Teaching Stats: R, Open Science, and–of course–AI

Each fall semester for the past 5 years, Amanda Woodward has welcomed a sea of 360 undergraduate University of Minnesota (UMN) Psychology students, some eager and many anxious to begin the semester's work in PSY 3801, Introduction to Psychological Data Analysis–a.k.a. s·t·a·t·i·s·t·i·c·s.  The importance of learning statistics as a student of Psychology cannot be overstated; statistics are used to identify patterns in behavior across multiple individuals, to discern those which may be unique and which may be shared, to accept or reject hypotheses about how people will behave, and to test the effectiveness of therapeutic interventions.  Statistics give psychologists a standard yet powerful language to share their work and review that of others.  Woodward is introducing the next generation of researchers and practitioners to a foundational 'language' of their profession.

How Woodward teaches statistics has been shaped by numerous recent developments in the wider discipline.  The move by many researchers to use R for computing statistics is one of the most significant changes in how psychologists approach data analysis in the past 10-15 years. R is a free programming environment built specifically for data analysis, and it is now required for UMN Psychology undergraduate statistics.  Although some students are initially surprised by the need to take on computer programming and statistics all at the same time, Woodward has found that 80-90% of her students ultimately report that R has helped them learn statistics and that they would prefer to use R rather than not. R requires students to write out what they are doing in their analysis, thus they document each step of their process, unlike in hand calculations or in point-and-click software packages.  Some students even report R being a 'cool' skill to have learned.  One student ended up getting an internship with Meta due to his knowledge of R picked up in Woodward's Intro Stats!

As part of learning R, students are introduced to the importance of open science research standards.  Open science is a catchall phrase for a series of steps researchers are now using to be as clear as possible about how their studies were performed, their assumptions made, and the possible presence of inconclusive results. R is a key tool of open science.  Using R for statistics requires the transparency needed for other researchers to review what can be complex statistical scripts, determining exactly how certain statistical findings were made, testing the replicability of such analyses against other datasets, and documenting clearly the assumptions made in analyzing the data.  Again, by using R, students must write out exactly what they are doing in calculating their statistical task, resulting in a deeper understanding and positioning them to better judge how others are using statistics. This more transparent approach to sharing the methods behind the conclusions is what R and open science are all about.  Students benefit from learning under an open science umbrella as there is less focus on 'significant' results and more on accurately detailing what was hypothesized and what was concluded, including inconclusive findings.  Recent studies indicate that student attitudes about statistics and about open science correlate strongly; the more students understand the specifics of statistics and how they are used in forming research conclusions, the more positive they are toward statistics and science.

Although R has had a seismic impact on the profession in terms of both research and teaching, the almost overnight widespread adoption of AI tools, particularly by undergraduate students, is proving to be yet another instructional challenge–and opportunity.  Woodward is committed to the use of R in her classroom because of the benefits noted above.  Using R means using computers, and as of 2024, using computers means using Artificial Intelligence (AI).  AI tools have little problem producing R scripts or solving statistical problems. For that matter, Woodward allows students to use AI on technical assignments, although they must cite the use of AI, explain what the generated code means, and the code must adhere to standard processes established in class. AI is not plug and play; AI is plug and evaluate. In addition, Woodward does not allow the use of AI in her summative assessments, which are case studies to be completed in R Studio, requiring students to address different data analysis tasks, compute statistics, and document their work.  Woodward is personally concerned about AI globally in terms of how we as humans configure it and respond to its use. As an instructor, though, she sees AI as another tool to teach her students as they learn statistics.

In summary, teaching statistics in 2025 means teaching students to:

• compute statistics and analyze data;
• use R and learn basic computer programming;
• grapple with AI as a both challenge and an opportunity; and,
• adhere to the tenets of open science by using the tools and knowledge of our advanced technological world in efficient and ethical ways.

For a glimpse into the future today, check out Woodward's syllabus for Introduction to Psychological Data Analysis Fall 2025.