07-31-2012 04:23 PM
I have a dataset (226841 records) having 2 columns(Claim_num and ACCT). there are about 235 ACCT having claims ranging 1 to 24,648)
I would like to get optimal sample size from this set to audit.
proc surveyselect data=cag_out n=100
strata cag / ALLOC=PROP nosample;
I getting error
ERROR: Variable PROP not found.
ERROR: Variable NOSAMPLE not found.
ERROR 22-322: Syntax error, expecting one of the following: a name, ;, -, :, DESCENDING, NOTSORTED, _ALL_, _CHARACTER_, _CHAR_,
ERROR 202-322: The option or parameter is not recognized and will be ignored.
any help is appreciated
08-02-2012 11:19 AM
All of the common statistical power/sample size procedures assume an infinite underlying (theoretical) population. Working with fixed populations complicates the estimation. Your 'population' is large enough to be functionally infinite. However, you still need to optimize on something else.
Change your question to something that power works for. Most likely, that would be to get a sufficient precision (s.d.) on an estimator of interest (cost?).