Constructs should do work: Reply to Forbes (2026), Lorenzo-Luaces and Buss (2026), Markon (2026), and Paulus (2026)

Forbes (see record 2027-91416-003), Lorenzo-Luaces and Buss (see record 2027-91416-005), Markon (see record 2027-91416-004), and Paulus (see record 2027-91416-002) provide commentries on the set-theoretic reconceptualization of psychiatric taxonomy presented in Fisher (see record 2026-72441-001). In the following, I respond to critiques concerning construct validity, the role of latent variables, severity measurement, and the predictive value of discrete versus continuous approaches. I argue that latent variable models presuppose that constructs should exist independently of their external relationships, that factor scores compress configural information that is clinically meaningful, and that the putative superiority of continuous over discrete measurement is a property of the linear modeling ecosystem rather than a mathematical absolute. I conclude that set-theoretic and dimensional approaches address fundamentally different questions and that the former is better suited to the actionable demands of clinical classification. (PsycInfo Database Record (c) 2026 APA, all rights reserved).