🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
argument_from_correlation_to_cause
The argument from correlation to cause reasons that because two phenomena are correlated (they vary together), one must cause the other. This is a defeasible argumentation scheme that can be reasonable when supported by a plausible mechanism, temporal precedence, and dose-response relationships, but it becomes fallacious when correlation alone is treated as sufficient evidence of causation. The classic confounds are reverse causation, common causes, and coincidence.
Countries with higher chocolate consumption per capita have more Nobel Prize winners. Therefore, eating chocolate must enhance cognitive function and scientific achievement. This ignores that both chocolate consumption and Nobel Prizes correlate with national wealth and research funding.
Studies show that cities with more ice cream sales also report higher rates of drowning. Therefore, eating ice cream must increase the risk of drowning. In reality, both rise together in summer because hot weather drives people to both swim and buy ice cream.
Data shows that employees who wear fitness trackers have higher productivity scores. Therefore, giving all staff fitness trackers will boost their output. This ignores that motivated, health-conscious employees are likely to both seek out fitness trackers and perform well at work — personality drives both.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Is a causal claim being derived primarily from a correlation?
Type: binaryHave confounding variables been considered or controlled for?
Type: binaryHas the temporal order (cause preceding effect) been established?
Type: binaryCould the correlation be spurious or explained by a third factor?
Type: binaryThe argument from correlation to cause reasons that because two phenomena are correlated (they vary together), one must cause the other. This is a defeasible argumentation scheme that can be reasonable when supported by a plausible mechanism, temporal precedence, and dose-response relationships, but it becomes fallacious when correlation alone is treated as sufficient evidence of causation. The classic confounds are reverse causation, common causes, and coincidence.
The human brain is a pattern-detection machine that naturally infers causation from co-occurrence. When two things consistently appear together, the causal interpretation feels intuitive and is hard to resist even when we know better.
Apply the Bradford Hill criteria: Is there a plausible mechanism? Does the cause precede the effect? Is there a dose-response relationship? Could a third variable explain both? Could the causation run in the opposite direction?
Correlation-to-cause arguments fill health journalism ('studies link X to Y'), marketing research, social science reporting, and pop psychology books.
Use these tools to detect, analyze, or train this aspect.