Just search for sample size calculators on the internet, and be sure you use one that allows you to adjust for the expected proportion in population.

What are you trying to do? For prevalence one usually wants to estimate with some margin of error with some level of confidence, so a margin of error calculator should be sufficient. If you want to test a hypothesis it will depend on the prevalence value you want to test against (e.g., power to detect a difference from 15% will be different from power to detect a difference from 50%). I'd note that the test for a single proportion usually defaults to 50% so isn't always very useful - the 95%Ci for prevalence might be more practical).

See the attached Google search. Best wishes David Booth notice that the default mentioned above 50per cent gives the largest n and hence the safest choice if you have no idea about the value of p.

Dear colleague, the usual formula in descriptive studies or even in clinical trials is the following:

ῆ = ((1,96²)×p×(1-p))/d²

So that :

ῆ : estimated sample size

p : prevalence (16% or 0.16 in your case)

d : half the desired confidence interval, i.e. 0.025 if the confidence interval is 5%.

So you will have, unless I am mistaken,

ῆ = (3,84×0,16×0.84)/0.025²

ῆ = 82,57 @ 83.

ῆ >= 83.

For more precision (2%, d = 0.01), you should have a higher estimated sample size ῆ >= 5161, which could constrain your study.

Good luck.