Leisure pursuits: uncovering the ‘selective tradition’ in culture and well-being evidence for policy

ABSTRACT This article reconsiders the value of leisure studies to well-being evidence for policy-making. It presents secondary analyses of Office for National Statistics data from the Measuring National Well-being (MNW) debate (2010). It finds that leisure is more important to the nation than accounted for in official reports. Using Raymond Williams’ framework of ‘lived culture’ and ‘recorded culture’, the article interprets this analytical discrepancy as demonstrating the culture of ‘selective tradition’. Raymond Williams’ work is used to understand the sidelining of leisure in the MNW programme in two ways. Firstly, the article addresses survey design and analysis. It finds a methodological bias against free text data as evidence, affecting which aspects of cultural and social life appear central to well-being. Secondly, the article takes a discourse approach to cultural sector advocacy, which, in arguing the value of arts-related activity, relegates aspects of ‘culture as ordinary’ by default. These two selective traditions mean that ‘everyday participation’ is not valued in discussions of well-being evidence, advocacy or in the final MNW measures, despite its value to the public. This article prioritises everyday understandings of well-being over those of experts, offering an ethical and practical contribution to leisure studies and policy-making for well-being.


Introduction
This article aims to reconsider and reinstate the value of leisure to well-being for policy. Measuring well-being is promoted as an improved way of evaluating governance decisions and tracking social progress nationally and internationally (Davies, 2015). The 'rise of well-being' in politics emerges from developments across intellectual fields, including psychology, social policy, economics and social statistics (Bache & Reardon, 2013, p. 908). While various policy areas have been increasingly claiming their contribution to personal or societal well-being over the last twenty-five years, these appeals are rarely evaluated on their own terms (Oakley, O'Brien, & Lee, 2013).
Policy areas such as culture have sought a clear identity in arguing their value to well-being (Oman & Taylor, 2018) through a 'special case' rhetoric (O'Brien, 2013) embedded in claims to a historical tradition (Belfiore & Bennett, 2008). Alongside this, increasing investment in forms of research to generate well-being evidence for advocacy has resulted in areas of social life, such as leisure, being sidelined (Oman & Taylor, 2018). As such the role of evidence in different policy areas demands investigation in order to further understanding of what works for well-being.
'Lived' and 'recorded' culture: contributing to well-being and leisure studies The study of well-being has evolved rapidly across the social sciences over the last three decades (Fleche, Smith, & Sorsa, 2012, p. 8), which reflects a more general trend towards the metrication of social life (Espeland, 2015), including culture (O'Brien, 2013). This, in turn, sits in a longer history of social categorisation and the measurement of everyday life (Davies, 2015;Poovey, 1998). Much of the research on quality of life, happiness and well-being begins with some debate on its definition. Arguably, this is because '[c]enturies of philosophical inquiry have failed to result in agreement about what the "good life" is' (Schoch, 2006, p. 5).
Well-being may be 'an idea whose time has come' (Bache & Reardon, 2013) in politics, yet there is no consensus of what well-being looks like, exactly. Growing emphasis on replacing economic measures of human progress, such as GDP, with something that reflected improved quality of life resulted in asking what to measure instead. In 2008, the Commission on the Measurement of Economic Performance and Social Progress (CMEPSP) was established to reflect on what 'might be required for the production of more relevant indicators of social progress [and] to assess the feasibility of alternative measurement tools' (Stiglitz, Sen, & Fitoussi, 2009, p. 7). The commission impressed that whilst tracking well-being should become an aim of all national statistics offices, each country should devise its own measures to reflect its own needs and cultural specificities (ibid, p. 18).
In the UK, the ONS' MNW programme responded to the call for each nation to begin measuring well-being, taking account of the lack of consensus on overlapping well-being definitions and disciplines by way of a national debate. 2 The then Prime Minister, David Cameron, launched the programme as 'a re-appraisal of what matters', claiming that, 'in time, it will lead to government policy that is more focused not just on the bottom line, but on all those things that make life worthwhile' (Cameron, 2010). The national debate asked people what 'the good life' means, based on everyday understandings, experiences and values, and brought those into dialogue with interdisciplinary expertise via advisory groups and literature reviews (Matheson, 2011). What was expressed in the debate therefore presented an evidence base for understanding how people experience well-being in a very ordinary way. The question is whether this resulted in a national measure of well-being that captured the relevance of all culture as it is lived, and if so, what is the role of leisure?
In 'The Analysis of Culture', Raymond Williams diagnoses 'lived culture' as one of three levels of culture (1961/1971). 'Lived culture' is 'only fully accessible to those living in that time and place'. '[R]ecorded culture', however, is that 'of every kind, from art to the most everyday facts: the culture of a period'. The third level is 'that which connects the two, the culture of the selective tradition'. Williams states the importance of 'clarify[ing] the meanings and values implicit and explicit in a particular way of life, a particular culture ' (1961/1971, p. 57). Implicit in this is the role of the selective tradition: what gets recorded and how this affects how culture is lived. As such, Williams' framework for analysing culture offers guidance to move understanding forward on the role of leisure in well-beingand what gets measured to inform policy.
Williams' work on culture makes visible the meanings and values which reside in 'What we Do', at the same time elaborating on the institutions which manage our quality of life. He draws attention to traditions in recording 'what matters', in particular how ideas of 'continuity' are determined by certain groups and formations which define the tradition. Thus, it is vital to incorporate 'analysis of elements in the way of life that to followers of the definition are not "culture" at all: the organisation of production, the structure of the family, the structure of institutions which express or govern social relationships [and] the characteristic forms through which members of the society communicate' (1961/1971, pp. 57-58). Williams is often misunderstood as proposing to raise the profile of everyday culture over elite culture. However, instead, Williams stated that 'our real purpose should be to bring all cultural work within the same world of discourse ' (1968/1989, p. 133), and, I argue, by extension, all work that involves generating knowledge for policy and what the good life might mean.
In the culture of the selective tradition, the areas of social life that are recorded become those seen as valuable, and thus essential to study further, contributing to the continuation of knowledge. The practice of developing government indicators involves the 'systematic removal of the persons, places and trajectories of the people being evaluated by the indicator and the people doing the evaluation' (Espeland, 2015, p. 56). Cultural analyses of the social sciences indicate that particular methods, like the survey, create a version of the social which is more amenable to being understood by particular techniques, thus perpetuating the value of the survey itself as a field of activity (Law, 2009). These sites of knowledge production are therefore not neutral, but performative, in that selection reproduces realities in which particular value systems are visible, and others are erased (Oman & Taylor, 2018). Government surveys are not the only cause of erasure in research methodologies, of course. For example, a recent report noted the bias towards demonstrating the value of singing to well-being over other activities (Crossick & Kaszynska, 2016, p. 108). This discursive performativity creates a reality in which funding increases for singing programmes and generates a research and evaluation economy to prove the value of this investment (Oman & Taylor, 2018), promoting a particular rendering of cultural value.
The contemporary use of culture in leisure policy has historical routes (Belfiore & Bennett, 2008). For Victorians, the arts and culture were considered 'elevating and refining to the working man' who was stigmatised as deficient (Bennett, 2000(Bennett, , p. 1414. Public cultural institutions were established 'to resolve problematic class behaviours', with Henry Cole advocating in 1884 that 'museums should go into competition with the Gin Palaces', cited in Bennett, 2000Bennett, , p. 1414, as 'the rapt contemplation of a Raphael' would keep wayward husbands from the taproom (contemporary magazine [1858], cited in Bennett, 2000Bennett, , p. 1414. This 'deficit model' of cultural participation arguably dominates contemporary cultural policy (Miles, 2013). Funding models foreground sanctioned participation in the 'arts and culture' to address perceived deficiencies in quality of life, rather than seeing the value of everyday culture (Miles & Sullivan, 2010). In the Leisure Studies literature, Stebbin's binary of 'casual leisure' and 'serious leisure ' (1997, 1999) indicates the latter is more 'important to the wellbeing of the individual and society', rather than largely non-productive leisure activities, such as 'hanging around' (cited in Blackshaw & Long, 2005, p. 248). 3 Bad choices and undesirable leisure pursuits remain a target for change, with personal and social 'happiness by design' (e.g. Dolan, 2014) dominating the discourse of behavioural and happiness economics.
In policy terms, 'casual leisure' is often demonised. For example, a description of the 1999 reversal of Bhutan's national television ban, include a story of soaring crime and drug-taking (Layard, 2007, p. 78). Richard Layard, the UK government's 'happiness tsar' explains that 'a third of parents now preferred watching TV to talking to their children', warning that the introduction of television as leisure coincided with the 'deteriorat[ion of] family relationships, the strength and safety of communities and the prevalence of unselfish values' (ibid, p. 78, 77). Bhutan was the first nation to announce it was measuring national happiness, and designed measures to demonstrate the interdependence of aspects of well-being. The index includes a whole domain called 'Cultural Diversity and Resilience', including 'native language', 'cultural participation', 'artisan skills' and 'conduct' (Helliwell et al., 2012, p. 115). Yet Layard's 2007 description of Bhutan's happiness and leisure policy is, of course, informed by the selective tradition he is familiar with that preconceives what is good leisure and what is not.
As is the case with a selective tradition, Layard's understanding of a particular leisure pursuit as bad for well-being appears to work as the understanding. This permeates into lived culture through recorded culture. Judgements made at the national level can be imposed on a personal level, and arguably reproduce effects of existing inequalities, stigmatising practices, especially those thought to be working class and 'deficient'. 4 Yet, arguments which profile the social benefits of activities, such as the arts and sport, are repeated in perpetuity, as they have become enmeshed with a generalised anxiety that culture and sportas activities which are already fundedmust demonstrate their value if they are to retain State investment. Some argue that this requirement has come about because cultural participation has hitherto been organised and rationalised as a good which has value (see Belfiore, 2006). Therefore, the culture of the selective tradition is not only visible in the study of culture, but in categorisation more broadly. Furthermore, the study of cultural activity reproduces inequality through obvious fiscal support of instrumental research and practice. Yet, as described in a paper emerging from evidence presented to the ONS in the debate, an account of well-being must be 'sensitive to important changes in wellbeing and insensitive to spurious ones. In practice, distinguishing between the two is quite a challenge and often relies on judgement based on a priori expectations' (Dolan & Metcalfe, 2012, p. 411). So, are we measuring all lived culture when measuring well-being, and what is the role of leisure?
The MNW debate is presented as an attempt to circumvent these value judgements by way of finding out what matters to people. Yet, as the next section describes, the very way in which lived culture becomes recorded (and thus valued) is based on a priori expectations which evolved from what is already measured. In the process of turning accounts of lived experience into a framework that is familiar to, and efficiently workable for, public sector statisticians, representations of what is valued in everyday life shifted the emphasis of the evidence base. The prominence of some findings over others, and the consequent effects on reproductions of 'the evidence', are described below, alongside the "qualitative analysis of survey data"? through free text to explain how this approach supports the analysis of culture and the measurement of well-being.
Free text fields: a methodology for lived culture Free text fields are most familiar in surveys as the 'Other' field where a respondent can write in their own words an answer that does not fit a predefined list. Also called open text, these fields contain rich and detailed descriptions of qualities of life, including the emotional impact of social and political phenomena on individual lives (Oman, 2015). These fields are common in paperbased and online surveys, yet survey analysis almost exclusively comprises dominant quantitative methodologies which count ticked responses to categories chosen by survey authors.
In this article, I return to the 6,787 MNW Debate questionnaire responses articulated in the free-text fields labelled 'Other'. 5 The survey comprised two versions of a questionnaire which had five main questions and a series of categories that were selected using a tick box. For example, question one: 'What things in life matter to you? Please choose all that apply' offered categories such as 'Health' and 'Personal and cultural activities, including volunteering'. Against four of the questions was an 'Other' field that enabled free text responses. There was also a final 'Any Other Comments' field. Notably, there were more free text responses to two of the four questions than ticked response to a specific ONS category. This suggests that: firstly, people often preferred describing well-being in their own words to only sticking to the categories offered by the ONS; secondly, free text responses should not be overlooked by default in survey analysis on the assumption they cannot be representative, as they provide opportunities for quantitative analysis if used this frequently.
Thus, returning to free text presents an opportunity to ruminate on gaps in the well-being evidence, and a second chance for participants to be heard. Traditional survey methodologies use random sampling techniques and the efficient analysis of large numbers, aiming for results that are representative of the population. However, what people say in free text can offer a more contextual rendering of the lived experiences of well-being. As such, they should receive due attention. Furthermore, if large numbers of people who complete surveys also use free text in them, it can make it possible to extrapolate policy-appropriate quantitative evidence through coding these data (Oman, 2015(Oman, , 2017. Despite this, free text is rarely used systematically in survey methodologies which support policy decisions. The determinants of subjective well-being differ significantly by the mode in which survey questions are delivered (Dolan and Kavestos, 2016) and the mode of response (free text or tick box) has similar effects (Oman, 2015(Oman, , 2017. 6 I re-analysed the MNW Debate free text fields, paying close attention to the values people expressed in their own words and inductively coding by hand. The most frequent theme which emerged in response to the questions 'what things in life matter to you?' was 'leisure and spare time', with 40 per cent of responses describing its importance in some way. Table 2 outlines my analysis of the free text response themes, showing the top 14. I have organised the themes by frequency of responses to question 1 to compare against the ONS' tabulation of its own categories for the same question (Table 1).
Taken at face value, quantifying the thematic re-analysis of the free text fields (Table 2) presents contradictory results to that in Table 1 from the ONS. This is despite the 'Findings from the National Wellbeing Debate' report explaining that 'time/use and personal activities' was a major theme in the wider debate (Evans, 2011, p. 25). The 'Findings' report often acknowledges the 'wider debate', citing textual responses in discussion areas, yet publishes the survey tick box responses in a far more prominent fashion, using statistics, bullet points and tables. This representation of the findings appears to have impacted on how the findings appeared elsewhere in other reports from ONS and in the media.
The sections labelled 'Key Findings' in 'The National Statistician's Reflections on the National Debate', explains that '[t]he main questions from the consultation questionnaire are listed below with the most common answers from a predefined list' (Matheson, 2011, p. 5 (2015). hierarchy appears as bullet points, presenting as if the key findings overall. This list then makes its way into media reporting of the debate, stating: People in the UK believe their well-being should be measured in terms of health, friends and family and job satisfaction, according to a report. (BBC, 2011) This rendering of what is important also finds its way into evidence in Parliament: Responses mentioned a range of factors affecting people's sense of personal wellbeing, such as health, good relationships with friends and family, job satisfaction, economic security, education, and the condition of the environment (present and future).  Table 1 of Evans' 2011 report (Self et al., 2012, p. 15) 1 , thus demonstrating that findings act as an evidence base to draw from, if only partially. 7 The different emphases across the two fields of qualitative and quantitative responses in the same survey were not acknowledged, and thus do not inform what is drawn on as the 'evidence base'. The mode of data collection (such as online survey or focus group) appears to change the headline findings (Oman, 2015), as did the way of answering (free text or tick box). This is no methodological oddity. Therefore, publishing results as 'findings' from only one response register provides an incomplete picture of what matters to people. Presentation of analyses of all modes of response should be includedand comprehensiveif there is to be an ongoing move towards governance using well-being measures which represent what matters to people, as the declared aim of the MNW Programme.
Free text: happy talk and talking about things we like to do The miscommunication of what matters for well-being requires a detailed investigation of categorisation, as selective culture. A total of 6,787 people chose to define well-being in their own words, rather than sticking to the limits of the ONS' predefined list of tick box options. This is greatly important to understanding what matters and of 'What we Do' as one of the domains of national well-being.
Policy-makers, academics, cultural producers and commentators have their own specialist registers with which to record and articulate leisure and culture and these have historically dominated debates on the relationship between 'lived culture' and 'well-being'. Attending to these 'free' data restores public voices to the discussion. Forty per cent of free text field responses to the question 'What things in life matter to you?' talked about their personal and cultural activities (using ONS terms), making 'leisure and spare time' the most frequent category in my thematic analysis of this question (  (Cases 114,159;161;167;215;241;256;320 and 336). Those descriptions which might be seen as more obviously 'arts and culture' included 'art, music and dance' (Case 258).
Respondents also used more generic descriptions. 'Sport' was used three times, with 'sports programmes' (as in provision of, rather than broadcast) once. 'The arts' (Case 140) was mentioned once, but people also talked about 'living creatively' as a spiritual pursuit (Case 262), 'beauty [and] creative ability' (Case 293), the importance of '[c]reating safe spaces for people to express themselves' (Case 268) and 'being able to interpret and interact' with the world (Case 61). Interestingly, one participant responded 'liberty, freedom, democracy and creative expression' (Case 311), suggesting they saw creative expression as a moral right in some way; another said 'fairness, prosperity, opportunity, leisure time' (Case 71).
Ordering this list of activities by their prevalence in free text is not something the ONS or a quantitative social scientist would be likely to do. For a start, there are too many activities with only one occurrence for it to offer much meaning quantitatively. However, doing so demonstrates that the value of cultural practices and leisure pastimes to well-being does not map on to the established hierarchy. People's experiences and expressions present their own scale. Few of these activities would fall under the remit of institutions which receive investment to support personal activities by way of the Department for Digital, Culture, Media and Sport (DCMS), being more 'everyday' in their nature (see Miles & Gibson, 2016;Ebrey, 2016 for definitions of how everyday participation is largely unfunded).
The way that culture is carved up to be managed ordinarily manifests through institutions and their relationship to resources from central or local government, the market or private subsidy. Yet only 0.5 per cent of those that highlighted the importance of free time to the question 'what things in life matter most?' used the word 'cultural', and both respondents referred to it as instituted within local service delivery or as an amenity, Having access to good local services at minimal cost that enrich cultural and physical life e.g. libraries, parks … (Case 114) it is important to me that community is fostered andd [sic] that collecctive [sic] concerns are given proper consideration consideration [sic] and funding. Local amenities, cultural centres such as theatres and community arts venues need encouragement and finacial [sic] aid, particularly in times of crisis. (Case 308) Case 308 was also the only description of well-being to contain any mention of a cultural institution that was not a library. Four people (one per cent) mentioned libraries specifically, once again as facilities, rather than cultural institutions, while four people also talked about heritage and the historic environment, with one explaining that their heritage and historic environment was most important (Case 115), two regarding its preservation (Cases 215 and 309) and one person describing their interest in local history and its research (Case 137).
Thus, people's expressions of free time in free text suggests that 'taking all things as equal' should be the selective tradition which would best understand 'What we Do'. These findings are corroborated by previous research into many of the pre-set categories of activities in DCMS' Taking Part Survey (TPS), 8 whether arts-related or more broadly recreational (Miles & Sullivan, 2010;Taylor, 2016). Miles and Sullivan concluded in a report to DCMS THE GOVERN: that it is participation per se that matters for health and well-being rather than participation in high rather than low culture … there is nothing intrinsically more valuable about certain types of activity. (Miles and Sullivan 2010, p. 19) Despite this, it is unlikely that investment based on contribution to well-being would find its way to the 'personal and cultural activities' cited as most important by debate participants. This section has shown that when research begins with what people say is important, rather than default categorisations, it gathers different answers. Presenting a snapshot of these practices points towards the measurement of what is simplest to understand as measurable, and public investment in what cultural elites have dictated is most important to fund. These 'states of affairs' are the product of cultural histories of governance for personal, social and national well-being that are enmeshed with an established moral system of good and bad leisure. The supposed benefits of enlightening those who would prefer to be sitting in a taproom, rather than in front of a Raphael, is a hierarchy that finds its way into the cultural sector's advocacy for its importance to well-being policy, as the next section describes.
Revisiting claims that culture is 'invisible' to policy-makers: making the case for leisure The MNW Programme affected policy sectors with emotional and professional investment in its findings. The identity of cultural professionals is often premised on their work being good for well-being (Belfiore, 2006). During the MNW programme, ONS regularly communicated its progress, and its impact on the well-being measures, by way of what they called 'releases'. Several of these releases cited the importance of 'cultural and personal activity' to MNW Debate participants and referred to DCMS' TPS as a potential source of data for the 'What we Do' domain (Matheson, 2011;Seddon, 2012b). However, the first iteration of the ONS' national well-being measures, released for further public consultation in October 2011, omitted TPS without an alternative indicator of how people spend their spare time in the domain 'What we Do'.
For many with an interest in cultural policy, the omission reinforced beliefs that this sector is of low importance to government (such as Gray & Wingfield, 2011;Stevenson, McKay, & Rowe, 2010). Culture is often portrayed as inferior to other policy sectors and this has created a common-sense history of feeling marginalised. The exception of culture from the measures was met with rallying calls to 'shout loudly about the contribution made by the arts to personal wellbeing' in the ONS' consultation (Jennings, 2011). It appears the sector did just this, with many of the consultation responses referring to the arts, specifically (ONS, 2012). Despite the fact that 'the most commonly requested additions during the consultation were measures to reflect arts, culture and sport' (Self & Randall, 2013, p. 9), the next iteration of the measures (July 2012) continued to exclude a measure for the arts and culture.
One response in The Guardian described the omission as 'pointed', explaining it resulted from culture being 'invisible' and trivial to policy-makers: Why then does the government cling to the notion that arts and culture are just about 'leisure' and 'recreation', and only happen after the real business of the day is done. This is nonsense and does not reflect lived experience. (Holden, 2012) Holden's article reinforces the cultural sector's belief that government does not understand cultural value and overlooks its importance at policy level. The author refers to culture as a way of life, like Williams. Yet, the distinction between 'arts and culture' and 'leisure and recreation' betray the fact that not 'all cultural work' is visible 'within the same world of discourse' (Williams, 1968(Williams, /1989. Given that State funding for the arts and culture doubled in the New Labour years (DCMS, 2008), it seems difficult to justify that this area of policy is invisible. Instead, the arts were foregrounded for their contribution to 'core policies including education, health, crime, regeneration and the well-being of the population at large' (DCMS, 2003). Yet, in 2004, Holden bemoaned that 'instead of talking about what they do', the cultural sector had to describe its contribution to wider policy agendas (Holden, 2004, pp. 13-14). There is widespread feeling in the arts and cultural sector that it is both sidelined from government policy on well-being, whilst simultaneously voicing reluctance to justify government expenditure through evidencing any impact it might have on this policy aim. Yet, with increased investment in the area came increased investment in evidencing cultural value.
Also in 2004, DCMS commissioned a specific survey to demonstrate the value of its investment in 'culture'. The TPS was 'a response to the growing focus on increasing engagement with culture and sport in government' (Keaney, 2008, p. 99). Thus, cultural policy history reveals a narrative which prioritises (and subsidises) arts and culture, rendering it more visible through measurement designed to show its value. In this 'culture' of the selective tradition, certain aspects of lived culture were recorded, and thus manifested in a value system of knowledge for policy, whilst others were not, and thus remain absent from the wider discourse.
Other public consultations which offer a broad evidence base to understand the value of culture see the same binary muddying analytical representations. In the 2006 Arts Debate, Arts Council England (ACE) consulted public opinion on the role of culture. Reports acknowledged that people spoke of cultural participation in broader terms of recreation, as they did not identify with the arts and culture as defined by government yet reports from the consultation tried to impress the relevance of the latter. For example: They may prefer to call their chosen art forms 'entertainment' in order to distance themselves from more conventional art forms, but there is an acceptance that their lives are touched by the arts when a broader definition is applied. (Creative Research, 2006, p. 132) The statement that people need to distance themselves from what the authors call 'art forms' is an over-simplification. Furthermore, the discourse that participants should 'accept' their lives are touched by art hardly advocates the value of work funded by ACE over those seen as entertainment. While there is much evidence of inequalities of cultural consumption (for example, Warwick Commission, 2014; Taylor, 2016), there is also work which evidences that people have little interest in the arts and actively decide to not participate (see, for example, Miles, 2013). However, the 'deficit model of participation' that is central to most cultural policy (Miles & Sullivan, 2010, p. 19) assumes people should 'accept' the State-funded cultural assets available to them. Those who don't are problematically called 'non-users', rendering these forms of participation 'hidden', even 'ghostly engagements' (Miles, 2013, p. 186). From a discourse analysis perspective, the erasure of forms of leisure is, in itself, a 'cultural practice'in that removing some 'cultural work' is, the 'selective tradition ' Williams diagnoses (1961' Williams diagnoses ( /1971. To counter the value narratives of cultural policy, the next section returns to the matter of how people expressed what mattered to them about 'What we Do' in free text. The matter of 'lived culture' expressed as leisure Whilst economic security is important, and everything to do with income and employment -I don't think these economic-related indicators should be the key focus of measuring national well-being. I think this exercise should break away from the traditional areas that already measured excessively, and focus instead on measuring the areas that genuinely make people happysocial connections, participating in local community and personal activities, having access to green open space for a healthy walk with your family/friends, having cheap/no cost access to cultural activities that lift the spirit e.g. libraries, parks, physical activity. (Case 199, Questionnaire 2: Question 5, 'Any other comments') The above free text account makes explicit the relationship between well-being and Williams' 'lived culture'. While previous sections have engaged with the coded dataset, here I centre on one participant in greater detail. This respondent does not stand out as saying something different from others. Rather, they echo the feelings of other survey participants. For example, the participant (Case 199) who stated that they would 'hate for someone to have to worry about having enough to eat or somewhere to live', but that 'doing other things including recreation activities improves your mental well-being which in turn affects your general well-being but you can only do those things if you are in a good financial position'. Here, it is clear that 'Personal Finance' does not stand-alone from 'What we Do', with many seeing the former as facilitating the latter.
The quality of leisure and spare time is particularly important: how, where and with whom they are spent, and the availability of a variety of places to enjoy oneself. Case 199 is one of many examples of free text fields that was coded using multiple themes because the participant described so many aspects of well-being. In this case, these included: 'security', 'financial stability', 'family', 'friends', 'spare time' and 'access to services' (see Figure 2 for all the themes).
Case 199 also drew a comparison between 'domains' of life which are 'measured excessively' and those which 'genuinely make people happy'. The respondent clearly categorised the practice of economic measurement as unwanted and unwarranted, while social, personal and cultural participation are described as sincere articulations of positive well-being effects. They reference a culture of the selective tradition in which the economic is valued, and thus recorded, while 'lived culture' is not.
Case 199 suggests that the MNW Programme should be a moment of change in the ordering of practices, thus echoing the promises of the National Statistician and Prime Minister at the launch of the MNW Programme to 'measure what matters' based on what people say. As such, they speak the ONS' discourse back at the survey authors. The language 'mirrored' the ONS' questionnaire and tick box options. The participant used the 'Other' field to not just elaborate upon the options presented by the ONS, but, arguably, to articulate the importance of some domains over others. In the ONS' analyses, the importance of a tick box was decided by the number of times they were ticked, not by any ordering of the respondent. This respondent adopted ONS language in free text recalling Williams' diagnosis of lived culture presenting new meanings that demand analysis.
The wilful and skilful mirroring of language in Case 199's intervention may not have been equally available to all participants. Other free text entries were expressed far more simply, yet, as the 'Findings from the National Debate' acknowledged, the wider debate offered many opportunities for joined-up reflections into policy decisions and predictions (2011, p. 6). Case 199's description of contributors to well-being is testimony to the value of recreation and access to services and activities which facilitate that. Arguably, such findings indicate several policy priorities, which should include the preservation of public space from private development, as well as subsidised cultural resources, including parks and libraries.
The current indicator of leisure and spare time covered by the MNW measures is restricted to 'engagement in arts and culture', which was later included in the national well-being measures in 2013. This in turn is based on an indicator which originates from DCMS' TPS (DCMS, 2016). Yet this indicator does not measure this indicator does not capture attendance at all the activities that TPS measures, only attendance at an art gallery, museum or heritage site in the last 12 months.
Academic well-being literature does discuss leisure (Haworth, 2010;Jeffres, Bracken, Jian, & Casey, 2009); however, it tends to focus on specific activities, particularly those which might fall under 'arts and cultural engagement' (Galloway, 2008). Spare time, free time, and recreation for enjoyment, however, have been largely neglected in literature which claims to inform policy (see, for example, Stoll, Michaelson, & Seaford, 2012). Even Leisure Studies tend to use the familiar lenses of arts, culture and sport (Blackshaw & Long, 2005). Therefore, there does seem to be a hierarchy of diminished visibility for cultural practices outside of 'the arts', and thus broader ideas of leisure get sidelined. What is clear is that both contingents of research for policy on 'lived culture' and well-being are clearly wedded to what they [the experts] do in such a way that 'What we Do' as an area of research remains obscured.

Conclusion
This article presented secondary analyses of qualitative data from the ONS' MNW programme . The National Statistician explained that the MNW Debate was designed to understand what is valuable to well-being for people (Matheson, 2011), thus bringing it into the same 'world of discourse' as experts (Williams, 1968(Williams, /1989). Yet reanalysis of qualitative debate data and policy documents calls into question the degree to which the ONS measures well-being as 'that which makes life worthwhile' (Cameron, 2010), or whether 'the bottom line' was to find an efficient way to present the most appropriate-looking statistical measures.
The UK Prime Minister stated it was important to listen to what people said mattered to 'put instincts to the test' (Cameron, 2010). The public debate generated 34,000 responses of primarily qualitative data, yet it is quantitative data that was and remains most readily presented as 'evidence'. Not all debate responses are available for analysis. However, reanalysis of the free text fields demonstrated that the mode of response affects headline results of 'what matters most' to people. In presenting this reanalysis, this article demonstrated that leisure is more important to the nation than is apparent in ONS reports or where the MNW debate is cited as evidence in Parliament and the media.
Using Raymond Williams' framework of 'lived culture' and 'recorded culture', the article interpreted two examples of the 'selective tradition' that contributed to the analytical discrepancy in presentations of the MNW evidence (1961/1971, p. 66). It firstly addressed methodological approaches to survey design and analysis, discovering a bias towards certain aspects of cultural and social life appearing as central to well-being. Long-form descriptions of well-being in the qualitative data are subjective and not easy for experts to order. When analysed, free text may also outside of the categories anticipated in the deductive coding frame. Thus, expert methods fall of ordering the world are such that people's descriptions of 'what matters' become 'spare' to evidence which might inform governance. As Williams argues, the organisation of disparate meanings, values, and practices into a dominant cultural and social order happens by way of cultures, traditions and institutions (Williams, 1977) and methods have been perfected to reproduce and naturalise modes of ordering (Law, Ruppert, & Savage, 2011). This article considered values and preferences, as expressed by the public in the debate, those representing the MNW programme, and those representing the cultural sector and found they differ.
Cultural sector advocacy prioritised the role of the arts and culture to well-being and is another example of the selective tradition. However, debate respondents talked of leisure pursuits, and more 'ordinary' cultural practices, but rarely referred to the arts and culture. In bringing all forms of culture 'into the same world of discourse' (Williams, 1968(Williams, /1989, the role of the selective tradition in cultural policy also emerges. Claims to the 'special case' of policy areas (such as culture), are arguably detrimental to the evidence base in their neglect of the broader context of well-being value (which includes leisure).
Reviewing how the 'selective tradition' transforms a broad category of lived culture into a naturalised hierarchy that portrays the value of the arts over recreation also reveals how this manifests as a moral value that is reproduced through policy decisions and investment. Similarly, in looking at the ways in which policy-makers and public-sector statisticians are accustomed to presenting, translating and consuming evidence, it is easy to see the habituation involved in ordering findings in such a way as to foreground certain forms of data over others. It is not the aim of this article to argue the value of qualitative approaches over quantitative ones. Similarly, it is also not the aim of this article to argue the value of one form of culture over another. Williams' attempts to conceptualise these aspects of culture have in fact led to some misunderstanding that he was endeavouring to do so. In returning to the well-being data, and the work of Williams, this article has revealed processes, politics and pragmatics involved in knowledge production for cultural and well-being policy-making.

Notes
1. According to the National Statistician, 'In total the debate generated 34,000 responses, some of which were from organisations and groups representing thousands more' (Matheson, 2011, p. 2). The figure 34,000 appears repeatedly, but when greater specificity is described, it is often contradictory. The live events, for example, involved 2,750 people (Self et al., 2012, p. 2) and 7,249 people (Evans, 2011, p. 19). Oman (2015) contains more detailed discussion of how the debate was represented. 2. Some debate participants remarked that what constituted 'the nation' was unspecified (Oman, 2015). 3. For a longer discussion on these moralising aspects of leisure studies, please see Blackshaw and Long 2005. 4. There is much work which addresses the issues of class, geopolitics and stigma, that there is no room to repeat here. Key texts include Skeggs and Loveday, 2012;Bennett et al., 2009. See also Tyler's, 2018 special issue of the Sociological Review. 5. See Oman (2015) for clarification on how 'Other responses' are distributed across the two versions of the questionnaire. pp. 71 and 72. Questions 1 & 2 saw a higher response rate using tick boxes, while later questions preferred free text. 6. Survey respondents in the national debate were able to tick many responses and write in free text. However, data were not stored in a way allowing for more detailed analysis. See Oman (2015) for more detail of how free text was used more than tick boxes. 7. It actually refers to another report: Seddon (2012a), which cites the statistic back to Evans (2011) on p. 18. 8. TPS originated in 2005 and is the main evidence source for DCMS and its sectors. The survey's main objectives are to: provide a central, reliable evidence source that can be used to analyse cultural and sporting engagement, providing a clear picture of why people do or do not engage.