Determining minimum test length for reliable assessment using generalizability theory: Evidence from an economics multiple-choice test

Ossai Elizabeth Ngozika

African Educational Research Journal
Published: March 2 2026
Volume 14, Issue 1
Pages 128-133
DOI: https://doi.org/10.5281/zenodo.18837663

Abstract

This study investigated the reliability of a multiple-choice Economics test for senior secondary school students using Generalizability Theory, with the goal of determining the minimum test length required for dependable assessment. A stratified random sample of 350 SS2 students was drawn from 3,869 students in government secondary schools in Nsukka Education Zone, Enugu State, Nigeria. The Economics Multiple Choice Test, consisting of 50 items, was administered twice to capture occasion-related variance. Data were analyzed using EduG software version 6.1e. The Generalizability study estimated variance components for students, items, occasions, and their interactions, while the Decision study simulated changes in test length to identify the minimum number of items needed for acceptable reliability. Results indicated that student item interactions and residual error were the major sources of score variability, whereas item and occasion effects were minimal. Relative reliability reached acceptable levels with 51 items, but absolute reliability required 52 items. The study demonstrates that carefully designed test length is critical for reliable assessment and recommends using Generalizability Theory to guide test development, ensuring accurate and trustworthy evaluation of student performance.

Keywords: Generalizability theory, test reliability, multiple-choice test, economics education, test length.

Full Text PDF

This article is published under the terms of the Creative Commons Attribution License 4.0