Dear Mr. Muhammad,

I think, several scholars already answered your question about the sample size. As it was presented in the previous discussion.

My opinion is, the main point of your view is about how to ensure the quality of the research findings. It is related to the trust and confidence issues of the findings.The trust refers to the transparency how people can see of the research process and the confidence is related to the finding validation how it represents the real world.

Several data collection issues may appropriate to the confidence point, e.g., the key informant, sample spread, unit analysis, and the sample size issues. For instance, the key informant is about how to choose the right respondents who have skills, knowledge, and experiences in the field.

I am sorry, if my opinion has described the other matter. But, I hope it will help you.

There are many sources that give you formulae for computing sample sizes.  Most of them boil down to solving the bound on the error of estimation for n (sample size).  The simplest case for large samples (at least 30) would be:

Bound = (z) / ((SD estimate) / (square root of n))

Thus,

n = (SD squared) x (Bound squared) / (z squared)

Your Bound could be the maximum absolute error you were willing to accept and the z would depend on the level of significance you wished to attain.  For example, using the normal curve, you could use "2" if you wished to have approximately 95% confidence in your result.  You can find similar formulae in almost any sampling book for all types of parameter estimates.

Are you using a qualitative or quantitative approach?

With quantitative methods, you have received good advice so far. You basically are performing a power analysis. Free programs and apps are available to help you. There are also ways of estimating sample size needed based on the number of item/questions your are asking. These approaches give just rough estimates.

With qualitative methods, you want to primarily consider two criteria. First, are you talking to people who sufficiently represent each of the categories you identified? Of course, you want each group represented in your data. Second, with each additional person you interview, are you uncovering new information, or are the same ideas being repeated? If no new information is being provided, that suggests that you have interviewed enough people.

Hope that information is helpful. Good luck with your project? 

In addition to answers already provided, there are certain considerations in your case because you have varying sampling frame for 7 categories of experts, times 30 projects. If each project has one project director, then you have N=30, whereas if each project has on average 50 site enginers, then you have  another N=1500. All these will be added up when total population is counted. Whether quantitative or qualitative, an ideal sample may be representative of actual population. As the research question(s) is not mentioned, based on whatever research questions are formulated, qualitative design oftentimes requires less number of samples than quantitative design provided that sufficient information has been transcribed based on small samples. On the other hand, if your design is quantitative, you can choose a sample size based on power analysis. On a side note, you may have to choose stratified random if you need to make sure that all categories of experts are represented in your sample. Hope it helps.

Thank you all of the experts for your valuable guidelines in response of my questions.

I agree that it is important to know whether you are conducting quantitative research (e.g., a survey) or qualitative research (e.g., a set of case studies).

For the survey approach, you basically have multi-level data, with the different positions grouped within the 30 projects.

Actually, I will collect qualitative data. 

For a qualitative project, the best strategy would probably not be to work with all 30 sites, since that would generate relatively thin data from each location. Instead, you would be better off using purposive sampling to select a meaning subset of locations for comparison purposes. If you search for the terms "purposeful sampling" and "purposive sampling" on the internet, you will find a number of useful sources.

In general, I would recommend a comparative case study approach, where you select case that you expect to be systematically similar in some ways and different in others. An excellent source on this kind of research is Yin's book, Case study research: Design and methods.