Podcast Episode: Mixed methods research and design: their development and use
Category: Social work (general)
What follows is a transcription of the audio recording. Due to differences between spoken and written English, the transcript may contain quirks of grammar and syntax.
JB - Julia Brannen
Introduction: Julia Brannen, Professor of the Sociology of the Family at the University of London, Institute of Education, is a leading international scholar in the fields of family life, work life and intergenerational issues. In this lecture at the Glasgow School of Social Work on 7th June 2012, Professor Brannen explores what we mean by mixed methods - that is mixed qualitative and quantitative research methodologies. She outlines the history of this approach to research and describes some recent examples of its use, in particular in studies of food practice.
JB I am going to start to talk more generally about mixed methods research and the different aspects of it methodologically, and then I am going to talk a bit about the development, historically, of mixed methods research. And then I am going to focus on some possible developments, or well existing developments, in research which have the potential from drawing on mixed methods, and I am going to outline three different types of developments with some examples from work that I have been involved with or others work or classic studies in the past from the past.
I think a starting point for mixed methods research is that it has been talked about very much in terms of there being a divide between qualitative and quantitative research - what are often referred to as the “paradigm wars”. And here there is an assumption that each type of research derives from a particular set of philosophical assumptions, for example the extent to which the data represents reality or is constructed. And these assumptions have been deemed to be incompatible - and so they are like different ways of seeing the social world, and these different ways have assumed to be somehow don’t … can’t be fitted together, and therefore mixed methods research would seem to be something that one didn’t do, in principal. But I started off with research questions and I tried to turn those research questions into, and think about those questions, and how they related to the kinds of methods - what were the types of questions? Were they hard questions? What questions? How many questions? Those kinds of things. And they were the things, for me, and I think they still remain the research questions are the driving force of what kind of methods we choose.
However, some people have argued, and I think this is the case, the actual methods themselves strongly influence how we frame our questions. So some people come to research with a particular kind of way of seeing in mind. So they immediate are drawn to qualitative methods. So they don’t think first “am I interested in how has the social world come to be as it is” - they think “oh I am going to use qualitative methods”, and that drives them down that route. And another aspect I think, which is becoming more important in terms of thinking about the use of mixed methods - we are in a sort of re-making methods and calling them, you know, “innovative” methods. In my view, most methods really there is nothing very innovative about them - but I think what is changing is how we apply our methods to what kinds of dimensions of the social world we think about.
So increasingly now there is work on the study of emotion, the study of embodiment - and so being methodologically creative helps us to find methods which will somehow engage with those ways of studying the world. Of course there is also, besides the fact that the methods drive the kinds of ways in which people get attracted to methods and the research questions determining methods, there is a third route, which is the practical route, so that certain kinds of data can only be collected in certain kinds of ways. So one is driven by the practical constraints of your situation.
There is also, in the literature, if you go back to Denzin 1970, he was the first to write about this as a methodologist - mixed methods was akin to having different investigators and different teams working from different paradigms. And also, the whole idea of … I won’t go into triangulation now, but the whole idea of having different perspectives on situations, so you would look at different perspectives of, say members of the family on a particular social phenomenon. And because of all these different aspects, people like Alan Bryman have talked about mixed methods more broadly as a multi-strategy research. But you can also have multi-strategy research in which you have a number of different research strategies linked to a number of different research questions in a more complex design. So you may have different teams from different disciplines and so on. You can have your mixed methods strategy in a long-term kind of way, you know, doing different methods over different years. So you might have a quantitative aspect to your research programme with one team of researchers, and then after that you might employ more qualitative methods with a different set of researchers.
The important thing really is from the point of view of the literature, there are two camps - those that people like me use it rather broadly, and those who say mixed methods research has to involve a qualitative and a quantitative aspect to it. But there is this multi-strategy research in which you might have different approaches within an overall quantitative approach and you might have different qualitative methods within a qualitative approach. You might also have … I mean some people say “well is the mixed method strategy the same thing as a mixed method design”? And again, I think that really depends, because if you are doing a case study, for example a case study of a Social Work department, Social Services - you might employ lots of different methods within that. But you might want to call it, and in fact, you know, for all intents and purposes it is, a case study design. So you could have a survey of the employees of Social Services, plus in-depth work with some of the managers and with the clients and whatever.
What does mixed method research imply? And I think some of the kind of benefits, if you like … and I think one of the benefits which has been somewhat played down, is that it provides an opportunity to engage in dialogue between researchers of different methodological persuasions - and that they can engage in across … “the divide”, if you like. And this means that lots of different kinds of logic come into play, so that you may have quantitative researchers who apply an inductive logic when they collect a lot of data and they look at patterns within the data and they draw out from the data some kind of patterns, or they might have a very deductive logic in which they have tested a particular hypothesis - but whatever their logic is, it may be that the logic of another team, or another group of researchers - qualitative researchers - may have a more abductive logic in which they are kind of coming to their research with a grounded theory, for example, in a sort of sensitising way, looking at their data with kind of ideas, but then refining those ideas.
So the point is, whatever the logic that you apply in terms of your data, you have to move back and forwards between theory and data, and whatever your methodological persuasion, you have to engage in a dialogue with the other researchers, whatever their methodological starting point.
So this involves looking for points of consensus and points of conflict and confronting where your differences lie and where your interpretations from your data may conflict with each other. It also means recognising, within a mixed method team, different kinds of generalise-ability, or different kinds of inference. So if you have got qualitative material, you are not going to generalise statistically across a sample because it’s not possible to do that, it’s unlikely to be random in any statistic … or representative in any statistical sense. But your form of inference will be to extrapolate, to draw some conclusions from your sample - either in terms of theory or in terms of saying “under what conditions does X, Y and Z hold true”?
What this means is that we need to work together to create communities of practice, as in inter-disciplinary teams. This trend is increasingly being recognised by the Research Council, by the SRC, which is encouraging the strategic research initiatives and research centres around interdisciplinarity.
So I want to say a little bit about the issues about integrating or combining methods - and a point of caution here, because the literature quite often talks about how you integrate your different analysis in … you different data analysis in that analysis phase. And I think it is very important not to just focus on the analysis phase, because it is important to see it as a whole process - how do the methods link together in the research design? What are the implications for the method, for the data that you are collecting? And also, of course, how are you going to fit them together or indeed are you going to even try to integrate them in the analysis? So these are some of the typical ways in which people have used different methods - qualitative and quantitative methods, combined.
First of all they have used them in a sense for very different purposes - so you might have a survey looking at the frequency of a particular social phenomenon - how many and what are the characteristics under which X, Y and Z happen, and you might have a small qualitative study going alongside it, or following on from it, to look at the processes involved and more in-depth at what is going on.
The second is the traditional triangulation approach, where the idea is that you choose a number of methods - you hope or think that each method is going to somehow measure or be helpful in analysing and understanding the same social phenomenon but from a different … through a different method - and you hope to corroborate your findings, that you will come up with the same results from each of the methods. Well I think this is increasingly seen as very naïve in the sense that methods do determine the kind of data that you create, and there may indeed be corroboration going on, but this isn’t necessary that the data are the same. The third way in which data are combined is that you are using different methods or hoping or expecting to find some tensions between your different data analyses, and to help you to raise new questions or to go back to your data and to look at it more closely. And this doesn’t mean these are mutually exclusive, these approaches.
But another approach of combining methods is to think about methods as bridging the micro and macro levels, or indeed micro, macro and meso levels of society. So in a cross-national study we employ different methods to analyse the work family situations of parents in different European countries. So at the macro level we looked at the public policies, the general discourses in the society about work and family life, we looked at the social trends data, the official statistics. And at the meso level we took organisations, in fact one of them was social services, in each country - and we looked at the organisation and how the policies and practices in those organisations affected what it meant to be a working parent. And at the micro level we looked at the biographies of these working parents and their current experiences of being a working parent. So we had different data, different levels, and this helped us to contextualise our working parents - not only in the societies in which they lived, but in relation to their social class, the kind of jobs they did, the organisations they worked in and the families and resources available to them.
So it is quite often, as Jennifer Mason’s article in Qualitative Research 2006, very usefully argues, is there is quite often a meshing of data rather than integration. Integration always reminds me, in some distant past, from a mathematical idea in which things are somehow kind of tightly bound together in a un-problematic way.
I just want to reflect now, rather differently, on how mixed methods has kind of developed - and I think it is true to say that methodology in general has become a bit of a growth area in the social sciences, while I suppose, individual disciplines have, you know, kind of suffered in current climates. Methodology itself has kind of surpassed itself and it’s a trans social science kind of phenomenon. And certainly the publishers of methodology books are doing very well out of it. Theory, by contrast, I think has rather suffered.
Of course there have been many examples of mixing methods throughout the history of the social sciences, but people didn’t write about it as such, as I am going to suggest. Those of you who may have heard of the classic study by Thomas Zanieki done in the 20s called ‘The Polish Peasant in Europe’ - this is a mixed methods study par excellence. It included a massive written autobiography, letters, archival material, newspaper articles, diaries, statistics, you name it. And this was a watershed at the time because it included subjective data which was very challenging to the accepted ways of thinking about the world at the time. A second example which I’ll mention again later is a study of … longitudinal study done between 1940 something and ‘65 of juvenile delinquents by the - and this included participant observation, ethnography, open ended questions embedded in questionnaires, interviews with respondents, with their families, with key informants and with social work, including social workers. And this was supplemented by archival material - and what I think is amazing about these studies is they were large scale, carried out by teams of researchers and conducted over many years. When you think about the studies that are done today, there is nothing really quite measures up to that in terms of its methodological challenges.
Mixed methods research was something that wasn’t talked about - it was something that you learned on the job, you didn’t go on training courses or read text books on it. And this really contrasts with the situation today in which we have a very distinct and increasingly separate field of methodology, in which people make quite self-conscious choices at the research design stage. And I think in this country at least there has been a move away from theoretically driven research, well really going back to Mrs Thatcher and her abolition of the Social Science Research Council back in the 80s. And we have also seen a pool of researchers, especially in the late 90s, coming into the field who are often not allied to any one discipline.
Another trend which has been going on again since the 80s has been a marketization of research which places importance on efficiency and competence in delivering research. So it has become, or it became cost efficient for researchers to be armed with a variety of methodological skills. And in this process we have seen a closer relationship developing between researchers and those whose interests the research serves, particularly with policy makers who typically don’t often come up with very researchable questions, but nonetheless want their questions answering. And this in turn has led to a fourth trend, which is about the rise, I guess, of qualitative research. Qualitative research has come into its own in a society which seeks to govern the subjectivities of its citizens. So if, as government has done and is doing, wants to transform clients of services into consumers, then the subjectivities of those consumers become very much something that they are interested in. So they don’t just want information about consumer numbers and their characteristics, which, you know, a questionnaire approach might provide - they want to know how they think and they want focus groups, they want also consultative methods and so on. And with the rise of identity politics, this trend has come more and more to the fore.
So in this process I would argue that social science is increasing being redefined as a skill based economy in which training and capacity building are essential - and we have seen, you know, as we are all here today, know a huge amount of investment by ESCR in training and research methods, with a plethora of Masters courses and also short term courses. We have got a National Centre for Research Methods which has sub-centres on particular methods, for example, one of which is based where I am working, which is … they kind of alternate - every 3 years there is another sort of tranche of these sub-centres - they call them “nodes” come on board, focusing on different methods. And also, at an international level, the European Commission framework funding was about capacity building, although they haven’t shown a great deal of interest in terms of the way they organise themselves and methodology. But it was all about fostering international collaboration. And in this process we have seen a market, an international market for mixed methods research text books. And I think, as I have already implied, in this field there is a range of … the mixed methods field has embraced a range of disciplines, but quite often those that are at the centre of the mixed methods … writing about mixed methods, come from disciplines that are not very strongly bounded. So I am thinking about the fields of education, health services, nursing research, business studies. And if you look at … there is an American handbook of mixed methods research. It is very much dominated by people coming from those fields. And there is now a Journal of Mixed Methods Research which is based in the United States.
So that is a bit about its sort of recent development, and now I am going to give you some a bit more … put a bit more meat on the bones and talk about some particular examples. Because I have been engaged in a lot of biographical research related to my interest in intergenerational families, I have kind of looked into this a bit. And as I have suggested, some of the classic studies which use biographical methods, in fact quite often use lots of different kinds of material - they collected lots of different kinds of data. So “The Polish Peasant” I have talked about, and I mentioned that this was a landmark study because it collected subjective data. And I think a very important thing to remember is at that time this was rather frowned upon because social science was about trying to find general laws that govern social phenomena. And this was very important - this kind of positivistic approach was important in America, because back in Europe ideologues were coming to power in the 30s, whose approaches were totally unscientific as well as being totally unethical. So this kind of positivistic thing about being very scientific was kind of partly a reaction to that.
So, you know, you have to understand methods and the kind of epistomological underlying assumptions in relation to the historical moment. And I mentioned the longitudinal studies of the Gluecks which was carried out in the 30s, but also the 40s. Typically, longitudinal studies at that time were often about deviant groups - people who didn’t fit into the normal fabric of society and what they were trying to do was to look and predict what would happen to these people in their future lives. So they followed it up for a number of years, quite a long time, and looked at all the events in their lives, and as I said, interviewed all sorts of other people, and collected data about them and their criminal histories and official records and so on. They didn’t integrate - they don’t actually refer to this qualitative material when they write it up. So they were looking for causal explanations basically. And Glen Elder’s study first published in 1974 called ‘The Children of the Great Depression’ is a cohort study which also, very impressively, collected qualitative and quantitative data. But his were of ordinary children, and he was interested in what it meant to be born in a particular period. So if you were born in the depression, how did that affect your life subsequently? And he shows, amazingly, that actually if you are in the cohort born in this period, if you were born 10 years later, your kind of future life course was nothing like as bad. But he uses the qualitative material as a basis for gaining … for illustrations, and he doesn’t see it as having been systematic enough to codify. But he had … and he incorporates into his writing, and I think this is really important, he incorporates in a way that I think a lot of modern cohort studies don’t do, he incorporates an understanding of the wider historical period. You know, this isn’t just a literature review, this is actually trying to understand what was going on at the time, in relation to the particular life course of these particular individuals. So it’s really an implicit integration of data, the use of context or primary qualitative data used as an illustration.
So I want to move now to look more at the explicit integration of data, and one of the first examples is a study by Tamara Haravan in the 80s which explicitly sought to combine different kinds of data. Her interest was looking at the relationship between work and family life in one particular industrial community in the United States, and she draws on company files and employee files from the communities she studied, she looked at parish records, insurance records and the census of 1900. And it was one of the first biographical studies to integrate this in an explicit way. And she talks about integrating the data, and she observes the empirical analysis reported here - although attempting to weld both types of evidence, at time presents two different levels of historical reality, each derived from a distinct type of data. So she doesn’t seek to claim more than, you know, she is able to. And indeed, you know, this notion that things cannot be neatly integrated is a good example I think where she shows what you get from one data set and what you get from another - and that is perfectly justifiable when you are writing up your analysis.
So I want to turn now to some recent studies. In particular I want to talk about what I think is, and what is, you know, when I first wrote about this wasn’t really much talked about - but is increasingly becoming the case through the initiatives of the SRC, is the linking of qualitative samples to large scale data sets, or reanalysing data from large scale data sets. One of the first examples of this, and where they derived a sample from a large scale data set - there were some people called Loudon & Sampson in a book by Elder & Geilor, and they went back to … amazingly went back to these Glueck studies, the ones I talked about, the delinquents back in the 40s. And they went back, and back in the 80s it must have been, they found these people - they were relatively easy to find, I suppose, some of them were, because of their criminal records. But some of them actually were not easy to find, because contrary to what the research had predicted at the time, they hadn’t followed a pathway into crime - but they managed to find them and they interviewed them. They asked them about their lives, looking back. And of course nobody had asked them that before - they had taken part in this study in a very timed point, but no one had asked the reflect on the whole of their life course. And I think the book that they wrote is a lovely book really, and suggests really that people do value that - especially people who have taken part in studies and then get their chance to tell their story backwards. And they find a lot of inconsistencies between the original findings and what happened to these groups some decades later. And they argue that the approach was useful in 2 ways - first of all it enhanced the original data analysis, quantitative analysis, but it demonstrates how over time, what complex processes take place in individual life course pathways, and that there is no substitute for looking at individual cases, because you know, if you look at any one point in time you are looking at statistical probabilities on groups. In looking over time you have to plot the pathways of each individual into an out of crime, in this case. And they chose a very good strategy, which was they looked for the negative cases that didn’t fit the quantitative results. So they found, they focus on a group of people who were predicted to follow a pathway in crime and who didn’t, and those who were kind of predicted to be, you know, okay for their future and turned out to have got into a load of trouble afterwards. So they argue that without the qualitative case material and obviously this very valuable approach to following them up years later, that often studies can mask the complex processes at work, and giving a semblance of continuity, which isn’t uniformly present. And they also make the crucial claim for biographical research is it’s so important to interpret data in relation to the historical context. So they look at the data and they seek to understand the criminality of these young people in relation to the level of crime at the time and so on. And so it’s not only about them, it’s also about the kinds of roles that they were in at those particular moments in time.
This next study is a study by Paul Thompson who is a sociologist and originator of polydata, the archival qualitative research, who wanted to study what it meant to grow up in a step family. And so he went along to the people who had … were in charge of one of Britain’s cohort studies, the National Child Development Study, who were all recruited at birth into the study in 1958 - and in the 90s he drew a sample from this study of people who, according to their data, had grown up in a step family. And he specifically asked for people who had grown up in a step family between the ages of 7 and 16. And he took a sample of 50 people - and he went back to them and he interviewed them about their lives and including about growing up in a step family and he found that one fifth of them had a completely different story to tell - that they had actually become, or been part of a step family far earlier than 7. And this leads him to … and there were also other awkward, much more awkward inconsistencies I think with the data which I am not sure that the people who were in charge of the cohort study were very pleased about - because you know, it rather threw into question some of their data which, well happens, and I think that data set has had its problems. But it leads him in an interesting article in 2004 in the journal that I am a co-editor of, to reflect on the accounts that people give retrospectively and contemporaneously, if I have got that word right. So they do, at the time it was probably very difficult to admit or more difficult to admit to growing up in a step family than it was subsequently, or their parents were interviewed about … yes, they wouldn’t have been interviewed, their parents would have been interviewed - and so they probably gave, you know, a somewhat edited account. And there was one family in which the mother of the child had claimed that her son was growing up in a step family, and it turned out actually that the father as in fact his real father, because it was only all these years later that she could say that, but not at the time because she married the father somewhat later than she had the baby. And Paul Thompson, in that article also reflects on what it is to take part in a longitudinal study. He says “could it be that to maintain membership of a longitudinal study you have to have a stable and coherent life, or you have to make it appear so”? And I think that if this a risk, that you get only people whose lives feel … they feel able to sort of “fit” the normative, what is an acceptable life, if you like, they are only prepared to take part if they can present an acceptable life - this does strengthen the case for doing more qualitative studies linked to cohort studies, but it means that we have to go out and find the people who don’t take part in the cohort studies, the people who are missing from the cohort studies, to supplement what isn’t there. Because there are attrition rates, but there are also whole groups of people who don’t take part in these studies. So there are benefits to using these large data sets as sampling frames, but we also need to target the non-participants … those who are likely to drop out.
And I am now going to talk about a current study which is quite a nice study in the sense that we are interviewing children from the ages of one and a half to 10, as well as their parents. And it really shouldn’t be me talking about it, it should be my colleague, Rebecca O’Connell. But just to tell you a little bit about how we have derived … well it was one of these bids that the SRC puts out occasionally in conjunction with what was the Food Standards Agency - and they wanted a study, a qualitative study linked to something called the National Diet and Nutrition Survey, which we put in a bid for and won it, much to our astonishment really, because it looked like it was going to be an extremely hard thing to do, because we couldn’t find out an awful lot about that survey at the time and what data would be available to us and what year and so forth, and it was all a bit worrying. Anyway, we decided, you know, when you get offered money these days you don’t turn it down, and we have done a qualitative study which the sample is drawn from this survey. However, and this is all of you who possibly will use such large scale surveys, this is probably a difficult case - but the government, by then the Food Standards Agency, we had a change of government and this survey had reverted to … well not reverted … was passed over to the Department of Health - and these were the people wanting us to do the research. And they had commissioned an independent research organisation to collect the data for their survey. So we had to negotiate with the Department of Health and with the research organisation. So they were kind of tri-party negotiations to get at this … to draw our sample. And it was very, very complicated and took a long time, and we couldn’t identify the people for various incredibly convoluted, technical reasons, to start with. Anyway, we certainly learned a lot - one about the context in which policy makers work, what it is like to work in an independent research organisation and you know, how to kind of … it wasn’t exactly mixed methods, but it was a community of practice learning to work together to access or to draw a qualitative sample from a quantitative one.
Because the survey was all about nutrition and involved nutritionists and people going to measure the height and weigh children and collect diary data on the portions of food that children ate, we also had to talk to nutritionist - but in the end we didn’t have the data at the individual level on nutrition … not that it would have been easy to have analysed it, and there was only limited analysis available to us. So we didn’t seek to assess the nutrition of the children in our qualitative study - we chose to ask rather different questions. Our idea had been that … we found in the literature some evidence that mothers in particular, working families, were supposed to be more likely to have children who were overweight. So our starting point was that - but to question that and to say, to pose a different question, to say “well how does food fit into working family life”? - a much more broader question - and not assuming that it’s only what parents provide for their children that makes them overweight or whatever. But how does food fit into ordinary, everyday, working lives. So we weren’t expecting to have comparable data with the survey, but we did look and reanalysed … or analysed for the first time, because the people in charge of the survey hadn’t done this - we looked at whether there was an association between being a working parent and having children that were overweight or badly nourished. And there was no association. However, they didn’t … the survey team didn’t ask parents their working hours, or the mothers, which was a major flaw in their design and unfortunate. So we looked at some other surveys and we found no such association, although we found a bit of an association in one survey, but only for mothers in higher status jobs, interestingly.
One of the things we looked at, because some of the American literature talks about people with … families with poorer diets tend not to have family meals - they don’t eat together. So this was something we looked at in our qualitative study. And when we looked at our 48 cases we found that it didn’t seem to relate to parents working time or working hours, either fathers or mothers, but it was more about timing, about the synchronisity in people’s lives. And we found that in the working week, about a third of the families did eat together and in two thirds they didn’t and this was sometimes because husbands worked later than the women, but more often than not, it was combined with other factors like, you know, children considered that their young children needed to be put to bed early or the children didn’t like certain kinds of foods and had, you know, were considered to prefer to have child meals at that time. So it was … the organisation and making of family meals was seen by parents as not necessarily about nutritional and health matters, or food quality - but it related to the way in which, the routine practice of everyday life, how you organise things and how timetables fit together.
So our claim for looking at this is one example in which we complete the picture - you remember I talked about complimentarity - well this helps to … is part of the picture, only one little bit, about why people eat the way they do. The Department of Health offered us the opportunity to follow up these families for another two years which we are currently beginning to do. One of our other research questions was about “how do children negotiate food practices”? Because we weren’t … if we are looking at how food fits into family life, we also saw children as active participants in that process, and not assuming that it was all down to what parents do and don’t do. So this required us to develop a theory and we used the term or the concept to practice, which is our unit of analysis, rather than looking at the individual. So practices are about relations between people or relations between people and things. And we also use our different methods - and I should have started this section by saying I am going to use this as an example of mixing different kinds of qualitative methods. So it’s an example of multi-strategy research, if you like, it is not qualitative versus quantitative. It’s different kinds of qualitative methods, including visual method. And what emerges here is comparing the perspectives of children and their parents, but using these different methods. So the parents’ accounts are verbal accounts from interviews, children’s accounts depend on a variety of different kinds of mainly visual methods. The children, as I said earlier, are aged between two and ten, and we chose visual methods to appeal to this broad age range because it is a broad age range really, and visual methods have been used quite widely with children and they have been found to be quite popular with children - and they also help children to voice their views and experiences, and in some cases can help a little bit to overcome that power imbalance between the researcher and the participant. And they are particularly helpful in studying food, because food is a very material thing, and it’s visual and sensual in other ways too. And visual methods are something which can break the frame somehow, they can somehow reveal perhaps what may not be spoken.
So these were some of the methods we used, semi-structured interview of the parents, topic guide for the children, depending on their age and their interest and so forth - and we gave the children paper plates to draw and colour on, the foods that they liked and didn’t like - and we usually gave them, when we arrived in the house. If we were interviewing the mothers first, this kept the children engaged while we interviewed the mothers. And we have got lots of their drawings, or increasing numbers on our website now, which they seem very pleased about, although they are very annoyed when we don’t put their real names next to them, which raised all sort of ethical issues around consent and so forth and anonymisation.
We did have one puppet, which one of us used, and that was a kind of helpful prop in getting the children to engage, although it frightened some of the children so we didn’t use it an awful lot. We used timelines with kind of a drawing of the sun in the middle and the moon at night and a bed in the morning, to help them recall what they had had on a typical, or yesterday, for their mealtimes. And we also used another timeline for a weekend day, and we gave them traffic light stickers with green, red and orange, on which they could mark how much say they had on each of the occasions they had a meal or something to eat. So the green was they had a lot of say in what they had, and obviously red was no say at all.
We also found in our pilot study that younger children engage with photographic images well, and we decided to use some vignettes of photographs as visual prompts to get them to talk about particular issues, particularly about normative views about what is appropriate in this situation, but also to get them to reflect on whether things had ever happened to them. And we had one photograph of a man and a boy fighting over a packet of crisps in a supermarket, and we asked the children what was going on there and why did they think the man didn’t want the boy to have them, and we got some quite strange answers sometimes. And we had another picture of a girl refusing food when she is being offered some food, from her meal on a fork, and we asked again what they thought was happening and should the child eat the food and did it ever happen to them.
And we had a shopping trolley which was an outline of a supermarket shopping trolley and we asked the children to fill up the shopping trolley using their coloured pens to draw things that they liked and put in the shopping trolley - and we noted the order and asked the children about what were they putting in the trolley. Again we got some interesting responses to that - sometimes it related to how the supermarket was laid out and they put in all the fruit and vegetables first, and sometimes they chose things that they really liked. But an awful lot of fruit and veg went into their trolleys, I think partly because of the colours.
And finally we did … a sub-sample of children were asked to take part in a photo elicitation exercise. This gave children a more active research role in the context, especially not only at home, but you know, we asked them … well the parents had to seek permission first, but they could take these to school or to their after school club, to take pictures of food, anything related to food in those settings. And we then went back later when the photographs were developed - we asked them to describe what was in the photographs. And of course some children … well only seven children didn’t take part in any of the activities - but at least … well 41 took part in at least one activity.
What do these different methods add and how might we combine these different kinds of data? And I suggested at the beginning that one way in which mixed methods, different kinds of data are brought together is to corroborate or confirm an account. Secondly, to elaborate or expand on one dataset might help to expand what was found in another, and thirdly one kind of data analysis might contradict another dataset. And I just finish now with talking about Zoe and her participation in the photograph exercise.
She is white British, seven years old, she lives with her older brother who is ten and her two parents who are both full time - and they are a low income household in the south of England. Zoe’s account corroborates her mother’s account. Her mother was asked what influenced her children’s requests for different kinds of food, and she is talking about the influence of television advertising, and she says “well they have always said if they see something on the telly “oh mum, mum, can we buy those”? “No”, “Oh why” - “why do you think? Too expensive or they are not … a yoghurt is a yoghurt, they don’t need it in the shape of a strawberry and pay double the price”.
However when Zoe completed her shopping trolley, she selected a very modest list of items compared with many of the other children - she selected apple juice, cookies, a bowl of fruit, oranges and apples, chocolate biscuits and chicken. And when asked to talk about the vignette of the boy and man who were fighting over the crisps in the supermarket, Zoe demonstrates her appreciation of the household constraints, the financial constraints in her family. And the interviewer says “what do you think is happening in this picture of the boy with the crisps and the father”, and she says “he, the boy wants the crisps and he is having a look at them, and he, the man, is trying to snatch them off him”. “Who is that, is this is dad do you think?” says the interviewer, “why would his dad be taking away the crisps from him in the supermarket”? And Zoe says “because it might be too much”, “yes”, says the interviewer, “what too much?” “Too much pennies”.
Zoe also elaborates on her mother’s interview, when her mother is asked whether the family eats together the mother said “I quite often don’t eat with the children, I get them out of the way, washed, bathed, homework done, food - and by the time I sit down …” And she goes on to say “my dining table is covered with washing, clothes, there is nowhere for them to sit”. Zoe took two photographs which not only confirm, but elaborated her mother’s account through her photographs - one photograph shows the kitchen table as the mother talks about, pushed up against a wall and with clothes hangers visible in the corner. Another shows her wearing her school uniform and eating on the sitting room floor. So they elaborate, but they elaborate on her mother’s account, because they supply new data - because Zoe also took photographs at her breakfast club - and at her breakfast club she listed all the things she ate there. But it was only when she showed us the photograph that she explained what made her experience of eating there different from eating at home. And she says “somewhere that people go”, that’s breakfast club, “and they can buy their breakfast and eat it at tables”. And she went on to say that she preferred to eat at tables because it was cleaner than eating on the floor. So breakfast club made up for what, from Zoe’s perspective, was less than optimal practice at home. And she also, through this account, elaborates on the importance of the social and physical aspects of eating.
And finally, data generated by different methods can contradict each other. In Zoe’s case, the photograph of a meal at home prompted a discussion that complicated the description of family meal times that her mother had presented. Zoe’s mother suggested that she didn’t cook separate meals for the family and that the children got what they were given - “they have the same, I can’t be doing with four different meals, it’s literally have what’s there, and tough”. But Zoe’s description and the photograph tells us a little bit more. The picture that she showed of a spaghetti Bolognese, and she described what was in it. And when asked whether she ate all the meal, she suggested that her mother did in fact adapt the meal to suit her individual taste. Such modifications of shared meals were quite common in our studies, which for example Jenny said she only ate lamb chops if they had melted cheese on top, and David talks about the different food being cooked by parents, if it was spicy food, so they wouldn’t have the spices. So these are micro-adaptations, if you like, but they do suggest a tension between the symbolic and the practical commitment to shared meals and the need or desire to recognise individual differences, and therefore individual identities. They do generate insights, particularly about power and agency in resources in family everyday life. In this case, I have suggested, Zoe is an emblematic case in which we can look at particular situations under particular social conditions.
Just to add really here, issues around power and truth are still relevant when we weave the data together, even though we didn’t seek to present a true picture of family food practices. But there is a kind of middle position here of producing a story of family life, recognising that different members have different versions of that story - but it’s a sense of real life in all its complexity and messiness.
Just to conclude, I have talked about the value of multiple methods and different types of data and collecting different types of data - and I think it is important to extend mixed methods to areas of research like biographical research. It’s important to think about existing datasets and situating small scale qualitative studies in those studies, and you know, with ESRC’s 44,000 household study called ‘Understanding Society’, this is a major such opportunity, because the kinds of questions inevitably that the quantitative team ask are limited. And therefore we could better exploit these datasets.
And lastly, I hope I have suggested how mixed methods can be used creatively, including visual material, especially in addressing the perspectives of, in this case, parents and young children. Thank you
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License