How we change what others think, feel, believe and do
Use it when there are smaller sub-groups that are to be investigated.
Use it when you want to achieve greater statistical significance in a smaller sample.
Use it to reduce standard error.
Strata can be natural groupings, such as age ranges or ethnic origins.
A high school student who is studying year-ten attitudes in the school uses registration tuition classes as strata and studies a random selection of students from each of these classes.
In a company there are more men than women, but it is required to have each group equally represented. Two strata are thus created, of men and women, with an equal number in each.
Stratification aims to reduce standard error by providing some control over variance. If you know that there are groups that must be included, for example men and women, then you can deliberately sample these in a due proportion.
Proportionate stratified sampling takes the same proportion (sample fraction) from each stratum.
Disproportionate stratified sampling takes a different proportion from different strata. This may be done to ensure minorities are adequately covered. If you do this, and want to make an estimate about the population, you will have to weight within-group estimates using the sampling fraction.
If the groups are homogeneous (ie. have the same proportions of each attribute), and hence within-group variation is lower than the population, then stratified random sampling will give a statistically more accurate result than simple random sampling.
Stratified sampling is sometimes called quota sampling or stratified random sampling.