A content analysis of school leaders’ conversations about ‘off rolling’ on Twitter and its relevance to teacher education

ABSTRACT

This article seeks to illustrate the usefulness to teacher educators of small-scale qualitative social media content analysis by reporting a study of online microblogs (tweets) posted by senior school leaders’ relating to ‘off rolling’ (practices that bypass legal permanent exclusion procedures in English mainstream schools). The outlined method enables analysis of social media content in the absence of costly opinion mining software and its associated affordances and limitations are discussed. The article also outlines the implications of the study’s findings for teacher education. In this instance, it served to illustrate a disparity between different types of professional discourse which future teachers should be prepared to navigate in their teacher training. Given the illegality of ‘off rolling’ and an increasingly incontestable socio-political discourse around inclusive education, it was hypothesised that Twitter affords school leaders a forum for the articulation of views that rest uneasily with this discourse, current legislation and statutory guidance around off rolling. A tendency to discount ‘off rolling’ as a rare occurrence was found despite growing anecdotal evidence that it is viewed as a legitimate, if illegal, practice within some school cultures. This finding is relevant to teacher educators and future teachers who aspire to an inclusive education system.

KEYWORDS:

• Teacher education
• content analysis
• off rolling
• school exclusion
• social media research in education

Introduction

A substantial number of texts within the teacher education literature concern teacher attitudes towards inclusion or advice on the instilling of inclusive values in future teachers, for example, Warnes, Done, and Knowler (2021) and Florian (2014) respectively. Our investigations of off rolling suggest, however, that efforts in initial teacher training (ITE) to inculcate suitably inclusive attitudes, values and practices may not be preparing newly qualified teachers for the school cultures that they might subsequently encounter in English mainstream secondary schools. We have previously highlighted the issue of disproportionality (the over-representation of specific groups within the school population) in official exclusion data (DfE 2019a, DfE 2019b; Done and Knowler 2020a, Done and Knowler 2020b, Done and Knowler 2020c) and would maintain that this is an issue which should feature prominently in teacher education given its significance for future teachers interested in social justice and their practice. This article focuses on one study that has supplemented and informed the analysis of empirical data generated through survey-based investigation of illegal exclusionary practices involving senior school leaders, parents and Special Educational Needs Coordinators (SENCos) (Done and Knowler 2020a, 2020b, 2020c; Done et al. 2021). These studies have identified multiple definitions of what constitutes an illegal exclusionary practice and, therefore, varied levels of understanding of off rolling. The most commonly identified definition is that of the national school inspectorate (Ofsted 2019a) which assumes that schools place excessive emphasis on academic performance and exclude students to enhance their position in performance league tables. Other scenarios are less well understood even where in legal terms they constitute off rolling. These investigations indicate that off rolling is best conceived as a process rather than an event (triggered by, for example, the school’s handling of an incident of bullying), and that off rolling is far more common than efforts at quantification suggest (Ofsted 2019b).

The study of tweets by senior school leaders related to off rolling reported here of tweets by senior school leaders related to off rolling can also be framed as an exploration of Twitter’s value and relevance in researching sensitive topics within teacher education. As Carpenter and Krutka (2014) and Carpenter and Harvey (2019) note, while there is research on educators’ Twitter use, the limitations of that research and the complexity of online spaces means that much remains to be explored. This observation is especially pertinent where school practices are problematic or controversial, and where dialogue around such practices in staffrooms is limited and subject to self-censure because it contradicts statements linked to management of a school’s public image. Accordingly, social media platforms may offer an opportunity to gain insights into how educators discuss issues like exclusion and off rolling in less formal environments.

Anderson and Kanuka (2003, 174) argue that content analysis is an appropriate method for such e-research and, for Trimble and Treiberg (2015), it is a methodology that can generate valid, rigorous, reliable and replicable findings. These are generalised statements and, below, we consider the specific affordances and limitations of microblog research from technical, ethical and methodological perspectives. Despite potential limitations, we argue that the qualitative content analysis of microblogs offers teacher educators and future teachers that are research active a flexible starting point in preliminary investigation of under-researched areas of practice. In the study reported below, explanations of off rolling, and why it might happen, differed from familiar accounts, particularly those of Ofsted (2019a). A conceptual framework for data analysis was then chosen in response to the explanations found in the microblogs which implied systemic issues such as the accountability pressures experienced by senior school leaders following the marketisation of the education sector in England and globally (Ball 2003). Prior to outlining the affordances and limitations, and the ethical implications, of e-research involving microblogs and its potential usefulness in teacher education, the following section describes off rolling and the microblog study.

Off rolling

Ofsted has defined off-rolling as a form of ‘gaming’ that entails ‘the practice of removing a pupil from the school roll without a formal permanent exclusion or by encouraging a parent to remove their child from the school roll, when the removal is primarily in the interests of the school rather than in the best interests of the pupil’ (2019a, 50). The paucity of published research on this practice has led to media reliance on anecdotal evidence found in commissioned reports (Daniels et al. 2003;  Edwards et al. 2006) or in annual reports of bodies such as Ofsted (2019a), the Office of the Schools’ Adjudicator (OSA 2018) and, most recently, IntergratED (2020). Ofsted (2018, Ofsted (2019b) has striven to quantify the prevalence of off rolling and commissioned a survey undertaken by YouGov (2019) which found that schools may exaggerate the severity of behavioural difficulties in order to justify exclusionary practices intended to improve their academic performance data. Ofsted’s recent research on unregistered or illegal alternative provisions (Roberts 2018) suggests that departure from the roll of a mainstream school can result in some pupils eventually entering settings that are highly inappropriate. Non-governmental bodies such as the Education Policy Institute (EPI 2019) and IntergratED (2020) have highlighted the issue of disproportionality, finding that children classified as having special educational needs (SEN) are likely to be significantly over-represented in the sub-population of off rolled pupils just as they feature prominently in official school exclusion data (DfE (Department for Education) 2019a). Below, the methods used to estimate the prevalence of off rolling are briefly outlined, including the EPI’s (2019) breakdown of ‘unexplained exits’ from schools in England.

Prevalence

Estimates of the prevalence of off rolling are based on school census data acquired in October, January and May every year to record which pupils are on roll at which schools. Pupil moves are calculated through comparison of these termly school censuses. In 2017 approximately 553,000 pupils reached the end of their secondary education, having spent some or all of that time in a mainstream school. The vast majority, approximately 516,000, completed their secondary schooling in such schools. Others finished Year 11 elsewhere in the state system, for example, in special schools, alternative provision, or university technical colleges. However, many pupils could not be accounted for, and Ofsted was particularly concerned that 5,800 pupils with special educational needs and disabilities (SEND) left school between Years 10 and 11 (January 2017 to January 2018), stating that a significant proportion ‘may have been off rolled’ (2019b, 53).  When all Year 10 pupils were considered, it was found that 19,000 had left school during this period whilst 9,700 remained unaccounted for (2019b, 50), and it was this concentration of school leavers at this point in their education that prompted suggestions of ‘gaming’ (Ofsted 2019a, 50). Whilst in early years settings the ‘gold plating of regulations’ relating to health and safety may account for off rolling (Ofsted 2019a, 27), and the formal exclusion or off rolling of pupils with SEN across the age range may be attributable to schools’ failure to manage disruptive behaviour (50), Ofsted identifies the ‘pressures of performance tables’ as a key factor in off rolling by secondary schools (27).

Disproportionality

Statistical modelling now used by Ofsted permits identification of exceptional (statistically higher than expected) movements by school and pupil category, and pupils classified as having SEND are disproportionally removed from school rolls (Ofsted 2018). Disadvantaged pupils are similarly over-represented as a sub-category (Ofsted 2019a, 50). Evidence provided by local authorities to the OSA (2018, 35) indicates increases in ‘elective’ home education of up to 70% between 2016 and 2017 and, anecdotally, many such cases are described as ‘inappropriate’ responses to pressure from schools, implying that Ofsted’s figures underestimate the scale of off rolling. It seems likely that schools returning from Covid-19 lockdown conditions will continue this practice, albeit for more varied reasons than that referred to by the OSA (2018, 36) which assumes that ‘coerced’ home education is intended to enhance school performance data.

Analysis of ‘unexplained exits’ from school by the EPI (2019) revealed that pupils most likely to experience such an exit were as follows: those who had previously undergone an official permanent exclusion (1 in 3) or fixed term exclusion (1 in 5); pupils in contact with the social care system (1 in 3); those with a high number of authorised absences(approximately 2 in 5 of whom in the 2017 cohort had experienced at least one unexplained exit); pupils eligible for free school meals (1 in 7); those from black ethnic backgrounds (1 in 8); and those in the lowest prior attainment quartile (1 in 8).

Microblog study rationale

Given the data presented above, the reported study was premised on the occurrence of off rolling as a reality in educational practice with potentially undesirable or negative outcomes for parents and pupils. Opportunities for senior leaders to discuss the practice are limited given its illegality and the paucity of published research around off rolling to date is presumed to be attributable to the sensitivity of the topic (the professional risks associated with disclosure) or to constitute evidence of complicity. As Ball (2003) argues, educational cultures characterised by high levels of high stakes external scrutiny demand image management at school level. However, it is conceivable that, for some, and beyond issues of legality, a sense of shame at having to work in this way is inhibiting professional dialogue. Teachers could experience conflicting loyalties where a protected school culture which condones off rolling contradicts their own more inclusive pedagogic values.

Context

The study of senior school leaders’ microblogging is one study within a larger ongoing research project exploring off rolling, including the exploration of senior school leader perspectives through a qualitative survey (Done and Knowler 2020b). The overarching questions which are common to both the reported strand (analysis of social media contributions) and the larger project (exploration of varied perspectives on off rolling among school leaders, parents, pupils and educational professionals) are as follows:

1. How do senior leaders in mainstream secondary schools in England understand and explain the practice of off rolling?  

2. What are the challenges and dilemmas surrounding this practice for senior leaders? 

3. What is the personal and professional impact on senior leaders of involvement in cases that might be defined as off rolling? 

4. What do senior leaders say about how the practice of off rolling might be avoided? 

Prior to embarking on this multi-stranded research project, we theorised that off rolling is potentially a function of tensions between policy imperatives that prioritise both academic standards and inclusion (Done 2019; Done and Knowler 2020a, 2020b). The initial qualitative survey, however, revealed an apparent lack of understanding as to what constitutes off rolling among senior school leaders (Removed for review). Given the prominence of the topic in the national media and Ofsted’s policies prior to the Covid-19 pandemic, which culminated in a ‘naming and shaming’ policy where off rolling was found at inspection, the authenticity of responses indicating lack of awareness was questioned. It was hypothesised that the somewhat less formal communications appearing on Twitter around off rolling and school exclusion might evidence a wider range of perspectives and values.

Microblogging study method

A search for tweets was undertaken based on the criteria listed below in the study described here to illustrate the affordances and potential limitations of microblog research. A search protocol was adapted from Small (2011) to search for Tweets containing specific hashtags related to the practice of ‘off rolling’ and/or permanent exclusion, e.g #offrolling #illegalexclusions #pupilmovement #PEx #PExD.

• Tweets containing specific hashtags related to the practice of ‘off rolling’ and/or permanent exclusion, e.g. #offrolling #illegalexclusions #pupilmovement #PEx #PExD

• Tweets mentioning media events related to ‘off rolling’, e.g. publication of key reports and policy such as new Ofsted framework, Timpson Report, Children’s’ Commissioner

• Tweets of prominent ‘Edutweeters’ accounts mentioning ‘off-rolling’ and posted between 1 May 2019 and 31 December 2019. Such ‘tweeters’ have in excess of 5000 followers and/or a ‘blue tick’ indicating the account is ‘verified’ by Twitter. Blue ticks usually indicate that the tweeter has some influence, as exemplified by @tombennett who is the DfE Behaviour Tsar and ResearchEd founder.

This manual search and collation process by two members of the project team over a two-week period generated a non-random sample of 797 tweets and included a historical search of tweets posted 1 May 2019 and 31 December 2019 and related to:

a) hashtag #offrollling

b) Prominent accounts and senior leaders

c) News events – Timpson Review (May 2019), Ofsted framework (Summer 2019), Ofsted Blogs on Off Rolling (10 May 2019, 6 June 2019), EPI Research Report (18 April 2019).

Following the collation of tweets according to the criteria outlined above, tweets were organised into a database and a qualitative content analysis was undertaken. Initially, all tweets collated within the specified period were reviewed (n = 797), and those outside the study focus were excluded from the data set, including those of parents and news accounts. The remaining tweets and related threads totalled 322% and 68% of these were responses to another tweet commenting on off rolling in relation to a different issue; 8.7% (n = 64) of tweets were prompted by comments on media stories related to off rolling.

An important feature in the development of a coding framework was discussions around the number of codes to be applied to the tweets prior to establishing agreed explanations for each code prior at the analysis stage. This was intended to support the coding process by ensuring clarity on how the codes should be applied to tweets. Two researchers independently utilised the coding framework to colour code the sample tweets and their related threads. An initial analysis was checked for consistency of code application and, where there was agreement between coders, these tweets or excerpts of tweets were collated in a database. Where there was disagreement between the two coders, dialogue with a third member of the team determined whether a coded tweet would be included in the analysis. This process resulted in a database containing eight sections where coded tweets were then analysed together.

Drawing on O’Leary (2014), two researchers then independently undertook subject-based coding, permitting the identification of relevant codes around keywords within tweets and reducing the risk of individual researcher bias. Emergent findings were then shared within the team of three researchers, enabling the completion of a second stage of analysis which explored the meanings of noteworthy and relevant tweets at three levels: examining the intent of the tweeter, the context of the tweet, and the feedback or interaction with the other followers or tweeters. A coding scheme was then developed and tested in March 2020 containing codes and their explanation, for example, ‘Legal/Statutory’ (tweets referring to the legality of exclusion practices) and ‘Policy’ (tweets that criticise education policy and call for change or amendment). Coding allowed the identification of further categories conceived as either categories or meta-categories rather than overarching and subordinate themes. This process reflected our reluctance to seek latent meanings in tweet content or engage in a procedure of repeated abstractions that risked an unwarranted departure from explicit tweet content or the misreading of the tweeter’s intent. Categorisation implies a recognition of the bounded and limited nature of the data set and, therefore, of the knowledge claims that could be considered legitimate.

The grouping of categories generated three areas which were then related to the selected conceptual framework:

1) The meta-category of Harm included awareness of the potentially damaging effects of exclusionary practices and off rolling upon school individual pupils, families and teachers. Tweets within this meta-category were taken to be indicative of the historical shift towards an ascendant neoliberal educational culture and not only current pressures related to, for example, funding for SEND (The Guardian 2019) and excessive external scrutiny (Ball 2003). This shift was variously presented as follows: bringing about a breakdown in trust between professionals; bringing about a general loss of confidence in the education system and the practices it relies on; evidencing a repeated failure of education policy to address inequities within that system and to deliver an authentic inclusivity; and as replacing dominant values of care which, in terms of professional identity, played out as the compromising of personal ethics and integrity. These developments were viewed as the displacement of context-specific professional judgement with a consequent neglect of children’s welfare in decision-making processes.

2) The meta-category of Discounting included tweets which appeared to downplay the extent and significance of off rolling by, for example, portraying such practices as extremely rare events. This portrayal assumes the existence of an education system in which the objective of inclusion has been fulfiled and, by implication, questions the attention afforded to evidence which undermines this assumption. Discounting can thus be read as a rhetorical device and was manifested in tweets which noted: a rectifiable inconsistency in practice; an unjustified focus on a few ‘bad’ examples; misunderstanding of bureaucratic procedures (where and how to ‘note’ the exclusion); and variation in school inspection ratings and judgements. Suggestions that the foregrounding of off rolling has served to divert attention from efforts already made to ensure the inclusivity of schools, thereby depriving such schools of a much-deserved recognition, were also viewed as indicative of discounting.

3) The meta-category of Conflicts included tweets on the definition of off rolling which, when taken at face value, indicated high levels of confusion, possibly due in part to Ofsted’s (2019a) adoption of the term to imply illegality whilst in other circumstances or guidance it continues to signify the legitimate and legal removal of a child from roll. The hypothesis that an online platform like Twitter might produce greater authenticity in senior school leaders’ expressed views on off rolling appeared to be confirmed in the apparent confusion around definitions of off rolling; however, it was clear from some tweets that there are school leaders who are clearly aware of the prevailing legal grounds for formal exclusion and the relevance of behavioural issues and permitted disciplinary responses. By implication, practices falling outside of this legal prescription are likely to be understood as illegal. This range of tweets concerning definitions could, however, be taken to indicate an uneven understanding of, for instance, the relationship between behavioural issues and SEND. This meta-category also included tweets which conveyed strong views around professional autonomy or matters of professionalism, particularly in the context of parental advocacy. The literature relating to parental involvement and SEND (e.g. Broomhead, 2018) may, for some senior school leaders, be experienced as an erosion of their professional autonomy and denigration of their professional judgement. The latter was evidenced in tweets that expressed resentment at what was perceived as interference through policy or guidance in school decision-making around punitive measures (in response to behavioural issues) in contrast to tweets objecting to the use of isolation rooms (taken to be indicative of a normative shift in educational culture).

To summarise, the practice of off rolling is variously understood and explained by educationalists and there was no reason to suggest on the basis of the findings reported above that the understanding of senior leaders in mainstream secondary schools in England is any less varied. The analysed tweets do indicate that Ofsted (2019a) definition of off rolling as performance data manipulation or ‘gaming’ is not particularly helpful as it serves to obscure less easily understood or detected instances of off rolling, and it localises the practice as a function of individual school leaderships and cultures as opposed to systemic factors such as the marketisation of education and issues around behaviour, SEND, funding and professional autonomy; hence, our findings aligned with the general recognition that this practice is not only an issue for senior leaders even though it is senior school leaders that must comply with relevant legislation and statutory guidance or risk punitive actions upon inspection where off rolling is suspected by Ofsted.

Social media research

As a microblogging social networking platform which has been available publicly without charge since 2006, Twitter permits users to publish tweets (limited to 280 characters per communication); ‘like’ or indicate support for or agreement with other tweets; and retweet (share to their own Twitter feed) or respond to other’s tweets. Twitter also includes a search function to allow individuals to identify ‘like-minded’ users or tweets which pertain to topics or events relevant to them. Educational professionals can use Twitter to interact with other stakeholders, share information, request advice, or for solidarity; it provides a platform for responding to policy updates or sharing salient aspects of their experiences in the field. Accounts can be operated by both individuals and organisations, and tweeters can link to other multimedia sources to share videos or images.

Limitations

Despite Twitter’s apparently simple format, data is prioritised or hidden from view through the application of a complex, and publicly unavailable, algorithm. This becomes highly significant when considering Twitter as a data source for teacher education enquiry. Limited claims can be legitimately made on the basis of collected tweet data; for example, if a researcher conducts a search for tweets containing a particular word within a stated time frame (as in the reported study), it is misleading to claim to have accessed all tweets that meet these parameters as the availability of these tweets is controlled in varied ways. Firstly, being logged into a Twitter account when conducting a search prompts Twitter to prioritise data it considers relevant based on the accounts that a user follows, their likes and retweet history; secondly, users can utilise privacy settings to ensure their tweets are viewed only by certain people; and thirdly, deleted tweets or those under review by Twitter for breaching user guidelines will be omitted, including content deemed to breach copyright. Such content is identified through automated screening and reporting from other users; however, definitions of offending categories are subjective. Even factors such as the time of day when the search is conducted can impact the volume of tweets returned due to ongoing maintenance and development of the site (Twitter 2020).

Additionally, sarcasm and irony in tweets can undermine claims around the obviousness of intended meanings in Twitter-derived data; phrases such as ‘Good Job OFSTED!’ can rarely be taken at face value. Consequently, consideration of the wider conversation and context beyond the immediate text may be required in analytic processes, implying possibilities for misinterpretation.

Affordances

Despite these limitations, microblogs offer a unique source of potentially rich data and, whilst the purpose of tweeting varies between users and tweets, Twitter is reactive and immediate in that data can be collected soon after real-world events (Carpenter and Harvey 2019). It is this immediacy, combined with the disconnection that social media user experience (such that ‘public’ is taken to be an abstract concept) that can potentially generate candid and more authentic responses (Carpenter and Harvey 2019). The reported microblog study findings confirm that Twitter provides senior leaders with a platform where they feel able to engage in dialogue around off rolling in sometimes candid and controversial ways. The opportunity for other stakeholders, for example, SENCos, teachers, parents and non-teaching education professionals, to join these dialogues creates a rich and more immediate insight into the experience of off rolling that differs from insights garnered from more conventional data collection methods.

The overarching aim of the reported study was to ascertain whether microblog content around off rolling differed from that found in professional or research contexts. An attendant risk in interview-based teacher education enquiry is that local schools are likely to be known to academic researchers, amplifying teachers’ and head teachers’ concerns about divulging sensitive information despite assurances of anonymity. Such concerns may also apply in lengthier qualitative survey responses. Additionally, where research is undertaken by ITE students, it is doubtful that senior school leaders will be eager to share contentious or risky information. However, trainee teachers may be presumed to be less familiar with the sedimented discourses surrounding exclusionary practices in schools and, therefore, able to offer fresh insights. Methodologically, our own choice of qualitative content analysis was intended to demonstrate rigour and, therefore, results that were credible and trustworthy (Bengtsson 2016), whilst recognising that such credibility is, to a large extent, dependent on the quality of the data gathered and not only the method of its analysis.

Content analysis

Content analysis is held to be unique in implying both a quantitative (Krippendorff 2004; Neuendorf 2002) and a qualitative methodology (Berg 2001). Bengtsson (2016) argues, contra Berg (2001), that content analysis can support either inductive or deductive analyses. In practice, we would argue, distinctions such as inductive or deductive and manifest or latent content can be difficult to sustain depending on the coding frame adopted. Manifest analysis must stay close to the text whilst latent analysis seeks underlying meanings and is an interpretative analytical process (Berg 2001). The analytical strategy adopted in the reported study reflected this concern and, in order to maintain rigour and a systematic approach, the analytical process was staged such that the initial manifest analysis was followed by a grouping of the descriptive categories based on manifest and quantified content into meta-categories.

Although qualitative content analysis can be criticised for its lack of abstraction, the reported study was not intended as a theory-building exercise (as in grounded theory) nor was the depth of understanding associated with phenomenological methods sought. Qualitative content analysis was considered to be appropriate to the research aims, overall purpose and timeframe (Bengtsson 2016). It is ideally suited for use in preliminary investigation by time-pressed teacher educators or those in ITE seeking to gain experience in generic qualitative analytical processes. Both groups will encounter numerical data-driven school environments; hence, experience of methods and analytical strategies that seek nuanced understandings of decision-making in practice contexts may be useful.

The staged analytical process was designed to be undertaken without recourse to costly purpose-specific software whilst, simultaneously, evidencing the possibility of replication of findings. As previously stated, in qualitative methodologies it is the trustworthiness and credibility of the analysis and findings that is important (Lincoln and Guba 1985), and the independent coding procedure recommended by O’Leary (2014) was adopted accordingly. Common criticisms of qualitative content analysis, for example, that it is inherently reductive and neglects the context in which texts are produced depend, as Bengtsson (2016) argues, on the scope and purpose of a study and practical considerations. As one strand of a much larger research project, the findings from the reported study were not intended to provide a definitive interpretation of senior school leaders’ views on off rolling but, rather, to generate tentative insights that might enrich our understanding. It is worth noting here that the concept of ‘voice’ that we mobilise when describing the opportunity for senior school leaders to articulate their views afforded by the preceding survey-based study (Done and Knowler 2020b) is not a unitary one; on the contrary, it is anticipated that ‘voice’ will comprise multiple and complex views.

Ethical considerations

The use of Twitter as a data source implies brevity of response and an immediacy unassociated with qualitative analysis, but it also raises ethical considerations around privacy, consent and anonymity. As a relatively novel method of enquiry, the ethical implications of using Twitter as a data source are only now being considered and a key issue is that of privacy. Whilst it could be argued that to place a tweet in the public domain is, effectively, tantamount to permission for that tweet to be used for any research purpose, it is contended here that further consideration is warranted where researchers are striving to ensure ethical practice. This position accords with the British Educational Research Association (BERA 2018) ethical guidelines that insist upon reflection as to ‘whether online communities perceive their data to be either public or private’ (para. 12). As noted above, the processes that dictate the visibility of tweets are highly complex and can be easily misunderstood by users. Boyd and Crawford (2012) maintain that, ultimately, the researcher must take responsibility for decisions around privacy. Townsend and Wallace (2016, 12) state that, even where the research topic is sensitive, the Twitter default setting is ‘public’ and hashtags are created to encourage the wider dissemination of tweets.

Anonymity, however, remains an issue in qualitative content analysis. Unlike quantitative methods involving frequency counts of tweets containing a particular hashtag, individual tweets are more likely to be identifiable; hence, anonymity in qualitative content analysis may be compromised when the content of a tweet is reported verbatim and a simple online search can reveal the source. For this reason, direct quotation of tweets in our microblog study data set has been avoided despite the lack of full transparency of our coding processes and the impact on the perceived validity of findings. To contact every ‘tweeter’ included in the data set to gain their consent to participate in research would have been impractical and, beyond the time commitment implied and possibility of negative responses, other issues can arise given Twitter’s collaborative nature. A Twitter ‘thread’ can include numerous individual tweeters and, where one tweeter does not consent, the resultant ‘gap’ in the conversation causes the narrative to lose cohesion. The entire conversation in these circumstances could be excluded from the data set, risking impacting the volume of data collected.

As previously stated, Twitter was selected as a data source for the investigation of off rolling as it was hypothesised that informal dialogue outside of professional educational settings might illuminate the opinions of those in leadership roles which would otherwise be hidden. Fiesler and Proferes (2018) argue that tweeters’ awareness of their presence in a public domain does not imply that they would articulate similar views as participants in a research study. This raises a further issue of the impossibility of ensuring that the information a user claims in their Twitter biographical detail is truthful; indeed, anyone can claim to be tweeting as a senior education leader. Where a ‘tweeter’ has not deliberately created a false persona, it is still conceivable that they consider their Twitter persona(s) to be ‘distinct from their “real” selves to a degree’, prompting more ‘forthright’ tweets despite them remaining ‘responsible for and potentially traceable through that account’ (BERA, 2018, 3). It is common for user’s ‘bios’ to state ‘all views my own’, indicating a deliberate distancing of themselves from any professional organisation that they may be affiliated to. Whilst this positions Twitter as an attractive source of voices and opinions that may be inaccessible to other types of enquiry (e.g. a survey sent to schools), researchers should consider any potential harm to the professional life of a ‘tweeter’, even where such consideration does not appear to have been made by the individual sharing their opinions in a public forum. Again, robust anonymity is an important safeguard (Fiesler and Proferes 2018).

Concluding remarks

A method of qualitative content analysis adapted from Bengtsson (2016) has been illustrated through a recent small-scale study of senior school leaders’ microblogging on Twitter related to off rolling. Findings indicated that tweeters experience challenges related to off rolling that go beyond lack of understanding and unclear boundaries between legality and illegality in relevant guidance. Off rolling is recognised as harmful by many, but the impression given is that few can imagine a different way of working within the current political climate, policy landscape and educational culture; hence, the relative paucity of suggestions on how off rolling might be avoided. Looking forward, it is possible that disruption to routine practices following multiple Covid-19 pandemic-induced partial school closures during 2020 and 2021 will lead to new policy initiatives and a post-lockdown educational culture that prioritises inclusion and wellbeing. It seems more likely, however, that the conflicts evidenced in this data set, particularly around resources and professional autonomy, will be intensified. Evidence of a personal impact on senior school leaders engaged in off rolling was limited when compared to findings from an earlier parental strand of the aforementioned larger ongoing research project investigating off rolling in English schools (Done et al. 2021).

We have argued that the outlined method of content analysis of social media microblogs may be useful to teacher educators, and teachers or future teachers engaged in research into sensitive topics, even though Twitter algorithms make searching difficult to replicate despite generalised statements around reliability in methodological texts (Bengtsson 2016). Manual as opposed to computer-assisted analysis is time-consuming and, as previously acknowledged, what tweeters say and what they do may be very different, particularly in a high risk educational culture where image management is deemed a matter of professional survival (Ball 2003). Nevertheless, such disparities suggest a potentially valuable avenue for educational research that is related to the reduction of social marginalisation through education and to issues in ITE itself such as preparation for school cultures which deviate from the ideal of inclusivity. Where only small-scale investigation is possible, a manageable data set can be generated that, nevertheless, facilitates nuanced analysis.

Acknowledgments

This study was internally funded and ethically approved by the University of Exeter. The authors would also like to thank the reviewers of this article for their invaluable guidance.

Disclosure statement

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

Additional information

Notes on contributors

Alice E Potter

Alice E. Potter is a doctoral researcher in the Graduate School of Education at the University of Exeter.

Helen Knowler

Helen Knowler is the NASENCO Prgrsmme Director and a Senior Lecturer at the University of Wolverhampton. She is also an Honorary Research Fellow at the University of Exeter and engages in collaborative research investigating exclusionary pressures in schools in England.

Elizabeth J Done

Elizabeth J. Done supervises doctoral students in the University of Plymouth’s Institute of Education, specialising in inclusion, poststructuralist theory and critical analysis. She is an Honorary Research Fellow in the Graduate School of Education at Exeter University and was formerly Professor in Health Psychology at A.U.W. in northeast Africa.