A statistical method that allows for ending an experiment as soon as a response with the desired precision is obtained, with each outcome being examined as soon as it occurs and being added to all the previous outcomes.