[__blockcipher__]

I think the acceptable value for N depends on the size of the effect being studied.

For example if it’s a massive effect, maybe 10 people is sufficient, but if it’s tiny enough to get swallowed up in statistical noise until you have 1000, then you need N >= 1000