Some more comprehensive simulations of what happens to ‘fragile’ p values (those between 0.01 and 0.05), when the actual power differs from the minimum detectable difference that an 80% power calculation was depended upon to set the sample …