Please Complete Your Census
Craig Dalzell - 5 May 2022
If you haven’t done so already, please compete and submit your census. If you have already done so, thank you – and please encourage your friends and family to submit theirs if they haven’t.
There are many reasons to ensure that you do – not least because it’s a legal obligation and should be considered as much a part of your civic duty as paying taxes, voting and serving on a jury. It is by far and away the most important statistical exercise conducted by our nation and goes a long way towards informing public policy directly and helping to correct and anchor other forms of data gathering. I make this plea as an informant of public policy and an unabashed stats geek. Neither I nor Common Weal have been paid by the Scottish Government to write this article (we wouldn’t accept such a payment even if it was offered). They haven’t even asked us to write this. We’re doing this because we understand how important it is that it gets done properly. Our own ability to create and analyse policy depends on access to high quality, accurate data and the census is one of the most important tools in our box of facts and information.
I completely understand that many, many barriers have been put up in front of you. For the first time, the census has been conducted primarily online with a presumption of "digital first" and paper copies available only on request. The last census or two was conducted largely by paper and you would have been posted out a large document to fill in but this time you would have received a smaller letter directing you to the census website. It’s easier to miss and easier to lose. In years previous even to the mailshots, the census would have been conducted by in person visits – it looks like the Scottish Government is going to have to roll out more of them than it expected to.
The decision to delay the census by a year due to the pandemic (the only British nation to do so) has proved to be disastrous. Scotland lacks an independent broadcaster and the Scottish Government was evidently unable to compensate and properly advertise the census in isolation to the other nations, particularly given how centred much of that media is on England. Those who remembered previous census years know that they always came in 2011, 2001, 1991 etc all the way back to 1801. This is the first and only full census in Scotland that has been conducted outwith that pattern (except perhaps the 1941 census that was cancelled due to WWII – Northern Ireland has experienced some census irregularities in certain years such as 1926).
More than an inconvenience to respondents, this may well cause significant problems for data users in Scotland going forward. By delaying the census till 2022, Scotland has greatly reduced the ability to compare ourselves to the other nations in the UK. A one year gap might not seem like much but it plays havoc with statistical models that rely on regular time spacing between data points. If we do want to compare data like-for-like with England or elsewhere then we’’ll have to apply some statistical modelling to estimate either what Scotland’s data would have been like in 2021 or what England’s would be in 2022.
How this is done and the assumptions that are made in doing so are likely to cause a lot of statisticians a lot of headaches – especially as it could be another year before the data starts to become publicly available meaning that statistics that must use census data and are thus still using 2011 data will be twelve or more years out of date. This isn’t even a problem covering just a few years though. The census is an important historical statistical tool. For centuries to come, Scottish data now has a “blip” that must be accounted for – one that will show up in every single study that attempts to look at the Scottish population at this time if global pandemic.
Statisticians are smart at this though – those problems can be worked around. A more serious problem is incomplete data due to folk not responding. Not every population study needs a full headcount of the country. Scotland produces annual population estimates based on more limited surveys and, as I explain in our Demographics of Independencepaper, even a 1,000 person poll can be shown to be mathematically accurate regardless of population size so long as it is properly weighted. But that last clause is important. You can’t properly weight a limited sample of the population if you don’t have some idea of what the whole population actually looks like. The census acts as a regular anchor to correct any deviations that may well creep in to data based on more limited estimates.
National polls and estimates often don’t have the resolution to accurately reflect what’s happening at local levels. A 1,000 person poll across all of Scotland will, on average, only sample 31 people in each of Scotland’s 32 Local Authorities. That simply isn’t enough to determine which towns get what share of public resources, how many new houses your community is likely to need built in the next few years, where the next GP clinic should go or how any number of other public services can be run in your area.
As you read this, you’ll also (I hope) be voting in the Scottish local elections. Once those votes are counted and the new councils are formed, they’ll be tasked with working out how to run your area according to the resources they have but that task will ultimately be determined by how much they know about what’s going on in that area. If you don’t appear in the data, your politicians can’t speak for you, can’t make the case for more money for your area and can’t invest in the services that you need. So please, if you haven’t done so, fill out your census form and submit it despite all of the barriers that have been needlessly put in front of you. You can start the process at census.gov.scot (and there’s plenty of information there to help you, including a phone number to call if you get stuck). And I’ll look forward to seeing the results of your efforts the next time I go diving through the stats to inform our next policy paper.