Falsifying Causal Graphs With Outlier Events

2026-07-13

AI Summary

This study addresses the critical challenge of validating candidate causal graphs in the absence of the true causal structure. The authors propose a statistical test grounded in the principle of anomaly propagation, leveraging the prior that weak anomalies cannot induce strong anomalies. Remarkably, this method enables falsification of candidate causal graphs using only a single anomalous sample. It represents the first approach to causal hypothesis testing that operates without requiring access to the true graph, offering rigorous control over false positive rates and strong statistical power. Empirical evaluations demonstrate that the proposed method effectively identifies causal graphs inconsistent with observed data and maintains robust performance even under extremely sparse anomaly regimes.

Abstract

True causal relationships are rarely known, and inferring causal graphs from data is hard. A fundamental challenge is how to assess whether a given causal graph is good in the absence of a ground truth. We propose falsifying candidate causal graphs based on whether they can explain the propagation of an outlier event. Our approach leverages a key principle: weak outliers rarely cause strong ones. While this principle has previously been used in root cause analysis to identify root causes without prior knowledge of the graph, we turn it on its head and use it to falsify candidate causal graphs whose implied outlier propagation is inconsistent with the data. To this end, we present the first statistical tests for the hypothesis that a candidate graph is the true causal graph, and show they have false positive control, power guarantees against incorrect causal graphs, and can operate with a single outlier sample.

Problem

• causal graphs
• falsification
• outlier events
• causal inference
• ground truth

Innovation

• causal graph falsification
• outlier propagation
• statistical hypothesis testing
• root cause analysis
• causal discovery