Assessing knowledge construction in online learning communities

The Teacher Professional Development Institute employs Internet technology to support distributed professional learning communities. The Secondary Teacher Education Program (STEP) at the University of Wisconsin-Madison will soon employ a similar model in its program. Work and learning within STEP evolves around instructional design teams comprising pre-service teachers and various advisors. Because much of this work is project-based rather than curriculum driven, traditional assessment is not appropriate. Our approach involves taking advantage of the on-line environment's ability to collect data on collaborative process. To help determine standards of evidence appropriate for judging groups as knowledge-construction entities, we articulated a theoretical framework based on four accepted views of social knowledge construction. Assessment was conceptualized as abductive inference, case identification given particular theories and relevant evidence pertaining to them. The theories predict indicators that can be represented as nodes linked by subjective probabilities within a Bayesian network. Observable features of on-line interactions provide input to the net. Outputs are probability distributions indicating the degree to which a group meets normative standards prescribed by theory. The network can produce general assessments and diagnostic profiles for groups and subgroups within large communities.