Inferring a correlation (maybe even a causal connection) between A and B, from the premise that some percentage (usually a high one) of instances of A are also instances of B, while failing to compare that percentage against the percentage of non-As which are Bs. People commonly declare their washing their car enhances the chances of there coming up a rainstorm. "Every time I wash my car, it rains", they say. Well, maybe so, but the question is, how likely is it to rain if they don't wash their car? Checking into the meteorological data may reveal that it rains all the doggone time, whether they wash their car or not-- so much for their causal inference.