Abstract for: Building Validated Causal Loops: A New Methodology Bridging Ishikawa Diagrams and Temporal Causality Analysis

Traditional system dynamics practice relies heavily on expert judgment to construct causal loop diagrams, often without systematic empirical validation. Understanding the behavior of complex socioeconomic and technological systems requires models grounded in accurate and defensible causal structures. This paper proposes a five-step methodology that integrates structured hypothesis generation via transformed Ishikawa diagrams with formal temporal causality testing to construct empirically validated causal loop structures. The framework exhaustively enumerates plausible causal relationships; tests temporal precedence and predictive relevance using appropriate causality methods and determines causal polarity to distinguish reinforcing and balancing feedback loops. The methodology is demonstrated through a macroeconomic case study involving nine interacting variables. From 72 hypothesized relationships generated through Ishikawa diagrams, only 28 (39%) were empirically validated, revealing five statistically significant feedback loops (p < 0.001). The validated causal structure was subsequently operationalized in a hybrid System Dynamics–Agent-Based model achieving strong empirical performance (R² = 0.89, MAPE = 6.7%). By embedding empirical validation directly into the conceptualization stage, the proposed approach improves structural validity, eliminates spurious causal assumptions, and provides full traceability from data to model structure. The methodology offers a repeatable protocol for causal loop construction that bridges system dynamics and data science, enhancing model credibility for high-stakes policy and decision-support applications.