Theory Matters

Abstract

Abstract: Following an extended perspective of evidence-centered design, this study provides a methodological exemplar of the theory-based construction of process indicators from log data. We investigated decision-making processes in web search as the target construct, assuming that individuals follow a heuristic search (focusing on search results vs. websites as a primary information source) and stopping rule (following a satisficing vs. sampling strategy). Drawing on these assumptions, we describe our reasoning for identifying the empirical evidence needed and selecting an assessment to obtain this evidence to derive process indicators that represent groups differentiated by search and stopping rule combinations. To evaluate our approach, we reanalyzed the processing behavior of 150 university students who were requested in four tasks to select a specific website from a list of five search results. We determined the process indicators per item and conducted multiple cluster analyses to investigate group recovery. For each item, we found three clusters, two of which matched our assumptions. Additionally, we explored the consistency of students’ cluster membership across items and investigated their relationship with students’ skills in evaluating online information. Based on the results, we discuss the tradeoff between construct breadth and process elaboration for deriving meaningful process indicators.