Going solo: lifelong nonparticipation amongst the NCDS cohort

ABSTRACT Some Western societies, it has been claimed, are experiencing an unparalleled downward trend in participation with manifold grave consequences predicted. In the UK, for instance, politicians and commentators, arguably influenced by Robert Putnam’s warnings of a collapse in community, have spoken of Britain’s broken society and disintegrating social ties with opting out, or nonparticipation, presented as a pressing social problem. Set against this background, and engaging directly with Putnam’s thesis, we explore the scale, characteristics and causes of an ‘extreme’ variant of nonparticipation – lifelong nonparticipation – amongst members of a national birth cohort, the UK’s National Child Development Study (NCDS) (1958). Joining structured survey data collected over the lifecourse, with biographical interview data collected from cohort members at age 50, we identify lifelong nonparticipation as a minority disposition associated with distinctive demographic traits being, for example, highly gendered and related to lower educational attainment. In terms of causes, time pressures arising from work and caring duties or, more precisely, the feeling of being ‘pressed for time’, appeared critical. The implications for policy and practice are considered.


Introduction
America, Robert Putnam (2000) famously argued, is experiencing a collapse of 'community', its citizens voting, giving, joining, attending church and socialising less and less. All across the country, and in all parts of society, a trend towards individualism, quickening with each generation, was identified. Replacing dense, multi-stranded, well-exercised social ties built on principles of cooperation and mutual support were weak, single-stranded, special-purpose, fluid relationships orientated around self-interest (Putnam, 2000, p. 184). This decline in participation in social and civic life, both in scale and in intensity, had, Putnam (2000, pp. 27-28) claimed, grave and far-reaching consequences affecting everything from health, happiness and well-being to education, the economy and neighbourhoods; our very democracy even, was said to be at risk. While Putnam's assertions about deep and widespread declines in participation have been disputed by some (see Paxton, 1999;Rotolo & Wilson, 2004;Stolle & Hooghe, 2005), less contested have been his claims about the costs and negative effects of disengagement in social and civic life. There is, for instance, growing evidence of a relationship between lack of participation and health (Pollack & von dem Knesebeck, 2004). Focusing on individual or multiple participation behaviours, studies have found associations between increased mortality and lack of active participation in organised groups and associations (Dalgard & Lund, 1998), between increased functional dependency, higher levels of depressive symptoms, poorer self-rated health and lack of participation in volunteering (Morrow-Howell, Hinterlong, Rozario, & Tang, 2003); between poorer self-rated health and lack of participation in volunteer projects, religious groups, the meetings of formal and informal groups (Lee, Jang, Lee, Cho, & Park, 2008), clubs and religious services (Veenstra, 2000) and between higher levels of depressive symptomology and lower levels of social leisure engagement (e.g. taking part in social, religious or recreation activities) (Duncan, Withers, Lucier-Greer, Ferraro, & Reed-Fitzke, 2017). Responding to concerns about the consequences of nonparticipation, such as those voiced by Putnam (2000), and noting accumulating evidence on this point (Pollack & von dem Knesebeck, 2004), we explore here lack of involvement in three participation contexts amongst members of a national birth cohort, the UK's National Child Development Study (NCDS) (or 1958 Birth Cohort) (Power & Elliott, 2006).
The NCDS provides a rich longitudinal dataset allowing individuals' participation behaviours to be tracked over time and situated in a whole life context. Recently, Bowling, Pikhartova, and Dodgeon (2016) used this data to explore associations between past participation in clubs/groups/ associations and cognitive abilities at age 50 and found that less participation was associated with poorer cognitive outcomes. Relatively few studies have, however, adopted Bowling et al.'s (2016) concern for participation across the lifecourse, yet by attending to it a deeper appreciation of participation propensities, behaviours, barriers and outcomes can be achieved (Bowling et al., 2016;Brodie et al., 2009). Participation over the lifecourse formed a particular concern for Putnam. He argued that a 'long civic generation', a cohort of people born between 1910 and 1940 that had been more engaged in social and civic affairs throughout their lives than either previous or successor generations, was dying out, and was being replaced by waves of 'civicdropouts', individuals who, throughout their lives, participated far less in such affairs.
Besides allowing the study of participation over time, the longitudinal data of the NCDS makes it possible to identify and study individuals demonstrating particularly intriguing or unusual longterm patterns of participation. For example, and of interest here, it enables a focus on individuals that Putnam might be expected to understand as pronounced 'civic dropouts', individuals who, as adults, have never participated in multiple aspects of social and civic life, people we term 'lifelong nonparticipants'. Studying such extreme or deviant cases can be especially enlightening, offering insights potentially relevant to more typical cases (Patton, 1990). With this in mind, we consider here the scale, sociodemographic characteristics and causes of lifelong nonparticipation in clubs/ groups/associations, volunteering and religious meetings amongst the NCDS cohort. To do this, we combine NCDS data with data from an associated biographical interview study completed with a subset of the NCDS cohort, the Social Participation and Identity Study (Elliott, Miles, Parsons, & Savage, 2010;Elliot, Savage, Parsons & Miles, 2013). In presenting our findings, we engage directly with Putnam's claims about the scale and causes of nonparticipation. These claims, in addition to broader findings on the scale, correlates and causes of nonparticipation in, particularly, associations, volunteering and religious meetings, are discussed in the next section. Subsequent sections present our methodology and findings and reflect on the implications for policy and practice.

Nonparticipation
Data on nonparticipation in social and civic life are typically culled from studies on participation in specific contexts, such as in organised sports, volunteering and religious groups. These data show that rates of participation and, by extension, nonparticipation, vary markedly between types of activity and organisation. For instance, nonparticipation in voting (Verba & Nie, 1987) is typically lower than nonparticipation in volunteering (Cnaan, Jones, Dickin, & Salomon, 2011). Putnam (2000, p. 186) drew on data describing participation in a range of contexts, including religious meetings, organised clubs, groups and associations, volunteering, political activities and workplace groups to show high levels, and growing rates, of nonparticipation, indeed the emergence of an 'anticivic epidemic'. For example, he reported that in the mid-1970s two-thirds of Americans attended club/group/association meetings but by the late 1990s two-thirds did not; that in the mid-1960s the average American devoted 3.7 h a month to clubs/groups/associations but by the mid-1990s this had fallen to 2.3 h and that while on an average day in the mid-1960s 7% of Americans spent some time in community organisations by the mid-1990s this was just 3% (Putnam, 2000, pp. 61-62).
Moving beyond individual participation contexts, and influencing the approach we adopted, some scholars have considered involvements across several domains and, by extension, provide more comprehensive accounts of both participation and nonparticipation. Putnam (2000, p. 465) measured 'regular community involvement' understood as weekly attendance at religious meetings and monthly attendance at club meetings. He found that whereas 22% of Americans demonstrated 'regular community involvement' in 1975, this had fallen to 9% in 1999, while the proportion who never attended either church or club meetings had risen from 11% to 20% (Putnam, 2000, p. 465). Harrison and Singer (2007, pp. 55-56) measured participation in 'passive activities' (e.g. socialising with a neighbour, attending a parents' evening at school, visiting a place of worship), 'active engagement' (involving charity work or the organisation of community events), 'community-based engagement' (e.g. engaging in a public consultation) and 'political activities' (e.g. protesting, writing to a newspaper, canvassing for an MP). They found that over a third of the population could be defined as 'community bystanders'individuals who were less likely to participate and who engaged least in all the activities considered (Harrison & Singer, 2007).
Nonparticipation (and participation) in social and civic life has been associated, although not always consistently, with certain psychological and attitudinal variables including morality, empathy, efficacy (Smith, 1994;Wymer, 1997), altruistic attitudes and sense of civic duty (Smith, 1994). It has also been associated with multiple sociodemographic variables (Brodie et al., 2009), education perhaps most consistently, but also ethnicity, age, gender, income, socioeconomic background and health (Jones, Millward, & Buraimo, 2011;Plagnol & Huppert, 2010;Rai, 2008;Verba & Nie, 1987). Generally speaking, studies find that the young, women, those with non-white backgrounds, less education, lower incomes, poorer health and lower socioeconomic status tend to be overrepresented amongst those who do not participate (Brodie et al., 2009;Jones et al., 2011;Putnam, 2000;Verba & Nie, 1987); however, the demographic profile of nonparticipants varies across participation contexts (Brodie et al., 2009). For example, women are less likely than men to participate in sport through clubs/teams/groups , whereas men are less likely than women to be church members and to attend church-related activities (Levin, Taylor, & Chatters, 1994). Various theories have sought to explain the observed relationships. Some posit that these factors make participation less easy, or limit the likelihood of individuals becoming either aware of participation opportunities or being invited to participate (Bussell & Forbes, 2002;Ellis, 2003;Musick & Wilson, 2008;Rai, 2008).
Working patterns and caring duties have consistently been identified as important influences on participation and, by extension, nonparticipation, with both affecting the time individuals have available to dedicate to groups and activities (Brodie et al., 2009;Bussell & Forbes, 2002;Musick & Wilson, 2008;Wilson & Musick, 1997). Surveys often find that lack of time is the most commonly cited reason for nonparticipation, and the main factor limiting participation amongst those who do take part (Brodie et al., 2009;Jones et al., 2011). Putnam (2000, p. 191) identified time pressures, which he grouped together with economic pressures, as one of four key reasons for declining rates of participation, noting that for full-time workers, the better educated and dualcareer families the 'time bind' formed a real impediment to community involvement. However, he also noted that evidence suggests that heavy time demands can be associated with more rather than less social and civic engagement, that employed people participate more than unemployed people for example and that people who report feeling rushed appear to forgo time spent 'doing nothing' and watching TV rather than time spent participating (Putnam, 2000, p. 192). The other three factors he identified as participation inhibitors were suburbanisation resulting in increases in long solitary commutes, social segregation and community fragmentation, the impact of TV watching on leisure time and, most importantly, generational change. On this latter point, and as noted, he argued that waves of 'post-civic' cohorts, which had never acquired a participatory disposition, were progressively replacing a 'long civic generation' that had always taken part. Presented as explanations for declines in participation, and discrete impediments to involvement, we believed it interesting and topical, given the continuing socio-political currency of Putnam's analysis, to explore the degree to which these four factors could explain long-term nonparticipation amongst the NCDS cohort. Consequently, we reflect on these factors when presenting our findings.

Methodology
To investigate lifelong nonparticipation in clubs/groups/associations, volunteering and religious meetings amongst the NCDS cohort, we integrated two linked secondary datasetsa quantitative dataset from the NCDS and a qualitative dataset from the associated Social Participation and Identity Study (SPIS) (2008). Ethics approval for the study was provided by a University Research Ethics Committee (reference number ERN 09-256).
The NCDS is a rich source of structured, longitudinal survey data. It began as the Perinatal Mortality Survey collecting data on some 17,000 babies born in 1 week in March 1958 in England, Scotland and Wales (Power & Elliott, 2006). Originally, it focused on the social and obstetric factors associated with stillbirth and death in early infancy (Parsons, 2010). Since then, it has gone on to collect over nine data collection sweeps and one dedicated biomedical sweep, data on members' physical and mental health, health-related behaviour, sociodemographic characteristics, employment, income, education, housing and attitudes (see Brookfield, Parry and Bolton (2018) for detailed information on the NCDS). Relevant to our interests, it has also consistently collected data on participation in clubs/groups/associations and religious meetings. Further, in the adult sweeps, it has twice at the time of writing collected data on volunteering, in sweeps 4 and 8 when members were aged 23 and 50.
The SPIS (2008) was administered by the NCDS research team. It was timed to coincide with the NCDS's eighth data collection sweep, completed when cohort members were aged 50. Adopting a stratified sampling approach, 220 geographically dispersed and socially diverse participants were recruited from the NCDS to the SPIS. Participants in this sub-sample took part in a face-to-face qualitative interview, steered by a topic guide. The interviews were audiorecorded and transcribed, with the transcripts made available for secondary analysis. The topic guide covered identity, life history, neighbourhood and belonging, family and friendships and the experience of being part of the NCDS (Elliott et al., 2010). Relevant to our interests, it also explored spare time interests and activities, involvements in clubs, organisations, groups and political parties, religious participation, plus participation in giving, volunteering and charitable work (Elliott et al., 2010).
Steered by Patton's (1990, p. 170) argument that 'more can be learned from intensively studying extreme or unusual cases than can be learned from statistical depictions of what the average case is like', we purposively sampled the SPIS interview transcripts for a set of cohort members who presented in their NCDS data records 'extreme', yet particularly policy relevant, participation narratives. Analysis of the NCDS data directed, then, the SPIS data selected for study. Of interest here, we sampled the interview transcripts associated with every individual (n = 21) in the SPIS who, in every adult NCDS data collection sweep to which they responded, reported not being members of and not joining in with social, leisure, sports, community, interest, political and religious clubs/groups/associations and not attending, with any regularity, religious meetings. This encompassed data from sweeps 4 to 8, sweep 9 not being available at the time of our study. In addition, in the sweeps where it was explored, these individuals reported not taking part in volunteering. We termed these individuals 'lifelong nonparticipants' and they form the focus of this article. In addition, we sampled the interview transcripts associated with every individual (n = 20) included in the SPIS who always reported participating in clubs/groups/ associations or regular (monthly) attendance at religious meetings, and a random sample (n = 8) of transcripts associated with cohort members who reported frequent participation in clubs/ groups/associations, or volunteering, at age 50. We termed these individuals, respectively, lifelong participants and frequent participants. Finally, although not considered here, we sampled the interview transcript associated with the only cohort member included in the SPIS who presented in the NCDS data as unemployed at age 50. Table 1 provides further information on the sampling criteria. To maximise the number of cases available, we did not exclude cohort members who presented incomplete NCDS data.
The quantitative NCDS data were analysed in STATA (Hamilton, 2012) and Excel. Personalised timelines detailing the participation commitments of each member of our sample at each data collection point were created. At the group level, descriptive statistics were developed to identify headline sociodemographic characteristics (e.g. gender, education) by participant 'type'. We adopted a pairwise deletion approach to handle missing data (Kang, 2013). An inductive thematic analysis (Pope, Ziebland, & Mays, 2000) was performed in NVivo (Bazeley & Jackson, 2013) by the first and second authors on the qualitative interview transcripts. Analysis began with open coding on a handful of the transcripts (Corbin & Strauss., 2008). Completing this analysis in tandem, the two authors developed an initial coding scheme that subsequently steered analysis of the remaining transcripts. As analysis of these transcripts progressed, additional codes identified within the data were, in discussion, added to the scheme. To ensure inter-and intra-coder reliability, the authors regularly reviewed and critically discussed their own and one another's analysis. Again working in tandem, the authors progressively combined codes to form categories and combined categories to form broad themes. The broad themes included types of participation, barriers to participation and reasons for not participating, benefits of and motivations for participating and routes into participation. We turn now to the results of these analyses. All names are pseudonyms.

The scale and characteristics of lifelong nonparticipation
Just 21 cohort members, or 9.5% of the SPIS sub-sample, emerged in the NCDS survey data as lifelong nonparticipants. However, analysis of the SPIS interview transcripts associated with these individuals indicated that just seven, equating to only 3% of the SPIS sub-sample, were 'genuine' lifelong nonparticipants. Most, 14 in fact, of the 21 NCDS-identified lifelong nonparticipants actually emerged in the SPIS data as occasional, past, ad-hoc and informal participants. Explored in more detail later, these individuals reported attending exercise classes and social clubs, helping- Table 1. Types of participation.

Type of participation
Sampling criteria N

Lifelong nonparticipation
Reported in every adult NCDS data collection sweep responded to not being a member of and not joining in with social, leisure, sports, community, interest, political and religious clubs/groups/ associations and not attending with any regularity religious meetings, and, reported in the sweeps in which it was addressed, and which were responded to, not taking part in volunteering.

21
Lifelong participation Reported in every adult NCDS data collection sweep responded to being a member of and joining in with social, leisure, sports, community, interest, political or religious clubs/groups/associations or attending religious meetings at least monthly.

20
Frequent participation Reported in sweep 8 at age 50 membership of, and joining in once a week or more with, at least three social, leisure, sports, community, interest, political or religious clubs/groups/associations or volunteering at least once a week.

8
The only unemployed individual included within the SPIS was also included in the sample.
out neighbours, participating in sports clubs and neighbourhood groups and assisting in groups and activities associated with their children. These diverse forms of participation in social and civic life had been 'missed' by the NCDS survey instrument that, as shall be noted, has habitually focused on 'formal' participation taking place in membership-based groups. 'Genuine' lifelong nonparticipation, as demonstrated by the seven individuals identified as lifelong nonparticipants by the SPIS data and the NCDS data, individuals we shall term 'SPISidentified lifelong nonparticipants', presented distinctive sociodemographic traits (Table 2). These matched traits commonly associated with 'general' nonparticipation but differed in some crucial respects (Harrison & Singer, 2007;Smith, 1994;Verba & Nie, 1987).
Lifelong nonparticipation was highly gendered. Almost 90% of the seven SPIS-identified and almost 70% of the 21 NCDS-identified lifelong nonparticipants were female. Interestingly, lifelong participation was also highly gendered with 75% of the lifelong participants being female. We believe this was a consequence of including regular religious participation in our sampling criteria since this form of participation tends to be associated with women (Levin et al., 1994). In contrast, the frequent participants were a gender-balanced group. Educational attainment was related to lifelong nonparticipation. Of the three types of participants studied, lifelong nonparticipants were the most likely to report no qualifications. Approximately 29% of the SPIS-identified and 14% of the NCDS-identified lifelong nonparticipants reported no qualifications. In comparison, just 5% of the lifelong participants and none of the frequent participants reported no qualifications.
Relationships between nonparticipation, general health and socioeconomic status, often identified in previous research Smith, 1994), were not clearly evident in respect of lifelong nonparticipation signalling, perhaps, that it has distinct predictors. For example, at age 50, and based on a cohort member's own classification, the lifelong nonparticipants were the least likely of the different types of participants studied to be located in the bottom two classes (IV and V) of the Social Class based on Occupation classification (previously the Registrar General's Social Classification). Looking at social class at birth, with this being based on the classification of a cohort member's mother's husband (Centre for Longitudinal Studies, n.d), the SPIS-identified lifelong nonparticipants were no more likely than the frequent or lifelong participants to be located in the bottom two classes. In terms of general health, lifelong nonparticipants were not especially likely, nor any more likely than the other types of participants studied, to report poor general health at age 50. None of the SPIS-identified, and just 5% of the NCDS-identified, lifelong nonparticipants reported poor health at this age. That being said, suggesting a more ambiguous relationship with health, lifelong nonparticipants were the most likely to be identified as 'sick' in relation to economic activity at age 50, and were the most likely to demonstrate 'poorer' mental health. Some 67% of the SPIS-identified, and 39% of the NCDS-identified, lifelong nonparticipants had a Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS) score of ≤44 at age 50, indicating a greater than 80% likelihood of psychological distress as defined by the Centre for Epidemiological Studies Depression Scale (CES-D), a clinically validated measure of depression (NHS Health Scotland, 2015, p. 14). For the frequent and the lifelong participants, the equivalent figures were 13% and 11%, respectively.

Causes of lifelong nonparticipation
For the seven SPIS-identified lifelong nonparticipants, confirming past research (Low, Butt, Ellis Paine, & Davis Smith, 2007), and supporting Putnam's (2000) theory about the importance of time pressures, the pressures of work, particularly the impact of long hours and the effects of work-related tiredness, emerged as the most commonly cited reason, put forward by six individuals, for not joining in. Two individuals who ran their own businesses spoke frequently of the amount of time their work took up; 'it's probably a bit over the top but I prefer to be in control, so I tend to take on more than, you know. I'm not very good at delegating' (Karen, SPIS-identified lifelong nonparticipant). Again reflecting previous findings (Brodie et al., 2009), and again supporting Putnam's (2000, p. 191) theory about the impact of the 'time bind', caring duties, which encompassed caring for children, older parents, even pets in the case of one individual, and which had evolved over cohort members' lives according to family formation and development, constituted further demands on time that impeded involvement in activities outside the home. For example, Dilys', in addition to working part-time, cared for dependent children and provided daily care to both her mother and mother-in-law. The combination of paid work and care produced practical, intractable barriers to her participation, 'I just don't have time' (Dilys, SPISidentified lifelong nonparticipant). Interestingly, however, our analysis of NCDS data revealed that the lifelong nonparticipants were, of the three types of participants studied, the most likely to be childless and were not especially likely to report caring for elderly parents or in-laws.
Amongst the 21 NCDS-identified lifelong nonparticipants, the pressures of work emerged as the main reason for little or no participation. Long working hours, a long commute, connecting to Putnam's (2000) theory about the impact of suburbanisation on participation, working unsociable hours and/or shift work were all said to limit opportunities, and an individual's willingness, to participate. However, the emphasis placed on work by this group, and the SPIS-identified lifelong nonparticipants, is intriguing as our analysis of NCDS data revealed these individuals resembled the lifelong and frequent participants in terms of working patterns. Working full-time, in either an employed or a self-employed capacity, was the most common working pattern across all types of participants. Finding that the most intensely involved individuals were able to successfully weave participation, work and caring duties into their lives, while the least engaged individuals identified work and caring commitments as insurmountable barriers to participation, yet presented no greater and quite often fewer responsibilities in these areas, might suggest that time pressures were experienced differently by our three types of participants. Moreover, it might suggest that it was these differing experiences of feeling 'pressed for time', rather than time pressures per se, that structured participation behaviours. Research by Southerton and Tomlinson (2005) suggests that there are indeed differences between people in how time pressures are experienced. They found, for example, that women are more likely than are men to report feeling 'pressed for time' (Southerton & Tomlinson, 2005). The highly gendered nature of lifelong nonparticipation might help explain, then, the emphasis placed on 'lack of time' when the SPISidentified and NCDS-identified nonparticipants described their reasons for not taking part.
Relative to time pressures stemming from work and caring duties, other factors appeared less important in explaining, at least from the point of view of our lifelong nonparticipants, nonparticipation. For the SPIS-identified lifelong nonparticipants, a lack of interest in, or preference for, 'joining in' was cited by four individuals: 'I prefer, I'm not a sociable sort of person' (Morgan, SPIS-identified lifelong nonparticipant). Pertinent here, 83% of the SPIS-identified and 44% of the NCDS-identified lifelong nonparticipants scored below one-half SD of the sample mean score (i.e. below 'average') for extroversion, determined using International Personality Item Pool (IPIP) items. In comparison, just 25% of the frequent and 24% of the lifelong participants did. Moreover, while none of the SPIS-identified and just 11% of the NCDS-identified lifelong nonparticipants scored above one-half SD of the mean (i.e. above 'average' on extroversion), 41% of the lifelong and 50% of the frequent participants did (Table 2). Supporting findings from the SPIS interviews, findings from the NCDS data suggested, then, that in comparison to the frequent and lifelong participants, the lifelong nonparticipants were a noticeably less extrovert, less 'sociable' group. Financial anxieties, a factor inhibiting participation for Putnam (2000), helped explain nonparticipation in two SPIS-identified lifelong nonparticipants. Of note, the interviews were completed at the time of the 2008 Global Financial Crisis, the repercussions of which were affecting some cohort members: 'we've felt the credit crunch crunching over the last few months, so we probably do stay at home a lot more' (Alison, SPIS-identified lifelong nonparticipant).
For the NCDS-identified lifelong nonparticipants, a wider range of factors were presented as explanations for limited participation including a preference to spend time at home, a lack of interest in 'joining in', health problems and, for a couple, an active dislike or suspicion of groups: 'these associations, everybody wants to be in charge, in control. So I stick to my own devices, it's easier' (George, NCDS-identified lifelong nonparticipant). Brodie et al. (2009) noted that suspicion and lack of trust might deter participation in some people. Looking beyond self-reported reasons for nonparticipation, while engaging with the two reasons Putnam (2000) suggested were most responsible for declining participation, TV watching and generational change, the SPIS data indicated that the most disengaged individuals differed to the most engaged individuals in crucial respects on these matters. When asked about their freetime activities, the SPIS-identified lifelong nonparticipants were more likely to mention watching TV than were the frequent participants. Four out of seven of these nonparticipants mentioned watching television in their free-time: 'so I work quite long hours, so I tend to flop when I come in. . .so it's mostly watching television' (Karen, SPIS-identified lifelong nonparticipant). In contrast, just one of the eight frequent participants mentioned watching television. When not participating, these most involved individuals seemed to continue to be engaged in somethinghousehold chores, talking to a partner, getting ready for work, studying and so on. However, the volume of information provided within the SPIS interviews by the most involved individuals on free-time and participation activities seemed, at times, to lead to less probing on privatised leisure activities affording less opportunity, perhaps, for TV watching to be mentioned. The lifelong participants were more likely to mention television watching than the frequent participants, but their viewing seemed infrequent and selective. This contrasted to the NCDS-identified lifelong nonparticipants who appeared to be more regular, less selective viewers: 'I'll watch whatever's on the box' (Janet, NCDS-identified lifelong nonparticipant).
The impact of generational change was, for Putnam (2000), the single most important factor behind declining participation. The SPIS interviews allow some consideration of this issue at the level of the individual. Although not a required question, some interviewees were asked how their interests compared to those of their parents, individuals liable to be drawn from Putnam's 'long civic generation', while some interviewees' volunteered information about their parents' participation commitments. Two of the five SPIS-identified lifelong nonparticipants who provided relevant information described parents with solitary interests like gardening and collecting. Further, three of the five depicted disengaged parents: 'I don't have a, not a great deal of interest, but my parents had even less. . .[my mother] was very much, her role is to look after her husband and that was that, so she didn't have any life outside looking after my father' (Dilys, SPIS-identified lifelong nonparticipant). In contrast, the more socially involved individuals appeared to have more socially involved parents. Only one frequent participant provided relevant information but reported that both parents had been members of the Labour Party. Amongst the lifelong participants, just over half provided relevant information and, in almost all cases, parents were described as participants, to some degree, in social and civic life, typically the church but also social clubs, volunteering, music circles and choirs. Providing some counter-evidence to Putnam's (2000) claims of generational change, these data, limited though they are, indicate continuity rather than difference between the generations in participation practices.

'Below the radar' participation
Most (n = 14) of the NCDS-identified lifelong nonparticipants emerged, as mentioned, as informal, ad-hoc or past participants in the SPIS interview transcripts. For multiple reasons, including question wording and order (see Brookfield, Parry and Bolton 2018 for a full discussion), the NCDS appeared unsuited to capturing informal, low-level, ad-hoc participation. These forms of participation occurred, in respect of the NCDS, 'below the radar'. The SPIS interviews, again for various reasons, including a semi-structured format that allowed cohort members to provide responses in their own words (see Brookfield, Parry and Bolton (2018) for a full discussion), were able to capture these varieties of participation.
For those individuals presenting 'below the radar' forms of participation, it appeared that selfinterest, direct or indirect personal gain -'selective incentives' to use Mueller's (1975) termand personal satisfaction or enjoyment, usually prompted, and captured what was gained from, participation. For Putnam (2000), and commentators like Bellah, Madsen, Sullivan, Swidler, and Tipton (1985), it was precisely this type of self-orientated participation that was replacing traditional, close-knit social ties built on principles of mutual support and cooperation, and in so doing was propelling forward a decline in community. Daniel had set up, and then ran, a homeowners' association to manage the residential community in which he lived in order to save money: 'we could either pay a lot more each year and somebody would run it for us [. . .] some businessman would run it for us, or run it ourselves'. Paul joined the National Trust, along with his wife, because they enjoyed visiting the Trust's properties: 'we look at the houses, we like the gardens, we like the countryside'. Around half of the group had previously attended gyms and exercise classes and/or participated in sports clubs and teams to improve their fitness, because they enjoyed sport/exercise and/or because they liked the camaraderie of a team. Similar to past studies that find individuals can volunteer for organisations that benefit family members (Bussell & Forbes, 2002), while around a quarter of the NCDS-identified lifelong nonparticipants were childless, several of those who had children reported helping-out in organisations and activities associated with these children. They helped-out at Parent Teachers' Associations, Brownie packs and playgroups. Family formation emerged, as it has in previous studies (Bussell & Forbes, 2002), as a unique influence on participation behaviours across each type of participant we studied. As others have commented (Botterman, Sodermans, & Matthijs, 2015), the role of parent opened up a range of participation opportunities but, equally, it featured a set of duties and responsibilities that limited the time and energy individuals had available to participate. Family formation appeared to inhibit forms of participation exclusively associated with the self and promoted those associated with cohort members' children. For the 'below the radar' participants, involvement in activities related to their children always ceased when their children moved on to other pursuits, a scenario observed by others (Bussell & Forbes, 2002). However, for the more engaged individuals, the frequent and lifelong participants, commitments could continue even after their children moved on to other organisations and activities.
It was rare for the 'below the radar' participants to engage in groups or activities that benefitted people beyond their immediate circle, with participation that helped unknown others extremely uncommon. The odd exceptions to this included Daniel who had volunteered as a marshal on a sponsored walk, and planned to do so again, and Janet who, once a year, assisted her husband in organising and running daytrips for disabled children. Although a 'major undertaking', which involved time, effort and money on her part, she felt this was 'actually good, it's quite moving as well, it makes you feel very humble'. Relative to these individuals, participation amongst the frequent and lifelong participants was more likely to be externally focused. All the frequent participants, in fact, were, or had been, involved in organisations and/or activities that benefited people outside their immediate circle. For example, these individuals were, or had been, sports coaches and parish councillors, played or had played active roles in charities and were, or had been, actively involved in community, youth and disability groups.

Conclusions and implications
Linking NCDS structured survey data collected over the lifecourse with SPIS biographical interview data collected from a subset of the NCDS cohort at age 50, we explored the scale, characteristics and causes of lifelong nonparticipation in clubs/groups/associations, volunteering and religious meetings. We found that 'genuine' lifelong nonparticipation was uncommon. Rather more common, revealed by the SPIS interviews but missed by the data collection instruments of the NCDS, was informal, one-off, special-purpose, self-orientated participation. This is precisely the type of participation that Putnam (2000) claims is replacing well-exercised social ties built on principles of cooperation and mutual support. However, while more prevalent than no participation, this form of participation did not dominate. It seemed the SPIS cohort had not succumbed to the 'anticivic epidemic' of which Putnam (2000, p. 186) and certain politicians and commentators have warned.
Like nonparticipation more generally (Brodie et al., 2009;Putnam, 2000;Smith, 1994), lifelong nonparticipation was gendered and related to lower educational attainment. However, contrary to expectations Putnam, 2000), lifelong nonparticipants were unlikely to come from more deprived backgrounds or to report poorer general health, although they did appear to demonstrate poorer mental health. Such findings suggest that persistent nonparticipation might have unique predictors.
In exploring the causes of lifelong nonparticipation, we considered our findings against the key reasons for declining participation identified by Putnam (2000). Matching his thesis, watching television, financial anxieties and, in particular, time pressures arising from work and caring duties or, more precisely, the feeling of being 'pressed for time', all seemed important. However, on the issue of generational change, the primary cause of declining participation according to Putnam (2000), we found some evidence of generational continuity. One generation appeared to have passed down its participation behaviours to the next. The more involved individuals described engaged parents while the least involved depicted disengaged parents.
The social costs (Putnam, 2000), from underperforming schools to impoverished democracies, and the negative health outcomes (Pollack & von dem Knesebeck, 2004), from increased mortality (Dalgard & Lund, 1998) to higher levels of depressive symptomology (Duncan et al., 2017), which have been associated with nonparticipation suggest that society and individuals could benefit from higher levels of engagement in social and civic life. Our findings suggest that this could be fostered by interventions that make it easier for individuals to manage competing demands on their time, particularly those arising from work and caring duties. Efforts could be directed towards increasing individuals' capacity to reconfigure their working hours and to work more autonomously. They could also be channelled towards improving the supply of high-quality, affordable childcare - Young (2018) found that the UK has the highest childcare costs of all OECD counties. Additionally, they could be directed towards creating volunteering and participation opportunities that facilitate more ad-hoc engagement that can be initiated and paused as working patterns and caring commitments allow. Our findings on the importance of family formation as a trigger for participation suggest that the arrival of children might be the most opportune moment to encourage the adoption of a participatory disposition. Information and advice about participation, and invitations to participate, could be targeted at individuals in this life stage to encourage a transition from occasional, informal 'helping-out' to a regular participation committment. Noting our findings on generational continuity, developing pro-participation behaviours amongst these individuals might promote participation in the next generation, helping to establish a self-renewing pro-participation cycle. This is the first study to integrate NCDS and SPIS data in order to understand long-term patterns of participation in social and civic life. Both are high-quality datasets that follow carefully constructed, transparent methodologies. Bringing the two together produced a more complete picture of how, why and where participation in social and civic affairs slots into, and interacts with different components of, an individual's life over time. Usefully, the SPIS interviews teased out many forms of participation overlooked by the NCDS that has traditionally focused on formal modes of participation practised in organised settings. Future qualitative initiatives linked to the NCDS could fruitfully engage with changing forms of participation over time, for example, in relation to technological change. Such initiatives could enrich, and bring greater nuance to, NCDS findings. In terms of limitations, while including all individuals in the SPIS whose NCDS data records indicated lifelong nonparticipation, the sample employed within the study was small, and contracted further as analysis of the SPIS transcripts indicated just seven individuals were 'genuine' lifelong nonparticipants. A second limitation concerns missing data. To maximise our sample, we included individuals with incomplete NCDS data records, while in the SPIS interviews all individuals were not asked for, and did not volunteer, the same information. The findings should be interpreted in the context of these limitations.