(Title Image: Hull Daily Mail)
Time for another look at a minor issue concerning independence – largely a follow-up to last year’s post on national statistics.
In 2021, another census (or, more accurately, three separate censuses) will be undertaken in the UK. With the issue popping up in the media in the last couple of months and with the ONS undertaking work preparing the census form itself, now’s as good a time as any to take a look at it.
How the current census works
A full census has been undertaken every ten years in the UK since 1801 with one or two interruptions – Northern Ireland during the partition years and also during the Second World War. The last census took place in 2011 and I went through some of the key findings for Wales on the old Oggy Bloggy Ogwr during 2013.
The goal of any census is to directly collect key demographic and sociological data to provide an as accurate as possible record of who’s living in any given area at a given time – a system that goes back as far as the Romans or, more famously, the Norman Domesday Book.
The data collected during a census provides governments and other decision-makers with the information needed to properly plan public services (such as health and education), budget for population and demographic changes (the Barnett Formula is at least partly based on Wales’ relative population compared to England) and determines where things may need to be built and when.
Some of the key information collected includes household size and composition (age, gender), nationality and race, occupations and other defining characteristics (religion, sexuality etc.).
While data findings from census returns are usually published within 2 years, the information contained on individual census forms isn’t supposed to be publicly released for 100 years.
The census is usually completed on a paper form on a nominated day (27th March in 2011) by one person in each household (with separate arrangements for places like prisons, hospitals, halls of residences etc.) – though an online version was available for the first time in 2011.
In 2011, every paper census form was electronically scanned at a rate of 15 questionnaires-per-minute at a central facility in Manchester. It’s also an offence to refuse to fill out a census form, with people liable to a £1,000 fine. 120 people were convicted in 2011.
As hinted earlier, there are three separate censuses in the UK. Scotland and Northern Ireland are responsible for collecting their respective censuses, while EnglandandWales has a single census overseen by the Office of National Statistics (ONS).
As far as I can tell, as it hasn’t been included in Schedule 7A of the Government of Wales Act 2006, the census is devolved to Wales – but the option of legislating for a Wales-only census has never been taken up by the Senedd. Existing EnglandandWales laws (Census Act 1920) still apply, with the Senedd being consulted on any changes.
Issues with the Census
Nationality & Race – This is the issue most people reading this will have heard about recently. As things stand, people from mixed or ethnic backgrounds other than white can only describe their nationality as British and not Welsh. The obvious solution to this would be to record ethnicity/race and nationality separately – and to all right-minded people they are separate; anyone who considers themselves to be Welsh, regardless of ethnicity, is Welsh. Hopefully, this will be fixed for next year. There’s a similar issue with gender identity.
Privacy issues – Collecting information which identifies individuals leaves room for abuse, even if the strictest measures are taken to control and protect information collected through the census. In some dystopian scenario, the information could be used to persecute minority groups and the government will know roughly where and how many of them there are.
Outdated collection methods – By and large most people will fill out the census on a single paper form, though the 2021 census will be moving “digital-first”. While a census is probably the best way to get a precise picture of what the country looks like, in the age of big data it may be surplus to requirements to a certain extent, which leads nicely into….
Value for Money: Is a formal census needed anymore? – The 2011 EnglandandWales census cost £482million. In 2014, it was decided a 2021 census would go ahead despite speculation the 2011 census would be the last one of its kind. Its future beyond 2021 is being questioned again. Things like population projections are becoming more sophisticated and are often published annually. One alternative to a single household-filed census would be to use administrative data (health records, DWP, electoral register, student administration) to get the data a census would normally collect. That said, the value of the information collected by the census has been estimated to be worth anything between £720-945million (pdf).
How censuses work around the world
Republic of Ireland – Ireland undertakes a traditional household census every five years, run by the Central Statistics Office; the next one is due to take place in 2021. The Irish census collects by and large the same information as the census in EnglandandWales with some differences in how homelessness is defined and counted. The Irish census is said to cost €55million (£46.2million) spread over three years.
Switzerland (pdf) – Since 2010, the Swiss census has primarily used administrative data from federal and cantonal government registers, with 200,000 households selected to fill out structural sample surveys, 40,000 selected to fill out a survey on specific themes (i.e. health, transport) and a further 3,000 surveyed on current affairs topics (i.e. internet use, race relations, eating habits, environmental attitudes). The new census is undertaken annually and the new method has been said to have saved up to CHF 100million (£78.6million).
The Netherlands (pdf p8-11) – From 2011, the Dutch have collected census data from public registers and sample surveys already undertaken on behalf of statistical authorities (i.e. labour force surveys). As a result, the Netherlands is said to have one of the fastest turnaround times in Europe for census data and it’s also saved money; the new register-based method only cost €1.4million compared to hundreds of millions for a full traditional census.
New Zealand – The census in New Zealand takes place every five years, with the latest one being undertaken in 2018. It’s a traditional paper census like EnglandandWales and cost NZ$126million (£62million), though there have been concerns that as many as 10% of people didn’t take part, as well as concerns about cost overruns.
Denmark – Traditional censuses were phased out in Denmark in 1970 and in 1968 a Central Population Register was set up using municipal registers (it’s compulsory to register with a municipality in Denmark) to provide census-like data. Additional registers provide information on households and detailed population statistics are published annually.
The Census & Independence
Wales does have a mini census of its own in the form of the Welsh Government’s National Survey – which is similar to the sample surveys carried out in Switzerland and cover a range of topics including internet use, public service satisfaction etc. The sample size is said to be 12,000 people – who receive a shopping voucher for taking part – and it’s (somewhat surprisingly) carried out as a face-to-face interview.
An FOI request from 2017 says the National Survey costs £1.8million at the time (pdf), saving £1.3million on the five separate surveys which were rolled into it.
You would assume that the responsibility for carrying out a census or processing census-like information would fall on any Welsh successor to the Office of National Statistics (see also: IndyWales & National Statistics).
Option 1: A Traditional Census
Given that there are far fewer residences in Wales than the UK as a whole then you would assume, logically, the cost of and time needed to carry out a Welsh census – presumably every ten years as now – would be proportionally less than the UK as a whole.
That isn’t necessarily the case. Scotland’s 2011 census cost £65.2million across ten years (pdf – p14). On a proportional basis a Welsh census would – based on those figures – cost £37million, which is higher than our proportional share of the 2011 EnglandandWales census costs (around £25million or 5.2%).
There are about 1.4million households in Wales and even adding a 5-10% uptick for student halls of residences, prisons etc. would mean that electronically processing a Welsh-only census at the same rate as the 2011 census would only take around 10-12 weeks. Data analysis and publication would, naturally, take a bit longer.
One of the main advantages of a “proper” census is it provides far greater detail on the status of minority groups than mere estimates and, subsequently, better informs decision-making – the Welsh language is an obvious example. Traditional censuses also include information that will be of particular use to historians and genealogists – that state registers alone might lack – because they give an idea of social and demographic trends over a longer period of time and captures information about individuals that may otherwise be lost or not recorded at all.
Option 2: A Mixed Census
A little like Switzerland, under this option there would be a sample household census (perhaps collecting information like that in the National Survey) alongside data taken from public registers. The process is likely to be significantly cheaper than the current census, though it may leave data gaps and perhaps not be 100% accurate, only providing a very good estimate.
It’s hard to say how much it would cost, but it’s worth comparing the National Survey’s £1.8million annual cost to the possible £25million+ cost of a full census. Also, register-based information is already collected, it just needs to be properly processed – which would come at a price but not a prohibitive one.
Option 3: A Register-Only Census
Under this option, only public registers and statistics would be used with no household survey. It would presumably be the cheapest option and provide a quick turnaround time, with figures easily published annually – as in the Netherlands and Denmark.
I would lean towards Option 2. The National Survey provides useful social information and it would be a shame to abandon it completely – though whether the sample size should be expanded or not from 12,000 is a matter for debate. It makes more sense to use the data already in the hands of public bodies – which is almost real-time – instead of a big survey at one point every several years.
This would in itself raise a number of related practical questions: Should it be mandatory for people to register with a municipality when they move within or first arrive in Wales (as in Denmark)? Should the National Survey move from in-person interviews to online (unless requested)? How could a revamped census meet the needs of historians? What type of data should be collected from public registers and why?