Sample Boosts


Interesting to hear Penny Young (@PennyYNatCen @BBCr4today) being challenged on BBC Radio 4 about the use of an “ethnic boost” in the UK Citizenship Survey. The interviewer appeared to be pre-primed to challenge the use of an ethnic boost in a random population sample because it “gave more importance to the views of ethnic minorities”.

Admittedly Penny introduced the term without explanation but then went on to explain that the purpose of the boost was to ensure that a sufficient number of interviews were obtained from certain subgroups of the population to enable reliable (as measured by standard error) estimates of key characteristics of those subgroups. For full population estimates, responses from subgroups who have been over sampled (i.e boosted) are given a lower weight in proportion to how much they were over represented in the initial sample. So for full population estimates the boosted minorities’ views are NOT given higher “importance”.

Much as Penny tried to explain the interviewer did not want to be convinced. But then who wants a lecture on basic sampling theory at 7am?

Leave a Reply

Your email address will not be published. Required fields are marked *