For a network structure to be successful, which of the following are necessary?

Introduction

Efforts at changing public educational systems in support of better achievement for students are commonplace across the globe. Many countries have experienced or are experiencing prime ministers, premiers, or presidents that define themselves as the ‘education’ leader and as such enact a series of reforms targeted at improving their nation’s schools [Levin 2007]. Typically, the subsequent gap between policy intentions and implementation, however, impedes these large-scale change efforts [Berman and McLaughlin 1978; Datnow et al. 2006; Fielding 2001; Hubbard et al. 2006; McLaughlin 1990; Spillane 2000].

This gap between intention and implementation is certainly the case in the United States under the No Child Left Behind [NCLB] Act of 2001. NCLB enacted a series of large-scale reforms targeted primarily at eliminating the persistent achievement gap by 2014. This legislation is built on several major assumptions. The first assumption is that schools can achieve unprecedented levels of equitable achievement among all subgroups of students in a short period of time without substantially changing current funding structures. Second, there are known methods of reform that will accomplish the task. Third, a formal structure of testing, threat of sanctions, and loss of both fiscal and human resources are the strongest incentives to produce improvement [Orfield et al. 2004]. Ironically, under NCLB the students who have been traditionally ‘left behind’, and for whom presumably the law was enacted, are now attending schools that disproportionately suffer the effects of sanctions and loss of resources [Sunderman et al. 2005].

The ‘teeth’ of the law are embedded in the accountability provisions that mandate Adequate Yearly Progress [AYP] toward established benchmarks. Failure to meet these targets results in a district or school being designated as ‘in need of improvement’ [INI] and a series of progressive sanctions that ultimately lead to a district or school being restructured. Perhaps most important, the number of districts and schools in INI is growing. During the 2004–2005 school year, individual states identified more than 9,000 schools as INI, representing a nearly 50% increase over the previous year [Stullich et al. 2006]. In California, the most populous state in the U.S., 2,218 schools were designated INI during the 2006–2007 school year, representing nearly 1/3 of all the schools in the state receiving federal funding. Exacerbating the issue, many schools designated as INI continue to remain in this status rather than move off accountability sanctions. For example, while 57% of the 90 Michigan schools in the final, more drastic stage of restructuring during 2005–2006 moved out of INI status the following year [Center on Education Policy 2007], only 5% of the 700 schools facing this sanction in California during 2006–2007 improved enough to be removed from INI status [Center on Education Policy 2008].

Given the pressure to achieve at increasingly high levels or risk facing progressive sanctions, educators have ratcheted up the number of change efforts underway in districts and schools to exit INI status [Mintrop 2004; Mintrop and Trujillo 2005]. It is at this juncture of sanction, urgency for change, and generally failed efforts at improvement that we examine the leadership network of the Dos Mundos School District,Footnote 1 an urban fringe district in the third year of INI. We have undertaken this case study to explore the underlying social networks of communication and knowledge transfer that are in place between district central office and school site administrators to examine whether these underlying networks influence the determination and direction of change [Tsai 2002]. Through our examination of the social networks in Dos Mundos we seek to understand the relationship between these networks and the capacity for districts to change when facing accountability sanctions.

Past research suggests that organizational improvement is closely linked to the ties within and across units [McGrath and Krackhardt 2003; Tenkasi and Chesmore 2003]. In the field of education, this has led to the development of professional learning communities and emphasis on collaborative structures [McLaughlin and Talbert 1993; Newmann and Wehlage 1995]. Most often, these types of communities are developed to increase communication and collaboration between teachers within schools; however, the literature from outside of education suggests that these ties, whether formal or informal, may be necessary throughout an entire district rather than isolated within schools to generate district-wide improvement. Recent research has found that informal social structures, in particular, provide opportunities for communication, information sharing, transfer and development of new knowledge between individuals, levels, and units within an organization [Ahuja 2000; Tsai and Ghoshal 1998]. To date, there is little understanding of the underlying networks within low-performing districts or between district and school leaders. A more in-depth investigation of the networks within districts ‘in need of improvement’ may uncover important characteristics of these networks that may facilitate or impede efforts at change. Therefore, our study is guided by the following research questions:

1. 1.

   What is the underlying social network structure of Dos Mundos and what does this structure suggest in terms of supporting or constraining efforts at change?

2. 2.

   How do site and district leaders perceive their leadership network and its effect on the flow of communication and knowledge throughout the district?

3. 3.

   To what extent does position within the formal and informal network structure affect leaders’ perceptions of the flow of communication and knowledge, and the district’s capacity for change?

To better understand how the underlying network of interactions within an organization may affect efforts at change, we drew upon social network theory and methods. Social network analysis [SNA] is a systemic approach used to quantify and visualize the ties and overall structure of formal and informal networks. Given the increasing number of districts and schools in improvement status in the U.S. there is an urgency to better understand the complexities of these organizations and provide strategies to educators working within these organizations. Understanding the structure of social networks may assist leaders to be better able to meet organizational goals once they understand the frequency and interaction patterns of communication and knowledge transfer between individuals and groups [Ahuja 2000; Tsai and Ghoshal 1998]. A systematic examination of underlying networks provides a lens through which to view organizations’ capacity for change.

We begin by briefly outlining the importance of a strong relationship between district and site leaders in enacting change as theoretical support for examining the bounded network of central office and site leaders. We then provide a review on social network theory and analysis and discuss how this body of research may be a useful lens to understand how networks may facilitate or inhibit change. Lastly, we discuss our study methodology, findings, and implications.

Importance of district–school relations

In an effort to understand the context in which contemporary schools operate, a numbers of scholars have shifted their focus from the school site as the unit of reform to exploring the important relationship between central offices and sites in engendering change [Datnow and Castellano 2003; Elmore and Burney 1997; Harris and Chrispeels 2006; Hightower et al. 2002; Honig 2004; Hubbard et al. 2006; Marsh 2002; McLaughlin and Talbert 2003; Togneri and Anderson 2003]. This line of inquiry acknowledges that schools are embedded within a larger district context and that this context may have a direct impact on the success of change efforts.

While the district context is important to our study, schools and districts also operate within, and are influenced by, the larger state and national educational systems. Districts and the schools they serve reside in a nested system of state and federal policies and a range of ‘non-system actors’ [Coburn 2005]. Educational policies are linked in formal and informal ways from the Federal government to the state to the community and ultimately to the district and schools [Datnow et al. 2006]. These linkages can be conceptualized as bridges between aspects of policy domains and influence both the structure and focus of districts as staff respond to the various policy pressures. Linkages may be formal such as specific mandates [e.g., NCLB], structural as in funding formulas, or relational as in the way that individuals work, interact, and network [Lasky 2004]. Although our study focuses on the relational linkages between district and site leaders, we consider the impact of formal and structural linkages of the greater policy system to understand our results.

Recent research identifies the key role that districts play in school improvement [Massell and Goertz 1999; Spillane 1996]. One attribute of reforming districts is a system-wide approach to improvement [Marsh et al. 2005]. District administrators are in the position to support systems whether they are moving toward goals or ‘stuck’ in dysfunctional patterns [Rosenholtz 1991]. Datnow and Castellano [2003] further this idea arguing, “supportive conditions at the district level are important to successful implementation and sustainability of whole school reform [p. 203].” Targeted support is especially critical for low-performing schools in improving outcomes [Massell 2000; Massell and Goertz 1999]. Beyond supportive conditions, scholars have also suggested that explicit, shared theories of action, mutual ‘sense-making’, and clear consistent communication around change efforts between central office administrators and site leaders results in greater systemic coherence and goal attainment [Agullard and Goughnour 2006]. Finally, central office staff are strategically positioned to ‘broker’ resources, knowledge, and ideas across the district [Burch and Spillane 2004]. These studies suggest the importance of a bridge between central office and site leaders for reform efforts to be successful. Absent that connection, successful efforts at reform appear remote [Leithwood et al. 2004].

Successful reform efforts, therefore, may require a shift in the way that change strategies are conceptualized and enacted within a school district. This shift entails a move from a singular focus on individualized segments of the organization to engaging the entire system in a network of connections. For example, a successful networked approach has been demonstrated in the United Kingdom through the National College of School Leadership’s [NCSL] Network Learning Group of 104 schools [Earl and Katz 2007; Earl et al. 2006]. This network yielded positive outcomes in a variety of areas including developing leadership, expanding boundaries, strengthening communities, and positively influencing student achievement [Earl et al. 2006]. The key conditions of these successful networks included frequent and pervasive communication, shared understanding and purpose, joint challenging work, and relationships built on trust that enabled the transfer of tacit and explicit knowledge. Although the NCSL project represents a much broader network than may exist in a single school district, the implications from this work are consistent with the scholarship on district–site relations and potentially hold importance as a way to create and understand networks within school districts. Above all, the balance of this work suggests the need for a more interconnected systems approach to organizational change [Fullan 2005; Hargreaves and Fink 2006; McLaughlin and Talbert 2003], requiring that district and site leaders, “…think systemically about schools and their development and see educational organizations in terms of their interdependent parts” [Smylie et al. 2003, p. 155].

Studies of successful school districts that applied more systemic approaches to change suggest a range of specific strategies that schools and districts can take in building stronger intra-organizational ties [Chrispeels 2004; Honig 2004; Togneri and Anderson 2003]. These strategies include creating structures for increased collaboration between central offices and sites [McLaughlin and Talbert 2003], enhancing communication channels [Agullard and Goughnour 2006], distributing leadership [Leithwood et al. 2007; Spillane 2006], and providing opportunities for input on decision-making [Brazer and Keller 2008]. While, these strategies identify formal structures that may help build inter-organizational ties, the broader literature on social networks suggests that in addition to these more formal structures district leaders should invest in the development of informal social relations.

From a network perspective, the work of an organization is captured well by Hubbard and colleagues [2006], who in their book on district reform define an organization as existing “in the interrelationships between activities of individuals” [p. 263]. It is the interactions between and among individuals that comprise the culture and structure of an organization. The assumption that undergirds this definition is that changes in organizations are often socially constructed [Hubbard et al. 2006]. Therefore, attempts to modify formal structures in support of greater collaboration, communication, shared leadership, and decision-making often require change in existing social relationships [Bartunek 2001; Borgatti and Foster 2003; Stevenson et al. 2003]. It is the organizational interdependence of action [Giddens 1979], reflecting a network of ties, that may ultimately moderate, influence, and even determine the direction, speed, and depth of a planned change [Krackhardt 2001; Mohrman et al. 2003]. According to Mohrman et al. [2003], because change processes emerge and are maintained through interpersonal relationships, “…lasting change does not result from plans, blueprints, and events. Rather change occurs through the interaction of participants [p. 321].” This research suggests that the careful exploration and analysis of the network of social relations in an organization may be an important first step in understanding and determining an appropriate change strategy, as well as identifying the potential for acceptance or resistance of individual members of the organization [Tenkasi and Chesmore 2003].

Social network theory and analysis

One of the basic conceptual foundations in understanding social network theory is the concept of social capital. A number of theorists have written on social capital; each foregrounding a different aspect of the concept and offering nuanced understanding of the idea [see, for example, Bourdieu 1986; Burt 1992; Coleman 1988; Lin 2001; Putnam 1993]. Lin [2001] notes that the common denominator between all major theorists includes the understanding that social capital consists of, “The resources embedded in social relations and social structure which can be mobilized when an actor wishes to increase the likelihood of success in purposive action” [p. 24]. Social capital is therefore an investment in the social relations in a system through which the resources of other individuals can be accessed, borrowed, or leveraged. This differentiates social capital from human capital, which refers to investments in training, development, or certifications of individuals, or physical capital that is contained in infrastructure and equipment [Bourdieu 1986; Coleman 1988; Dika and Singh 2002; Lin 2001].

The idea of social capital has been increasingly used in education and has been connected to a variety of positive outcomes such as educational attainment [Dyk and Wilson 1999], educational aspirations as tied to institutional agents [Stanton-Salazar and Dornbusch [1995], and home–school connections [Horvat et al. 2003]. However, while the effects of social capital are generally in the direction expected in terms of educational outcomes, researchers have suggested that additional empirical studies are necessary given the variation in both definition and outcome [Dika and Singh 2002]. In response to this critique we have narrowed our focus and drawn heavily on the scholarship of network theorists who argue that much of social capital research must be strongly rooted in individual interactions and networks [Lin 1999; Portes 2000]. Therefore, an understanding of the ties between individuals is foundational for our work.

Social capital can be operationalized as the resources embedded in social systems, accessed and used by actors for action [Lin 2001]. Social capital is therefore concerned with the resources that exist in social relations [sometimes referred to as ‘ties’] between individuals as opposed to the resources of a specific individual. This implies that actors must be aware of the assets in their network and take action through social ties to access these resources [Portes and Sensenbrenner 1993]. It is the quality of those ties between individuals in a social system that creates a structure that ultimately determines opportunities for social capital transactions and access to resources [Burt 1992; Coleman 1988, 1990; Granovetter 1973, 1982; Lin 2001; Putnam 1993, 1995]. Strong social ties support the transfer of tacit, non-routine, or complex knowledge [Hansen 1999; Reagans and McEvily 2003; Uzzi 1997], joint problem solving [Uzzi 1997], and the development of coordinated solutions [Uzzi 1997]. Strong ties also have been associated with low-conflict organizations [Nelson 1989]. Less dense networks tend to be better suited for the transfer of simple, routine information [Hansen 1999] and can provide for brokering opportunities between actors [Burt 1992; Granovetter 1973]. Interestingly, both strong and weak ties are necessary within a social structure as they facilitate access to different kinds of information [Haythornthwaite 2001; Tenkasi and Chesmore 2003].

Recently scholars have also been examining the concepts of social capital at an organization level. This larger organizational notion of social capital has been termed “communal social capital” and is defined as the “benefit that accrues to the collectivity as a result of the positive relations between different groups, organization units, or hierarchical levels.” [Ibarra et al. 2005, p. 360]. These studies primarily pertain to the influence of network structure on such factors as organizational performance, socialization, communication, knowledge transfer, innovation, and productivity [Song et al. 2007]. Studies in the business literature have suggested that strong ties within and across units are associated with initiating and sustaining successful large-scale change efforts [McGrath and Krackhardt 2003; Tenkasi and Chesmore 2003]. In fact, Cooper and Markus [1995] found that inattention to the strength of network ties during a significant organizational change effort of a large corporation resulted in a failed strategy.

Networks are typically identified by the content that is exchanged between actors or flows through the social ties [Scott 2000; Wasserman and Faust 1998]. Any number of resources may move through both formal and informal social networks. For example, communication networks might encompass information exchange, knowledge transfer, and advice networks. The content of the network flows through relationships creating a structure that defines the purpose of the network and in turn how well the resources flow between actors [Wasserman and Faust 1998]. For example, the social structure of a work-related knowledge network may differ significantly from the structure of a more normative social network, such as friendship. In both examples resources flow through ties [the first being knowledge, the second friendship], but the overall structure of the network may look quite different. These network structures may facilitate the transfer of resources if the necessary relationships are in place and are accessible, but they may also constrain resource exchanges if the ties do not have sufficient connectivity to move the resource [Hite et al. 2005]. In many cases, the underlying social structure determine the type, access, and flow of resources to actors in the network leading some scholars to suggest that the old adage ‘It is not what you know, but who you know’, is more accurately, ‘Who you know defines what you know’ [Cross et al. 2003].

In traditional social science research the attributes of the individual actor matter most. By contrast, in SNA, attributes, while important, take a secondary role to the ties with other actors in the network. SNA can reveal the underlying network structures that are important in understanding resource exchanges between individuals and groups within an organization [Cross et al. 2002; Song et al. 2007]. Understanding these network structures may be useful for educational organizations enacting change as these underlying networks may be leveraged to better create, use, and diffuse knowledge and innovation [Cross et al. 2002]. These resources may be of particular use as districts attempt to innovate and adapt to meet demands in high-stakes educational contexts.

Enhancing the strengths of ties between hierarchical levels within an organization is also critical to building an organization’s capacity for change [Kogut and Zander 1996]. Intraorganizational interaction and communication is important for the diffusion of new ideas and stable ties between sub-groups within an organization as they facilitate knowledge transfer, cooperative relationships, and innovation [Ghoshal et al. 1994; Song et al. 2007; Tsai and Ghoshal 1998]. Informal networks within and between units that have dense ties generally achieve at a higher levels of performance than those with sparse connections [Reagans and Zuckerman 2001]. This suggests that the overall structure of the network is important not only to individual resource advantage, but in achieving larger organizational goals [Guzzo and Shea 1992].

In addition to the benefits of dense connections between levels in an organization [including the administrative layers of a school district], units within the larger organization [such as schools] benefit from being connected to other similar units as a way to communicate information and gain novel knowledge [Tsai 2001]. These lateral connections are critical for the development of complex information, knowledge exchange, and innovation. In addition, dense lateral ties increase a unit’s ‘absorptive capacity,’ defined as a group’s ability to assimilate and replicate new information from external sources [Cohen and Levinthal 1990]. Absorptive capacity builds cumulatively with prior knowledge, meaning that the better equipped a unit is with prior related knowledge often predicts how well new information is absorbed into that unit. The ability of a sub-group to absorb information is directly related to the output of information from that group [Balkundi and Harrison 2006; Szulanski 1996]. This input–output relationship creates a reciprocal process that is facilitated by ties within and between lateral units. Therefore, investing in a sub-group’s ability to take in information from other sub-groups may lead to that unit outputting additional new information into the larger organization. This suggests the importance of dense lateral ties as a way to build absorptive capacity and support overall organizational goals [Tsai 2001]. In an educational system this might be accomplished by creating clusters of site administrators who interact and share work-related knowledge.

The balance of scholarship in this section suggests that dense interconnected networks of communication and knowledge transfer at all levels of the system may maximize the uptake of complex knowledge thus increasing the potential for organizational change. Social network studies in the field of education have examined principal networks [Friedkin and Slater 1994]; school and teacher networks [Bakkenes et al. 1999; Granovetter 1986; Penuel et al. 2007; Penuel et al. 2009]; teacher professional development networks [Lima 2007]; departmental structures [Lima 2003, 2004; Spillane 2006]; school–parent networks [Horvat et al. 2003]; and between school networks [Mullen and Kochan 2000]. In addition, two studies have examined the leadership network of an entire school district [see Hite et al. 2005, 2006]. Building off of these previous studies, a careful exploration and analysis of the network of social relations may offer a nuanced understanding of facilitative and constrictive conditions for change in an INI district.

Methods

We used an exploratory case study design to examine the communication and knowledge networks among principals and central office staff in the Dos Mundos School District, using Social Network Analysis [Scott 2000; Wasserman and Faust 1998] and semi-structured interviews [Patton 1990]. A case study approach is most appropriate when the phenomenon of interest has a level of complexity that requires multiple data sources and methods to gain an in-depth understanding [Yin 2003]. We selected Dos Mundos, as it is a mid-size school district that is both undergoing and poised to enact a significant change effort this school year through its strategic plan developed primarily in response to its INI status under NCLB. In addition, the district was selected as it reflects the diversity of student population found in many schools across California and in urban settings across the U.S. Dos Mundos serves students in kindergarten through eighth grade in eighteen schools, and is located in a rapidly growing ‘urban fringe’ area north of Los Angeles, California. The district has 16,000 students of whom 40% are Latino, 30% are African American, and 27% are White. Of the total student population 66% are classified as low socio-economic status and over 15% are English Language Learners. The district is in the third year of INI and facing greater sanctions from failure to meet Adequate Yearly Progress benchmarks under NCLB that may result in significant intervention.

Data collection

Social network analysis

In order to assess the social networks in Dos Mundos we developed an online survey that comprised 20 distinct networks and demographic questions. Network questions were generated from previous network research [Cross and Parker 2004] and original items. Specifically, we asked respondents to quantitatively assess their relationships with each of the other 58 administrators on a 5-point interaction scale ranging from 1 [no interaction] to 5 [1–2 times a week]. For this paper we will narrow our findings and discussion to two of the key networks, communication and knowledge transfer as those have been repeatedly identified in the literature as associated with change efforts. For the communication network we asked every administrator to respond to the survey prompt, “Please rate every administrator in the Dos Mundos network as to whom you communicate with in regard to work related tasks…and at what frequency?” The knowledge network was taken from the prompt, “Please rate every administrator in the Dos Mundos network as to whom you go to for work related knowledge…and at what frequency?”

Our study involved only those in formal leadership positions in the district, including the Superintendent, Assistant Superintendents, Directors, Supervisors from the Central Office and Principals and Assistant Principals at the school sites. For this study, we focused only on the administrators, rather than teachers, in an effort to understand the school–central office leadership networks. Our recent work has suggested that schools become more centralized and rigid after being placed under sanctions, and as a result of this response and the high degree of pressures placed on principals, these actors may be more likely to mediate knowledge and information flow between the school and district [Daly 2009; Finnigan and Stewart 2008]. Future studies will extend the work to include teacher leaders and site-level curriculum experts to provide a more complete network picture, as these actors often are important in the uptake, spread, and depth of change [Lima 2007; Spillane 2006].

Twenty-three district office and thirty-five site administrators responded to the survey representing 98% of the total administrative population in Dos Mundos. A high level of participation of individuals within a network is critical to ensuring validity. Table 1 provides the overall demographic qualities of the sample. We used a bounded/saturated network approach [Lin 1999; Scott 2000] meaning that we included all the members of the Dos Mundos leadership team [site and central office administrators]. We provided respondents with a list of all leadership team members and asked them to respond to each relationship and the frequency of that interaction. We used a bounded approach to network data collection as it provides for a more complete picture of the leadership network and potentially more valid results [Lin 1999; Scott 2000].

Table 1 Demographics of Dos Mundos’ Leadership network

Full size table

Interviews

While the SNA provides details regarding the structure of the network, our interview data offers insights into the nature of communication and knowledge transfer in the district. We conducted eight 45-min interviews with administrators in different formal and informal positions in the organization. We selected four site administrators and four central office leaders to represent the different formal positions in the network. We interviewed these actors using a semi-structured interview guide [Patton 1990; Spradley 1980]. The interview guides were designed to better understand the administrators’ perspectives of communication and knowledge exchange between and among the central office and site leaders. We asked participants to describe the patterns of communication and knowledge flow between and among the group of leaders and to provide examples of the types of exchanges that occurred. We also asked interviewees to describe and explain the overall structure of the network from their perspectives. We ended the interviews by asking what they might suggest to improve the flow of communication and knowledge within and across the system.

We selected these actors from different formal positions in the district based their location in the informal structure determined by how central the actor was in both the communication and knowledge network networks. Centrality generally refers to how many ties an actor either initiates or receives in relation to the specific network being examined. We divided the centrality scores of both networks into quartiles and then randomly selected two central office and site leaders from the 1st [least central] and 4th [most central] quartile. This allowed us to select respondents who represented different formal [by position] and informal [by centrality] positions in the network securing a range of perspectives.

Data analysis

Social network data analysis

We conducted a series of network measures using the UCINET software [Borgatti et al. 2002] to better understand the structure of the network. Given the extensive literature on the importance of tie intensity in networks, the fact that respondents are more accurate at identifying ongoing patterns than determining occasional interactions [Carley and Krackhardt 1999], and that we were interested in stable structural patterns [Krackhardt 2001], we dichotomized the data to include only the most frequent communication and knowledge ties between actors. In order to be considered a frequent tie individuals would have communicated or shared knowledge once every two weeks to a couple of times a week [4 and 5 on the rating scale]. In using frequency as a proxy for tie intensity we built on the work of Borgatti [2007] who outlined four distinct types of relationships studied in social network analysis: Proximities [e.g., similarities or distances], Relations [e.g., kinship or roles], Flows [e.g., the exchange of money or goods], and Interactions [e.g., talking with, helping]. Borgatti argues the most appropriate measure of intensity of tie is specific to the type or relationships the tie represents. For example, the strength of proximity-based tie should reflect “how proximal” two nodes are, either representing very small distances or very high correlations between profiles. For interactional ties, the type of relationship we research in this study, strength of tie can be measured either by quantity [how frequently do you interact] or quality [how ‘good’ is the interaction]. In this study we focus on the quantitative aspects of the interaction measured by frequency in order to provide more comparability across respondents, as the qualitative aspect of ‘quality’ is necessarily subjective. However, it should be noted that using frequency alone has limitations such as not acknowledging the role quality [measured through expressive relations] may play on communication and knowledge networks [Marsden and Campbell 1984].

We first ran a density measure to determine the percentage of frequent ties within the communication and knowledge network in total. The density of a network is the number of connections between actors divided by the number of total possible connections and can be thought of as how tightly knit a network is. A dense network, meaning one with a high percentage of ties, is thought to be able to move resources more quickly than a network with fewer ties [Scott 2000]. We also block-partitioned the data into central office and site level administrators to compare the density of communication and knowledge flow between ties in each of these subgroups.

Centrality measures were taken on each of the actors to determine the total amount of ties an actor initiates and receives in the communication and knowledge networks. Centrality has been thought of as an index of activity [Freeman 1979]. Highly central actors in a network have increased access to resources and a high potential to create new linkages that may enhance social capital and build organizational capabilities [Stuart 1998; Tsai 2000]. Those who are less central to the organization may be on the periphery and receive less information and often do not have the opportunities to gain from the resources and information held by those in more central positions. Moreover, these less central individuals are more likely to receive only the resources deemed necessary by those in a more centralized position [Burt 2000], thus potentially restricting their perspective of the overall organization. Centrality therefore can be considered a point of intersection in which the person in the center of the intersection is able to disproportionately and more quickly amass resources, thus allowing this central individual to influence the network by determining where the resources flow [Raider and Krackhardt 2001].

Given our interest in the relations between district and site administrators we conducted an External/Internal ratio analysis, often referred to as an E-I index, that indicates the extent to which the overall organization is characterized by interunit, as opposed to intraunit, strong ties [Krackhardt and Stern 1988]. The scale ranges from −1 completely internal [intraunit] ties to +1 completely external [interunit]. This network measure assesses the relationship between external and internal ties based on a specific actor attribute [in this case work location, meaning either central office or site] by comparing the numbers of ties within groups and between groups. A group [in this case central office administrators or site administrators] is considered externally focused if relationships are more uniformly dense over the entire network [interunit], or internally focused if there is a small dense core of relationships within a segment of the network [intraunit]. High E-I indices [more externally focused] have been associated with large-scale successful organizational change [Krackhardt and Stern 1988; McGrath and Krackhardt 2003] and greater unit cooperation [Nelson 1989] while low E-I scores potentially limit how well an organization negotiates external pressures [McGrath and Krackhardt 2003].

Lastly, we conducted a core periphery [CP] measure to better understand the structure of the entire network. A CP network structure is defined as one that has a dense cohesive central core of actors with a less connected periphery [Borgatti and Everett 1999; Wasserman and Faust 1998]. The CP network structure essentially determines to which degree actors belong to one group [core] or another [periphery]. The CP measure compares a given overall network to a theoretically perfect CP model [highly centralized] and reports the correlation between the actual and idealized centralized structure providing a measure of the centralization of a network. This network measure is useful as CP structures have been associated with how well certain types of knowledge are transferred [Cummings and Cross 2003]. In addition to the aforementioned network measures we conducted interviews around the social networks of communication and knowledge with central office and site administrators in Dos Mundos to enable us to acquire a deeper understanding of the characteristics and structure of the networks in this district.

Interview data analysis

We used the interview results to explore the perceptions of administrators in different structural positions [more and less central] in the network. The eight interviews were audio-recorded and transcribed verbatim. The first cut of the interview data consisted of an inductive analysis that allowed important themes to emerge “out of the data rather than being imposed on them prior to data collection and analysis” [Patton 1990, p. 390]. Interview data were analyzed using a constant comparative analysis method [Boeije 2002; Glaser and Strauss 1967], as well as checking and rechecking emerging themes [Miles and Huberman 1994]. We grouped responses to the prompt and compared the different perspectives of district and site administrators, as well as those in the least and most central positions in the network. The themes that arose from this preliminary analysis were then re-examined, looking for patterns across district office and school sites and structural position as connected to the social network survey findings. This process of constant comparison “stimulates thought that leads to both descriptive and explanatory categories” [Lincoln and Guba 1985, p. 341]. In order to ensure the trustworthiness of interpretations, member-checking procedures were carried out as emerging themes developed and were shared with participants [Miles and Huberman 1994].

Findings

After analyzing the data several major findings about the leadership network in Dos Mundos emerged: [1] sparse communication and knowledge ties exist between site and district office leaders; [2] the district office administrators are central in the underlying social network and rely on technical exchanges; [3] frequent within district office ties and infrequent school–school connections were evident; and [4] the overall structure of the Dos Mundos Leadership network is centralized. In the following section, we discuss these key findings based on our analysis of both network and interview data.

Sparse communication and knowledge ties between site and district office leaders

After dichotomizing the data to reflect frequency of interaction, we measured the density of the communication and knowledge networks. Fifty-eight administrators in the bounded network represent a potential of 3,306 ties across district and site administrators. We first examined the communication network in Dos Mundos in terms of the rating given by administrators regarding the frequency of communication with other administrators around work-related tasks. Based on their responses, we found that the district had an overall density of 22%, meaning that 22% of all possible ties in the network [n = 3,306] occurred at a high frequency [ties that occur at least once every two weeks and as many as a couple of times a week]. This suggests a pattern of low interaction [less than once a month if at all] between and among central office and site administrators in the remaining 78% of the possible work-related communication network ties.

Next we examined the density of the knowledge network in terms of the frequency that respondents rated going to other administrator in the district for work-related knowledge. This network had an overall density of 13% [out of 3,306 possible ties] indicating an even less dense network in comparison to communication. This suggests that 87% of the remaining work-related knowledge transfer ties occur, if at all, less than once a month. This level of knowledge exchange system-wide may limit the amount of shared knowledge in the organization and perhaps inhibit efforts at complex change.

The measure of degree for the communication network had a mean of 5.379 with a standard deviation of 4.563 and a range of 0–27, indicating that out of a network of 58 administrators the average leader is connected to five other administrators representing only 8% of the administrative team. More importantly, the large standard deviation and range indicates a high degree of variability between actors. The majority of actors in the network have limited communication ties to other administrators that may make system-wide communication more difficult as the average administrator is communicating with less than 10% of the total leadership network.

The knowledge network had a comparatively lower mean score to the communication network of 4.069 and standard deviation of 3.617 with a range from 0 to 21º. This variability indicates that a majority of actors are not in the structural position to either offer or receive work-related knowledge. In both the communication and knowledge networks district administrators had higher degree scores in comparison to site administrators ranging from 13 to 21 suggesting that the district office is more central in the network. Figures 1 and 2 created with Netdraw [Borgatti et al. 2002], a network visualization tool, graphically display the communication and knowledge network structures with the lighter color nodes representing the central office administrators and darker color the site leaders. Nodes are sized according to overall centrality [with larger nodes having higher centrality scores] and connecting lines indicate a frequent tie between actors. The type of shape, either circle or square, refer to the core-periphery model that will be discussed later.

Fig. 1

Communication network [To whom do you communicate with in regard to work related tasks…and at what frequency?]

Full size image

Fig. 2

Knowledge network [To whom do you go to for work related knowledge… and at what frequency?]

Full size image

District office administrators are central in the network and rely on technical exchanges

As is shown in both Figs. 1 and 2 while the central office administrators have the highest degree of centrality, thus indicating disproportionately large organizational influence, their centrality does not necessarily constitute ready access to less central parts of the network, namely the school sites. In fact, the majority of district office communication and knowledge comes from other central office administrators, as indicated by the ties between lighter colored nodes. The figures above illustrate the relatively frequent ties between central office administrators in both the communication and knowledge networks and comparatively less frequent ties between site administrators. This high frequency communication and knowledge sharing between central office leaders is useful in supporting consistency and coherence in terms of work-related activity, but it also may constrain opportunities for novel, non-redundant information to enter the central office. In addition, results suggest that the majority of site administrators are not central in either of the networks. Furthermore, they have limited high frequency work-related communication or knowledge ties with the central office or one another. With weak communication and knowledge ties of site administrators, this network structure may constrain the overall reform strategy in the district in terms of enacting a district-wide coherent change effort.

The school leaders described these communication patterns as one-way in nature, flowing to them from the central office. Furthermore, they reported that typically within communications is a request to produce some type of technical information to be reported to the district office. A more central principal of eight years in the network captured the general theme of centralized communication offering, “It is like we are cogs in the machine just responding to district mandates and not having the opportunity to share our views to the district office or each other [P2MFootnote 2].”

Closely related to the centralized, technical communication patterns was the finding that site leaders perceived erosion in their ability to make decisions at the site. A less central principal stated,

Decision-making has become more autocratic with less of a shared decision-making process in place. This has forced me to give up some of my creativity and much of the control I once thought I had in terms of decisions and directions to take at the school. It now comes from the district office. [P1L].

While it was not clear whether this was a district response to the NCLB sanctions, our research in other districts suggests that this increased centralization and decreased site-based authority may be closely linked to the ‘threat’ from accountability sanctions [Daly 2009]. In combination, these themes suggest asymmetrical centralized communication patterns between the central office and site with district mandates and directives often overriding site leaders’ ability to make school-level decisions. This finding was consistent regardless of the participants’ centrality in the network.

Site administrators noted that communication and knowledge sharing tended to be more ‘technical,’ meaning that it was less about larger issues related to teaching and learning and more about budgets, prescriptive curriculums, and fidelity to district-adopted programs. In support of the idea, a more central principal shared,

The district office, who has been bottling level 5 leadership to us, are now only giving us more level 1, micro-management inflexible directives without culling them or without anything else and without a whole lot of direction or connection to our work at school [P3M].

Another less central principal offered, “The district belief is now that it is about the instructional materials you use, scripts allowed, time on task, and test scores, and less about developing relationships [P1L].” This more technically oriented, first-order communication pattern may constrain leaders from taking more adaptive [Heifetz and Linsky 2002], second-order approaches to change and will likely inhibit improvement.

The view of the factors affecting the underlying networks was somewhat different from the central office perspective with more of a focus on the lack of cohesiveness across the district. One less central assistant superintendent captured the overall feeling of central office administrators: “Part of our struggle with communication between the district and sites is that we’re fragmented, broken into two worlds, and that has created challenges for us [D3L].” Another more central district office administrator concurred offering a similar frustration,

I am sitting with these people [site administrators] day after day, year after year, supporting them as many of the complaints that come out of the schools end up on my lap, and I try to resolve them, and I call them and share and communicate frequently, but I feel that we are not connecting. I just want to make it work [D4M].

Administrators in the district indicated that these challenges and minor frustrations might have to do with the nature of the work. Central office administrators see themselves as having a district-wide focus, while they perceive site administrators typically advocating only for their individual sites and often, “…missing the big picture, the overall plan that makes the district function as a whole [D4M].” Ironically, according to our findings it is likely difficult for site administrator to see the big picture as they are not as central in the larger network and have limited connections to district office leaders who have a broader perspective.

Frequent within district office ties and infrequent school–school connections

When we block partitioned the district office administrators from school site leaders an interesting network pattern emerged. In terms of the communication network, the central office administrators have a density of 30% frequent communication ties between one another compared with only 16% of the ties between site administrators. In addition 23% of the knowledge ties are frequent between central office administrators compared with 7% for site administrators. This data suggests a comparatively dense communication and knowledge network within the central office and sparse ties among site leaders with 93% of the site ties occurring less than once a month. This may significantly impede the ability of site administrators to communicate and share work-related knowledge that may inform practice; thus, potentially limiting the diffusion of innovation across schools within Dos Mundos.

In addition, the E-I index [ranging from −1 completely internal ties to +1 completely external] was run on the administrator’s primary work location [central office or site]. The overall E-I index for the communication network was −.551 [p