The challenge of model complexity: improving the interpretation of large causal models through variety filters

While large causal models provide detailed insights to the analysts who develop them, general users are often challenged by their complexity. Commonly, these models overwhelm the cognitive capacities of human beings. The inaccessibility of large causal models is particularly regrettable when they deliver valuable expertise and information that should be shared with other researchers and practitioners. To address this issue, we propose a set of tools—so‐called variety filters—to reduce model complexity and promote the accurate interpretation of their results. These filters encompass interpretive model partitioning, structural model partitioning and algorithmic detection of archetypal structures (ADAS). We demonstrate the efficacy of the proposed variety filters using the World3–2003 Model—a simulation model of remarkable complexity. The filters drastically attenuate the complexity while enhancing the comprehension of the model. Based on our findings, we derive implications for the use of complex models and their interpretation.