The article discusses how to accurately estimate the effects of interventions in complex systems using causal models. By assuming linear causal relationships with equal error variances, researchers can identify causal graphs from observational data. However, when the causal structure is unknown, confidence intervals for causal effects can be overly optimistic. To address this, a framework based on test inversion was developed to provide confidence regions for total causal effects that consider uncertainty in both the causal structure and the size of effects.