CCSMHSS Analysis And DIF Detection Based On IRT

Objective: College students,as an important force in the future construction of the motherland,colleges and universities also attach great importance to the mental health status of students while cultivation them.As the most direct and efficient evaluation tool,the mental health assessment scale generally has the problem of a large number of items.Since it is difficult for the traditional classical measurement theory to analyze each item more in-depth,the analysis of the measurement results is relatively simple.Therefore,this research aims to adopt the Item Response Theory,using its more precise discrimination,difficulty,information and differential item functioning,to further screen out more representative measurement items in the "Chinese College Student Mental Health Screening Scale".The purpose is to improve the accuracy of the scale,simplify scale,and analyze the Differential Item Functioning of the new scale.It explores the fairness of the test by studying the different responses of different groups of subjects with the same trait level on the same item,provides guidance value for being able to better interpret the measurement results,and offer analysis methods for colleges and universities to use mental health assessment.Methods: This research randomly selected 2000 psychological evaluation data of grade2020 undergraduate freshmen from a university in Sichuan Province,deleted test papers of invalid answering and answering time less than 100 s,and obtained a total of 1976 valid data.By using item response theory,and adopting the ltm package selection grade response model in R language,we selected items,and obtained a simplified new scale;then we used lordif package and MULTILOG 7.03 to perform DIF analysis on the new scale.Results: After item analysis,the new scale contains 70 items,and the R language was used for DIF detection.The results show that there are 21 items with gender DIF: 7,10,12,14,16,25,33,34,38,40,41,42,45,46,47,50,52,57,58,63,69;there are 10 items with DIF in the origin of student: 17,24,28,29,30,32,37,52,64,67;there are 13 items with only-child DIF: 1,2,6,11,12,22,27,32,33,34,52,56,64.Conclusion: By using the item response theory that has been proved by a large number of studies to have absolute advantages in the revision of the scale,we simplified the "Chinese College Student Mental Health Screening Scale" and obtained,a new scale with high-precision,high-discrimination and relatively good amount of informatio.We also conducted DIF analysis on it,obtained items with DIF in the three groups of gender,birthplace,and whether or not the only child,and conducted a detailed analysis of these items with DIF.This research provided theoretical guidance for the revision of the mental health scale,gave more diversified thinking directions for the interpretation of test results,It also conducted exploration for subsequent computer adaptive tests such as mental health measurement,psychological trait measurement and artificial intelligence consultation.