Larger sample sizes result in improved power of finding an important treatment effect when it’s there, as well as tighter CIs, and reduced publication biases when an effect is significant; but pre-experimental planning and statistical inference given a borderline significant p value are quite different things.
1
—
—