International Edition
Latest News
Technology

Correction: JMIR Study Fixes Miscalculated F Statistics in AI Adoption Research

JMIR Publications corrected a 2023 study on artificial intelligence adoption in health care after the authors identified miscalculated F statistics and degrees of freedom in their manipulation check, the journal announced Thursday. The correction notice, published November 20,…

JMIR Publications corrected a 2023 study on artificial intelligence adoption in health care after the authors identified miscalculated F statistics and degrees of freedom in their manipulation check, the journal announced Thursday.

The correction notice, published November 20, 2026, in the Journal of Medical Internet Research, replaces the original manipulation check section and Table 1 with item-level validation metrics from the raw dataset. The underlying partial least squares path models remain sound and unaffected by the recalculation, the authors stated. Glorin Sebastian, Amrita George, and George Jackson Jr. originally reported a significant mean difference between communication strategy conditions using a three-group ANOVA with incorrect parameters F1,144=43.79; P. Participants in the pathos condition (mean 3.37, SD 0.90) reported scores different from those in the ethos (mean 3.52, SD 0.97) and logos (mean 3.43, SD 0.89) conditions, the original text stated. The corrected analysis reveals a four-group one-way ANOVA conducted on a reverse-scaled relevance item — “This product is irrelevant to me” — produced a statistically significant difference across control, pathos, logos, and ethos conditions (F3,197=2.86; P=.04).

The overall composite involvement scale did not reach significance, the correction notice states. The study examined whether rhetorical strategies — ethos, pathos, and logos — in promotional advertisements could overcome barriers to AI product adoption among patients. Researchers collected responses from 150 participants via Amazon Mechanical Turk in the original experiment.

Pathos-laden promotions improved AI product adoption by nudging users’ trust (n=52; β=.532), the original study abstract indicates. The correction has been resubmitted to PubMed, PubMed Central, and other full-text repositories because it was made after initial submission to those databases. The article is distributed under the Creative Commons Attribution License, permitting unrestricted use with proper citation of the original work published March 13, 2023.

About the author: Technology Desk

Technology Desk at Archynewsy.