Data management/Multiple response questions

The information at different levels of the hierarchy can be linked via index or key variables. Thus, in Breedsurv, three variables (WARD, VILLAGE and HHOLD) are used together to uniquely define the homestead and link the HOUSEHOLD and PRODCAT files. Use of index variables ensures efficient use of storage space (because data contained in the HOUSEHOLD file does not need to be repeated for each record in the PRODCAT file).

Recognition of the different layers in a hierarchy at which data are collected and the use of different files to represent those layers can also ensure that errors in the choice of denominator for the calculations of percentages and other statistics do not occur.

A hierarchy has not been established for homestead member within cattle activity in this case; both variables are stored together as 'flat' data in the PRODCAT file (CS11Data2b). This was because no additional information was collected about the different family members other than whether or not they were engaged in the activity. Some extra care may be needed, therefore, in analysing these data.