Work out minimum detectable effect instantly with clear inputs, formula shown and shareable results.
The minimum detectable effect is the smallest true effect a design can find reliably. With 100 per group at 80% power and alpha 0.05, it is d = 0.397 - so smaller genuine effects will usually be missed. Halving that threshold to 0.199 needs 397 per group, since sample size scales with the inverse square of the effect.
Minimum detectable effect
MDE = (z(alpha/2) + z(power)) x sqrt(2/n)
Sample scaling
halving the MDE requires four times the sample
Because it answers the more useful question: given the sample I can actually recruit, what is the smallest effect worth designing for?
Because the effect enters squared in the sample size formula, which is the fundamental cost of chasing small effects.