I would like to thank Isabelle Darmon, Alan Warde and the two anonymous reviewers for their helpful comments, and Lucy Lyall Grant for her translation.
- 1 On the history of canteens and the literature and works carried out on this topic, see the insightf (...)
1Eating out does not carry the same meaning across cultures (Mennell, 1985; Beriss and Sutton, 2007, Holme et al. 2012). It depends on commercial supply and, more generally, relates to the specific relationship that a culture has with food. The seminal work of Warde and Martens on eating out in England has brought to light six main characteristics of this activity: ‘Eating out is a specific sociospatial activity, it involves commercial provision, the work involved is done by someone else, it is a social occasion, it is a special occasion and it involves eating a meal. These six themes constitute the shared understanding of eating out’ (Warde & Martens, 2000: 46). Excluded from this definition, then, are meals eaten in the workplace during working hours, and more specifically, canteens. The definition underlies ‘the importance of the distinction between eating out as entertainment and eating out as a necessity’ (Warde & Martens, 2000: 36). In France, conversely, the literature is more closely focused on the frequenting of canteens,1 while restaurants are mostly understood through their entrepreneurs and the history of their establishment (in a city, a community, etc.) (Hassoun 2010, 2014). These French studies as a whole shed more light on the social relations within the world of work than on family and individual practices of eating out. Nor do they enable us to understand how this consumption fits into everyday food habits or lifestyles and consumption patterns. Setting eating at work and eating out at the restaurant against each other immediately raises the question of eating out in a formal setting.
- 2 INSEE : See Appendix 1 for the list of acronyms
2This paper focuses on a cross-national comparison of the understanding and practice of eating out in France and Britain. It aims at defining the contours of eating out in each of the two countries. What sort of meals and practices do the expressions ‘repas hors domicile’ or ‘meals out/eating out’ refer to? As stressed by Isabelle Darmon and Alan Warde in the introduction of this issue, earlier attempts of comparisons using harmonised surveys took ‘the risk of monitoring specific and largely decontextualized habits’. This is why rather than carrying out secondary analyses drawing upon categories which are taken for granted, we took an opposite approach in terms of methodology, based on the analysis of survey classifications. Classifications lie at the heart of the analysis carried out here. Such methodology has proved to be very effective. In France, it is mainly socio-professional classifications that have attracted scientists’ attention, whether they looked into their genesis (Desrosières & Thévenot, 1979), or, later, with the exponential growth of econometric models and the changes in recruitment conditions for INSEE2 statisticians, into their decline in favour of qualifications and income (Pierru & Spire 2012). Previously, Luc Boltanski pointed out the role of econometrics for the shaping of food classifications in the 1963 Household Budget survey. He showed how classifications based on a division adapted to the nature of their specific calculation techniques led to a naturalist interpretation of the results (Boltanski, 1970: 36).
3The paper draws upon a corpus of classifications from consumer surveys from official statistics in France and Britain, which seem fairly comparable in their objectives and collection methods. The comparative approach adopted here is conducted in a dual movement (Sobal 1998). We analyse separately the topic of eating out in each national series of questionnaires. The classifications are analysed diachronically in their national context to allow for comparison over time and so as to understand how their parameters evolve. As Uprichard points out, ‘The qualitative changes to the classification or variables are just as important as the quantitative change within them’ (Uprichard, 2012: 95). First, we shall ask whether the topic was taken into consideration in the surveys. What do questionnaires focus on (lunch, dinner, week day, weekend, contents of the meals, places or prices…)? What were the main changes over time? Although the period of time under consideration here is relatively short (approximately 60 years), it was rich in events that, more or less directly, affected eating habits and routines. The aftermath of the war and the end of rationing, economic expansion, the mechanisation of housework (especially for cooking), were all events or trends that rapidly and durably changed people’s habits.
4As demonstrated by Desrosières and Thévenot (1998), the classifications in use in statistical survey questionnaires are linked to statistical, political and cognitive representations of society.As far as expenditure surveys are concerned, an international classification (COICOP) has been in existence since 1998: it aims to calculate the harmonised European index of consumer prices and purchasing power parities between countries (Boeda, 2008: 10). This classification divides the ‘catering services’ category into two headings: 11.1.1. ‘Restaurants, cafés and the like’ and 11.1.2. ‘Canteens’. Nevertheless it corresponds only at the most aggregate level, leaving countries free to add as many sub-categories as they wish. We shall see how French and British classifications effectively diverge from this framework. By analysing in detail all the criteria and how they have evolved in questionnaires, we will be able to define the issues at stake (for the authorities, at least) in eating out and to delineate national profiles.
- 3 On the setting up of these surveys during this period, see Desrosières (2003). For a detailed list (...)
5Our research is principally based on four series of surveys from official statistics: the series of INSEE surveys ‘Consommation alimentaire’ (Food Consumption) (1969-1991) and ‘Budget des familles’ (Household Budgets) (1965-2010) for France and the ‘National Food Survey’ (1940-2000) and ‘Family Expenditure Survey’ (1957-2000) for Great Britain, these two surveys having been combined in 2001 into the ‘Living Costs and Food Survey’. Surveys commissioned by private bodies were not taken into account as they proved more difficult to use in a comparative approach because we know little about how they were produced. In addition to so-called ‘permanent’ official statistics, i.e. reconducted over time, other, occasional surveys were selected if they seemed relevant to addressing the question of eating out. In terms of chronological markers, this paper covers the period 1940-2005. The interwar years will be the topic of later studies. The juxtaposition of French and British chronologies in the form of a timeline [Appendix 2] reveals a strong continuity for the English surveys and a relative discontinuity for French surveys. In particular, between 1941 and 1965, myriad occasional surveys existed in France with relatively varied aims and procedures, carried out by newly created bodies such as INH, INSEE, INED, CREDOC, etc., which do not feature in the timeline3. They are currently being examined in order to verify the type of information collected and if they contain questions concerning eating out.
- 4 The National Archives [TNA] MAF 156 Ministry of Food and Ministry of Agriculture, Fisheries and Foo (...)
6In Great Britain, the surveys have been extremely stable since 1940. It can be summarised as two main surveys carried out annually. The National Food Survey was established during the war by the Ministry of Food to monitor the population’s diet and the effectiveness of measures taken to help feed the population. Initially known as the Wartime Food Survey, it became the National Food Survey in 1950 when it began covering a national sample of several thousand households that had to fill in a logbook of their consumption over 7 days. In parallel, 1957 saw the start of the Family Expenditure Survey (1957-2000) conducted by the Department of Employment (1957-1988), the coordinating body of England’s official statistics set up in 1941 (Desrosières, 2000 : 203; Moss, 1991), then the Central Statistical Office (1989-1995) and Office for National Statistics. The survey was not limited to food expenditure, and covered all household expenses, to be used to set the United Kingdom Retail Price Index. The survey comprised both a questionnaire on a household scale and a consumption diary over 14 days concerning each member of the household over 7 years of age. Since there was much overlapping in the two surveys, they were merged in 2001 to become the Living Costs and Food Survey (initially entitled the Expenditure and Food Survey), which continues to this day on an annual basis under the authority of the Office for National Statistics. The corpus of British surveys is completed by two others. One, conducted in 1949, stems from the Wartime Food Survey and only covers meals out. No official publication of this survey exists, but we were nonetheless able to find traces of it in the Ministry of Food’s archives4. It appears that since the results were not considered reliable, they were not published. However, they can be compared with those of another survey carried out the same year on the same subject by the Central Statistical Office. This second survey was repeated in 1951 and 1956 and the results published in 1961 (Kemsley & Ginsberg, 1961).
7The French corpus is not yet completely stabilised. For this paper, it began later with the INSEE ‘consommation alimentaire’ (food consumption) survey (1964-1991), conducted annually between 1964-83 with the exception of 1968 and 1975, and then biannually, and the INSEE ‘Budget des familles’ (Household Budget) survey (1965-2010), annual between 1965 and 1972, and then carried out every 5 years. The two series of surveys contain representative samples of ordinary households. When it was launched in 1965, the food consumption survey was part of a planning objective with regard to the Common Agricultural Policy that required ‘National Food Balance Sheets’ (Desabie & Rempp, 1964). Its aim was to study all food provisioning for the home, whether originating from a purchase, a gift received, taken from household production (home consumption) or the shop’s stock (self-provision) and expenditure on eating out. The information was collected from a questionnaire and a consumption diary over 7 days filled in by the household member responsible for food provision. The aim of the Household Budget survey was to assess the annual growth of the household consumption budget for goods and services (Desrosières, 2003: 107). It did not only cover food consumption but concerned all household expenditure on goods and services. It comprised a questionnaire recording all sources of income and the household’s main expenditure over the last 12 months as well as a diary to be filled in by all household members over 14 years of age, in which all expenses over 14 days were noted.
8The analysis concerns the classifications resulting from the data gathered from the questionnaires and/or the diaries for the sections on eating out. Although not sufficient to exhaust all the classifications at work in a survey (Lemel, 1988), it involves less aggregate data or ‘dirty data’, to borrow Uprichard’s expression, as opposed to data that have been cleaned up with the aim of reconstituting series that are comparable over time: ‘by leaving the data alone, and using the categories in their raw form as much as possible, descriptions of both qualitative and quantitative changes to social classifications systems are rendered possible’ (Uprichard, 2012: 94). However, it is not a question of raw data as declared by the survey subjects in their diaries, but rather the classification based on which these data were recorded and coded by the INSEE. This material was party collected via the Quetelet and Economic and Social Data Service websites, which dispose of a set – albeit incomplete – of digitalised archives of the surveys. The INSEE Food Consumption survey is available almost in its entirety from the Quetelet portal (only the years 1967, 1973 and 1974 are missing), as is the Household Budget survey from 1972 onwards. Likewise, the National Food Survey is partially available on the ESDS site from 1974 and the FES from 1961. For the missing surveys, the corpus was completed from the library when questionnaires were published as appendices to official reports and publications. For the British surveys from the 1940s and 1950s, the corpus was also completed using The National Archives (TNA).
9The growing autonomy of eating out in the bibliography of the Food Consumption survey reflects the expansion in volume and in value of a new consumer trend. Between 1965 and 1985 (dates of the surveys), three publications were specifically devoted to this (Thi Nguyen Huu, 1970a, 1970b, Villeneuve 1974). From the 1980-81 survey onwards, a more regular publication began. However, it was not yet autonomous since the results were published in reports devoted to meals (at home and outside the home) (Mercier 1984, 1985). It was from the 1987 survey that meals out became autonomous in publications, which began to treat eating out independently from meals eaten at home (Manon, 1991, 1992, 1993, 1994). Carried out at the INSEE, the official body in which the proportion of economists is growing (Pierru & Spire, 2012), this mainly involved assessing the share of the budget devoted to the catering industry, both commercial and collective.
- 5 We should make it clear that the categories specific to beverages (called “drinks” in the diary as (...)
10The Food Consumption survey’s data collection classification has evolved in the sense of an increased number of lines of 9 categories in 1967 to 16 in 1991. From the beginning, the diary was presented in the form of a table with 3 columns in which the person responsible for procuring provisions recorded ‘all food purchases made by yourself and members of your household’, specifying the designation and the amount spent, including drinks consumed in cafés and meals eaten in canteens5. On the right-hand page of the diary there was also a column entitled ‘Meals eaten outside the home (restaurant, canteen, in another home) by the members of your household’, with a few lines left blank to add details. Next to this table was a coding classification for the investigators combining two main criteria: the location and the fact that the meal was paid for (Table 1). In this way, commercial catering premises (restaurants and cafés) were distinguished from collective premises (canteens) or domestic locations (‘at the home of the employer’, ‘in another home’) and whether or not the meal was paid for (‘as a guest’, ‘as a paying guest or customer’ or ‘packed lunch’). Snacks formed a classification entry by product rather than where they were eaten, in order to distinguish an eating out expense from an earlier expense (packed lunch) accounted for in the ‘procurement of provisions’ section of the diary.
Table 1: ‘Meals eaten outside the home’ section
1967 – Consumption Logbook
Type of meal eaten out:
1: Restaurant, paid for by household member
2: Restaurant, not paid
3: Canteen, company restaurant
4: School canteen, university refectory
5: Snack in a café
6: Snack brought in, packed lunch
7: Meal eaten in employer’s home
8: Meal eaten as a guest in another home
9: Meal eaten as a paying guest in another home
10: All meals eaten outside the home
Indication of day of week and of midday or evening meal
Source: INSEE Food Consumption Survey 1967
11Analyses devoted to meals out for the 1967 surveys and after 1971 mainly concerned the breakdown of meals out, clienteles, expenditure and their significance in household budgets. During this initial period, the authors pointed out the growing number of meals eaten in canteens (school, university or company): the number quadrupled in fifteen years (from 8,600,000 units per week in 1956 to 34,600,000 in 1971) (Villeneuve et Bigata, 1975: 11). This observation led the authors of successive reports to focus on midday meals in particular because the most rapid developments concerned this meal. To these recurring themes, Villeneuve and Bigata added two extra chapters to their analysis of the 1971 survey: invitations between households and above all, the ‘aspects of eating out connected to family life’. They discussed the effects on meals eaten as a family of the increasing number of meals eaten outside the home, whether this involved adults’ or children’s meals (Villeneuve & Bigata, 1975). The authors analysed in detail ‘the habitual family composition around the kitchen table for the midday meal’, the ‘frequency of the head of the family and his children meeting during the midday meal’ and ‘the housewife’s role as responsible for the family meal’. In short, does the rise in meals eaten outside the home threaten family cohesion? The authors did not have the resources to answer this question, which was not taken up in subsequent reports. However, their perplexity in light of this growing trend was an indication of the rapid progression of new consumer habits linked to new routines. The arrival of women onto the labour market, the increasing distance between home and workplace, etc. led households to rethink family routines and domestic organisation.
12In the 1980-81 survey, the location of meals was broadened to include medical and social institutions (e.g. sanatorium, crèche, mess). The category ‘skipped meals’ also appeared, only to disappear the following year. At the same time, the lines ‘meals eaten in the home of the employer’ and ‘meals eaten as a guest in another home’ were merged. Mercier observed that while the share of this category had not evolved since 1974, on the contrary, the structure of the budget item indicates a ‘reduction in the number of meals eaten in the home of the employer in favour of meals eaten as a paying guest in another home, including children’s meals with child carers either in the child’s home or the child carer’s home’ (Mercier, 1985: 21). In 1981, Mercier observed an increase of + 37 % in the average number of meals out per person between 1967 and 1981 (from 1.9 to 2.6). At this date, eating out represented 13.8 % of the total food budget (compared with 11.5% in 1971 (Mercier, 1985). Under the instigation of Marie-Annick Mercier, in charge of the INSEE survey, the increasing significance of meals out in household budgets gave rise to more regular publications on eating out (Mercier 1984, 1985).
- 6 On the emergence of fast food outlets in France, cf Fantasia (1995).
- 7 Interviews will enable us to understand why it was not taken directly from the latest classificatio (...)
13In 1982, the list of medical and social institutions grew longer (hospitals, holiday camps, etc.) without, however, becoming the subject of specific analyses. Meals eaten in another home with payment in kind also appeared. Eventually, in 1985, the list of premises was expanded to include the category of ‘fast food’6. This is remarkable insofar as the classification records no details for the ‘restaurant’ category, which combines very different premises in terms of price, menu and use. From 1971 onwards, Villeneuve and Bigata specify with regard to this category, ‘the budget item is relatively well defined, but we should draw readers’ attention to its wide heterogeneity; a meal in a restaurant can range from a simple ‘assiette anglaise’ (assorted cold meats) eaten rapidly in a small snack bar to a copious business dinner or festive meal’ (Villeneuve & Bigata, 1975: 9). It is interesting to note the two English references to the ‘assiette anglaise’ and ‘snack bar’ in the French text, leading one to believe in a British influence that, however, has no equivalent in contemporary British classifications... Another addition to meals connected to medical and social institutions was meals delivered to homes for the elderly. Because the survey ended in 1991, this was taken up by the INSEE’s Household Budget survey. Although this survey began in 1965, for this paper we consider the classifications from 1984-85, the date at which the classification concerning meals eaten outside the home was copied from that of the 1967 Food Consumption survey7.
- 8 In 1994 the CREDOC carried out a survey for The Directorate General for Competition Policy, Consume (...)
- 9 N.B. this category appeared in the Food Consumption survey as early as 1985.
1419958 marked a turning point in redefining the classification exclusively concerning meals from the commercial sector (whilst still retaining canteens). This radical change is explained by the INSEE’s concern to ensure that these categories corresponded to those used in national accounting. By measuring the product of each branch of activity, national accounting adopts a different method to aggregate expenses linked to food, in particular considering meals eaten outside the home as ‘services’ independent of food consumption (Lambert, 1992). These expenses were not, therefore, integrated into households’ food budgets but allocated to the social activities from which they stemmed: food consumed in cafés and restaurants was included in the ‘cafés, hotels & restaurants’ column, food consumed on transport etc. was included in the column corresponding to ‘transport’ and so on. In the Household Budget survey, this was not exactly the case because expenses in commercial premises and in canteens remained considered in their own right. This explains, on the other hand, the disappearance of meals eaten in medical and social institutions (included in the expenses linked to these institutions) or invitations to eat in other people’s homes. Thus, all meals eaten at home, especially invitations from others (treated elsewhere) and snacks (made from domestic production) disappeared from the classification. At the same time, the list of commercial premises was expanded with two new categories: the category ‘cafés, bars, etc.’, no longer merged with restaurants, and the ‘fast food’9 category, divided into two sections based on the distinction of whether the food was eaten on or off the premises. Lastly, the snack disappeared at the same time as fast food appeared, without its being discernible whether this was purely symbolic or if it reflected market trends. In 2005-2006, commercial premises became even more specific with the addition of the category of ‘cafeteria’. The snack that had disappeared returned in the anglicised form of ‘sandwich’, with the commercial connotation absent from ‘snack’. This example indicates the INSEE’s increasing interest in assessing the share of the budget devoted to eating out. In particular, it was observed that in 2005, ‘households devote a share of the budget to eating out that is barely greater than it was in 1960. Nevertheless, this increase is remarkable if we compare it with the drop in total food expenditure at home: almost 12 points over the same period’; the budget item ‘restaurants’ had doubled since 1960, becoming the most significant expense in the eating out budget (61%). In parallel, the share of cafés was divided by three to represent 16% of the budget item, and the share of canteens remained at 23% (INSEE, 2009).
15Thus, through successive modifications, the classification of French surveys progressively focused on commercial premises, whilst still retaining school and company canteens. The 2005-2006 survey gives the most detailed list of commercial premises. On an initial level, it sets ‘restaurants’ apart from all the other premises grouped together in the catch-all category of ‘cafés, bars and equivalent premises (buffets, refreshment areas, tea-rooms, fast-food outlets, etc.)’. This second category is then divided into 6 sub-categories: cafeteria, café/brasserie/snack bar/tea-room, fast food outlet (eaten on the premises) fast food outlet (to take away), other premises outside the home, other, unspecified premises. In total, there are 7 levels of distinction, reduced to two in the 2010 survey, i.e. the aggregate level of the previous survey: meals eaten in a restaurant and meals eaten in a café, bar or equivalent premises. In line with the COICOP, the category ‘school or workplace canteen’ was retained.
16This initial exploration of French classifications enables us to understand the complexity of the mechanisms at work in drawing up classifications. On one hand, they are based on economic and social trends that have to be recorded by the nomenclature. However, to accommodate longitudinal comparison, this is subject to a certain inertia. They are also based on restrictions linked to the institutions themselves. The Household Budget survey is now aligned with national accounting categories, reducing even further any room for manoeuvre. To conclude, used in a survey with an economic objective, they are the result of a tradition of administrative statistics describing the population and the economy for the requirements of the State (Desrosières, 1985, 2003). The topic itself of food consumption appears not to be especially appealing beyond the calculation of budgetary shares.
17In the National Food Survey’s bibliography, there are no reports exclusively devoted to eating out. However, each report contained an analysis of meals out that calculated the total average number of meals out per week and per person, as well as the average number of midday meals. The aim of this information was resolutely nutritional: from the very beginning, it was a question of assessing the contribution of meals out to average daily calorie intakes. Only two reports contained a section devoted specifically to meals out. In 1954, ‘Appendix B: Meals outside the home in 1952’ delivered a series of findings, the objective of which was less to shed light on meals out than on food eaten at home: ‘The Survey collects certain information on meals out in order to calculate the proportion of nutrient requirements represented by the domestic diet’ (Ministry of Food, 1954: 53). The report published in 1995 contains a section specifically devoted to eating out (Section 4 Special Analysis – Eating Out) which presented the results from the questionnaire put to a sub-sample of the survey on meals out, established following the Food safety Act (1991).
18With regard to classifications, during this period we can observe an increase in the number of questions but above all, a change in the nature of the type of information collected and a growing precision in defining meals out. In the early 1950s, the National Food Survey questionnaire examined who ate the meal (age and gender) and the ‘type of meal’. This was to apply weights to calculate the indexes of meals eaten at home insofar as the nutritional composition of meals out differs from that of domestic meals. In 1952, almost half the population never ate a single meal out, and a fifth of the population ate less than 5% of its meals outside the home (i.e. one meal per week) (Ministry of Food, 1954: 54). The second subject of concern that was also discussed in an appendix to the 1954 report involved school meals: ‘school meals are of such importance that they merit special attention’. These two types of information continued to be gathered in exactly the same way until 1975. This was the year that saw the introduction of the category ‘Packed meal from home’, and with it, the idea of eating outside the home food brought in from one’s home. In 1986, information about packed meals was completed with their content (‘packed meal content’).
19The 1990s were marked by an acceleration in modifications made on the one hand to the classifications of the diary, and on the other, to the subject of eating out itself, for which a special diary was established due to the increasingly large proportion of meals out. ‘A number of factors suggest that eating out is an increasingly important part of modern society. The growth of fast food outlets and their success indicates that eating out quickly and conveniently is a major activity. Women work outside the home more, and many people eat out at least once a week. There is more of a social element in eating out as well, and friends and families eat out as a group today rather than visit each other for home prepared food. It is estimated that expenditure on food bought and consumed outside the home has risen from about 20 per cent of expenditure on food ten years ago to around 30 per cent today. If we accept then than food bought and eaten outside the home provides a significant proportion of total food intake, it is clear that the NFS can no longer provide reliable data about total dietary intakes of nutrients, additives etc’ (Yeomans, 1991: 97). According to the report of the 1990 survey, ‘Information on the source of mid-day meals for children aged 5-14 has been collected continuously since 1972. Analysis of the data shows that the number of packed lunches has quadrupled over this period, while there has been a marked decline in meals at home and, to a lesser extent, school lunches’ (p.11). In 1990, the majority of meals were eaten at home, ahead of school meals and then packed lunches. The same year, the diary classification regarding school-aged children’s meals was expanded with the category ‘other lunches bought outside the home’. The classification also integrated ‘meals on wheels’ (meals delivered to the elderly in their homes). In 1991, the methodology of the survey was modified slightly, ‘such that a light tea or supper (when taken in addition to another evening meal) is no longer regarded as a meal’ (MAFF, 1992: 11). The following year, the mention ‘snack’ appeared next to that of ‘packed meal: ‘did anyone take a packed meal or snack from home to eat out?’ Above all, however, 1992 saw the addition of a diary specifically devoted to meals out, entitled ‘Eating Out Extension’, reproduced twice in 1994 and 2000.
20The information collected in this special diary completed rather than replaced that of the ‘classic’ diary. According to the special report published by the ministry, ‘it was initiated to fill a gap that existed in the coverage of government dietary and nutrition surveys and to provide a more complete picture of the national diet’ (MAFF, 1995: 39). The findings were organised into the following divisions: expenses, consumption, and nutritional analyses. The extension was drawn up because of the Government’s refocusing on health and safety concerns from a nutritional perspective, in line with the 1990 Food Safety Act (Maclean 1991: 4-5).
- 10 N.B. School meals and Meals on Wheels appear in their own right in the sub-category “Multi-course M (...)
21This modification was accompanied by changes to the definition itself of ‘meals out’. From 1992, the perimeter no longer included food brought in from home: ‘Eating out is defined as the consumption of food and drink outside the home that has not been obtained from the household supplies’ (Rimmer 2001, p.1173). The movement occurred simultaneously in the NFS and Family Expenditure Survey. The NFS ‘Eating Out Extension’ distinguished eating out expenditure according to the product (‘Average expenditure on food and drink consumed outside the home’) and the premises (‘Expenditure on food and drink eaten by outlet type’) (Table 2 for the aggregate classification and Appendix 3 for the full version). The first category was divided into three sub-categories distinguishing consumption according to the type of meal (‘composite meals and snacks’), the type of food or dish (‘individual food and dishes’) and the third category exclusively concerned ‘drinks and confectionery’. Although the number of meals eaten outside the home remained in the diary, the extension concerned average expenditure. The information collected involved the number of courses10 (from ‘single’ to ‘multi-course meal’), the main ingredient of the dish, including for sandwiches, for which there was a detailed list of fillings, and the ethnic nature of the food consumed (‘Chinese or Indian meal, ethnic foods’). The version of the extension published in 2000 was simplified, giving a single list of products (meat and meat products, cheese and egg dishes and pizza, yoghurt, bread sandwiches, etc), and no longer including information on the number of courses and the premises where the food was consumed. Specific meals such as school meals and meals on Wheels also disappeared, making way for a list exclusively concerning food or ingredients. In other words, the collection unit moved from the meal to the food.
Table 2: NFS - Eating Out Extension: aggregate level
EO Extension 1994
EO extension 2000
Average expenditure on food and drink consumed outside the home
Average consumption of food and drink eaten out
COMPOSITE MEALS AND SNACKS
- Multi-course meals
- Single course meals
INDIVIDUAL FOODS AND DISHES
Expenditure on food and drink eaten by outlet type
List identical to the 1994 list
-Restaurants, public houses and takeaways
- Other outlets
Source : MAFF National Food Survey 1994
22In 2001, the NFS was merged with the ‘Expenditure and Food Survey’, which used the same classification for the 2001-2002 edition. In 2005-06, the classification was completely overhauled. It incorporated a first section entitled Free Food (‘for the sake of completeness from a nutritional point of view’) that disappeared in 2007. Groups of food were replaced by categories grouping meals according to their ethnic origin (Indian buffet, Chinese/Thai buffet, all other ethnic meals) and the four other categories concerned products: salad buffet, sandwiches and rolls, meal (not specified) and soft drinks. The 2007 version made the task of identifying specifically what has been consumed even more difficult. With the exception of drinks, for which a detailed list was given, solid foods were only documented by their external characteristics: was the food hot or cold, eaten on the premises or not, or purchased outside the home to be eaten at home. The food itself was grouped by class: confectionery, ice cream or the overall term of ‘food’ (hot food, cold food, etc). Where the food was eaten was mentioned only via the statement ‘on/off premises’, with the exception of meals eaten in the workplace.
23We can thus observe the progressive and remarkable breakdown of ‘meals’ into consumption occasions until the 1980s, the unit remained the ‘meal’ out; in the 1990s, the Eating Extension introduced the breakdown of the meal into food intake whose unit became the dish or product based on a detailed list of foods. In the 2000s, this detailed list disappeared, replaced by more aggregate categories and above all, a distinction based less on the food consumed and more on its ‘external’ characteristics (hot/cold, on premises/take away, etc.). The types of food consumed noted in Table 4 apply to any food intake independent of the status it is given by the person eating it (meal, snack). Interviews with those responsible for the EFS will enable us to gain a better understanding of its objectives.
24The aim of this paper was to present original material for an exploratory comparative analysis of the contours of the definition of eating out in France and in Britain, and the modifications made to these parameters. Despite the European framework to harmonise classifications, this initial exploration of the material highlights some similarities and numerous differences. In both countries, after successive modifications, classifications progressively focused on meals eaten on commercial premises (with the exception of workplaces). However, this move occurred in very different ways in the two countries. The ‘restaurant’ category is extremely revealing from this point of view. In the French surveys, within establishments in the commercial sector, it gradually became distinct from ‘cafés’ ‘cafeterias’ and ‘fast food outlets’. British surveys do not have a ‘restaurant’ category but use the category ‘catering services’, encompassing the entire commercial sector. On the other hand, the list of products collected allows for premises to be better specified since fish and chips or Chinese/Thai buffet provides both information about the nature of the food eaten and where it was consumed. Instead of the premises unit that continues to prevail in France, implicitly underpinned by a price scale, British surveys use a nutritional unit, explicitly underpinned by the type of food or dish eaten.
25The resolutely nutritional focus of the British surveys from their very beginning gives us a partial understanding of how they developed. This focus can be understood in the light of the history of administrative statistics. Desrosières, in particular, has shown the way in which public health has occupied a crucial place in British statistics. Whereas in France, the ‘‘moral statisticians’, gathered together around the Annales d’Hygiène Publique (Public Health Annals), remain external to French official statistics, their British counterparts from the Public Health Movement are at the centre of official statistics in the General Register Office’ (Desrosières, 2000: 185). The NFS thus fits into the framework of nutritional observation measures under the responsibility of the Ministry of Food, and the survey is part of Britain’s long-standing interest in nutrition (Smith, 1997). Conversely, in the French survey, the focus on catering services and their accounting into household budgets confirms previous demonstrations of the growing role of econometrics for shaping classifications (Boltanski, 1970; Pierru and Spire, 2012) and, most importantly, for shaping the knowledge and the social representations of eating out. As the attention moved from the frequency of eating out (which mainly concerned midday-meal and eating at work) to expenditure (which mainly concerned restaurants), meals out were reduced to an economic question and nothing was said about their place in the modes and rhythms of life.
26However, the work to explain this remains to be done. It is currently involving gaining a better understanding of the chronology of the main evolutions of the surveys against the economic and social contexts, the public policies they reflect and also the transformations of the commercial landscape.