Utilising Bacchi's what's the problem represented to be? (WPR) approach to analyse national school exclusion policy in England and Scotland: a worked example

ABSTRACT This paper examines and compares national policies on school exclusion, using a specific framework for public policy analysis developed by Carol Bacchi [(2009). Analysing policy: What’s the problem represented to be? Frenchs Forest: Pearson]. This framework is known as ‘What’s the problem represented to be?’ or ‘WPR’. Bacchi’s framework has been applied to areas of policy as diverse as health care, labour markets, immigration and higher education, but its use in school-based education has been much more limited to date. Noting too, that the WPR framework itself is rarely interrogated or critiqued in the literature, our small-scale study therefore addresses two main objectives: (a) to offer an illustration, a worked example, of WPR in education as a contribution to methodological debate, and (b) to critically appraise the WPR approach, and its potential to acknowledge, challenge and disrupt normalising discourses within one key area of school-based policy – school exclusion.


Introduction
Carol Bacchi's (2009Bacchi's ( , 2012 'What's the Problem Represented to be?' (WPR) approach to policy analysis provides a novel way of thinking about and analysing policy. The approach challenges the dominant 'problem-solving' paradigm in policy research and instead proposes a new 'problemquestioning' paradigm (Bacchi 2009, p. xvii). The approach is not concerned with 'intentionality' and identifying 'gaps between the terms of a specific policy and what is actually delivered' but rather with the 'deep conceptual premises' on which policies are built (Bacchi 2009, p. xix). WPR provides a systematic methodology to critically question 'the taken-for-granted assumptions that lodge in government policies and policy proposals by interrogating (problematising) the problem representations it uncovers within them' (Bacchi 2009, p. xv).
While Bacchi's WPR approach has gained some traction in social policy analysis internationally, and within the broad field of education (see, for example, Loutzenheiser 2015, Bills and Howard 2017, Lomer 2017, Skovhus and Thomsen 2017, Beutler and Feneca 2018, Magnússon et al. 2019, most published studies focus on the findings; few studies provide a detailed account of the process of analysis or offer critique of WPR itself. Our article therefore has two main objectives: (a) to offer an illustration, a worked example, of WPR in education as a contribution to methodological debate, and (b) to critically appraise the WPR approach, and its potential to acknowledge, challenge and disrupt normalising discourses within one key area of school-based policyschool exclusion.
Before turning to describe the WPR approach in more depth, and our process of analysis, we first outline the background to our small-scale study.

Background
Official figures show that the propensity to use permanent school exclusion as a disciplinary measure differs vastly across the UK. National school exclusion statistics published on the Department for Education (DfE) and Scottish Government (SG) websites in 2020, show that in 2018/19, 98% of all children permanently excluded in the UK were from schools in England (N = 7894); compared with a total of only three cases in Scotland in the same period. 1 Given the known negative consequences of school exclusion, including for example: child criminal exploitation (Temple 2020), poor academic outcomes (Gill et al. 2017), becoming NEET (not in education, employment or training; Timpson 2019), substance abuse, and mental ill-health (Parker 2015, see also Daniels et al. 2003), the impetus to compare the warrants of permanent school exclusion, and the different responses to challenging student behaviour across jurisdictions, is strong. While individual risk factors associated with school exclusion (such as: inequalities associated with poverty, gender, special or additional support needs, ethnicity and often the intersection between these factors) are well-documented (e.g. Riddell andMcCluskey 2012, Gazeley 2013), much less is known about the system-level factors that lie behind the numbers.
Our own analyses of interview data collected from senior level policymakers and practitioners as part of the Excluded Lives research project: 'Disparities in rates of permanent exclusion from school across the UK' (2017/18), indicated that different policy discourses may help to explain the disparities between the four UK jurisdictions (see: Cole et al. 2019, Daniels et al. 2019, McCluskey 2019. In the study reported here, we were seeking to explore the findings from the Excluded Lives project further by comparing national school exclusion policy discourses, and their potential effects, in the two UK jurisdictions with the largest disparity in school exclusion rates; England and Scotland. As we began to consider an appropriate design for our study, we undertook a search for existing analyses of national school exclusion policies in the UK. The literature is sparse, particularly in the Scottish context. Macleod (2014) is an exception. In England, Dunn's (2015) critical discourse analysis of changes in school exclusion policy over the period 1997-2006 is also helpful. To our knowledge, similar policy analyses have not been conducted for the successive government periods in England, and none to date have adopted Bacchi's WPR approach. aim to address. She suggests that public policies, such as those guiding school exclusion, all play a role in constructing, reproducing and reifying the very phenomena they set out to address. She argues that 'there are no problems separate from the proposals purported to address them' (Bacchi 2009, p. 15); and hence, that policies can be understood as governing strategies in themselves. This all seemed helpful to us as we set out to undertake this analysis. It positions WPR not simply as method but as methodology. It follows Foucault's (1994, p. 456) concern with 'seeing on what types of assumptions, of familiar notions, of established, unexamined ways of thinking, the accepted practices are based'. It promises to provide a structure within which to open up these governing strategies to critical scrutiny.
For Bacchi (2009, p. xvi), WPR 'mounts a challenge to the current dominant intellectual paradigm that focuses on solving "problems"' and the 'what works' agenda. Given the ongoing debate surrounding the 'what works' agenda, examined in detail, for example, in a recent special issue of this journal in 2016, the potential for application of WPR to policy on an educational topic so much debated for so long, and so often in the public eye, seemed both timely and potentially generative.
In the next section, we provide a brief account of the WPR framework before describing our 'worked example'.

The WPR framework
The WPR framework consists of six questions (see Table 1) and starts by asking in a relatively straightforward way about the proposals for change or solution set out in a given policy and from there, uses this change or solution to trace how the underlying problem is constructed. Bacchi (2009, p. x) provides the following example: … if training courses are offered to women as part of a policy to increase their representation in better paid occupations or in positions of influence, the 'problem' is represented to be women's lack of training. WPR questions 2-6 then direct the analyst to probe the solution or proposal for change, asking about the: 'rationales for the proposal, deep-seated presuppositions underpinning the proposed change, possible silences in the understanding of what needs to change, and the effects that are likely to accompany this particular understanding of the "problem"' (Bacchi 2009, p. x).
Bacchi later added a helpful step 7 to make clear the need to both consider the relationship between the researcher's own positionality and the policies under scrutiny and apply the set of six questions to any re-problematisations or policy proposals put forward by the researcher. The questions are set out in Table 1 as though sequential and consequential but may be better understood as iterative. As we began, we were aware that there might be more than one problem representation in a document, and that there might be tensions and contradictions between problem representations (Bacchi and Bonham 2014). Table 1. WPR approach to policy analysis.
Question 1 What's the problem represented to be in a specific policy or policies? Question 2 What deep-seated presuppositions or assumptions (conceptual logics) underlie this representation of the 'problem' (problem representation)? Question 3 How has this representation of the 'problem' come about? Question 4 What is left unproblematic in this problem representation? Where are the silences? Can the 'problem' beconceptualised differently? Question 5 What effects (discursive, subjectification, lived) are produced by this representation of the 'problem'? Question 6 How and where has this representation of the 'problem' been produced, disseminated and defended? How has it been and/or how can it be disrupted and replaced?
Step 7 Apply this list of questions to your own problem representations.
In the worked example below, we provide an account of the step-by-step process through which we set parameters for the scope and depth of this small-scale study, and then conducted the analysis. We illustrate each of the steps with some tentative findings.

Worked example
Our study was guided by the following two research aims: (1) To test whether Bacchi's WPR approach offers a useful method for conducting comparative critical policy analysis of national school exclusion policy in two neighbouring jurisdictions with contrasting school exclusion rates; (2) To compare how the 'problem' of challenging behaviour in schools is represented in national school exclusion policies in England and Scotland.
To address these aims, we first had to identify which policy documents to include in the analysis and set our design parameters.

Setting the parameters
As we were interested in assessing the suitability of Bacchi's WPR approach, we chose to focus on the most recent 'prescriptive texts' (Bacchi 2009, p. 34), initially selecting seven documents for analysis. These included the current school exclusion statutory guidance or equivalent in England and Scotland, and their immediate antecedents, along with the most recent school behaviour guides in both jurisdictions. Given the intertextual nature of policy (Ball 1993, Gale 1999 we were aware that the selected documents were likely to reference other policies (for example regarding special or additional support needs), legislation, interventions and strategies, which would all have their own problem representations, and that by excluding those other documents we could only seek to provide a partial picture of what the 'problem' was represented to be. We also acknowledge that formal policy documents comprise only one aspect of larger policymaking and implementation processes. An extensive analysis was, however, not possible within this one small-scale study.
In order to provide a clear worked example for this paper we made a decision to further reduce the scope of the comparative analysis to the most recent school exclusion guidance in England and Scotland only (DfE 2017, SG 2017).
Having selected our documents, we then set further boundaries, mindful that Bacchi (2009) herself acknowledges the emphasis on different WPR questions may sometimes vary depending on the aims of the study. We did not attempt, in Foucault's terms, a full genealogy of national school exclusion policies in England and Scotland (which we align with Bacchi's WPR Q3), nor did we examine how the representation of these policies had been produced, disseminated and defended (aligned with WPR Q6). We identified questions 1, 2, 4, 5 and step 7 as most closely aligned with the study's aims, allowing us to interrogate the solutions these texts suggest, how the 'problem' of challenging behaviour is represented, and the underpinning presuppositions, silences, and effects of these problem representations.

Process of analysis
Initially, the selected documents (DfE 2017, SG 2017) were analysed independently by the two authors, to produce discourse profiles for both jurisdictions, noting similarities and differences between the two jurisdictions. We began the analysis by focusing on the first of Bacchi's questions: What's the 'problem' [in this case challenging behaviour] represented to be in a specific policy? To do this, we explored the solutions proposed by the current policies, and 'worked backwards' to '"read off" the implied problem from the proposal' (Bacchi 2009, p. 48). It should be noted here, that while we began our analysis looking for the proposed solutions within the policy documents, it became clear during the analysis, and as wefamiliarised ourselves with the method, that it would be helpful to look at 'exclusion from school' as a 'solution' in its own right and interrogate how different understandings of the purpose of exclusion may affect how challenging behaviour is represented and therefore addressed.
Next, we compared and discussed our results in order to check our understanding and application of the method, and the consistency of our findings. We created tables for both of the policy documents to compare our independent representations and annotated the tables with comments. We then discussed in person the differences in our identified representations and returned to the documents to re-interrogate and/or defend our understandings. When we returned to the policy documents, our discussion led us to question the extent to which each of our identified representations deeply permeated the policy text, or merely appeared as passing comments. Relatedly, we then questioned the extent to which each of the policy solutions wereoperationalised in ways in which Local Authorities and schools could translate into concrete actions. Having compared our tables, we produced amalgamated lists of agreed representations for both of the policy documents. We then repeated the same exercise, focusing on exclusion from school as the proposed solution.
The iterative process of independent analysis followed by discussionutilised for WPR Q1 was repeated for each of the subsequent WPR questions. In the section that follows we provide a step-by-step account of the process of analysis for each of the chosen WPR questions/steps (questions 1, 2, 4, 5 and step 7) before turning to critically appraise the WPR approach.
WPR Q1: what's the problem represented to be in the most recent English and Scottish national school exclusion policies?
As we have outlined the procedure for WPR Q1 in depth above, to avoid repetition, we move straight to look at Table 2 which consists of our amalgamated list of identified, and agreed, representations for the two policy documents presented in this paper (DfE 2017, SG 2017. Due to space, we have included a single example of a proposed solution to illustrate each of the representations. It is important to note that some of the representations were reflected in a number of solutions mentioned in the text. While there are some overlaps in the representations identified in the two guidance documents, for example challenging behaviour potentially indicating unmet needs, there are also differences, for example the lack of focus on how school and community factors may be affecting behaviour in the English policy, compared to the Scottish policy. In order to consider 'exclusion as the solution', we then identified the stated purposes of the English and Scottish guidance and compared the respective criteria for school exclusion (see Tables 3 and 4).
We were struck by the difference in emphasis and tone in the titles of the English and Scottish policy documents (see Table 3). We noted that in England the stated aim of the policy is to offer 'greater confidence to head teachers on their use of exclusion' (DfE 2017, p. 6), making clear their legal responsibilities, whereas in Scotland the aims relate to 'the prevention of and the management of exclusion from school in the context of national government and education authority responsibilities and desired outcomes for children and young people' (SG 2017, p. 5).
Further comparison elicited contrasting notions of challenging behaviour and of exclusion; in which challenging behaviour could be seen as requiring punishment (England) or welfare-based intervention (Scotland) and exclusion framed as a legitimate sanction (England) -'The Government supports head teachers in using exclusion as a sanction where it is warranted' (DfE 2017, p. 6)or an undesirable outcome (Scotland), which needs to be 'managed as a supportive way forward for the child or young person with transition planning put into place, in order to bring some resolution to the situation and avoid unnecessary gaps in educational provision' (SG 2017, p. 47). WPR Q2: what deep-seated presuppositions or assumptions underlie the problem representations outlined above?
Having identified problem representations for each of the policy documents, the second step, was to consider how it is/was possible for such proposals to be made, what meanings and 'deep conceptual premises' (Bacchi 2009, p. xix) needed to be in place for such proposals to emerge, and what this could reveal about inherent power relations. … may result from ineffective learning and teaching 'Risk and health and safety assessments should also be carried out to determine any potential concerns arising from the child or young person's behaviour, and should identify any steps deemed necessary to support the child or young person in preventing harm to themselves or others (22) … is a threat to community safety Table 3. Purpose of DfE (2017)  'This document from the Department for Education provides a guide to the legislation that governs the exclusion of pupils from maintained schools, pupil referral units (PRUs), academy schools (including free schools, studio schools and university technology colleges) and alternative provision academies (including alternative provision free schools) in England' (3) 'This statutory guidance has been updated in a small number of areas, in particular to provide greater confidence to head teachers on their use of exclusion and to provide greater clarity to independent review panels and governing boards on their consideration of exclusion decisions' (6) 'The purpose of this guidance is to refresh national policy on the prevention of and the management of exclusion from school in the context of national government and education authority responsibilities and desired outcomes for children and young people' (5) 'This guidance is an opportunity to refresh and realign the use of exclusion in learning establishments. It includes a refreshed focus on prevention, early intervention and response to individual need in line with the principles of Getting it Right for Every Child (GIRFEC). It emphasizes the need for learning establishments to place a greater importance on inclusion through effective learning and teaching; promoting positive relationships and behaviour; and employment of preventative approaches which reduce the need to consider exclusion' (5)

[Bold emphasis added]
Turning first to the Scottish guidance, the presupposition seems to be that challenging behaviour is not simply a 'within child' problem, and that preventing school exclusion requires a holistic approach, and recognition that 'all behaviour is communication' (SG 2017, p. 22). In contrast, despite some, limited, recognition of the known impacts of unmet needs, the English statutory guidance does not seem to have shifted too far from the 'individualised discourses of behaviour and responsibility' identified by Dunn (2015, p. 285) prevalent 15 years ago. Interestingly, the undercurrent of neoliberal responsibilisation (O'Malley 2009) also emerges in Scottish policy, where we noted that one of two sets of grounds for exclusion states that a school may exclude where the 'parent of the pupil refuses or fails to comply, or to allow the pupil to comply, with the rules, regulations, or disciplinary requirements of the school' (SG 2017, p. 31). As of 4 August 2021, the Independent Provider of Special Education Advice (IPSEA) listed on its website that it is not permissible to exclude a student in England for 'something which their parents did or did not do'.
In order to examine this seeming paradox further, future research may benefit from a thorough analysis using WPR Q3 and Q6 (see Table 1).
WPR Q4: what is left unproblematic? Where are the silences? Can the 'problem' be conceptualised differently?
In WPR Q4 the focus moves more specifically to detecting what the problem representations do not say. It builds to some extent on WPR Q2 and allows the analyst to be inventive, to imagine a world where specific problems are reconceptualised and re-problematised and perhaps not thought about as problems at all.
In the English guidance (DfE 2017), attention to the influence of school related factors (e.g. pedagogy, curriculum, physical environment of the school) is largely missing. The guidance doesrecognise that 'there are certain groups of pupils with additional needs [those who are Looked After Children and those with Education, Health and Care Plans] who are particularly vulnerable to the impacts of exclusion' (DfE 2017, p. 11) and calls on schools to avoid permanently excluding students from these groups. The guidance also encourages schools to engage and co-operate 'proactively' with parents, 'foster carers or children's home workers, the local authority that looks after the child and the local authority's virtual school head' (DfE 2017, p. 11). However, it is worth noting here that this is the only reference to the impacts of exclusion. Headteachers have the power to exclude.
'A decision to exclude a pupil permanently should only be taken: . 'in response to a serious breach or persistent breaches of the school's behaviour policy; and . where allowing the pupil to remain in school would seriously harm the education or welfare of the pupil or others in the school' (10) 'It is unlawful to exclude for a non-disciplinary reason' (9) Power to exclude rests with the Local Authority who can delegate to schools.
Regulation 4 provides that an education authority shall not exclude a pupil from school unless the authority: . 'are of the opinion that the parent of the pupil refuses or fails to comply, or to allow the pupil to comply, with the rules, regulations, or disciplinary requirements of the school'; or . 'consider that the circumstances to allow the pupil to continue his attendance at the school would be likely to be seriously detrimental to order and discipline in the school or the educational well-being of the pupils there' (31) 'Children and young people must not be sent home on an 'informal exclusion' or sent home to 'cool-off'' (25)

[Bold emphasis and italics added]
In contrast, the Scottish guidance refers directly and frequently to the positive effect of school connectedness, strong teacher/pupil relationships, and the need for curricula input that teaches pro-social skills. It notes the negative impacts of exclusion, particularly within those groups of pupils where exclusions are more prevalent: those with an assessed or declared disability; looked after children and young people; children and young people from deprived areas; and those with an additional support need (particularly if that support need is social, emotional and/or behavioural). (SG 2017, p. 16) It provides a checklist of questions to ask and actions to be taken before proceeding to exclusion, for example: 'Has the incident that precipitated the consideration of exclusion been reviewed with all staff who were present to explore fully what happened?' And 'has another professional from within the school who is not directly involved, been consulted on the situation in order to provide a different perspective?' (SG 2017, Annex C, p. 53). Though there is a '[t]hings to consider' list for head teachers in the English guidance which asks: 'Did I consider factors that could have contributed to the pupil's behaviour (e.g. SEND [special educational needs and disabilities] or bereavement) and have I taken these factors sufficiently into account?' (DfE 2017, p. 51), again we note that there is no mention of how school factors may contribute to the pupil's behaviour, and few examples of individual circumstances to be taken into account. Overall, our analysis indicated that though the English guidance does make reference to 'early intervention', 'multi-agency assessment' and assessing the appropriateness of provision and support for pupils with special educational needs and disabilities (DfE 2017, p. 10) the Scottish guidance provides far more practical information and signposting to services to help with preventing exclusion than the English guidance.
We felt WPR Q4 worked well and suggest that future research may find it similarly helpful to capture not only the number of times each identified representation appears in a text as a way to assess the prominence and dominance of presuppositions, but also the extent to which policy solutions are operationalised and expected to connect to practice, actors and actions. The comparative nature of the analysis was also a notable strength in this part of the analysis, helping to reveal and instantiate silences which may not have been noticeable in a single document analysis, for example the lack of consideration of school level influences on exclusion in the English guidance.
WPR Q5: what effects are produced?
The aim of WPR Q5 is to consider the political implications of how particular problems are represented, asking: What effects (discursive, subjectification, lived) are produced by this representation of the 'problem'? According to Bacchi (2009), discursive effects set boundaries around what can be recognised as relevant. Subjectification effects denote the subject positions that are made available within particular discourses/knowledges, e.g. 'welfare recipient', 'higher education student', 'asylum seeker'. Lived effects capture the material 'impact of problem representations on people's embodied existence' (Bacchi 2009, p. 70). While identifying the three categories of effect it is important for the researcher to consider who is likely to benefit or be harmed by the problem representation and question how 'the attribution of responsibility for the 'problem' affect[s] those so targeted and the perceptions of the rest of the community about who is to 'blame'?' (Bacchi 2009, pp. 70-71).
Our analysis of WPR Q4 suggests that silences and absences have discursive effects, in the same way as presence, setting boundaries around what counts as worthy of consideration when dealing with a student who is presenting challenging behaviour, and deciding on a course of action. The lack of focus, for example on school related factors in the English guidance (DfE 2017) may impact on the extent to which teachers are likely to turn the lens on themselves, their own identities, pedagogy, curriculum, school environment, systems and policies. Though we see ways in which both the English and Scottish government documents produce subjectification effects where students are able to occupy the subject position of 'student in need', our analysis suggests that the English guidance opens up more room for a student to enter other positions such as being seen as 'harmful' (see, for example, criteria two in Table 4). The dividing effects seem to be less prominent in the Scottish guidance than in the English equivalent, because of its stated aim of keeping ' … all children and young people fully included, engaged and involved' (SG 2017, p. 7). The ways in which behaviour is understood, and students are positioned and categorised , have 'effects in the real' (Bacchi 2009, p. 18) for both students and their teachers. In the current case, the higher exclusion rates witnessed in England may be understood as the material or lived effect of the arguably more punitive school exclusion policy in place in England, though further research is needed to confirm this. The identified silences and effects identified in WPR Q4 and Q5, point to potential areas for policy development and change, and encourage us to think differently about how challenging behaviour should be understood and what the solution to the problem should be.
Step 7: positioning ourselves within the analysis An important consideration throughout the study was the impact of our own positionality on our analysis and the problem representations we identified. Werecognise that what we were able to see in the texts was filtered by the parameters we had set, the theory we had chosen, but also our own personal and professional backgrounds and beliefs (Fairclough 2003). One of the authors, for example, had been involved in the development of the Scottish Government (2017) guidance, which raised an interesting dilemma, given she was therefore analysing problem presentations which were already partially her own. Bacchi's approach provided a response to this challenge in step 7, by asking the researcher to apply the list of questions to their own problem representations (Bacchi and Goodwin 2016). Moreover, the intentionally iterative nature of the analysis and the involvement of two researchers, with differing levels of familiarity with the texts and broader education systems (being based in different jurisdictions), in the analysis, proved helpful for interrogating and challenging the presuppositions and positions of both researchers. Nonetheless, bias is likely to remain. A further round of reflection was also prompted by the comments received from the reviewers on our initial submission to this journal, and it may be helpful in future studies to build this form of external review into the analysis process from the outset.
Overall, we came to the view that following Bacchi's method allowed us to take a considered and methodical approach to our analysis, aiming to ensure that, as far as possible, prior assumptions could be opened up to view and limitations acknowledged.

Discussion
In this small-scale study we set out to explore the usefulness of Bacchi's WPR approach for comparing how the 'problem' of challenging behaviour in schools is represented in national school exclusion policies in England and Scotland, and the potential effects of differing policy discourses.
Using WPR, we have been able to identify overlaps in the problem representations in the two policy documents we analysed (DfE 2017, SG 2017), for example in terms of the representation of challenging behaviour as potentially resulting from unmet needs, while also showing that the heft of such representations differed significantly. Like Macleod (2014), referred to earlier, we found that a holistic, welfare-based, approach focused on prevention and intervention permeated the Scottish guidance more deeply. We feel confident that WPR analysis was able to facilitate our aim of comparative analysis allowing detection of meaningful differences in the policies of England and Scotland. This analysis has suggested that Scotland's welfare-based approach is more than mere 'political spectacle' (Edelman 1988). It has allowed us to demonstrate that Scottish exclusion policy, in contrast with that of England, can be seen to attempt to challenge and support schools whilst also recognising the complexity of many young people's lives. This aligns with Bacchi's (2009, p. 44) goal to 'take the side of those who are harmed … to intervene to challenge problem representations that have these deleterious effects, and to suggest that issues could be thought about in ways that might avoid at least some of these effects'. Our findings suggest that English policy may limit the capacity of schools to approach discipline in proactive ways. While the English guidance explains how to do school exclusion, in terms of the legal process, unlike the Scottish guidance, it does not provide advice on how to do early intervention or prevent exclusions (see also de Friend 2019).
While we have speculated that the policy differences identified in England and Scotland may go some way to explaining the disparities in permanent exclusion rates between the two jurisdictions, further micro level analysis is needed to explore how national policy is enacted at the local level. As noted earlier, Bacchi's (2009) WPR approach is not concerned with identifying potential gaps between policy and practice, and therefore this would require WPR policy analysis to be combined with other forms of data collection and analysis (see, for example, Clarke 2019). This is not intended as a criticism of WPR per se, but it is a point of note. At a more practical level, it is also worth noting that WPR proved to be very labour intensive. While it would be desirable to examine a larger data corpus than we were able to in this study, others may wish to test out the viability of the approach on a small number of documents first.
Notwithstanding these potential disadvantages, we came to the view that WPR worked well as a method for conducting comparative critical policy analysis of national school exclusion policies in two neighbouring jurisdictions with contrasting school exclusion rates, and helped us to build on our previous research. WPR enabled us to render visible some key aspects of policy which had previously been submerged or obscured, and to identify some of the major differences, imbalances and gaps in these policy approaches to school exclusion in two neighbouring jurisdictions. It allowed us to highlight contrasting problematisations of challenging behaviour in national policy discourse in the two jurisdictions, and to isolate potential system-level factors that lie behind the disparity in permanent school exclusion rates in the UK. It has brought clarity to the need for two key lines of further research, firstly the need to extend this current analysis to encompass a larger corpus of policy documents, and the other UK jurisdictions; and secondly, to examine how best to combine Bacchi's WPR approach with other methods to explore how policy discourses mediate school exclusion processes in practice.

Conclusion
In this paper, we were guided by two objectives: (a) to offer an illustration of WPR in education as a contribution to methodological debate, and (b) to critically appraise the WPR approach, and its potential to acknowledge, challenge and disrupt normalising discourses within one key area of school-based policyschool exclusion. We have sought to illustrate the stages of our analysis, to 'show our working', in light of the paucity of previous work which has done this. In summary, we would argue that our study demonstrates that the WPR approach can provide a new and effective way to undertake comparative analysis; that this approach can be seen to point clearly to imbalances and gaps in different policies in ways that help to begin to problematise and reconceptualise problem representations. While acknowledging that it is not enough to examine school exclusion in isolation from the many other moving parts of education and its construction of challenging behaviour, WPR has helpfully isolated some key elements of discourse here for closer examination.
In conclusion, we would suggest that our small-scale study indicates that Bacchi's (2009, p. xix) framework provides a systematic approach to interrogate in a new way the 'deep conceptual premises' on which different policies are based, and the effects of different problem representations. The WPR approach also has the distinctive benefit of building reflexivity into the analysis process, and is well suited to comparative studies. Our worked example has revealed some important practical points to consider before adopting WPR more generally, but it has, in our view, provided evidence overall of a promising new method of analysis of school exclusion policy, with the potential to offer new avenues of interrogation of policy within the field of education more broadly.