Some of my earlier studies were about this sort of phenomenon: namely, that people are more willing to derive a cause from a correlation when they can picture the underlying mechanism. Even though the actual data remains the same, we are much more willing to leap to a causal conclusion when we can envision the fluent process by which an outcome is generated. There’s no problem with that, unless the underlying mechanism is flawed. When we are wrongly convinced that we understand a fluent process, we are more likely to draw a flawed causal conclusion.
