Building Collaboration: Examining the Relationship between Collaborative Processes and Activities

Abstract

This study investigates if the collaborative process differs among a group of public programs involved in varying levels of interorganizational activities. Thomson and Perry (2006) suggest five process dimensions underlie collaboration: governance, administration, norms of trust, mutuality, and organizational autonomy. While these dimensions are clearly unique, it is unclear if any of these dimensions are necessary or sufficient for varying degrees of interorganizational involvement. Inventorying the interorganizational activities of pairs of government-funded preschools as ranging along a continuum of no relationship, cooperation, coordination, and collaboration, I conduct a qualitative comparative analysis (QCA) to assess the relationship between collaborative processes and activities. The findings suggest that the collaborative processes dimensions differ depending on the level of involvement. The QCA results also reveal substitutable combinations of process dimensions that underlie respective degrees of interorganizational involvement, offering insight to public managers about different skill sets they can focus on when managing interorganizational activities.

Introduction

Public agencies are asked to engage interagency and cross-sector collaborations with increasing frequency. While years of practice and observation certainly have increased our understanding of what “interorganizational innovations” are and what they are used for (Mandell and Steelman 2003), scholars agree that there is still much to learn. We are still unraveling how public managers actively engage interorganizational relationships, through negotiating and managing collaborative processes. Broadly defined, collaborative processes are activities and ongoing interaction that provide both structure and meaning to collective action (Ring and Van de Van 1994). Scholars identify these interactive features of collaboration; and although disagreement exists about the composition, most models include the consideration of trust and communication, shared decision-making, shared goals or vision, and power (Ansell and Gash 2008; Bryson, Crosby, and Stone 2006, 2015; Emerson et al. 2012; Thomson and Perry 2006). These qualities separate collaborating from other forms of interactions, such as those housed in market- or hierarchal-based relationships (Powell 1990).

While elucidating features that underlie collaboration, the scholarship does not clarify whether all of the identified dimensions are necessary for varying degrees of interorganizational relationships. Are some process dimensions more important than others depending upon more or less involvement between partners? In their 2015 review of the field, Bryson, Crosby, and Stone voice that while many explanatory frameworks of collaboration exist, most fail to consider how elements of collaboration may vary based upon the degree of collaborative involvement. This study examines this question by assessing how collaborative processes differ based upon the degree of interorganizational activity between public program partners.

To frame this study, I review literature about variation in interorganizational forms and discuss the collaborative process dimensions as described by Thomson and Perry (2006). The Thomson and Perry model is an advantageous lens because it provides a parsimonious framework that considers structural, social capital, and organizational autonomy dimensions. This empirically based conceptual model offers a way to tap the beliefs, actions and protocols that comprise the dimensions (Thomson, Perry, and Miller 2007). To examine potential linkages between collaborative processes and degree of interorganizational activity I examine public–public program collaboration (PPPC) between 16 dyadic relationships comprised of administrators of the Head Start and Virginia Preschool Initiative (VPI) programs in five regions in Virginia. I use qualitative comparative analysis (QCA) to examine this question, discuss the findings, and end with a discussion and implications.

Defining PPPC and Variation in Interorganizational Forms

PPPC is two or more publicly funded program service providers interacting, and developing rules and shared norms, in order to address a particular problem domain and to increase public value (Bardach 1998; Wood and Gray 1991). What is contested is whether the entities interacting with one another collaborate by simply engaging with each other or if they collaborate when they achieve (or work toward) mutually identified goals (Bardach 1998; Cook 1977; Imperial 2005; Mandell 2014; McGuire, Agranoff, and Silvia 2011). In this study, I lean towards the latter approach that looks at interorganizational activities as an outcome of collaborative processes. I follow in a tradition of other scholars who acknowledge interaction ranging from more shallow forms compared with deeper forms that may include creating joint processes or new organizational configurations (Alter and Hage 1993; Mattessich, Murray-Close, and Monsey 2001; Selden, Sowa, and Sandfort 2006; Sowa 2008).

Interorganizational and network forms matter in terms of shaping member interactions and reaching collaborative goals (Agranoff 2007; Alter and Hage 1993; Mandell and Steelman 2003; Mattessich, Murray-Close, and Monsey 2001; Selden, Sowa, and Sandfort 2006; Sowa 2008; Thomson and Perry 2006; Thomson, Perry, and Miller 2007). For example, Mandell and Steelman (2003) identify a continuum of interorganizational forms ranging from intermittent commitment to network structure, suggesting that the forms reflect variation in time commitment and the nature of the goal to be accomplished. Sowa (2008) focuses on service delivery collaboration that ranges from shallow to deep integration, arguing that shallow forms include simple financial contracts between agencies and that the deepest forms create value beyond the program clients to the community as a whole. My focus also centers on a full range of interorganizational activities between two publicly funded preschool programs, but differs in drawing attention to the underlying collaborative processes that support these activities. This consideration allows for deeper understanding of how practitioners attempt to accomplish work together and the applicability of identified process dimensions for a full range of interactions.

Mattessich, Murray-Close, and Monsey (2001) offer a simplified continuum of interorganizational relationships from the least to the most integrated. With the focus of this study on dyadic relationships, this continuum provides a useful lens for examining the nature of the interorganizational relationship between Head Start and VPI administrators. Mattessich, Murray-Close, and Monsey (2001) define the degrees along the continuum as cooperation, coordination, and collaboration, respectively. Each step up along the continuum involves greater integration of missions and tasks, risks and rewards, and authority and accountability (p. 61). The Mattessich et al. continuum has been used by other scholars who focus on collaborative service provision, particularly for early childhood education (Selden, Sowa, and Sandfort 2006; Sowa 2008). Whereas the primary focus for those works was collaboration, I consider collaboration to be the most involved form on a full range of interorganizational activities that are worthy of study given their relevance to the working reality of public administrators.

According to Mattessich et al. (2001), cooperation includes informal relationships with minimal risk assumed or reward gained among the partners. Cooperation takes place in the exchange of information. With coordination comes some increased risk and resource exchange among partners; however, each partner still maintains independent authority. Finally, collaboration includes comprehensive planning and often a distinct collaborative governance structure separate from the individual organizations that comprise it. To extend the continuum, I add no relationship to represent situations where two programs reside in the same region, but have little to no interorganizational relationship.

Process Dimensions of Collaboration

Identifying variations in interorganizational forms illuminates myriad ways that public programs work together; equally important is recognizing the underlying dimensions that comprise collaboration. These dimensions help distinguish collaboration from other types of relationships, such as hierarchies or markets (Powell, 1990). Thomson and Perry (2006) and Thomson, Perry, and Miller (2007) identify five distinct dimensions of collaboration. These five dimensions include two structural dimensions, two social capital dimensions, and one organizational dimension.

The governance dimension, a structural dimension, includes serious consideration of collaborative partners, setting up decision-making processes, and participating in group-brainstorming sessions; in other words, this dimension describes the structural framework required for collaboration to occur. The second structural dimension, the administration dimension, includes the nuts and bolts of collaborating, or getting the work done. Thomson et al. (2007) identify members knowing their roles and responsibilities, agreeing on goals, and coordinating efforts to be essential components of the administration dimension. They conceive of structure in line with Bardach’s (1998) assertions that behaviors and processes underlie collaboration. Thus, following Thomson et al.’s (2007) definition, the structural elements of collaboration are less about reformulating or changing organizational boundaries than about setting the stage for ongoing interaction. As the structure develops between collaborative partners, rules and norms develop (McGuire 2006; Wood and Gray 1991).

If governance and administration set the stage for interaction to occur, social capital develops out of the relationships and interactions that do occur between collaborative members. Thomson and Perry (2006) and Thomson et al. (2007) find empirical support for two distinct social capital variables that comprise collaboration. The first dimension is mutuality, which reflects the “capital” part of social capital, including aiding in communication and goal achievement. They find sharing information and resources, and achieving goals better to be aspects of the mutuality dimension. Other scholars also emphasize the importance of information exchange and achieving goals as outcomes of collaborative relationships (Agranoff 2007; Imperial 2005; Isset et al. 2011; Powell 1990). For public networks, the mutuality dimension is vital to increasing public value by working together.

The second social capital dimension that Thomson et al. identify is norms of trust and reciprocity that develop between collaborative partners. These norms develop when partners learn to count on each other and develop commitments, which in turn may act as a social lubricant that enhances mutuality. Thomson’s analysis supports mutuality and norms as unique dimensions, but these two are arguably reciprocal and continually aid in the support or demise of each other. Vangen and Huxham (2003) discuss this occurrence as a “trust-building loop” (p. 8) and Ansell and Gash (2008) enfold these social capital dimensions in their discussion of the collaborative process as an iterative, nonlinear progression of trust building, communication, commitment, and outcomes.

One final dimension that Thomson and Perry (2006) and Thomson, Perry, and Miller (2007) discuss is the organizational dimension, which they label as autonomy. This dimension reflects the tension that occurs between juggling organizational and collaborative identities. When collaboration occurs, organizations must contend with fulfilling both organizational and collaborative goals. Galaskiewicz (1985) explains the tension as organizations wanting to maintain autonomous control while recognizing that they need to collaborate with other organizations for survival. Sedgwick (2016) identifies tensions between balancing both identities for publicly funded preschool programs that simultaneously balance collaborative goals of school readiness for all community children while at the same time protecting narrower program goals, such as safeguarding program enrollment. In particular, this dimension may become heightened when collaborative partners could potentially supplant rather than complement each other in their collaborative arrangement (Bassok 2012).

While these existing frameworks provide insight to different interorganizational forms and the underlying processes that embody them, they lack consideration for the relationship between variation in underlying collaborative processes and variation in degree of interorganizational involvement. It is unclear from the existing literature if the five process dimensions, governance, administration, mutuality, norms of trust, and autonomy are always present in the varying degrees of interorganizational activity, ranging from cooperation to collaboration, or if some combinations suffice for more or lesser involvement. This research attempts to clarify these relationships. First, I briefly discuss the contextual backdrop for this research, the Head Start and the VPI programs.

Head Start and the VPI

Head Start was enacted in 1965 to provide poor children with preschool opportunities that would increase their readiness for public elementary education (Harmon 2004). It grew out of the 1964 Economic Opportunity Act’s (EOA) Community Action Program (CAP) model that envisioned impoverished communities organizing their poor to combat local poverty issues (Vinovskis 2005; Zigler and Muenchow 1992; Zigler and Styfco 2010). Local Head Start grantees directly receive funds from the federal government to operate their programs. Most programs are housed in community action agencies, followed closely by local education authorities (LEAs), private/public nonprofit agencies, and other government agencies. In Virginia, a regional Head Start administrator leads a team of administrators, including educational, community, and mental health specialists. A Head Start region typically includes a combination of cities and/or counties that reside in a geographical area.

Section 22.1–199.1 of the Code of Virginia established the state-funded preschool program that became known as the VPI. The code clearly established its relationship with Head Start, stating that “the General Assembly hereby establishes a grant program to be disbursed by the Department of Education to schools and community-based organizations to provide quality preschool programs for at-risk 4-year-olds who are unserved by Head Start programs and for at-risk 5-year-olds who are not eligible to attend kindergarten” (Code of Virginia, 22.1–199.1, emphasis added). The Virginia General Assembly enacted the program legislation in 1995, which made the first year of funding available in FY96 (Joint Legislative Audit and Review Commission 2007).

From the enactment of VPI, the legislators attempted to clearly articulate a separate program boundary from the existing Head Start program, but practice proved to be more challenging than envisioned. Eligibility for VPI includes locally defined risk resulting in more flexibility and the ability to enroll children from families with incomes above Head Start’s strict income eligibility. In theory, these programs serve distinct populations with little overlap. Over the years, however, tensions between programs have flared at times over issues of “stealing” program enrollees from each other (whether real or perceived). Recognizing that collaborating could smooth enrollment issues and maximize program coverage for both Head Start and state-funded preschool programs, the 2007 Head Start Reauthorization act requires Head Start programs to coordinate with other publicly funded preschools in their regions and to establish Memoranda of Understanding (MOU) between programs. Although these MOUs are required and signify, at least symbolically, a commitment to an interorganizational relationship between Head Start and other publicly funded preschool programs, not all Head Start programs have signed agreements with the other public preschools in their service areas, and there exists wide variation in degree of interorganizational activity (Bassok 2012; personal communication with informants).

Interactions between program administrators of Head Start and the VPI run the gamut along a continuum of interorganizational activities. Ranging from yearly phone calls to simply exchange enrollment lists to working together to provide a blended classroom experience,¹ many programs find ways to cooperate, coordinate, or collaborate. Other examples include sharing buses to transport preschool children, sharing joint professional development opportunities, or reviewing curricula together. Some program administrators review preschool applications annually and jointly decide upon program placement for enrollees. Clearly, solely exchanging enrollment lists falls on the “cooperation” end of the continuum while creating a classroom that blends the funding streams and regulations of two standalone preschool programs exemplifies program “collaboration.” All of these examples of activities offer opportunities for public programs to increase public value by working together and simply differ in the degree to which programs engage with one another. Moreover, studying programs where minimal collaborative interaction occurs presents a prime opportunity to gain insight to what, if any, collaborative process dimensions exist.

Thus, the variation in degree of interorganizational activity among the Head Start and VPI programs sets the stage for investigating underlying collaborative process dimensions. This research is primarily exploratory; however, I propose that “stronger” or “presence” of governance, administration, norms of trust, and mutuality will be part of the outcome set for dyads with more involved degrees of interorganizational activity; whereas, “stronger” or the “presence” of organizational autonomy is proposed to be part of the outcome set for dyads with lesser-involved interorganizational activity given that strong organizational self-interest potentially reduces deep collaboration. I now turn to a discussion of the research design for this project.

Research Design

To investigate interorganizational activity that occurs between Head Start and VPI, I conducted a multiple case study (Stake 2006; Yin 2009) using in-depth interviews as the primary method of inquiry. I selected five Head Start regions across Virginia that represented varying degrees of interorganizational activity and variation on important demographic factors, such as population density and poverty.² Most dyads in this study had a history (10 plus years) of working together or attempting to work together, however, a change in a VPI or Head Start Administrator resulted in six dyads having relatively short histories of working together (less than 1 year).

Each region is comprised of multiple dyadic relationships between one regional Head Start administrator and a corresponding VPI administrator. Thus, regions in this study contained anywhere from two to seven Head Start–VPI dyadic relationships, with one regional Head Start administrator engaging in multiple dyadic relationships with the respective VPI counterparts located in their region. Supplementary appendix I displays the selected regions and variation on key factors. I conducted face-to-face and telephone interviews with 26 Head Start and VPI administrative personnel representing 16 complete dyadic³ relationships between respective Head Start and VPI administrators between fall 2012 and spring 2014.

After transcribing interviews and parsing text for structural and emergent themes (Saldaña, 2009), I conducted a QCA to examine the relationship between collaborative process dimensions and degree of interorganizational activity. QCA is a technique that uses Boolean minimization processes to link causal conditions to an outcome. For this analysis, the degree of interorganizational activity is considered the outcome and the collaborative process dimensions are examined as potential causal conditions. QCA allows for examination of equifinality, or when different combinations of conditions can lead to the same outcome (Rihoux and Ragin 2009). The technique handles causal complexity and focuses on comparing subset relationships between conditions and an outcome. I used fsQCA software (Ragin, Drass, and Davey 2006) to perform one of the types of QCA analysis, a crisp-set analysis.⁴ For crisp set, conditions and the outcome are dichotomized into “absent” or “present” to conform to a “0” or “1” coding necessary for Boolean logic.

Measures

To operationalize the outcome variable, degree of interorganizational activities, administrators were asked to complete a short inventory of the activities that they participate in with their VPI or Head Start counterparts.⁵ These activities range from minimal involvement, such as exchanging lists of enrollee names with one another, to highly involved activities, such as creating blended classrooms that contain both Head Start and VPI funded students. These activities were ranked from the least to the most involvement and assigned categorization to fit the modified Mattessich et al. (2001) continuum.⁶ After inventorying activities, each dyad was identified as having no relationship, cooperation, coordination, or collaboration based upon the number and degree of activities undertaken by the program pair. Table 1 shows a listing of typical activities ranging from the least to the most involved, and how each of the 16 dyads was categorized.⁷

To conduct the QCA, the original interorganizational activities outcome variable⁸ was created by breaking the dyads into two categories, those with more involved interorganizational activities and those with minimal to no interorganizational activities. Those dyads that partake in coordinating or collaborating degrees of interorganizational activity are labeled as “strong interorganizational activities” (SIORACT) and coded as “1.” Cooperating and no relationship dyads are coded as “not strong interorganizational activities” and coded as “0.” While mutually exclusive, “coordination” and “collaboration” both include activities that require more involved interactions than “cooperation” and “no relationship” that could be grouped for assessing underlying process dimensions. Eleven dyads comprise “strong interorganizational activities,” and five comprise “not strong interorganizational activities” (table 2).

To operationalize the collaboration process dimensions, I asked informants interview questions based upon the indicators created by Thomson, Perry, and Miller (2007). Given this study’s qualitative design, I loosely based questions about the five dimensions on 17 indicators developed by Thomson et al. (2007).⁹ For governance, four indicators tapped this dimension including taking each other serious as partners, brainstorming with each other, and the existence of formal (regular set meetings) as well as informal aspects (telephone calls or email exchanges). I denoted a dyad as having “strong” governance (SGOV) if they exhibited all four indicators, “medium-strong” if they exhibited three indicators, “medium-weak” if they exhibited two indicators, and “weak” if they exhibited one or less indicator.¹⁰

After this step, the next was to create the threshold by which a dyad was assigned as having strong governance “present” versus “absent.” For this particular dimension, I assigned dyads that exhibited “strong” or “medium-strong” governance as having strong governance “present”; those falling below this were deemed as “absent.” In general, the cutoff line to assign a dimension as “present” was between medium-strong and medium-weak for the four dimensions that included three or four indicators (governance, administration, organizational autonomy, and norms). With mutuality having two indictors, the cutoff point was assigned between strong and medium for that dimension. Supplementary appendix V contains all of the conditions and outcome and the decision logic that connects to establishing the final thresholds for each.

I calibrated the other dimensions using similar logic to governance as discussed above. For administration, three indicators comprise this dimension: if dyadic partners agree there are clearly defined roles, if they agree there is clear coordination of efforts (or identifies a collaborative coordinator), and if the partners agree on goals. For dyadic partners to exhibit the “presence” of strong administration (SADMIN), they have to agree to at least two of the three indicators.

Three indicators comprise the norms dimension. Trusting each other, committing to each other, and having relationship connectedness¹¹ indicate whether dyadic partners exhibit the normative dimension. For dyads to be considered having the presence of strong norms (SNORM), they had to have medium-strong or higher strength on the dimension, meaning that they exhibited strong relationship connectedness and answered in the affirmative to at least one of the other two indicators (trust and commitment). For dyads to be considered having an absence of strong norms, they did not exhibit strong relationship connectedness, and answered in the affirmative to one or less of the other two indicators.

For the mutuality dimension, two indicators tap the dimension: belief that one’s own program goals can be better achieved by working together and belief that the other program appreciates what your program brings to the collaboration.¹² Dyads that answer in the affirmative for both indicators are deemed as “strong” and considered to display the presence of strong mutuality (SMUTUAL). Dyads that answer in the affirmative to one or less of the indicators are deemed as medium strength or less and are considered to exhibit the absence of strong mutuality.

Finally, for organizational autonomy, three indicators capture this dimension, including partners finding it challenging to balance both programs’ goals, a partner expressing that their goals are affected (or changed) by working with the other program, and partners expressing concerns over current preschool enrollees being enrolled (or “stolen”) by the other program.¹³ To be considered having the presence of strong organizational autonomy (SAUTO), the dyads have to exhibit medium-strong or higher autonomy. Dyads that expressed having the “stealing kids” issue and answer in the affirmative to one or less of the other two indictors are denoted as medium-strong or higher.¹⁴ Dyads that express having the “stealing kids” issue or answer yes to one of the two indicators are designated as having medium-weak organizational autonomy. Finally, dyads that did not express having the “stealing kids” issue but answered that organizational goals were affected by the other program were labeled as weak.¹⁵

The process of operationalizing each collaborative dimension as a QCA condition involved a judgment call on whether each dyad exhibited the respective indicators. Some operational challenges did arise; for example, it was much easier to affirm participation in brainstorming sessions, an indicator of the governance dimension, compared to interpreting challenges about balancing both program and collaborative goals as an indicator of organizational autonomy. For example, some extremely connected dyads expressed balancing goals as challenging, but not insurmountable, and with dialogue usually worked out. Conversely, some of the least connected dyads suggested that balancing both program goals was not challenging at all because their lack of connectedness to the other program did not provide a platform for their goals to intermingle. While those with the most “balancing challenges” may have been intended to reflect higher organizational autonomy, I did not find this to always be the case. This finding is not surprising given Thomson et al.’s (2007) discussion that autonomy remains a challenging dimension to operationalize. In addition, while only occurring for a minority of coding assignments, occasionally information obtained by dyadic partners was not consistent. In these cases, I erred on the side of being conservative and reduced the strength of the dimension in part because how such a lack of consensus concerning the nature of the relationships reflects on the relationship. This is where my familiarity with the cases, as advised by best practice guides to QCA (Rihoux and Ragin 2009; Schneider and Wagemann 2010), allowed for assigning dyads to accurate levels for the respective process dimensions.

By translating the interview text into operational indicators and setting condition thresholds, I create an “in-set” and “out-set” for each condition and outcome. Simply put, an in-set is the number of dyads that have the conditions or outcome present; the out-set has the conditions or outcome absent. Table 2 relays this information.

Analysis

QCA identifies patterns of causal combinations that meet the sufficiency criteria. In other words, fsQCA helps to identify those causal combinations that whenever are present among the cases observed, the outcome variable is also present (Schneider and Wagemann 2010). Two analyses are run when conducting a QCA. First, the analysis is run for when the outcome variable is present. Secondly, the analysis is run for when the outcome variable is absent, or set negated (Ragin 1987). This allows the researcher to gain understanding of causal combinations that underlie those cases that present the outcome variable, but it also offers additional insight into causal combinations for cases where the outcome variable is absent.

Many experts recommend testing for necessary conditions first, since simply running the QCA could obscure existing necessary conditions (Rihoux and Ragin 2009; Schneider and Wagemann 2010). Necessary conditions are those conditions that when identifying all cases where the outcome is present (“1”), the condition is also present. For the first analysis of Strong Interorganizational Activity, I identified two necessary conditions, strong governance and weak organizational autonomy. For Not Strong Interorganizational activity, or those dyads that cooperate or have no relationship, I found three necessary conditions: weak governance, weak administration, and weak norms.

After checking for necessary conditions, I ran the truth table analysis. A truth table is all possible combinations of conditions. For crisp set QCA all of the conditions are dichotomized; thus, the total possible combinations are two raised to the total number of conditions (2^x). However, just because a combination of conditions is theoretically possible does not mean that any empirically observed cases (dyads) fit the combination. For this first analysis, which includes five conditions, fsQCA produced 32 possible combinations of conditions (2⁵) in a truth table, of which only six combinations contained observed cases for Strong Interorganizational Activity. For Not Strong Interorganizational activity, the same held true, with only six causal combinations containing observed cases. These “unobserved” patterns are also called logical remainders. The logical remainders were removed from the final analysis. Supplementary appendix VI contains all of the truth tables for the QCA.

Next, I identified which combination of conditions would generate the possible causal recipes. This decision is based upon assessing the number of observed cases (dyads) per each combination of conditions and the “consistency” of the combination. Consistency is the measurement of sufficiency, or the percentage of observed cases that have the combination of conditions and have the outcome variable present. Best practice guidelines for QCA suggest a consistency cutoff point of .75 for including condition combinations in the logical minimization process (Schneider and Wagemann 2010).

The fsQCA software produces three different solution recipes after running the analysis: parsimonious, complex, and intermediate. Out of the three, the intermediate solution reduces complexity while still maintaining meaningful details. It includes any theoretical assumptions that the researcher builds into the analysis. For this study, I included the assumptions that strong governance, strong administration, strong mutuality, strong norms, and weak organizational autonomy would be present in the outcome sets of those dyads with strong interorganizational activity present. As recommended by Ragin (2000), I use the intermediate solution created by fsQCA to identify recipes for both Strong Interorganizational Activities and Not Strong Interorganizational Activities.¹⁶

Findings

For the first analysis, looking at the five collaborative process dimensions as causal conditions for strong interorganizational activity, the solutions have both strong consistency and coverage. These measures are important in assessing the parameter of fit of the overall causal recipes. Solution consistency indicates the percentage of cases that display the outcome when the causal recipes are present. For this first analysis, a solution consistency of 100% denotes that the outcome (Strong Interorganizational Activity) is always present for the cases that display the two recipes. Recipe consistency measures for each recipe the percentage of cases that display the outcome variable when the recipe is present; for this first analysis, both recipes produced individual consistency scores of 1.0.

Solution coverage indicates the percentage of cases with strong interorganizational activities that are explained by the recipes provided. For this first analysis, solution coverage of 100% denotes that having strong interorganizational activity present is completely “explained” by the two recipes. Unique coverage identifies how much of the outcome variable is uniquely explained by each recipe. Higher unique coverage scores indicate less overlap between recipes; lower unique coverage scores indicate more overlap between recipes. Table 3 displays the results of the first analysis.

The results for Strong Interorganizational Activity yield two separate recipes. First, those dyads with strong governance and strong administration, but weak organizational autonomy have strong interorganizational activity. This recipe is perfectly consistent, with all 10 of the dyads that display this combination of conditions also displaying strong interorganizational activity. The recipe uniquely explains 45% of those dyads that have strong interorganizational activity present; this score indicates that 55% of the dyads that have this causal recipe also share the second causal recipe.

The second recipe shows that those dyads that have strong norms, strong mutuality, and strong governance, but weak organizational autonomy also have strong interorganizational activity. As with the first recipe, this second one is also perfectly consistent: all six dyads that display this combination of conditions also display strong interorganizational activity. This recipe uniquely explains 9% of the dyads that have strong interorganizational activity indicating that 91% of those dyads with strong interorganizational activity also display the first causal recipe.

As discussed above, initial examination of necessary conditions reveal that strong governance and weak organization autonomy are both necessary for strong interorganizational activity. This finding is reflected in the two recipes that both include these two conditions. Thus, the overall recipe for strong interorganizational activity can be rewritten as:

SGOV*sauto (SADMIN + SMUTUAL*SNORMS)¹⁷ → SIORACT

This new recipe suggests that dyads have to exhibit strong governance and weak organizational autonomy and either strong administration or strong mutuality combined with strong norms to also display strong interorganizational activity.

For example, for one of the strong interorganizational activity dyads, both Head Start and VPI administrators acknowledge strong administrative components, but a lack of complete mutuality. On both sides, while discussing clearly defined roles and coordination to work through tough collaborative issues, both program administrators expressed some frustration. The Head Start director discussed the sense that for this relationship, Head Start was merely a “funding stream,” and not appreciated for all of the program elements it brings to the table. For the VPI administrator, she expressed a lack of appreciation from Head Start for what the VPI program provides to all of the preschool students in the blended Head Start-VPI classrooms.

The second analysis examines Not Strong Interorganizational activity, or sioract. For this analysis, one recipe is produced, and the overall solution coverage and solution consistency are very strong, both at 100%. These findings indicate that whenever the recipe is present the outcome is also present, and the reverse also, that is whenever the outcome is present, the causal recipe is also present (table 4).

As mentioned earlier, the same three conditions are identified as necessary. Thus, for the five dyads that do not display strong interorganizational activity, the conditions weak norms, weak administration, and weak governance are necessary and when combined in a causal recipe, are also sufficient. Since there is only one recipe for not strong interorganizational activity, the unique coverage and consistency match the overall solution coverage and consistency at 100%.

For example for some of these less-involved dyads, different circumstances lead to weakened governance, administration, and norms. In one dyadic relationship in small central city, a change of the VPI administration has prompted recent conversations with the regional Head Start program about opening up opportunities for working together, but at the time of the interviews, little development had occurred to consider each other serious partners, establish clear roles, or create a trusting relationship. In another instance in the eastern shore area, while some attempts have been made to interact and discuss program enrollment, both Head Start and VPI administrators express differences about who coordinates efforts and the dyad currently lacks a formalized memorandum of agreement. Also, for this dyad some issues of mistrust and commitment abound due to enrollment issues and concern that the respective partner does not appear fully committed.

The findings from these two analyses update the collaborative process dimensions as envisioned by Thomson and Perry (2006) and Thomson et al. (2007). For these particular 16 dyads, to land on the higher end of the interorganizational activity continuum, collaboration or coordination, the dyad has to exhibit strong governance and weak organizational autonomy. This suggests that one of the structural dimensions and the organizational autonomy dimension are necessary ingredients for strong interorganizational activity. Dyadic partners who take each other seriously, brainstorm on collaborative issues, and/or have informal and formal aspects in place are more likely to engage in strong interorganizational activity than those that do not. Also, when partners are less focused on organizational issues (such as the “stealing kids” phenomenon) and more focused on overall collaborative issues they tend to engage in deeper collaboration than those dyads with strong organizational autonomy.

When strong governance and weak organizational autonomy are in place, two paths for strong interorganizational activity exist. First, dyads could also exhibit the other structural dimension, strong administration. Those dyads that also have clearly defined roles, coordinate with one another and/or agree on goals engage in more involved interorganizational activities. The second recipe suggests that dyads that display strong norms combined with strong mutuality also pursue more involved interorganizational endeavors. Dyads that trust each other, commit to each other, appreciate each other, and believe that goals are achieved better by working together also engage in deeper involvement.

These paths are not mutually exclusive, meaning that many dyads exhibit both of these patterns. These distinct recipes suggest substitutability to the combinations of dimensions that dyads with strong interorganizational activity can exhibit; however, the unique coverage scores suggest that these dimensions tend to occur together.

Dyads that do not engage in strong interorganizational activity display weak structural dimensions, governance and administration, and weak norms. These dyads lack the structural framework that comes with taking each other seriously as partners, brainstorming together and clearly defining roles and coordination. They also fail to develop trusting relationships or commitments to each other, which promotes them to engage in weaker degrees of interorganizational activity.

Additional Analyses

These first two analyses provide support that the underlying process dimensions, or the lack thereof, differ for those dyads engaged in more involved or lesser-involved interorganizational activities. They shed little insight to identifying processes at each respective degree of interorganizational activity, or the processes that comprise cooperation, coordination, and collaboration. The next part of the analysis addresses these issues.

To conduct these additional analyses, I redefine the in-set for each outcome variable. I ran cooperators, coordinators, and collaborators separately with those dyads identified as the respective degree of interorganizational activity coded as a “1” and all other interorganizational activities set as “0.” This created four opportunities to examine the underlying conditions for the respective interorganizational activities.

To compare collaborating dyads to all of the less-involved dyads in the study, collaborators are coded as a “1” and all others coded as a “0.” For coordination degree of interorganizational activity, I ran the analysis twice, once removing collaborating dyads from the analysis and comparing coordinating dyads with those lesser-involved dyads. The second analysis of coordination was to remove the lesser-involved dyads and compare the coordinating dyads to the collaborating dyads. Finally, for the last analysis, I combined cooperating dyads with “no relationship” and coded them with a “1” and compared them to the more involved dyads, coded as a “0.” Table 5 displays these new in-set definitions for the outcome variables. No changes were made to the causal conditions. Supplementary appendix VI contains the truth tables for the additional analyses.

Collaboration

First, I analyze collaboration degree of interorganizational activity. These are the dyads that engage the most with each other; specifically, they blend their programs to the point that Head Start and VPI preschool students sit in the same classroom together. Similar to the earlier analyses, I also checked for necessary conditions with the outcome variable set as collaboration. I found strong governance, strong administration, strong norms, and weak organizational autonomy to be necessary conditions. The only collaborative process dimension not deemed as necessary is mutuality.

For the analysis of collaboration, or COLLAB,¹⁸ the overall solution consistency is not overly strong, at 60%. Examining the individual dyad conditions and outcomes, the low solution consistency stems from two dyads that engage at coordination degree of interorganizational activity (for this analysis “Not COLLAB”) also sharing the same conditions as three dyads that engage in collaboration. They all have strong governance, strong administration, strong norms, strong mutuality, and weak organizational autonomy. A couple of factors potentially affect the low solution consistency. First, the in-set for this analysis is relatively small—there are only five collaborating dyads—so any addition or subtraction of a dyad from the measure of sufficiency greatly affects the solution consistency percentage. The addition of two non-collaborating dyads that share the recipe with the three collaborating dyads drops the solution consistency to .60. Even if all five collaborators shared the same recipe, the addition of the two non-collaborators would result in a solution consistency of .71, still below the .75 rule of thumb. Secondly, the two coordinating dyads that have all of the ingredients for the most-involved degree of collaborative activity lack upper administration support from their LEAs to blend classrooms. For one coordinating dyad of the northwestern valley, LEA administrators fear the consequences of blending two publicly funded preschool funding streams in the event that one of the grants would disappear from their school district. For the other coordinating dyad of the northwestern valley, having relatively few Head Start classrooms compared to VPI classrooms prompts upper LEA administration to downplay blending. Table 6 displays the causal combinations for collaboration outcome.

The analysis of collaboration yields one recipe. Typically, dyads that engage at the collaboration degree of interorganizational activity are strong on both the structural and social capital dimensions, and weak on organizational autonomy. Three of the five dyads that engage collaboration share these features, and is reflected in the solution coverage score of .60. The other two only lack strong mutuality. For this degree of collaborative activity, perhaps the necessary conditions are more telling since all five collaborators share four of the five dimensions (minus mutuality). However, what prompts someone to engage in a deeper degree of interorganizational activity by blending programs is not completely explained by these collaborative process dimensions since two dyads that also have these features opt to engage in coordination. As discussed above, lack of higher administration support in both of these dyads contribute to not engaging in collaboration degree of interorganizational activity.

Coordination

To analyze the coordination degree of interorganizational activity I chose to run the analysis twice. First, I compared those dyads with coordination to those with cooperation and no relationship, and removed collaborating dyads from the analysis. For the second analysis, I compared those dyads that coordinate to only those dyads that collaborate and then removed the rest from the analysis. I created two separate groupings because to group those that are the “most” involved (collaboration) with those that are the “least” involved (cooperation and no relationship) into one solidified “negated” group does not reflect a logical ordering. After running both analyses, the first testing of coordination (comparing coordinators to cooperators and no-relationship dyads) yielded identical results as the very first analysis run (Strong Interorganizational Activity, see table 3). For space considerations, this analysis has been removed from the findings, but can be made available upon request. The findings suggest that those coordinating dyads resemble collaborating dyads to the degree that even when separated out, coordinating dyads compared to lesser-involved dyads share the same causal conditions. Contributing factors to this similarity could be that coordinating dyads comprise 6 of the 11 “strong interorganizational activities” dyads and also share two of three Head Start Regional Directors with collaborating dyads.

Comparing coordination to lesser-involved dyads shows how similar coordination and collaboration may be with respect to collaborative process dimensions; it is now time to turn to how they may differ. For the second analysis I compare coordination degree to collaboration degree only. Necessary conditions for coordinating dyads include strong governance and weak organizational autonomy. The recipe has a solution consistency and coverage of 67%. This is not surprising given the same issues as discussed above that there are several dyads at the coordination and collaboration degree that share identical conditions; however, this test will uncover what differences may exist. Table 7 displays the causal combinations for the second analysis of coordination, or COORD2.

The first recipe for coordination compared to collaboration shows that dyads with strong administration and strong governance, but weak organizational autonomy and weak mutuality exhibit coordination. This suggests that those dyads that have both structural dimensions in place (as well as focus on collaborative issues compared to organizational ones), but fail to see how the programs achieve goals better by working together or fail to appreciate each other tend to coordinate. This recipe is 60% consistent and uniquely explains 50% of those cases at the coordination degree of interorganizational activity.

For one coordinating dyad in the southwestern valley, the VPI administrators share that they sense from Head Start a tension that stems from the threat of VPI “stealing” Head Start enrollees. This tensions results in VPI perceiving a lack of appreciation as a collaborative partner. Another VPI administrator who is part of a coordinating dyad in the northwestern valley discusses that attempting to blend Head Start and VPI goals results in her feeling as if working together does not always help her VPI program achieve its goals better. She explains: “I think that it tends to give us more things to work on. And I think you achieve goals better when you don’t have as many.” For these coordinators, while they have strong structural dimensions, they lack strong mutuality.

The second recipe shows that strong governance, strong norms, and strong mutuality, but weak organizational autonomy and weak administration can also lead to coordination compared to collaboration. This suggests that the social capital dimensions may be in place for a dyad, along with the necessary conditions of strong governance and weak organizational autonomy, but that a dyad lacks strong administration. This causal recipe suggests that a dyad may trust each other and see the benefit of working together, but lack a clear understanding of roles with each other and thus, tend to participate in the coordination degree of collaborative activity. This recipe is consistent at 100%, keeping in mind that it reflects one coordinating dyad (however, there are only six coordinating cases overall.)

For this coordinating dyad in the southwest valley, both Head Start and VPI representatives express some issues of role clarity with recent changes to the Head Start administration. Also, for this particular dyad, the VPI representative downplayed a clear collaborative role, and instead focused on her role as part of the VPI program. In addition, both VPI and Head Start administrators emphasize philosophical differences in programs that downplay goal alignment. Historically, Head Start is known more for its emphasis on socio-developmental focus with strong parental involvement, whereas VPI aligns with a more traditional academic focus of public school. To this end, a Head Start director from southwestern valley states:

The other thing that I think presents us with a real challenge is that whole philosophy [emphasis]. You know, I think it is very important for children to be academically ready for school, but I truly believe [emphasis] that it is a waste of time if they’re not socially and emotionally ready. Because look, here’s what I say: “Nobody is in prison because they couldn’t say their ABCs.”

While some administrative indicators are absent, the current Head Start and VPI administrators call on their historical relationship to emphasize trust and commitment when coordinating efforts.

Given the necessary conditions of strong governance and weak organizational autonomy for dyads at the coordination degree of interorganizational activity, the above recipes can be simplified as:

SGOV*sauto (smutual*SADMIN + SMUTUAL* SNORM*sadmin) → COORD2

This recipe suggests that when comparing coordinating dyads to collaborating dyads, coordinators share similar features of having both strong governance and weak organizational autonomy as necessary conditions. They differ from collaborators in that they generally have weak mutuality but have strong administration or they have strong mutuality combined with strong norms, but have weak administration. This finding suggests that coordinating dyads need not have both strong social capital dimensions and strong structural dimensions in place to engage in coordination between programs. Once they have met the necessary conditions of strong governance and weak autonomy, they can have strong administration if they fail to achieve mutuality. Another path is to establish strong social capital dimensions if they have weak administration. There is more than one path to take for coordinating dyads with workable substitutes of dimensions.

Cooperation and No Relationship

The next analysis was to revisit those dyads at the lower end of the interorganizational activity continuum, cooperation and no relationship. However, upon examining this test, pulling out cooperation and no relationship and comparing it to collaboration and coordination is the logical inverse of the very first analysis conducted for this QCA—the presence and negation of strong interorganizational activity. I did run the analysis to confirm, and indeed did find that the analysis produced the same necessary conditions: weak governance, weak administration, and weak norms. These were the same conditions produced in the QCA causal recipe, thus showing that these conditions are necessary, and when combined, sufficient for those dyads that cooperate or have no relationship. This relationship yielded 100% coverage and consistency for five cases that cooperate or have no relationship. The takeaway from this analysis is that those dyads lack the two structural dimensions, governance and administration, and do not trust or commit to each other.

snorm*sadmin*sgov → COOPNO

Limitations

Some limitations should be noted. First, some challenges did arise with assigning the collaborative process dimensions as QCA conditions. First, while I attempted to create objective operational definitions and logic criteria for assigning dyads to specific attributes, these assignments are understandably subjective. I identified and linked segments of interview text to support the decision-making process for assigning dyads to attributes of each indicator to add to the construct validity.

Secondly, knowledge about the dyads was based upon information gained from both the Head Start and VPI administrators who comprise the dyad; however, the information is arguably biased towards the VPI administrator’s perception of these dimensions. The VPI administrators were sharing information about their individual relationship with their respective Head Start program, whereas the Head Start administrator was asked to discuss his or her relationship with all of the VPI programs within his or her region. I asked follow-up questions to Head Start directors to ascertain specific elements of each of the relationships; however, it was challenging at times for the Head Start informant to reflect on each specific relationship. I followed up with Head Start directors with emails when specific process dimensions questions were lacking for a specific dyad. I also observed two group meetings for the regions with a larger number of dyads to help triangulate the relationship dynamics. While this is a limitation, I do not believe it influenced the findings substantially. I am confident that the Head Start administrators adequately differentiated each of their VPI partners to provide an accurate appraisal of their relationship, even if doing so required additional questions. Moreover, as with most research, these data are not perfect; however, assembling two administrators’ perceptions assured representation of their dyadic activities and processes with minimal missing data.

Thirdly, due to the nature of building an understanding of collaborative processes and activities from two different perspectives, occasionally I would receive contradictory information about a specific indicator, or in one case, the dyadic relationship shifted during the course of the study and follow-up information revealed contradictory evidence to earlier information obtained. In these cases, I erred on the side of being conservative and assigned an “average” attribute that fell between the extremes of the contradictory information. However, contradictory evidence was atypical for this study, and generally speaking, I found the information obtained from both Head Start and VPI administrators to support each other.

Discussion

Conducting the QCA proved fruitful for gaining an understanding of how collaborative process dimensions link to the degree of interorganizational activity. In particular, the logical minimization process that is the heart of QCA identifies those collaborative process dimensions that present for dyads at various degrees of activity. For this particular study, certain dimensions stood out for various degrees and will be discussed below.

When looking overall at strong interorganizational activity, strong governance and weak organizational autonomy stand out as necessary conditions. Strong governance epitomizes the approach that partners take towards collaborating with each other and allows the framework for deep involvement to flourish (Agranoff 2007; Mattessich, Murray-Close, and Monsey 2001; Thomson and Perry 2006; Thomson et al. 2007). These findings suggest the importance of structural elements, such as brainstorming and meetings, to solidify deep interorganizational involvement. While Thomson and Perry (2006) discuss all of these dimensions as processes, these findings provide evidence that basic structural building blocks, such as formalizing meetings, brainstorming with collaborative partners, and setting aside organizational goals, are necessary ingredients for undertaking more involved interorganizational activities together.

While having these conditions in place are necessary for more involved interorganizational activity, they are not sufficient. Dyads that display these conditions tend to also have two different additional combination of process dimensions: they either also possess strong administration or they have developed strong norms and mutuality within their dyad. Put another way, dyads need not have all of the dimensions in place to interact in more involved ways; they can either put energy (or have the requisite skills set) into administrative aspects, or they can work on the social capital dimensions.

At the other end of the continuum, dyads that engage in less involved interorganizational activities have weak governance, weak administration, and weak norms. These dyads tend not to brainstorm with each other about collaborative issues and have little formal or informal structure in place. They also are typically unclear about their roles and coordination between partners. Finally, they tend not to trust their partner and have uncertain commitment to collaboration. Like other scholars, these findings support the importance of structural dimensions, but also highlight the importance of trust to developing deep collaborative relationships (McGuire 2006; Vangen and Huxham 2003).

While these first analyses add to our knowledge about how the presence or absence of conditions link to more or less involved activities, they do not offer us specific insight to how processes link to specific degrees of interorganizational activity. Running the analyses that created specific in-sets for each degree helps to illuminate the linkages.

For those dyads that are the most involved, collaborators, four of the collaborative process dimensions prove to be necessary conditions: strong governance, strong administration, strong norms, and weak organizational autonomy. For the causal recipe, strong mutuality also is included. These dyads have strong structural dimensions and strong social capital dimensions. They focus more on collaborative issues than organizational ones when working together. This analysis supports that deep collaboration requires the underlying collaborative process dimension as envisioned by Thomson et al. (2007).

While mutuality was not deemed necessary, the majority of the collaborators exhibited this condition also. Similar to Thomson et al. (2007), the mutuality dimension in this study is the most varied and may be a challenging concept to tap. Particularly for this set of dyadic partners, the questions to tap mutuality may not adequately reflect feelings of mutual benefit or appreciation for one another. Informants were asked if they achieved their goals better by working together and if they felt a sense of appreciation. For some, the standalone nature of their individual programs, Head Start and VPI respectively, downplayed the necessity of working together to achieve programs goals. In other words, while some of them found working together as a way to improve preschool provision in their community, others dismissed this as a good, but unessential, way to meet their goals.

An interesting outcome of the QCA was comparing coordinating dyads to lesser-involved dyads and finding that the coordinators share the same causal recipes as strong interorganizational activity overall.¹⁹ Six of the 11 cases that comprised the in-set for strong interorganizational activity included those dyads that coordinate, so perhaps this drove the similarity. Coordinated groups, compared to lesser-involved groups, have strong governance and weak organizational autonomy. Then, they either build up a requisite amount of administration or they build a trusting relationship with mutual appreciation of each other.

Eleven of the 16 dyads examined in this study engage at the coordination or collaboration degree of interorganizational activity; thus, while comparing these more involved partners to lesser-involved partners is important, differences between these two strong interorganizational relationships is particularly illuminating. We know that both collaborators and coordinators have strong governance and weak organizational autonomy as necessary conditions. But how do they differ? Pulling out the causal recipe for coordinators compared to collaborators reveals that two paths lead to coordination assuming the necessary conditions are in place: coordinated dyads have either strong administration combined with weak mutuality or strong norms combined with strong mutuality combined with weak administration. To interpret this further, once having the necessary conditions in place, we find that coordinated dyads tend to develop clear roles and coordination but lack the belief that working together enhances goal achievement. Other coordinated groups develop trusting relationships and do believe that working together will help to accomplish their goals better, but they lack the requisite coordination and understanding of their roles.

These causal recipes present interesting takeaway knowledge for PPPC. What may keep a dyadic partnership at the coordination degree are strong structural dimensions but lack of belief that working together is beneficial. For others, there exists strong social capital and deep connections between partners, but an unclear administrative path for deep collaboration. Coordination between Head Start and VPI consists of coordinated communication activities, sharing space, or transportation; sometimes it includes developing joint professional development opportunities together. To participate in these types of activities, dyads need not have all of the collaborative process dimensions in place; they can still achieve these types of activities if they have strong governance and weak autonomy and either strong administration or strong social capital dimensions.

As with all small n studies, attempting to generalize these findings too broadly can be problematic; however, given this study’s focus on interorganizational activities between pairs of public preschool administrators, I do tap generalized activities and processes that most likely make up many different types of interorganizational relationships. In other words, the basic building blocks of interorganizational relationships studied in this research are probably highly similar to the basic building blocks that underscore a whole host of other collaborative relationships.

Also, while the findings of these analyses present a more static state of interorganizational relationships, certainly these relationships are dynamic and have and do change over time. In fact, as evidenced by some of the examples from the dyadic relationships, underlying process dimensions can shift with changes in personnel or other dynamic organizational fluctuations. For example, in southwestern valley, a Head Start building that suffered from a roof collapse created an opportunity for a conversation between VPI and Head Start administrators about blending a classroom. While organizational changes and events are not always predictable, findings from this study suggest process dimensions that public managers attend to in order to manage the dynamic environment.

Interorganizational activities were treated as the outcome variable to conduct this analysis; however, the relationship between processes and activities arguably feed one another. Undertaking activities together is part of building up collaborative processes just as building collaborative processes forge opportunities to undertake activities together. Thus, I am not pointing to having processes solidly in place before agreeing on an activity; rather, these findings point to process dimensions that public administrators need to work on and build as they engage in activities with one another.

Conclusion

The goal of this study was to illuminate similarities and differences in underlying process dimensions for dyads participating in varying degrees of interorganizational activity. Indeed, the findings support more than one path (including different combinations of processes) that collaborators, coordinators, and cooperators may take when working together. This approach to studying interorganizational activities and process dimensions for dyads could also be applied to networked relationships. While a survey approach with close ended-questions for both activities and outcomes may work better in a multi-partnered setting, each network could still be assigned to a degree of interorganizational activity and an “average” score on the process dimensions. This could further illuminate if the findings presented here hold for a variety of interorganizational forms.

While the findings support that deep collaboration typically involves most of the process dimensions, governance, administration, norms of trust, mutuality, and weak organizational autonomy, partners that do not plan to undertake that degree of involvement need not build up all of those processes in order to engage one another. For coordinators, partners that focus on collaborative interests and approach working together seriously can either build up strong social capital dimensions or they can clearly designate roles and coordination of efforts. Cooperators lack strong structural dimensions and do not trust each other. Without these dimensions they are able to engage in simple information exchange for program-based and professional development opportunities. However, those administrators wishing to deepen a relationship between partners will have to relinquish program focus, approach each other seriously and either build up trust and mutuality or provide clear administrative effort.

Supplementary Material

Supplementary data is available at the Journal of Public Administration Research and Theory online.

References

Footnotes