Practical Prediction of Overall Performance from Formative Assessment Results of Engineering Students

In this paper, a new model was employed for probabilistic prediction of overall performance of engineering students. The model employs formative assessment results to estimate the summative assessment mark of an individual student. In the present study, statistical evaluation of predictions was conducted using data of seven (7) examination events involving 441 students studying bachelors and master's degrees of civil engineering programmes. The other key variables of the data comprised different class sizes, and heterogeneous classes containing students of varied academic performance levels. It was found that the model gave realistic predictions with a good to excellent level of accuracy. The range for summative performance results of students whose formative assessment marks fall between 50 to 70%, can be accurately estimated. The model may be used to inform policy frameworks targeted at promoting students' performance and throughput.