CONSTRUCTION AND USE OF MULTIDIMENSIONAL STUDENT MATHEMATICS ENGAGEMENT SCALE IN PREDICTING MATHEMATICS ACHIEVEMENT AMONG SENIOR SECONDARY SCHOOL STUDENTS IN EKITI STATE, NIGERIA

Abstract

A recurrent argument among educational psychologists globally is the number of dimensions that Student Mathematics Engagement Scale (SMES) should have. Some authors argued that it should be one dimension, while others posited two or three dimensions. Currently available SMES are either 1-dimensional or 2-dimensional and rarely more. With the introduction of more robust statistical methods, it is possible to have a multidimensional SMES; however this has not been fully explored as extant literature shows. This study was, therefore, designed to construct a reliable 6-dimensional SMES and to examine its predictive power on mathematics achievement among secondary school students in Ekiti State, Nigeria.

The study adopted a survey design. Three phases and three sets of samples were involved. Phase 1: Pilot testing of the initial pool of 100 items measuring SMES. Phase II: Calibration and selection of SMES items and Phase III: Usage of the SMES. All the three senatorial districts in Ekiti State were sampled, and four Local Government Areas (LGAs) were randomly selected from each senatorial district. For Phases 1 and III, three each of private and public Senior Secondary Schools (SSS) were randomly selected from each of the sampled LGAs. The sample sizes were 1008 and 1032 SSS2 students for Phases 1 and III respectively. For phase II, four each of private and public SSS were randomly selected from each of the sample LGAs. Sampling without replacement was adopted to avoid selection of same school twice. A total of 1600 SSS2 students participated in phase II. A 50-item Mathematics Achievement Test was constructed with a reliability index of 0.83 (KR-20 formula). Data were subjected to Exploratory Factor Analysis (EFA), Parallel Analysis (PA), Confirmatory Factor Analysis (CFA), Polytomous Graded Response Model (PGRM) and Multiple regression analysis.

Twenty-four factors, comprising 64 SMEI, were extracted through EFA. The 64 items from EFA were reduced to 45 items through PA and further reduced to 37 items through CFA. The 37 items were subjected to PGRM for item calibration and reduced to 35 items. Dimensionality analysis of CFA and PGRM showed that the 35 items loaded on six factors and denoted sub-scales. These factors were:Personal Agency Engagement, Positive Affective Engagement,Negative Affective Engagement, Positive Behavioural Engagement, Negative Behavioural Engagement and Cognitive Engagement. The reliability index of the6-dimensional SMES was 0.90, while the reliability index of each of the sub-scales of the SMES ranged from 0.68 to 0.87. Regression analysis of the sub-scales showed that only Negative Behavioural Engagement predicted students’ achievement in Mathematics (β = -0.12, t = -2.952, p<0.05). This implies that students who exhibited negative behavioural engagement tend to perform poorly in Mathematics.

A robust 6-dimensional Students Mathematics Engagement scale was constructed, and its sub-scales were used to predict students’ achievement in Mathematics. Mathematics teachers should be encouraged to use this scale for measuringsecondary school students’ level of engagement in the subject.