Study On The Rater Effects In Chemistry Experiment Skills Test

Experimental Skill Test is an important component of chemical academic quality evaluation,while the effect of raters affects the objectivity and accuracy of the results of chemical experimental skills.Rasch model based on item response theory has the characteristics of Item independence and sample independence.And it can provide objective,equidistant scales.Hence,in this paper,author used the Rasch Model to monitor the rater effects in the Chemistry Experiment Skills Test,and proposed a series of strategies to reduce the rater effects.Based on extensive literature research,The experimental operation test paper was prepared which was checked by educational and didactic experts.The raters were trained and two rounds of test were conducted.The subjects tested were freshmen and sophomores of a normal university of chemistry.In the first test,seven raters scored 225 examinees in groups and this work lasted one day.Through the analysis of variables such as Infit,Outfit,and t,as well as the White charts and the severity wave charts,Separation is not between 0 and 2,which means that outer consistency among 7 raters is poor;the four raters Infit MnSq are not within a reasonable range of 0.4 to 1.2,indicating that themselves consistency is poor;There were 74 significant deviations between the raters and the examinees;Most raters occurred serious rater drift.Therefore,the second round test was performed after the rater retraining and a trial test.And we improved test strategies.39 examinees were scored by 6 raters.The inner consistency and outer consistency of the raters have been greatly improved.The phenomenon of rater drift has been significantly reduced.The study found that there was no significant rater effects difference between genders.Conclusion: 1.Rasch model can monitor rater effects in chemical experiment skills tests.2.In order to reduce the rater effects on the chemical experiment test,the following strategies can be implemented: Selection and training of raters before testing;Trial test are needed before testing;Examinees shouldn’t be tested in big groups;Adequate break are necessary;Raters should sit as far as possible.