N.B. This research proposal is a response to a specific call from the ESRC/Food Standards Agency relating to the NDNS

Recent research suggests that the nation's diet will be more likely to improve if healthy eating policies take into account changing patterns of family life (Jackson & Pickering, 2009). This study aims to map and understand the effects of a major social change that research indicates affects the quality of children's diets, namely the rise of maternal/dual parental employment in the UK. The study will take as its starting point that children's nutrition and food practices take place not only in their homes but in a range of contexts. The study will interrogate the 2009 National Diet and Nutrition Survey (NDNS), the Health Survey for England (HSE) and the Avon Longitudinal Study of Parents and Children (ALSPAC) to examine in relation to diet the associations found in other studies between childhood overweight and parental employment. This secondary analysis will be followed by an intensive study of 48 working families sampled from the NDNS and selected according to (high and low) income level and the quality of children's diets. This part of the study will seek to provide explanations for statistical associations found (or not found) in the NDNS survey data. It will employ ethnographic methods, including interviews and photo elicitation, in order to understand the social processes which influence healthy and unhealthy diets of children both within and outside the home. The quantitative analysis of secondary data will hypothesise and examine associations between diet and parental employment status while the qualitative part of the study will inductively explore the contextual meanings of 'food use' in working families, the embodiment of food practices, and their embeddedness in different social contexts (inside and outside the home). The implications of the research for policy and practice include informing the design and evaluation of health interventions so that they may be tailored more effectively to the needs of employed families. Through its use of a variety of research tools the study will inform the methodology of future studies.

Background

Between 1995 and 2003, the prevalence of obesity among children aged 2 to 10 rose from 9.9% to 13.7% (Health Survey for England, 2006). As UK studies suggest, early childhood overweight is associated with the propensity for overweight and obesity in adulthood (Gardner et al., 2009). Given the lack of robust evidence concerning the effectiveness of interventions later in the life course (Summerbell et al., 2003), public policy is concerned with children's health (DH/DCFS, 2008a). Research has shown that some groups of children are at greater risk of being overweight and obese than others (Waldman, 2008): those from lower social economic groups (e.g. Kinra et al., 2000; Armstrong et al., 2003; Power et al., 2003; Stamatakis et al., 2005); and some ethnic minority groups (Health Survey for England, 2004; Rennie & Jebb, 2005; Law et al., 2007).

In addition, research evidence suggests an association between maternal employment and children's overweight status (Scholder, 2007; Hawkins et al., 2008). Analysing the Millennium Cohort Study, Hawkins et al. (2008) found associations between maternal employment and overweight among preschool children, a finding supported by some US research (e.g. Anderson et al. 2003; Crepinsek & Burstein, 2004). Specifically, Hawkins et al. found that children's likelihood of being overweight increased with the number of hours their mother worked per week. However, this finding was only significant among higher income families (annual income in excess of £33,000) (Hawkins et al., 2008). These results mirror US studies showing links between childhood overweight and maternal employment in highly educated, well off, white families (Fertig et al., 2003). These findings thus serve to complicate the general picture of a positive association between household income and children's weight (see e.g. Health Survey for England, 2006). Hawkins et al. (2008) hypothesise a link with diet, suggesting that longer maternal working hours may impede young children's access to physical activity and healthy foods. Analyses of ALSPAC suggest that the dietary patterns of children aged three (North & Emmet 2000) and seven (Northstone & Emmet, 2005) whose mothers were in paid employment were significantly associated with a 'junk' dietary component, although these studies do not explore diets in relation to working hours. In apparent contradiction, a survey by Sweeting & West (2005) of older children and their working parents (N=2,146) suggests a link between non-working mothers and poor diets : 63% of 11 year old children whose mothers were at home full-time were classed as eating "less healthily", compared to 52% of those whose mothers worked full-time. However, this latter analysis did not control for income and maternal education which are known to affect children's diets (Gregory et al., 1995; Northstone & Emmett, 2005).

Early nutritional experiences have both immediate and long term consequences not only for younger children's weight but also for their educational attainment and mental, social and economic wellbeing (NHS/HDA 2004; DfES, 2006; FHF, 2007; Feinstein et al., 2008; Golley et al., 2008). A possible link between parental employment and children's diet is important since the rise of maternal employment is one of the key recent social changes which has impacted upon children's lives in the UK (Layard & Dunn, 2009); the economic activity rate for women aged 16-59 rose from 59% in 1971 to 74% in 2007 (ONS, 2007; Walling, 2005). The most striking change in employment rates has occurred among mothers of young children (Berthoud 2007). In 2004, 57 per cent of women with a child of pre-school age were economically active compared with 55 per cent in 1997 (Aston et al., 2005); more than two thirds of working-age women with dependent children (68%) were in employment in the second quarter of 2008 (ONS, 2008).

However, these rises in parental employment have not been accompanied by any significant increase in public policy or workplace support for employed parents. Within the household, although recent studies have highlighted increases in men's care contribution to family life (e.g. Gershuny, 2001; O'Brien & Shemilt, 2003), including food provisioning (e.g. Drydon et al., 2009), gendered sociocultural expectations around carework have not kept pace with changes in parental employment patterns. Women remain disproportionately responsible for food work (Murcott, 2000; DeVault, 1997): the UK Time Use Survey suggests that in 2000 females spent roughly twice as much time as males on the activities of shopping, preparing food and washing up (ONS, 2000). Some US literature suggests that 'time poor' working families may face significant challenges in meeting the demands of feeding their families (e.g. Jabs et al., 2007). Parents' feeding decisions may represent attempts to reconcile competing symbolic and practical tasks. For example, whilst they recognise that some 'convenience' food is not 'healthy', mothers may find it helps them 'meet their priority of feeding their families in a time-scarce environment' (Jabs et al., 2007:24; Warde, 1999).

Theoretical approaches and methodologies

Research suggests that because parents play such a critical role in determining the diets of their children, as meal providers and role models (e.g.Wardle and Cooke, 2008), 'pressures on parents' food choices have great importance for the nutrition and health status of their children' (Devine et al., 2006: 2592). However, children exert their own pressures on parents (Norgaard et al., 2007), form their own preferences and practices and use food to forge (and reject) connection with others. Food is an important way in which people construct identities through consumption (Valentine, 1999), especially for children (e.g. James, 1998; Wills et al., 2008; James, 2008; James et al., forthcoming). In short, food practices are negotiated (DeVault, 1997; Jackson & Pickering, 2009:4). This study will be guided by such ideas and by a practice-based theory of family life (Morgan, 1996) in which children's agency is given due weight (Christensen, 2004). While home and family are of central importance in terms of feeding younger children and establishing eating patterns (Birch, 1998), they also interact with other environments (Bronfenbrenner, 1979) which provide food and influence food practices (NICE, 2006:149). School is a key eating environment for older children and when mothers or both parents work, for younger children nurseries, childminders, after school clubs, and the homes of friends and relatives are environments where a large proportion of the daily diet is consumed. Much of the limited supply of childcare in the UK is provided in the private sector by large companies or small businesses (private nurseries and childminding) where regulations around food consumption are limited (More, 2008), in contrast with the tighter regulations in schools (cf. Belot & James, 2009). A holistic view of children's nutrition and food practices is needed to understand the range of intersecting contexts in which children are nourished and their food practices nurtured.

A focus in some public health policy on individual consumer 'choice' has tended to confound food selection with preference, obscure the socio structural conditions in which food practices evolve and neglect the cultural and emotional factors that influence nutrition (Attree, 2006:67). Food plays important roles beyond providing sustenance, fulfilling non material cultural goals (de Garine, 2004:19) for adults and children (Alcock, 2007). It is an expression of care and identity (Kaplan, 2000). It is also political (Lien, 2004); food mediates power relations, including those based on age and gender (Murcott,1982,1983a; Charles & Kerr 1988). Understanding food practices requires approaches that go beyond rational-choice paradigms to confront 'habitus' (Bourdieu, 1977) - the situatedness of practices in everyday routines and social relations. Providing empirical data concerning the relationship between parental employment and children's diets and the embedded practices and processes which shape them, the proposed study will fill a clear gap in knowledge.

The main aims of the study are to address these key research questions:

* How does parental employment influence and shape family food practices in particular the diets of children (aged 1.5 to 10 years)?

* How do parents' experiences of negotiating the demands of 'work' and 'home' affect domestic food provisioning in families?

* What foods do children of working parents eat in different contexts - home, childcare and school - and how do children negotiate food practices?

The specific objectives of the study are:

1. To examine the relationship between parental employment status and the diets of younger children via secondary analysis of the 2009 NDNS Survey, the Health Survey for England (HSE) and the Avon Longitudinal Study of Parents and Children (ALSPAC)

2. To seek understandings of the food practices of children in working families, including families with high and low incomes and those whose children are classified on the NDNS as having 'healthy' and 'unhealthy' diets, by applying a range of in-depth qualitative methods

3. To develop the methodology in this area through the use of a multi-method research approach

4. To inform healthy eating advice for employed families, by presenting to policymakers and practitioners issues to consider for bringing about improvements in children's diets

Research design: A mixed methods approach

Hawkins et al., (2008:37) have suggested that '[f]urther research is needed to examine factors along the causal pathway between maternal employment patterns and childhood overweight, which can help inform policy and interventions. For example, little is known about differences in children's diet or physical activity levels by maternal employment status. Focussing on diet, this study will employ the different logics which underpin different methods (cf. Murcott, 1995:734; Brannen, 1992, 2005; Bryman, 1988; Greene et al., 1989). Through secondary analysis of survey data it will hypothesise and examine associations between diet and parental employment status. Through an anthropological and sociological approach it will inductively explore the contextual meanings of food use in working families, the embodiment of food practices, and their situatedness in different social contexts and practices. Together these approaches will help to provide a fuller picture (Brannen 2004; Mason, 2006).

Exploiting the knowledge and capacities of the multidisciplinary research team, the study will draw on a mix of disciplines (anthropology, sociology and social statistics) and research fields including the 'new social studies of childhood', family studies, childcare research and the study of food practices to achieve these objectives and complement the FSA's existing work in nutrition, dietetics and physiology. Its sample will aim to focus on families with at least one young child aged 1.5-10 years. The study will have four main phases. The timetable below is proposed based upon NDNS data becoming available in December 2009:

Phase 1: October - December 2009 Access to large scale data sets (NDNS, HSE, ALSPAC), identification of relevant variables, preparation of data set for analysis in Phase 2, design and piloting of qualitative research instruments for use in Phase 3.

Phase 2: November 2009 - Feb 2010 Secondary analysis of large scale surveys. To examine how parent employment relates to children's diets, secondary analysis will be carried out on the 2009 NDNS, the Health Survey for England (HSE) and the Avon Longitudinal Study of Parents and Children (ALSPAC), as all of these collect data on children's diet and mothers' employment. The NDNS is the only rolling UK survey to collect nationally representative and detailed dietary information on children and adults. So far, different waves have concentrated on specific age groups. A key advantage is that NDNS contains linked data on food, nutrient intake, nutritional status and contextual information on individuals. The first NDNS (1992/93) collected data on the diets of 1,340 children aged 1.5 to 4.5 and collected detailed data on the mother's hours of work. The third survey (1997) collected data for children aged 4 to18, with a full dietary record for 1,701 children; the survey asked about mother's employment, but only classified hours of work as full or part time. Some data have been analysed by social class (Gregory & Hinds, 1995), but not by mother's employment. Unfortunately, the sample sizes are quite small, especially for analysing sub-groups (Scientific Advisory Committee on Nutrition, 2008:21), such as employed mothers. Parental employment appears to have been asked in NDNS 2009.

HSE is an annual survey carried out on behalf of the NHS. The survey includes data on diet and nutrition for children and hours of work for mothers (although only full or part time). In 2007 there was a boost sample of children, so that a total of 7,504 children were included (Craig & Shelton, 2008b). The relation between diet and maternal employment has not been analysed. However, even a sample this size has limitations: as the survey report noted, while 'exploratory analysis indicated that there may be associations between a child's perception of how healthy they perceive their diet to be and how healthy their parents perceive their own diet to be... numbers of parent and child pairs ... are too small to produce reliable conclusions' (Craig & Shelton, 2008a: 292).

From the early 1990s ALSPAC followed approximately 14,000 children from birth into their teenage years. It collected detailed data at different age points on children and their families, including data on parental employment and on diet and nutrition (Emmett, 2009). This unique dataset provides opportunities to explore associations between employment and children's diets in a large sample across the childhood years. The level of detail on mothers' hours of employment varies at different points in time: at 47 months mothers were asked about employment, but not about hours of work; at 61 and 85 months mothers were asked about hours of work. However at 73 months mothers were not asked any questions about their employment. This study would look in detail at the effect of hours of work at different ages. Other studies have related diet and nutritional information for children and socio-economic data for adults: diet in 3-year olds (North & Emmett, 2000) and 4-year olds and 7-year olds (Northstone & Emmett, 2005) was analysed in relation to socio-economic factors, including mother's employment, although diets in relation to working hours was not examined.

These three datasets each has some information to contribute, but each has its limitations - in terms of sample size or in terms of details on hours of work. By analysing the data in combination, it should be possible to explore the relation between mother's hours of work and children's diet in more detail than has been done before. To identify a composite measure of diet quality using variables from the NDNS, HSE and ALSPAC datasets the advice of a public health nutritionist will be sought. The nutrition literature contains a wide variety of diet quality indices (Emmett, 2009; Kant, 1996) which have different advantages and disadvantages. The aim is to derive and interpret children's dietary patterns and assess their association with parental employment status using appropriate multivariate analysis techniques. The viability of the NDNS 2009 sample we wish to select will also be tested - single and dual parent earner households with children aged 1.5-10 years. Given the possible small sample size of the latter age group in the NDNS 2009 it may be necessary to carry out the analysis on all children in the survey who are under 18 years.

Phase 3: January - December 2010 Selection of participants, fieldwork and preliminary analysis of the qualitative study of employed parents and children's diet. The 2009 NDNS offers the most up to date analysis of children's diets and will provide the sampling frame for working families. This part of the project will seek to understand the relationship between parental employment and the food practices in which parents and their children aged 1.5-10 engage both in the home and in contexts outside the home. Forty eight dual and lone parent households and high and low income groups will be selected from the NDNS 2009 survey. The sample of households will be drawn from a number of urban and suburban areas in which these data are clustered in several different parts of England. To assist recruitment and thank participants for their time commitment, vouchers will be given to each household. Whilst we would hope to secure all 48 households in this way, we may need to be pragmatic if this is not possible. Further families will be found through snowballing from the selected NDNS participants. These participants will be screened for household employment and age of children, income and broad dietary indicators through completion of a short telephone interview (Casey et al., 1999). The diet data will be collected in the same format as in the NDNS to make the additional sample comparable. Four groups of working families divided by income level and by the quality of children's diets as assessed by the NDNS data will be selected:

(a) 12 households (one child aged 1.5-10) in which the child scores high on a composite measure of diet quality and where the parent /s they live with are employed and in low status jobs;

(b) 12 households (one child aged 1.5-10) in which the child scores low on a composite measure of diet quality where the parent/s they live with are employed and in low status jobs;

(c)12 households (one child aged 1.5-10) in which the child scores high on a composite measure of diet quality where the parent/s they live with are employed and in high status jobs;

(d) 12 households (one child aged 1.5-10) in which the child scores low on a composite measure of diet quality where the parent/s they live with are employed and in high status jobs.

Since the home is at the centre of family food practices the study will necessarily focus on food in the domestic domain. However, because other environments (workplaces, preschools, schools, after school clubs) interact with the home (e.g. Brannen & Storey, 1998; Burgess & Morrison, 1998), supplementing and influencing food practices within it (and vice versa), data will be collected from parents and children about food consumption in these contexts. Further, because food practices are embedded in social relations and social processes, they are not necessarily easily accessible to reflection (Eisner, 2008). The use of creative and visual methods facilitates investigating layers of experience that cannot easily be put into words (Gauntlett, 2007; Bagnoli, 2009). A flexible range of research tools (e.g. Mooney & Blackburn, 2003; Edwards et al., 2005) will be employed as appropriate to bring practices to the level of discourse and to address the contexts in which food is consumed. These will include 4 approaches: (a) in-depth semi-structured interviews with parents and children aged 1.5-10 that includes story telling with children; (b) drawing methods with children (e.g. Backett & Alexander, 1991; Backett-Milburn & McKie, 1999; Hill et al., 1996; Morrow, 2001); (c) a task based exercise in which family members will be asked to suggest working-family-friendly meal ideas for inclusion in a cookbook; (d) photo elicitation interview methods (PEI's) (Collier, 1967; Radley & Taylor, 2005) carried out by parents and children (Punch, 2002) in which they will photograph foods and meals (consumed in and outside the home) for discussion in the interview. Adopting this range of qualitative tools will enable choices to be made about the most appropriate methods for eliciting children's perspectives based on their maturity, competencies and preferences (Hill, 2006; Christensen & James 2000). Rather than only serving as 'records' of food eaten (as in a realist approach), photographs, drawings and storytelling will be employed to aid reflection (Harper, 1998, 2002; Pink, 2001) upon the meaning of food events to participants and the relevance and importance of different social, environmental and temporal contexts. Photography is a particularly appropriate method for achieving our objectives, not only because it may bring the public into the private (and vice versa; cf. Moss, 2001) but also because food is a material substance which appeals to several senses and is amenable to visual representation. Households will be visited twice. At the first visit, the PEI and other exercises will be explained to parent(s) and children; at the second visit, parent(s) and children will be interviewed, employing photographs, drawing, storytelling and the cookbook task as appropriate. Analysis of photographs is a collaborative process between participants and the researcher. As in qualitative research more generally, analysis is an iterative process in which emergent questions guide the collection of future data (Pink, 2004; Jenkings et al., 2008). Qualitative data analysis software (e.g. NVivo/NUD*IST/Atlas.ti) will be employed to assist in the inductive development of key themes, patterns and categories (Hammersley & Atkinson, 2007) specifically relating to the negotiation of domestic food provisioning and the experiences of participants relating to food practices across different sites.

Phase 4: Jan 2011-September 2011. Analysis, writing up and raising methodological questions for future research including the NDNS survey. The fourth phase will involve integrating the data and results generated from the secondary analysis of NDNS, HSE and ALSPAC and from the qualitative study. The ways in which the different data sets will be integrated will be considered from the outset of the study (Greene et al 1989; Brannen 1992, 2005; Bryman, 2006). Mixed methods can provide for an articulation between different layers and types of explanation - macro and micro - each of which cannot be fully explained without reference to the other (Kelle, 2001). However, it is recognised that translation of research questions across different methods of data collection changes their significance and is likely to affect responses and create problems of interpretation. The two layers do not map on to each other readily; they are different forms of explanation albeit they may complement one another (Bryman, 2007; Murcott 1995). We may expect to find some dissonance (Perlesz & Lindsay, 2003) between findings generated by different methods, e.g. reports of the nature of diets in the survey and what is said in the interviews. On the other hand, we may expect the two analyses to complement each other, e.g. explanations for dietary habits may not be viewed in relation to health or quality issues. The analysis of the NDNS and the qualitative study will provide guidance for future research about the value of a mixed method approach for this area of research, which methods and which survey/interview questions are useful and for which purposes.