Determinants of Pupils’ Poor Academic Performance in Basic Education Certificate Examination in Mathematics: A Multinomial Logistic Regression Model - International Journal of Academic Research in Business, Arts & Science | IJARBAS

Issue: Issue: 1,, Volume: Volume: 1,, Year: Year: 2019.

Determinants of Pupils’ Poor Academic Performance in Basic Education Certificate Examination in Mathematics: A Multinomial Logistic Regression Model

Date of Publication : 07, Aug, 2019

Author: 1 Foster Kwashie Dugble (MPhil)

Co Author: 2 Eliot Kosi Kumassah (MPhil) , 3 Damianus Kofi Owusu (MSc)

The aim of the present study is to examine and to model in multinomial logit some selected demographic, homes-related, school-related, teachers-related and pupils-related factors as predictors on pupils’ Basic Education Certificate Examination (BECE) performance in mathematics as categorical response variable (upper grade, average grade, and lower grade). A combination of systematic and simple random sample of 62 pupils was selected from a cohort of BECE candidates of University Junior High School in Cape Coast municipality. The instrument for data collection was the interview guide. It was then analyzed to calculate the significant parameter coefficients from maximum likelihood method at 95% Confident Interval using IBM SPSS version 21. Two multinomial logit models were developed. The findings show that the Age of pupils, and Class size were significant in the two models. The findings in the first model shows that the incidence of upper grade in BECE mathematics is largely dependent on the Age of pupils, and Class size with younger pupils exhibiting significantly upper grades than older pupils, and with pupils in less class size showing significantly upper grades than those in large class size respectively. In the second model, the occurrence of average grade in BECE mathematics is as well largely dependent on the age of pupils, and class size, with younger pupils exhibiting significantly average grades than older pupils, and with pupils in less class size showing significantly average grades than those in large class respectively. Another significant predictors in the first model are; Gender, School location, and Homework, and in the second model, Parental educational level. The overall prediction accuracy (robustness) of the model is 71.0%. The study concluded that the pupils who lack the benefit of the factors especially (school location, Class size-wise, self-homework undertaking, and Parental education) are with a high probability of recording poor performance (lower grades) at BECE mathematics.

 

DOI: 10.5281/zenodo.3363501

Sharing is caring!

Keywords : Multinomial Logit Model, Odds Ratio, Academic Performance Basic Education Certificate Examination,

Serial: 4 Download Page: 49 - 74