Human Factors Evaluation Methods for Embodied Interaction Mediators: A Systematic Literature Review and Exploratory Meta-Analysis

Human factors constructs such as workload, comfort, affect, trust, and agency are widely studied in HCI and HRI, yet their evaluation remains fragmented across embodied interaction mediator (EIM) communities. This systematic review synthesized 62 empirical studies of task-oriented robots, social robots, VR/AR mediators, and wearable mediators using a four-dimensional taxonomy of physical ergonomics, cognitive ergonomics, affect, and belief-related states. Two exploratory three-level meta-analyses were conducted. Meta-analysis I examined task-oriented-robot cognition studies (k = 8; g = 0.73, 95% CI [0.44, 1.02]), while Meta-analysis II examined target affective directions (k = 7; g = 0.67, 95% CI [−0.05, 1.40]). Neither showed a reliable objective–subjective difference. The findings are hypothesis-generating and reveal a Method-Construct Alignment Gap, supporting a construct-driven heuristic linking target constructs to primary and complementary evaluation methods.