If $P$ is the population proportion of some characteristic under study, $Q=1-P$ and $n$ is the sample size, then the standard error of observed sample proportion is