Suppose that I am conducting a questionnaire study that is trying to measure level of awareness of subjects about a programming language and find the relation of those level of awareness to working conditions and methods etc.
To improve my precision I decided to go with stratified sampling. If I have 1 criterion for stratification such as geographical distribution (to make sure I don't over-represent subjects from areas that have less programmers), then I end up with 6 distinct strata (country provinces).
I know how to go about analysing these to find margin of error, standard error etc but I realised that is not good enough and I need to introduce more criteria for stratification, such as level of education (so i don't over-represent a group who are not very present between programmers based on their education), level of seniority etc.
I have the proportion (in
%) for all these criteria but I don't know how I go about sampling when I have more than one criterion?