Impute index of multiple deprivation

[nikhilwoodruff]

We don't have this in the FRS microdata, and per discussions around UK public service imputations it seems important to have imputations take into account of statistics splitting service usage by index of multiple deprivation, so we should try and impute this for households. I think the best method here is to map the constituencies to their average IMD (IMD is of small local areas, smaller than constituencies), then use the constituency weights to get average IMDs for each national weight sum.

[nikhilwoodruff]

I found only one data source with index of multiple deprivation rankings for the entire UK rather than just England or Scotland. IMD are only published in relative form it seems, so we'll have to use this data.

The data source seems to have some councils that are out of date/combined/from pre-2025 local authority districts (9% of our local authorities can't be matched to an IMD value), so for those local authority districts I've matched them to their geographically closest LAD which does have an IMD. This results in the below, which looks sensible to me.

I sense-checked some of these, e.g. the Northern Irish LADs with very high IMD values (least deprived) are genuinely non-deprived areas (per the BBC, "Lisburn and Castlereagh was the only council area that has no areas among the 100 most deprived").