Scarcely visible? Analysing initial teacher education research and the Research Excellence Framework

In the UK, the Research Excellence Framework is a mechanism used for ranking the quality of research in higher education institutions. While there has been analysis of the entire Research Excellence Framework, and of the Education unit of assessment more generally, analysis of how research on initial teacher education featured in the Research Excellence Framework has been minimal. In this article, we report on Phase I of an 18-month project that mapped the extent to which initial teacher education-focused research was included in the 2014 Research Excellence Framework. Employing a novel


Introduction
Since the 1990s, successive UK governments have sought to use initial teacher education (ITE) as a focus for their 'reforming' policies. (For the purposes of this article, we define ITE as the period of education and training culminating in a qualification to teach in schools.) This has resulted in policy churn, with little discernible positive effect, at least at the classroom level (la Velle and Reynolds, 2020). Added to this are high-stakes performative technologies, such as inspections undertaken by the Office for Standards in Education, Children's Services and Skills (Ofsted) (see Clapham et al., 2016) (Furlong, 2013;Matthews and Kotzee, 2022;Pollard, 2014).

and the Research Excellence Framework (REF), with similar examples becoming increasingly widespread internationally. Both REF and Ofsted inspections carry significant outcomes for higher education institutions (HEIs) and teacher educators in the UK
Teacher educators (those employed by HEIs to teach in their ITE programmes) in the UK must balance often competing requirements. On the one hand, they may need to be 'research active' and, on the other, they must negotiate the demands of a postgraduate programme requiring both academic input at master's level and supervision of candidates' teaching placements in schools (see Czerniawski et al., 2018;Murray, 2012). We suggest that although the REF only concerns the UK, similar exercises exist in other countries and are becoming increasingly prevalent globally. Therefore, unpicking the extent to which ITE research features in frameworks such as the REF, and its effect on those researching ITE, is of interest to all those concerned with teacher education.
In this article, we report on Phase I of an ongoing project that examines the extent to which ITE-focused research was 'returned' in REF 2014 by HEIs that also offered ITE programmes. Due to the devolved nature of higher education in the UK, the article focuses only on HEIs in England. Using an innovative model, described in detail later on, we provide evidence of a dearth of high-level research in teacher education in England and discuss the consequences of this.
We categorised and analysed outputs from 12 identified HEIs. More than 1,600 outputs were analysed, with the aim of selecting those that had a focus on ITE. A total of n = 5,526 outputs from n = 76 HEIs were returned for the Education unit of assessment of REF 2014. The outcome of our analysis was stark: only 5.5 per cent of the outputs in our representative sample focused on ITE. At the time of writing, Phase II preliminary analysis of a sample of the outputs for REF 2021 data is under way. Although various developmental changes to the REF have been seen in the REF 2021 exercise, there is a high degree of continuity in terms of the criteria for assessment. Our preliminary 'snapshot' analysis of a sample of the outputs for REF 2021 indicates that 4.8 per cent are ITE-related. We argue that these findings are of interest to all those in England, the UK and internationally who are concerned with teacher education and research in this area.
To understand this, we draw on Ball's (1993) suggestion that policy has two facets: text and discourse. As we go on to discuss, there are key policy texts and discourses embedded within the REF that have significant implications for teacher educators' practices, as both academics and researchers. Similarly, governmental policy discourses, such as those in the 2022 ITE Market Review pledging to make 'England the best place in the world to become a great teacher' (DfE, 2022: n.p.), highlight the importance of ITE for the UK government (in respect of England and Wales). Nonetheless, the reality of teacher education being research informed, at least in REF terms, is brought into question by the analysis presented in this article.
We suggest that the lack of visibility of ITE-focused research in REF 2014 means that this area of enquiry is marginalised, echoing concerns raised by the Stern (2016)

Initial teacher education
There is significant literature exploring the ITE landscape in the UK (see, for example, Murray, 2012;Whiting et al., 2018) and the Department for Education (DfE, 2022) provides significant policy and guidance documentation in this space. Rather than attempting to go over 'old ground', there are two points that are worth stressing here.
First, as the DfE (2021) Initial Teacher Training Census shows, ITE courses make a significant financial contribution to the UK HEI sector. For the 2021/2 cohort alone, we estimated that 33,946 entrants (DfE, 2021) paid around £314 million in fees (GOV.UK, 2022). This financial contribution is important when we also consider how organisations such as the Universities' Council for the Education of Teachers (UCET, 2020) has been advocating that teacher educators should be more strongly supported as researchers (see also la Velle and Kendall, 2019). Similarly, the British Educational Research Association (BERA, 2014, 2018a) has a long-standing commitment to practitioner research. Nonetheless, despite this financial contribution and the remonstrations of UCET and BERA, there are significant challenges facing teacher educators attempting to undertake and publish ITE-focused research (see la Velle and Kendall, 2019), which is returned to the REF.
Second, the move towards a school-based model of ITE has seen a reduced emphasis on teaching as an autonomous, intellectual and research-informed profession (Duggan and la Velle, 2019). Greater importance has incrementally been given to accountability, through measures such as Ofsted inspections. In these, Ofsted undertakes judgement of ITE providers' performance in three areas -overall effectiveness, quality of education and training, and leadership and management -and it subsequently ranks the provision as 'outstanding', 'good', 'requires improvement' or 'inadequate'. Both a 'requires improvement' and an 'inadequate' ranking carry significant implications for the provider and can ultimately result in that provider losing its accreditation to offer ITE courses.
It is important to stress that Ofsted's expectation for high-quality ITE provision does not in itself preclude the undertaking of high-quality research in the ITE space. However, what the intersection between performative demands and the capacity for ITE academics to undertake research in their subject area illustrates is the complex landscape of competing policy discourses and texts that shape teacher educators' work. On the one hand, there are the high-stakes regulatory and disciplinary frameworks mediated by Ofsted and policy such as the 2022 Market Review (DfE, 2022), both of which seemingly fail to acknowledge the importance of ITE-focused research. On the other, there are the pressures that academics face to undertake and publish 'high-quality' research that can be submitted to the REF. As we go on to discuss, these competing demands, and the policy texts and discourses that drive them, have significant implications for those undertaking the work of both teacher educator and ITE researcher.

Research Excellence Framework
It is now over 30 years since the UK first introduced a national research assessment exercise. The REF aims to analyse and measure the quality of research outputs (journal articles, book chapters, monographs, books and so on) from UK HEIs (Tymms and Higgins, 2018;Watermeyer, 2016). Similar systems are rare internationally, with the Excellence in Research for Australia (ARC, 2022) being one example. In Europe, although systems exist for research evaluation and performance-related research funding, these tend to be much smaller in scale than the UK's REF and are mainly institutionally based (Silvertsen, 2018).
For REF purposes, each academic discipline is organised into a unit of assessment (UOA), of which Education is one, and, for REF 2014, 154 UK HEIs entered 36 UOAs with a total of 1,911 submissions. Overall, 52,061 academic staff were 'returned' (that is, their work was included in their HEI's REF submission) with 191,150 outputs (REF, 2014). For 2014, REF submissions encompassed three elements, weighted as shown: research outputs (with a maximum of four outputs per researcher; 65 per cent); impact (a descriptive template and case studies; 20 per cent); and environment (a template for describing research culture, with data; 15 per cent). REF 2014 provided detailed explanation as to how outputs were defined and the assessment criteria that were used to grade the overall submission, each separate component of that submission and each individual output. Ultimately, HEIs were ranked in relation to the number of outputs submitted, how many staff were returned and, using the criteria of originality, significance and rigour, how many outputs were graded as 4* ('world leading'), 3* ('internationally excellent'), 2* ('internationally recognised'), 1* ('nationally recognised') or unclassified ('below the standard of nationally recognised work').
It is important to stress that the REF has faced substantial criticism since its inception. For example, the Stern (2016: 11) Review emphasised significant problems associated with its 'costs, burdens and distortions'. Watermeyer (2016: 199), meanwhile, reported how some academics viewed the REF as 'an infringement to a scholarly way of life'. The Stern Review also highlighted how some of the REF operating principles made an impact on which research was, and was not, returned. One of these rules was that HEIs had to return all employed staff with a contract of greater time than 0.2 full-time equivalent, each of whom were required to nominate up to four outputs produced between 2008 and 2013. As Stern (2016: 15) makes clear, the requirement to return a fixed number of outputs per individual may have encouraged a focus on 'safer' publication strategies and 'short-termism' in individual researchers' research strategies. Moreover, Stern (2016: 15) continued by stating that the four-output requirement may have resulted in a 'distortion of career choices', as the outputs from researchers who did not produce the 'requisite number of outputs within the census period are not visible to the REF'. What Stern (2016) suggests is that, despite the high-stakes nature of the REF and efforts to improve equity, equality and transparency for REF 2014, inequalities still resulted, in part, from the operating rules that were in place.
In this article, we set out to draw attention to the gap in our knowledge about the extent of ITE-focused research submitted to the REF. To do so, we drew on our significant experience in the sector and on the wider literature to develop the overarching research question: In REF 2014, what is the proportion of outputs in the Education UOA that have a principal focus on initial teacher education? In the following sections, we map out the ethical, methodological and analytical approach we designed to answer this question.

Ethical considerations
For reasons of ethical research practice (see BERA, 2018b), the HEIs used in our methodology are anonymised from the point at which the final sample was selected. However, as all the data on which these results are based are in the public domain, identification of each HEI is possible.

Methodology
The data set for HEIs in England that provide ITE programmes and are also involved in educational research is large. In many cases, candidates complete their undergraduate degree followed by a Postgraduate (or Professional) Certificate in Education (PGCE) or Postgraduate Diploma in Education (PGDE), both of which are offered for qualification to teach in primary (elementary) or secondary (high; post-primary) schools. In the academic year 2021/2, 69 HEIs offered PGCE/PGDE courses (DfE, n.d.). To achieve a manageable and representative sample of HEIs, this large number was narrowed down using the following sampling approach.

Choosing initial sample
The sampling process was initiated by identifying those HEIs that have had a sustained involvement in ITE and education research, and that (1) submitted returns in Education for two successive research assessment cycles (Research Assessment Exercise [RAE] 2008 andREF 2014) and (2) offered PGCE/PGDE ITE routes for the academic year 2021/2. These criteria provided a list of 51 HEIs as potential candidates for analysis. The DfE (n.d.) website provided an alphabetical list of ITE providers, which was searched manually for provision of primary or secondary PGCE programmes, or both. Many of these HEIs also had other ITE routes, such as undergraduate courses, School Direct and Post Compulsory Education and Training, demonstrating a broad portfolio. From this initial process, we then moved on to employing a set of metrics to identify the final sample.

Identifying final sample
A range of publicly available metrics was also analysed in relation to each HEI to illustrate the size and scope of the ITE provision and the associated research. Metrics that were considered of importance were: First, the grade point average, a measure of the overall quality of the research, was used to identify potential HEIs for the sample. It is calculated by multiplying the percentage at each star-rated quality level by its rating, so the percentage of 4* is multiplied by 4, 3* by 3 and so on, down to 0 (unclassified), with these figures totalled and divided by 100 to give a score between 0 and 4 (see Jump, 2014). Second, research power, which is calculated by multiplying the HEI's overall rounded grade point average by the overall number of full-time-equivalent staff it submitted to the REF, was analysed for each HEI. Third, the number of research outputs submitted to the Education UOA by each HEI was identified. Fourth, the Ofsted inspection grades for each HEI were identified. These grades were taken from the most recent Ofsted inspection for each HEI, with the values given by Ofsted (n.d.) for every phase assessed, and then used to calculate a mean value. Finally, the recorded number of students for each HEI who had accepted places on ITE programmes for the academic year 2014/15 (UCAS, n.d.) was identified.
Each HEI was then independently ranked for each metric. From each ranked list for the five metrics, the top five and the bottom five HEIs were selected. Then, the median was calculated for each metric, and HEIs were selected that met the median value plus one value each above and below the median. This allowed the collation of a subset of data comprising the top, middle and bottom ranked HEIs for each metric (Table 1). Some of these metrics were more useful than others in identifying the sample. For example, the Ofsted grade metric was not a good discriminator, as too many HEIs fell into the middle value. However, this information was used later to inform analysis and discussion. In contrast, Table 1 indicates the values selected for the student numbers by HEI in 2014/15, which was particularly useful. Each HEI from the top, middle or bottom subset for each metric was given a combined weighting (Table 2). Scores were given as follows: 3, for each time it ranked in the top group; 2, for each time it ranked in the middle; and 1, for each time it ranked in the bottom group. Note that some HEIs from the initial selection received no weighting, if they failed to fall into a top, middle or bottom bracket for any metric.

Pilot study and categorising outputs
To stress test this methodology, we undertook a pilot study of HEI A, chosen due to its large number of REF outputs (n = 790), from 249 authors. As it was the pilot, HEI A was not included in the final analysis. Using the information collected from the pilot, different categories were explored, manually using keywords to filter the outputs. This resulted in five categories used for the final study (Table 3).

Results and discussion
In the following sections, we discuss: the methodology, and its strengths and weaknesses; some detailed and headline findings; understanding the REF as text and discourse; and reflections and next steps. Before doing so, however, it is important to reiterate that this is a new methodology, which may need further development. Nonetheless, our analysis paints a stark picture of how significantly ITE-focused research featured in REF 2014. As Table 4 indicates, the HEI A pilot indicated that only 1 per cent (8 out of 790) of these outputs were categorised ITE England. Although the picture is slightly better regarding the 11 HEIs that we identified as our stratified sample, even here, only 5.5 per cent (45 out of 823) of outputs were categorised ITE England.

Discussion of methodology
Although novel, we argue that the methodology we have used to ascertain these findings has several strengths. Not only is the approach scalable to be applicable across the entire Education UOA, it can also be amended to reflect changes in the REF process. Significantly, the approach has the capability to be amended for use to analyse data from similar research frameworks beyond the UK. We have also illustrated how the method was both co-constructed and used by a team of academics from different HEIs. Moreover, the integral quality-control process moderating how outputs were categorised appears to be effective. The method relies, at least in the case of the UK REF, on publicly available, anonymised, secondary data, which means that research teams do not have to generate new data. The richness of the data set is reflected in the employment of a cross-sectional sample, the use of a range of metrics and the production of a stratified analysis, providing nuance to the overall findings. Despite these strengths, inevitably, such a new approach has areas that can be further developed. For example, although more than 1,600 outputs were categorised, the 2014 Education UOA consisted of more than 5,500 outputs in total. As a result, potentially, a significant number of ITE-focused outputs were returned that were not included in our sample. The sheer scope of the task at hand, even with the reduced sample of HEIs, raised significant challenges. Even with a sample of just 12 HEIs, designing the approach and undertaking the analysis was a lengthy and time-consuming process. Taking on this work for the entire Education UOA, potentially 6,000 outputs, is a significant and challenging task. Of course, the use of technology to mediate a meta-analysis of outputs would reduce the time that the analysis Analysing initial teacher education research and the Research Excellence Framework 8 would require; however, such software would need to be funded and developed, and integrated carefully into the methodology.
It is also worth discussing some specific areas that could be further strengthened, which emerged when the approach was used in practice. For example, and somewhat surprisingly, not all data for all the metrics were available from all the HEIs. Although not significantly skewing the analysis, the absence of these data for some HEIs did require some thought and discussion as to how to address it. In order to maximise consistency, the research team engaged in internal moderation of how outputs were categorised. Because we were unable to ascertain which outputs might have been double weighted by the HEI submitting them to the REF, we did not 'weight' outputs (books versus journal articles, for instance) differentially. It is accepted that this could slightly skew the overall Research Power metric that was used in this study.

Discussion of findings
A detailed discussion of the findings of the REF 2014 analysis is the focus of further publications. Nonetheless, there are some elements that are indicative of the form that such future analysis might take, and they are therefore worth mentioning briefly here. The headline finding that only 5.5 per cent of outputs analysed in the main study were categorised as 'specifically relating to ITE' (ITE England) is indeed stark. However, drilling down into our analysis reveals a significant number of other findings. For example, the 45 outputs classified as ITE England were published in close to 40 (n = 36) different journals/publications. This suggests that authors writing about ITE are using a wide variety of methodological and theoretical frameworks. In some HEIs, outputs were written by only one academic, or by a small group of academics, in relation to one specific subject area. An interpretation of this is that an HEI might have a relatively large return of outputs focused on ITE research simply because that is an area of interest for one academic, or a group of academics, rather than being due to a strategic focus on this area of enquiry.
Our analysis also indicated that 50.8 per cent of outputs were categorised as not relating to teaching and learning in English schools. A detailed discussion of this is outside the scope of this article. However, what this does suggest is that Education research, in REF terms, encompasses a wide variety of areas of focus, but that research related to activity in schools in England is the largest single subject area (Teaching and learning in England and English Schools/other activity categories). Of those outputs, 30.4 per cent were categorised as English Schools/other activity; that is, their focus was on activity in English schools, but not on teaching and learning (for example, school governance), while 12.2 per cent were related to teaching and learning in English schools, but not to ITE. It is also worth stating that these percentages are from the sample of 11 HEIs, and did not include the HEI A pilot data. Given that only 1 per cent of HEI A's outputs were focused on ITE and that HEI A had by far the largest return from an individual HEI, had we included this in our analysis, the headline figure of 5.5 per cent would have been even lower, at 3.2 per cent. One final point of interest is that HEIs F and H did not have any ITE England outputs and a low number of total outputs. However, HEI F had very low student numbers, while HEI H had the highest number of students in the 2014/15 cycle.
We have decided not to focus on individual HEIs by name. However, analysis of individual HEIs has produced some further interesting findings. For example, HEI K had almost 20 per cent of its outputs categorised as ITE England; however, because of the small size of the return, this was only six outputs, which were all concerned with the same subject area. HEI K also had a relatively small cohort of ITE candidates. Although this is a tentative finding, there does seem to be some correlation between the number of candidates following ITE programmes and the number of outputs returned. When we compared other HEIs with HEI K, we saw that those with larger ITE cohorts returned significantly fewer outputs. Moreover, ITE-focused research did not feature heavily in 'research-intensive' HEIs with large REF returns.
Finally, there is the issue of the nature and focus of the outputs themselves. Although almost 50 per cent of the outputs are ITE England, Teaching and learning in England, English schools/other activity or ITE outside England, 50.79 per cent were categorised ITE outside England. Discussion around this distribution of output focuses is outside the scope of this article. Nonetheless, a detailed analysis of the topics, methodologies and results in articles presented to the Education UOA of REF would clearly be of interest to education researchers, both in the UK and internationally.

Discussion of the REF as text and discourse
When we consider the REF as policy text and discourse, perhaps the paucity of ITE-focused research in the 2014 exercise can be explained. Our findings highlight that, despite organisations such as UCET advocating greater support for teacher educators as researchers, and BERA's (2014) assertion that teacher education should be a research-informed activity, neither of these aspirations are reflected in our analysis. Our findings also chime with la Velle and Kendall's (2019) analysis that there is a wide range of challenges that prevent teacher educators from undertaking and publishing research, challenges only exacerbated by the tensions that academics feel between being, on the one hand, a teacher educator and, on the other, a researcher (Czerniawski et al., 2018).
This duality in teacher education has become increasingly evident, with the result that some academic staff are primarily researchers, and others are primarily teacher educators (see Munn, 2008). Similarly, teacher educators often have heavy workloads, teach long hours in universities and schools and undertake high levels of student-teacher support (see Murray, 2012). These are conditions that Czerniawski et al. (2018) suggest have significant impact on teacher educators' capability to undertake, write and publish ITE-focused research. Teacher educators face a complex set of competing pushes and pulls on their time. For example, in England alone, HEIs offering teacher education range from long-established, research-intensive universities, to newly established teaching-intensive universities. Schools of education within these differing types of HEI vary greatly in how they 'instantiate the discourses and practices of teacher education' (Czerniawski et al., 2018: 136). As a result, teacher educators face varying expectations as to how they undertake their teaching, management and research activities.
Moreover, teacher educators must navigate their way through increasing competition, accountability and external evaluation requirements, all of which restrict time and access to research-related activities. Hoult et al. (2021: n.p.) describe this as teacher educators holding a 'mixed identity' as researcher and as educator, which, in turn, produces an 'uneasy relationship with REF'. Added to this are the practicalities of being a teacher educator and teacher educator researcher, highlighted by Murray (2012). The high levels of stress and workload that teacher educators face are in part due to the requirements to work across both the higher education and school sectors: requirements that must be negotiated just to fulfil the 'day job'.
Consequently, narrating the REF as both text and discourse powerfully illustrates many of the elements of disciplinary power described by Michel Foucault (1977). We argue that the marginalisation of ITE-focused research in REF 2014 highlights the framework as a disciplinary and panoptic technology that mediates self-discipline and renders bodies -institutional ones such as HEIs, and individual ones such as teacher educators -docile. Added to the texts and discourses of the REF are those rehearsed by government, through, for example, the ITE Market Review (DfE, 2022). All this adds up to a highly complex working environment, where teacher educators must negotiate competing policies -policies that not only compete, but seemingly are in complete opposition to one another.

Reflections and next steps
As we have outlined, we are not immune from the pressures facing academics, and we have undertaken this work without funding. Consequently, in Phase II of the project, we are only able to undertake a similar analysis of a stratified sample of HEIs from the 2021 data set as to that undertaken with the 2014 data in Phase I. Again, it is important to stress that both the 2014 and 2021 analyses are 'snapshots' of only a selected sample. What has become clear, therefore, is that there is significant further analysis which could be undertaken with both the 2014 and 2021 data.
It would be most useful to direct analysis towards the entire Education UOA, rather than just a sample. Similarly, a detailed analysis would enable some of the questions raised by the REF 2014 analysis to be addressed. For example, do HEIs with a larger cohort of ITE candidates produce less research than those with a smaller cohort? Are teacher educators producing research in areas other than ITE? What is the research-teaching workload balance of teacher educators who have articles submitted to REF?
Similarly, an analysis of the focus of the ITE articles that were returned and of the methodologies and theoretical frameworks used would be of interest. There is also the opportunity in the future to analyse ITE research that is based internationally; for example, 7 per cent of outputs from one HEI returned in 2014 were focused on ITE outside England. What these developments require, however, is capacity. The irony that a group of researchers investigating the lack of ITE research in the REF found it extremely challenging to undertake that research due to lack of capacity was not lost on the team.

Conclusion
We have demonstrated that ITE-focused research was, to use a phase from the Stern (2016) Report, almost invisible in REF 2014. Of course, there is more to research than the REF. Nonetheless, the extent to which ITE-focused research features in such an exercise does signal how HEIs value (or not) such enquiry. Much has been written about how performative technologies such as Ofsted have an impact on educators' work.
We argue that this article takes the literature on ITE research forward by exploring the implications of the REF as a policy text and discourse, and as a high-stakes regulatory system for teacher educators undertaking and submitting ITE-focused research. For example, the 2014 requirement for academics to submit up to four outputs may have meant that teacher educators researching in the ITE space were not returned to REF only because they had fewer than four outputs, not that they did not have any outputs at all. Similarly, the 2021 requirement that HEIs return academics with 'Significant Responsibility for Research' (SRR) could well have had an impact on how much ITE-focused research was returned, as it is unclear how many teacher educators were employed on SRR contracts.
Our analysis suggests that ITE-focused research was marginalised in REF 2014. Such marginalisation has significant implications for teacher educators and the ITE space, as REF performance is directly related to research funding at the institutional level, and to performance management at the individual level. Consequently, if ITE-focused research is to feature in the 'gold standard' REF rankings, and in similar accountability structures increasingly prevalent internationally, then teacher educators and teacher education departments must be supported to undertake, publish and submit ITE-focused research to further iterations of the REF. Failing to do so brings into question the extent to which policymakers and governments, who claim that they value ITE highly, do so in reality. Consequently, unless significant changes are made to the processes and practices inherent in REF, it seems likely that ITE-focused research will remain one of those areas rendered almost invisible by it.