PROC SURVEYREG assumes missing values are missing at random and deletes them. How can I include item-nonrespondents in the variance estimation?

 

Beginning in SAS 9.2, you can specify the NOMCAR option in the PROC SURVEYREG statement. This option requests that the procedure treat missing values in the variance computation as not missing completely at random (NOMCAR) for Taylor series variance estimation.

 

Prior to SAS 9.2, you can use the SREGSUB macro. By default, the macro treats the non-missing values as a subgroup of item-respondents, and thus includes the missing values in variance computations.