when you attemt to generate the theory in research area, theoratical sampling provide a structure of data collection as well as data analysis. in this sampling first, collect the data and thereafter find the relationship between that the representative exist in that category.
I recommend you to read the book by Kathy Charmaz which provides details of theoretical sampling:
Charmaz, K. (2006) Constructing Grounded Theory: A Practical Guide Through Qualitative Analysis. London: Sage.
You can also read the following papers which used theoretical sampling:
(a) Lord, L., Jefferson, T., Klass, D., Nowak, M. and Thomas, G. (2013) 'Leadership in context: Insights from a study of nursing in Western Australia', Leadership 9(2): 180-200.
(b) Dacin, M. T., Munir, K. A. and Tracey, P. (2010) 'Formal dining at Cambridge colleges: Linking ritual performance and institutional maintenance', Academy of Management Journal 53(6): 1393-1418.
Theoretical sampling, I recommend to use it fundamentally when the object of study has characteristics that do not allow the researcher to access their units of analysis directly, while their selection procedure is informal and a bit arbitrary. Even so, these are used in many investigations and from them inferences are made about the population. The truth is that the theoretical samples have many disadvantages. The first is that, since we are not probabilistic, we can not calculate with precision the standard error, that is, we can not calculate with what level of confidence we make an estimate. However, when we work with nominal variables in which the subjective determines or weights, say attitudes, ways of life or behavior in minority groups utrast similar situations, this type of sampling is very useful. I have nothing to find of its use, generally because of the characteristics of my subjects of study it is the one I use the most, only it is the most questionable and on which we must base its use very well so as not to sin of poverty in scientific rigor.