Understanding adult education in community contexts: A critical realist perspective

Third-sector community organisations are important sites for learning, especially for the most excluded groups in society. However, scant attention has been paid to the various factors shaping educational provision in community contexts, and how these interact to shape the provision available to marginalised populations. This article presents new evidence addressing this gap, through drawing on interview data from practitioners working in the UK homelessness sector. It identifies a range of factors shaping educational provision in these community contexts. These are: service user need and demand; staff roles and capacity; organisational purpose and structures; national policies; support from other adult education providers; non-governmental finance; and volunteers. With some modification, and placed within an overarching critical realist framework, it is argued that these findings are consistent with Boeren’s comprehensive lifelong learning participation model. The article concludes that so long as government policy and related funding continues to ignore and fails to support educational provision in these settings, it will remain piecemeal and highly contingent on the contribution of volunteers and short-term funding.


Introduction
Third-sector community organisations are important sites for learning, especially for the most excluded groups in society (Quinn et al., 2005;Reisenberger et al., 2010;Tett, 2010;Golding, 2012). Indeed, adult education originated within the context of community organisations, driven by a commitment to social justice and empowering excluded groups of learners (Hamilton & Hillier, 2006;Barton et al., 2007). Learning in these and other educational settings can take different forms, ranging from formal structured programmes of learning involving the acquisition of qualifications to informal learning which can occur whether or not an adult intends to learn something new (Boeren, 2016). Informal learning in community contexts can be 'less obvious' than that found in formal educational institutions, with provision 'often hidden or embedded in other services or support provided' (Barton et al., 2007, p. 34;Golding, 2012). Less formal education provision in community contexts may offer an alternative to more rigid, standardised forms of educational provision, which result from increasingly prescriptive education policy and can deter marginalised learners (Appleby & Bathmaker, 2006;Hamilton & Hillier, 2006;Hamilton, 2009;Duckworth, 2013). In contrast, there is arguably a risk of 'philanthropic amateurism' (Anheier, 2014, p. 214) in such contexts, where volunteers are expected to tackle social problems despite not being qualified or experienced in aspects of the support they are providing. Nevertheless, community organisations play an important role in facilitating adults' access to learning, particularly where more informal learning opportunities targeted at 'hard to reach' groups are concerned (Tusting, 2003;Quinn et al., 2005).
However, few studies explore the extent or nature of this important source of educational provision in any detail, nor do they consider the range of factors shaping it. This article helps to fill this gap by focusing on one particular 'type' of third-sector community organisation-those supporting homeless adults. More specifically, it explores literacy and numeracy support within broader 'Education, Training and Employment' provision in these settings. Homeless adults represent one of the most marginalised populations, excluded not just from educational and economic opportunity, but also from accessing a secure home, often combined with multiple and complex needs. Arguably, the challenges this article deals with are most stark for this group of (potential) learners, however the findings are relevant to those working in a range of community contexts.
The remainder of the article proceeds as follows: first, the context around adult learning provision in the homelessness sector is presented; second, a theoretical framework is outlined. The methodology is then described, before key findings are presented and discussed, and recommendations and conclusions are made.

Homelessness, adult education and the community sector
Homelessness is a complex phenomenon which takes various forms, including rough sleeping, living in hostels and temporary accommodation, and 'sofa surfing'. Homelessness of all forms has been increasing in recent years across developed nations (FEANTSA, 2018). In England, for example, rough sleeping doubled between 2010 and 2015 (Fitzpatrick et al., 2016). Homelessness results from the complex and nonlinear interaction of various factors at individual, interpersonal and structural levels (Fitzpatrick, 2005). It affects a range of household types, although the focus in this article is on single homeless adults. This group is typically excluded from statutory support and services, and is thus a key target for third-sector homelessness organisations, which step into this gap.
Data are lacking regarding the educational profile of homeless adults, however the majority have faced poverty and social exclusion throughout their lives (Bramley & Fitzpatrick, 2017), which is associated with poorer educational outcomes and exclusion from learning opportunities. Whilst those who might be formally considered 'high skilled' may find themselves without a home (Barton et al., 2007), a small but growing evidence base suggests that many homeless people have poor literacy and numeracy skills (Luby & Welch, 2006;Olisa et al., 2010). Dumoulin and Jones (2014), for example, found that in a sample of 139 homeless adults, 51% and 55% respectively had poor literacy and numeracy skills (i.e. below Level 1).

Adult education in community contexts 675
However, previous studies have found that homeless people can be excluded from opportunities and support offered by adult colleges and other private training providers (Dumoulin & Jones, 2014;Luby & Welch, 2006;Olisa et al., 2010;Reisenberger et al., 2010). Perhaps reflecting this, many third-sector homelessness organisations have developed Education, Training and Employment (ETE) support alongside other interventions to address the diverse range of complex needs of their service users, and a high proportion offer literacy and numeracy support as part of this (Homeless Link, 2016).
UK policymakers have sporadically acknowledged the importance of homelessness organisations as sites for learning. In the mid-2000s under New Labour, after identifying homeless people as a group in need of improving their basic skills, homelessness organisations became sites of Skills for Life provision (Barton et al., 2007). More recently, the Conservative-Liberal Coalition government (2010-2015) funded STRIVE (Skills, Training, Innovation and Employment) pre-employment pilots, which took place in London at two national homelessness charities (DCLG/BIS, 2014). STRIVE was a small-scale 'pre-employment' programme, providing opportunities for homeless people to build their confidence and develop their IT, maths and English skills. Commenting at the pilot's inception, the then Skills and Enterprise Minister, Matthew Hancock, said: It is wrong that until now excellent education projects led by [Homelessness charities] have been denied government fundingtoday we are putting that right. There is no doubt that [National Homelessness charities] are the best placed to reach those in need of help, but we are backing them in this vital task. (Varvarides, 2014) However, despite policy rhetoric around the value of engaging homeless adults in education, statutory funding for learning and skills flowing into homelessness agencies has been minimal. In England in 2016, only 3% of accommodation projects and 7% of day centres received any 'employment and education' funding (Homeless Link, 2016). This is against a backdrop of policy commitment to austerity, which has had a profoundly negative impact on local services (including both adult education and homelessness organisations) (Jones, 2019a). The current government's commitment to this agenda is uncertain.
As with policy and practice, education provision for homeless adults is also a neglected research topic. In 2000, a review of research on homelessness in Britain concluded that: 'most research addressing homelessness and education is limited to children', which the authors took to suggest that 'there is little expectation that underachievement at school can be compensated for afterwards' (Fitzpatrick et al., 2000, p. 34). Moreover, existing research on adult education in homelessness and other community settings has focused on the learner and their learning experiences, rather than the extent and nature of provision available in these settings or the factors shaping this (Castleton, 2001;Luby & Welch, 2006;Juchniewicz, 2011).
This evidence gap is important for those concerned about homeless adults' access to learning opportunities. Exclusion from formal adult education and a preference to engage with non-governmental, charitable organisations can mean that the support homeless people are able to access depends on what these local services offer. Such institutions and the learning opportunities they provide are a key factor influencing 676 Katy Jones whether or not homeless adults are able to engage in educational provision (Boeren, 2016). Without knowing about the factors shaping educational provision in these settings, it is difficult to identify ways to improve it. The following section considers existing theoretical developments in relation to the activities of third-sector organisations, and identifies Boeren's (2016) comprehensive lifelong learning participation model in combination with a critical realist underpinning as a potential way forward.
Theoretical framework: a critical realist take on the comprehensive lifelong learning participation model Most theories concerning the development of third-sector organisations tend to focus on particular factors influencing their activities. Social origins theory (Salamon & Anheier, 1998), for example, emphasises the embeddedness of the third sector in broader political and social contexts, and resource dependence theory (Pfeffer & Salanick, 1978) highlights organisations' dependence on resources outside their control (including finance and information) in order to function and survive. Whilst useful, a tendency to focus on particular factors operating at one level (e.g. policymaking) fails to encompass the range of factors influencing these settings and neglects to consider how factors operating at structural levels interact with the agency of individuals and organisations.
In educational research, Boeren (2016) offers a way forward, although her focus was on 'formal' adult education providers, rather than the community contexts of concern here. According to Boeren et al.'s (2010) integrated model of participation in adult education, whether or not adults participate in learning is the result of a number of factors operating at three key levels; namely, that of individuals, institutions and countries. Influenced by Giddens (1984), Boeren (2016) develops the model further, emphasising the interaction between the range of factors influencing adult learning participation. As Giddens (1984, p. 171) explains, to understand social phenomena it is necessary to recognise the 'duality' of structure, as: Human societies, or social systems, would plainly not exist without human agency. But it is not the case that actors create social systems: they reproduce or transform them.
Whilst, for example, an adult's decision whether or not to engage in education is made by the individual, choices are influenced by the social structures and entrenched inequalities in which they are located (Bourdieu & Passeron, 1990;Barton et al., 2007;Duckworth, 2013). Lower socio-economic status has a profoundly negative impact on early educational outcomes (Cassen et al., 2015), and those leaving compulsory education with no or low qualifications are less likely to participate in further learning (Kuczera et al., 2016). Thus, 'lifelong learning' serves to widen educational inequalities rather than narrow them, as those who are better qualified on leaving school go on to achieve higher qualifications in adulthood (Makepeace et al., 2003;Wolf & Evans, 2011;Aldridge & Hughes, 2012;Golding, 2012). Such structural inequalities are both reflected in and reproduced by government education and skills policies (and the institutions through which these are enacted) (Duckworth, 2013) and the structure of opportunities in the labour market. Failing to compensate individuals for unequal life chances, educational systems contribute to the production Adult education in community contexts 677 and reproduction of social inequalities (Willis, 1977;Bourdieu & Passeron, 1990;Archer, 2013;Duckworth, 2013).
Drawing on Boeren's (2016) integrative model is helpful in prompting consideration of the range of factors which might impact on homeless people's (potential) participation in learning opportunities, helping to explain their exclusion from formal adult education provision. Where policy results in standardised provision which does not meet the needs, speak to the interests, or lend itself to the inclusion of homeless people, the available evidence suggests that this group of (potential) learners will be unlikely to engage in adult learning provision. In addition, where homeless adults are perceived to have limited desire to engage in learning opportunities (perhaps communicated through poor attendance or lateness), educational institutions may be less likely to seek to cater for this group, particularly when their funding is predicated on measures such as attendance levels, course completion and qualifications obtained. Whilst Boeren (2016) focuses on formal and non-formal learning, it is argued here that her model has wider applicability, helping to demonstrate the various factors influencing adult learning participation in community contexts. However, to reflect a key difference in community learning contexts identified in this research, it will be argued that some modification is needed to reflect the important role of volunteers and non-government funding in shaping the extent and nature of provision.
Furthermore, this article also argues that locating Boeren's model within an overarching critical realist framework can help to enhance its explanatory power and identify opportunities for individual agency and change, albeit recognising the constraints imposed by dominant social structures. A critical realist perspective regards organisations as complex entities, which can produce (and reproduce), resist or challenge social structures (Elder-Vass, 2010;Edwards et al., 2014;Vincent & Wapshott, 2014). An organisation's activities are influenced by larger social structures, the actions of the actors working with them and the individuals who draw on their services (Elder-Vass, 2010). Vincent and Wapshott (2014) identify the importance of acknowledging how factors operating at multiple levels interact and influence the activities of any given organisation. However, critical realism is also underpinned by a commitment to 'analytical dualism', which holds that neither structure nor agency can be 'wholly explained in terms of the other' as both are interdependent (Shipway, 2011, p. 84). This is consistent with Gidden's structuration theory, which influenced Boeren's work (Stones, 2001;Fitzpatrick, 2005). However, whilst recognising this interdependence of structure and agency, it is important to recognise that structure precedes action: structure precedes action which, in turn, leads to a more or less attenuated structural outcome. . . which, in turn, provides the preconditions for action. (Stones, 2001, p. 180) However, whilst 'social structures are dependent on human actors to reproduce them', critical realists recognise the ability of individual agents to make changes in the world.

Method
The limited existing research on adult education in homelessness organisations has focused on the learner and their learning experiences, rather than the extent and 678 Katy Jones nature of provision available in these settings or the factors shaping it (Castleton, 2001;Luby & Welch, 2006;Juchniewicz, 2011). This is an important gap to address as institutions and the learning opportunities they provide are a key factor influencing whether or not adults are able to engage in educational provision (Boeren, 2016).
To address this gap, this study adopted a qualitative methodology, involving semistructured interviews with practitioners working across Greater Manchester's homelessness sector (Greater Manchester is an urban conurbation in the North West of England consisting of 10 local authorities). Whilst often neglected in research focusing on homelessness, practitioners are well placed to understand both the services and support offered by the organisations in which they work, alongside the factors (both internal and external) that shape it. Importantly, though constrained by pre-existing social structures (Fitzpatrick, 2005), individual agents can make changes in the world, and as agents with a pivotal role in enacting change, homelessness practitioners are well placed to offer insights into what could enhance the support available in their organisations.
A key aim of the research was to capture the perspectives from those working in multiple organisations across the homelessness sector in order to identify common themes. A purposive, non-random sampling strategy was employed (Mason, 2002). All staff and volunteers working in organisations identified in a desk-based review of homelessness provision in the area were invited to participate. In total, 27 semi-structured interviews were conducted with practitioners working in 12 third-sector organisations (a large proportion of organisations operating within the Greater Manchester area). The sample contained broadly three different organisational 'types': activity centres, residential projects and social enterprises.
Fieldwork took place between August and November 2015, following approval from Lancaster University's ethics review panel. Face-to-face interviews were conducted, mostly in private rooms within the participants' place of work. With consent, all interviews were digitally recorded. The sample includes 12 'strategic-level' workers (those with responsibility over the strategic direction of the organisation's activities, such as chief executives and directors), six 'managerial-level' workers (those in charge of managing other staff in the organisation) and nine 'operational-level' workers (those with front-line roles and responsibilities-such as 'support workers' or 'project workers'). Despite significant recruitment efforts, no volunteers were included. This is slightly problematic given the pivotal role volunteers appeared to play in educational provision. However, this perhaps reflects the perceived importance of volunteer time-the organisations contacted may have played a 'gatekeeping role' here, keen not to 'burden' volunteers through making additional asks on their time beyond their valued volunteer role.
Interviews were semi-structured in order to allow meaningful comparison across participants whilst at the same time allowing for flexibility in the discussion. The interview topic guide was informed by both the research questions and the literature review (a copy of the topic guide is available on request). Interviews began with a discussion of the role and professional background of interviewees and the organisation in which they worked. Participants were then asked about the nature and extent of literacy and numeracy support offered in their organisation, and were asked to consider who and what influenced this support.

Adult education in community contexts 679
Whilst there are limits to what can be ascertained through the accounts of practitioners (the knowledge of actors can only ever be partial), the reasons and accounts individuals give for their actions 'form the logically indispensable starting points. . . of social scientific inquiry' as it is through these actions that structures are reproduced or transformed (Bhaskar, 2014, p. 156). Given the absence of research focused on these settings to date, practitioner accounts can be used to identify the structures and mechanisms which shape the support they provide (or do not provide) (Corson, 1998, cited in Shipway, 2011. To date, little attempt has been made to theorise what shapes support in this context. As such, analysis began with a commitment 'to work out a. . . reliable explanation for these patterns of events via the development of more adequate accounts of the powers, entities and mechanisms which created them' (O'Mahoney & Vincent, 2014, p. 9). An inductive thematic approach underpinned data analysis, starting with a 'commitment to theoretical pluralism' until initial investigations helped to identify the key factors at play (O'Mahoney & Vincent, 2014, p. 18). Once interviews were conducted and transcribed, an initial framework for analysis was created based on emerging themes resulting from detailed and repeated reading of the interview transcripts. Systematic thematic analysis of the interview data was then conducted using QSR NVivo10, allowing for the creation of new themes as they arose (contact the author for the coding framework developed as part of this process). Whilst initial analysis was grounded in the interview data, it will be shown that the findings in many respects verify Boeren's (2016) model as a helpful framework through which to understand the various factors shaping educational provision in these settings. However, it will be argued that the model requires modification to reflect the complexity of such organisations, and can be enhanced through being placed within an overarching critical realist framework.

Findings
Interviewees described a range of educational activities offered or facilitated by their organisations, including reading groups, creative writing groups, literacy and numeracy courses and one-to-one support, and embedded learning ('learning on the job' in a social enterprise). In a small number of cases, educational provision and support formed a regular part of the services offered. However, most commonly, learning opportunities were short-term, ad hoc and occupied a precarious position. Seven key factors appeared to impact on the role and nature of literacy and numeracy support offered. These were: 1. needs and demands of service users; 2. roles and capacity of staff; 3. organisational purpose and structures; 4. national policies relating to adult education, austerity and welfare reform; 5. support from other adult education providers; 6. non-governmental finance; 7. time and expertise of volunteers.

Katy Jones
As shown in Figure 1, the first five of these factors neatly map onto Boeren's (2016) model: all can be considered either individual (factor 1), institutional (factors 2 and 3) or country-level factors (factors 4 and 5) impacting on adult learning participation. However, the latter two are harder to place within the three-tiered model. This reflects the complexity of third-sector organisations-the fact that the provision of learning opportunities is not typically a key aim of such institutions, and the finding that educational provision in community contexts is highly contingent on factors beyond an organisation's control (through a heavy reliance on volunteers and shortterm funding opportunities). In the following sections, evidence is presented to support this.
Individual factors: the impact of service users on learning provision Unsurprisingly, as organisations designed to support homeless adults, interviewees explained that service user needs were an important factor shaping provision. Supporting earlier research findings (e.g. Luby & Welch, 2006;Dumoulin & Jones, 2014), staff believed that a significant proportion of their service users had a literacy or numeracy 'need': There are very few who have even finished school. . . there's such a huge need. (Strategic-level worker, residential project and social enterprise) Staff considered developing literacy and numeracy skills to be an important part of building a person's confidence and self-esteem, helping to 'empower' people, and enabling them to live independently with more control over their lives-both in and outside the labour market: Figure 1. Factors shaping educational provision in homelessness settings (adapted from Boeren, 2016) Adult education in community contexts 681 I think with being able to read and write the key thing is that you can start to do things for yourself. . . from benefits to getting a job to understanding what's going on around you. (Operationallevel worker, day centre) Interviewees recognised several barriers to homeless people's participation in opportunities to improve these skills, ranging from individual motivations and confidence, to exclusion from formal education provision. In contrast, the educational provision offered in their settings was reportedly shaped by the needs and demands of their service users. Educational activities were described as flexible and responsive, reflecting an appreciation of the multiple and complex needs of many homeless people, alongside fluctuating motivation and self-confidence. Attempts were also made to recognise and overcome individual barriers to learning: [W]e have to make adaptations. . . if someone has got mental health issues by all means we'll have regular breaks, we'll make sure that you're in a good frame of mind. Take into consideration what people's barriers are and work round it individually. (Operational-level worker, residential project) Most participants talked about the importance of offering opportunities for service users to engage in learning within the context of their own, or similar, organisations. Being supported to learn within a 'familiar', 'comfortable', 'trusted' and 'safe' environment was a key reason for this: People tell us that they don't feel judged here. . . they feel valued and respected and all the rest of it and that's what we want to do. Because some people don't feel that anywhere else. (Strategiclevel worker, day centre) This was juxtaposed with the exclusion they believed their service users experienced from a variety of formal or 'mainstream' services (including, but not limited to, adult education). Many talked about the barriers to accessing formal adult education for the people they were supporting, and identified a reluctance amongst service users to access support in unfamiliar settings. This supports the notion that learning in such alternative contexts, outside of the formal education system, can offer important opportunities to those who are unlikely to engage with 'formal' provision (Tusting, 2003;Quinn et al., 2005): [T]here's a real clear need for it to not always sit in college and learning environments. . . increased access to opportunities in different spaces. . . to kind of make it more accessible, and not always sending people to. . . really kind of what they may view as intense adult learning centres and activities like that because that can also be a real barrier for some people. (Strategic-level worker, day centre) Knowing the extent to which provision genuinely matched the interests and needs of service users is limited in the absence of consultation with homeless service users themselves. Practitioners arguably have a stake in arguing for services to be located within their own organisations, especially within the context of a challenging funding climate. However, evidence from other research involving homeless participants substantiates these claims (Luby & Welch, 2006;Juchniewicz, 2011).

Institutional factors: the impact of practitioners and organisational structures
Interviewees also identified a number of 'institutional' factors shaping educational provision in their settings. Namely, the roles and capacity of staff working in 682 Katy Jones homelessness organisations and the structures and aims of the organisations in which they worked. Whilst the service mix and support varied across organisations, interviewees tended to self-identify as 'generalists'. With high levels of autonomy, their roles were guided by the diverse needs and aspirations of service users, to respond to whatever they needed help with, whether that was housing, issues with drugs and alcohol, benefits, health or moving into work. This could be difficult, particularly in small teams with high case loads. Regarding literacy and numeracy, staff described helping people where there was a need and they were able to: You just do everything you can to help someoneif we've got chance to [support people with literacy] and we've got time to do it then we will. (Operational-level worker, day centre) However, interviewees commonly felt that they lacked the skills and expertise to support people appropriately. Most had backgrounds in social work, probation and youth work. Only one had received training relating to adult education. Only two organisations directly employed tutors (in one case only one role was funded on a part-time basis and the tutor's role was primarily to support service users to develop their ICT skills). Thus, whilst staff were 'professionals' in terms of 'supporting homeless people', their ability to provide educational support was limited: I don't have the knowledge base to teach, so people aren't getting what they need. (Operationallevel worker, day centre) Educational provision was also shaped by the structures and functions of organisations. Whilst sharing a common mission to support homeless people, organisational forms and functions varied. In the activity centres, for example, staff worked with a relatively large number of service users. Here, service user needs were most varied, ranging from people in immediate 'crisis' to those with more settled accommodation. Due to higher 'footfall', more 'potential' learners were in contact with the service, however attendance at learning activities could be more sporadic. Staff roles were more diversified in larger activity centres-and here there were the odd examples of skills tutors and planned learning activities. In the residential projects, in contrast, a smaller number of staff were working with a relatively small number of service users. This created more opportunities to identify educational needs and aspirations, develop trusting relationships and support people informally. Structured courses were sometimes more possible in these settings, as service users spent more time on the organisation's premises. In the social enterprises sampled, a key concern was the successful operation of the 'business'. Here, opportunities to develop skills through formal training were typically guided by the needs of the enterprise, such as workplace health and safety certification. In addition, whilst here there were perhaps more opportunities to identify educational need and provide opportunities for development 'on the job', there was less time for structured courses.

Country-level factors: the impact of national policy on literacy and numeracy provision
Consistent with the third element of Boeren's model, a number of country-level factors also appeared to impact on educational provision in homelessness settings. Central to interviewee narratives was the impact of the national policies of austerity, Adult education in community contexts 683 welfare reform and adult education and skills policy, each of which are considered briefly below (see Jones, 2019a, b for more detailed discussion).

The impact of welfare reform and austerity
Interview data reveal how the wider welfare system impacted on the services offered by homelessness organisations as staff responded to the needs of their service users, many of whom were struggling to adapt to an increasingly restrictive and punitive welfare regime (introduced against a backdrop of austerity). Whilst a range of educational activities had taken place in the past, increasing amounts of staff time were consumed by helping service users to learn about and understand benefit changes, and advocating on their behalf to challenge adverse decisions about benefit entitlement: There's an element of crisis work that has become a priority. . . the number of people in situations where they've been going for week after week without money. . . that kind of work has taken a priority over the last year or so. (Strategic-level worker, activity centre) Welfare reform also prompted training around digital skills in order to equip service users with the skills they need to navigate a new cost-saving 'digital by default' system for administering people's social security benefits. Furthermore, some interviewees described needing to plan course provision around the conditions service users were expected to meet in order to access benefits. For example, provision was flexible, allowing for missed sessions and lateness in recognition of service users' need to attend appointments at the public employment service. This, it was felt, was not as well catered for in more formal adult education settings: It affects the attendance. . . because they do have appointments on what they need to stick to. (Operational-level worker, accommodation project) Alongside welfare reform, the broader impact of austerity on provision in these settings featured across the accounts of interviewees. Many of the organisations had experienced significant funding reductions over recent years. Here we can see the fragility of provision in these settings and its vulnerability to inconsistent government funding.

The impact of adult education and skills policy
Reflecting the lack of state-funded learning across the homelessness sector identified above, most organisations were not receiving any statutory education funding. As the major funder of adult education activities, the absence of government funding perhaps explains why support in such settings occupies such a marginal position. There were some examples where organisations had been able to draw down European funding to provide learning opportunities. However, one interviewee explained how accessing funding which recognised the challenges of working with their 'client group' was difficult: If we go to a hostel and two people show up, and the funding that we've used for that is based on a guided learning hour calculation. . . it's not sustainable for us. So we need to find funding that recognises how much it costs to do that well and that's a real struggle at the moment. (Strategiclevel worker, activity centre) 684 Katy Jones A lack of government funding was felt to reduce the extent of the educational provision they were able to facilitate. However, perhaps more positively, the absence of such funding may help to explain the nature of provision that does exist. Without the need to satisfy government standards and outcome measures, which tends to result in more rigid, standardised forms of adult literacy and numeracy provision (Hamilton & Hillier, 2006;Barton et al., 2007;Duckworth, 2013), these organisations had greater freedom to develop support in a way which was guided by the aims and interests of learners rather than predetermined frameworks with little relevance to them.

External adult education providers
The impact of national policy on provision in these settings can also be seen through exploring the interaction between the homelessness and adult education sectors.
Recognising the limits of their own capabilities, interviewees described attempts to identify and draw on resources from the wider community. Several described hosting external adult education providers within their settings. However, no such activities were underway at the time of interview, and interviewees had observed a notable reduction in engagement and outreach in recent years. Whilst many had previously hosted tutors from local education providers, they were disappointed at the recent withdrawal of this following funding cuts: We used to have the [adult education provider] in. They used to regularly do stuff at [the organisation]. I'm going back several years. . . particularly literacy classes. . . but all that funding's gone. (Operational-level worker, activity centre) One participant mentioned 'Skills for Life', the most significant adult literacy and numeracy policy over the past few decades, reflecting that: All that concern with Skills for Life has gone. . . back then, you couldn't turn a corner without somebody telling you the stats about young male illiteracy levels and stuff like that. I don't hear it anymore. (Strategic-level worker, activity centre) More generally, interviewees felt that opportunities for learning within the wider community were becoming increasingly limited, thus limiting the potential brokerage role they could play. Several described a lack of affordable learning opportunities in their local areas, and restrictions in the types (subjects) of learning opportunities available, which did not meet the needs or appeal to the interests of the homeless people they were supporting: [T]here are a number of other colleges who. . . fund courses but they tend to be the same old same old. . . Want to do a level 2 in customer service? Want to do a level 2 in cleaning? Well no, we don't reallyit's all a bit mundane! (Strategic-level worker, social enterprise)

Additional factors influencing educational provision in community contexts: volunteers and alternative funding sources
In addition to the above, a further two key factors were identified which affected both the nature and extent of literacy and numeracy provision in these settings. First, there was a heavy reliance on the skills and experience of volunteers: Adult education in community contexts 685 [T]here is this teacher, or ex-teacher, and he suggested [developing literacy and numeracy learning opportunities] to us and we were like yeah, wonderful. (Strategic-level worker, residential project and social enterprise) Often, time donated by volunteers was integral to the ongoing provision of learning activities and support. Although highly valued, dependence on volunteers to support the ongoing provision of learning opportunities could make services inconsistent: He is a volunteer. . . it's hit and miss. If [he] doesn't want to come then we can't force him. (Operational-level worker, residential project) One interviewee also felt there was a lack of volunteers with the skills necessary to teach literacy and numeracy.
Particularly where volunteers are not trained in teaching, this perhaps exposes a high incidence of 'philanthropic amateurism' (Anheier, 2014, p. 214) mentioned above. Importantly, some respondents were concerned about the appropriateness of relying on support offered by sometimes inconsistent and inexperienced volunteers: Providing that one-to-one support requires a real kind of commitment from people which is difficult to guarantee. . . the last thing we want is those people having yet another bad experience of education. (Managerial-level worker, day centre) Second, in the absence of government funding, interviewees described drawing on alternative third-sector funding sources (e.g. large grant-making trusts and one-off grants from local authorities) to fund learning activities. These time-limited income sources were subject to a high level of competition from other organisations and causes. As noted above, several organisations also operated as social enterprises. Educational provision was highly contingent on whether funding was obtained from these other sources-and this was not in plentiful supply.

Discussion
This article has identified multiple factors shaping educational provision in third-sector homelessness organisations. Arguably, the challenges this article deals with are most stark for homeless learners (as they represent one of the most marginalised populations), however the findings are relevant to those working in a range of community contexts.
With some modification, the findings presented above are consistent with Boeren's (2016) comprehensive lifelong learning participation model, despite its focus on formal adult education. Individual-level factors (the needs and demands of service users), institutional-level factors (the roles and capacity of staff and organisational structures) and factors operating at the broader national and policy levels (support from adult education providers, austerity and welfare reform) all impact on the extent and nature of support available in these settings.
However, reflecting the complexity of third-sector organisations (and the fact that educational provision is not their main purpose), adding volunteers and non-governmental finance to the model helps to capture a more complete picture of the range of factors impacting on education in these community settings. A key reason why 686 Katy Jones learning opportunities were typically short-term, ad hoc and occupied a precarious position resulted from homelessness organisations' dependence on volunteers and short-term funding opportunities. Placing these factors outside of those identified in Boren's original model (rather than incorporating them into the existing factors) is important as it helps to emphasise that ultimately, these factors fall largely outside the control of the homelessness organisations offering these services (Pfeffer & Salanick, 1978).
The findings support Boeren's (2016) argument that multiple aspects of her model of adult learning participation interact. For example, the needs of learners are impacted on by national policies of welfare reform, the services offered by homelessness organisations are shaped by the need to respond to these and by the broader context of austerity in which they operate, and the inaction of national policymakers to support work in the sector has contributed to the limited support available. However, it is further argued that the explanatory power of Boeren's (2016) model can be enhanced by being placed within an overarching critical realist framework which not only emphasises the interaction of different factors but also acknowledges the dominance of structural factors in explaining social phenomena.
Although the agency of homelessness practitioners is demonstrated through their responsiveness to the needs of service users, their actions were constrained by the wider social structures in which they were operating. Organisations have demonstrated a propensity to develop literacy and numeracy support, but while government policy and funding does not recognise, reward and support such provision, it is likely to remain piecemeal and highly contingent on the contribution of volunteers and short-term grant funding. Despite sporadic policy announcements about the importance of engaging homeless adults in basic skills support, any tangible support is hard to identify. The lack of government funding in this area explains in large part why only a limited level of support is available, yet also (and more positively) why the support offered is flexible and designed to suit and fit around the needs of homeless learners. Thus, it is arguably factors operating at the national policy level which most convincingly explain why provision in these settings is not more substantial.
The dominance of structural factors is further supported as the interaction between policy and educational provision in homelessness organisations appears to be one-directional-such institutions appear not to impact on 'country-level' factors. Whilst it is possible to conceive that lobbying from larger, higher-profile national organisations has resulted in at least some funding for basic skills provision (e.g. through the STRIVE pilot), the homelessness sector more generally does not appear to have much influence over adult education policy and funding. That said, it is important that the work that goes on in these settings is not dismissed. Within the context of profound inequalities, spaces in which those excluded from mainstream learning opportunities can learn, develop and be empowered to move towards an independent life are vital. Moreover, in line with critical realism's commitment that individuals not only reproduce but have the power to 'transform' social structures and make changes in the world, the value of such activities should not be underestimated.

Implications for policy and practice
The findings presented above have a number of policy and practice implications. Reflecting the range of factors impacting provision in these settings, recommendations are made at multiple levels.
In the homelessness sector, organisations could explore ways of using existing activities more effectively to develop learning opportunities for their service users, for example through social enterprise activities and other forms of service user involvement (e.g. in newsletters). Opportunities for collaboration between different homelessness organisations could also be explored. For example, through exploring the possibility of co-funding models, or promoting existing educational activities across the local homelessness sector rather than to service users in single organisations.
The adult education sector should reflect on the perception of homelessness practitioners that homeless people are often excluded from mainstream adult education services (which echoes previous research focused on education and homelessness, including Luby & Welch, 2006;Reisenberger et al., 2010;Dumoulin & Jones, 2014). Whist it is important not to downplay the work already undertaken by local authority adult education services through their community learning offer, formal adult education institutions should ensure that relevant outreach opportunities are communicated effectively to the homelessness sector, and through a collaborative approach should identify and (where possible) remove barriers to learning participation in their organisations for those with multiple and complex needs, including those experiencing homelessness.
Finally, and most crucially, the government must ensure that educational opportunities, including those to develop literacy and numeracy skills, are adequately funded across the homelessness sector. Policymakers should reflect and act on the fact that despite sporadic pronouncement about the importance of ensuring homeless adults are supported to develop literacy numeracy skills, this research exposes a dearth of government funding in this area. Funding must recognise the challenges involved in supporting homeless people to improve their literacy and numeracy skills, and build on existing provision developed in response to service user needs, capabilities and aspirations.

Conclusion
This article has presented new empirical evidence exploring the factors shaping educational provision in homelessness settings. Alongside contributing to the small evidence base on homelessness and education, the article has also demonstrated how an adapted version of Boeren's (2016) comprehensive lifelong learning participation model can help to explain the extent and nature of provision in these community settings. Individual, institutional and country-level factors, alongside the availability and resources of volunteers and non-governmental finance, interact to shape provision in these settings. This is the first time such a theoretical framework developed within the educational research tradition has been applied to phenomena occurring in the homelessness sector. Furthermore, the article has demonstrated the need to place this model within an overarching critical realist framework to enhance its explanatory 688 Katy Jones power. Whilst recognising that factors at different levels interact to result in the shortterm, ad hoc and precarious nature of provision, a critical realist framework explains why structural factors (i.e. the lack of consistent policy commitment) appear to dominate the explanation for the typically marginal position of education in homelessness settings.