Nishani - it would depend very much on what you want to achieve. A sample of 21, using a quantitative approach, would be very difficult to generalise if that is our intention i.e. conducting a descriptive survey. The size of the topic should not matter too much in preventing you from conducting in-depth interviews. 21 would be a good sample size for conducting qualitative interviews.

Thank you very much. I don't wish to generalize.I want to get an idea about their skills & attitudes. I hope to give a questionnaire for all 21 and conduct interviews with 6-7 staff members

The following papers may further help: 

• Bock, T. and Sergeant, J. (2002) Small sample market research, International Journal of Market Research, 44, 2, pp. 235-244.
• Sandelowski, M. (1995) Sample size in Qualitative Research, Research in Nursing and Health, 18, 2, pp. 179-183.
• Robinson, O. C. (2014) Sampling in Interview-Based Qualitative Research: A Theoretical and Practical Guide, Qualitative Research in Psychology, 11, 1, pp. 25-41.
• Lenth, R. V. (2001) Some Practical Guidelines for Effective Sample Size Determination, The American Statistician, 55, 3, pp. 187-193. 

it depends  on your research aim. However, I think qualitative approaches can be effective for small and non-representative samples. 

As your purpose is not to generalize, I can suggest a trick for use of quantitative analysis. Actually it is amount of the data we need, rather than people we need. Normally we measure small number of measures (for variables) on large sample of subjects, now you can do large number of measures on small sample of subjects. And then analyze the data specific to each participant in terms fo number of measures or it can be frequencies and chi square analysis and the like.

as for validation Mr. Nishani, I highly suggest doing a test re-test method to make sure their answers are not manipulated nor not the same.

for example, give them the questionnaire with similar answers and a week from that time give them the same questionnaire with misplacing of the order. finally, you compare both of the tests results to see if there any false in it since you are conducting it on a small group (21 participants).

If you primarily want to do descriptive statistics, such as percentages and averages etc., then a survey could be useful for your purposes. But in addition to the lack of generalizability, the small sample size would make it almost impossible to run tests where you would get statistically significant results.

I suggest you to do a qualitative study. You may use observation or interviews. Otherwise, if you decide to use quantitative then you need to consider getting several readings for every participant which probably considered as time series. At the end your research question is more important in deciding your methodology because it determines the type of data you are looking for and then you can decide about methodology. Sample can be expanded if you think about other groups similar to the people in your organisation. Wish you all the best

For small sample non-parametric tests can be tried out.

Dear Nishani

From your profile it would suggest that you are more of a qualitative researcher who has experience of interviews. Is the study for post graduates or under graduates!

For undergraduates you could have a potential sample of 10, for post grads 20 would be fine.

I have used Grounded theory for analyzing semi-structured interview at both levels. But would expect a greater application of GT at post grad level.

Good luck.

Small samples usually lend themselves better to qualitative methods (see papers attached), but you can power quantitative studies to see whether you would have enough participants (see link) and this can be done beforehand to see how many participants you will need; or afterwards to see how many you would have needed to get a significant effect size in your data set.

http://www.gpower.hhu.de/en.html

Article Sample Size and Saturation in PhD Studies Using Qualitative Interviews

Article How Many Interviews Are Enough?

Article How many qualitative interviews is enough? [THIS PAPER IS OP...

Article Are We There Yet? Data Saturation in Qualitative Research

Article Code Saturation Versus Meaning Saturation: How Many Intervie...

You could use small-n analysis (mahoney, 2000)

Better to go for qualitative and triangulate it with the descriptive study.