A researcher is experimenting with several regression equations. Unknown to him, all of his formulations are in fact worthless, but nonetheless there is a 5 per cent chance that each regression will--by the luck of the draw--appear to come up with `significant' results. Call such an event a `success'. If the researcher tries 10 equations, what is the probability that he has exactly one success? What is the probability of at least one success?
Here you may look at an
answer
to this question--but try it yourself first!