Adaptive item sequencing in item-based learning environments
Kelly Wauters · Lirias · 2012
In order to make computer-based learning environments more efficient, researchers have been exploring the possibility of an automatic adaptation of the learning environment to the learners needs and preferences (Wasson, 1993). In this dissertation, we focus on dynamic, item-based adaptive learning environments in which the item difficulty level and the learners ability level are assessed to enable adaptive item sequencing and adaptive feedback. Adaptive item sequencing is well-established in computerized adaptive testing (CAT; Wainer, 2000), which often makes use of the item response theory (IRT; Hambleton, Swaminathan, & Rogers, 1991). IRT expresses the probability of observing a particular response to an item as a function of certain characteristics of the item (e.g., item difficulty) and the person (e.g., knowledge level). We will explore how applying IRT can be helpful for adaptive item selection in learning environments, which could yield an increase in motivation, as is found in testing environments (Wainer, 2000) and enhance learning. In addition to that, we investigate how the use of IRT can help in providing detailed feedback to learners about his or her (progress in) knowledge level and the characteristics of the items. Furthermore, we examine the adequacy of some alternative measurement methods, incorporating an IRT model and otherwise.Hence, the aim of this dissertation is to unravel the possibilities and challenges that come together with extrapolating the ideas of CAT and IRT to item-based learning environments. By doing so we could provide the optimal learning path for each learner by selecting the problem that maximizes learning based on the learners current knowledge level and the difficulty level of the problem and adapting the feedback accordingly. After a general introduction (chapter 1), this dissertation provides a critical description of the possibilities and problems that may occur when applying IRT for adaptive item selection in learning environments, starting from existing research literature about electronic learning and testing environments (chapter 2). This analysis results in the identification of three concrete components of such personalized instruction: assessing the difficulty level of each item, assessing the learners current ability level and optimizing the interaction between the item and the learner.Part I of this dissertation (chapter 3) focuses on the first challenge, which is the assessment of the difficulty level of each item. We consider both response data (i.e. the correctness of the response to an item) and judgment data (i.e. learner feedback and expert rating) for the estimation of the item difficulty. Part II of this dissertation tackles the second challenge, which is the assessment of the learners current ability level. This part embodies two aspects of ability estimation in learning environments: the estimation of the learners initial ability level (chapter 4) and the tracking or following of the intra-individual change in ability level (chapter 5).Part III of this dissertation sheds light on the third challenge, which is the optimization of the interaction between the item and the learner. Attention is drawn to the item sequencing algorithm (chapter 6) and to the notion of adaptive item feedback (chapter 7).To conclude (chapter 8), we discuss the findings of this dissertation and define new research lines. This dissertation adds to the realization of adaptive item sequencing in item-based learning environments through the evaluation of the adequacy of IRT and alternative measurement methods and through the study of a specific item sequencing algorithm and an adaptive feedback approach.