중학생의 성별과 과학시험유형에 따른 더닝-크루거 효과와 과잉확신편향 분석 연구

Abstract

Objectives This study aims to investigate potential gender differences in middle school students' self-assessment of their abilities during academic exam situations. Focusing on the potential emergence of overconfidence bias when predicting their own exam scores, we examined how the Dunning-Kruger effect interacts with an inherent trait-gender, and the illusion of competence based on the type of science exams (multiple-choice/shortanswer questions) as observed in prior research. The objective is to collectively ascertain whether such gender- based distinctions exist.
 Methods Individual data were collected through a survey of 2nd-grade students at a coeducational middle school located in Gyeongsangbuk-do, during the week preceding the midterm and final examinations. Utilizing the actual test scores achieved, the study employed t-tests and regression analysis to scrutinize the gender-based distribution of academic performance, the variation in performance between genders across different test types, and the gender-associated dissimilarities in perceptions of competence.
 Results The Dunning-Kruger effect, characterized by the illusion of competence, was observed as a negative phenomenon, evidenced by the bias score—a metric reflecting the disparity between actual scores and anticipated scores. Among both male and female students, a positive correlation was established between higher expected scores and higher actual scores. Additionally, a noteworthy negative correlation emerged between actual scores and biased scores, signifying that students achieving higher actual scores exhibited lower biased scores. Specifically, male students demonstrated an overestimation tendency within the ‘low’ and ‘middle’ quartile groups, whereas female students displayed an underestimation tendency within the ‘highest’ quartile group. Expected scores exhibited no dependency on either test type or gender, yet significant effects were discerned for actual scores and bias scores in relation to gender. This confirmed that the magnitude of overestimation was greater among male students than their female counterparts, whereas the extent of underestimation was comparatively less pronounced in female students compared to male students.
 Conclusions This study successfully validated the manifestation of the Dunning-Kruger effect based on students' gender. Male students exhibited a propensity for overestimation driven by overconfidence bias, and it was substantiated that this overestimation was more pronounced in short-answer type questions compared to multiple- choice questions, irrespective of gender. Recognizing the inherent gender distinctions among students, it is recommended to propose a self-directed learning approach aimed at enhancing metacognitive abilities. Additionally, future research endeavors should focus on mitigating test-related anxiety factors through follow-up investigations.