This repository is under review for potential modification in compliance with Administration directives.
Hidden Confounding Variables
Every month, ETAI will be sharing a term or concept of the month that is related to ethical issues in biomedical research. The term of the month is part of our module’s milestones relating to engagement and reflection on ethics with the AI-READI team.
Term of November 2025: Hidden Confounding Variables
Hidden confounding variables are factors that influence both the outcome of interest and the measured variables in a study or model, but whose presence or effect is not known to the researcher or system developer. Because they go unrecognized, these hidden confounders can lead to misleading or spurious relationships—for example, a computer vision system distinguishing cows from whales based on background color rather than the animals themselves.