1990 POPULATION CENSUS Detailed data based on a 2 per cent representative sample (1992)

III. THE METHOD OF SAMPLING

Reduction and adjustment were carried out independently in the different area units. Consequently, the subsample of each area unit contained - all dwellings of the initial subsample remained after reduction, and - all persons living in those dwellings. As indicated above, the purpose of reduction and adjustment was to ensure that the sampling fraction be 2 per cent for the dwellings as weíl as for the persons of all area units in consideration. The result for dwellings was, however, somewhat biased, as ctoellings and other accomodation units could not be distinguished in the course of the sampling process. Sample estimate of a totál of dwellings was therefore usually lower than the preliminary figure of the same data, the difference being the sample estimate of the number of other accomodation units. In the case of Budapest e.g. the difference amounted to 5327, while the preliminary figure and the 2 per cent sample estimate of the stock of dwellings were 7B9177 and 7B3850, respectively. The method of adjustment slightly differed from the generál procedure in the case of people living in institutional households; here only 20 area units (19 counties and the capital city) were considered in place of 78 such units. The reason for this was that proper representation for more detailed breakdown would not have been possible as the corresponding cell counts were rather — or in somé cases extremely — small. 239