Approximated Social Grade on the 2021 Census

Approximated Social Grade (ASG) has been produced for the last three censuses – 2001, 2011 and now 2021.  ASG uses an algorithm, developed by members of the MRS Census & GeoDems Group, that assigns a likely grade to each household based on the characteristics of the Household Reference Person (HRP).  The HRP is the Census equivalent to the Chief Income Earner used for social grading in market research.

The 2021 ASG approach is broadly similar to that taken for 2011, however the 2021 development employs different modelling techniques using different variables.  Therefore, it would be misleading to use Census outputs to examine trends in Social Grade profiles between 2011 and 2021. Further details of the 2011 Social Grade model may be found here.

Development Report
This document summarises how Social Grade has been mapped to the 2021 Census.

England & Wales

The Office for National Statistics has released a series of Census 2021 output tables and datasets on Approximated Social Grade (ASG).    The outputs on ASG includes cross-tabs on HRPs and characteristics of usual residents.  All ASG datasets and tables are for households containing Household Reference Persons aged 16 to 64.  

The datasets are downloadable from the Nomis website, at various levels of geography from total England & Wales down to Output Areas.  See Census 2021 data on Nomis here. 

Alongside this release, ONS has published a statistical bulletin on the new ASG data – the bulletin is available from here. 

Northern Ireland

The Northern Ireland Statistics and Research Agency has released two output tables on ASG.  NISRA applied the same algorithm as ONS and the tables are also for households containing Household Reference Persons aged 16 to 64.

Further information and to download the outputs see here 

Applying the Social Grade approximation to other data

The Social Grade approximation can be modelled onto an external dataset, such as a market research survey, using the same algorithm that has been applied to the 2021 Census. 

This first entails ensuring that the dataset contains all the necessary variables, presented in a required order and coded exactly as prescribed. 

The process is explained in this document and uses a set of R scripts and data files that may be downloaded here.