Assessment of distractor efficiency of MCQS in item analysis.

Abstract

Objective: To assess the distractor efficiency of the multiple-choice questions and find out the structural flaws in the items that negatively impact the distractor efficiency. Study Design: Retrospective Observational study. Setting: Department of Pathology, Rawal Institute of Health Sciences (RIHS) SZAMBU, Islamabad. Period: April 2021 to June 2021. Material & Methods: This study was conducted at Rawal College of Medicine, RIHS Islamabad. The data was based on the item- analysis report from a sendup exam MCQ paper (2020) of 3rd year MBBS class. Distractor efficiency of total of 140 MCQs was analyzed. Distractor efficiency was categorized as low in MCQ with 3-4 non-functional distractors, medium with 1-2 nonfunctional distractors and high if there are no non-functional distractors. These MCQs were investigated with reference to item writing guidelines proposed by Haladyna et al. The flaws identified were grouped as, within option flaws, alignment flaws between options and stem/ lead-in and other flaws. Results: Out of 140 MCQs, distractor efficiency was high in 58 (41%), moderate in 75 (54%) and low in 7 (5%). The item writing flaws identified in moderate to low DE items were Alignment flaws between distractors and stem/ lead-in were linguistic cues 8(10%), logic cues 10 (12%) and irrelevant distractors 5(6%). Item flaws within the distractors were non homogenous length 2 (2.4%), non-homogenous content 5(6%) and distractors with true and false statements 3(4%). Flaws that were categorized in other than distractors were low cognitive level items 13 (16%) and un-necessarily complicated stems were 8(10 %). Conclusion: This analysis found out distractor efficiency of multiple-choice items was moderate to high. The major flaws that negatively impact distractor efficiency include the distractors with low cognitive level, unnecessary complicated stems, logic cues, linguistic cues, irrelevant distractors and distractors with non-homogenous length and content.