공변수 IRTree 모형을 적용한 대학전공선택검사 공학계열 문항의 성별에 따른 응답 경향성 분석

Abstract

This study utilized the IRTree model to analyze specific response style in participants' item responses, irrespective of the content of the items being measured. The research instrument consisted of 30 items assessing engineering field activity preferences in the Korean College Major Interest Inventory (KCMII-2). A total of 9,069 data records were analyzed, excluding non-responses and repeated identical responses. The study explored the response process and response styles of participants in the engineering field activity preference items and investigated whether there were significant differences in response processes and response styles based on gender. The research findings indicated that among the graded response model (GRM), linear IRTree model, and nested IRTree model, the most suitable model was the nested IRTree model. The nodes within the model represented positive preference, positive extreme response, and negative extreme response for each item. Examination of the latent ability distribution for each node revealed that male students had higher latent abilities for positive preference and positive extreme response, while female students had higher latent abilities for negative extreme response. The IRTree model analysis, incorporating gender as a covariate, demonstrated that gender had a significant impact on the latent abilities at each node. In conclusion, this study suggested that applying the covariate IRTree model to various educational and psychological measurement tools can lead to more meaningful research results and their practical utilization in the future.