When I taught Quantiative Methods last spring, a colleague picked up a paper of mine from the printer and came looking for me. “You’re doing replications with the undergrads?” she asked. “Why?” I looked at her and without thinking simply replied, “Why not?” Replicating studies is considered a best practice of sorts in graduate level methods training. None of the reasons given there – teaching disciplinary norms, emphasizing the importance of transparency in research, etc. – fails to hold in the undergraduate context. If anything, our undergraduates have more need of those objectives than our grad students, who will have them repeatedly reinforced across multiple classes. For most of us who teach undergrads, one methods class is all we’ve got, and we need to make it count.
For most of us, part of the objective of a research methods course is to introduce students to the ways of thinking and doing that characterize social science, and social science research especially. It’s a key point of socialization into the discipline, where they go from being students of politics to being students of political science. Replication, even in the highly simplified form of a class assignment, brings home many of the key points we try to make about the need for transparency in data and methods and about best practices in selecting and reporting models. How is this measured? Are higher numbers better, or lower ones? Does including this variable matter? All of these are questions that require close reading and an interaction with the source text that many of them have never experienced before.
Replication exercises are thus a form of active learning. They ask students to apply what they’ve learned in the statistics part of the class to what they’ve learned in the reading, and to try to reproduce the same set of results. It challenges them to connect pieces from disparate parts of the class and to write about statistics in ways that are typically new to them. From a pedagogical perspective, depending on their design, they can serve almost all of the same functions that a standard research paper assignment does, but with far less grading and prep work on your part (no draft literature reviews! Hooray!!) Or, again depending on design, they can simply be one form of problem set given as a homework exercise.
In my next post (next Friday), I’ll have one of my replication assignments to share, and in part 3 (mid-April), I’ll share some best practices for developing your own replication assignments. But in the meantime, consider whether you can combine teaching writing and research skills with statistics skills by assigning a replication in your own course.