Women and Equalities Committee
Oral evidence: Race Disparity Audit, HC 562
Wednesday 20 December 2017
Ordered by the House of Commons to be published on 20 December 2017.
Members present: Mrs Maria Miller (Chair); Tonia Antoniazzi; Philip Davies; Rosie Duffield; Kirstene Hair; Eddie Hughes; Jess Phillips; Tulip Siddiq.
Questions 1-43
Witnesses
I: Dr Richard Norrie, Demography, Immigration and Integration Research Fellow, Policy Exchange; Professor Shamit Saggar, Professor of Political Science and Public Police, Institute for Social and Economic Research, University of Essex; Andy Shallice, Policy and Information, Roma Support Group; and Dr Debbie Weekes-Bernard, Policy and Research Manager, Joseph Rowntree Foundation.
Written evidence from witnesses:
Joseph Rowntree Foundation (RDA0005)
Witnesses: Dr Richard Norrie, Demography, Immigration and Integration Research Fellow, Policy Exchange; Professor Shamit Saggar, Professor of Political Science and Public Police, Institute for Social and Economic Research, University of Essex; Andy Shallice, Policy and Information, Roma Support Group; and Dr Debbie Weekes-Bernard, Policy and Research Manager, Joseph Rowntree Foundation.
Q1 Chair: Good morning. This is the first of our oral evidence sessions on the Government’s race disparity audit. The audit publishes Government data on how people of different ethnicities are treated across public services, in areas including health, education, employment and the criminal justice system. We are really grateful to the people and organisations that have already submitted written evidence to the inquiry. That evidence can be seen on our web pages.
Our focus today is on the quality and usefulness of the data revealed by the audit. In future sessions we will consider the policy issues that the data has raised and what the Government’s response should be. We are tweeting about the inquiry with the hashtag #RaceAudit.
Before we start, I thank, on behalf of the whole Committee, each of our witnesses for taking the time to come along and, I know, to prepare for the session. We are enormously grateful to you. I know that some of you have not actually been along to a Select Committee before, so we hope that this is a good experience for you. Before we start with Jess’s questions, could I ask each of you to say your name and which organisation you represent?
Dr Norrie: My name is Richard Norrie and I work for Policy Exchange.
Dr Weekes-Bernard: My name is Debbie Weekes-Bernard and I work for the Joseph Rowntree Foundation.
Professor Saggar: I am Shamit Saggar and I am the associate pro-vice-chancellor at the University of Essex.
Andy Shallice: I am Andy Shallice from the Roma Support Group.
Chair: Brilliant. I note that the acoustics in the room are bizarrely appalling for such a relatively new building, so I ask you all to use your best projected voice. There is amplification, but it can be very difficult to hear sometimes, so I apologise for that. We will start with a set of questions from Jess, and then other members of the Committee will come in with other questions through the morning.
Q2 Jess Phillips: Do any of you feel that there are any significant gaps in the data of the first release of the audit? Do you know if there are plans to update it in the next release?
Dr Norrie: Yes, there are significant and substantial gaps in the data. You can define gap in two ways: gaps between groups and within groups, but also things that are missing in the data. In terms of differences between groups, I would start with employment. There is a strong difference between ethnic minority and ethnic majority, and there are differences among ethnic minorities, too.
Q3 Jess Phillips: So the difference in the rates of employment is not included in the report?
Dr Norrie: That is there, of course. In terms of what is missing, the health statistics are all on mental health; there is nothing really on the incidence or prevalence of specific conditions or on outcomes. One of the key things about the exercise was to look at how different peoples are treated by public services. There is nothing on cancer survival rates by ethnic group, which would be very interesting.
Professor Saggar: Richard is right, but I think it goes back to the second part of the question to do with plans. My understanding is that the disparity audit is the beginning of something rather than the end of something. You will find in the chapter dealing with health the anomaly of looking at some things rather than others, which clearly is not satisfactory. That is okay if you then want to get on to looking at health outcomes, health experiences and access to GPs. Essentially, a lot of this is going through administrative data, and taking a view on which is the richest form of administrative data, so you can get that out the door and into the first cut of the report.
I understand that the disparity website design will be updated, both on what it already covers and on additional items. I do not think it is the case that this is the finished product.
Jess Phillips: The beginning and the end.
Professor Saggar: You would have to take two views on the plans. Presumably, the Government should take a view on whether we are done or whether we will update this.
Q4 Jess Phillips: What do you feel was missing? Do you feel that anything very specific was missing? There was quite a lot of criticism when it was first released.
Professor Saggar: There probably was, and there probably would have been. I was one of the people who served on the advisory groups, chipping in ideas about where you should start and where you should not start. I thought that the view taken by that group most times was that there will always be things left out and you will always slightly upset a particular sort of perspective or a constituency. In the end, that was not an argument not to go ahead and have the disparity audit. The labour market section is quite good, but it only begins to get into quite a lot of other things that we know in social research about people’s experience in the labour market by ethnicity. One could go on.
Andy Shallice: There are three distinct areas, I think. Rowntree would support this in their submission: there is no reference to early years provision at all, in terms of access, progression and benefit from early years provision. There is nothing about higher education participation and attainment, and very little on access to social security state benefits. We would add that migration status, which perhaps we will come to, is also absent.
Dr Weekes-Bernard: It is important to say that we thought it was a useful site, because it gave us the opportunity to see lots of data in one place. For those of us who work in that sort of area, seeing lots of data in one place about ethnicity gaps is really important, because it reinforces how important it is to have information about quite shocking gaps in a lot of the public policy areas that were revealed in the site.
One thing we thought was quite interesting was that there wasn’t anything about poverty rates. I suppose you might say, “Well, Joseph Rowntree would say that,” because we are an anti-poverty organisation, but a lot of the datasets that the website drew on are the sort that we draw on when looking at the groups who are more prone to poverty, or more at risk of falling into it. The fact that many of those datasets were there, but that the poverty rate was not and poverty was not mentioned at all was, we thought was slightly odd.
The pages that look at employment pay and benefits, housing and education are really interesting and reveal some interesting data. But they are also the sorts of drivers for high poverty rates. I agree with Shamit, because lots of organisations may look at this site and say, “Well, you haven’t looked at a variety of things that we are personally quite interested in.” Those areas also revealed a lot of information that would help us know more about poverty, so the fact that poverty was not mentioned seemed quite odd.
Poverty rates for minority ethnic groups are really high, and poverty is rising. Although we think it is important to find out about different groups and household income, where people are in the labour markets, and what people spend their disposable income on—the high cost of housing, for example—if we look at all those things together, that will help you to understand what some of the drivers might be for why half of Bangladeshi families are living in poverty, for example. You should include poverty because you are kind of talking about it, even though you didn’t really mention it.
Q5 Jess Phillips: Are there plans for that to be included?
Dr Weekes-Bernard: I don’t know if there are plans to include it, but we would definitely recommend that it be included. We already have information about something like this, but we are not actually naming poverty or saying that is there, and it would be a simple thing to do. The datasets that they used are the sorts of datasets that we use. They use households of below average income, for example. We use that to get a calculation of poverty rates. If we can do that then—it is not difficult.
Dr Norrie: On the way in, as I was swatting up, I noticed that there is a reference to poverty in the actual report that was published to accompany the launch of the dataset, but I couldn’t see any poverty statistics on that website. At Policy Exchange we run a website called integrationhub.net, and we set out to do something very similar to what this website does, and to tell the story of integration in this country and look at the differences between ethnic groups. We have data on poverty on there, but only on relative poverty. It would be good to see absolute poverty rates as well.
Q6 Jess Phillips: A number of written submissions that we have received mentioned the lack of intersectional analysis on the data that was in the audit—obviously when looking at BAME women and BAME disabled people, as well as BAME poor people, one could easily assume that they would be affected worse by all of the categories in the audit. Is it possible to disaggregate and analyse the data in that way, using the datasets provided in the audit?
Professor Saggar: It rather depends on which bit of data you are looking at. It is broadly made up of a lot of administrative data, which has rich, large sample sizes, low response rates, plenty of coverage and reliability. On the other hand there are surveys, some of which are quite sophisticated. When you get into surveys, you tend to run them with smaller numbers, and then you run into issues about the limitations of disaggregating and cross-sectioning.
It really depends on which bit of data you are considering at any one time. The labour force survey is quite rich and will allow you to do quite a lot of that to most people’s satisfaction, and to tell a story within a story. The “Understanding Society” study that we run at Essex is another example—it is the largest household longitudinal study. It is huge and there are lots of opportunities to cut this by that. Then you have other surveys that make it much more difficult to do that. There is no point in trying to pretend otherwise, and that everything can be broken down and broken down, but some things that are large can be reliably broken down.
Q7 Jess Phillips: And what would they be? Can we break it down by gender, for example? That is quite binary.
Professor Saggar: Yes. I always stick to the rule of the three Gs: gender, generation and geography. For example, the labour market circumstances of a first-generation Pakistani woman living in Greater Manchester are often very different from a second-generation Indian male living in suburban London. That is gender, generation and geography. You can do that straight from understanding society. They are completely different stories, and yet both are ethnic minorities in the UK today. One is at the top end of the labour market doing fantastically well and so on, on average, and the other is mostly not. You want to do that, otherwise you will say, “These are two south Asian people who have the same experiences.” Actually, they don’t—not at all.
Dr Norrie: The more you cut into the data, the smaller the number of observations you will have, and so the noisier the estimate will be. As you know, I was swotting up before I came in, and I noticed something on one of the pages about the prevalence of a possible personality disorder. That is from the adult psychiatric morbidity survey, which found that 17.7% of black women had a possible—
Jess Phillips: Seventy per cent?
Dr Norrie: Seventeen, and I think it is a very generous estimate. It was basically a very open diagnostic, whereby one of the items was, “Are you normally an impulsive sort of person?” which would include many people who do not have anything. The point I wanted to make was that the confidence interval—the estimate of error—could range from anywhere between 9.6% and 30.5%. That is what happens when you have relatively small numbers of black women.
Chair: Small datasets and large confidence.
Dr Norrie: Yes, exactly.
Andy Shallice: The major issue is that, within the education datasets, there is no sample. It is all children, in terms of their attainment and attendance from five to the end of statutory school age. For the Roma Support Group and for many Gypsy and Traveller organisations, overall within the UK we are relatively small communities, so in many of the samples no data at all is revealed about the housing or health experiences of Gypsy, Roma or Traveller communities. But within the school-age population, with pupils, there is an ability to look at differences between gender and ability and disability. We feel that there is much more to be gained from analysing the data, and particularly, as Shamit suggested, the importance of locality. That arises in the education datasets and not elsewhere in the information that we are talking about today.
Q8 Jess Phillips: So there is a distinct lack of the intersection of where you live at the moment in the data.
Andy Shallice: Apart from education.
Q9 Professor Saggar: Can I follow on from what Andy is saying? It is important to reinforce this point. If the purpose of the audit and the interest of the Committee is to get into real stories or pockets of deprivation, disconnectedness with the labour market and opportunity and so on, and it has something to do with ethnicity but maybe not entirely—these administrative datasets are really poor at looking at small groups, because if you are not plugged into the system of schools, hospitals, GPs and so on, you are not going to show up. When you have the one dataset that does that, which is what Andy is saying, you should really exploit it. It is the one thing that really sheds light on a group that we are concerned about but do not yet know a great deal about. That is my view.
Dr Weekes-Bernard: On the intersection, it is really important to try to get at some of those bits of data. In our submission, I talked a bit about apprenticeships, for example. We have some information on the website about apprenticeships, which is good, and it is really interesting, because it shows you how many ethnic groups are taking part in apprenticeships. We also know that there is a discrepancy between the number of individuals who are applying and the number who are getting them. It would be really interesting to be able to have that information by ethnicity, but also by gender. Broadly, when it comes to apprenticeships data, we know that young women are more likely to be found in the five sectors that are the lowest paid, and which have traditionally been seen as female-dominated sectors.
This is probably a policy point, but if we are trying to use apprenticeships to enable people to gain access to better work, the fact that we have got more young women engaging in particular types of apprenticeship, in which they are less likely to be paid well, is not a good policy development. We are really interested in finding out how that is broken down by ethnicity. We already know that there is a difficulty with young people from a minority ethnic background applying and then not getting apprenticeships. It would be really interesting to find out, if possible, the extent to which young women from a minority ethnic background are applying and not getting access to apprenticeships, or whether they are applying and finding themselves in sectors that are more likely to be poorly paid.
It is not difficult to get that sort of information. If we can use the site to drill down a bit deeper into some of those areas, that would be a useful thing to do. I also think—this is something I mentioned in the submission—that using something like this, which as far as I can see had some cross-departmental support, as a reason for trying to get better administrative data would be a really good exercise.
Chair: We are going to come on to administrative data.
Q10 Jess Phillips: Is it possible, from the data presented, to disentangle the effects of ethnicity on the outcome from other effects, such as socioeconomic status or geography—you used the example of a woman living in Manchester?
Professor Saggar: In some datasets and on some questions, it is the case. For example, we know a great deal about the labour market and about the factors that shape and condition your participation and attainment—your outcomes. To cut a long analytical story short, what you do with all that data is control for the things that normally condition how well you do in the labour market. Then when we get down to looking at some minority groups—not necessarily all of them—we have an unexplained gap. In the technical world, we call it the residual. In other words, it can’t be explained by the factors that normally explain how you do in the labour market. That is a very profound finding in social research. It is reflected in the work of the audit, but it could be amplified and made more of. You can take that away.
In the US, it is generally known as an ethnic penalty. In this country, perhaps we are a bit more polite about it—I don’t know. There is a penalty that is not explained by anything that is used to explain it. Since we are on it, some analysist and think tanks come along and say, “Well, it’s mostly social capital. It’s a lack of awareness, networks, connectivity with the labour market and proxies for the labour market, so it’s about the group.” Others say, “Well, it’s probably to do with discrimination, if you think about it.” We can’t really resolve that in the data, but Debbie made the point that if you put it on a one-stop shop website, the data to support people’s positions on that questions is at least there for people to look at. In my view, that must be a good thing. It shouldn’t be shared just by boffins and specialists.
Q11 Jess Phillips: I suppose the question that comes about, if we can control, is this. Where I live, there is Bangladeshi poverty, but there are similar levels of poverty in the white community. The race equality audit does not necessarily try to understand whether it is poverty that stops them progressing, rather than their ethnicity. At the moment, I am not sure the audit helps me disentangle that for the kids who grow up in one particular council estate in Birmingham where Bangladeshi and white kids live.
Professor Saggar: It is important to address that, but we are stuck with the fact that different datasets are able to do that more than others. The opposite of what I described is DLHE—the Destination of Leavers from Higher Education initiative data. It is really good at telling us who gets into university, where they go, what degree they get and which job they get, but there is no contextual information about parents and siblings.
Dr Norrie: I was surprised that no use of the census was made in this exercise, because it has perfect geographical insight. The largest datasets in the survey data will only really go down to local authority level, so those cannot really show you any differences in your local area. They are not really helpful in that regard.
Q12 Chair: Why would the census not be in there?
Dr Norrie: I guess because it is a bit dated now. That is the problem: it is brilliant, but it goes out of date very quickly.
Dr Weekes-Bernard: It does, but one thing it also does is it has—
Q13 Jess Phillips: What date was the last census?
Dr Weekes-Bernard: 2011. It has really great ethnic categories. It has all the categories you would want if you wanted to be able to look at any of these public policy areas, and it breaks down the categories in a really good way. I know that that is a problem that we have with our administrative data: we just don’t have consistency across ethnic categories. If we used the census as a starting point, it would be really useful in helping us to get better and more fine-grained analysis on various things.
Your point about your constituency and the difference between your white constituents who are in poverty and your Bangladeshi constituents who are in poverty is really important. There are places where you can look at income quintiles. For example, the website talks about how much each ethnic group spends on housing. That is a really good way of trying to find out people’s poverty risk, because those who have lower incomes spend much more of their disposable income on housing and fuel and a variety of other things.
While it is great that we have that data around certain minority ethnic groups spending more of their income on housing, if we had that broken down by income distribution—income distribution is included on the site elsewhere—we could combine that information to see how many individuals from a minority ethnic group in a particular income band are spending on housing. We would then be able to get at how many individuals are from poorer backgrounds and to say a bit more about those individuals. For example, we would be able to say more about what might be happening with some of your constituents.
Something I think is also really important, which Shamit talked about, is that ethnic penalty issue. There are different drivers for people who come from different ethnic groups, in relation to their poverty risk.
Chair: I am going to have to move us on to our next section, because I am very mindful of time—we started late, so I apologise for that. Tonia will ask the next set of questions. I know we have a time limit on this, so I ask people to make their answers brief.
Q14 Tonia Antoniazzi: How useful is the category of ethnicity when looking at inequalities—I know that is something you have touched on—and would data on country of origin, languages spoken or parental heritage be more useful in uncovering trends?
Professor Saggar: It would be good to have both, because then you will get the most accurate and nuanced picture. It is remarkable, by international standards—certainly by most west European standards—that we have in the UK this quite impressive ethnicity data on average on lots of different things. However, within that, we have a big focus on ethnicity but relatively little focus on country of origin or country or birth. That is funny, because in the old days it was the other way around—we used country of birth as a proxy for ethnicity, so we have inverted that.
Given that this country has experienced significant amounts of immigration in recent times—from all over the world, not just Europe—surely, when looking at, for example, the British Bangladeshi or Indian communities, you would absolutely want to know whether they were born in India or Bangladesh, or in east London. That makes a huge difference. Studies demonstrate that being UK-born is a big conditioner of, basically, being ahead by a long way.
I am entirely with you, but the solution to this is to try to find a way of having both, as it were, as a way of explaining disadvantage or exclusion. Ethnicity is really important, but being able to say where people are born, for example, in addition to their ethnicity, is even better. Some bits of the datasets we use allow us to do that, but many do not.
Andy Shallice: I do not want to be too sectional about this, but there has been a lot of praise for the 2011 census. Migrant Roma from central and eastern Europe do not feature on that census as a separate category. We would strongly support the importance of home language or first language spoken and nationality being included in ethnicity throughout the datasets. We are looking forward to conversations about that at the next census in 2021.
Dr Weekes-Bernard: I agree with Shamit. I think it is important to have both. There are some real differences between first and second generation migrants in terms of how they fare in society in relation to poverty and a variety of other things. I think it would be really useful to have ethnicity as well as country of origin, but English language is also incredibly important, because having English as your first language really increases your capacity to do well and to get work, to be frank. If you do not have English as a first language, it does discriminate against you when you are seeking work. It would be really important to have all of those things that you have listed. It would be great if we could have them all.
Q15 Tonia Antoniazzi: Is it possible or desirable for our public services to use the same categories when collecting their ethnicity data?
Witnesses: Yes.
Professor Saggar: Otherwise it would be a mishmash. It would be all over the place.
Chair: That is a pretty straightforward answer.
Professor Saggar: Time is against us, but I will be brief. This has come up a number of times. In the 1970s we used the “new Commonwealth” and “Pakistan” as proxy for being black or brown. Then Pakistan was thrown out of the Commonwealth in the 1971 civil war with Bangladesh and our census category took a bit of time to catch up with it, so this lack of consistency has been around for a long time, to do with wars in other countries. It is extraordinary how we do not tweak these things in real time.
Dr Norrie: One of the problems is that there is only so much space on the website. The more variables you bring in, the larger the story gets. That was one of the problems we had with the integration hub. We would have liked to have delved more into variations within ethnic groups, but it would have been—
Jess Phillips: People are of mixed heritage as well, aren’t they? I have friends who are 30 different things.
Q16 Tonia Antoniazzi: Where we aggregate into broad categories, such as white and non-white, is there a risk of creating the appearance of disparities that do not in fact exist?
Professor Saggar: Yes, massively.
Dr Norrie: The other thing is that with “white”, if you look at employment, for example, you might misread that, because a lot of eastern Europeans come in and work very hard, and that will make the difference larger. If you interpret that to be “white British and non-white British”, that is not the case, but it is migrant labour affecting parts. There is a difference between white British and other ethnic groups, too, but these things are just too broad and do not really do anything. We also want to know the differences between ethnic minority groups, too, and that is very important.
Dr Weekes-Bernard: Non-white really does upset lots of different ethnic groups, because they have very different experiences of public services. If we talk about education for example, Chinese and Indian students are doing exceptionally well at GCSE compared to black Caribbean students and white working class students. It is important to not have enormous white versus non-white definitions, because they do not really tell us anything. They just tell us that there is a difference between white and everyone else, but it does not tell us anything about what is happening to everyone else, because there are huge differences in that group.
Professor Saggar: If one thing is apparent in the broad sweep of research that has been done in this area in the last 10 or 20 years, it is that the term ethnic minority is very misleading, because there is so much variation from within—the examples mentioned earlier on. To simply say black or brown, ethnic minority or non-white is no longer fit for purpose, from a policy point of view. We have big policies designed in the past that have been loaded up on being an ethnic minority and they have then funded public money to groups—funnily enough—which are doing quite well, and, by definition, less money to those doing less well. That is clearly not tenable.
Q17 Eddie Hughes: The website has been designed to be used by those people who do not have your statistical experience. Has that attempt at simplicity hampered in any way the ability of expert researchers to use it? Richard, has the integration hub handled this differently?
Dr Norrie: We have handled it differently. Our website is not as well designed.
Eddie Hughes: Not as well? I thought you were going to say “better”.
Dr Norrie: We have some things over this one. For instance, we made some beautiful data maps, which you will be able to zoom in on and see everything about your constituency. I really like the design. It is very responsibly put together, with lots of important qualifiers, good descriptions of the data and very good descriptions of what the different measurements actually measure. I also like the different functionalities, such as the ability to download every dataset. It is a very well-designed website. It is a very good description. Chris Bryant made the point during the debate that you need to delve into it in more depth to understand what is going on, and that is going to take a bit more expertise, but overall it is a very good description of what is going on.
Andy Shallice: I have just one minor point. There is understandably a lot of emphasis on differences between ethnic groups in the datasets, and maybe the narrative accompanying the data does not concentrate sufficiently on changes that are taking place within particular communities. The pretty astounding example that we looked at was the seeming outperformance of Gypsy and Roma children on free school meals—they outperform their peers who are not in receipt of free school meals. That is an unusual finding, and we think that there is something that lies behind that, but obviously the dataset is about disparities between ethnicities rather than changes within particular ethnic communities.
Chair: That echoes something that Shamit said a bit earlier on.
Q18 Eddie Hughes: Do you want to comment, Shamit, with regard to the accessibility of the data?
Professor Saggar: Can I answer your question from a slightly different angle? I said at the beginning that this is the start of something rather than the end. As a thought experiment, it might be worthwhile thinking about the fact that it is a form of comparative information not dissimilar to the way in which we live our lives. We often consult comparative information—league tables—on how schools are doing, on gas and electricity prices and on the best broadband contracts. We are consumers. The idea behind this is that you collect this administrative data, you put it in one place and people should be able to use it. It is part and parcel of being a modern citizen—a modern consumer. If there are disparities in the distribution of outcomes between ethnic groups, as citizens we should be able to hold the Government, or indeed individual Departments and Secretaries of State, to account. That is the logic of what lies behind this—it was actually in the PM’s speech. You publish, or you comply, as it were. There is an idea that transparency is a good unto itself.
This is not dissimilar to the way we set up Ofsted a generation ago. There is not a single parent in the country who would like the information that Ofsted presents about the schools in their locality taken away from them. We have got used to that. I see this as part of that. I see it as part of the same sort of approach. Of course, we are dealing with race and ethnicity, which is a sensitive subject, but in its own day, which school you sent your kid to and which school you didn’t send your kid to was pretty sensitive stuff as well. If you put education on top of race, you see the sensitivity.
My view is that this is a great start but you should try to move in the direction of the way in which we present comparative information to citizens and consumers in general. We do that in this country to a considerable extent. There are some doubts about how much people use it. I understand that not everyone consults a financial website when purchasing motor insurance or life insurance or whatever. But we do have this in things like Ofsted and parents using parent power and transparency to find out who is doing well—which school locally is underperforming when it should be doing better and which school is overperforming when our expectation is that it should be doing less well. They can hold schools to account. I see this as part of that.
Q19 Eddie Hughes: If it is a first step towards that and the Government have deliberately chosen not to publish analysis or a narrative with the data—it is just info and it is up to you to delve into it—was that a mistake, or was that the right approach?
Professor Saggar: It is the right approach, because then you can look at change over time.
Q20 Eddie Hughes: It is the right approach to start off with, but it will develop? It is a work in progress?
Professor Saggar: Even with the indicators that we have in the first cut, with all the limitations we have been discussing, imagine you did a reasonable update—once a year or once every two or three years—on these things, you could quickly see who is moving ahead and who is falling behind on a range of indicators. You could then attribute that to the group, if you wanted, or to the Minister in charge of welfare or jobs or whatever. Anyone can take the credit for it. However, looking at improvements or change over time is what I would like to see come next, as well as more data.
Dr Weekes-Bernard: Just to disagree with Shamit slightly, I think it is important to have data with analysis, and possibly a bit of narrative. That is only because I know that, when it was initially announced that the site would be put together, there was some disquiet among a variety of organisations that felt that some of the information that was likely to come out of the audit would not be new and would sort of reflect, I suspect, some of the work that a variety of different think tanks and organisations have been conducting over years around gaps and disparities. I think there was more of a concern that people wanted to see what the Government would do next, and whether action would be taken on clear disparities that were likely to be revealed by the audit.
I think it is definitely important to have the audit and the data. However, I also think it is useful, if it is supposed to be something that ordinary individuals are able to access by themselves or in an organisation or however we expect them to consume it, to perhaps see somewhere that something is likely to be done or that something is likely to follow from what are actually quite stark disparities on the site.
Q21 Eddie Hughes: A kind of “so what?”
Dr Weekes-Bernard: Yes. As researchers, we often get a lot of that. I think it is a fair question from individuals, particularly if we want to allow people to be a bit more informed by data. To be able to go to the site and see something and say, “Oh my god, I didn’t know that”, or perhaps, “I am glad I know that now”, I think it is important to have something accompanying the data, so that people get a sense that we are not producing data just for the sake of producing data, and that there are plans to do something with it, or that there is some analysis that helps them not to feel as though the world is going to end, because the disparities are quite terrible in places.
Andy Shallice: I completely agree with Debbie. I have three other points. First, the importance of quantitative data shouldn’t rule out qualitative information that we draw from communities, neighbourhoods and population groups about the things that they want to talk about on the relationship between them and public services.
Secondly, organisations should be held to account. The difficulties of holding organisations to account as a result of the publication of this data are revealed when you start asking local authorities to explain, for example, the over-representation of Gypsy and Roma children in the exclusion data. It is not always easy to even have a dialogue.
Thirdly, and in that sense, the first response to the Race Disparity Audit from the Department for Education it is to be applauded. It has said it wants to concentrate on the issue of exclusion of children, which has obviously been one of the main talking points that has arisen from the data. I hope that that doesn’t start from a blank sheet but builds on comments made in this country for 20 or 30 years about the reason why some children are more likely to be excluded than others.
Dr Norrie: There is the problem of causation and correlation. The data are all merely descriptive and ambiguous, as I said in my submission. It is very difficult to build policy on that ambiguity. There is a risk that, the more you start adding commentary to the website, you start introducing, for example, certain political persuasions. I kind of like the dataset as it is.
Eddie Hughes: Just so I am clear, you are suggesting that whoever writes the narrative might be influenced and therefore interpret it—
Dr Norrie: It is possible. The intention behind our website was just to put the facts out there so as to better inform debate, because it is quite a contentious area, and it is a debate that needs to be grounded on fact.
Chair: We are going to move on to the next set of questions.
Q22 Jess Phillips: The Race Disparity Audit, as you have said, Shamit, relies heavily on administrative data that is a by-product of systems developed for operational purposes, such as people registering for health services. How useful is that data in trying to understand race disparities?
Professor Saggar: Some of it is potentially very useful, but the general problem with many administrative datasets is what I call the lack of contextual information. You capture a claimant of a welfare payment. You capture someone in Jobcentre Plus or their GP clinic, or a school leaver. It is very rich in terms of describing them, but it often lacks much about things around them, such as their parents, their households or their siblings, all of which are often quite important, and particularly things like education. That is your big trade-off. You have very high numbers in administrative datasets, so it is really powerful. You can disaggregate along the lines that we talked about. You can drill into it in quite a considerable number of ways. We can talk about the pluses, but the downsides are important.
As I said a moment ago, what about people not plugged into the administrative system? There could be all sorts of groups. We don’t even know who they are because they are not plugged in, so we can just surmise who they are. They effectively fall between the cracks. The sweet spot in all of this effectively is to have some big administrative datasets accompanied by some robust surveys that are as big as possible, such as the UK household longitudinal study, which is a big survey. It longitudinally looks at the same 40,000 households every year, and it really does track quite a lot of stuff. If you can combine the two things, you are going to begin to understand a great deal about the role of ethnicity in all these things, which is the purpose of this exercise.
Jess Phillips: Any other views? Andy?
Andy Shallice: It gets to the hub of some of the problems that we have with this. If Roma parents have had their children attending a primary school for a number of years, where they see the benefits of education and their children prospering, they are quite likely to define themselves in answer to a question about their ethnicity as “Roma and Slovak” or “Roma and Romanian”. If they are attending the jobcentre, it is quite unlikely that they will declare themselves as Roma. They are more likely to say that they are Romanian, Slovak or Polish. The degree of trust between an individual and the service that is collecting the data is critical.
Q23 Jess Phillips: Does anyone feel that it is unreliable in its collection?
Dr Norrie: Are you talking about one specific dataset?
Jess Phillips: Specifically administrative data, some of which is guessed.
Dr Norrie: I could not tell you about any guessing or anyone writing in something for someone else. One of the things I spotted in data on prosecutions and convictions was that in 2016, there were roughly 300,000 prosecutions, but in 60,000 of those, there was no disclosure of ethnicity. That is roughly 20%. We do not know what happened or who those people are.
Jess Phillips: So in 20%, ethnicity is missing?
Dr Norrie: Yes. The one thing I will say is that the conviction ratio is 81%, which is the same for non-white ethnic minorities. For whites, it was slightly higher, at 86%. It is possible that you will have a substantial group within that 20% who might sway the ratio.
Jess Phillips: Is there a potential negativity bias in the data? I am thinking particularly of the white Irish community. I would never write “white Irish” on a form, yet my family are Irish. If my family were Pakistani, I would write “Pakistani” on the form. Are we potentially missing positive stories in this data of how integration has worked well in the past?
Professor Saggar: The degree to which we know about how people self-identify with the categories that they are given is still quite a grey area.
Q24 Jess Phillips: I would say it is quite white.
Professor Saggar: Yes. Sometimes people in response to surveys on administrative data go to pick catch-all categories, and others go to some things that are very specific. There is very little that is known in general about the efficacy of different ways of categorising in terms of trying to count from this end and how people respond to that. We know for example that the second generation of south Asian immigrants, or the offspring of south Asian immigrants, are often less likely to automatically identify with the country of origin. After all, Pakistani is not an ethnic category—it is a national category—so why would they? I mean, they know their parents came from there. In the absence of all other things, people say, “Yes, I’m Pakistani.”
Q25 Jess Phillips: People say they are British-Pakistani, don’t they? I don’t say I’m British-Irish.
Professor Saggar: We must remember the overarching five main categories where we mean “ethnic minority” are made up of a series of kind of ethnic categories and national categories. It is a bit of a mishmash, so you are going to find problems of efficacy and reliability down the line at some stage.
Q26 Jess Phillips: Leading on from that, are all administrative data in the audit equally robust, or are there sets that we need to be more cautious of? Are there any that shouldn’t be relied on at all?
Dr Norrie: I don’t know.
Dr Weekes-Bernard: I’m not sure.
Professor Saggar: We could have a look at that—
Q27 Jess Phillips: That fills me with confidence that you are fairly confident in it!
Dr Norrie: Mostly, with administrative data, you have to work with what you’ve got. There is a question about how big a level of non-disclosure is a problem. You report your statistics with the caveat “of all those who declared their ethnicity”. That is really the best you can do. The Government could obviously try and get that disclosure rate up. That would be worth doing.
Andy Shallice: I have just a tiny example: primary schools, in terms of recording on the school census. In neighbouring primary schools, both with catchments of Roma children, one returns all Roma children and the neighbouring school returns all white European children. So schools with a similar catchment population, both Roma, have determined that some children are Roma and some children are white European. It is a tiny example, but one where the agency determines the ethnicity or the description more than the children or the families.
Jess Phillips: I have to say that when we were asked for our ethnicity at my kids’ school, there were loads of people who wouldn’t declare, for all sorts of ethical reasons—who refused to declare.
Professor Saggar: If I could make a very short point in response, extending Andy’s point, there is an opportunity to focus in this sort of exercise on those areas of public services where we think there are higher than normal levels of discretion in the hands of professionals. That might be teachers or doctors, or it might be Jobcentre Plus workers or whatever.
If I am in the business of making a discretionary decision about a welfare payment to you—as opposed to it being mandatory that I have to do A, B and C—that is an area where, as a general principle, we should be scrutinising a bit more, because that could be a place, a location, where some of the biases, intentional or unintentional, are most likely to cluster. It is sort of extending Andy’s point, if you see where I’m going with this. There is a general point about how much discretion we give professionals, particularly in public services. We must have a view on that, whatever it is. Within that, if we think there are opportunities, even unintentional, for people to be biased or indeed discriminate, that may be a place where you would take the audit next and concentrate disproportionately on those sectors of those fields and actively choose to put other things on the back burner. That is sort of what I would suggest is a sensible thing to do.
Q28 Kirstene Hair: The UK is seen as being one of the most advanced countries in terms of its data collection on race, yet stark disparities continue to exist regardless. What evidence is there that the presentation of data through the Race Disparity Audit will lead to a reduction in disparities?
Dr Weekes-Bernard: Do you mean simply by presenting the data?
Professor Saggar: Do you mean: will that lead to a reduction in disparities in itself?
Q29 Kirstene Hair: Yes.
Professor Saggar: The theory is that we hold Ministers and others to account because they are outliers in running the Prison Service, as compared with running the schools or something like that, and that the logic is that there is public pressure and political pressure, debates in the House of Commons and newspaper op-eds that say, “This is really bad, you have got to do something about it.” That is the theory: that it applies soft pressure—political pressure, if you like.
I doubt that that in itself it will lead to a reduction in disparities, but it will be a helpful start. If you had administrative data that coded most of these things tolerably well, where in the modern world is there an argument to say that we should not publish that or that we should bury it? That does not seem to me a very strong argument. In itself, it is a start, but to what extent pressure can be used as a tool to get to a genuine closing of those disparities—especially big outliers that are unexplained—is really in the hands of the Committee and the Government.
Dr Weekes-Bernard: Having the data is obviously important, because it shines a light on disparities, as Shamit says. It draws attention—hopefully policy attention—to them. I hope to see it being used to develop evidenced solutions and target solutions to the communities where we can see clear gaps in a variety of public services. At JRF, we spend a great deal of time commissioning research to find out where the individuals are who are experiencing a higher rate of poverty than others. We investigate all sorts of characteristics, and we use that information to make policy requests. We also talk to other stakeholders who bear some responsibility—not just Government and local authorities, but employers and a variety of others—and tell them what the evidence is showing. We should be able to use that evidence to develop evidence-based policy solutions. Just publishing it without a plan would obviously not achieve anything.
Dr Norrie: I would say no to your question, but the good news is that it shows that a lot of the gaps are narrowing, which is a descriptor. For instance, if you look at employment, the take-up in people going into work is mainly amongst ethnic minority people. I downloaded another dataset today—not the one we are talking about—and had a look at some of the stats. Between 2001 and 2017, the number of women in employment who are white grew by 14%, but amongst ethnic minority groups the proportion was much, much higher, and amongst Bangladeshi women it was as high as 564%. You can really see a massive growth there—a narrowing of the gap between the majority and minority.
Andy Shallice: I suppose I would like that question to be asked in 12 months’ time, rather than two months after the data has been published. It might be interesting to see how different Government Departments respond to your Committee when you ask whether it has caused them to consider changes in policy, emphasis or budgetary decision making.
Q30 Kirstene Hair: Finally, the EHRC has submitted that it is not necessary to have a Government audit of race to uncover race inequalities in the UK, because they are well known. Do you agree? If so, what is a better way of addressing those inequalities?
Professor Saggar: Broadly, I do not agree. Part of the argument is that they are already known. Well, they are known in some circles, but not in most, frankly. You have to be close to the research or part of a single-issue pressure group or equivalent to know about any of this, never mind all of it—never mind the nuances and the issues we have been talking about. If you look at it from that point of view, of course this is a better way to start.
I think the EHRC is perfectly well equipped to take forward some of this agenda—I do not think that it is mutually exclusive in that sense—but this is about more than just bureaucracy looking at the issue. It is also about people themselves holding their MPs, Parliament and system of government to account—not exclusively that, but there are lots of ways in which that can happen. It is very much the beginning of something, not the end.
I do not particularly think that these things are well known, if I may say so. I have been studying this myself for the best part of 20 years and I think that the tent has grown marginally, in terms of who is aware of these things.
Andy Shallice: I have a lot of sympathy for the EHRC assessment, except in the sense that this gives some groups that have been virtually invisible—in terms of either Government policy or the administrative collection of data—an opportunity of making a claim that more light needs to be shed on their fate and their relationship with public services. Gypsy, Traveller and Roma are certainly critical in that sense.
Dr Weekes-Bernard: I think it is important to have the site, as I have said already. Some material isn’t new, because we know that there are gaps in particular areas, such as the criminal justice system, although I know that we had a review by David Lammy very recently and that there has been a lot of work on looking at some of those disparities over time.
Shamit is right: there are single-issue organisations that have been looking at this work for a long time, so might know more about this information. However, there are lots of other people who have absolutely no idea, to be frank. They may think that they know something about those disparities, but actually having the evidence in front of them may either confirm what they know or surprise them.
I also think it is important to have the site because it is useful to us as researchers. We work on poverty and there are things in the website that we think are useful and that can help us, but we also think there are gaps—I have mentioned some of those and put them in my submission. There are other individuals who don’t work in these areas at all but who talk about these things and think they know, but actually don’t, so I think it is helpful for those individuals.
Dr Norrie: It is very good to have everything under one roof, because you can just explore and learn and not have to go looking or trawling through lengthy documents. It is very digestible.
Q31 Chair: Can I close by asking a very quick question about the quality of the information provided? With a one-word answer, how would you assess the quality of the data that is there? Good? Bad? Needs work?
Dr Norrie: Good.
Dr Weekes-Bernard: Needs work.
Professor Saggar: A-minus.
Andy Shallice: From a sectional point of view, I have to say it is of minimal relevance, but it at least gives us the opportunity of casting the light where little light has been cast to date.
Q32 Chair: So it gives an opportunity for real improvement, in terms of the GRT community?
Andy Shallice: Certainly.
Q33 Chair: That is wonderful. Thank you very much for your time. I am sorry that we started late. We had a lot of ground to cover, which you did extremely expertly and succinctly. I thank you for that, on behalf of the Committee, and I look forward to getting your responses when we eventually publish our inquiry report. Thank you very much.