Observational Scaffolding for Learning Analytics: A Methodological Proposal

Temporal analysis of learning data is attracting the interest of researchers, and a growing body of Learning Analytics (LA) research applies lag sequential analysis. However, lack of methodological frameworks that guide the data gathering and analysis poses multiple conceptual, methodological, analytical and technical challenges. While observation as a technique has been already used in LA, systematic observation methods and designs have not been applied so far, and parameters often used in the observational domain (such as order and duration) are still under-researched. In this paper we propose a methodological framework, and illustrate its potential by applying it in the analysis of a Knowledge Forum dataset. Results show the potential of the proposed method to uncover behavioral patterns prospectively (lag +1 to lag +5) or retrospectively (lag −1 to lag −5), and to reduce this information through polar coordinate analysis. Moreover, as illustrated in this paper, observational methods offer a rigorous framework for LA datasets, enabling the replicability, validity and reliability of the results.