Simpson’s paradox—two examples and a bit more

Abstract

When data that are not homogeneous are combined they often produce a variety of results that are deceptive and likely useless. Discussions and examples illustrate.