CHECKING THE ASSUMPTIONS FOR USING PARAMETRIC TESTS IN RELATION TO LOW SOCIO-ECONOMIC DISTRICTS EARLY ADOLESCENTS MOTIVATION AND ENGAGEMENT LEVELS IN LEARNING

Abstract

Low participation in learning of secondary students is a matter affirms examination, mostly in low socio-economic districts in Sri Lanka. One of the main factors contributing to this situation may be students’ motivation and engagement in learning. Therefore, this study tried to find out the levels of motivation and engagement among early adolescents. Motivation and Engagement ScaleJunior School was employed to collect data from Monaragala and Nuwara Eliya districts in Sri Lanka. Hundred male and 100 female students were chosen using stratified random sampling method. Confirmatory factor analysis did not provide a robust factor solution and it was decided to conduct exploratory factor analysis. Accordingly, four factors were identified: Positive Motivation (PM), Positive Engagement (PE), Failure Avoidance and Anxiety (FAA), and Uncertain Control (UC). It was decided to identify their motivation and engagement levels in relation to these factors using parametric tests. Therefore, the assumptions of using parametric tests were checked; normality, homogeneity of variance, data type, and independently distributed errors. In relation to the assumption of normality, the results of the normality tests using skewness and kurtosis were checked; all the scales showed substantial normality. This is further evident from the histograms and box-plots for all the scales. The PM, PE, and FAA scales show results from tests of homogeneity of variance based on gender. Only PM and FAA scales variances demonstrated homogeneity of variance for ethnicity. The test based upon grouping by school indicated that only the PM scale possessed homogeneity of variance. Of the four basic assumptions, normality, data type, and independently distributed errors were fulfilled; the assumption of homogeneity of variances was not. Therefore, it was decided to conduct both parametric tests and non-parametric tests.