Conference sessions in Peer learning and evaluation
Herding in the classroom - an experiment
Presentation at DEE 2017,My project looks at how students are influenced by their peers' responses to in-class questions. In particular, I study whether students' knowledge about the distribution of answers provided by their classmates to a multiple-choice question affects their subsequent answer. I carried out an experiment in an upper-level undergraduate economics field module where students were asked to respond to at least 2-3 multiple-choice questions in each lecture session. Answers were submitted using a web-based audience response system. Students were shown the distribution of answers from the entire class of about 20 students, and then asked the same question again. I recorded whether or not students changed their answers, whether the answers were correct in the first instance and eventually, and various demographic characteristics of the students, and whom they were sitting with. Any movement towards the modal answer is labelled as herding. My analysis looks at what determines students' proclivity to herd, including individual characteristics, the characteristics of their immediate peers and the type of question. Any systematic patterns in the herding behaviour can be used to influence the organization of peer learning groups - if students are more likely to learn from peers with particular characteristics, lecturers may want to design "think-pair-share" or other kind of group learning activities with these characteristics in mind.
Self vs. peer evaluation: Are students more accurate at evaluating the work of their peers than their own?
Presentation at DEE 2017,One of the most important objectives of courses in higher education is to develop the independent learning skills of the students. In order to become effective self-directed learners they need to acquire good self-evaluation skills. Unfortunately, numerous studies have found evidence that undergraduate students are rather poor at judging the quality of their own work (Guest and Riegler 2017; Nowell and Alston 2007; Grimes, 2002). The data indicates that this divergence between the student and the tutors’ perception of the academic work is not random. Most studies find that on average students systematically over-estimate their own performance. Does this evidence indicate that large numbers of students find it difficult to effectively internalise the standards on their course? This might be the case. However, an alternative explanation is that students find it particularly difficult to objectively evaluate the quality of their own work because of the emotional investment involved. They may find it easier to judge the appropriate standard when looking at the work of others. To test this hypothesis the paper compares the ability of students to self and peer evaluate an assessed economics essay that has been completed out of class. The data comes from a group of 2nd year undergraduate students.
Birds of a feather – social interactions of university students in a classroom
Presentation at DEE 2017,Social networks can facilitate the communication among students and help them in learning. However, there are rare studies combining observation of students’ behaviour in groups with their academic performance. Academic performance can boost one’s reputation among peers, giving them a significant degree of social influence. Even if the process of building reputation in a network is usually slow, compared to the spread of deviant behaviour, it gives a potentiality for various effort making strategies. For example, Sacerdote (2000) shows that an increase in average peer achievement may lead to an increase of a student’s own achievements. While obtaining new knowledge and developing social capital are definitely good sides of peer effect, reputation may also have drawbacks. For example the significant differences in reputation may arise possibilities of ‘moral hazard’ and free-riding strategies. Students with lower academic performance may be willing to stay close to high performing ones, wishing for their help or just treating them as “additional information assets”. In turn, students with high academic performance may be reluctant to keep such ties, considering them abusive. In this study we want to investigate the patterns of social interaction among students during a semester. We analyse the spontaneous social networks closely related to the final evaluation of students: the peer selection for team work and the neighbourhood selection on the ‘test days’.
