Sampsize

Sample size for:

Sample size for a prevalence survey, with finite population correction

If you enter a precision of 5%, Sampsize will return the sample size needed for 95% (default) or any other confidence interval where the upper limit equals prevalence + precision and the lower limit equals prevalence - precision.

The population size is the total size of the population from which a sample will be drawn for the survey. If the population size is small, the correction for finite population will result in a reduced sample size.

Entering 50 for the estimated prevalence will result in the highest sample size. Entering 0 for the population size will result in Sampsize using an infinite (very large) population size.
If the population is infinite, sampsize will also return the exact binomial confidence limits given the returned sample size and entered prevalence. These limits should be close to the specified limits. If the exact bounds are wider than the specified ones, then the formula for estimating the sample size is likely inappropriate, and an alternative course of action is to test different combinations of sample size and number of observed events (such as the number of patients with the studied disease) and calculating an exact confidence interval given the combination.

Binomial Exact Confidence Interval

Example: You flip a coin 10 times and it comes up heads only once. You are shocked and decide to obtain a 99% confidence interval for this coin. You enter 10 in the field "N total", 1 in the field "Number of successful events", and 99 in the field "Level of Confidence Interval", which is expressed as a percentage. Sampsize will return a 99% confidence interval from 0.05% to 54%, which tells you that observing only one head may be due to a random sampling effect given a 50% probability of heads to show up.

Sample size needed to observe at least n events given probability of occurrence equals prevalence

Example: A study on the hepatitis C virus aims at describing some characteristics of the virus strains. It is assumed that the prevalence of virus positive individuals is 10% in the population. The investigators wish to calculate the sample size needed to be 95% sure that at least 5 patients will harbour a virus. The values 10 in the "Prevalence" field (prevalence is expressed as a percentage), and 5 in the "Minimum number of events" field should be entered. The default value may be left in the last field "Probability to observe the above numer". Sampsize returns an estimated sample size of n = 90.

Goto index


Precision	%
Prevalence	%	Enter 50 if unknown
Population		Enter 0 if unknown
Level	%	Level of the confidence interval


Prevalence		% Enter 50 if unknown
Minimum number of events
Probability to observe the above number of events in the sample		% (default 95%)


N total
Number of successful events
Level of Confidence Interval