Where the population is divided into groups (strata) sharing a common characteristic, for example, relating to age, gender or socio-class, and a random sample is then taken from each strata. It reduces sampling error as, unlike simple random sampling or systematic sampling, it ensures a cross section of the population is selected. It is a good approach to use when one or more of the stratums in a population occur less frequently than others, as it ensures that every stratum is included in the sample.