The Royal Statistical Society, together with the Statistics User Forum, has today (12 December 2013) submitted its official response to the proposals put forward by the Office for National Statistics (ONS) to change its methods in the collection of population statistics. In the consultation document, released in September, the ONS set out two approaches:
1. To take a radically new approach to the census and generate the information from administrative data held by government, supplemented by surveys.
2. To carry on with a decennial census, but do more of it online.
The RSS response begins by stressing the importance of collecting quality population statistics which affects the future planning in almost every national and local body which makes business decisions on a daily basis. Although it recognises the work undertaken to engage a wide range of users, the response points out the need to understand that some users of the data are moving into a different decision-making context, for example in health and education where decision making is increasingly taken within new governance structures or at different levels. These users must be engaged in the future for the statistics to remain relevant to their needs.
On the question of producing population statistics to the required levels of quality, the response points to the review undertaken by a team headed by Professor Chris Skinner from the London School of Economics. Professor Skinner concluded that before administrative sources could be used to compile accurate statistics, further work would be needed to ensure there was enough evidence that this could be done satisfactorily. The RSS response agrees with the conclusions of the review and its recommendations, but additionally proposes ONS should set out a clear strategy for the use of linked administrative data which goes wider than just population counts by sex and age.
The response goes on to deal with cultural and legislative issues that need addressing. The use of administrative sources would require data linking across government, for which further legislation would be needed. Related to this. The general public may be uneasy about personal data being used for statistical purposes; the RSS response proposes further and extensive work is undertaken to address issues of public acceptability.
In conclusion, the response states that it would be too risky not to continue with a ten-yearly census, but stresses the work on an alternative approach should continue in earnest, in parallel with preparations for a census in 2021. The response notes the move away from a traditional census in other countries has happened over a number of decennial cycles, which indicates the complexity of the work ahead.
John Pullinger, RSS President said: ‘The ONS should make a clear statement that it will not replace the ten-yearly census at least until real numbers using an alternative approach have been produced and accepted by those who use them. To continue with the ten-yearly census (with a strong online focus) may be unambitious, but to abolish it without a well-tested alternative would be reckless.’