Questionable mediated effect of a positive attitude toward AI on AI literacy: A comment on Wang and Niu (2026)
Year: 2026
Type: preprint
Abstract: Drawing on the covariance matrix reported by Wang and Niu, I reassessed the empirical basis of their mediation model linking a positive attitude toward AI to AI literacy. I demonstrate that the observed covariance structure can be reproduced by a theoretically different latent-variable model that does not rely on the directional mechanism proposed in the original article. In this alternative account, a positive attitude toward AI, AI self-efficacy, motivation to use AI, and self-rated AI literacy are conceptualized as indicators of a broader construct, core self-evaluation. The results illustrate a broader inferential problem in psychological research: multiple theoretically plausible models may provide equally adequate explanations of the same correlational pattern. Consequently, observational associations alone offer limited grounds for preferring one explanation over competing alternatives. Researchers should therefore routinely evaluate alternative models when drawing inferences about psychological mechanisms from nonexperimental data.
Source: PsyArXiv (OSF Preprints)
Author: Kimmo Sorjonen
Language: English
Subfield: Education
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7221611747