In general, sample size calculation may rely on information obtained after e.g. pilot studies. You can use descriptors of your population (e.g. mean, standard deviation or others) to estimate the proper sample size you need for getting given level of confidence intervals, probability at various distances to the mean. Basing on these outputs you can therefore decide what is the best solution fitting the trade-off between say your fieldwork and the results you wish to obtain.

Dear Mr. Paolo Giordani,

Thanks for your answer, I think that your suggesstion is very useful for my work. In the case of determining the sample size, is there some concrete solutions?

Thank You Very Much,

Viet

Yes, once you have this basic info about your population (mean, standard deviation, etc.) there are a couple of software that can easily run the analysis. I know PASS 11 does it, but even if you google "Sample Size Calculator" you can find a lot of freeware solutions... There was recently a similar question on RG suggesting good items (I can post it to you... if I'll find it again!)

Thank You So Much, Mr. Paolo Giordani!

Cheers,

Viet