If we start from the premise that we want assessment to encourage and support learning, then one measure of the assessment’s effectiveness is better performance on later summative tasks. Mundeep Gill and Martin Greenhow (Gill, M. and Greenhow, M. (2008) How effective is feedback in computer-aided assessments? Learning Media and Technology, 33(3), 207-220) report on work where the introduction of computer-aided assessment had positive impact (by this measure) in all areas but one.
The problem hinged round the presence of correct units with students’ numerical answers, so we might accept an answer of 10 metres, 10 meters, 10 m, but not 10 M, 10 kg, 10 m s-1 or just 10. Like most physicists, this is something on which I have extremely strong views – I regard the unit as a crucial part of the answer. The problem is that the answer-matching for many e-assessment systems doesn’t allow you to check for correct values and correct units at the same time (let alone whether the answer has been expressed to the correct precision etc.). Fortunately this is something that OpenMark handles well – we can check numbers, units etc. etc. and give appropriate targeted feedback on any aspect(s) of an answer that are incorrect.
In the system Mundeep and Martin were using this was not possible, so units were provided for ths students outside the input box; all students had to input was a number. Unfortunately, over the two year period of the investigation, students were observed to be more likely to omit units from their written work. This is another one of those unintended consequences – not exactly a positive outcome for e-assessment. Thank you Mundeep and Martin for your honesty in reporting this; others would be advised to take note.