Using latent class analysis to justify a latent continuum in item development.

We illustrate the use of exploratory latent class analysis (LCA) in item development or item adaptation. In particular, we focus on the use of LCA as a method for justifying the assumption that the target construct reflects a latent continuum, a key assumption in both classical test theory (CTT) and item response theory (IRT) analyses. We show how LCA statistics, in particular contrasts of predicted probabilities for each item based on the logic of Kelley's discrimination index, are informative about whether items behave in a manner consistent with the assumption of a latent continuum. LCA is particularly valuable because it makes relatively minimal assumptions and can cope with important design issues, such as planned missingness or mixed item types, that confound CTT. Moreover, LCA output is understood more readily by nontechnical stakeholders such as content experts, teachers, or clinicians than the results of most IRT analyses. For simplicity, we focus on binary scored test items, although extensions to polytomous data are not difficult. To illustrate, we use simulated data that mimic two pilot studies of a prototype mathematics assessment. Our code is available for readers who want to run their own examples. (PsycInfo Database Record (c) 2026 APA, all rights reserved).