjects “paper-people” design method- ology. Each participant was asked to assume the role of a collaborative leader and was presented with a series of four experiments. Each experiment was separated by sets of distractor questions. These distractor ques- tions were related to collaboration but not related to partner selection. They were incorporated so that the respondent would be unable to discern the motivation for the study and game the experiment.

Within each experiment, respondents were ran- domly assigned to read one of four vignettes and report how likely they would be to select the described individual as a collaborative partner. The first sentence in each vignette identified the potential partner by a gender-neutral name. This was done to personalize and humanize the potential partner while not intro- ducing gender into the selection decision. The name varied by experiment. The first sentence also identified the resource that the potential partner could contrib- ute to the partnership. This study featured three dif- ferent types of resources: money, information, and prestige. Money and information were incorporated based on their prevalence in the collaboration and resource dependency literatures, and prestige was included based on the institutional theory perspective. The resource varied by question within each experi- ment. The second sentence in the prospective partner description drew on both transaction cost theory and the literature on collaborative personalities in present- ing a statement regarding whether the respondent had worked with the potential partner in the past and a short description of the potential partner’s team player style (see table 1 for sample vignettes). The prospec- tive partner’s personality style and familiarity were discussed in the same sentence because it is likely that they interact. The result is a 4 × 3 × 2 experimental design (table 2). The phrasing in the vignettes was con- sistent within and between experiments. For example, the description of each resource was consistent across experiments, and the description of the Parker team player styles and the familiarity statement was consist- ent within experiments. This allowed for the compari- son of aggregated group responses between questions and across experiments.

The design also helped decrease the risk of both manipulation-independent and manipulation-pro- duced confounding variables. The likelihood of

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 7

manipulation-independent confounding variables was addressed using random assignment. Randomized group assignment randomly distributes the effect of an unmeasured variable across the groups. If the groups are not systematically different, then the group mean differences can be attributed to the experimen- tal manipulation. While random assignment cannot

guarantee that the groups do not differ on a key vari- able, it greatly reduces the risk.

The threat of manipulation-produced confounding variables was addressed via the experimental design. This experiment used a controlled setting in which only one characteristic of the potential partner was varied at a time. Group differences are attributable

Table 1. Sample Vignettes

Experiment Description of Potential Collaborative Partner

Experiment 1 N: Jamie Jamie works for an organization that has access to monetary resources that could help your group pursue its mission and goals. Although you have never worked directly with Jamie you have heard that Jamie is task- oriented, responsible, organized, and good with technical information, but can get lost in the details and fail to see the big picture.

P: contributor R: money F: no

Experiment 2 N: Jordan Jordan works for an organization that has access to the information that could help your group pursue its mission and goals. Although you have never worked directly with Jordan you have heard that Jordan is enthusiastic, considerate, supportive, and a consensus builder who works hard to ensure all group members are actively involved, but can place more emphasis on the process the group uses to complete its tasks than on actually completing them.

P: communicator R: information F: no

Experiment 3 N: Lee Lee works for an organization that has a reputation that could help your group pursue its mission and goals. Although you have never worked directly with Lee you have heard that Lee is goal-oriented, open to new ideas, imaginative, and is willing to help others work toward accomplishing the group’s objectives, but can fail to periodically re-visit the group’s mission, give enough attention to the basic tasks of the group, and consider the individual needs of the group members.

P: collaborator R: reputation F: no

Experiment 4 N: Taylor Taylor works for an organization that has access to monetary resources that could help your group pursue its mission and goals. You have worked directly with Taylor in the past and have found Taylor to be honest, candid, open, willing to take well-conceived risks, and play the devil’s advocate by questioning the group’s goals and direction, but can become self-righteous and take it too far and not always know when to back off.

P: challenger R: money F: yes

Note: N, name; P, personality; R, resource; F, familiarity.

Table 2. 4 × 3 × 2 Experimental Design

Experiment Description of Potential Collaborative Partner

Experiment 1 N: Jamie N: Jamie N: Jamie N: Jamie P: contributor P: contributor P: contributor P: contributor R: money R: information R: reputation R: money F: no F: no F: no F: yes

Experiment 2 N: Jordan N: Jordan N: Jordan N: Jordan P: communicator P: communicator P: communicator P: communicator R: money R: information R: reputation R: money F: no F: no F: no F: yes

Experiment 3 N: Lee N: Lee N: Lee N: Lee P: collaborator P: collaborator P: collaborator P: collaborator R: money R: information R: reputation R: money F: no F: No F: no F: yes

Experiment 4 N: Taylor N: Taylor N: Taylor N: Taylor P: challenger P: challenger P: challenger P: challenger R: money R: information R: reputation R: money F: no F: no F: no F: yes

Note: N, name; P, personality; R, resource; F, familiarity.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx8

to the effect of the varied partner characteristic and the effects of the partner’s characteristics are isolated from potential institutional confounders. Therefore, the threat of construct validity problems resulting from attribution uncertainty was minimized.

The questionnaire was tested prior to its adminis- tration to ensure that the wording of the questions was clear and to verify that the vignettes comported with the team player styles as described by Parker (2008). Results of the pretest indicate that the wording of all questions, including those in the experimental sections, was clear and adequately described the underlying constructs.

Participants A total of 229 participants responded to at least one of the experimental questions. In addition to the experi- mental and the distractor questions, the 18-question Parker Team Player Survey (Parker 2008) and demo- graphic questions were asked. All participants were graduates of a highly ranked master of public admin- istration program in the United States. They had, on average, 16.4 years of work experience, held their cur- rent position for 4.7  years, and worked within their current organization for 7.9  years. Approximately 62% of the respondents were male. There are suffi- cient data to determine the team player style for 215 of the respondents. Based on the results to the question- naire, it appears that the individuals in this sample not only see collaboration as a positive problem solving approach but are also active collaborators themselves. This is consistent with the literature that posits that virtually all public managers have and/or will work in a collaborative arrangement (O’Toole 2014). More than 90% of them agreed or strongly agreed that col- laboration is valuable, that it is an important strategy to address complex problems, and that it is benefi- cial in bringing individuals with different skills and perspectives together. Additionally, 82.4% agreed or strongly agreed that they were good at collaborating and 73.4% agreed or strongly agreed that their organi- zation valued collaboration (tables 3 and 4).

Issues of Generalizability and External Validity The objective of this research was to gain a better understanding of the underlying process of collabo- rative partner selection. The contribution this article makes is about theories, not about a population. While certainly some research is conducted as a way to rep- licate a real-world scenario and then to use the subse- quent findings to draw inferences about what would happen in the real world, generalizability and external validity (EV) are not the intention of many experi- ments (Levitt and List 2007; Mook 1983). Mook (1983) argues that

the distinction between generality of findings and generality of theoretical conclusions underscores what seems to [him] the most important source of confusion in all of this, which is the assump- tion that the purpose of collecting data in the laboratory is to predict real-life behavior in the real world. Of course, there are times when that is what we are trying to do, and there are times when it is not. When it is, then the problem of EV confronts us, full force. When it is not, then the problem of EV is either meaningless or trivial, and a misplaced preoccupation with it can seri- ously distort our evaluation of research . . . But even where findings cannot possibly generalize and are not supposed to, they can contribute to an understanding of the processes going on. (381, 382, emphasis in original)

This study was not intended to replicate a real-world circumstance. Rather, an understanding of the under- lying process of partner selection and the theoretical conclusions that this understanding produces were the objectives.

The design of this study utilized an artificial sce- nario to examine theories. While a single leader could be the sole decision-maker regarding the selection of collaborative partners, this is undoubtedly not always the case. In the real world, this decision could be made in a number of ways, including group consensus, by mandate, and so on. In fact, the home agency may send a representative who is not selected by a member of the collaborative at all. In any case, the actual choice mechanism was not intended to be studied in this arti- cle. The findings in this article do not speak to circum- stances where the partnership decision was made at the level of the home organization. Instead, this research sought to examine how individuals make trade-offs between potential partner characteristics. Although the design of this study did randomly assign partici- pants to conditions, the sample used in this study was not random, and it was not intended to be so. While random samples are critical in survey research where the objective is to extrapolate the findings to the population from which the sample was drawn, rep- resentativeness is not always an important concern in experimental work where theoretical contributions are the goal (Mook 1983).

Methods Two-tailed t-tests were used to examine the pairwise comparisons and uncover the trade-offs that some- one serving in the role of a collaborative leader makes when selecting a potential partner. Since sample inde- pendence is assumed when conducting t-tests, and a respondent could have answered both of the questions

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 9

included in some of the between-experiment compari- sons of interest, any respondent that answered both questions was dropped from the analysis of that com- parison. Also, t-tests assume homogeneity of variance. For comparison where this assumption did not hold, Welch’s t-tests were performed.

The analysis of the 44 planned comparisons in this study could be problematic because the probability of a Type I error is compounded when repeated tests are conducted. However, there is some controversy in the academic community regarding whether to adjust the critical alpha level against which the test statistics are compared when determining whether to reject the null hypothesis that there is no difference between the two group means. For example, O’Keefe (2003) argues that “adjusting the alpha level because of the number of tests conducted in a given study has no principled basis, commits one to absurd beliefs and policies, and

reduces statistical power. The practice of requiring or employing such adjustments should be abandoned” (431). Whereas others, fearful of the increased likeli- hood of making a Type I error, argue that alpha level adjustments must be made so as not to capitalize on chance (Rosenthal and Rosnow 1984). The purpose of this article is not to settle this methodological issue. Therefore, both the unadjusted and Bonferroni–Holm (Holm 1979) adjusted results are shown and discussed.

The Bonferroni–Holm (Holm 1979) correction is a common approach for adjusting the alpha level due to a large number of comparisons. This approach requires that the p-values of all the t-tests be ordered sequen- tially, from smallest to largest. The first p-value is com- pared to a critical value equal to α/c, where c is the number of comparisons. For example, using the critical alpha value of α = 0.05 and 44 comparisons, the first (smallest) p-value is compared to αcrit = 0.00114. If the

Table 3. Respondent Demographics

Demographic na Percentage Mean SD Range

Gender Male 135 61.9 Female 83 38.1 Age (years) 25–34 50 22.9 35–44 59 27.1 45–54 42 19.3 55–64 38 17.4 ≥65 29 13.3 Years since MPA graduation 217 16.4 14.0 0.0–53.0 Years worked in current position 126 4.7 4.7 0.0–25.2 Years worked for current organization 128 7.9 7.6 0.3–40.0 Years worked in the public sectorb 87 17.6 11.4 1.8–46.2 Years worked in the nonprofit sectorb 24 11.1 12.5 0.5–42.3 Years worked in the private sectorb 39 6.2 6.7 0.5–40.0 Team player stylec,d

Contributor 49 22.8 Collaborator 40 18.6 Communicator 32 14.9 Challenger 29 13.5 Contributor/challenger 12 5.6 Collaborator/challenger 12 5.6 Collaborator/communicator 11 5.1 Contributor/communicator 8 3.7 Communicator/challenger 7 3.3 Contributor/collaborator 4 1.9 Collaborator/communicator/challenger 4 1.9 Contributor/collaborator/communicator 3 1.4 Contributor/collaborator/communicator/challenger 2 0.9 Contributor/collaborator/challenger 1 0.5 Contributor/communicator/challenger 1 0.5

aA total of 229 respondents completed at least one question on the survey. bOnly includes those who reported having experience in the given sector. cOf the 229 respondents, 215 individuals answered a sufficient number of questions to determine their team player style. d Per Parker (2008), if the two highest style scores were within three points of each other, the individual’s primary style was considered to be both.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx10

first p-value exceeds that αcrit, the process is stopped and all subsequent comparisons are deemed to be sta- tistically insignificant. However, if the first (smallest) p-value is found to be statistically significant, the second smallest p-value is compared to a critical value equal to α/(c − 1) [αcrit = 0.05/(44 − 1) = 0.00116]. If that com- parison is found to be statistically significant, the third smallest p-value is compared to a critical value equal to α/(c − 2) [αcrit = 0.05/(44 − 2) = 0.00119], and so on. This process is continued until the p-value of one of the t-tests exceeds the Bonferroni–Holm adjusted αcrit values.

Results and Discussion

The results from this experiment demonstrate that the potential partner’s personality matters a great deal when selecting someone with whom to collaborate. As seen in table 5, people generally prefer to work with contribu- tors and collaborators more than with communicators. Challengers were reported as the least desirable.

Contributors and collaborators are more focused on getting things done, while communicators and challengers are more concerned with the collabora- tive process. Therefore, the pattern of preference for contributors and collaborators seems to indicate that collaborative leaders prefer to work with “doers.” As mentioned previously, collaboration is hard work. Understanding this, collaborative leaders seem to grav- itate toward selecting partners that are either focused on task execution (contributors) or goal accomplish- ment (collaborators). Since the purpose of collabora- tion is to accomplish something that the individual entities could not realize alone, someone that needs to

collaborate will want to work with an individual per- ceived as directly contributing to task completion or goal attainment.

The potential problem with the strong desire for “doers” is that the collaborative team can become unbalanced since the lack of any one personality style can be detrimental. As Parker (2008) argues, properly functioning teams should include a balance of team player styles because “the effective team is equally concerned with getting the job done and how the job gets done” (68). The lack of a communicator can cause a failure to generate and/or maintain the “ties that bind,” resulting in a poor internal climate as partners are more focused on getting the work done than on working together, which could be dangerous for the collaborative’s viability. While the partners presumably joined the collaborative to work together in order to reach their goals, their continued participation in the effort is likely influenced by their feelings of inclusion in the actions and direction of the group. If collabora- tive partners do not feel included and cannot get along well enough to work together effectively, there may be a failure at the operational level of the network (Mandell and Keast 2008).

The absence of a challenger may be even more detri- mental. Challengers often have the reputation of being difficult to deal with because they challenge the status quo and the group to be its best. In so doing, they may make the other group members feel uncomfortable and can come across as demanding or disagreeable. Studies have shown that agreeableness is positively associated with group performance (Bradley et  al. 2013), so if a challenger goes too far, the network’s effectiveness can suffer. However, by not including challengers, the

Table 4. Respondent Responses to Survey Questions

Questiona Mean

Percentage (%) of Respondents that:

Strongly Agree Agree

Neither Agree nor Disagree Disagree

Strongly Disagree

Collaboration is: beneficial because it brings individuals with

different skills together 4.37 46.0 49.1 1.8 2.3 0.9

an important strategy for addressing complex problems

4.36 44.6 50.0 3.2 1.4 0.9

beneficial because it brings people with different perspectives together

4.36 45.5 48.7 3.2 1.4 1.4

valued by me 4.33 44.6 47.8 5.4 0.9 1.4 a flexible problem solving approach 4.05 23.4 61.3 12.2 3.2 0.0 something I am good at 4.04 24.3 58.1 15.3 1.4 0.9 valued by my organization 3.98 32.9 40.5 18.9 6.8 0.9 not beneficial because no one is in charge 1.86 0.6 4.1 8.1 55.4 32.0 a waste of time 1.51 0.9 0.9 4.1 36.5 57.7

a222 respondents answered the questions in this section of the survey.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 11

collaborative may open itself up to the problem of groupthink, or what Freshwater et al. (2014) refer to as “collaboration dysfunctional consonance,” which occurs when individuals “uncritically accept the sta- tus quo, and work harmoniously together to ensure that the dominant practices are perpetuated” (63). If no one is there to question the decisions being made, collaborators may deny or suppress the tensions that “stimulate critical feedback to counter groupthink” (Sundaramurthy and Lewis 2003, 407). This can exacerbate “faulty attributions, threat rigidity, and escalating commitment to a failing course of action, eventually resulting in failure” (402).

While the potential partner’s personality is impor- tant, it is not the sole factor in the selection decision. The results of this study show that as other variables are considered, as they must be in any selection deci- sion, the strict preference for contributors and collabo- rators becomes less clear. In addition to the potential partner’s personality, the collaborative leader’s person- ality is also important. Past research has found that

people like to work with people that are like them- selves (Byrne 1971; McPherson, Smith-Lovin, and Cook 2001). However, in this study the findings were mixed (table  6). The analysis of the data from this study shows that collaborators, contributors, and chal- lengers did not have a preference for selecting a part- ner like them; the t-tests for cases where there was a personality match compared to cases where there was a mismatch were not found to be significant. On the other hand, communicators did reveal a preference for working with other communicators.

Since a communicator places a great deal of value on ensuring everyone is involved and vested in the group, it follows that he or she would want others to focus on the collaborative process as well. However, too many communicators could significantly hamper the efforts of the collaborative. In his discussion of ineffective communicators, Parker (2008) identifies a number of ways that the role of a communicator could be taken too far. First, having too many communicators in the group could result in a collaborative that focuses on

Table 5. Results of t-Tests Sorted by Aggregated Team Player Stylea

Group 1 Group 2 p-Value Traditional Significance

Holm Adjusted Significance

Contributor Challenger <.0001 *** *** (75.0) (65.1) Collaborator Challenger <.0001 *** *** (72.7) (65.1) Contributor Communicator <.0001 *** *** (74.9) (68.5) Collaborator Communicator .0009 *** ** (72.7) (68.4) Communicator Challenger 0.0371 ** (68.3) (65.1) Contributor Collaborator 0.1252 (74.8) (72.7)

aMean in parentheses. *p < .05, **p < .01, ***p < .001.

Table 6. Results of t-Tests Conducted on the Match Between the Respondent’s and the Prospective Partner’s Team Player Stylea

Resource Group 1 Group 2 p-Value Traditional

Significance2

Holm Adjusted Significance2

Communicator Match Mismatch 0.0246 ** (73.3) (66.6)

Collaborator Match Mismatch 0.1083 (75.7) (71.7)

Contributor Match Mismatch 0.1457 (76.9) (73.0)

Challenger Match Mismatch 0.5811 (65.8) (64.1)

aMean in parentheses. *p < .05, **p < .01, ***p < .001.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx12

the process so much that the means become the end. This could be exacerbated by the second source of inef- fectiveness: the avoidance or suppression of dissenting views. In the absence of a voiced dissenting opinion, the network might not only experience the groupthink problems discussed earlier, but it also might not have a mechanism to balance the communicator-driven focus on process with a more task- or goal-oriented approach. This lack of balance is the third source of ineffectiveness, which could result in the aliena- tion of the task- and goal-oriented individuals on the team. A  challenger may withdraw from the network as their well-intentioned attempts to play the devil’s advocate are rebuffed. Contributors and collaborators may renege on their commitment to the collaborative because they feel that nothing is being accomplished. Thus, too many communicators may cause the one thing that a communicator does not want to have hap- pen: conflict resulting in alienation.

The interactions between the potential partner’s per- sonality, their familiarity with the collaborative leader, and the resources they bring to the table revealed some interesting patterns. When examining the probability of collaboration based on the collaborative leader’s familiarity with the potential partner, no statistically significant relationships were identified (table 7). There are a number of potential explanations for this non- finding. First is that familiarity does not matter. A col- laborative partner needs to bring a new member into the group to fill a certain role, and the transaction costs associated with an unfamiliar versus a familiar poten- tial partner are not of significant concern. Thus, per- sonality, which is constant across these comparisons, may be the driver of the partnership decision rather than familiarity. An alternative explanation could be that the wording of the potential partner vignettes was not able to sufficiently replicate the emotional connec- tion between the respondent and a familiar potential partner. A  third potential explanation is that there is an interaction effect between familiarity and the

potential partner’s other characteristics. In this study, the familiarity of a potential partner was varied while the resource, which was money, was held constant. Perhaps a potential partner with money is so appealing that respondents do not care whether that person is familiar or not. Money may be the driver of the deci- sion and perhaps, had a different resource been used in these questions, different findings could have emerged. Future work could examine this issue further.

Significant differences, however, were discovered when analyzing the interactions between resources and personality. When the team player style was held constant and the resource was varied (table 8), it was found that respondents preferred to partner with con- tributors with financial or informational resources more than with those having reputational resources. Additionally, collaborators who could contribute mon- etarily were preferred over those whose impact on the network was related to their prestige and reputation.

An explanation for this could be that contribu- tors are task oriented and tasks most often take either money or information to complete. Hackman (2011) identifies “access to the information a team needs to accomplish its work” and access to “ample material resources” (116–17), which would include financial resources, as two factors that are important predictors of team performance. In order for these informational and financial resources to be beneficial and improve performance, they must be used. A contributor, par- ticularly one who brings these resources to the table, would be particularly focused on using them to accom- plish the work of the collaborative. Reputational resources, on the other hand, would be of less inter- est when working with a contributor because the ben- efits of such resources are less directly related to task accomplishment.

The results for collaborators are somewhat less clear. Collaborators with money are preferred over those with reputational resources. However, it is some- what surprising, given the patterns uncovered in the

Table 7. Results of t-Tests Sorted by Familiaritya

Group 1 Group 2 p-Value Traditional Significance

Holm Adjusted Significance

Unfamiliar contributor Familiar contributor 0.2415 (78.1) (74.2) Familiar challenger Unfamiliar challenger 0.2805 (68.7) (64.8) Unfamiliar communicator Familiar communicator 0.2999 (70.8) (66.7) Unfamiliar collaborator Familiar collaborator 0.5611 (75.8) (74.0)

aMean in parentheses. *p < .05, **p < .01, ***p < .001.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 13

aforementioned findings, that no other statistically sig- nificant results were revealed for the other compari- sons involving collaborators. While the comparison of means shows that respondents were more likely to partner with collaborators with money than with information and with collaborators with information than with reputational resources, these relationships were not statistically significant. Perhaps collaborative leaders see money, more so than reputational effects, as directly supporting attainment of the network’s mis- sion and goals. Collaborators are, after all, focused on the big picture, the achievement of which will take the long-term, strategic utilization of financial resources. However, information and reputation are also impor- tant resources that can be successfully used by a collaborator.

Information can be, and likely is, used to achieve the collaborator’s strategic purposes. The accumulation and sharing of information are among the most critical activities of an organization (Scott and Davis 2007). This may be more the case for networks than for tra- ditional, hierarchical organizations since barriers to information sharing often increase in interorganiza- tional structures (Agranoff 2012). Information, such as the “interpretation of aggregate trends or patterns, and gossip or ‘intelligence’ about political developments”

(Bardach 1998), is integral to the network’s ability to adapt to changes in the external environment and con- tinue toward mission accomplishment. Even reputa- tional resources, while having a lower mean likelihood than either money or information and being found to be significantly less desirable than monetary resources, were found to be similarly important as compared to information. Mission success takes long-term effort, which will necessitate the support of external parties for the network. A visionary partner, such as a collabo- rator, who can attract external support for their grand plans, could be quite useful.

The comparisons between resources among com- municators and among challengers were not found to be statistically significant. Communicators and chal- lengers are process-oriented individuals. While the resources they bring to the table would be useful, the utility of individuals with either of these two process- oriented personalities is not dependent upon that resource. Communicators will still be able to engen- der a feeling of unity and challengers can still play the devil’s advocate regardless of the resources they bring to the network.

Some have suggested that relationship-oriented individuals (i.e., communicators) are more desirable in circumstances where information is being shared

Table 8. Results of t-Tests Sorted by Team Player Stylea

Style Group 1 Group 2 p-Value Traditional Significance

Holm Adjusted Significance

Contributor Money Reputation 0.0155 ** (78.1) (68.5)

Information Reputation 0.0279 ** (77.4) (68.5) Money Information 0.8254 (78.1) (77.4)

Communicator Money Information 0.3082 (70.8) (66.8)

Reputation Information 0.5680 (69.1) (66.8) Money Reputation 0.6496 (70.8) (69.1)

Collaborator Money Reputation 0.0393 ** (75.8) (68.4) Money Information 0.1544 (75.8) (71.7)

Information Reputation 0.3615 (71.7) (68.4)

Challenger Money Information 0.5400 (64.8) (62.4)

Reputation Information 0.6464 (64.4) (62.4) Money Reputation 0.9060 (64.8) (64.4)

aMean in parentheses. *p < .05, **p < .01, ***p < .001.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx14

because they work to establish a trust and goodwill among collaborative partners (Gratton and Erickson 2007). However, a significant interaction between com- municators and information resources was not found. Instead, when information as a resource was held con- stant, respondents preferred contributors more than either challengers or communicators and preferred col- laborators more than challengers.

Analysis was also conducted to compare the impact of personality when the resource was held constant (Table 9). For those offering monetary support to the network, prospective partners who had a contributor or collaborator personality were preferred over com- municator and challenger personalities. This further

supports the notion that the mix of monetary resources and the “doer” personalities is a particularly attrac- tive combination for collaborative leaders. A  similar pattern was uncovered for perspective partners with informational resources. Again, it appears that contrib- utors and collaborators were more enticing potential partners.

The preference for contributors and collaborators did not hold when leaders considered potential part- ners with reputational resources. In this case, only the preference for communicators over challengers was found to be statistically significant. While only signifi- cant at the 0.10 level, this is nevertheless very interest- ing in that it is the only time there is a stated preference

Table 9. Results of t-Tests Sorted by Resourcea

Group 1 Group 2 p-Value Traditional

Significance2

Holm Adjusted Significance2

Money Collaborator Challenger 0.0001 *** *** (77.2) (63.1)

Contributor Challenger 0.0015 *** * (78.5) (65.1)

Collaborator Communicator 0.0813 * (77.3) (70.2)

Contributor Communicator 0.0837 * (77.2) (70.1)

Communicator Challenger 0.1142 (70.3) (63.5)

Collaborator Contributor 0.9750 (75.4) (75.3)

Information Contributor Challenger 0.0003 *** ** (77.4) (62.4)

Contributor Communicator 0.0230 ** (77.8) (67.6)

Collaborator Challenger 0.0534 * (71.6) (64.0)

Contributor Collaborator 0.1090 (78.1) (72.3)

Collaborator Communicator 0.5852 (71.6) (69.4)

Communicator Challenger 0.6260 (65.8) (63.5)

Reputation Communicator Challenger 0.0987 * (72.3) (65.0)

Contributor Challenger 0.1141 (68.5) (64.4)

Communicator Contributor 0.5447 (70.5) (67.8)

Collaborator Challenger 0.6364 (69.1) (66.9)

Contributor Collaborator 0.7234 (69.7) (68.1)

Collaborator Communicator 0.9737 (70.7) (70.5)

aMean in parentheses. *p < .05, **p < .01, ***p < .001.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 15

for communicators. This speaks to the role of commu- nicators in the group. Communicators want everyone to feel included and part of the enterprise. This desire to reach out to others may extend beyond the members of the collaborative. As reputational resources are used to build social capital and engender support for the network, an individual with a welcoming personality and who builds consensus may be able to optimally use reputational resources. Conversely, the challenger, who can alienate others by taking the devil’s advocate role too far, may be an undesirable personality type to build external support for the network.

Conclusion

This article focused on using an experimental design to better understand the factors that help determine how a collaborative leader confronts the trade-offs between personality type, resource access, and familiarity when picking a potential partner. The findings of this pre- liminary study show that it is actually the combina- tion of factors, especially personality and resources that interact to make a potential partner more or less desirable. Respondents consistently preferred partners with accomplishment-oriented personalities (contribu- tors and collaborators) and those with resources that contribute to getting things done (financial and infor- mational resources). While personality appears to be the biggest single factor in the partnership decision, its impact is moderated by the resource the poten- tial partner brings to the table. An explanation for this phenomenon is that networks are under a great deal of pressure to get things done and to ensure their sustainability. This is best achieved by demonstrating their value through task accomplishment and mission success.

There are a number of theoretical and practical lessons that can be learned from the results of this study. First, the findings highlight the importance of the potential partner’s personality in the partner selec- tion process. The impact of personality has not, to my knowledge, been investigated previously in the con- text of collaboration. The findings in this article, along with findings that indicate that a group’s performance is influenced by the group’s personality (Kramer, Bhave, and Johnson 2014), suggest that incorporat- ing personality factors into collaboration management research may prove fruitful. Secondly, this study makes a contribution via its uncovering of the importance of the interaction between a potential partner’s per- sonality and the resources he or she can bring to the table. The objective of the article was to identify the trade-offs that collaborative leaders make when select- ing a partner. What was discovered is that it is not a circumstance where one factor (personality, resources,

or familiarity) trumps the others. It is not solely the potential partner’s personality, as suggested by the personality theorists, or the resources, as suggested by resource dependency theory, that matters. Rather, it is the interaction between these two factors that best pre- dicts collaborative partnering.

Future studies can build upon this study and address some of its limitations. In terms of future research, fur- ther investigation of the impact of familiarity would be helpful in extending our understanding of collabo- rative partner selection. Familiarity, as operationalized in this study, did not prove to play a pivotal role in the selection process. However, as noted previously, this could be because the vignette did not adequately capture the feeling of familiarity or because familiarity truly does not play a significant role in the collabo- rative partner selection decision. It is also plausible that familiarity was not found to play a significant role in the partnership decision in this study because not all familiar partners are the same. Therefore, the quality of past interactions could be addressed. In her study, Malatesta (2012) found that a relationship his- tory can be good or bad and that the quality of the relationship is more important than the length of the relationship. She argues that the transaction cost gains resulting from partner familiarity cannot overcome the increased transaction costs resulting from partner mis- trust. Thus, studies that look at both transaction cost theory and social capital theory could be an important next step in this stream of research.

As mentioned above, many of the important con- textual aspects of collaboration that affect the part- nership decision were purposefully not studied in this article. An important next step in this stream of research would be to explore these issues. For example, the findings of this study paired with the findings from those who have explored the impact of institutional constraints such as sectoral constraints (e.g., Herranz 2008), mandated/voluntary collaboration (e.g., Saz- Carranza et al, 2015), and modes of governance (e.g., Provan and Kenis 2008) would be of great value to the collaborative management literature. Further, in practice, collaborative partners are sometimes chosen by methods other than direct selection by collaborative leaders. Future studies could examine (1) how man- agers of home organizations decide whom to send to the collaborative as their representative, (2) how stake- holders of the collaborative view partner selection and network composition, and (3) how collaborators view the characteristics of fellow collaborators whom they did not choose for themselves.

Also, it would be interesting to determine if there are other important interactions. For example, studies that address the impact of gender, of the sector of employ- ment, and of other resource types (e.g., technology,

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx16

staffing, facilities) could further this work. A  study investigating the role of gender on the partner selection would be particularly intriguing. Although the data in this study do not allow for an in-depth study of the role that gender plays in this decision, the interactions between the potential partner’s gender and personal- ity with the collaborative leader’s gender would be of great value to the field. The study of the impact of sectoral difference on partner selection would also be very interesting. Perhaps the sector in which the col- laborative leader works, the sector in which the pro- spective partner works, or the interaction between the two could significantly impact the partnering decision. Further, collaboratives are often intentionally designed to represent the diverse interests of stakeholders. While the diversity of the potential partners’ perspec- tives was not included in this study, the investigation of the importance of perspective representativeness in the partnership selection decision would provide an important contribution to the field. Finally, since there is evidence that there is a preference for working with people with accomplishment-oriented personalities who have resources that can be directly tied to getting things done, there is a concern that the absence of other personality types or resources may hinder the effective- ness of the collaborative. Thus, a key next step would be to investigate differences in effectiveness between balanced teams that have all personality types and resources represented and unbalanced teams that are missing certain personalities or resources. Finally, this study focused on the trade-offs a single, collaborative leader makes in choosing a potential partner. However, there are times where the partnering decision is made by multiple individuals. In future studies, it would be interesting to examine how the group decision-making process operates given the norms, ground rules, and institutionalized frameworks that often exist in col- laborative settings.

This article also offers some practical implications for those working in collaborative settings. The find- ings of this study provide theoretic insights into the partnership decision that can be generally applied to collaborative arrangements in which partner selec- tion is determined from within. Particularly, it can help collaborators recognize their own partnership preferences, understand the preferences of others, and beware of the impact that these preferences may have on the collaborative. For example, the partner selec- tion decision logic seems to make it more likely that the collaborative will suffer from groupthink. Because few people want to work with challengers, and com- municators want to work with other communicators, there may be a tendency for the group composition to become relatively homogenous. Collaborative leaders must be aware of this potential problem. Ideally group

membership would be balanced between the four team player styles, since each plays an important role in keeping the group on task, focused on their goals, working well together, and striving for the best pos- sible outcome. When the group is unbalanced, one or more of these important perspectives is lost, and the group may not function optimally. For example, com- municators are important for group dynamics, but a network predominately composed of process-oriented individuals may be harmoniously unproductive. And although challengers, if they go to the extreme, can have a deleterious effect on group dynamics and effec- tiveness, the group can become detrimentally single- minded without someone who is willing to question the prevailing wisdom.

The findings from this study shed light on how man- agers choose those with whom they want to collabo- rate. This decision is of the utmost importance to the success of the collaborative, because, as Collins (2001, 44)  commented, the leader must “first get the right people on the bus . . . before [he or she can] figure out where to drive it.”

Funding

This research was partially supported by a grant from CH2M Hill through a program at the University of Kansas' School of Public Affairs and Administration.

References Agranoff, Robert. 2012. Collaborating to manage: A  primer for the public

sector. Edited by Beryl A. Radin. Public Management and Change Series. Washington, DC: Georgetown Univ. Press.

Agranoff, Robert, and Michael McGuire. 2001. Big questions in public net- work management research. Journal of Public Administration Research and Theory 11:295–326.

Agranoff, Robert, and Michael McGuire. 2003. Collaborative public manage- ment: New strategies for local governments. Washington, DC: Georgetown Univ. Press.

Ansell, Chris, and Alison  Gash. 2008. Collaborative governance in theory and practice. Journal of Public Administration Research and Theory 18:543–71.

Bardach, Eugene. 1998. Getting agencies to work together: The practice and theory of managerial craftsmanship. Washington, DC: Brookings Institution Press.

Bell, Suzanne T. 2007. Deep-level composition variables as predictors of team performance: A meta-analysis. Journal of Applied Psychology 92:595–615.

Berardo, Ramiro, and John T. Scholz. 2010. Self-organizing policy networks: Risk, partner selection, and cooperation in estuaries. American Journal of Political Science 54:632–49.

Bradley, Bret H., John E.  Baur, Christopher G.  Banford, and Bennett E.  Postlethwaite. 2013. Team players and collective performance: How agreeableness affects team performance over time. Small Group Research 44:680–711.

Bryson, John M., Barbara C. Crosby, and Melissa Middleton Stone. 2006. The design and implementation of cross-sector collaborations: Propositions from the literature. Public Administration Review 66 (Suppl 1):44–55.

Byrne, Donn E. 1971. The attraction paradigm. New York: Academic Press. Calanni, John C., Saba N.  Siddiki, Christopher M.  Weible, and William

D. Leach. 2015. Explaining coordination in collaborative partnerships and

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx 17

clarifying the scope of the belief homophily hypothesis. Journal of Public Administration Research and Theory 25:901–27.

Cattell, Raymond B. 1951. Determining syntality dimensions as a basis for morale and leadership measurement. In Groups, leadership, and men, ed. Harold Steere Guetzkow, 16–27. Pittsburgh: Carnegie Press.

Christensen, Robert K., Steven W.  Whiting, Tobin  Im, Eunju  Rho, Justin M. Stritch, and Jungho Park. 2013. Public service motivation, task, and non-task behavior: A performance appraisal experiment with Korean MPA and MBA students. International Public Management Journal 16:28–52.

Cigler, Beverly A. 2001. Multiorganization, multisector, and multicommunity organizations: Setting the research agenda. In Getting results through col- laboration: Networks and network structures for collaboration, ed. Myrna P. Mandell, 71–85. Westport, CT: Quorum Books.

Clay-Williams, Robyn, and Jeffrey Braithwaite. 2015. Reframing implementa- tion as an organisational behaviour problem: Inside a teamwork improve- ment intervention. Journal of Health Organization and Management 29:670–83.

Cohen, Susan G., Gerald E. Ledford, and Gretchen M. Spreitzer. 1996. A pre- dictive model of self-managing work team effectiveness. Human Relations 49:643–76.

Collins, Jim. 2001. Good to great: Why some companies make the leap ... and others don’t. New York: HarperCollins Publishers, Inc.

Crosby, Barbara C., and John M. Bryson. 2005. Leadership for the common good: Tackling public problems in a shared-power world, 2nd ed. San Francisco: Jossey-Bass.

Daft, Richard L. 2008. Management, 8th ed. Mason, OH: South-Western. DiMaggio, Paul J., and Walter W.  Powell. 1983. The iron cage revisited:

Institutional isomorphism and collective rationality in organizational fields. American Sociological Review 48:147–160.

Eglene, Ophelia, Sharon S. Dawes, and Carrie A. Schneider. 2007. Authority and leadership patterns in public sector knowledge networks. The American Review of Public Administration 37:91–113.

Emerson, Kirk, Tina  Nabatchi, and Stephen  Balogh. 2012. An integrative framework for collaborative governance. Journal of Public Administration Research and Theory 22:1–29.

Feiock, Richard C., In Won Lee, and Hyung Jun Park. 2012. Administrators’ and elected officials’ collaboration networks: Selecting partners to reduce risk in economic development. Public Administration Review 72:S58–68.

Feyerherm, Ann E. 1994. Leadership in collaboration: A  longitudinal study of two interorganizational rule-making groups. The Leadership Quarterly 5:253–70.

Freshwater, Dawn, Jane Cahill, and Chris Essen. 2014. Discourses of collabo- rative failure: Identity, role and discourse in an interdisciplinary world. Nursing Inquiry 21:59–68.

Friedman, Thomas L. 2009. Hot, flat, and crowded: Why we need a green revolution—and how it can renew America. Release 2.0, updated and expanded edition. New York: Picador.

Graddy, Elizabeth A., and Bin  Chen. 2009. Partner selection and the effec- tiveness of interorganizational collaborations. In The collaborative public manager: New ideas for the twenty-first century, eds. Rosemary O’Leary and Lisa B. Bingham, 53–69. Washington, DC: Georgetown Univ. Press.

Gratton, Lynda, and Tamara J. Erickson. 2007. Eight ways to build collabora- tive teams. Harvard Business Review 85:100–9.

Gray, Barbara, and Donna J. Wood. 1991. Collaborative alliances: Moving from practice to theory. Journal of Applied Behavioral Science 27:3–22.

Gulati, Ranjay, and Maxim  Sytch. 2008. Does familiarity breed trust? Revisiting the antecedents of trust. Managerial and Decision Economics 29 (2–3):165–90.

Hackman, J. Richard. 2011. Collaborative intelligence: Using teams to solve hard problems. San Francisco: Berret-Keohler Publishers.

Herranz, Joaquín Jr. 2008. The multisectoral trilemma of network manage- ment. Journal of Public Administration Research and Theory 18:1–31.

Hinds, Pamela J., Kathleen M. Carley, David Krackhardt, and Doug Wholey. 2000. Choosing work group members: Balancing similarity, competence, and familiarity. Organizational Behavior and Human Decision Processes 81:226–51.

Holm, Sture. 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics 6:65–70.

Hu, Qian, Sana Khosa, and Naim Kapucu. 2015. The intellectual structure of empirical network research in public administration. Journal of Public Administration Research and Theory 26:593–612.

Human, Sherrie E., and Keith G. Provan. 2000. Legitimacy building in the evo- lution of small-firm multilateral networks: A comparative study of success and demise. Administrative Science Quarterly 45:327–365.

Huxham, Chris, and Siv Vangen. 2005. Managing to collaborate: The theory and practice of collaborative advantage. New York: Routledge.

Klijn, Erik-Hans, and Chris Skelcher. 2007. Democracy and governance net- works: compatible or not? Public Administration 85:587–608.

Koljatic, Mladen, Mónica  Silva, and Eduardo Valenzuela. 2006. The devel- opment of cross-sector collaborations in a social context of low trust. In Creating a culture of collaboration, ed. Sandy Schuman, 55–67. San Francisco: Jossey-Bass.

Kramer, Amit, Devasheesh P. Bhave, and Tiffany D. Johnson. 2014. Personality and group performance: The importance of personality composition and work tasks. Personality and Individual Differences 58:132–37.

LePine, Jeffery A., Brooke R.  Buckman, Eean R.  Crawford, and Jessica R. Methot. 2011. A review of research on personality in teams: Accounting for pathways spanning levels of theory and analysis. Human Resource Management Review 21:311–30.

Levitt, Steven D., and John A. List. 2007. What do laboratory experiments measuring social preferences reveal about the real world? The Journal of Economic Perspectives 21:153–74.

Linden, Russell M. 2010. Leading across boundaries: Creating collaborative agencies in a networked world. San Francisco: Jossey-Bass.

Malatesta, Deanna. 2012. The link between information and bargain- ing efficiency. Journal of Public Administration Research and Theory 22:527–51.

Malatesta, Deanna, and Craig R. Smith. 2014. Lessons from resource depend- ence theory for contemporary public and nonprofit management. Public Administration Review 74:14–25.

Mandell, Myrna P., and Robyn Keast. 2008. Evaluating the effectiveness of interorganizational relations through networks. Public Management Review 10:715–31.

Margetts, Helen Z. 2011. Experiments for public management research. Public Management Review 13:189–208.

McGuire, Michael. 2006. Collaborative public management: Assessing what we know and how we know it. Public Administration Review 66 (Suppl 1):33–43.

McGuire, Michael, and Chris  Silvia. 2009. Does leadership in networks matter? Examining the effect of leadership behaviors on managers’ per- ceptions of network effectiveness. Public Performance & Management Review 33:34–62.

McGuire, Michael, and Chris Silvia. 2010. The effect of problem severity, man- agerial and organizational capacity, and agency structure on intergovern- mental collaboration: Evidence from local emergency management. Public Administration Review 70:279–88.

McPherson, Miller, Lynn Smith-Lovin, and James M. Cook. 2001. Birds of a feather: Homophily in social networks. Annual Review of Sociology 27:415–44.

Mook, Douglas G. 1983. In defense of external invalidity. American Psychologist 38:379–87.

O’Keefe, Daniel J. 2003. Colloquy: Should familywise alpha be adjusted? Against familywise alpha adjustment. Human Communication Research 29:431–47.

O’Leary, Rosemary, and Nidhi Vij. 2012. Collaborative public management: Where have we been and where are we going? The American Review of Public Administration 42:507–22.

O’Toole, Laurence J., Jr. 1997. Treating networks seriously: Practical and research-based agendas in public administration. Public Administration Review 57:45–52.

O’Toole, Laurence J., Jr. 2014. Networks and networking: The public adminis- trative agenda. Public Administration Review 75:361–71.

Journal of Public Administration Research and Theory, 2016, Vol. xx, No. xx18

Parker, Glenn M. 2008. Team players and teamwork: New strategies for devel- oping successful collaboration. San Francisco: John Wiley & Sons.

Pfeffer, Jeffrey, and Gerald R. Salancik. 2003. The external control of organi- zations: A  resource dependence perspective. Stanford Business Classics. Stanford, CA: Stanford Business Books.

Powell, Walter W. 1990. Neither market nor hierarchy: Network forms of organization. In Research in organizational behavior, ed. Barry M. Staw and Larry L. Cummings, 295–336. Greenwich, CT: JAI Press.

Provan, Keith G., and Patrick Kenis. 2008. Modes of network governance: Structure, management, and effectiveness. Journal of Public Administration Research and Theory 18:229–52.

Romzek, Barbara, Kelly  LeRoux, Jocelyn  Johnston, Robin J.  Kempf, and Jaclyn Schede Piatak. 2014. Informal accountability in multisector service delivery collaborations. Journal of Public Administration Research and Theory 24:813–42.

Rosenthal, Robert, and Ralph L.  Rosnow. 1984. Essentials of behavioral research: Methods and data analysis. New York: McGraw-Hill.

Ryu, Sangyub. 2014. Networking partner selection and its impact on the perceived success of collaboration. Public Performance & Management Review 37:632–57.

Sawyer, Keith. 2007. Group genius: The creative power of collaboration. New York: Basic Books.

Saz-Carranza, Angel, Susanna Salvador  Iborra, and Adrià  Albareda. 2015. The power dynamics of mandated network administrative organizations. Public Administration Review 76:449–62.

Scott, Richard W., and Gerald Fredrick Davis. 2007. Organizations and organ- izing: Rational, natural, and open systems perspectives. Upper Saddle River, NJ: Pearson Prentice Hall.

Smith, Craig R. 2009. Institutional determinants of collaboration: An empirical study of county open-space protection. Journal of Public Administration Research and Theory 19:1–21.

Sundaramurthy, Chamu, and Marianne  Lewis. 2003. Control and collabora- tion: Paradoxes of governance. The Academy of Management Review 28:397–415.

Vincentini, Francesca, and Paolo  Boccardelli. 2014. Team composition and project-based organizations: New perspectives for human resource man- agement. In Management, valuation, and risk for human capital and human assets, ed. Meir Russ, 37–58. New York: Palgrave Macmillan.

Wernerfelt, Birger. 1984. A resource-based view of the firm. Strategic Management Journal 5:171–80.

Whiting, Steven W., Philip M.  Podsakoff, and Jason R.  Pierce. 2008. Effects of task performance, helping, voice, and organizational loy- alty on performance appraisal ratings. Journal of Applied Psychology 93:125–39.