Hello Kiross, You are using the correct manual formula and this, as you said, can be slightly different from BayesSQUEAC calculator sample size. If you want to share the data from your example (alpha. beta, precision and mode) we can have a closer look to it, but mainly 2 things might be happening: 1) The BayesSQUEAC calculator applies a modest population correction and it may show a lower sample size than using the manual formula given the same prior and required precision. This may happen with weak priors 2) There is a rule that states that the likelihood must be at least as strong as the prior. When a strong prior is used you me see difference due to this rule I do hope this helps Jose Luis

Dear Jose, Thanks. the data i am referring are already available on the following link; http://www.coverage-monitoring.org/resources/. i did check for the following reports; Darfur April2013, Afghanistan Jan2013, Philippines Feb2013 and Pakistan TMK Jan2013. As per the manual calculation, the sample size for the above reports is 80, 16, 16, and 64 respectively while using the calculator, it is 70, 32, 21 and 58 respectively. it seems the calculator estimation is higher except for Darfur case. Tahnks Kiross