A Quantitative Investigation of the Correlation Between Academic Program Majors and Educational Objectives: A Data-Driven Approach

Abstract

In tertiary education, program education objects (PEDs) are a core component around which all program's activities revolve. This paper presents a data-driven approach to uncover an important aspect of this component that is its correlation to program majors (PMs). It does so by applying three well-known data correlation metrics, namely Pointwise Mutual Information, Correlation Coefficient, and Odds Ratio, to a dataset extracted from self-study reports of a set of Engineering programs. The collected dataset has undergone a preprocessing step to transform it into a suitable representation. This involves data cleaning, data annotation using a set of PEDs labels, and data projection to break down each multi-PEDs label data instances into a number of single PEDs data instances. The results obtained from the application of the three correlation metrics show a remarkable consistency among the three metrics in their evaluation of the correlation between PMs and PEDs. In a subsequent step, a ranking procedure of the PEDs within each PM, based on the obtained PMs-PEDs correlation strength, is applied and then a majority vote among the ranks of the three metrics is performed to obtain an overall rank of the PEDs within each PM. The obtained results show that each PM has a unique pattern of PEDs ranks, which suggests that PM nature plays a key role in determining the PM-PEDs correlation pattern. As a general conclusion, although the obtained results need further investigation on their causality correlation, the obtained quantitative correlations are very beneficial to the academicians particularly when designing new programs or reviewing existing ones.