The multiplexity of professional learning communities: exploring the co-evolution of teacher social networks

ABSTRACT

Research on teacher professional learning communities (PLCs) has found that part of their benefit comes from increased collegiality. And yet, while the nature of such collegiality is theoretically multifaceted and includes a combination of instrumental, expressive and substantive relationships, this multidimensional nature of PLCs is almost never studied directly. In this paper, I fill that gap by examining the dynamics of relationship formation and evolution in a multiplex social network of 70 teachers during the first year of a small school reform. Using longitudinal social network models I analyse the within and between network dynamics of three types of discussion networks: course-related, personal, and, reform-related. Results show that the formation and evolution of ties within particular discussion networks were shaped by preference for similarity in demographic characteristics and attitudes. They also show that co-evolution between networks was shaped by reform-related discussions operating as a bridge connecting personal and course ties.

KEYWORDS:

• Professional learning community
• social network analysis
• school reform

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. The emphasis on small schools can be at least partially traced to what was at the time an emerging body of research suggesting that smaller school size was associated with positive academic (Lee and Smith 1997) and social (Cotton 1996) outcomes for students as well as teacher efficacy and satisfaction (Lee, Dedrick, and Smith 1991).

2. The overall survey response rate across all three waves was 80% (time 1 = 75%, time 2 = 94%, time 3 = 94%). In each wave, however, some teachers did not fill out the network portion of the survey. Missing survey data were imputed using the k-nearest neighbour procedure in the r package DMwR (Torgo 2010). This approach was chosen over a multiple imputation one because Siena models require a single value for attributes. Missing network data in the first wave was also imputed using the k-nearest neighbour in UCINET (Borgatti, Everett, and Freeman 2002), utilising the extension for networks developed by Žnidaršič, Ferligoj, and Doreian (2018). The Siena program then internally handled network imputation for the next two waves (Ripley et al. 2017).

3. For the first two constructs, individual items were rated on a scale from 1 to 6, where one represents ‘strongly disagree’ and 6 represents ‘strongly agree.’ For the perception of success construct individual items were rated from 1 to 4.

4. The exact name for the effect used here is transitive triplets (see Ripley et al. 2017 for more details).

5. I used the square root for in-degree popularity, out-degree popularity, and out-degree activity because this provides a better model fit (Ripley et al. 2017).

Additional information

Notes on contributors

David Diehl

David Diehlis an assistant professor in the Department of Human and Organizational Development at the Peabody College of Education at Vanderbilt University. His research focuses the social and organizational dynamics of educational institutions such as schools, classrooms, and universities. One line of his work examines how attributes of educational organizations (e.g., school size, climate, demographic composition, internal structure like type of curricular differentiation) shape the social networks and social experiences within them. A second line examines how ideas about schools and schooling diffuse through policy and education fields and are translated and reconstituted by teachers within schools.