Making the census count – and accountable

October 3, 2022 | Blog
Home > Making the census count – and accountable

by Richard Rodger (Edinburgh University)

The author is Professor Emeritus of Economic and Social History at Edinburgh University and a Fellow of the Academy of Social Sciences. He has published widely on British and European urban history, including a recent article in the Economic History Review.


For a country obsessed with its history, we are rather casual with our records. We authorize transcriptions of our censuses, but do not cross-check the accuracy of the digitized version.


Take this example from Rochdale in 1871; see figure 1. The enumerator’s handwriting is very precise. Transcription should be straightforward. But it is not a full (never mind an exact) transcription as it appears in the digitized census.


Gone are:

  1. house numbers
  2. the relationship to head of family
  3. precise age of those under 1 year old
  4. occupation
  5. birthplace below county level
  6. disability – blind, lunatic, deaf, etc.


Figure 1.   Census and transcription for Rochdale, 1871


Worse still, editorial information has been added which undermines the claim to an ‘accurate transcription’. For example, a ‘birth year’ column is introduced but is an automated addition with only a 50 per cent chance of being accurate. Mary Ann Fitton declared she was aged 57, which the digitized census calculates to be 1814 (i.e. 1871 minus 57), which is assigned as her year of birth. But suppose her date of birth was in between June and December 1813? She would be 70 when the census was taken on 2 April 1871. The column purports to provide accuracy. Instead, it manufactures inaccuracy which misleads those who then search through record linkages for death certificates, wills, and other documents.


In Scotland, matters are worse. Paying subscribers are not even permitted to see an image of the original document. Worse still for Scotland, it is not possible to consult the enumerators’ books page by page, and so searches are at the level of an individual record and so more time-consuming, and thus more expensive to search. And of course, the digitized Census of Scotland for 1911 is not yet available on the same basis as in England, and as for the 1921 Census in Scotland, well …


Just as frustrating is the inaccurate spelling of street names in the digitized censuses. This could quickly and cheaply be cross-checked and corrected by a panel of local historians who could also vet the surnames which are often mis-spelled (approximately 1 in 20 according to my recent sample of 13,000 individuals) in the digitized version. As the heaviest users of the digitized census, local historians would willingly provide a reservoir of free labour to improve the quality of what should be a priceless historical asset – OUR censuses.


Surely it is time for some form of audit and quality control to be introduced into this aspect of government expenditure.


To contact the author: