Under normality, the sample mean has a smaller standard error than the 20% trimmed mean or median. If observations are sampled from a distribution that appears to be normal, does this suggest that the mean should be preferred over the trimmed mean and median?