The Simpson's paradox is a paradox in probability and statistics in which a trend appears in several groups of data but disappears or even reverses when the groups are combined. This result is often encountered in social-science and medical-science statistics and is particularly problematic when frequency data is unduly given causal interpretations.

Example: There exist treatment A and treatment B for kidney stones. Treatment A is more effective when used on small stones, and is also more effective when used on large stones, yet treatment B is more effective when considering all stones at the same time.

Different levels of overview:

simple.wikipedia.org

en.wikipedia.org

plato.stanford.edu