OPINION12 April 2021

The census is dead, long live the census!

Are the days of the traditional census coming to an end? Don't be surprised if not, says Guy Goodwin.

People and data

The 21^st of March, census day in England and Wales, has come and gone and most of us will have dutifully filled in our individual census forms. This may be the last census to be collected in this way, with the Office for National Statistics (ONS) taking a decision in 2023 on the future approach.

At its simplest, the census is an official count of the population. It’s recognised in countries across the world as important to support national and local planning, resource allocation and delivering services including housing, transport, education and healthcare.

So, the right question is not whether we need a census but how to deliver it – via a census form or by linking together administrative data? At a moment in time or on a rolling basis? This time round we’ve had a digital-first census where we filled in our census forms online.

If it is the last traditional census, it’s been an eventful one. In 2020, the Scottish census was delayed to 2022 due to the coronavirus and workload pressures. Traditionally, the Registrars’ General across the four nations of the UK work together to ensure consistency of outputs including to align on a common census day.

There was an inevitable outcry from those who want comparability in statistics across the UK, and perceptions of political interference. It is a legitimate point of debate, though, as to whether a census is best delayed during a pandemic, given our circumstances may be different now than in a year’s time when the effects of the pandemic may have eased.

One of the most difficult topics for the census authorities has been how best to record sex and gender identity, given the ongoing public policy debate around birth sex and self-identification. There seemed an inevitability about this ending up in court[ 1 ], with the national statistician being accused of bias by some on both sides of the debate.

The good news is we should get a better count of how many trans people there are because we don’t know currently with any certainty[ 2 ]. If it is under 1% of the population, as many expect, then the risk of outputs by sex being unduly affected by how trans people fill in the question should prove unfounded.

The combination of the pandemic and the increasing diversity of the population bring unanswered questions for those that argue the traditional census and survey infrastructure we have used to underpin our social statistics is nearing ‘end of life'.

An administrative census would link together the likes of DWP, GP, council tax and other datasets, but it’s unproven as yet as to how best to adjust for over-coverage in these sources or how they would capture the increasing complexity of our lives, for example around family formations. Administrative records are usually set up for specific administrative purposes, as witnessed by the use of the ONS Covid-19 infection survey (rather than GP or hospital records) to inform us how many people have the Covid-19 virus, including asymptotic cases.

The main argument against the traditional census has always been its cost, not its burden or bias. The current census requires 20 minutes of our time every 10 years for benefits that we largely take for granted, while the case around user needs versus costs has always stacked up well. We could change from this, of course, but don’t be surprised if the decision in 2023 is to continue with what’s worked well, at least most of the time, in the past.

Guy Goodwin is chief executive at the National Centre for Social Research (NatCen)