Developing and Validating A Students’ Underperformance Indicator Instrument For The Ethiopian Secondary School Leaving Certificate Examination: Evidence From Eastern Ethiopia

Abstract

This study aims to develop and validate a multi-item instrument designed to measure the students’ underperformance indicators on the Ethiopian secondary school-leaving certificate examination. Accordingly, a cross-sectional retrospective design was adopted. Data were collected from 602 students enrolled in TVET and remedial programs across four major cities in eastern Ethiopia: Chiro, Harar, Dire Dawa, and Jigjiga, using stratified random sampling. These participants were among those who failed to meet the 50% threshold for the 2022/23 national examination and enrolled in either TVTE or remedial programs. In the first instrument development stage, literature reviews, expert opinions, and content evaluations were conducted, and a questionnaire consisting of 30 items, divided into three components, rated on a 5-point Likert scale was developed. In the subsequent validation stage, the total sample was divided into two halves (n=301), with one half designated for either EFA or CFA and analyzed using the JASP for Windows 10 application. The results of the EFA indicated a sampling adequacy (KMO = 0.93; Bartlett’s test, χ² = 4572.82, p < .001), leading to retaining all 30 items with a three-factor solution, all having item loadings of ≥ .60. The CFA supported the proposed structure, depicting acceptable fit indices: (χ²/df = 1.05, RMSEA = 0.013, SRMR = 0.041, CFI = 0.995, and TLI = 0.995). The reliability and validity of the instrument were established with Cronbach’s alpha ranging from 0.909 to 0.916, composite reliability greater than 0.70, and average variance extracted remaining ≥ 0.50. Discriminant validity was also confirmed. Generally, the finalized instrument exhibited strong psychometric properties, making it a valid and reliable tool for assessing students’ underperformance in Ethiopia's secondary school national examination context.