1970 HUNGARIAN CENSUS OF POPULATION Information on the data collection and processing (1977)

I. THE COMPREHENSIVE REWIEV OF THE DATA COLLECTION AND PROCESSING - 8. The electronic data processing of the census of 1970

Considering this, a less complicated solution presented itself which has proven, however, reliable enough. Thus the correction of the registered material of the flats and buildings could be concentrated to 2 programmes, the full-scale control having been also taken into consideration. In the first step, a full-scale correction was performed with the data of dwellings and buildings too. It was identical with the control performed in checking of the personal data. In the second stage, the correction of the absurd code-numbers and of the data referring to the communities followed. These data are to a lesser extent depending on the others; consequently, it was expedient to separate them from the largescale operations of correction. In the same stage, it came to the checking of the connections existing between the dwellings of the same building, as well as some corrections have also been carried out by the machine. In lack of kitchen or kitchnette, if there was a summer kitchen, the latter was registrated as kitchen, and so on. According to the connections of the control activity, the actual values of a criterion affect the eventual actual values of another. The connections are unidirectional and irreversible. The serial number of the error concerned each possible combination of a criterion with another. If a connection of such kind does not hold, we are still unable to say which of them is erroneous, since both notions may be erroneous with the same probability. It seemed expedient that the erroneous criteria should be defined after the checking of the connection. If some groups of criteria prove so faulty that the correction seems to be useless and impossible, the machine is able to replace it, taking the best sample of the census district into consideration. This method is known in the literature as the "colddeck techniques". Such criteria may be ever "prefabricated" eventually, according to the respective mean values obtained in the former census. If the connections in question give no sufficient instructions for the correction, these "prefabricated" values are attributed to the faulty record. If the whole record is charged with errors, the whole has to be replaced by a faultless sample (so-called "hotdeck" techniques). We have been prepared for eventual changes of this kind to be performed by the machine, but practically they never occurred during the correction procedure. This method has completed - in a separate step - the data concerning the stock of flats and buildings by household data. The data completed offered opportunity for the correction of some connections (for which no information was available previously). These corrections could be executed on the basis of the connections existing between the number of households, the inhabitants and flats and so on. When taking the aspects of the processing into consideration, several complementary and concentrated code-numbers were attached to the records of the dwellings and buildings. In the following, the stocks of dwellings and buildings have been separated, which served as the basis of the tabulation. The correction of the sample survey of 25 per cent representativity The automatic control and correction of errors as described above referred to the data obtained by the 100 per cent full-scale survey.- The correction of the code-numbers of the representative survey sample based on 25 per cent representativity was somewhat different. Namely, the criteria conform to those of the full-scale survey have been copied by the machine, out of the records of the 25 per cent sample which underwent a control from the point of view of the completeness. Thus the stocks amounting to 75 per cent of the basic survey were completed by means of the copied data which resulted in a fullscale stocks of 100 per cent. The corrections performed in the copied records of the sample of 25 per cent representativity - after the conclusion of the procedures of automatic 73