This research seeks to apply the concepts of collaborative learning and open learner modelling in order to find out whether seeing their own group learner model helps learners improve their learning in a computer-based collaborative learning environment. There is previous work on giving back information about learning performance as a group but very little, if any, empirical work on the benefits of a group open learner model (GOLM). A major benefit of collaborative learning is to encourage learners to learn further from what they cannot achieve when do it by themselves but they can manage with another. Combining this with viewing and judging the information about learning found in a learner model, it was expected that this would increase their learning awareness in order to improve their learning performance. Without such group learner models, learners might not improve their learning performance in the collaborative learning environment as much as they might. To find out whether opening the group learner models helped learners to improve their learning performance we developed a system called 'GOLeM', and we focused on the learner's score on learning concepts and their degree of confidence in their answer. GOLeM was used as a learning environment to test for evidence in relation to two comparisons of individual performance. The first was a comparison of individual performance between participants in a non computer-based individual learning environment and a computer-based collaborative learning environment. The second respect was to compare the results of learning in two different computer-based collaborative learning environments which were only different in terms of whether or not the learners could see their group learner model. The content of number-conversion is chosen for the domain knowledge. Dialogue games and sentence openers are used to implement a chat-tool to exchange beliefs between peers. Bar charts and textual explanations are used as external representations of learning performance as a group. The system was implemented and tested in two versions: paper-based, for the plausibility of the content and the user interface; and computer-based, for comparing the learning results among three different learning environments regarding the two respects above. To make sure what we built was valid – in terms of suitable content applied to the right target group of learners, we did several tests. These tests consist of a questionnaire with multiple choice questions applied to a small group of participants some of whom have a background in computing, and some have no background in computing. The questionnaire was examined for the suitability of its content and for the target group. A modified questionnaire was used with 122 participants who have a background in computing to validate in relation to the difficulty level and item discrimination. Five questions were selected as representative of the domain knowledge for a paper-based design and applied to six pairs of learners for the suitability of the questions and the number to be used, time taken, user interface, etc before developing the computer-based version. Regarding the comparison between participants in a non computer-based individual learning environment and a computer-based collaborative learning environment, the results show there is a significant difference at the 5% level in terms of learning concept-score and degree of confidence in favour of individual learning performance of learners in collaborative learning environment. Considering the comparison of learning between the two computer-based collaborative learning environments, participants who are able to see their learning performance as a group learner models both before the group test and after each item of the group test, have a slightly higher concept-score and improved degree of confidence than those who cannot see these learner models. Moreover there evidence regarding the participant's self-assessment and peer-assessment, their opinion of the helpfulness of seeing the group learner model and their satisfaction in using this system confirms that further study in this area is justified. It leads to the conclusion that in these specific circumstances, learners benefit more from learning and seeing their group learner model. However the evidence that we have here is not sufficient to answer whether it is likely to be true that other systems like this will always lead the better learning. As a result, we plan to continue our work in both similar and different directions to improve the strength of the conclusion that providing group learner model in a computer-based collaborative learning environment helps learners to benefit from learning. The thesis mainly contributes to both CSCL and AIED communities for further study of GOLeM itself. Regarding the AIED community, GOLeM can be used for the further study on the benefits of seeing learning performance as a group learner model both before and after performing a group-test. Regarding the CSCL community, using this GOLeM with either a larger or a wider variety of groups of learners focusing on knowledge contribution during the group-test for the concrete evidence to support that social interaction has an impact on collaborative learning. The evidence that we have found suggests that being able to see a GOLM improves learning. Though this evidence is not statistically significant, this thesis has provided the most thorough empirical examination of the benefits of a GOLM so far.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:495207 |
Date | January 2008 |
Creators | Tongchai, Nilubon |
Publisher | University of Glasgow |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://theses.gla.ac.uk/195/ |
Page generated in 0.0015 seconds