Correlation
and Causation
Another favorite of misusers of statistics
is the old: if A follows B, B must have caused A. So take as a piece of
statistical evidence that the poor do worse on IQ tests (or SATs) than the
rich. Well, this means that the poor aren't as smart, and their lack of
intelligence is causing their poverty, right? Nonsense. It could as easily be
(and is) the case that the tests are written by the relatively wealthy for the
relatively wealthy, and as a result the wealthy understand the assumptions and
so on and do better. Statistics certainly can't make that decision-- to figure
out causation of a thing like this, we have to look carefully at the evidence.
You can find correlations (meaning as one
thing goes up, another does (for positive correlation) , or as X goes up Y goes
down (for negative correlation) ) between all kinds of things: The price of
rice in India has a strong correlation to teachers' salaries in Quebec. Does
this mean that one is causing the other? (Or is inflation responsible for
both?)