Volume 1

Census 1951, Great Britain : one per cent sample tables.

  • General Register Office Northern Ireland
Date:
1952
Catalogue details

Licence: Public Domain Mark

Credit: Census 1951, Great Britain : one per cent sample tables. Source: Wellcome Collection.

    12/190
    1 Time table.—As already indicated, a paramount feature of the situation was the need for speed ; it was felt that if the full fruits of an advance production were to be realized, a time limit would need to be prescribed, and a target date of one year from census day was provisionally set by which the analyses should be completed and the classified results made available in published form. Such time limit left no margin for frills; during the months immediately following an enumeration, the central offices are invariably swamped with a deluge of millions of schedules and other documents received from the many local census officers throughout the country and all available staffs are preoccupied with the preparation and publication of the preliminary reports. So that the period left for the treatment of the sample was in effect much less than a year, neces- sitating strict economy both in its basic planning as well as in all the operational stages. Efforts have been rigorously bent towards accommodating performance with intention and, in spite of unforeseen obstacles which were more or less bound to attend what is an entirely new experiment, it is hoped that the publication of the more important analyses at a date only a few weeks outside the target period, with the remainder following shortly thereafter, will have met the need for an up-to-date presentation of the census characteristics of the population that will serve for most practical purposes until the results of the full Census are available. Scale of the Sample.—Two opposing considerations arise in the determination of the size of the sample to be used. Other things being equal, it is obvious that the larger the sample, the more precisely will it represent the total experience from which it is drawn, but that on the other hand the smaller it is, the less will be the time and labour involved in its extraction and treatment. Theoretically, therefore, the choice needs little more than an optimum reconciliation of these opposing factors in the light of prevailing circumstances. Within this generalization, however, two sizes have outstanding merit where decimal numeration is in current usage, viz., a Io per cent. sample or a I per cent. sample, because they not only readily lend themselves to simplicity in the initial selection of the sample items but are of even greater service in the interpretation of the results, since such sample records can be grossed up by sight to represent the whole experience simply by adding o or oo to any sample figure—an inestimable advantage to users where masses of statistical tables are in question. The total population enumerated in Great Britain in 1951, as provisionally published in the Preliminary Reports, amounted to 48-8 million persons, so that a Io per cent. sample would involve treatment of nearly 5 million records or a I per cent. sample about half a million. Whatever the scale chosen, the sample material would not only be subject to the same processes as attend the treatment of the full census records—detailed schedule examination, coding of occupations, industries, birthplaces, etc., and the punching and machining of tabulation cards—but all these operations would have to be performed by new staffs in the stage of learning their jobs when the rate of output would be at a low minimum. It was at once evident that a Io per cent. sample was beyond the range of practical considera- tion ; it would involve far more than could be accommodated within the allotted time and would equally impose an undue delay in the work on the main Census. A I per cent. sample, on the other hand, would be a somewhat small one imposing restrictions on the permissible tabulations. Even so, it was recognized that a programme within these dimensions would not be secured without the greatest care and foresight and in such circumstances it was deemed wise, as a first sample experiment, to aim low and adopt 1 per cent. as the chosen scale. Even if it did not provide all that might be asked of it, it would be a valuable advance record such as we have never had before ; it should go far to meet most of the demands that would arise and would provide a sound basis of experience in this field for future guidance. Selection and processing of the Sample records.—In the arrangements for selecting the records to represent the sample, the time factor was again a dominating one. Within the projected pro- gramme, it would have been impossible to have delayed operations until all the census schedules had been received and assembled at the national census offices in Southport and Edinburgh, whatever advantages might have accrued from the greater degree of control that would have been secured over the selection and extraction operations. The only practicable course was to arrange for the sample extracts to be made coterminously with the collection of the full census records in all local areas, thereby ensuring their production at the earliest possible moment, and with the added advantage of spreading this laborious feature of the operation as widely as possible. Instructions were accordingly prepared and simultaneously sent to each of the 2,251 local census officers, directing him and his team of enumerators to transcribe on to specially provided forms certain prescribed portions of the total census material for which each was responsible, designed in such a way as to secure a contribution from each district amounting as nearly as may be to I per cent. of the total records of the area, and as free as possible from avoidable bias. When the local extracts had been completed and scrutinised in accordance with the plan, they were forwarded to the appropriate census head offices—at Southport in respect of England and Wales and Edinburgh in respect of Scotland—and they were there subjected to a further process of scrutiny and overall check to’ ensure that the instructions had been adequately followed in accordance with intention. The population identified by the sample at this stage amounted to 488,395 for Great Britain as a whole, a figure which is as near 1 per cent. of the provisional total of 48,841,131* as could have been anticipated. For the separate local census areas the fit was a Mio Pa * The figure from the published Preliminary Census reports, viz., 48,840,893, is here adjusted in respect of errors which have since come to light, Vv
    digitised image 1
    digitised image 2
    digitised image 3