An adjustment automatically takes account of any data present on the elements included in the collections to be used for the adjustment. For the observations to be included in the analysis, data must exist for every element included as an information source. If this is so, then the maximal number of cases for which there exist observations on each such element comprise the cases taken into account, and the number of cases is the sample size. If the sample size is greater than one, then you need to prepare for an exchangeable adjustment (see §6.6.3, §16.1). If the sample size under these rules turns out to be zero, then no data is available for the adjustment, and thus the results pertaining to the observed adjustment are not available. In these circumstances, [B/D] must be told what hypothetical sample size to assume. By default this is unity, but the obs control can be used to set any positive hypothetical sample size.
It might be that there is data available for every information source, but that you do not wish it taken into account. In this case the usedata control can be switched off, and the data (if any) are ignored. The obs control may be used, as above, to set a hypothetical sample size.
Occasionally you might wish [B/D] to operate upon some subset of the data. If so, you should switch off the autoselect control (which effects the automatic selection of data for many [B/D] commands) and then use the SELECT: command to select the desired subset of data.