Stratified Random Sampling is a sampling method (a way of gathering participants for a study) used when the population is composed of several subgroups that may differ in the behavior or attribute that you are studying.
For example, you want to find out whether workers who did a lot of overtime work had higher performance scores. If you had existing data suggesting that workers who had children were less likely to work overtime than those who did not have children, you would divide the employee population into two groups: parents and non-parents. You would then randomly select an equal number of people from each subgroup.
In this case, conducting a stratified random sampling assures that you will be able to get sufficient data about each subgroup to make a meaningful analysis.