The General Linear Model and Minimal Research Compendiums: An Approach to Assure Statistical Validity in Digital Humanities Research

Abstract

Abstract There has been little research on what specific statistical techniques humanities scholars should focus on in their training to assure reproducibility. This article addresses this gap by illuminating the importance of the general linear model (GLM) in statistical research. Drawing on best practices in the social sciences asking to make the GLM the cornerstone for statistical training, this article provides examples of how the GLM works and its assumptions. It then details how to create a “minimal research compendium” focused on a series of steps in R to assure reproducibility when using the GLM and how the GLM underlies more advanced statistical techniques. Overall, the goal of the article is to encourage humanities scholars to join scientists in developing a culture that supports best practices for reproducibility and statistical accuracy.