The question is as follows:. How frequent are all blips that might be classified as X overall, and from that, what is the expected rate of spurious correlation? To arrive at this one must have access to quite a bit of comprehensive data collected in an unbiased fashion.