Official Report: Minutes of Evidence

The Chairperson (Dr Aiken): I welcome Siobhan. Due to the current circumstances, we are slightly more spread out than usual. Jim is doing that for all the best reasons, but, rest assured, his voice will carry. The Clerk's briefing paper on NISRA is on page 16, and a briefing paper on the overview and priorities of NISRA is on page 17. NISRA officials will remain to answer questions on the next agenda item, which is the Census Order (Northern Ireland) 2020.

Ms Siobhan Carey (Department of Finance): It is nice to be here. Thank you very much. I am the Registrar General and chief executive of the Northern Ireland Statistics and Research Agency (NISRA). At NISRA, we are all very passionate about what we do, and I thank the Committee for the opportunity to provide you with an overview but, most importantly, to answer any questions that you might have for us.

I have colleagues from my senior team here today. Dr Tracy Power is the director of analysis; Dr David Marshall is director of census; and Kathie Walker is the deputy registrar general. They are sitting behind me, but I may ask one of them to come forward to answer any really specific details that I might not be fully on top of.

NISRA is an executive agency of the Department of Finance and incorporates the General Register Office (GRO). The GRO provides for the register of births, deaths, adoptions and gender recognition and administers the laws in relation to marriage and civil partnership. The statistics and research agency is the principal source of official statistics and social research for Northern Ireland. We produce all the main economic and social statistics, as well as conducting the census of population and housing every 10 years. The next census is in one year and three days' time: 21 March 2021. It is subject to the legislation going through, but that is the date proposed.

The agency has 486 permanent staff and a number of additional staff, of whom 388 are statisticians, and we are split more or less equally between delivering core services for the Department of Finance from its headquarters in Colby House and being outposted to over 30 separate public-sector locations, including all Northern Ireland Departments and a range of non-departmental public bodies and councils. That puts analysts close to policymakers and decision makers. In addition, we employ around 300 household interviewers working across all of Northern Ireland.

Our analysts provide government and citizens with external accountability through the publication of national and official statistics — information that they can trust, produced to a clear code of practice. They provide evidence to support policy development and decision-making and to monitor and evaluate policy outcomes. They also support the Programme for Government (PFG) through the development of indicators at the population level and then the collation of evidence on the underpinning report cards using the outcomes-based accountability (OBA) model, which, I am sure, you are all very familiar with.

While we work across the Civil Service, I will focus on the work of the Department of Finance element. Our main activities on the statistics side are the production of statistics on the economy, labour market and society. That means that we collect data from businesses and households, and the data that we collect serves the needs of Departments here. It also contributes to the production of statistics at a UK level. The labour force survey that we do contributes to the UK statistics on the labour market. Last year, we administered 11 different business services, collecting 65,000 returns from businesses, so there are quite a lot of forms coming in. Over the past few years, we have moved most of our business surveys online, and now 60% of the returns that we get are filed electronically. That is compared with 27% in 2016, so we have made a big push to move that data collection online. That reduces the burden on businesses, of which we are very mindful, and it makes our processing more efficient and cost-effective.

Returns from households and businesses are the raw material for the statistics that we produce, and we are really grateful for the cooperation that we receive from businesses and the public. We also provide an end-to-end survey research service of the general population. Our 300 interviewers approach some 45,000 households a year on 13 different surveys, seeking their cooperation on wide range of topics covering the labour market and health and well-being. Those surveys provide the data for 21 of the 49 Programme for Government indicators. In view of the evolution of COVID-19, we are pausing fieldwork pending further advice, and I am sure that you will understand why we are doing that.

The statistics that we produce are subject to regulation, and that is done by the Office for Statistics Regulation, which is part of the UK Statistics Authority. The Office for Statistics Regulation conducts assessments of the statistics produced for compliance with the code of practice under the themes of trust, quality and value. Those assessments and all the correspondence are published by the authority on its website.

This is an exciting time for anyone involved in data and analysis. Data are now everywhere. The provision of public services via digital channels provides much richer possibilities for evidence-based decision-making and developing greater insight from data. Alongside that trend, the Digital Economy Act 2017 came into effect in 2018, and that allows for the sharing of data for the production of statistics and research. While it is a permissive gateway in that it allows data to be shared, it does not compel Departments to share data for the production of statistics and research.

The investment that we secured from the Economic and Social Research Council (ESRC) to support the use of linked administrative data is starting to make a difference. The administrative data network is a partnership between NISRA and its academic partners — Queen's University and the Ulster University — and delivers new, innovative analysis to support government decision-making. Our first publication from the NISRA investment is on drug-related mortality, and that will be published next week.

Last year, we published our five-year corporate plan. That sets out the things that we need to do to meet the increased demands for effective data analytic support and the provision of timely and fit-for-purpose data to underpin evidence-based decision-making. It sets out our ambition for how we want our work to be increasingly relevant and for the kind of workplace that we aspire to be: innovative, collaborative and efficient. A key part of that is about how we disseminate our analysis, making it more thematic, more relevant and more accessible.

We publish statistics on a number of topics of interest to the public and government alike that attract considerable media coverage. Recent examples are the first publication on levels of loneliness in Northern Ireland and the number of deaths due to drugs and alcohol. One of our most-read publications 'NI: IN PROFILE' brings together key statistics in a single place, so that the interconnections between statistics generated by the different Departments and different sources are more visible. I brought some copies today that I can leave with you for afterwards. When the publication was first created in 2018, it highlighted the pace at which the population was changing and, particularly, how it was ageing. In 2019, we were less than 10 years away from having more people aged over 65 than children aged under 16. That was one of those nuggets that really resonated with people. The analysis has been updated, and you can see that on our website. It got quite a lot of coverage.

The Committee asked for our priorities, so, before inviting questions, I draw your attention to three specific areas. First, as I said, we are 368 days away from the census. Your next agenda item is on the census legislative scrutiny, and I will come back to that in a minute. Delivering a successful census is a top priority for us. The census is a huge logistical task, and it needs detailed planning. While it might seem a long time away, it is not something that we can pull together at short notice. In my view, it is a little bit like the Olympics: we have to be ready for the exact date. There is no point in our being ready a week after somebody else won the gold medal. That is very much how we focus on making sure that we are in peak performance and are ready for census day.

Secondly, the Programme for Government, as you know, has been refreshed as a result of the New Decade, New Approach agreement, and we have an important role in monitoring the population indicators in the current outcomes delivery plan. The technical assessment panel is chaired by Tracy and determines the thresholds for change that indicate whether an indicator is improving, worsening or staying the same. That is agreed in advance so you know how much of a movement you need to see to say whether it is the same, improving or worsening. Most of the indicators are based on NISRA surveys, such as the proportion of people experiencing crime or the employment and economic activity rates. We received recent investment to increase the sample sizes to improve the precision of the estimates and to support more detailed analysis by age groups, protected characteristics and lower levels of geography. That funding has one more year to run, and the investment has really made a difference in being able to disaggregate the main indicators to lower levels of geography and to smaller groups.

Not all the outcomes were able to identify suitable data sources for the indicators. Of the 49 indicators, 13 initially required data development across Departments. New sources have been developed for seven of those, and existing data have been repurposed or reported in a different way for a number of other indicators. A further two indicators are still in the development stage, and four measures have since been placed back on data development due to various emerging issues with data sources. As the new Programme for Government emerges, NISRA is ready to support the development of any new population indicators as well as to support policymakers in Departments in measuring at performance level whether, in outcomes-based accountability parlance, anybody is better off as a result of their interventions.

The third area that I draw your attention to is that, on 13 January, legislation came into effect that allowed for same-sex civil marriage and opposite-sex civil partnerships. The first ceremonies took place in the week beginning 10 February. The system changes required to implement that had to be carried out in a very limited time period. The other issue of note in recent times was a judgement to allow for marriages by humanist celebrants. That has been accommodated for civil marriages and civil partnerships but will require legislation to give full effect to the decision.

Finally and by way of context, one of the challenges that we face is keeping pace with the art of the possible. As more administrative data become available, there is an opportunity to revisit some of the analysis that we do. There is a vibrant tech community, particularly in the Belfast area, so recruiting and retaining staff remains an issue and presents us with a challenge. I apologise in advance if, given the events of the last few days, we are not quite as fluent as we might have liked to be. We hope that you will bear with us.

The Chairperson (Dr Aiken): Thank you very much indeed, Siobhan. Before I open it up to the rest of the Committee for questions, I would like to say something and to ask you a couple of questions. First, I congratulate you, because NISRA is one of the few areas of government and sources of information that is well respected and is quoted significantly in peer review publications. Indeed, one of the biggest problems that we have had in Northern Ireland is access to good, high-value data, and I commend NISRA for the work that it has done. It has made it both for trying to understand changes and what is happening. That area of data analytics is vital and the depth and detail of your knowledge has been significant. I would like you to take that from me, and I imagine that the rest of the Committee will echo my comments about the great work that NISRA has done. That was the good bit.

I have a couple of questions. Which Departments, if any, are not sharing data with you and why? The relationship between ESRC, Queen's and the Ulster University in partnership: who owns that data? Given the changes in data analytics and the rest of it, what have you been doing to look at best practice across these islands, particularly in how data analytics is used? One of the biggest problems that we have had with PFGs and outcomes-based government has been the measurement. One of the issues is that we are trying to measure 42-odd things, whereas places like Scotland are measuring six. What discussions have you had with other areas to look at those things?

Ms Carey: Which Departments are not sharing data? I would not say that any Departments are not sharing; it is about the speed at which they can share. We have data that we have asked for from probably all the Departments, but it may not flow at the speed that you would want. It is a permissive gateway, so the data owner can share it but does not have to. There are lots of discussions about making sure that the legal gateways are right. It is slow, and the Digital Economy Act 2017 is fairly new, so people are really finding their way with it. I do not think that it is an issue of people refusing to share.

The Chairperson (Dr Aiken): Is there an MOU with timings between you and other Departments? Should there be one that says that they need to share data within a specified time?

Ms Carey: I do not think that that is necessarily helpful. Most people are moving to share data as soon as they are confident that they can share it and that they have the legal gateway. This is all new legislation. Everybody is trying it out for the first time, and that is an issue across all the jurisdictions; in fact, we have been much more successful here in making data available than they have been in Whitehall. That is partly because the network is smaller, there is a more established level of trust, and there is more to be gained from making the data available for analysis. In the reinvestment that came from the ESRC, the three devolved nations were highlighted as being exemplars of how it could be done really well. The new investment was probably as much about refocusing how they can make that happen across the piece. I do not think that Departments are not sharing data; it is just that, when you want data, you want it now, and we probably get a bit impatient.

On the question of the relationship and who owns the data, all data has an owner. There is a clear data controller, and then, sometimes, we are processors. Each data set has an owner who has to sign off that they are happy for that to be done. I am not sure whether that was the intention of your question.

The Chairperson (Dr Aiken): It is just that some, particularly some universities, are putting a value on the IP of the data that they have. One of my concerns is that some universities have been bringing in data from other areas and putting their own value on it, whereas a lot of that data has come from government. We have already paid for it, as taxpayers, so we are being double-charged for data that we already own.

Ms Carey: The general principle, absolutely, is that all the data in government has been paid for by the taxpayer and, therefore, there are no charges for using it. I could not comment on what data they have in the universities, but certainly the approach within government is that it is a global public good and, therefore, should be made available.

The investment that we have from the ESRC brings the data into our safe setting. We link it, if necessary, to any data for the particular research purposes. We have a secure setting. Once researchers are approved and have gone through training, the project has been approved and the information owners have signed off that they are happy, they come in and use the data in our safe setting. All the outputs from that are checked before anything goes out.

Various methods can be used for monetising the data. We have not tended to put a pound sign in front of it, but, for our purposes, it is a global public good, so the more it is used, the more value it has.

Best practice in data analytics is very much an emerging area as more data moves online and becomes available. We work really closely with the Government Statistical Service in sharing best practice, sharing code, where we have it, or supporting each other. We also work closely with the Central Statistics Office. Some of our data that we publish comes from it, and we obviously do a joint publication on the census.

In terms of best practice in analytics, it is moving so fast that it is quite hard to keep up. The big investment that came into the system was from the Bean review, with the establishment of the Data Science Campus. That is based at the Office for National Statistics (ONS), but it is very much a UK investment. They have been really fantastic in helping us, through training, taking on projects on our behalf and advising on how best to maximise the opportunity.

The Chairperson (Dr Aiken): Thank you very much.

Mr Frew: Thank you very much for your presentation. I suppose that there are two aspects to this: the census and the really important work that you do to populate the surveys and the information for the Programme for Government. That will prove more pivotal in the years to come, as we try to align a Programme for Government with a Budget and vice versa, as well as with policies.

I will take you into the world of justice, for one minute. Justice is in honour of one of the outcomes — it is outcome 7, off the top of my head, but I am not sure — and, within that, the important information is that five indicators are used: the percentage of the population who believe that their cultural identity is respected by society; the average time taken to complete criminal cases; the prevalence rate; the respect index; and the reoffending rates. Each of the indicators has a different baseline, so some are recorded from 2013-14, some 2016 and another 2014-15. Is there a reason to explain why the baseline year changes?

Dr Tracy Power (Department of Finance): Do you want me to answer that?

Ms Carey: Yes, if you do not mind.

Dr Power: Do you want me to sit or talk —?

The Chairperson (Dr Aiken): You stay where you are.

Dr Power: I will stay where I am.

The Chairperson (Dr Aiken): Apologies, I have been corrected by the Clerk. Will you come forward so that we can pick you up on the microphone?

Mr Frew: Sorry for the disruption.

Dr Power: One of my roles is to chair the technical assessment panel that makes those decisions. We were starting from scratch the first time round and learning how to do that work. The work that we do is to look at each indicator in great detail, along with the producing statistician and the policy owner, so that they can get to understand their indicators well. As we started, we realised that we had a few important things to do. First, we had to make sure that the indicators were appropriate to be used in this way. They have to be reliable, and there is a checklist that we go through. Secondly, we had to determine what a threshold for change looked like. Remember: we are not talking about targets in the OBA terminology; we are talking about "Getting better", "Getting worse" or "Not changing". That is difficult. It sounds easy, but it is not. We had to determine at what point we will say that there has been a change either for the better or for the worse. We had to do that for all the indicators.

We were starting from a place where lots of different data sets were, perhaps, collected only every two years or were of such a quality that they could not be used for specific years. We had to make a judgement call on what the baseline would be, starting off. My preference was that all of them would have been at the same baseline year of 2015-16. Of course, you also get some finance years and some calendar years. You will find, as you look through the 49 indicators, that the majority are 2015-16. There are some that were unable to provide reliable data or any data for that year. That is why some indicators have different baselines.

As part of the work that we are going into on developing the one-year PFG and then the longer-term one that you will all know about, I will look again at the indicators. I want to get to a point where we have that single baseline year when we get to the multi-year PFG.

Mr Frew: Will that baseline year, even if it is synchronised, change? If it does change, what is the justification for changing it, and who will make that call?

Dr Power: It will be based on a technical assessment for change. You specifically mentioned justice, and one of the indicators in justice has a different baseline from the other indicators: the time taken for cases to go through the criminal justice system. The reason that it has a different baseline is that, at the time we looked at it, the data were not available for the subsequent year. When we go back to look at it, we will align that with the other indicators. It is a calendar year indicator — other indicators are mostly financial — so we still have that bit of difference as some will be, say, 2016 and some may be 2016-17, this time around. That work has yet to be done.

It is a valid question that you ask, but we had to start somewhere in trying to pull all that together. When you set a baseline year for indicators, in terms of things worsening or getting better, the year that you look at is not really that material. It is really about whether, over that time, things are getting better or worse.

Mr Frew: You are a statistician, but let me pull you into the murky world of politics [Interruption.]

The Chairperson (Dr Aiken): Oh, Mr Deputy Chair. I may allow this or I may not.

Mr Frew: If it was expedient for a Government to change the base year to make the figures look better, who makes that call?

Dr Power: I love that question. Interestingly, that is something that the panel that I chair has come across. I mentioned that we had statisticians in that group, but we also have the policy people.

Some policy colleagues absolutely see the point that you just made and say, "Tracy, we're not interested in the baseline year. You do what you have to do." There are other policy colleagues who would like to determine a baseline year based on what makes it look better. That is just their initial thinking. We talk it through and they get to an understanding. The decision is mine, so that takes the politicisation out of it.

Mr Frew: Good that the decision is yours, but would you not come under increasing pressure?

Dr Power: Oh, yes. I have been a statistician for 30 years. Most days I come under pressure from policymakers to do certain things.

Mr Frew: OK. That is interesting. I have another question about the indicators. There seem to be different tolerances and thresholds as to whether it becomes "no change", "negative change" or "positive change".

Dr Power: Yes, absolutely.

Mr Frew: For instance, you have the percentage of the population that believes that its cultural identity is respected by society. The current status is "no change". However, 2017 records 66·2%, whereas 2015, which is the baseline year, is 64%, so there is a change, but it has been recorded as "no change". Where are the tolerance levels?

Dr Power: That is also a very good question. There are 49 indicators from a mix of sources. Forty-six are from NISRA sources. Out of our sources, they come from administrative data sets or from surveys. If they come from administrative data sets, we have a call to make on the technical assessment panel as to what change is.

Say, for example, that the number of children attaining GCSEs who are on free school meals or non-free school meals is totalled, and the Department of Education has to make a determination. At that point, it would say, "The non-free school meal kids are doing very well. So, if you jump from the 85s to the 86s, is that really a change?" The Department, in conjunction with the panel, has to determine what we accept as a change. That is the easy bit.

When the data come from a survey, we have a sampling error around estimate, depending on the sample size. Some of our surveys with small sample sizes will have larger estimates, particularly as you get to the extreme of the distribution. If you have 50:50, it is like —

The Chairperson (Dr Aiken): Tracy, just to cut across, but when you are working with the Economic and Social Research Council (ESRC), obviously that must be in accordance with standard norms and statistical analysis.

Dr Power: Oh, yes, it is a mathematical calculation. We are assuming that our surveys are random samples. When you have a random sample, an easy calculation determines the error around the estimate. As the estimates get further to the extremes, that error gets higher. If you are interested in justice, the crime victimisation rate is really low. We have to have quite a large sample size to affect any change that we can pick up in that survey.

That links into what Siobhan said about investing in these surveys. That was the purpose of increasing the sample size.

Each indicator is looked at in itself and what survey it comes from. That is why we wanted the investment. The labour force survey, in particular, is a really good source of many of the indicators, but the sample size was just not big enough for some of our purposes. It is such an important survey. In the indicators, you have employment rates, employment rates by council, and economic inactivity rates.

We invested in that survey, which is coming to fruition. We hope to find that the change needed between the estimate and the baseline will be smaller. Therefore, if we increase the survey, it becomes more sensitive to the change.

That is why there is a different number for each indicator. The process has been to look at the survey and make that determination.

Mr Frew: I have one question on the census itself. There seems to be a massive change in our inner-city populations and in homes of multiple occupancy for foreign nationals and people coming here to work. Neither the Housing Executive nor the council seems to have got on top of the issue of how many people are actually living in some households. It seems to be very hard for the council to monitor and enforce the multiple occupancy measures. How will your staff be armed in order to get that information? There could be a language barrier, and people may not know what the forms are about. How can you be assured that there are not 12 people living in every household instead of four?

Ms Carey: I will ask David to come in. We will not be arming them. They will be well armed, but not with arms.

Dr David Marshall (Department of Finance): The census relies on a register of addresses, so we create a register of addresses, and colleagues in the office make sure that it is as accurate as possible. We gather information from the Housing Executive on its houses in multiple occupation (HMO) list, but we go wider than that: we have a field force that goes out to assess in each small area across Northern Ireland — each enumeration district — to find out whether we have all the addresses that we need. There may be addresses that were not collected from government that people have created.

On foreign nationals, we make the census available in translation booklets in up to 17 languages. It is translated into 17 different languages. There is a leaflet that goes along with the census form pointing people towards a call centre where people can talk through an interpreter. This time round, the census will primarily be made available online. From the earlier test, we recognise that people are using online tools to translate the questionnaire itself.

If we do not get a response from an individual household, we have people who will go door to door to follow it up. They will be able to direct people or to supply translation materials or in some instances, if they have language skills themselves, support people in filling out the census form.

Mr Frew: How do you know that it is the truth when people fill it out online? Secondly, when you are physically at somebody's front door and you look into a living room and see 10 bodies, six of whom are only visiting, but really they are living there, how do you ascertain that information? How do you claw the truth out?

Dr Marshall: Two different elements are used to quality-assure the figures. After the census is over, we run a second survey called the census coverage survey. Basically, we go out to a sample of the houses that filled out the census or an area where people did not fill out a census form. It is a bit like in agriculture when they ring fish: we capture the same data from the house to make sure that somebody was not missed. It is a similar process, if you can imagine. That quality-assures it and makes sure that our estimates are as accurate as possible.

As Siobhan said, we also have access to the administrative data from other systems. For example, we have access to administrative demographic data from the National Insurance system, the health system and the school system. We use that to layer on top of the census to make sure that we do not miss anybody or, if we do, it is not from the lack of trying.

Ms Dolan: Thanks very much for your presentation. I am very excited about the census, but that is just me.

Ms Carey: You are in good company. [Laughter.]

Ms Dolan: My questions are about the multiple deprivation measures for rural areas. There are seven domains; income is one of them but expenditure is not. I am from a very rural area in Fermanagh that has no public transport, so I have to have a car. Therefore, I have to pay for insurance and fuel. My area of Fermanagh could be more deprived than somewhere in west Belfast, but because expenditure is not taken into account when allocating public funding, west Belfast gets — I am not just saying west Belfast; Castlederg, I think, is the most deprived rural area. Therefore, public funding goes to the areas that are more deprived, and there is a bit of an inequality. Can you comment on that or is there a way of getting round it?

Dr Power: OK. The deprivation measures were done in 2010 and updated in 2017. In 2010, the data available were very different from the data that we had in 2017. We are very well aware of the fact that, given that the population is more dispersed in rural areas and that a lot of the domains depend on a certain density of population, it could look, if you were just looking at the top 20% of areas in 2010, as though there was no rural deprivation whatsoever. We were part of the Committee's response in trying to improve that. We looked at new data sources for 2017. The two new sources that we had, which are really important to know about, were broadband access, which is fine, but we had public —.

The Chairperson (Dr Aiken): I just want to put it on record that I do not think that any of us thinks that broadband access is fine anywhere.

Dr Power: No, no. It is an indicator of rural versus urban.

The Chairperson (Dr Aiken): Particularly rural.

Dr Power: That is the point. We had it then to measure that. The other source was public transport access, and we did a lot of work on that.

The income domain does not — you are quite right — include expenditure, but what we have done is adjusted it at council level for housing costs. If you look at the income domain, you will see that we do not have an expenditure part of the equation, but we have adjusted it for differential housing costs so that there is some element of what you highlighted to try to make it more comparable.

At the time, in 2017, we think that we did the best job that we could with the data available. If we are commissioned to update it at some point in the future, we will reopen that debate. What we have found is that the top 20% of areas on the 2017 measure does include rural areas. We think that that happened because we used better data that time round.

Ms Carey: It is better, but it is still not perfect.

Ms Dolan: Yes. It is a work in progress.

A recent DAERA report indicates that income is 9% lower in rural areas. That sort of information could be taken into account.

Dr Power: It is taken into account.

Ms Dolan: OK. That is grand.

Mr Wells: There is a proposal that this will be the last census. Does that proposal apply to Northern Ireland? If it does, what do we envisage happening in 10 years' time?

Ms Carey: There is a debate about whether this will be the last census. It is pretty much the same debate that took place 10 years ago. Obviously, things have moved on a lot in the last 10 years. The decision to take a census is a political one. The Office for National Statistics does the census for England and Wales, National Records of Scotland does it in Scotland, and NISRA does it here. The decision about whether this is the last census will be taken in each of the jurisdictions. The things that might work in its favour are that it provides much improved administrative data and that there is the ability to do a synthetic census, so you do not necessarily need to do a traditional one. David, do you want to add anything?

Dr Marshall: Some countries are beginning to move away from traditional census-taking. The plan for Northern Ireland is to move more towards online delivery of the census in 2021. There is a move in that direction. The United States is a running a 2020 census. It changed the way in which it ran the census in 2010 by splitting the census into two pieces: the census itself, capturing the core number of people and houses; and a subsequent survey with administrative data that is a lesser burden on the public. They have split it into two. The French have decided to move towards a rolling census whereby they do one part of the country at a time, because it is more cost-effective. There are different approaches.

In the United Kingdom, we need to take a decision after the next census. That will link in with what happens in this census. What happens in the Republic of Ireland might also influence decision-making because they are deciding whether to move away from the census as well. We will have to look at that too.

Mr Wells: It would be a four-country decision. It would not be a case of one country deciding to go on its own?

Dr Marshall: The Northern Ireland Executive could decide to run a census in Northern Ireland if they wished to do so. The legislation allows that to happen. It would be much more costly than running the 2021 census because we process the census in Northern Ireland alongside our colleagues in the Office for National Statistics.

If the Executive were to decide to do it, they would have to do it in the clear knowledge that it could be quite an expensive exercise.

Mr Wells: Are you content that, with more modern data-gathering techniques, you are getting enough regular data to replace the census?

Ms Carey: That is an ongoing judgement. The data maturing and the census next year will give us a good indicator of how good it is now, but there will always be a judgement call as to what is good enough and balancing that against the cost. If you look at how things have changed in the last 10 years, and if that accelerates, you would be making a decision in five years' time about how confident you are that what you could get is either good enough or whether you want to do another "traditional census", although it looks nothing like what censuses used to look like.

Mr Wells: How much would we save if we did not have one?

Ms Carey: The additional cost of conducting the census is £35 million over 10 years.

Mr Wells: Just for us?

Ms Carey: Yes. I will say a little bit about a question that often comes up and perhaps pre-empt some of your other questions: the cost and value of having good data to make decisions. It is a question that we often get, so we are well rehearsed in how we respond to it. People often look at the cost of statistics and think, "That is really expensive", but I like to think of official statistics as being the flip side of your public accounts coin. While your annual accounts tell you literally what you spent the money on and where it went, the statistics should tell you how that spending is changing the things that you want to influence and how you want to move the dial on and inform you about where you may need to spend the money.

I like to think of it as not just how expensive statistics are but the alternative question about how expensive it would be not to have good statistics that you can trust. It would mean making decisions in the absence of data. If you take the cost of having statistics just on the economy, for example, that represents an absolutely miniscule proportion of the block grant here.

Official statistics tell you whether what you are dealing with is a big issue or a small issue. They tell you where it is, the characteristics of the issues that you are trying to address and how you are doing in addressing them. Without statistics, it could mean that hospitals are built where they are not most needed.

I can give you an example: we have been working with colleagues across the NICS on data support COVID-19 planning. Official statistics give you the numbers of people over 70 who live on their own. It puts them in households, which you would not be able to get from any other data source. It shows how many people in various districts are over 70 and living alone or over 70 with underlying health conditions or who are living in households where everybody in the household is either over 70 or over 65.

The sort of analysis that we have been doing over the last couple of days to support the decision-making on where the vulnerable populations are is not available from any other source. I always say that, without data, you are just another person with an opinion, so I am in the camp that believes that data are of real value in helping people to make the right decisions at the right time.

Mr Wells: Well, you have pre-empted my next two questions, so that is me. You were well prepared for that one.

Ms Carey: Was I? We get it all the time.

Mr Allister: Chair, can you clarify whether we are having a separate session on the census?

The Committee Clerk: We are.

Mr Allister: So, it is not appropriate to ask census questions now?

The Committee Clerk: It is one that is specifically on the SL1.

Mr Allister: Right. So, if we have general questions about the census, now is the time to ask them. OK. I will not start with the census, but you said that the Programme for Government has been refreshed.

Ms Carey: It is being refreshed. In 'New Decade, New Approach', it says that there will be a new Programme for Government.

Mr Allister: But you have not been shown anything?

Ms Carey: I have not seen it, but I understand, Tracy, that there are discussions about it.

Dr Power: The Executive Office leads on the outcomes delivery plan with which you are familiar; that is not a NISRA lead. It will be for one year, for next year, 2021. It is my understanding that that will be available in the next few weeks. That is based on the priorities in the New Decade, New Approach agreement. From there, we will be working with TEO to develop a multi-year Programme for Government from April 2021 onwards. It will include whatever indicators politicians require at that time — whatever the outcomes are after a level of engagement .

Mr Allister: Have you seen a revised —?

Dr Power: No.

Mr Allister: May I ask you about marriage registrars? As I understand it, you are the Registrar General.

Ms Carey: I am.

Mr Allister: The registrars who conduct civil marriages are attached to our councils. Is that right?

Ms Carey: They are employees of the councils.

Mr Allister: They are employees of the councils. How does that fit in with your role?

Ms Carey: Kathie is the deputy registrar-general.

Ms Kathie Walker (Department of Finance): The registrars are all employed by local councils, which employ registrars and deputy registrars. We put in place the legislation that they work to, but it is the councils' responsibility to deliver registration to the public. We put the legislation, policy and guidance in place and support them. As far as costs, for example, are concerned, we pay for everything to do with the registrars, even though they are not our employees, but they are there to represent us as much as they represent the councils.

Mr Allister: You say that you support them. As you mentioned, same-sex marriage has come in. For most registrars, that was not the situation when they applied for and secured the job. How do you support a registrar who says that, in conscience, because of religious belief, they do not wish to be associated with or cannot perform a same-sex marriage?

Ms Walker: First, as I said, they are an employee of the council. They would have a job description and a contract with the council, not with us in the General Register Office.

Mr Allister: You said that you support them.

Ms Walker: We do support them. We provide their IT and their papers and we tell them what they can and cannot do.

Mr Allister: But what you are telling us is that this subject is not for you.

Ms Walker: This subject is not for me. The NIO has made it quite clear, in its consultation paper and in other papers, that as a person working for a public authority, there are no exemptions and, therefore, they have to carry out same-sex marriage.

Mr Allister: So, there are no conscience rights.

Ms Walker: They have no conscience rights.

Mr Allister: That is not the case in other countries.

Ms Walker: There are a few countries where that is different.

Mr Allister: Yes; Canada, South Africa.

Ms Walker: Yes, there are a couple. What the NIO has put in place is exactly the same as it is in the rest of the UK.

Mr Allister: Do you think that it is conscionable that somebody can be compelled, in breach of their own religious beliefs and conscience, to carry out that function?

Ms Carey: It is not really —.

Ms Walker: I cannot give you my personal opinion. I can only tell you —.

The Chairperson (Dr Aiken): I am sorry; Jim, we cannot ask for opinions. It is not a question of opinion.

Mr Allister: No, I am sorry: the question was whether she thinks that it is conscionable that those whom she supports —.

The Chairperson (Dr Aiken): No. My ruling is that you are asking a question about opinion there, Jim.

Mr Allister: This is the second time in the Committee that any attempt to raise the issue has been oppressed. This is a Committee with oversight over marriage. Here is the Registrar General for marriage, and you are telling me that I cannot ask such a question: really?

The Chairperson (Dr Aiken): Yes.

Mr Allister: That just adds to the feeling of resentment that many registrars have about the manner in which they are being coerced. Yet, when someone seeks to raise it, it is ruled out of order, when you have sitting here the Registrar General with oversight for marriage.

The Chairperson (Dr Aiken): Your view has been noted and recorded.

Mr Allister: Noted and ignored, I suspect.

The Chairperson (Dr Aiken): Do you have any further questions?

Mr Allister: I have further questions.

There has already been a question about how you test the veracity of answers given in a census.

Let me take a practical example. As, I am sure, you are aware, there is a very political campaign to big up the interest in the Irish language and Ulster Scots, and you will ask questions, Dr Marshall, about proficiency in, say, Irish. If there is a political campaign to up the level of perceived interest in that, how do you verify the answers? If someone answers that they are proficient in Irish or in Ulster Scots, whatever that is, how do you know whether that is right or wrong?

Dr Marshall: The proposed question for the 2021 census will have two elements for both Irish and Ulster Scots. It will ask about people's stated technical ability: whether they can understand Irish; and whether they can speak, read or write Irish. If they can speak Irish or, indeed, Ulster Scots, they are then asked how often they speak that language. That is, fundamentally, a question for householders to answer themselves. After the census has been carried out, we go out and, as I said to Mr Frew, run a coverage survey. We also run a quality survey after the census. We go out and ask a sample of people whether they answered the question and whether they understood it, and we assess whether they answered the question accurately. There is, therefore, a follow-up survey that compares the results from the census with a second question asked of the same people.

Mr Allister: You ask them on paper whether they can speak Ulster Scots or Irish, and they say yes. You ask them how proficient they are, and they say that they are very proficient etc. You then go out and knock their door. You ask them — in English — whether they can speak Irish, and the answer is yes.

Dr Marshall: We have asked the question on proficiency in Irish in the past three censuses.

Mr Allister: Yes, I know.

Dr Marshall: The figures from those three censuses are broadly very similar. So, unless there was some sort of process that we were unaware of, we feel that, essentially, the figures are accurate and fit for purpose. We have asked these questions in the social surveys, and we get broadly the same answers.

Mr Allister: If these are self-professed answers that cannot be independently verified, the process is open to potential exaggeration or abuse.

Dr Marshall: All surveys are open to potential exaggeration and abuse.

Ms Carey: They are self-reports.

Mr Allister: A way of doing it, of course, would to be ask an Irish speaker to complete the survey in Irish. Would that not be a way of doing it?

Dr Marshall: The option to complete the 2021 census online will be available for those who want to complete it in Irish or Ulster Scots. We will know the number of people who complete it in that way.

Mr Allister: Will that be publicised?

Dr Marshall: Yes, we will publish that at that time.

Mr Allister: Will there be any read-across between those who profess proficiency in Irish and those who deploy Irish in completing the survey?

Dr Marshall: If the figures allow it, we will, of course, do that analysis and make it available.

Ms Carey: It is a self-reported measure.

Mr Allister: Yes. Is your agency subject to FOI?

Ms Carey: Yes.

Mr Allister: Thank you.

The Chairperson (Dr Aiken): Thank you very much indeed.