Longitudinal Social Network Analysis of a Networked Improvement Community Supporting Ambitious and Equitable Teaching Practice

• In Event: Designing for the Next Generation of Teaching Practices With a Networked Improvement Community: A Longitudinal Study of Local Adaptation and Improvement

Abstract

Purpose: Applying principles for the development of a NIC (Bryk et al., 2015), researchers partnered with teams of teachers, coaches and principals from eight secondary schools to improve a set of evidence-based science teaching practices and school and district-based professional learning structures. The project began in a few schools then expanded district wide to all secondary schools. While many schools experienced an initial increase in students’ standardized test scores (grades 8 and 10), schools that participated for 4 years experienced steady gains in scores. More importantly, teachers began working with one another to improve rigorous and responsive classroom teaching practices. This paper describes the social network that supported these learnings.

Framework: Within a NIC, knowledge travels through at least three mechanisms: through people, tools and tool use, and joint engagement within designed settings that bring different stakeholders together (Bryk et al., 2011; Coburn, 2003). Social Network Analysis is a methodological approach that examines ties among individuals in networks ties and can indicate the impacts of newly engineered social structures and tools that support these interactions.

Methods: The NIC grew from two to eight schools and the types of networked activities across the network expanded. Each school participated in full-day Professional Learning Communities (PLCs). To support the network, we ran convenings, summer academies and workshops.

Data Sources: Science teachers, coaches, and researchers (N=21) were surveyed annually on who they seek advice from to improve science instruction. Longitudinal, quantitative social network analyses using separable temporal exponential random graph modeling (STERGM) was employed to analyze the network data (adjacency matrices). School membership was accounted for in the model. STERGM considers the formation and dissolution of ties (relationships between two individuals) across timepoints, rather than at one time.

Results: Network descriptive statistics are provided in Table 2 and sociograms across years in Figure 4. STERGM results showed that the formation of ties over time were lower than expected by chance (formation edge coeff=-5.04, SE=0.14, p < 0.001). However, once a tie within the network was formed, there were significantly greater reciprocal ties (mutual relationships) than would be expected by chance (reciprocal tie coeff=1.43, SE=0.31, p < 0.001), as well as significantly more ties forming in the same school (ties in same school coeff=2.21, SE=0.19, p < 0.001). Across the network measurement occasions, there was a negative dissolution edge estimate, indicating a significant proportion of ties dissolved less than would be expected by chance (edge coeff =-0.88, SE=0.31, p=0.005). Additionally, reciprocal ties were significantly likely to persist once a tie had been formed (reciprocal tie coeff=1.34, SE=0.60, p=0.028). Last but not least, ties within a given school were likely to remain consistent (the estimate was close to zero) (ties in same school=0.42, SE=0.37, p=0.258).

Scholarly significance: Findings from this longitudinal social network analysis show how the network changed over time. Of practical importance, we found that, although most teachers networked with individuals in their own schools, those relationships appeared to have significant persistence, which bodes well for links with practice.

Authors

• Jessica J. Thompson, University of Washington

• Elizabeth A. Sanders, University of Washington

• Nathan Abe, University of Washington - Seattle