Where a population is not homogeneous, but contains groups of individuals (the strata) for which the characteristic of interest is likely to be noticeably different, it is likely that a stratified sample will provide a better estimator of the characteristic than a simple random sample, in the sense that the estimates will be closer to the true value more often. The procedure is to take a random sample from within each group, proportional to the size of the group. This removes one major source of variation in the simple random sample, namely how many of each group is taken.