Putting The Learner At The Center Sharing Analytics With Learning Participants

I argue that the researchers and developers leveraging learning analytics should find ways to use analytics information to involve learners as active participants in their own learning. Inscrutable student models are used to build predictive models, to direct the actions of automated systems, to assess student learning, or to recommend student interventions. Knowing when and what to expose to nurture self-directed and self-regulated learning abilities involves, in part, having student models that are accurate, up to date, and relevant. Good student models can support applications that seek to influence a range of intrapersonal elements cognition, metacognition, emotions, engagement, motivation, creativity and those interpersonal as well: social relationships, identity, group roles, and collaboration. This "scrutable" modeling task is a central challenge for research in learning analytics, because detailed "by concept" models, agency, efficacy, and psychographic factors of individuals should merge with models of social roles and relationships to provide actionable information to students, teachers, and administrators.I describe what is important to measure and model about the learner and present methods that experimentalists have of collecting multiple forms of data and making inferences about learners. In online learning environments that support the collection of similar data from many more students than can be supported in traditional classrooms, composite models can be created that reflect "best guesses" of what to do next for a given student. Numerous challenges exist in the complexity of learning and the limitations of modeling for driving action, yet the area holds promise for improvements in students' overall experiences.