This study designs and evaluates a fairness measurement system that embeds teacher accreditation metrics into the evaluation logic of a foreign language Educational Technology platform.The system defines a composite F...This study designs and evaluates a fairness measurement system that embeds teacher accreditation metrics into the evaluation logic of a foreign language Educational Technology platform.The system defines a composite Fairness Metric Index integrating pedagogical quality,assessment literacy,and professional reflection,and is deployed with 264 English as a Foreign Language teachers across 18 institutions,covering 38,412 lessons and 126,507 feedback events.Compared with the platform's original performance index,the integrated model increases the correlation with human accreditation ratings from 0.46 to 0.71 and raises the conditional R squared of mixed models from 0.37 to 0.62.Group-wise parity loss between institutional and experience groups falls by roughly fifty percent,and agreement with an independent accreditation panel improves from kappa 0.51 to 0.74.These results indicate that embedding accreditation constructs into algorithmic scoring can simultaneously improve alignment with professional standards and reduce systematic unfairness in teacher evaluation.展开更多
文摘This study designs and evaluates a fairness measurement system that embeds teacher accreditation metrics into the evaluation logic of a foreign language Educational Technology platform.The system defines a composite Fairness Metric Index integrating pedagogical quality,assessment literacy,and professional reflection,and is deployed with 264 English as a Foreign Language teachers across 18 institutions,covering 38,412 lessons and 126,507 feedback events.Compared with the platform's original performance index,the integrated model increases the correlation with human accreditation ratings from 0.46 to 0.71 and raises the conditional R squared of mixed models from 0.37 to 0.62.Group-wise parity loss between institutional and experience groups falls by roughly fifty percent,and agreement with an independent accreditation panel improves from kappa 0.51 to 0.74.These results indicate that embedding accreditation constructs into algorithmic scoring can simultaneously improve alignment with professional standards and reduce systematic unfairness in teacher evaluation.
