Hello Arman,

If confidence that your samples will be representative of the intended target population is critical to your study, then, you should use probability samples:

1. Specify the characteristics of that population (e.g., do you want only undergraduate students who are enrolled in less than a "full-time" course load, and at what kind of institution?)

2. Enumerate the population (e.g., identify a complete and current listing of that population).

3. Randomly sample however many cases are needed from that population in order to achieve your research goals. For some, that may be the estimation of one or more population parameters with some target degree of precision and confidence. For others, that may be having sufficient statistical power to detect minimally meaningful effects (differences, relationships) for some type of proposed hypothesis test.

4. The sampling may be made more efficient (e.g., fewer cases needed) if you can specify one or more stratification variables that relate in some way to the outcome variable(s), then sample proportionately to the way those strata appear in the population.

Non-probability (convenience) samples are far easier to collect, but offer little or no confidence of representativeness for the target population.

No matter your approach, it would be wise to over-sample slightly, as there will almost always be instances wherein incomplete responses/participation occur, or participants do not follow your protocols appropriately, or persons solicited for participation choose not to participate.

Good luck with your work!

Thank you very much sir for the brilliant suggestion. Agreed with you but Is purposive sampling possible? Although it is non probability? How to justify?

Agree with the over sample concept. I always practice that to overcome problem in sampling number....thanks again for suggestion.

So, what if the samples recruited from three different universities...Three groups of total 600 part time undergraduate students from three universities in Thailand (ABC university, 123 university and XYZ university), what is/are the best sampling technique/s to employ and what are the possible sampling procedures researchers need to follow?

I totally agree with David Morse as you need to specify the characteristic of the population targeted in your study e.g. undergraduate, postgraduate, age, etc. For sampling technique, I would suggest you to apply probability sampling in your study (stratified sampling) just so you can have better confidence level.

However, if you wish to go for simpler option (maybe because of time constrain, etc), convenience sampling would be a better technique for your study. Hope this helps. Arman Haji Ahmad