A Tutorial on Unidimensional Unfolding: From Automatic Item Generation to Insightful Inferences

Unfolding theory maps individuals and stimuli into a common latent space, reflecting preferences through proximity. Unlike traditional psychometric methods, unfolding models are complex and computationally intensive, leading to their underuse despite their potential for nuanced insights. This study presents a tutorial on unidimensional unfolding, using ChatGPT to create items and providing R functions for analysis. Structured sequentially, we first describe what AIG is, highlighting the possible improvements allowed by large language models in general. Subsequently, the unidimensional unfolding theory, synonymous with the Coombs system, is introduced as the foundation for our AIG procedure. The following section is a description of our rationale on the approach to training ChatGPT in the creation of items via the binary tree structure, concurrently instructing the reader on how to analyze the data. We then present a practical case study focusing on the measurement of attention-seeking tendencies, with a comprehensive presentation and commentary on the necessary R code. The paper concludes with reflections on future directions for research endeavors.