 Hello, everyone, and welcome to this Eucadata Service webinar introducing data and resources related to food and food security. I'm Marguerite Ceraulo. I'm a Senior Communications and Impact Officer working for the Eucadata Service, and presenting with me today is Rebecca O'Connor, a Senior Research Officer at UCL. Right, so today I'm going to look at food as a topic and how you can find relevant sources of data on the Eucadata Service. I will also briefly look at some of the useful resources and documents that come with the data and show you how to access the data on our website with a live demonstration. Rebecca will then introduce you to her study of food, families and work using the National Diet and Nutrition Survey, which is an excellent source of data on dietary habits. And at the end, as I said, we'll have a question and answer session. Right, so for those who are unfamiliar with the Eucadata Service, we provide a single point of access to a wide range of social science data. If you've never used the service before and want to know more about it, you can find a range of video tutorials and guides on our website. Right, so food. So the food system cuts across many academic disciplines and covers a wide range of research topics, such as consumer culture and attitudes towards food, of course, a healthy diet and food safety, food security and better access to food, impact of extreme weather events on food systems, the need of a growing population for sustainable agriculture and medical challenges of diseases caused by either under or over nutrition. And the Eucadata Service does hold a variety of data sets that relate to these topics. So now I'm going through a couple of them. Just the key data sets that we hold, however, these are not all the data sets that we have relating to food. So that's just something to keep in mind. So we've got the 1970 British cohort study, which is a longitudinal study. And it has data about diet, including a four day diary. It has information about consumption of ready meals, convenience foods, takeaways, home cooked meals and drinks. And it has questions such as, you know, what type of drinks have you consumed in the last seven days, beer, cider or so on. Or, for example, the frequency of eating ready meals, is it, you know, once a week, twice a week and so on. The English longitudinal study of aging is also a good data set for food data. And it has data on expenditure on food and groceries, whether cut food consumption, you know, whether people have to cut food consumption due to financial difficulty. It has information about expenditure on eating or buying food outside the household and the food and drink consumed the previous day. Now, different type of data is held by the Farm and Bizarre, but the Farm Business Survey. And as you can see, it started in 1892, sorry, 1982, and it's continuing. And it's data about farms in England and Wales. And it has data about livestock crops, farm characteristics, farm finance, including cost receipts, assets, income, subsidies and so on. So that's a really good data set as well, depending on, of course, what topic you're researching. Right, we've also got the Food and Youth Survey and that's data collected by the Food Standards Agency. Started in 2010 and the survey I think comes out every two years. So data is collected every two years. And it has data about behavior and attitudes and knowledge towards food issues such as the food safety, healthy eating, shopping habits. And it has questions such as, you know, have you eaten fish or poultry in the last seven days or have you had takeaway food? And so you can find lots and lots of information about dietary habits. The Health Survey for England has information about diets, mainly fruit and vegetable consumption. I just had a look earlier and there's a lot of questions about, you know, how many times people eat fruit and veg during the week, if it's fresh vegetables, boiled vegetables and so on. And then, of course, we've got the National Diet and Nutrition Survey and I'm sure that Rebecca will tell you more about it later on. And that as well has a four-day food diary, eating habits, smoking and drinking, general health and even blood and urine samples. I have a little question for you. So according to the Food and Youth Survey 2014, how many people said they never eat raw fruit? So is it 11.6% of, of course, the people that responded to the Food and Youth Survey said it, 2.1%, 4.5%, 0.7%, let's see what you think. So most of you said 11.6% and then 30% of you said 4.5%. And I have to say those that did say 4.5% were right. As you can see, 4.5% of the respondents, which in this case were 3,452, overall, said they did never eat raw fruit, which I thought was quite a high number and, but yeah, so it's very interesting. So this is the type of data that you can find from the surveys. But we don't only have UK data and this is a question someone actually asked me just now, I can see. We also hold international data sets about food. And an example is the IMF International Financial Statistics. So this is different type of food. More than dietary habits is market prices of food, such as wheat, rice, fish, meat, cooking oil, et cetera. And then he has data about exports and imports of food of different countries. And it's the country aggregated data. The country level data and the World Bank World Development Indicators have data about food as well. And it's about food nutrition, food production, some data about agriculture and food deficit. But as I said, these are only a couple of the data sets that have information about food and food security. And if you've grabbed the slides on the handout, you'll be able to follow the link here that says discover all studies related to politics, which I sincerely apologize, I should have said food, clearly. But the link is right and I will take you to all of them. But I'll also show you in a second the food theme pages. And this is a graph about extra virgin olive oil, X tanker price in the UK, and it's in US dollars per metric ton. And this comes from the IMF International Financial Statistics. So this is, again, another type of data you can get from the UK data service. Right, so I'll go on to our website now and I'll just do a quick live demonstration of the food theme pages. So I'm going to get data on our website and then going to food. You can have access to the food theme pages. There's a table here with some, again, some key data sets, but more than the ones that I've talked about earlier that relate to food. So you can see a couple that I mentioned earlier, but this is a good start if you want to look at some of the data that we've got. And it will tell you the name of the study, the coverage and the topics included. And then we have different tabs. So we've got a Discover tab and this has all the links that you can click to discover all the data on our website and so on. And our website is being a little bit slow, so sorry about that. So you can click on a list of studies related to food and food security. And you can search for food because we have the source as well. And you can look at the variables on food and food security data. And then we've got a research tab and this just has a few examples of studies that have used our data that is related to food, of course. So for example, this is a study about the geography of women's diets in the UK. So you can see how researchers have used the data that we hold in research. And then we've got a last tab and it just with useful resources on food and external providers as well, of course, other websites that provide data on food. Going back to our homepage, of course, you can also use our Discover interface, which is our search functionality. And you can use that as well to search, to search for data. However, usually it is more useful to know what you're looking for beforehand because this is not a variable search, but it searches the titles of data sets. So that's why ThinkPages can be useful. Right, so going back to my slideshow. And now I'd like to talk about accessing and downloading data very briefly before I hand over to Rebecca. Access to a data can come with different forms of license and access conditions. So the different access arrangements reflect the risk of potential disclosure. All the data are anonymized, but if you had detailed information about, let's say, a person's job, which sector they're working, the area they live in, as well as their age and gender, this is a danger that in some cases you could identify who that person is. However, most of our data is under end-user license. So to get this, you just register with the UK Data Service and then you can download for free all these studies on their end-user license. If you're not registered with the UK Data Service, registration is quick and easy and takes about five minutes. And people from abroad can, because I know that there's people from the Netherlands, for example, attending today, you can register as well. It may be a little bit longer, as you have to request a username and login. From the UK Data Archive. But, I mean, of course, the data are freely accessible to you too. And we do not charge for our data unless it's for commercial use. So that's quite important too. Right, and now I'll hand over to Rebecca and she'll tell you more about her study. OK, thank you, Margarita, for your invitation to present today and thanks everyone for joining us. I'm going to talk for about 15 or 20 minutes about a recent project that used the National Diet and Nutrition Survey, the NDNS, for a mixed methods longitudinal study of food, families and work that was published as a book by Bloomsbury earlier this year. So I'm going to highlight some of the ways in which we use the survey as a sampling frame for a qualitative study, using the nutritional data to develop an index of diet quality and our use of the contextual or paradata to examine patterns of family meals. So the NDNS was established in 1992, but from 2008 it became a continuous cross-sectional survey and it's designed to collect detailed quantitative information through three or four day diet diaries on food consumption, nutrient intake and the nutritional status of the general population, aged one and a half years and over, living in private households in the UK. And the survey covers a representative sample of around 1,000 people per year. And in 2008 the Economic and Social Research Council collaborated with the Food Standards Agency to fund social scientific research which would build on what's already known about diet in the UK. So specifically the programme sought to generate innovative research to further explore and explain UK dietary decisions, the context in which they are made and the circumstances under which they change over time. So our bid, which was funded, was designed to address the issue of children's food in working families, reflecting a long history of policy relevant research at the Thomas Coren research unit where I'm based on children and young people and the work family interface. So the policy context for this included a concern with children's diets alongside rising maternal and dual parental employment that mean that the combined hours of working parents in the UK are among the highest in Europe. Some secondary analysis of the Millennium Cohort Study had recently found a link between maternal employment and children's overweight status and eating behaviours. But the diet data in the MCS are rather limited and so we wanted to see whether these associations were born out elsewhere to examine qualitatively patterns of food and eating and what facilitates and constrains them in families. And in the follow-on project to explore how children's and family's food and eating change over time. So these are the research questions that we posed. I'll let you read them yourselves. I've highlighted the first three because these are the questions we addressed using both quantitative and qualitative data sets. And I'm going to talk briefly today about the first and the third question but you can refer to the book and to publish papers from the study that address other questions as well. So the research design and strategy was as follows. The research was conducted in two phases, wave one 2009 to 11 and wave two 2011 to 14. We adopted a mixed methods longitudinal design. Secondary analysis was carried out on the NDNS and other surveys, including at wave one, the Health Survey for England and the Avon Longitudinal Study of Parents and Children and at wave two, Understanding Society and the Millennium Cohort Study. We used the NDNS as a sampling frame and drew a purpose of qualitative sample of 47 children aged two to 10 years and their employed parents from the NDNS and I'll talk a bit about that next. 36 families were followed up successfully two years later. We carried out semi-structured interviews with parents, usually mothers at both waves and at both waves also carried out a range of methods with children including drawing and in some cases photographs. And then the qualitative analysis was consolidated at the case and the thematic level whilst the quant and qual analyses were combined where appropriate and this kind of meshing of the data is something that I'll touch on a bit later. So this slide shows the sample we were aiming for and underneath notes the one that we achieved. The latter was slightly skewed towards the higher income range which does reflect the NDNS overall sample and I should note that we had to include a wider range range of children than we would have preferred and this was due to the relatively small sample size small size of the overall sample around 500 children a year. So it's also relevant to note that this was an ongoing survey at the time that we drew the sample and that had a number of implications. So in the first place there were understandable concerns about participants' research fatigue and our methods for contacting participants were had to be agreed with the data owners, the FSA and survey administrators, Natsyn. In addition the pool for the sample had to be drawn under secure conditions at Natsyn's data enclave. And importantly, although we proposed a sample on the basis of known diets, children with healthier and less healthy diets, most of the food nutritional data were not going to be available, analyzed and available in time for us to draw our sample. So what did we do? We worked with nutritionists, Carol Devine in the US and Alison Lennox here in the UK to design a diet quality index for nutritional scoring system to identify children with healthier and less healthy diets. The nutritional score was used both to sample and as an outcome variable for the secondary analysis. And we used this, the data that we used was from the dietary feedback that survey participants were given within three months of taking part in the survey. And it provided this information here about individual intakes of key nutrients. Here's an example of the scoring system that we did. Much more detail about, and the methodology for doing this is available in this paper that's available as an open access paper online. So our first research question concerned how parental employment influences the diets of children. And the nutritional score that we developed was used not only to sample children, but also as an outcome variable for the secondary analysis, including this question. The table shows the regression coefficients showing associations between direct measures of the fruit and vegetable consumption as well as the nutritional score. Maternal employment and other socioeconomic characteristics. And I'll just briefly touch on what we found. We found that 65% of mothers with children in our age group were in paid employment in 2009-10. That maternal employment was not related to the nutritional score, or to children's consumption of portions of fruit and vegetables. And that children of mothers who were not employed had very similar average nutritional scores to children of mothers who were. So our findings contrast with those of the Millennium Cohort study. However, that analysis was confined to families where mothers worked full time. If we had included hours of work, we might have found a similar relationship to that reported in the MCS. Unfortunately, hours of work for mothers weren't collected in the NDNS at the time. However, on the other hand, our finding of no relationship could be true. One advantage of using the NDNS compared to the Millennium Cohort study is that the very detailed diet data includes food eaten outside of the home. So it is possible that the scores better reflect children's overall diets that might include healthy foods provided in schools and childcare. And finally, an important impact of our study is that mothers working hours and education are now included and have been added to the annual survey from 2015. I'm going to move quickly on to talk briefly about our work on families and meals that we examined in both the qualitative and the quantitative data sets, albeit our definitions of family meals and the ways that we framed our questions were different, reflecting the different types of data. So we were interested in family meals for three main reasons. Firstly, they have normative status, and there is evidence that they are linked to a number of positive outcomes for children. Second, some research has found that mothers who work longer hours eat fewer meals with their children. And third, much of this research has been based on self-reported behavior in response to direct question and likely suffers from social desirability bias. So in the qualitative research, we asked parents and children about their last working day, including who ate with whom at which occasions, and then we asked about typicality. We grouped families according to whether they usually, sometimes or rarely ate together and examined the conditions under which this did or didn't happen. And the paper describes this in more detail. A wave two, we looked at how patterns had changed or not. In the quantitative research, we asked the question posed on the slide, and we used the survey paradata or contextual data that had become newly available in the archive. So this slide shows an example from a food diary. The contextual data that had been archived related to the information given about whom eats with a child, where the child eats at a table with a TV on, and what the child eats as well. And originally, these were intended as prompts for the diary to improve accuracy in recording the food's consumes, but then somebody had the very good idea that these were also useful data in their own right. I won't go into the intricacies, but it was quite complicated to define both a family and to define a meal. We were pragmatic. We adopted a food and time-based definition of a meal. So we took the evening as the time slot, and we summarized food into core and non-core types, and meals were defined as including at least one core food. And we compared those eating with the parents or carers versus those eating with others. Oh, I think I've gone one ahead. There we are. So this is what we found. Importantly, children who had more frequent family meals on average had higher nutrition scores. Child age was significantly related to family meals so that younger children were more likely to eat to family meals than older ones. And we didn't find a relationship between maternal employment and frequency of family meals, but again, if we'd had hours of employment, we might have done. That said, our qualitative research suggested that it wasn't hours of work that was important in explaining whether families ate together, but rather the issue of timing, how the different schedules of family members intersected. The qualitative research did confirm the quantitative finding that children ate few family meals with their parents as they get older, fewer, sorry, and elaborated some of the reasons, namely that fathers were working longer hours as they were promoted and so forth, and children were taking part in more extracurricular activities. I'm gonna skip over this slide quite quickly for reasons of time, but because this was an exploratory analysis, we also looked at family meals in these two other data sets and I put the questions. So this was responses to direct questions, but it shows that our analysis of understanding society and millennium cohort study did find a relationship between longer hours of maternal employment and fewer meals eaten as a family. Okay, so to finish up, here are a few reflections on our use of the NDNS and on data linkage, and the first in the chapter noted at the bottom consider these in much greater detail. So the opportunities included that the NDNS provided a basis for deriving a qualitative sample on the basis of known diets. We had the opportunity to develop a unique index of diet quality, a nutritional score that was useful for sampling and as an outcome measure, and the paradata or contextual data was a useful alternative or supplement to survey response and to direct questions, and it enabled us to problematize and unpick and unpack the idea of a family meal. One of the key challenges for us was the missing sociodemographic data about mothers, but this is now included. The time lag as well for the process dietary data could be particularly problematic when examining children's diets qualitatively since their diets may change quite rapidly. Six months between completing the diary and being interviewed by our team could be quite a long time. And while these data are absolutely the gold standard, there's the question of validity at the individual level. So one parent we interviewed said that they had completed their child's diary whilst they were visiting family in Scotland, and so there were an awful lot of chips compared to when they were at home. Finally, a couple of words about mixing methods and meshing data. To some, mixed methods research is seen as a belt and braces approach. We would argue very strongly that it isn't that. In our study, the quantitative and qualitative constituent parts of the study addressed differently framed questions. And in practice, whilst analyses sometimes corroborated each other, for example, children's less regular family meals as they got older, in other instances they complimented one another. So our analysis that I haven't talked about of who does the food working, working families was one of these cases, while in others they were dissonant. So here are some links that you might find useful. The first one is to the NDNS. The second, I think it's worth looking for our qualitative data that we anonymized and archived from the link study on the UK Data Archive. I've also put a link to the Food Standards Agency Food and News Survey that Marguerita mentioned in her introduction. It includes some really interesting data, including of the new wave for a measure of household food insecurity. And I've put a link also to a recent paper that I was involved in that did some secondary analysis of the Food and News Survey to look at links between the affordability of food and domestic food safety practices. And if you're interested in food and security and affordability, you might want to take a look at the website for the current study that I am leading that is of families and food in hard times in three European countries. It's also a mixed method study, this time funded by the European Research Council. Finally, there's a link to the very reasonably priced paperback book of food, families, and work. Thanks very much for listening.