Construct jangle or construct mangle? Thinking straight about (nonredundant) psychological constructs

Abstract

Psychological science aims to make the abstract measurable and quantifiable. As psychologists it is our challenge and charge to capture complex abstractions (e.g., authoritarianism, prejudice, depression) accurately and with transparency. But recent concerns have been raised about the proliferation of constructs across psychology sub-disciplines, with construct redundancy now rife. Critics charge that we do not take seriously construct validity, especially discriminant validity, which exacerbates the replication crisis. Here the author outlines the problem and discusses at a conceptual level how latent modeling can aptly capture constructs and their interrelations without error, isolating and helping to circumvent construct validity problems. At the core of the issue lies a mathematical reality that seems to be largely ignored in psychology: if correlations within and between indicators of constructs are roughly comparable, their latent factors will correlate near-perfectly and thus be redundant. Thoughts about how the field arrived at this juncture are discussed, along with a recommendation to avoid using cute labels (e.g., jingle-jangle fallacies) to represent very serious problems (better labeled as construct redundancy fallacies). Recommendations for thinking straight about constructs, their validation, and their uniqueness, are offered.