Needless to say my question could be more complex than that. I wish to study 4 farmers' markets. Respondents would be attendees at those markets, any adult attendee approached randomly by student helpers. In so far as I am concerned, while the population is anything but random (it almost defines the word "purposive") the sampling from the population is random. It has been suggested to me that's wrong and the way forward is quota sampling. That is being defined to me as 400 split 4 ways ie a sample of 100 per market. My view is that is not quota sampling which concerns sub division of the sample (eg by demographics) and is not a matter of the nature of the population. My view is that one simply takes results from the random samples at each of the the four markets then merges data from each market and proceed with multi-variate analysis. Thoughts?