Consider a study where we want to detect a small difference in means between two groups. If we increase the sample size $n$ while keeping the variance and $\alpha$ constant, what is the effect on the power of the test?