Student perceptions of smoke-free school policies in Europe – a critical discourse analysis

ABSTRACT Smoke-Free School Policies (SFSP) are primarily designed to ensure educational spaces remain free of second-hand smoke, whilst contributing to a reduction in adolescent smoking by challenging the practice per se. Evidence regarding the latter goal is inconclusive, however, with most studies suggesting SFSPs are ineffective in reducing smoking prevalence. A dearth of qualitative research limits our understanding of why this inefficacy persists and how it might be addressed. This paper addresses this lacuna through a critical discursive analysis of data from 56 focus groups, generated with adolescents across 17 schools in seven European cities. It reveals that, while smoking is banned on school premises in most European countries, young people experience wide variation in implementation. Despite this, participants framed SFSPs in remarkably similar ways. Among young people most likely to smoke, representations of SFSPs often undermined their efficacy, leading to the displacement of smoking (outside the school grounds) rather than a reduction in prevalence. We argue that, policy effectiveness could be improved if schools worked collaboratively with students to develop positive collective beliefs and understandings about SFSPs, but caution realism about the potentially limited power of schools to reduce adolescent smoking prevalence.


Introduction
Reducing adolescent smoking is a global public health priority. In Europe, a range of youth smoking prevention policies have been implemented of which Smoke-Free School Policies (SFSPs) are a key element (Papanastasiou et al., 2018). The goals of SFSPs are twofold, they aim: to prevent smoking on school grounds (thereby ensuring students are educated in smoke-free environments); and, ultimately, to reduce adolescent smoking prevalence by challenging smoking as a practice per se (Schreuders et al., 2017;Trinidad et al., 2005). While there is evidence that well-enforced SFSPs can reduce smoking at school (Kuipers et al., 2016), their effectiveness in reducing adolescent smoking prevalence generally is contested, with studies showing mixed findings (Galanti et al., 2014). It is unclear which aspects of SFSPs are most relevant in reducing adolescent smoking prevalence, and how these might be strengthened to improve the policy's broader goal (Lee et al., 2007;Piontek et al., 2007).
Previous research has highlighted several factors that may undermine SFSPs' effectiveness in reducing adolescent smoking. These include: visible teacher smoking (Murnaghan et al., 2009;Piontek et al., 2007;Trinidad et al., 2005); weak or overly punitive enforcement (Pentz et al., 1989; • How are SFSPs interpreted and represented by adolescents most likely to smoke? • How might these interpretations and representations mediate the impact of SFSPs on adolescent smoking prevalence? We draw data from 56 focus groups (FG) undertaken by the SILNE-R consortium; a multi-institutional research project exploring adolescent targeted tobacco control in seven cities in Europe (SILNE-R, 2018). Using the analytical lens of critical discourse analysis (CDA), we conclude that even well enforced SFSPs may struggle to influence youth smoking rates because of the way they are interpreted and represented by students.
In what follows, we outline the central tenets of CDA, before discussing our methodological approach. Thereafter, we discuss our findings, exploring the plurality of narratives which contribute to student understandings of SFSPs.

Critical discourse analysis (CDA)
CDA is premised on the ontological claim that, whilst an objective world exists, it cannot be straightforwardly known. Rather, subjects are involved in collaborative meaning making: conceptualising, representing, and subsequently contributing to the creation of social worlds (Fairclough, 2010). Importantly, localised communicative acts are dialectically related to externalities, including practices, institutions and expressions of power (e.g. ideology) (Fairclough, 2010). Consequently, researchers should situate localised communicative acts in the conditions of their production, tracing the relationships which shape spoken/written texts. Given this, CDA presents a sociologically valuable lens through which to explore SFSPs by reference to their social context.
Discursive analytical methods are increasing within tobacco control (Gilbert, 2008;Lamerichs et al., 2009;Thompson et al., 2007;Triandafilidis et al., 2017). Many studies utilising this approach seek to describe localised narratives or identify discursive strategies and effects, framing spoken texts as relatively atomised (Lamerichs et al., 2009). This approach highlights the 'work' speakers can do, with discourse, in localised spaces, but largely fails to address the role context can play in discursive formations. A limited (albeit growing) public health literature offers a more critical socio-political analysis (Gilbert, 2008;Triandafilidis et al., 2017). This paper contributes to this literature.
In accordance with CDA, we understand narratives regarding SFSPs -espoused within our focus groups -by reference to the materiality and practice of school, as well as legal discourse and liberalism. We highlight how pluralistic, contradictory, and complementary narratives together create representations of SFSPs, school and the self. We posit that these policy interpretations will guide responses to them (Piontek et al., 2007). Importantly, CDA frames the subject as (primarily) formed through, rather than as an active user of, discourse (Bacchi, 2000;Fairclough, 2010). Thus, when we speak of participants 'drawing on' dominant discursive trends, we do not suggest that they do so strategically, but rather pre-reflexively and in ways which reflect their immersion in social spaces.

Participants and recruitment
During 2016-17, seven SILNE-R research teams conducted 56 FGs with 319 adolescents aged 14-19. Participants were recruited from 17 schools located in seven European cities: Namur (Belgium), Tampere (Finland), Hannover (Germany), Dublin (Ireland), Latina (Italy), Amersfoort (the Netherlands) and Coimbra (Portugal). These cities were chosen as they are median-sized and approximated national averages in terms of socioeconomic demography and non-foreign population. The schools were selected to represent the different school types in each country (e.g. in Germany a 'Gymnasium' and a 'Hauptschule') and catchment areas with different socioeconomic status (SES). Participants were sampled purposively: teachers identified students whom they perceived to be smokers or at risk of becoming smokers (i.e. known to have smoking friends/family members), from classes of predominantly 15-year-olds. This is a well-established method for selecting participants in qualitative research on adolescent smoking (Sluijs et al., 2016). To minimise selection bias and coercion, researchers briefed teachers regarding the importance of ethical recruitment e.g. ensuring students participated voluntarily and were treated with dignity and respect. Half the FGs were with girls and half with boys; half involved schools serving predominantly low SES populations, and half predominantly high SES populations. This aspect of our purposive sampling was adopted as gender and SES can mediate perceptions and/or experiences of smoking (Amos & Bostock, 2007;Fletcher & Bonnell, 2013;Hefler & Chapman, 2015). In general terms, this observation was born out in our data. However, as discussed in more depth below, participants represented SFSPs in remarkably similar waysregardless of gender, SES, and location. Our analysis therefore does not focus on these distinctions.
A mini-questionnaire collecting basic participant demographic data was distributed at the beginning of each FG. Half the participants identified as current or ex-smokers and 43% as never smokers, although several self-identified 'never smokers' subsequently reported having smoked during FG discussions. Smoking status was not recorded for 7%. Ethical approval was obtained from all the institutions where research teams were based. Participants' parents were sent opt-out consent forms in all sites except Hannover, where the school required opt-in consent from parents (a standard practice in German schools). Participants were given an information sheet and asked to provide affirmative consent prior to their involvement in FGs, i.e. an opt in approach.

Data collection
Local research teams received facilitator training at a workshop (led by AA and SH), where they were familiarised with a collaboratively generated topic guide covering smoking experiences, familial and peer smoking, tobacco access, and smoking at school (including SFSPs). The topic guide was initially developed by the authors, received comments from the local research teams, and finalised during the workshop. Eight FGs of 3-9 participants were held in each of the seven cities in school but without school staff present. FGs were conducted, recorded, and transcribed in local languages before translation into English for analysis by the authors.

Analysis
In what follows we explore how participant narratives regarding SFSPs were dialectically related to (and shaped by) the materiality and practices of their schools, discourses of law, and the liberal conceit of individualism. Our findings highlight how these narratives might influence the effectiveness of SFSPs. We present our analysis as one possible interpretation of available data, which provides a fresh vantage point from which to consider the nature and efficacy of SFSPs.
We performed an immersive read of 24 transcripts and subsequently created a coding framework by deductively identifying salient themes within relevant theoretical and substantive literature and inductively identifying themes emerging from the data. This framework was iteratively developed following an immersive reading of the remaining transcripts. Transcripts were uploaded to the qualitative analysis software NVivo and coded by RH. Data were read critically, with a view to exploring the social and political conditions of discursive production. Common and recurring narratives were explored in light of the various practices, institutions, and power relations observable across field-sites.
In what follows, we briefly describe the policy and school environments that contextualise our data, before discussing our most salient findings. Quotes refer to country, sex, SES status and facilitator (F) or participant (P).

Context
All field-sites, apart from those in Germany and the Netherlands, had national legislation prohibiting smoking on school premises (WHO, 2017). Lower Saxony (where the German schools were located) had a state-level ban (Smoke-Free Partnership, 2018). However, there is some legislative ambiguity regarding what constitutes school premises, and our findings suggest significant variation regarding where and when SFSPs were enforced, how, and against whom. While participants from some schools (e.g. Irish, high SES) reported heavily enforced SFSPs, prohibiting smoking on all school grounds and peripheral areas, others indicated that their schools (e.g. Belgium, low SES) barely enforced SFSPs, effectively permitting smoking on school grounds and in buildings. Most participants reported policy implementation somewhere between these extremes -e.g. SFSPs enforced on school grounds, but not in school peripheries (e.g. Portugal, high and low SES). In general, participants' accounts echoed previous research which indicates that SFSPs enforced consistently by all teachers, against all students, in all school areas, prevented on-site smoking; whilst poorly enforced SFSPs did not (Schreuders et al., 2017).
However, despite this significant material variance -and the significant variation in students' social and geographic positions -participants from across field-sites represented and evaluated SFSPs in normatively similar ways. In what follows, we demonstrate this striking consensus, by exploring a range of findings regarding: the perceived purpose of SFSPs, the jurisdiction of 'school', personal autonomy, considerate smoking, and the age of capacity. We highlight how materiality, practice, and dominant ideological/discursive trends (e.g. liberalism) could be understood to exist in dialectic relationship with, and thus formative of, participant perspectives.

The Purpose of SFSPs
While a handful of participants across field-sites suggested that some teachers (particularly those involved in pastoral care) might enforce SFSPs to protect student health, most believed that SFSPs served less compassionate means. Many participants from across field-sites speculated that, when implementing and enforcing SFSPs, teachers were largely motivated by a desire to protect their school's reputation.

Ireland, girls, high SES
As these excerpts demonstrate, participants believed smoking was banned not because it was problematic per se, but because it could harm a school's reputation. This can be interpreted in two (arguably interrelated) ways: either students believed their schools feared judgment merely for failing to enforce the rules, or they believed schools feared stigmatisation because of student smoking. Certainly, and in concert with previous research (Turner et al 2006;Brown et al 2020;Fletcher & Bonnell, 2013), numerous participants indicated that, within schools, smoking was stigmatised -associated with poor educational and behavioural outcomes, by either teachers, peers, or themselves: P: The problem is that when they talk about it, you're all of a sudden framed. Well, not to that extent, but when correcting exams, they'll be like: "these are smokes and delinquents" . . ..

Ireland, boys, low SES
These excerpts indicate that smoking has been stigmatised in these spaces i.e. marked as undesirable (Triandafilidis et al., 2017), and may illuminate why participant's felt schools would 'look bad' if they failed to enforce SFSPs. The stigmatisation of smoking exists ambivalently with smoking behaviour, however, as young people may negotiate smoking in diverse ways (e.g. desisting or smoking covertly) to avoid projecting morally questionable identities (Barnett et al., 2016;Triandafilidis et al., 2017). Furthermore, stigma can deepen pre-existing social schisms. In countries where smoking prevalence is higher in marginalised communities, the burden of stigma falls heavily on the socially excluded (Graham, 2012). This can further polarise marginalised groups (spatially and socially), creating 'smoking islands' which normalise or reinforce smoking (Thompson et al., 2007). Thus, if students understand SFSPs are motivated by a desire to avoid the stigmatisation of schools, this could prove counterproductive, particularly for those vulnerable to smoking uptake/continuation. In short, it may encourage rather than discourage smoking and ingrain health inequalities (Barnett et al., 2016;Graham, 2012;Johnston et al., 2012).
Building on this, several participants argued that teachers were merely 'following orders' when enforcing SFSPs: acting out of an obligation to do so rather than a genuine commitment to their goals.

Portugal, boys, low SES
Again, this representation of SFSPs weakens challenges to smoking per se, problematising smoking where and when it is 'against the rules'. Representations of SFSPs as in the interests of reputation and rule-following occlude public health motivations, framing these policies as arbitrary, authoritarian, and moralistic, and potentially undermining their capacity to challenge smoking per se.

Bounded jurisdiction
Another dominant narrative regarding SFSPs pertained to the 'jurisdiction' of a school and its staff. Many participants, across sites, argued that a teacher's responsibility or 'right' to prohibit smoking was temporally and spatially bounded.

Belgium, girls, low SES
The proposition that school-staff might intervene outside the scope of their jurisdiction occasionally elicited an emotive response.

Netherlands, girls, high SES
This arguably demonstrates a particular conception of what constitutes 'school', as well as a normative judgment regarding the relationships 'school' should engender. School, according to many participants, is a socio-spatial entity made up of buildings, fences, and grounds. The jurisdiction of school staff ceases as one crosses a physical border e.g. the school gates. Conversely, school was sometimes conceived in temporal terms with teacher authority perceived as limited to 'school hours'. In both instances, it is evident that physical materiality and institutional practices exist dialectically with participant representation of institutional authority. Moreover, when taken together, these understanding of jurisdiction may work to displace smoking, rather than to challenge it.
These conceptions also appeared to guide participant behaviour. Spatial interpretations of school jurisdiction frequently appeared to create contexts in which students congregated outside the school gates to smoke -a practice compounded by a lack of enforcement in these areas. Consequently, several participants represented their schools as conducive spaces for smoking. Take for instance, this excerpt from a Portuguese participant, reflecting on his experience of a school which enforced SFSPs on school grounds, but did not police the immediate peripheryresulting in considerable peripheral smoking.
I thought that since I've switched to another school, I thought that in this school things were different, that my colleagues wouldn't smoke, that I was going to reduce, but after all when I got here everybody smoked and well, usually . . . in order not to be alone . . . I would go with them . . . so I smoke.

Portugal, boys, low SES
This demonstrates how some areas -often just outside the school sites -can become 'smoking islands', 'spatialized outcomes of behaviour regulation' (Barnett et al., 2016, p. 134) which resist attempts to challenge smoking by nurturing social networks and transgressive smoking subcultures. Participants indicated that these islands encouraged them to smoke more at school than they did at home (where smoking was usually condemned). Some schools prohibited peripheral smoking, however, usually leading to a dispersal of smokers -occasionally creating more distal islands.
In contrast, 'home' was not represented as temporally or spatially bounded. Numerous participants described going to great lengths to avoid parental censure regarding their smoking, in ways which transcended spatial and temporal demarcation. This was typified eliminating traces of thirdhand smoke prior to seeing parents. Take, for instance, the following excerpt, in which one participant from Ireland described their first cigarette. P: There was this girl who lived like up the road and she's so scary . . . I was so scared of her, and we were out one day and she '

Ireland, girls, low SES
This differentiation between teachers and parents, home, and school is illuminated by reference to the public/private spheres binary. This ostensibly liberal dichotomisation is used to demarcate the proper reach of the state: whilst public bodies are required to desist from paternalistic interventions, private relationships can function according to agreed normative standards (e.g. parental expectations). Through this lens, we can understand 'school' as engendering public relationships, subject to limitations ascribed to similar governing bodies. Thus by failing to address smoking beyond the school gates, SFSPs could neglect those most at risk of smoking (e.g. those permitted to smoke at home) and subsequently reinforce inequalities. By placing the 'private sphere' beyond the reach of institutional authority, SFSPs may fail to respond to the politicality of the personal. This may be compounded by the primary mechanism via which schools tended to sanction students found violating SFSPs, i.e. informing their parents. This approach -which is arguably designed to mitigate school limitations by drawing on the power of the 'private sphere' -only works if parents disapprove of, and react to, knowledge that their children smoke. This was not always the case, as some participants explained:

Italy, girls, low SES
Moreover, not all parents feel equally able to develop a positive relationship with their child's school, particularly socio-economically and ethnically marginalised parents (Kainz & Aikens, 2007). This may compound the disadvantage of adolescents already occupying less privileged social spaces. By reading interpretations of school and family jurisdiction as dialectically formed through the liberal dichotomisation of spheres, we can develop a sense of how dominant (liberal) ideological trends might function to shape participant representations and behaviours.

The autonomous self
In a similarly liberal move, which echoes findings from previous research (Turner, 2004a;Unger, 1999), smoking participants routinely justified their smoking behaviour by reference to the normative significance of their autonomy.

Ireland, boys, high SES
Participants suggested that their school lacked the authority to impede their autonomous decision to smoke. This conceptualisation of the 'self' can be best understood by situating it within prevailing ideologies of individualism and the related notion that personal choice is sacrosanct (Buchanan, 2008;Rajczi, 2016). This perspective ordains that rational individuals be minimally governed so long as they do not harm others (Buchanan, 2008;Rajczi, 2016). Even when the exercise of autonomy compromises one's physical self, the individual should remain free from paternalistic intervention. By invoking concepts like 'free will' and bodily ownership, participants appear to reproduce this highly individualised, fundamentally liberal, conception of the self.
Discourses of individualism informed a broad array of participant representations, which extended beyond discussions of SFSPs. Narratives of personal choice were invoked in discussions of smoking uptake and social relationships, obfuscating structural and social determinants of smoking. This prompted some participants to articulate seemingly contradictory narratives, simultaneously demonstrating consciousness of peer influence, while remaining committed to individualistic accounts of smoking.

Netherlands, girls, low SES
Moreover, reverence towards autonomy spread beyond conceptions of the self and influenced how participants represented their relationships with others. Many non-smoking participants claimed to respect their peers' 'right' to smoke, disavowing acts of interference.

Finland, boys, high SES
P: If someone really wants to do it and they are still a nice person, then I'd probably leave them alone. Because, you know, it's legal. If it were against the law, that would be different, but this kind of thing is legal. P: To me it's not that big a deal, because they don't force me to smoke. It's their thing. As long as I'm not hurt or affected by it.

Germany, girls, high SES
Discourses of autonomy and personal choice undermine the capacity of SFSPs to challenge smoking as a practice per se, as they challenge the right of public bodies (e.g. schools), to intervene in individual behaviours undertaken in ostensibly private spaces. In this sense, discourses of individualism complement, and are complemented by, the similarly liberal dichotomisation of spheres and the subsequent commitment to the bounded authority of schools. This is no accident, but rather reproduces liberalism as an ideology of limited states and 'free' citizens.

Considerate smoker
In keeping with liberal conceptions of autonomy -which stress the inviolability of the bodyparticipants routinely framed their smoking as acceptable only if it did not harm others. This framing often manifested in the adoption of a 'considerate smoker' narrative, a much theorised discursive trend (Poland, 2000;Thompson et al., 2007).

Ireland, girls, low SES
Here, participants argued that smoking should take place off school premises in the interests of nonsmoking students. These discourses simultaneously support the SFSP goal of purifying school spaces, whilst undermining their aim to challenge smoking. This tension is reflected in a growing literature addressing considerate smoker narratives, which argues that they reflect both a form of compliance with, and resistance to, smoking regulation (Barnett et al., 2016;Poland, 2000;Thompson et al., 2007). Individuals internalise regulatory norms and self-govern (e.g. remove themselves from smoke-free spaces) whilst simultaneously forming new, resistant subjectivities (e.g. smoking elsewhere). Consequently, smokers draw on considerate smoker narratives to represent themselves as responsible citizens, whilst engaging in behaviours represented as irresponsible (Poland, 2000). By articulating these narratives, participants demonstrate a continued negotiation of their liberal subjectivity in a way which supports ongoing smoking practices.

Legality and age
In narratives which similarly supported the SFSP aim to purify educational spaces of smoking, whilst undermining the problematisation of smoking per se, many participants argued that they should not be permitted to smoke on school grounds because they were underage.

Finland, boys, low SES
Participant understandings of SFSPs appeared to exist dialectically with legal norms regarding the age of majority, if not dogmatic law (age is used to regulate who can purchase cigarettes not smoking). This prohibitive perspective did not extend to smoking in general, e.g. participants did not feel their age should prevent them from smoking elsewhere. This representation of SFSPs undermines their potential to influence smoking per se, as it problematizes under-age smoking only, at least in or near school. The use of age as a classificatory boundary again arises from liberal conceptions of the rational individual, where one's autonomous desires should be respected unless compromised by a lack of capacity e.g. childhood (Rajczi, 2016). Some participants appeared to support the proposition that age undermined capacity, albeit only in particular contexts, arguing that younger adolescents lacked capacity to choose or the normative right to do so. P: Yeah, I already saw some people, little ones who were ten years old. They were in a group and started smoking, at ten . . . P: He's not even done shaping up and he's already smoking . . . P: It's a bit disappointing, even. They're young and they're already smoking and destroy their health.

Portugal, boys, high SES
These excerpts demonstrate how, rather than simply rejecting age as a legitimate category of discrimination, participants sought to 'move the goal posts' -setting the age of capacity significantly lower than legal norms. This represents a similar combination of acquiescence and resistance exhibited via the adoption of a considerate smoker status: it accepts the prevailing legal and liberal view (e.g. children lack capacity) while subverting its full realisation in ways that permit continued smoking (e.g. I'm not really a child). Thus, it demonstrates the complex dialectic relationship between legal discourses, dominant ideology, social practices, and local communicative acts.

Discussion
Based on analysis of FG data from 17 schools in seven European cities, we explored the diversity of narratives contributing to and reflecting conceptions of SFSPs amongst adolescents most likely to smoke. Drawing on the principal tenets of CDA, we argued that these policies existed in a complex web of dialectic relationships with materiality, prevailing ideologies, legislation, and school practices. We demonstrated that -regardless of differences in enforcement, gender, SES and geographic localeadolescents in these European cities interpreted and represented SFSPs in remarkably similar ways.
Participants conceived of teacher authority as bounded by the spatial materiality and temporal practices of school. Moreover, the motivations participants ascribed to teachers in enforcing SFSPs framed them as arbitrary -an exercise in rule following, motivated by reputation control. We have argued that participant understandings of SFSPs as associated with 'reputation control' may stem from a broader stigmatisation of smoking -the reproduction of which can produce and reinforce inequalities. Certainly, ample evidence suggests smoking was stigmatised within schools. In response to these findings, and in concert with the tenets of CDA, we have argued that socio-spatial and temporal aspects of school, the dichotomisation of spheres, and normative narratives regarding hypocrisy and discrimination are dialectically related to, and formative of, adolescents' narratives. Consequently, such narratives undermine the capacity of SFSPs to challenge smoking per se.
We also explored how participants conceived of themselves in relation to SFSPs. Participants sought to assert their right to choose, so long as they did not compromise the health rights of others. Furthermore, participants referenced the age of majority as a legitimate measure of who should smoke, whilst rejecting the notion that all under-age adolescents lacked capacity. Here, participants exhibited both compliance with and resistance to prevailing ideological and legal norms, drawing on liberal arguments regarding harm, autonomy, choice and capacity -as well as smoke-free regulatory frameworks -to make sense of adolescent smoking at school. The resultant narratives produced responses to SFSPs in ways which partially endorsed their goal of creating smoke-free educational spaces, while undermining their goal of reducing smoking prevalence by challenging smoking per se. Insofar as interpretations are understood to be behaviour guiding, we suggest that this provides some insight into why SFSPs can struggle to influence smoking prevalence among adolescents.
Accordingly, these representations of SFSPs may serve to displace -rather than holistically challenge and discourage -adolescent smoking. Our study has limitations. The relationship between SFSPs, their interpretation, and the behaviour they elicit are bound together in mutually productive relations, making it difficult to tease apart how one affects the others. Future research might consider qualitative longitudinal studies, spanning the implementation and enforcement of SFSPs, to explore these dynamics more closely. Furthermore, our study focused on adolescent smokers and those at higher risk of smoking. We therefore have limited insights about how SFSPs might influence adolescents at lower risk of becoming smokers. In addition, while we recruited a diverse range of adolescents across and within distinct geographical spaces, we only included participants in one city in each country. Thus, the generalisability of our findings is somewhat limited. However, our results arguably demonstrate analytic generalisability and allow for case transferability. By bringing rich and nuanced data, which demonstrated consistency across field sites, into dialogue with critical discursive analysis, we provided a compelling intellectual framework with which to understand this field of study. The relative uniformity of participant representations -regardless of gender, SES, locality or policy enforcement -provides compelling evidence that dominant discursive trends regarding school jurisdiction and individualism play a role in shaping responses to SFSPs in seemingly diverse spaces. Researchers working in similar fields may be able to apply our framework to explore the degree to which our higher-order conceptualisations shed light on their localised research settings (Polit and Beck 2010). In concert with others (Turner, 2004a;Unger, 1999), we would argue that the perspectives and attitudes of adolescents most at risk of smoking likely play a significant role in the impact and effectiveness of SFSPs. This highlights the need for future research to consider the broad socio-political conditions within which school policies are enacted.
Our analysis points to several practical steps that may improve SFSP outcomes: intervening in smoking which takes place peripherally to school grounds, communicating clear messages about the purpose of SFSPs, and considering how a student's home life may mediate their attitudes to schooltime smoking. Discourses of autonomy, consideration and jurisdiction may be more difficult to address -as they are formed by powerful discursive forces out-with 'school'. Schools might therefore consider deploying a critical pedagogical or participatory approach to challenge the discursive trends which limit their institutional reach and promote narratives which challenge smoking (Turner, 2004a(Turner, , 2004bUnger, 1999;Clark et al. 2002). A positive example of this approach is the Scottish 'Lothian Tobacco Free Schools Project' which aimed to implement and evaluate supportive, collaboratively designed and health-focused SFSPs (Lothian & Scotland, 2018). In adopting such an approach, schools could work to imbue SFSPs with more positive and health-salient messages, whilst simultaneously mobilising young people's autonomous capacities. By demonstrating respect for adolescent capacities, autonomy, and power, rather than working against these qualities, schools might be able to create more effective ways to influence student norms and behaviour. However, in light of our findings, we would caution some realism. As localised institutions, situated within a complex web of formal and informal governance structures, schools possess relatively limited power to challenge the influence of dominant ideological and legal norms or, indeed, the impact of familial relations -so often shaped by socio-economic factors. Whilst these more distal and macrosystemic forces persist, SFSPs may remain limited in their capacity to challenge young peoples' relationship to smoking per se.

Disclosure statement
No potential conflict of interest was reported by the author(s).