The National Compensation Survey (NCS) is the new BLS survey for measuring employee wages by skill level. NCS uses a rotating panel design, with three stages of selection used in selecting each panel: geographic area PSUs; establishments selected from industry strata; and occupations selected separately from each sample establishment. The establishments are selected with probability proportional to size (pps), with total employment the measure of size. For each selected establishment the number of sampled occupations are a function of employment size. This paper investigates two aspects of this sample design. First, whether some other probability selection scheme for sampling establishments yields lower variances than direct pps. Second, if there exists an allocation by size class for the number of occupational selections that yields lower variances than the current allocation.