Beschreibung:
Accurately measuring the causal effects of business process interventions is crucial for effective process improvement and evidence-based decision-making. When randomized controlled trials are not feasible, organizations can resort to event logs that capture process execution to conduct observational studies. However, these studies are prone to confounding factors that can introduce bias and compromise the validity of conclusions drawn. This study proposes adopting doubly robust estimation techniques, combining propensity score weighting and regression adjustment to mitigate bias to address this challenge. We demonstrate the practical implementation of doubly robust estimation and present a comprehensive step-by-step workflow that organizations can employ using a simulated claims management process dataset. This workflow offers practical guidance for measuring and improving business process effectiveness and efficiency. This study also explores the limitations and challenges of this approach, as it integrates process mining, business process management, and causal inference. By emphasizing the potential of doubly robust estimation as a reliable method for estimating the effects of business process interventions without relying on randomization, this work provides organizations with actionable insights and guidance for implementation.