Statistics Measure

 Suppose I was to provide two facts that seem unrelated but find that they fluctuate in the same way?  Say that more people die in boating accidents when it is a sunny day.  Did a sunny day cause boat accidents?  well unless it was sun glare, obviously not.

What is more likely is that more people like to boat on sunny days and more people boating most likely results in more people having accidents.  Each accident would of course have its own cause and the perceived correlations is meaningless.

There is a connection of course, the first event a sunny day led to more boating.  Statistics would indicate more boats lead to more accidents.  So there is a correlation, just not a cause and effect.

Statistics can be manipulated in many ways.  They can be useful to prove or at least indicate certain things but can also be manipulated to argue for lies.

Take a use from the last Presidential election where past statistics were used to argue a falsehood.  In past recent elections, Ohio was a good predictor of presidential winners.  This was possibly a good correlation but not a cause and effect.  So when in the last election Ohio went for the loser, his people claimed it indicated voter fraud since it went against its prior trend.  However that trend was always just correlation and never cause and effect.  Populations shift and potentially Ohio's no longer is a good reflection of the nation as a whole.  Essentially it was never a predictor, simply a coincidence.

Statistics measure, and some use them to predict.  However those predictions would be based on flawed logic, even when they worked.