Two reasons : 1) Because large effects are more robust than tiny effects to things like researcher bias, choice of statistical methods, and experimental error.

Two reasons : 1) Because large effects are more robust than tiny effects to things like researcher bias, choice of statistical methods, and experimental error. 2) because perturbing large effects are more likely to see real world difference for every day life than small effects. The conclusion from studies of breastfeeding versus bottle is it does not matter because it is a tiny and likely negligible effect (even if the effect is highly statistically significant).