The Practicing Connections Framework: A Learning Science Approach to Introductory Statistics

Introductory statistics courses often present a fragmented sequence of topics-data visualization, t-tests, ANOVA, regression-that can leave students with knowledge that is neither coherent nor transferable. This traditional approach often contrasts sharply with how most statisticians engage with data: through statistical modeling. Modeling frameworks unify diverse concepts and tools, yet introductory courses rarely place modeling at the center of instruction. This disconnect limits students' ability to apply statistical concepts flexibly, especially to advanced techniques like multivariate and computational models.

The Practicing Connections Framework, grounded in research from the learning sciences, provides a path forward. This framework posits that coherent and transferable knowledge is built when students actively practice connecting core concepts (such as the statistical model) with: real-world contexts and authentic practices, key representations (e.g., R code, graphs, General Linear Model notation), and other concepts and procedures.

The CourseKata interactive textbook, designed for introductory statistics courses, embodies this framework. Using the core concept DATA = MODEL + ERROR, CourseKata introduces the mean of an outcome variable (i.e., Y) as the simplest statistical model, builds an understanding of residuals and standard deviation as measures of error. As the course progresses, students encounter more complex models, where the value of another variable (X) is used to predict Y. This extension introduces students to group models (i.e., t-tests, ANOVA) and regression models, all within the same modeling framework. Each new variant extends the modeling core concept.

Evidence shows this method enhances transfer to advanced models, promotes equity, and shifts attitudes toward R programming positively. By centering statistical modeling in introductory courses, the Practicing Connections Framework aligns teaching with modern practice.

Pedagogy

Statistical Modeling

Introductory Statistics 

Education

Symposium on Data Science and Statistics (SDSS) 2025