Validating the text-to-measure pipeline: A procedure-based approach to creating measures of latent concepts with supervised machine learning

• In Event: Machine Learning Methods for Text

Brief Overview

We propose a standard procedure to measure a latent trait using a supervised learning method to solve the issue of extensive heterogeneity across extant studies in measurement and validation processes when constructing such measurements.

Authors

• Ju Yeon Park, The Ohio State University

• Jacob Michael Montgomery, Washington University in St Louis