Paul Crowley comments on Polyhacking

[Paul Crowley]

The true mean values should be close, but the medians etc can be very different.