Dear allstat members, To produce an audit from data management point of view, 10% of sample (n ) is taken from the dataset and expect an error rate of 1%. If we have say 200 patients , we select 20 patients , if we have 10000 patients we select 1000 patients for the audit and still expect 1% error rate. eg 1% error rate=error in data/number of patients Is there any method/way of determining the cut off value for number of patients required and still obtain an error rate of 1%. Please send response to me directly. I'll also welcome any references on the subject. Regards, George