The proposed paper will explore a related set of methodological and theoretical issues surrounding the identification of a sample of young people who were permanently excluded from a special school or a Pupil Referral Unit (PRU) (alternative educational provision) in England during the 2005-06 school year. The main focus of this small-scale longitudinal study, which is funded by the Department for Children, Schools and Families (DCSF) (formerly the Department for Education and Skills — DFES), is a group of 20 young people. The research team is tracking these young people over a three-year period in order to explore the events that led up to their permanent exclusion; to identify what types of alternative provision were available to them and their destinations post-exclusion; and to examine the impact of their permanent exclusion and related experiences on their behaviour, attitudes, achievement and attainment. Children who have been permanently excluded from a special school or PRU generally have complex and multiple disabilities. They are often described as ‘hard to reach’ (or ‘hard to find’, Macnab et al, 2007), by virtue both of their exclusion from school and their disability. However, as Brackertz (2007) has pointed out, ‘there is a lack of clarity about what is meant by “hard to reach” [and] the term is employed inconsistently.’ It can refer to minority groups; hidden populations (i.e. those who do not wish to be found or contacted); or ‘broader segments of the population, such as old or young people, or people with disabilities’ (Brackertz, 2007, p 1). In the research literature, considerable attention has been paid to the role of gatekeepers in providing access to some young people (Daniels and Macnab, 2004; Emmel et al, 2007). The introduction of the notion of ‘vulnerability’ into this equation adds to its complexity, and has presented the research team with considerable challenges in terms of recruitment to the study. The terms ‘hard to reach’ and ‘vulnerable’ imply a degree of homogeneity within groups. They also indicate that the problem lies within the individual, rather than in the nature of the research approach (and its intrinsic systemic constraints — level of funding, timescale, etc); or, indeed in the extrinsic systemic constraints that dictate the nature of the approach — e.g. data protection legislation and high staff turnover among key service providers — ‘socially situated, imperfectly knowledgeable actors’ (Garland, 2001, p 26). Furedi (2007) notes that from the late 1980s onwards, ‘vulnerability was no longer seen as something that springs from specific circumstances, for example poverty: rather it was considered to be an inherent condition of the individual’ (p 7) (see also Furedi, 2006). The inherent sense of powerlessness and fragility conveyed by the term ‘vulnerable’ is fundamentally at odds with the centrality accorded to young people’s agency and aspiration in recent English education policy documents such as Every Child Matters (Department for Education and Skills, 2004). Every Child Matters and Learning Behaviour: The Report of the Practitioners’ Group on School Behaviour and Discipline [The Steer Report] (Department for Education and Skills, 2005) form the backdrop to the study reported here. The paper will outline the assumptions upon which the research methodology was based; and the time-consuming and messy nature of the research process itself. It will also explore the complex interactions between a set of inter-related tensions: • ‘protecting young people from harm’ and giving them ‘far more say about issues that affect them individually and collectively’ (Every Child Matters) • child protection and information-sharing across professional boundaries • trust and the search for more perfect forms of accountablilty (O’Neill, 2002).

destinations and outcomes for a small group of pupils permanently excluded from special schools and Pupil Referral Units (PRUs) in England