Abstract

For Census 2021 in England and Wales we planned to make greater use of admin data than we have previously. Uses ranged from standard products we had planned from the start to ones we developed as a result of increased risks of low or biased coverage. One of our uses of alternative data sources was to create an independent estimate of the number of occupied households that should have responded: our Alternative Household Estimate (AHE). We created this using a simple prediction model, census responses and census paradata, and multiple administrative data sources. The results agreed closely with our independent coverage estimates created using our dual-system estimation model, and we have been able to use these to explain to users how we were confident of the final census data. This presentation touches on the range of uses we put administrative sources to, then concentrates on the creation and use of the AHE.