Intelligent marking in summative e-assessments

This paper describes our experiences of employing the algorithmic e-assessment approachto summative assessments in a large cohort first year engineering module at the Universityof the West of England, Bristol (UWE). The philosophy of this approach, with regards tomarking and feedback, is that the automated process should simulate the process employedby human markers. This includes the process of continuation marking, mal-rule detectionand adaptive marking. In performing these tasks we illustrate the importance thatalgorithmic e-assessment plays in the marking and associated feedback processes;algorithmic e-assessment is significantly more than simply „randomly generated questionparameters‟.We report on the results of using these features in an e-Assessment that was run for thefirst time this academic year at UWE, replacing a paper-based assignment. We highlight thebenefits to the students of this experience of e-assessment. Comparison is made betweenthe students‟ performances with some of the algorithmic marking features having beingdisabled. Running summative assessments in this way is not only efficient for the setter,resulting in a huge saving in marking time, but also results in increased student satisfaction;students receive prompt, detailed and specific feedback on their work. For these e-assessments we use the DEWIS algorithmic e-assessment system which was designed anddeveloped with the complete algorithmic philosophy in mind. Because of this, theimplementation of quite complicated procedures such as continuation marking (withmultiple branches) is relatively straight-forward