Key points are not available for this paper at this time.
Abstract Open science adoption among Generation Z students in a developing country, with distinct educational culture, presents a critical test of a wide range of technology acceptance frameworks. This study examines factors predicting open science resource utilization (measured through self-reported frequency of diverse uses in academic contexts) among 1,422 Vietnamese undergraduate students using Bayesian regression analysis, which enables probabilistic inference and robust model comparison, adding to the methodological rigor and novelty to the literature in this area. We tested a theoretical framework integrating a multitude of factors such as capabilities, perceived benefits, institution-centered, value-alignment, etc., through progressive model building. Model comparison via WAIC identified the main effects model as optimal. Results revealed Technical Self-efficacy (β = 0.254) and Open Science Self-efficacy (β = 0.205) as strong predictors, with Institutional Support (β = 0.117) and Attitude (β = 0.119) showing moderate effects. Language Self-efficacy showed a weak but credible effect (β = 0.089). Contrary to Western technology acceptance models, Perceived Benefits showed uncertain influence (β = 0.054, 89% HPDI includes zero), and Value Alignment demonstrated no credible effect (β = 0.013). Hypothesized interaction effects between Technical Self-efficacy and Perceived Benefits were not supported by model comparison, suggesting additive rather than synergistic adoption mechanisms. These findings challenge the digital native assumption and reveal that Vietnamese students adopt open science for practical reasons, relying on skills and institutional support rather than ideology. Interventions in resource-constrained settings should therefore prioritize technical skill development and visible institutional support, rather than relying primarily on value-based or normative appeals. This pragmatic approach has implications for advancing open science in the age of AI.
Le et al. (Sat,) studied this question.