A taxometric approach to base-rate estimation and idiographic classification in psycholegal research.

Abstract

Taxometric analysis employs multiple, nonoverlapping statistical procedures to estimate parameters that characterize latent categories (e.g., base rates). Consistency among these estimates can inform substantive inferences about latent variables and facilitate idiographic classification. We provide a sketch of a taxometric research program to estimate guilty-suspect base rates in criminal justice and legal systems and use this sketch to explore the possible benefits of taxometric investigations for science and public policy. We investigated whether taxometric analysis can accurately estimate base rates and facilitate idiographic classifications under conditions psycholegal researchers might face. We demonstrate taxometric analysis on simulated data to detect latent categories, estimate their base rates, and classify individual cases. Our simulations show that taxometric analysis can accurately estimate taxon base rates. Specifically, estimated base rates differed from simulated base rates by less than 3%. Further, idiographic classification rules derived from taxometric analysis accurately classified individual cases in additional data sets, with positive predictive values and negative predictive values exceeding .85. If legal categories of interest represent nonarbitrary classes, taxometric methods afford an analytic approach by which researchers can use fallible indicator variables to estimate their base rates and develop algorithms for legal classification. We discuss potential objections to the taxometric approach and identify important avenues for future research and development in psycholegal applications of taxometric methods. (PsycInfo Database Record (c) 2022 APA, all rights reserved).