Aggregating student peer assessment during capstone projects

• Autores: Desmond Adair, Martin Jaeger
• Localización: The International journal of engineering education, ISSN-e 0949-149X, Vol. 33, no. 1 (Parte A), 2017, págs. 216-224
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Student assessment of other student’s work has many potential benefits to learning for both the assessor and the assessed.However, sources of peer assessment provide, quite often, subjective evidence, which can be conflicting, uncertain and evenignorant. One of the key elements to providing an overall quality assessment of a student’s work derived from theassessment by his/her peers is the use of an appropriate method of combining or fusing these heterogeneous evidencesources. Since the development of the belief theory introduced by Shafer in the 1970s, many combination rules have beenproposed in the literature with two main methods selected here. The first is an evidential reasoning (ER) approach, thekernel of which is an ER algorithm developed on the basis of the framework and the evidence combination rule of theDempster-Shafer (DS) theory. It has been claimed also in the literature that Dempster’s rule generates counter-intuitiveand unacceptable results in practical situations, so an approach based on the Dezert-Smarandache (DSmT) theory offusion will also be explored, in particular the PCR6 rule of proportional conflict redistribution. Results for peer assessmentmarksallocatedby astudent cohort,consistingof 20 students,duringtheir capstone projects,and,aggregated usingeachofthesetwo approachesare compared witheachother andwith resultsobtained by the moretraditional AveragingRule (AR)approach. It is clear from the findings that when subjective evidence is aggregated then the simple AR approach as theaccepted combination method is in doubt. It also seems that the DS method of aggregation seems the best alternative totraditional averaging.