Research On Score Corection Method Of Online Multi-peer Mutual Evaluation Based On Learner’s Score Deviation

In recent years,peer evaluation has achieved fruitful academic results in the reliability and validity of evaluation,learning output,and knowledge contribution.The design of peer evaluation system has been gradually improved,and the scope of application has gradually expanded from second language acquisition to program design and In the study evaluation of other various professional courses.This research has compiled and cleaned up field experiment data for many years.Through the analysis of the consistency of external scores,internal scores,mutual evaluation scores and expert scores of learners in peer evaluation,it is found that there are a number of sources in the peer evaluation community.The learners’ bad scoring deviation behaviors have a certain impact on the reliability and validity of the data,which not only affects the quality of mutual evaluation data,but also weakens the teaching effect of the course.Therefore,this study starts with learner scoring deviation,identifying different learner scoring deviations in two dimensions: the reviewer as the object and the author as the object,and classifying the scoring deviation as motivated by the learner according to the characteristics of the learner’s scoring behavior.The motivational bias caused by competition and the extreme bias caused by the learner’s unconscious grading preference.Based on the experimental performance of different types of scoring deviations,this study created a normal assignment correction model for extreme deviations,and a behavior factor correction model for motivational deviations.The previous field experiment data was brought into two correction models for experimentation.Calculate the root mean square error,the average absolute error,and the Pearson coefficient of the corrected score and the expert score.It is found that the two correction models can reduce the negative impact of the learner’s score error on the mutual evaluation data to a certain extent,and both can improve the data Validity,and the correction effect of the normal assignment correction model is better than the behavior factor correction model.The research results of this study have enriched the research on peer evaluation in scoring correction,and at the same time provided suggestions for the improvement and optimization of peer evaluation scoring system,and provided relevant thinking for follow-up research.