Stratified sampling is a technique used in some instances to generate a 'fairer' sample than standard random sampling. Wherein a population is divided into subgroups called 'strata'. These strata are usually based upon a characteristic that makes its members distinct from the rest of the population.
For instance, if I wanted sample of 10 students in a school to take a survey, I would want a fair representation of the school as a whole. Random sampling would mean assigning each pupil in the school a number, then using a random number generator to give me any 10 of these numbers and the corresponding students would take the survey. This is good because I cannot influence the sample and so it is not biased, but it still may not be fair as by chance all the students could be year 7s. Their views may not be representative of the views of the other years and my results would be skewed. If I were to do a stratified sample however, I would be sure to get a more representative sample of the school as the pupils could be divided into each year group - these are my strata. I can then find the percentage of the school that each year group accounts for, and calculate how many students I need from that year in my sample, then randomly select pupils in that strata in the same way I would for a random sample.