In a recent webinar, Barbara Means detailed how a Digital Promise learning sciences team is studying how instructional practices differ in blended introductory statistics courses at broad-access two- and four-year colleges. The work focuses on how teaching changes when institutions adopt digital courseware designed to deliver full-course content, practice, and assessment.
AI and Multi-Source Data Inform Analysis
Researchers analyzed data from more than 90 classes across 28 colleges, drawing on surveys, interviews, classroom observations, and AI-enhanced audio. Generative AI tools classified instructional approaches by examining talk-time percentages and speaker-labeled transcripts, used alongside traditional quantitative and qualitative methods.
Findings Highlight Promise and Limits
The resulting classification showed clear differences in courseware use, student engagement, and the likelihood of earning a C or higher. The findings point to both the potential and current constraints of applying generative AI in large-scale educational research.
