Diagnóstico do comportamento dos aprendizes na educação a distância com base no estilo de aprendizagem

Abstract

Computer systems are increasingly being used by schools and universities to manage learners in the worldwide. The data generated by the use of systems such as Enterprise Resource Planning, Learning Management System and Virtual Learning Environments can contain valuable information. This information can be used strategically to support teachers in decision making. However, there are situations where only the learners’ historical data are not sufficient to support decision making. Decision making may be required even before obtaining historical data, in example, for newly registered learners. The learner Learning Style can complement historical data covering this gap through the intersection with the historical behavior patterns. Once collected, learning style can be used at any time in the future, it is considered stable so as not change significantly over time. Thus, this thesis aims to propose a evaluation of the potential use of Learning Style in the early learner behavior diagnosis in the distance learning to support decision making. The hypothesis that there is a relation between Learning Style and the learner behavior in Virtual Learning Environments arises from the understanding of the concept of Learning Style, where the cognitive and behavioral preferences are described numerically. The learners’ behavior in a virtual environment is identified through its interaction with the elements of the virtual environment. Thus, the data generated from these interactions can be related to the variables that describe the Learning Style, called dimensions, allowing that early behavior diagnosis can be achieved by promoting the support to decision making. Thus, this dissertation brings the efforts of the potential use of Learning Style to diagnose learners’ behavior in twelve experiments. The results of these experiments suggest that the Learning Style can explain the learner behavior related to the amount of interaction in the learning environment, the interval of time among interactions and evasion. The Sequential/Global dimension had clear results that it is possible to anticipate the learner behavior using only the style with accuracy between 50 and 80% for almost all experiments. In addition, the Active/Reflective dimension also presented results with the same level of accuracy, however satisfying only half of the experiments.