ON SAMPLE SIZE

[VARUN ARORA]

The issue of sample size has been raised time and often, generally students use a universal formula for sample calculation in their dissertations i.e.

n = 4pq/ (r*r)

where p is taken as prevalence, q is taken as 1-p and r is taken as allowable error which they usually take as 0.05 or 5%.

My first question to all those in teaching/evaluation profession is, how many of you have seen this approach being used universally by the students?

In fact, this approach is applicable for prevalence studies and that too in a finite population.

The exercise of calculation of sample size should be done with caution, depending upon the objectives, type of data and characteristic being examined and more importantly knowing the limitations.

In this series we shall be trying to address the problem of sample size calculation based on objectives, type of data and characteristic being examined and limitations. We shall be coming across a number of formulae and their variants.

Contributions from all the participants are welcome.