Derivation of Canonical Discriminant Models for the Academic Performance of Pre-major Engineering Students

Engr. Victor V. Hafalla Jr., REE, RME.


The study sought to derive discriminant models which discriminate between high, moderate and low achievers among pre-major engineering students of the University of Baguio based on derived factor constructs and using agglomerative hierarchical cluster analysis.  Results of the factor extraction using orthogonal rotation reveals four underlying factor constructs, namely, student’s prior academic performance, student’s UBMAT performance, parental influence and socio-economic status and high school section.  The four extracted factors account for 71.467% of the total variability in the data.  Results of the four agglomerative hierarchical clustering algorithms using the respondent’s factor scores produced comparable dendrograms with four clusters formed.  However one-way analysis of variance of the means on their college grade point average of the different groups suggests that groups 1 and 2 be combined.  Results from the Ward’s method combining clusters 1and 2 were utilized in deriving the discriminant model.  Results of the derivation of the discriminant model produced two canonical discriminant functions which were statistically significant to effect group discrimination.  Assessment of the model’s predictive accuracy also reveals it to be valid and accurate against a chance model.  The model’s hit ratio of 95.1% deems it to be a better model than those derived from previous studies.  Results of the study confirm the multidimensionality of tertiary academic performance and findings from previous studies on predictors of tertiary academic performance.  The derived discriminant functions may be used in profiling the students into high, moderate and low achieving students.  Furthermore, the derived functions may be used to gauge student capabilities prior to tackling the engineering course.

Source: UB Research Journal, Vol. XXXI, No. 2, July – December 2007