Tony Dent to Ed Humpherson: ONS Covid Infection Survey and Code of Practice for Statistics

Dear Ed,

It was good to see you in Glasgow, I am sorry that I missed your important session. I hope it went well. For the record my personal view is that there is little value in continuing with the present systems and I will be interested to review the discussion when I have the time.

Meanwhile, I am pleased to say that we have now received a reply from Peter Benton and I will include that in the record of the correspondence between BSC and the ONS on the CIS over the course of the past 10 months. I expect to send that record to your team investigating the CIS in the next few days, and I will also copy you on it.

I doubt that we will have any time on the 27th to discuss the wider issues of the code of practice, so I hope that we might set up an alternative time to specifically discuss that. I am sure you will recall that the two primary concerns of BSC were, firstly, strengthening the code with regard to Value for Money and secondly, providing some guidelines in respect of minimum response rates.

You have commented that you see little reason for amending the code for value for money and I accept that the code does reference the issue. However, the discussion on the 27th should serve to explain our concerns as to the current weakness of the code and I return to this further below.

In respect of our concerns about response rates, we are pleased that you propose to take the issue further with the GSS. The requirement for action here is urgent because poor response rates increasingly seem to be accepted without comment; a factor that is severely limiting the representation of our ethnic minorities and poorer members of our society and which must call into question the mantra “for the public good”.

I would draw your attention to two recently published important studies whose results, we believe, may be inaccurate as a result of poor response rates:

Firstly, Public confidence in official statistics 2021, as conducted by Natcen on behalf of the UKSA and
Secondly the CDEI report on Public attitudes to data and AI: Tracker survey, conducted by Savanta on behalf of the Department for Digital, Culture, Media and Sport.

Although the former is appropriately conducted by modern standards, the overall response rate was just 23% and we believe it was significantly lower within important subgroups of the population. Certainly, the headline results do not conform with other more qualitative information available (for example please see Public Understanding of Economics – J. Runge et al).

Response rates for the CDEI report are not presently available, but they have said that they will ensure some representation of ethnic minorities in the telephone ‘boost’ study for the next wave. Ethnic minorities were not represented at all within the boost sample for the first tracker and, therefore, the study results cannot claim to be representative of the UK population in the manner inferred by the headlines of the published report.

Clearly, providing some specific code of practice guidelines in this area might well prove to be controversial, but it is important that we recognise the limitations of a lot of the research presently conducted and not continue to turn blind eye to these facts.

As you know, our concern extends beyond the world of official statistics and BSC is supportive of the aim of the OSR to extend the reach of the code of practice; indeed we would like to see its influence reach into the commercial sector.

Strengthening the code in these two areas, linking value for money to guidelines that ensure improved inclusion and therefore representation can only serve to strengthen our statistics, both national and other.

I hope we can find time for consideration of these important issues in due course.

With all good wishes,

Tony
Director, BSC

Tony Dent to Ed Humpherson: ONS Covid Infection Survey and Code of Practice for Statistics

Related links