Association for Public Policy Analysis & Management

This study concerns how intra-organizational networks affect the implementation of policies and practices in organizations. In particular, we attend to the role of the informal subgroup or clique in cultivating and distributing locally adapted knowledge on which actors can draw to implement innovations.

Hypotheses

Our first hypothesis concerns the capacity of the organization to distribute knowledge equally to all subgroups: the more even the flow of knowledge to potential recipient subgroups the greater will be the systemic implementation of practices dependent on the knowledge.  Our second hypothesis builds on the assumption that superior knowledge flows will come from subgroups in which interactions are focused on relevant productive activities: the more knowledge is restricted to flow fromsubgroups focused on a particular practice, the greater will be the systemic implementation of practices dependent on that knowledge.

Data and Methods

We test our hypotheses with longitudinal network data in 21 schools. We begin by identifying cohesive subgroups from the network question asking teachers to list their closest colleagues in their schools. We then use Shannon’s measures of entropy to characterize the distribution of potential flows of expertise between subgroups within schools.  Our outcome is the average change in implementation of locally defined school-wide initiatives, essentially modeling change in implementation using OLS (essentially a differences in differences approach).

Results

We do not find support for the first hypothesis.  But we do find support for the second hypothesis that implementation of school-wide initiatives is greater the more knowledge is restricted to flow fromsubgroups focused on a particular practice.  We quantify the robustness of our inference, reporting that the correlations associated with an unobserved confounder would have to be more than three times greater than the correlations associated with our strongest covariate to invalidate our inference.

Policy Application

Our findings suggest one should characterize networks not purely in terms of their structure, but in terms of where resources are located in that structure.  Thus a network that can effectively distribute resources for one organizational task may not do so for others.   Our findings can also help managers cultivate knowledge flows from expert subgroups to contribute to organizational change.