harper_r1.pdf (304.85 kB)
Download file

Allowing for guessing and for the expectations from the learning outcomes in computer-based assessments

Download (304.85 kB)
conference contribution
posted on 24.05.2006, 14:51 by Ray Harper
Computer-based assessments usually generate a percentage mark. It is not self-evident how this relates to the final percentage mark or final grade for the work since this depends on (i) its relationship to the "baseline" mark expected for someone who only guesses, (ii) to the "expectations" for the piece of work in relation to the learning objectives and (iii) the grading scheme employed. For some question types it is possible to allow for guessing within the marking scheme for the question using negative marking but in general it is preferable to correct for guessing within a post-test grading scheme that allows for guessing. The relationship between the assessment learning objectives and essays where choice is available and topics can be avoided compared with computer-based assessments where no choice is available and topics cannot be avoided is considered. It is concluded that commonly maximum performance should not be set at a mark of 100% but that an allowance should be made for the maximum expected performance based on the learning objectives. The use of formulae in a spreadsheet to convert the marks into grades based on a statistical allowance for guessing or additionally allowing for the maximum expected mark is demonstrated. A spreadsheet pro forma containing all of the formulae for adjusting marks and determining grades can be obtained by selecting “Grading” from the menu at http://students.luton.ac.uk/biology/webol/.



  • University Academic and Administrative Support


  • Professional Development

Research Unit

  • CAA Conference


312164 bytes


HARPER, R., 2002. Allowing for Guessing and for the Expectations from the Learning Outcomes in Computer-Based Assessments. IN: Proceedings of the 6th CAA Conference, Loughborough: Loughborough University


© Loughborough University

Publication date



This is a conference paper.



Usage metrics