Deriving realistic examples of statistical problems for life science majors using ChatGPT


By Jacob Andros (Texas A&M University)


Information

At large state universities, introductory statistics courses typically contain hundreds of students from a wide variety of different majors. At Texas A&M University, one of the largest introductory statistics courses (Stat 302) is offered to students in any life science major that requires statistics-related coursework. These majors include animal science, biomedical science, ecology, food science, public health, and wildlife management, among many others. However, most of the exercises we do in class or on homework assignments contain fairly generic examples that have little to do with the life sciences. Because so many exercises are not applicable to life science majors, they often fail to demonstrate the usefulness of statistics in their specific fields of study.

While we have been working to develop new, major-specific problem banks for each chapter in Stat 302, we have found that LLM's like ChatGPT are able to quickly and reliably produce logical examples of basic statistical problems with contexts that relate to specific fields of study in the life sciences. Not only are the examples coherent and clear, they also choose numeric values that make sense in each specific context. This saves statistics instructors like us (who may not have any expertise in, say, ecology) from having to spend time researching a specific ecology topic and selecting reasonable numbers that make sense for the problem.


register