AN EMPIRICAL INVESTIGATION OF GOODNESS-OF-FIT OF 1, 2, AND 3-PARAMETER LOGISTIC MODELS TO ACHIEVEMENT TEST DATA IN ENGLISH LANGUAGE

Ifeoma, M and Ugwoke, M. E and Iluobe, Ovekairi Irene (2018) AN EMPIRICAL INVESTIGATION OF GOODNESS-OF-FIT OF 1, 2, AND 3-PARAMETER LOGISTIC MODELS TO ACHIEVEMENT TEST DATA IN ENGLISH LANGUAGE. Journal of Educational Assessment in Africa, 13. pp. 14-27.

Text 2017 AEAA PUBLISHED MARK UGWOKE.pdf Download (554kB)

Abstract

This study investigated the goodness-of-fit of 1, 2 and 3-Parameter Logistic Models of Item Response Theory (IRT) to Achievement Test Data in English Language. The increased use of achievement tests in selection, promotion and awards of certificates inevitably brings attention to the quality and fairness of achievement testing. To adequately address these issues require sophisticated mathematical methods. The traditional Classical Test Theory (CTT) approaches that evaluated psychological measures at the total test scores have been complemented by more recent IRT approaches that focus on item level data. The ex-post facto research design was adopted for the study. The sample consisted of 3,000 examinees’ responses which were randomly selected from Edo and Delta States of Nigeria in English Language Multiple-Choice Test Items conducted by the National Business and Technical Examinations Board (NABTEB), in 2014, 2015 and 2016. The test is valid and reliable because NABTEB conducts standardized tests. Three research questions guided the study and three hypotheses were tested. Data analysis was carried out using eirt -Item Response Theory Excel Assistance Version 3.1 Software. Pearson Chi–Square was used to test the hypotheses at 0.05 significant level. Findings from the study revealed that there was no significant difference among the 1, 2 and 3 -Parameter Logistic Models fit in 2014 and 2015 NABTEB English Language Multiple-Choice Test Items while there was a significant difference among the 1, 2 and 3 -Parameter Logistic Models in 2016. Based on the findings it was concluded that the 1, 2 and 3 -Parameter Logistic Models fits the data across the three years under study, none was empirically superior to others. It was recommended among others that examining bodies should make sure that selected models fits the data to be confident of results generated from such data.

Item Type:	Article
Subjects:	L Education > L Education (General)
Divisions:	Faculty of Arts > Faculty of Law > Faculty of Management and Social Sciences > Faculty of Education
Depositing User:	mrs chioma hannah
Date Deposited:	01 Jul 2024 10:56
Last Modified:	01 Jul 2024 10:56
URI:	http://eprints.gouni.edu.ng/id/eprint/4328

Actions (login required)

View Item