The Multinomial Logistic Regression Model’s Utility to Assess Parameters in Predicting Junior High School Students’ Preference for Selected Mathematics Topics

This study predicts the preference for three mathematics topics among Junior High School students. Four hundred (400) Junior High School (JHS) students, comprising two hundred and eighteen (218) males and one hundred and eighty-two (182) females selected from Junior High Schools in a school district in Ghana, participated in the study. The multinomial logistic regression model, consisting of three unordered outcome categories (i.e., Relations and Functions, Algebraic expressions, and Linear equations), with predictor variables comprising continuous, nominal, and ordinal variables were used for the study. For Relations and Functions, the results indicated that Math self-concept, Arithmetic ability, Motivation, Instructional strategies and methods, Asanti, Fanti, Ga, and Ewe, were statistically significant (p < .05). Hence, for a unit increase in the Math self-concept measure, a student is 5.82 times more likely to be in the Relations and Functions topic category than in the Linear equations topic category, controlling for the other variables. Again, a female student is 1.15 times more likely than a male student to be in the Relations and Functions topic category than in the Linear equations topic category, controlling for other variables. Similarly, for Algebraic expressions, the results indicated that Math self-concept, Math attitude, Motivation, Instructional strategies and methods, female, Asanti, Fanti, Ga, and Ewe, were statistically significant (p < .05). Thus, for a unit increase in the Math self-concept measure, a student is 2.63 times more likely to be in the Algebraic expressions topic category than in the Linear equations topic category, controlling for the other variables. Again, a female student is 3.75 times more likely than a male student to be in the Algebraic expressions topic category than in the Linear equations topic category, controlling for other variables. These significant predictor variables influencing students’ preference for mathematics topics, add to the body of literature on the factors affecting decision-making in mathematics teaching and learning.

Keywords: Relations and Functions, Algebraic Expressions, Linear Equations, Categories, Multinomial Logistic Regression Model.