Dear Rasel,

The answer was given many years ago and this is your own SATURATION. When you can not extract any new significant information for your research from a new school you start to deal with, just a repetition or small variations of the already got knowledge, you stop your study. So, the sampling cannot anyhow be determined before the study.

Bests,

Igor Gurkov

Dear Rasel Babu, first you have to decide the qualitative  data collection method to be used: focus group  discussion (FGD),  in-depth  interview or observation. For instance, if you decide to use  FGD, you have to categorize  the schools based on hemogeneity.  Then,  you will have heterogeneous groups. Therefore, if you have e.g. 2.heterogenous groups of school, you have to select minimum  of 2 and conduct minimum of 4 FGDs: 2 in each group. The max number of FGD will be decided based on idea saturation.

From my point of view, there are different options given, depending on what question you want to be answered. You can design a case study for each school and later derive typical elements. If your case study is designed to elaborate particularly important organizational behaviours, I would suggest to choose schools which are very "typical" in their individual behaviours. Heterogeneous case selection actually leads to fruitful insights. The cases still can be compared due to their homogeneous composition on a structural level: all are primary schools. The number of cases to be selected depends on the quality of answers regarding your research matter. If you once have the feeling that further investigations just repeat already known topics instead of delivering new insights, you can stop looking for new cases. I would suggest to determine the number of schools not prior, but rather during your investigation.

Dear Anne and Liben, thanks a lot for your responses. Very helpful!

The broadest approach in qualitative research is "purposive sampling" (also known as "purposeful sampling") where you select cases that suit your purpose. You have already described one set of purposive "inclusion criteria" (non-formal primary schools). Now you need to decide what set of comparisons within that set would be most useful.

Since you are doing a comparative case study, I would recommend Yin's book, Case Study Research: Design and Methods.

The following papers should be helpful to your topic: 

• Flyvbjerg, B. (2006). Five Misunderstandings About Case-Study Research. Qualitative Inquiry, 12, 2, pp. 219-245.
• Gerring, J. (2004). What is a case study and what is it good for? American political science review, 98, 2, pp. 341-354.
• Mason, M. (2010). Sample Size and Saturation in PhD Studies Using Qualitative Interviews. Forum, Qualitative Social Research, 11, 3, pp. 1-20.
• Sandelowski, M. (1995). Sample size in Qualitative Research. Research in Nursing and Health, 18, 2, pp. 179-183.
• Yin, R. K. (1981). The case study crisis: Some answers. Administrative science quarterly, 26, 1, pp. 58-65. 

My colleagues have mostly addressed the question well. I would add that sometimes the case you choose might be a critical case that illustrates something unique, exemplary, or unusual. Again, as previously indicated, it all depends on your research question and whether there has been previous research on similar cases.

For qualitative case study, the basis for sampling is the sampling criteria to be set up by the researcher prior to sample selection, so that the sample chosen will be able to meet the purpose of the study.

Hello, Rasel.

You can also take into consideration the methodologycal research tradition, since they also indicate things like the amount of case over which conclussions can be originated. For example,  the design of Grounded theory the basis could be different that the design with phenomenology. I think it's not exclusivelly on the technics for taking data, but more on the implications of the method and epistemological approach (subjectivity, reflection, reliability, etc). If you were to consider the significations of discipline and moral in elementary school children, you could take 3 schools, one child per school, or you could also take 3 childred from the very same school. Qualitative designs or research do not seek for generalizations as in experimental or descriptive positivism researches.

I agree with the colleagues and add one observation: the sample size will impact in the kind of analysis you will be able to do. For example, if you do interviews with a large sample with distinct groups, you can apply lexical analysis to compare this groups, using quantitavive analisys of qualitative data.

But, how collegues indicated very well, the proceding and choices depends of purposes of study, sample acess and avaible resources.

Dear Rasel you pose very interesting yest complex issue to be discussed. I am working on the features on the qualitative design right now. All I can tell you is that the literature is quite a mess. A lot of authors, a lot of theories, a lot of opinions on the qualitative design, yet no exact and clear answers. Some said X is a method, another called it practice, third mention that X is theory ... Somewhere I have read that depend on your research you, the the sample is different. Imagine, only one participant can be your sample. What I do, and this may help you: I have read many papers dealing with this issue and then select several relevant author that use case studies for example. Then  try to find their original work on net, since the adaptation of their theories from other experts are different. Then I have looked at steps in their researches and modify my article according to this. Of course I state in the methodology section something like this  ... this article leans on the   case study  research conducted from X. In my country , this paradigm (qualitative) is not known broader.  The quantitative paradigm is very much exploited. But as i wrote the new trends are going towards quantifying the qualitative data? Can you believe this?

The next I would like to comment is not you questions but the implications. I wonder why the journals insist so much on the methods and sampling and population and dimensions etc.? If someone describes the steps of the research and not taken in action that it is a case study, or discourse critical analysis etc. and yet conducted a decent research work, what is the lack of his research? Quantifying qualitatively gathered data? I personally hate everything that is number, and prefer to describe things and do not want to use tables, computer programs, chi square, t-test etc. I want to describe my research. What is wrong with that? However, if you want to be published in a high ranking journals you should stick to theories etc. Wish you like with your research.

Kind regards,

V. Janusheva

Once I did a piece of research on Excellent teachers, and It was published in a book.  However, one of the educators was so interesting, that It came out as a separate book!.  Which means, that sometimes a case can be so unique, and its relatability to other practitioners and researchers in related fields is so strong, that the sample can be limited even to one case or individual.  

Dear Rasel,

The answer was given many years ago and this is your own SATURATION. When you can not extract any new significant information for your research from a new school you start to deal with, just a repetition or small variations of the already got knowledge, you stop your study. So, the sampling cannot anyhow be determined before the study.

Bests,

Igor Gurkov