Berkson's Paradox (Berkson's Bias)
Berkson's paradox, also called Berkson's bias or fallacy, is a type of selection bias - a mathematical result found in the fields of conditional probability and statistics in which two variables can be negatively correlated even though they have the appearance of being positively correlated within the population.
For some people this is counterintuitive (a surprising or unexpected) result. This is why it is called a "paradox." It is considered a complicating factor when performing a statistical test of proportion.
For example, in middle school height may appear to be positively correlated with proficiency level at math - taller students are able to complete more complex math problems. However this isn't a true statistical relationships in that middle school aged taller students are better at math simple because they are older.