 I think people are still joining us, but we'll start quite slowly, so hopefully they will catch up as they come in. I'm Nigel Deenarona, I'm part of the team presenting, I'll introduce Lisa shortly. So we're talking about a new resource that we've made available to you through the event page, the evens teaching data set. We'll talk more about that afterwards and it will be available through the UK Data Service shortly. So what we're going to do is, first of all, have a bit of a welcome and introduction. I'm going to ask you a few questions about yourselves. Lisa is then going to talk about the even survey. So this is a cut-down version of a previous talk, so there is a link there as well if you want more detail about the even survey itself. I'll be doing a session on teaching the evens data and thinking about some of the ideas of what's there and at the end we'll kind of give you the opportunity to come back, but we're going to break in the middle, in the bottom of the bottom you'll see a Q&A thing. So if you have a question as we're going through, put it in and we will pause at the end of the section about the even survey so we can clarify any questions about that now. So on to the interaction. This is over to Lisa. I'll let her introduce herself first and then we'll start moving through her presentation. And there's a link here, so the slides will be available on the event page. So you can go and look at the fuller presentation on the Equality for National Survey. Thank you Nigel and good morning to you all. It's fantastic to be here with you and to see such diversity and expertise of disciplines and places of employment in the group today. So a very warm welcome to you all. I'm going to take a bit of time now as Nigel said to introduce the evidence for equality national survey and let's start by thinking what's new about evens? Why should we be bothered about this new data set that the UK data service is supporting us to promote and support the use of four key points? So evens is the largest and most comprehensive survey of ethnic and religious minorities in the UK. We can say ever and certainly in contemporary data. So we have 14,000 participants in the survey, around 10,000 of those identify as an ethnic minority. So a really large sample size compared to other surveys in the UK. So we have here in this data set novel data in terms of the topic, which are common on to robust data in terms of the method to present us with a really unique and rich resource for understanding the experiences of minoritized ethnic groups. And one thing to note with evens and that you can find out a lot more about in the other resources from UK data service and the evens team is that we use a non probability survey method for evens. I'm not going to go into very much of the detail of that today, but happy to answer questions. And there are the other resources to look at if you're interested in that. So we've got this new resource, but why did we want to collect these new data? Why did we want to know about these experiences? And it comes really from my work as a member of the Centre on the Dynamics of Ethnicity, working with a fantastic team, looking at ethnic inequalities over the last 10 years. And together with others who have been working on racial justice and ethnic inequalities really became very conscious of the persistence of ethnic inequalities in the UK across social realms and not really much change in that over the course of this work. And this was really enhanced during the pandemic. It became very evident quite early on during that crisis that ethnic minorities were having very distinct experiences from the general population and often worse experiences of that pandemic. So we were motivated to collect data to help us understand those distinct experiences and the patterns and processes driving ethnic inequalities. So I led the Evens project with a large team, including James Nassau and Laia Bucharest, who have argued in other work that the mortality in areas of high ethnic minority populations were a consequence of social and economic inequalities driven by entrenched structural and institutional racism and racial discrimination. So that's where we as a team within code and within the Evens project we're coming from, we're concerned about inequalities, we're concerned about racism. But why a new survey? That's the context, but we were also concerned that the data that existed prior to Evens was deficient for what we wanted to get at in a number of ways. So firstly, social surveys in the UK, wonderfully rich as they are and I use many of them a lot, do tend to represent quite a limited number of ethnic groups and very broad ethnic groups. So we might be able to say something about the Black population, we might be able to say something about the Asian population, but we as scholars and teachers really wanted a bit more nuance and detail in how we were thinking about people's identity. Secondly, existing surveys because of the way that they are designed and they sample tend to over-represent ethnic minorities that live in areas of residential clustering, particularly in the southeast, particularly in London. So really those surveys can't effectively tell us about what's it like to be an ethnic minority in the north east of England or in rural Lincolnshire or in the Scottish borders. So we hope to address that deficiency using Evens. Thirdly, because this was a survey that was specifically motivated by understanding experiences of minoritised groups, the topics and the questions are really concerned with that. So we have a whole bank of questions in this 30-minute survey that are focused on these issues of inequalities and minoritised experiences. Of course, we can tell something about this from a census and administrative data which have fabulous population covers and for human geographers like me, if you're interested in place, the census is such a wonderful resource, but they have limited topic coverage. So we can't know about attitudes, about details of finances, about political positionings, about experiences of racism with the census or indeed with administrative data currently. The final reason why we thought there was a gap in the market, so to speak, for Evens was methodological. So working with leading social statisticians at the University of Manchester, we really pioneered this non-probability survey approach as a way to try to improve the representation of small and minority populations in survey data. So that's why we produced Evens as a team within code, but how did we do it? What did we do to get these 14,000 people into the survey? So we used a non-probability approach, as I've said, with a 30-minute questionnaire which we developed in collaboration with our partners, and I'm going to come onto the partnerships later because that was really important part of the approach that we took. This was available online and on the telephone in 14 languages that are listed there, so we tried to be as inclusive as we could with the resources that we had available to us. And it was an open survey, so we didn't approach people and say and invite them to take part in our survey. This non-probability method just says, if you feel that you're an ethnic minority, commentate part in this survey and tell us about your experiences. So it's an open web link, open invitation survey recruitment method. So there's no inclusion or exclusion based on identification based on neighborhood. There was a requirement for residency in England, Wales and Scotland to be eligible, and we gave all participants a voucher of £10 upon completion of the survey. The data were collected during the middle of the COVID-19 pandemic February to October 2021, and we have some original questions in the full questionnaire, plus some that were borrowed from existing surveys such as a health survey for England, ELSA, Understand Society and others. We worked with Ipsos to administer the survey and the whole project received ethical approval, including a number of amendments from the University of Manchester Ethics Panel. So we did set some target quotas by ethnicity, age, sex and region to set to maximise our chance of having representativeness of the sample, and we had an initial screening questionnaire for eligibility in terms of GB residents. An important part of this methodology, so we have our targets, target quotas if you like, we say anybody commentate part, but of course we have limited resources to pay this incentive to people so we can't be open-ended, we can't recruit 50,000 people much as we might like to, and so we have to monitor when we reach our target for a certain group and then sort of say to anyone else who comes in, I'm sorry that this is full, so we were monitoring on a daily basis and on that we were adapting our recruitment approaches. So say we wanted to know more about the experiences of older Bangladeshi people in the northwest of England, we could target our recruitment efforts if we weren't meeting our quotas because we were doing this daily monitoring of the responses and we used some quite advanced statistical indicators to help us with that targeted recruitment. So key to this was working with partners and we had a number of fantastic partners that are listed here, aimed really representing a range of ethnic groups and regions of the UK, so ethnic minorities and youth support team in Wales, friends, families and travellers, Stuart Hall Foundation, Bemis in Scotland, Institute for Jewish Policy Research, NHS Race and Health Observatory, Business in the Community, Muslim Council of Britain, Running Need Trust, the Ubali Initiative, Race Equality Foundation, Operation Black Vote and the Migrants Rights Network. So all of these organisations worked with us to promote the survey and recruit to it. And here are some examples of the kinds of advertising and events that we did, the kind of imagery, the kind of branding that we had. So it's quite different from a typical survey approach where people are sampled from a sampling frame and invited to take part. And you can see at the bottom there are given some examples of particularly targeted advertising to particular groups, so friends, families and travellers there on the left. We really did work hard to try and get Gypsy, Roma and Traveller participants in the survey. And the bottom right there you see a bespoke advert for Jewish participants. So with the partnership with friends, families and travellers, I want to dwell on this a little bit because we took what I think is quite an interesting approach to this. And so through our monitoring we saw not unexpectedly, as this is common in research and including survey research, that we weren't getting the response we hoped for from Roma or Gypsy, Traveller populations. And we did some focus groups in collaboration with friends, families and travellers and found that really there were a lot of concerns in the communities around trust, around confidentiality and with digital literacy. So in partnership with FFT, we took a different approach to the recruitment of these participants and we trained community interviewers who then went into communities with digital devices and completed face-to-face the interviews for evens with people in those communities. And that took part just in a portion of the overall fieldwork time, so in the summer of 2021 and through that we got 309 participants through that community interview method. So for those of you interested in survey methods, I think this is something interesting and for those of you using the data as well, it's just worth being aware in teaching as well that the sampling approach for these groups is slightly different from the overall sample. So let's come to the sample overall. How did we do in terms of recruiting people across the groups? Well, we met on the whole the target populations, which were linked to the distribution of the population, I think groups in the population overall. In some groups we exceeded our targets, so for example in the Indian group and for other groups, so the Roma, we didn't quite meet our targets but still we were quite pleased with how we did. And you can see that there's also a general population sample here, which includes white British 4500 here and that's to enable comparisons across the survey sample. These are the unweighted survey sample sizes. I'll say a little bit about weights in a moment. And we can also see the sample broken down into religion, into region and into age group. So for those of you who are based in particular parts of the country, it is possible to look in more detail at particular areas. We got pretty good sample here in Wales, 900 pretty good sample in Scotland as well, just over 1100 and across other regions of England. On age, the main thing to be aware of in terms of the deficiencies of the sample is the relatively low numbers in the older ages. So with analyses we are recommending grouping the older groups into 55 plus because of the lower numbers and the distribution of those across the groups in those older ages. So something to work on if we do the survey again, but otherwise the distribution of the sample across these key target variables is pretty good. I want to say just a little bit about weights, not to get too technical but what the weights do is they enable users of the data to use those data as if they are representative of the general population of Britain. So when you're using survey data, you need to apply these weights, you need to tell whatever programme you're using to use these weights, which will just run in the background for you. So once you've told it to do it, you don't have to worry about it too much. So we use quite a sophisticated approach to weighting in evens. But the main thing to know for today really is that this accounts for coverage errors. So where we've missed people of a certain region or certain religion or certain age group, we can adjust that to what it should look like in relation to what we know about the population. But we also adjust the selection bias by which I mean the likelihood of people taking part in a survey because of aspects of their life, their characteristics. And that we draw on survey research more generally and use some quite sophisticated matching techniques to link the evens data to other data sets to account for that selection bias. So we're fairly confident that we have pretty good weights in there so that you can use evens to make claims about the population of Britain overall. So it is vital to use weights when using the evens data, including for teaching. And you'll see that Nigel has put some comments about that in the documentation for the teaching data set. And so in the teaching data set, if you just move on a slide, Nigel, please, we have two weight variables, one called BK weight or book weight. And these are the weights used in the evens book, which has been published last year. And we include that because as Nigel will come on to show you, you may very well want to replicate some of the analyses in that book in your teaching or in your work or just as an exercise to check your use of the data. So we include that weight that we used in the book. However, we have also provided updated weights since we published the book and they're included in the main data set and also in this teaching data set with the variable name of weight. So for all other analyses, apart from comparison with the book analyses, we recommend that you use the weight variable. The reason why we updated the weights was because we weren't happy with how we were adjusting for experience of racial discrimination and the adjustments for selection bias. So in sum, what we found was that using the original weights, we are underestimating experiences of racism. So what you'll see in the book is which is available as a free ebook, which is there's a link to that at the end of these slides. The figures there on experiences of racism actually underestimate what we think is the case. So the full evens data set has a lot of topics in it. And the ones that are included in the teaching data set are highlighted here in purple. So the teaching data set is a selection of the variables from the main data set. And it draws across many of the topics, but focusing on a number of themes that Nigel will come on to talk about. But in the main data set, we have quite a lot of detail on socioeconomic circumstance and financial circumstance and how that changed during the pandemic, which those of you from economics background may be interested in. We have quite a lot on identity. So perhaps sociologists among you will be interested in how people articulate their identity, what they mean by ethnic identity, how it relates to other aspects of identity and belonging. We have housing and demographic information in the survey and quite a lot on health and well-being. Nigel's going to illustrate that a little bit, both mental health and physical health, both in terms of diagnoses and in terms of self-reported and self-perceived health. We have questions about Black Lives Matters, about activism, about support for that, questions on social cohesion and belonging, including belonging to local area, attitudes towards the police, perhaps a criminologist among you will find that interesting, and COVID-19 compliance as well on a number of measures and a number of compliance initiatives. We have trust in government and we asked about that at national, UK and regional levels, regional, mayoral levels. And then we have quite a big module in the data set on racism and discrimination and that is included in the teaching data set. Let me say a little bit about that racism and discrimination module. So we have data variables on people being insulted for reasons to do with ethnicity, race, colour or religion, damage to property, physical attack, being treated unfairly in a number of social domains, racism from neighbours and then we have worry about being harassed which I think is a really interesting variable actually and some interesting results on that. Change in treatment experiences of racism during the pandemic and then also an interesting question shown here on the right hand side, how did people respond to experiences of discrimination or unfair treatment? Another feature of the data worth noting that's in the main data set that we won't linger on too much today is that we asked people when they had these experiences of racism, was it very recently or was it long ago? So in that sense we can sort of begin to construct a life course history of experiences of racism even though this is a cross-sectional data set, a data set that's just recording people's experiences at one point in time. So there's loads you can do with the data, both in teaching and in research and this is just a few examples of things that the evens and code team are working on at the moment, the life course experiences of racism across ethnic groups, connections between racism experiences and loneliness, the protective effect of religion for loneliness during the pandemic, social connectedness and loneliness, prevalence of mental disorders during the pandemic by ethnic groups, local belonging and ethnicity, articulations of ethnic identity and how we can think about official ethnic group categories, political trust across ethnic groups and whether it's and how this is related to compliance with COVID-19 measures and methodological papers on producing this non-probability survey. And you can find out a bit about what that I've already referred to which is available free as an ebook, you can also buy it as well if you like, if you want the lovely hard copy but you can get it instantly from the Policy Press website and also find out more about it from the evens web pages on the the code website and this has lots of thematic chapters covering lots of the topics that I've just mentioned. So that was a relatively brief coverage of what evens is about and how we came to do it and something about the survey overall and I think we're going to pause here to just check some of the questions which. Okay I think we're going to stop the questions there but keep them coming and we'll pick up again at the end of this and thanks very much for that interesting session. So I'm going to move on to talking about the teaching data set now. So just to say first of all this will be available, it will include the teaching data set. So this has around 67 variables I think around that. It's open access so if you're used to using UK data service data you will know that a lot of our surveys are safeguarded but we do produce open access versions to support teaching and to support other uses. There's a data dictionary which gives you a guide to the variables and categories. I didn't include frequencies in there partly because of the issue with weights so I didn't try to provide frequencies but those are fairly easy to generate and there's a user guide which talks about how to use evens in teaching which a lot of this material that's coming up comes from. So I think the benefits are firstly it's a topic of substantive interest in many disciplines. So I've taught in both sociology and human geography and I would have bitten my hand off of having a data set like this to teach with because students were interested, it linked into other courses, they reduce it, use it, doing it connected with substantive topics so there was a possibility for bringing this into a broader discussion and it offers this comprehensive range of information about ethnic minorities and the key areas we're kind of offering are racism, identity and national belonging, the impact of COVID on health, employment, education and well-being and political attitudes and trust and I think for methodological teaching it supports descriptive analysis, statistical models and statistical models including logistic and linear regression which are kind of highlighted in the user guide to some extent and I think the kind of learning outcomes from evens are both substantive and methodological so all of a sudden fantasize engages with recent data about the real world so this is a thing that is in our memories it's not historic and it's probably quite close to quite a lot of people. It links more naturally with other modules that students are engaged with and in terms of the requirements from QAA it addresses curriculum inclusion of equality and diversity, prosperity, themes and encourages a critical approach to the use of evidence. Methodologically I think it's useful for exercises across the kind of general introductory statistical training in social sciences. It encourages better interpretation of data and because of the way that this data was collected it stimulates a focus on thinking about data collection and processing and my experience I suppose with lots of students is they want to go and collect their own data and there are some nice pointers here in thinking about how to do that rather than asking all the people who you talked to on Saturday nights but thinking more purposefully about what you're collecting what it's going to tell you. So the user guide has some re-information on the three so it's got three summary information about the survey links to materials about the general survey as well as the teaching survey. Three sets of teaching guides on the three topics so race and identity belonging, health well-being and the impact of COVID-19 and each section then looks at research topics and potential research questions, areas to analyse and shows some sample outputs from the evens book so a useful way of starting teaching is to replicate what people have produced. It suggests the exploration of some associations between variables and where appropriate gives advice on building regression models and also we use two mental health scales one on depression the eight item depression scale which I'm going to go over in a bit and the seven item generalized anxiety and depression scale. So there are some issues before we move into it in the way that the data set has been put together so first of all where missing values are there in the main evens data set they've been set as a value category not applicable so that they can be incorporated in models so there's a number of routing areas within the survey that means there's quite a lot of missing values so if you for example say you've never experienced racism then you don't get to answer all those questions so that group of people who didn't experience racism won't necessarily appear when you go into the subcategories of racism so what you'll get in the teaching this set is not applicable. To reduce the number of variables I've reduced the specificity of country of birth to those that have 30 or more respondents and grouped others together into broader geographical areas there's still around 50 countries I think. Similarly with the party people intend to vote for there was quite a detailed breakdown in the main data set that again has been reduced. A couple of areas where you might expect to see you would see in most surveys in terms of occupational social class we didn't capture enough information on occupation to enable us to apply the methodology to develop occupational social class but there are a couple of potential substitutes what is highest level of qualification which is pretty detailed and income but that is household income so it would need to be adapted based on household size. The tenure is just rented or the categories of ownership and I think living rent free but it doesn't break down private and social renting and because relationships within the households aren't captured it's an individual survey there's limited information about household composition so you can tell the number of adults and children broadly speaking in the house or whether there's adults one two or more than two and whether there's dependent children. So let's have a look at some of the ideas for teaching so first of all there's a set of demographic data here so at the core of this I suppose is ethnicity, religion, region, age, sex and then you've got a number of other demographic characteristics so country of birth, year of arrival in the UK, household structure, number of children, number of generations which is quite interesting household income, length of residence and highest level of qualification and you could for example look here at how does the highest level of qualification vary between ethnic groups and I've cut out the Asian group here what you can see is as an Asian group there is a kind of average across the levels but there are quite significant differences between groups so when you look at postgraduate degrees that's quite high for the Indian group the average is 30 but when you move across to Pakistani it goes down significantly in Bangladesh before and you can see those kinds of breakdowns so partly when you're in terms of presenting this here it's not so easy to present the 21 ethnic groups and this breakdown so chunking this does one thing enables to give an example but it also shows the effect of grouping characteristic so the higher level ethnic group here has quite a different outcome that doesn't really reflect the diversity between the group. On racism identity and belonging we've got experiences of racism whether we feel part of the country, feeling part of Britain, England, Scotland or Wales, feeling British, English, Scottish, Welsh or people's own ethnic identity. Questions around local belonging and whether that's changed and that's the kind of defined in the classic way of within 15 minutes I think of the area you're living how you responded to racism and worry about racism which Lisa's already talked about and the importance of the ethnic background and religious background to who you think you are and in the example how do experiences of racism in the local neighborhood vary between ethnic groups and that's what I'm going to show you and then you might explore how that varies by sex age, migration history and region so this is picking up one of those racism variables and looking at differences and they are quite stark really you know these are ordered down but high levels of racism in the neighborhood so those from any other Black background, Gypsy, Traveller, Roma, Mixed White and Black African etc and you might you know want to think about why that is and maybe with some geography that would help to break it down or whether it's about age or sex so there's a number of interesting areas to explore just from that initial look at the breakdown. This is a really complex set of data I suppose from putting it together and trying to group it into things but it's about health, well-being and the impact of COVID-19. On the left hand side we've got information about subjective health, limiting long-term conditions and access to care. We've got things around COVID I think one of the kind of key findings in the book was about experiences of bereavement, differential experiences of bereavement but also around attitudes to vaccination, having had COVID, having had tests and whether people use the app. Economic stroke, employment impact, worrying about money and those kind of home effects like how many people or what percentage of time are people spending working for home, what was the impact on caring for their children and schooling at home, what kind of lifestyle adjustments do people make and those have largely framed in terms of financial impacts I think but the implication is because we have less money or because we could do less these are the kind of changes we make and also access to the outdoors which was a key factor. So do people have private gardens or private spaces they can go to so they have shared spaces in the house they live in or are they reliant on being outdoors and if they are are they near to green spaces parks etc. Then the last set of measures around mental health I'm going to go into a little bit more. So looking at this is an example from the book about experiences of COVID related bereavement and what we can see is kind of quite big differences between different groups here. So the South Asian, Indian, Pakistani, Bangladeshi have higher and Chinese have higher levels or are more likely to have experienced bereavements than the white group which is a comparator similarly for white Irish, Gypsy, Traveller, Jewish so looking down that list it does look like there was quite a difference between most ethnic groups and the white British less so on bereavement of any kind but still significant patterns and this is one of those examples where you could replicate this within class using the data and the weight from the book. So the last section is on political attitudes and behaviour so it's got interest in politics intention to vote. Trust in the government handling of COVID-19 and as Nisa said that's trusting government at different levels so people ask whether they trusted the UK government whether they trusted their own government whether they trusted their local mayor. Attitudes to Black Lives Matter and attitudes to and experiences of policing and whether people have taken action in terms of Black Lives Matter and if we look here you can see there's quite big regional differences between trust in the UK government and ethnic minority groups and white British group. I suppose maybe a bit surprisingly I think my minority groups tend to trust the government more and reflect my attitudes but I suppose I'm I did complete the survey but I'm probably not figuring enough in the waiting to move that along and then I want to talk about these measures of depression and anxiety. So there are two sets of scales so this scale is the depression scale it's a standard scale it asks you these sets of questions and I've used SPSS when I was looking at this I'm aware that factors operate quite differently in different packages so this is something you need to experiment with before but potentially you could generate one factor for all questions or you could generate a second factor or if you're working with students you could say let's put these together out you know how are we going to put these together and work out how to do that with the students a kind of positive thing I've done I think in in the past but overall it gives you a measure of depression which is relatively well recognised and is used in the evens book. The next one is slightly more complicated so over the last last two weeks how often you've been bothered by these different problems my screen sharey stops me seeing the answers at the bottom but in essence you derive a score based on all of those answers so you add up the columns and arrive at a single score which gives you generalized anxiety measure. Again a useful thing if you're thinking about how are people feeling potentially potentially very useful for new regression so that's me if there are any questions now I'll we can pick those up around the teaching data set or more generally. So as I said just to reinforce if you do have things you would like to share with us it will be great and to hear about the experience of this because if the bid for code mark whatever is successful there will be a new survey or it's a part of the components of that survey so to hear how you're using it maybe the kind of edges you bought up against that would be useful to include in the new version would be great material for Lisa to take forward into the into the centre if the bid is successful. Okay well we'll hang around for a while if people still have questions or want to talk but thank you very much for attending hopefully you've got something useful from the session. Thanks very much everyone enjoy the data set.