I am looking to conduct an epidemiological study on a population of 1000 individuals and need to find the appropriate sample size. So far, I have the parameters to calculate my sample size but I am a bit unsure about whether I am using the correct formula(s). The situation is as follows:

population: 1000
test sensitivity: ~100%
test specifity: 99,81%
Z = 1,96 (two-tailed 95%)
disease prevalence: between 0,4% and 8,4%

To calculate my sample size, I used (from [1]): $$n=\frac{N\cdot Z^{2}\cdot P(1-P)}{d^{2}(N-1)+Z^{2}\cdot P(1-P)}$$

With N = population = 1000
n = sample size
Z = 1.96
P = expected prevalence
and d = precision = P/5, which I got from [2]

If I plug in the numbers for 8.4% prevalence, I get a sample size of 506, but we suspect that the prevalence will actually be at around 0.4%, which gives me a sample size of 960 individuals. Now, my question would be: am I doing this right? Am I using the right formulas? Is my thinking for the sample size calculation correct? Thank you all for your help!

[1] NAING, L., WINN, T. & RUSLI, B. 2006. Practical issues in calculating the sample size for prevalence studies. Archives of orofacial Sciences, 1, 9-14.

[2] POURHOSEINGHOLI, M. A., VAHEDI, M. & RAHIMZADEH, M. 2013. Sample size calculation in medical studies. Gastroenterology and hepatology from bed to bench, 6, 14-17.


IMHO, the reason you are getting such a large sample size (compared to the population), is that you want a very tight precision. You set d (which I translate as a form of statistical error ) as 1/5 of the actual value you are expecting. When P is 0.4% that means 0.08% statistical error which is quite low.

If you indeed need such a high precision, then you are probably better of sampling the entire population. However, since 1000 is unusually round number for population, I am highly sceptical regarding your application.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.