Measurement of challenge and self-efficacy in learning

Abstract

Students’ perceived self-efficacy is an important factor in determining their educational success. Those with high self-efficacy are likely to engage readily in learning activities, and to persist with their studies in the face of adversity. Those with low self-efficacy may shy away from engagement in activities that they perceive as challenging, and may give up when problems are encountered. The perceived difficulty of learning tasks is also important. If there is insufficient challenge, students might not value the learning and thus disengage. However, if the material is too challenging, learners may lose confidence in their ability to master the material and may also disengage. Getting the level of challenge right is thus a key factor in designing learning experiences that will engage learners and build their confidence in their ability to master challenges in their future.This thesis investigates the possibility of providing educators with objective evidence of students’ self-efficacy and the perceived challenge of the learning activities in a course. The investigation includes both theoretical and practical perspectives.The theoretical perspective involves the derivation of a formal measurement model, together with its theoretical and conceptual underpinnings. When the assumptions of the model hold, the model produces objective linear measurement from ordinal data, accompanied by estimates of uncertainty, conjointly for students’ self-efficacy and the challenge of activities. A comprehensive framework of hypotheses is developed to test the assumptions of the model, from the theoretical quantifiability of the constructs, through to fitness for purpose. A software implementation of the model is developed and evaluated from theoretical and empirical perspectives. Evidence of construct validity is provided. The measurement model used required a reimagining of measurement theory from an information theoretic point of view. This point of view adds to the theoretical understanding of measurement when items with multiple categories are used.The practical perspective involves the development of conceptual and representational frameworks that are readily understood and interpretable by educators. The key elements are the use of multiple output representations, the use of appropriate analogies and metaphors, the mapping of statistical and information theoretic terms and constructs to equivalents that are more familiar to educators, and the reimagining of reliability as a measure of fitness for purpose. Two alternative approaches to the measurement of self-efficacy are investigated: one based on direct reporting, and one based on inference from engagement in activities. Both are found suitable for the measurement of self-efficacy.The model is robust under the introduction of random noise and violation of the core assumption of local independence. Relatively few subjects and items arerequired to achieve useful measurement accuracy and this can be achieved inrealistic educational settings. Overall, it was found that measurement of selfefficacy is both practical and useful in a realistic educational setting.