Seòmar agus comataidhean

Yes, yes and yes. It is entirely reasonable to ask why there were varying rates of return between Scotland and the rest of the UK; it is a perfectly reasonable question to try to get to the bottom of. However, we should also be comparing our experience with experience elsewhere, especially in the rest of the industrialised world and especially through sociodemographic comparisons, to see where there are similarities and differences.

We are not yet at the end of the process of understanding the differences, but it is unavoidable to conclude that people being in their houses during the pandemic was a significant contributory factor in the ability to reach people—especially those from more challenged sociodemographic backgrounds.

I am not sure whether Mr Cameron was one of the MSPs who went out and saw the census collection. He is indicating that he was not able to see it. MSPs saw the efforts that went into knocking on doors again and again to try to reach people. If people are not in, which was happening a lot, it is difficult to get them to take part in the process. This is an unscientific conclusion, but I draw it as a non-statistician, and not as a census professional, but one might conclude that there is definitely something in that. However, that does not make me revisit the question whether the timing and the decision in Scotland were correct or not. I think that the decision that was taken in Scotland—as it was in the majority of countries—to not go out and send thousands of people into communities to knock on doors and have face-to-face conversations with people at a time when we were telling them not to do that, was the right response.

To answer Mr Cameron’s question whether we should be trying to learn every lesson from the experience in Scotland, in the rest of the UK and in the rest of the world, especially in countries with which we can compare ourselves best, I say that we absolutely should do that. The reason why is that I think that we are dealing with a societal trend; I do not think that we are dealing with a specific moment in time. If it was about a specific moment in time, it might have been in countries where a census was conducted during a lockdown. The rest of us are dealing with an on-going trend, and we are going to have to work out how to get information from people, in this context as in many other areas, when they do not want to provide it, do not trust the process, do not understand it or do not have enough time, as people said were their reasons for not taking part.

I am in favour of reflecting on everything. However, one person’s sensitive question is another person’s less than sensitive one. Therefore point 1 is: what is a sensitive question? For point 2, I go back to the statistical response that we received when we asked people what their reasons were for not taking part in the census. I do not want to repeat myself at length, but I note that concerns about certain types of questions being a main contributory factor in people taking part or not came in at less than 5 per cent. Does that mean that one should not think about that? No—of course one should. Frankly, we need to think about everything.

Because of the very nature of what a census is supposed to provide—so that we can understand society in the 21st century—we ask a wide range of questions to understand the kind of country that we are in. I will leave it to the statisticians to go through them. The census is a million miles away from where it was 20 or 100 years ago, because we require much more information if we are, among other things, to provide the public services that we wish to provide in a way that reflects our society. That is why we have to ask the broadest range of questions.

To return to the central question of whether we should be prepared to think about all kinds of questions, my answer is that we absolutely should.

If it is helpful to the committee, I am happy to provide the background of the different types of messaging that we use across different platforms, from television through to social media. It was full-spectrum communication and was aimed at different target audiences. I do not have to tell Mr Ruskell that the audience that uses TikTok is quite different from the one that uses Facebook, which is quite different from one that watches certain television channels, which is different from other types of audience.

It is a reflection of the times in which we live that one has to communicate across all those platforms and more. No doubt the conclusion will be that we will have to do more of that the next time the census comes round. However, Paul Lowe made the important point that the lessons that we are learning from the process are not unique to the census. They are reflective of a societal trend and a challenge for anybody who wants to collect information about the public to help to provide the best public services, in the case of the census, to understand the labour market or to understand any number of other things about society at different stages.

How can we do that in a way that is genuinely reflective of the whole of society? Sarah Boyack has spoken about that before and she is right to highlight the point. There are variable rates of return. In shorthand, the more affluent an area, the higher the turnout; the lower the income demographics, the lower the rate of return. I am very much simplifying, but that is one of the most significant factors. Because of that variable rate, we must have mechanisms in place to ensure that the conclusions of the census or other statistical products are genuinely reflective.

The survey work that takes place after the census is really important. I have no reason to disbelieve that the committee understands that. I think that I am right in saying that that work—I am looking at my NRS colleagues before I overclaim—is the biggest survey in Scotland after the census. We are talking about a return of the best part of 30,000—off the top of my head, I think that it is between 25,000 and 30,000—and, as committee members know, it is normally about 1,000 for a representative statistical survey. Therefore, we are talking about an exercise that is 25 to 30 times the size of that.

Significant efforts are being undertaken to make sure that targeted information is obtained. I am sorry; I should have stressed the point that it is targeted within those harder-to-reach parts of the return from the census, to make sure that the overall picture provides not only the statistical certainty of population numbers, as we are confident it does, but that level of granular detail about people of different backgrounds in different communities, so that the provision of important public services, such as health and education, is done on the basis of reflective and high-quality data. I and my NRS colleagues are confident that that has been achieved in the 2022 census.

Dr Allan used the word “you”. I was not part of the decision, so it is difficult for me to think my way into—

Yes, I am content. I am looking back at decisions that were made at the time, which I did not play a part in, but it is obviously important to look at those decisions and try to understand the rationale behind them.

First, there is an international context. Out of 83 nations that planned to conduct censuses over that period, 59—71 per cent—delayed their census field collections. That includes not only Scotland but Germany, Italy and Ireland. I can provide the list to the committee if members wish to know about the other nations among that 71 per cent that made the same decision. Only 10 countries—12 per cent—in that period proceeded with their field collection as previously planned.

I am satisfied with the rationale, but I do not lose sight of the fact that the advice to the public was to minimise contact with one another. The wider context was of holding a census during the biggest pandemic in 100 years, so I am content that the correct decision was made. Now that we know that we are within touching distance of a 90 per cent return rate for the census and can be assured that the quality of the data is of the standard that is required to complete the census, I think that the right decision was made.

We could probably use all the time in the session to discuss that question, because it is the nub of trying to understand the experience of the recent census process here and what will be required at the time of the next census to make sure that we collect the appropriate quality of data from society.

Throughout the census collection period, I spent a lot of time with my professional colleagues, who are here and online, trying to understand the phenomenon of reduced collection rates in certain parts of the country. I will let them do some of the technical statistical explanation of that.

I should say that the issue is being evaluated currently, so you are asking us to take the temperature of that issue on the basis of what we understand thus far without having completed all the work.

I am sitting giving evidence to colleagues who are unusual in society, in that, as MSPs and candidates, we spend a lot of time knocking on doors. I am appreciative that Ms Boyack and a number of other MSPs took the time to go and see how the census was being collected. Having been out, she was able to see the phenomenon that is, I think, entirely consistent with what we as members of the democratic political community are aware of—namely, that there is a reducing rate of participation in elections, reducing turnout, reducing rates of data that we are able to collect when we do doorstep visits and higher numbers of people saying that they are not prepared to say how they are thinking about voting at election times. We hear a variety of reasons to explain why they will not take part.

Before the end of the census collection period, I said to colleagues at NRS that I thought that it would be particularly important to understand, in qualitative and quantifiable terms, the reasons why people were not participating—the reasons that they were giving, as opposed to others’ interpretation. The answers are really quite instructive. They are worth sharing with the committee so that they are on the record.

The answers come from 1,200 people who had not returned their census forms, making it larger than a standard opinion poll sample size. They were asked their main reasons for not completing the form—why they were not doing so. The biggest reason, for 35 per cent of people, was that they felt that they were too busy—that they did not have enough time. The next biggest reason, for 17 per cent of people, was that they were not aware of the census. The next biggest, for 14 per cent of people, was that they did not realise that they had to complete it. Lastly—all of which came in as reasons for 5 per cent of people or less—were concerns about privacy, trust in Government, the nature of questions, access to paper and so on. I imagine that members would recognise that kind of response from the times that we knock on people’s doors.

We should not lose sight of the fact that, by the end of the process, nigh on 90 per cent of people had returned their census form. The question is: how much more does one need to do in 2022, or in 2031 or 2032, by the time the next census comes around, to maintain that high level and high return rate?

My colleagues who are much more versed in the statistical side will be aware of this, but I note that New Zealand is about to undertake its census—next year, I think—and has set its target for a return rate of 90 per cent. My observation is that we are seeing a phenomenon here in Scotland that is not unique; indeed, it is occurring in other countries. The question is: what can we learn from our experience so that we can maximise the rate of return next time round?

I am sorry—I do not want to hog the microphone if colleagues from NRS want to make a contribution. A lot of evaluation work is being undertaken, which will no doubt be shared with the committee when it is published. Do colleagues wish to flag anything in relation to my answer to Ms Boyack?

Following the closure of the main census collect period on 31 May, on 22 August, the census coverage survey also came to an end. Although that may mark the end of live operations for Scotland’s census 2022, it certainly does not mark the end of the work that is required to deliver high-quality census outputs.

Scotland’s census is a highly complex programme that, in common with other modern censuses, consists of many elements. Although it is understandable that much of the focus so far has been on the public-facing elements of the census—particularly the census return rate—that is not the deciding factor in determining whether a census has or has not been a success. As the international steering group set out in the paper that it provided to the committee, and as Professor Sir Ian Diamond and Professor David Martin explained during the evidence session two weeks ago, it is the combination of three pillars that will deliver the high-quality census outputs that users require. Those are high-quality census returns, of which an almost 90 per cent return rate has been achieved; a coverage survey and peer reviewed statistical techniques; and the use of high-quality administrative data.

This was the first primarily online census and generally that worked well, with 89 per cent of respondents completing online. That exceeded NRS’s target of 75 per cent and clearly indicates a strong preference for the majority of citizens to use digital rather than paper completion. That shift in public preference should be taken into account for any future census exercise or similar significant public engagement. The census was also the most flexible one ever delivered, with options for completion digitally, by paper form and through assisted completion by telephone or field force.

Despite concerns, the month-long extension to the collection period led to a significant improvement of return rates at national and local levels. The national return rate increased by 10 percentage points since 1 May but, crucially, the extension also ensured that there was enhanced coverage across the country, with 30 of the 32 local authorities achieving return rates of more than 85 per cent and no authority achieving less than 83 per cent. Eighteen of those local authorities achieved a return rate that was greater than 90 per cent.

There are, however, emerging indications of shifts in public attitude in Scotland to the importance of the census, and there is a need to understand that. However, that phenomenon appears not to be restricted to the census, and is emerging in other areas such as completion rates in broader Scottish social surveys. The committee recently heard from Sir Ian Diamond that that trend has been seen in declining participation rates across recent years. As such, it will be important to understand and plan for such an event up-front in the design and risk management for any future census.

However, with a final return rate of 89.2 per cent, I hope that committee members, and indeed the public, are reassured by the words of the members of the international steering group who, in their submission to the committee’s inquiry, noted that they

“consider that the main census enumeration has provided the foundation for a high-quality set of census outputs, in terms of coverage of the population”,

as well as Sir Ian Diamond’s evidence that the census in Scotland will still produce “really good” data.

As recommended by the international steering group, NRS is working at pace to secure the necessary access to key administrative data sets for the purpose of census estimation and adjustment. That expansion and enhancement of administrative data use beyond the original plans for estimation of census response will put NRS in a strong position to deliver a high-quality set of census outputs for Scotland’s 2022 census.

The Scottish Government and NRS are extremely grateful for the time and expertise that the international steering group continues to provide as it moves through planned post-collection quality control and assurance work. In the coming months, NRS will continue to focus on planned post-collection quality control and assurance work to deliver the high-quality census outputs that users require.

Finally, I put on record my thanks to the millions of households who participated in Scotland’s census 2022.

I look forward to answering your questions.

It is for NRS to make decisions as to what it will publish. However, I want maximum transparency so that not only NRS but Government ministers and the people who hold us to account can understand the lessons.

And election literature.

It was £144.6 million.