The number of teachers leaving the profession in England is increasing, with the latest school workforce data showing a fall in the five-year retention rate, as well as a rise in the advertised vacancy rate (Allen et al., 2022). Attrition (excluding retirements) increased between 2011/12 and 2015/16. It fell slightly in the two years leading up to the Covid-19 pandemic and then more sharply during the pandemic; however, the latest data show that rates are now rising once again (McLean et al., 2023). Combined with failures to meet recruitment targets, this resulted in a total of 81,468 teaching vacancies being advertised by the end of the 2021/2 academic year, 59 per cent more than in 2018/19. By February 2023 the number was nearly double the total for the same period pre-pandemic (Allen et al., 2022). Worryingly, evidence suggests that teachers working in schools serving a higher proportion of socio-economically disadvantaged students are less likely to envisage remaining in the profession in the future (Allen and McInerney, 2019).
Recruitment and retention are not new problems in England, and demand for teachers has exceeded supply during most of the post-war period, with a particularly marked peak in the mid-1990s (Chevalier and Dolton, 2004). Shortfalls are partly influenced by economic conditions in a counter-cyclical manner and, relatedly, by relative wages (Chevalier et al., 2007).
Studies to date have explored how to stem leaks in the teacher supply pipeline (Allen and Sims, 2018; Barmby, 2006; Chiong et al., 2017; Menzies et al., 2015; Sims, 2020, 2021; Small et al., 2018), but there are fewer studies of how turnover affects students, and what to do about this. As discussed in greater detail below, studies tend to fall into two groups: those that emphasise the negative effects of turnover (Carver-Thomas and Darling-Hammond, 2017; Darling-Hammond and Sykes, 2003; Gibbons et al., 2018; Ronfeldt et al., 2013), and those that highlight the potential benefits of the ‘right sort’ of turnover (Adnot et al., 2017; Hanushek and Rivkin, 2010). The former draw on empirical data to demonstrate decreases in student achievement associated with higher turnover, while the latter tend to be more compositionally focused, demonstrating increases in student achievement when less effective teachers are replaced by those who are more effective.
Turnover and churn have been defined in the literature in numerous ways. Sims (2021) distinguishes between ‘attrition’, where teachers leave the profession, and ‘turnover’, where teachers move from one school to another. Movement between schools is also referred to as ‘churn’ (Worth and De Lazzari, 2017). Others argue that churn should include ‘within-school churn’, and that ‘newness’ should be seen on a spectrum (Atteberry et al., 2017). According to this conceptualisation, viewed from students’ and teachers’ perspectives, movements of teachers between groups of students reduces familiarity in a similar manner to other forms of turnover. Thus, while some teachers teach the same class in the same school for several years, providing a high degree of familiarity, others are veterans of a school but have been allocated to a new class every year. Meanwhile, others might be new to the school but have already worked in the same community. At the other extreme are the newest teachers, with the lowest degree of familiarity. These are teachers who have moved from a different geographical area, or who are newly qualified teachers. Empirical validation of this conceptualisation comes from evidence that student progress varies continuously along the spectrum of newness, such that the newer the teacher, the less highly students achieve (Atteberry et al., 2017).
This article follows Atteberry et al.’s (2017) broadened conceptualisation of turnover and churn, since recognising a spectrum of newness and its ‘disruptive effect’ (Ronfeldt et al., 2013: 4) can help explain the negative effects of turnover and churn by centring the relationship between teachers and their students. This is also in line with Blazar (2015: 223), who notes that research on teacher stability normally focuses on movements into and out of the profession, or between schools and districts, but that within-school movements represent ‘another potential facet of stability’. This approach aligns with the notion of ‘continuity of care’ (Haggerty et al., 2003; Hegde and Cassidy, 2004), a concept taken from the literature on public health which I draw on throughout this article.
For clarity, the article refers to ‘attrition’ (where teachers depart the profession), ‘between-school churn’ (where teachers move from one school to another) and ‘within-school churn’ (where teachers move from one group of students to another within the same school). All three terms are included under the umbrella term ‘turnover’.
This article argues that an improved understanding of how turnover (including attrition and both forms of churn) affects students, staff and institutions, points the way towards strategies that mitigate its negative effects. It begins by proposing three ways in which turnover might impact on schools and students. It then turns to how teacher allocation (as a result of teacher preferences and management decisions about timetabling) might exacerbate or mitigate these effects by determining the degree to which students are taught by a ‘new’ teacher. It draws on new data showing that teachers in England may be receptive to evidence-based allocation strategies, and concludes that efforts to support socio-economically disadvantaged students’ educational attainment should take into account continuity.
This article provides a synthesis of, and commentary on, existing evidence regarding turnover. It also draws on large-scale data from the school survey service Teacher Tapp to provide new evidence on teachers’ attitudes to class allocation.
Teacher Tapp is an app-based service that asks a series of closed response questions to a large sample of teachers every day. Teachers sign up to the app on a voluntary, self-selecting basis and, at the time of writing, the service reached over 7,000 teachers daily. Results are weighted based on gender, age, seniority, phase taught and school type/location. Because the sample is a convenience sample, findings should be interpreted as indicative. Data from the survey have been used in a number of recent studies of teacher well-being, mental health and working hours (Allen et al., 2020, 2021; Jerrim et al., 2022).
For the purposes of this study, I take advantage of the opportunity to randomly split respondents and ask slightly different questions to each group. By doing so, I am able to identify the extent to which teachers’ class allocation preferences might be influenced by research evidence. This comes with the additional caveat that by signing up to the survey app, teachers may already be demonstrating a particular interest in research and evidence.
The impact of teacher turnover
Carver-Thomas and Darling-Hammond (2017) suggest that turnover is problematic because it contributes to shortages, reduces achievement and imposes costs. However, they acknowledge that a degree of movement between schools is desirable, since it may serve to improve teacher–school matches. Ronfeldt et al. (2013) demonstrate the negative effect of teacher turnover on pupil achievement at primary school, finding that a one standard deviation increase in turnover results in a decrease in mathematics achievement of approximately 2 per cent of a standard deviation. Effects were slightly larger in mathematics than in English language arts, and larger in schools with a higher proportion of low-achieving and Black students. Importantly, the authors find that these effects extend beyond the students whose teacher departs. They therefore identify what they term a ‘disruption effect’, which includes the loss of institutional knowledge.
Research by Gibbons et al. (2018) explores the impact of turnover in the English context, noting that, on the one hand, turnover may facilitate the exchange of ideas between schools, help teachers develop new experiences and lead to better teacher–school matches, but that, on the other hand, it can also drain away relevant knowledge, cause discontinuities in teaching style and lead to time and resources being wasted on induction and ‘assimilation’. They find that a one standard deviation increase in teacher entry rates in the two years leading up to students sitting their GCSE examinations is associated with a 0.8 standard deviation reduction in points scores. The authors note that boys are more negatively affected by turnover than girls, and that there is some heterogeneity of effect depending on teachers’ seniority and subject taught, but that there is no consistent trend in terms of whether disadvantaged students (by prior attainment, socio-economic status and ethnicity) are more negatively affected.
Students are more likely to be exposed to the effects of turnover in some schools than in others. Nguyen et al. (2020) find that teachers are more likely to leave schools with low student achievement than those with high achievement, and Darling-Hammond and Sykes (2003) note that schools serving disadvantaged communities are more likely to struggle to fill posts with well-qualified teachers, making teacher departures particularly problematic in these schools. Analysis of data from three large school districts in the US also shows that poor, low achieving, Black and Hispanic students are more likely to be taught by novice teachers (Kalogrides and Loeb, 2013).
Studies emphasising the negative effects of turnover contrast with those that point to potential benefits (Adnot et al., 2017; Hanushek and Rivkin, 2010). However, Adnot et al. (2017) focus specifically on a programme that removed less effective teachers and replaced them with more effective teachers, and Hanushek and Rivkin (2010) focus on comparing the productivity of teachers who remain in schools with that of teachers who leave. Although these two studies are therefore compositional in nature, they demonstrate that turnover need not always be negative; yet, as Loeb et al. (2012) note, even where school administrators can accurately identify the least effective teaches, it is not always possible to dismiss them.
Drawing on other fields can provide a more comprehensive theorisation of the effects of turnover. In their multidisciplinary review on continuity of care in the field of public health, Haggerty et al. (2003) distinguish between three types of continuity: management continuity, relational continuity and informational continuity. The typology was developed as a means of understanding and responding to patients’ experiences of ‘fragmentation of care’, a concern that aligns closely with those relating to turnover in schools, particularly where students’ education depends on a series of different professionals over time. In the following sections, I argue that Haggerty et al.’s (2003) three forms of continuity are also important for students in a school context, and provide additional nuance regarding the nature of the ‘disruptive effects’ (Ronfeldt et al., 2013) of turnover, and how these relate to ‘familiarity’ (Atteberry et al., 2017).
Alongside Haggerty et al.’s (2003) categories of continuity, I draw on theories of collective action and social capital (Axelrod, 1984; Coleman, 1988; Gambetta, 1988; Olson, 2012), as well as school culture (Baars et al., 2018; Deal and Kennedy, 1983; Menzies et al., 2010) to suggest that the damaging effects of teacher turnover on students and schools are driven by the erosion of trust, student-centric and institutional knowledge, and collaboration and collegiality.
Cooperation between teachers and students in a school environment can be theorised as a collective action problem requiring trust. For example, teachers and students both stand to benefit from avoiding misbehaviour and sanctions, but often end up trapped in damaging cycles of tit-for-tat disruption and punishment. Game theorists have long recognised the role of trust in overcoming collective action problems (Axelrod, 1984; Gambetta, 1988; Olson, 2012); however, the emergence of trust depends on the expectation of future interactions. High teacher turnover reduces the likelihood of long-term, repeated interactions, and this diminishes the ‘shadow of the future’, or the time horizon over which a ‘discount factor’ is applied, reducing the incentive to collaborate (Axelrod, 1984). As noted below, given the potential for turnover to undermine trust, schools facing high turnover may therefore need to take additional steps to intentionally nurture relationships and build trust at the institutional level.
Ashcroft et al. (2016) argue that relationships can be seen as narratives, in that they have a starting point and develop around key events and interactions, which layer on top of each other over time. They suggest that continuity can be maintained by ensuring a web of relationships that function like a many-stranded rope, such that if one is ruptured, others can then maintain continuity. This aligns with Haggerty et al.’s (2003: 1221) ‘relational continuity’, whereby a ‘consistent core of staff provides patients with a sense of predictability and coherence’. Students might therefore benefit from developing strong relationships with a range of staff aside from their class and subject teachers, including those in pastoral roles and those running extra-curricular clubs, since this reduces single-person dependencies.
Activities outside the classroom, such as school residentials, have also been shown to accelerate the build-up of trusting relationships (Dudman et al., 2019). These might be particularly useful at the start of a year, when teachers and students first meet, yet at present most schools in England still organise residential trips at the end of the school year (Menzies et al., 2017).
In his work on trust and social capital, Coleman (1988: 104–5) emphasises the role of norms and effective sanctions in establishing cooperation across a community, including in ensuring that individuals act in the interest of the collective good, even when they do not know each other. If individual-to-individual trust is insufficient to guarantee collaboration in a high-turnover environment, one response is to establish trust at an institutional or community level. This approach finds expression in guidance from England’s Department for Education, which recommends that schools should use routines and sanctions to create a predictable environment that minimises disruptive behaviour (Bennett, 2017). Research on school culture also emphasises the role of consistency in institutional practices (Baars et al., 2018), and suggests that if a norm of order is established at a school level, rather than relying on ad hoc relationships, this can shift the prevailing culture, encouraging pro-social behaviour through predictability (Menzies and Angus, 2021). Meanwhile, in a public health context, Haggerty et al. (2003: 1220) argue that ‘management continuity’, comprising the right combination of consistency and flexibility, is needed in order to establish continuity of care. The authors note that this is crucial where different service providers must work together, and this may be analogous to the activity of teachers in a school, particularly at secondary school level, where students have a larger number of teachers.
Student-centric and institutional knowledge
As Atteberry et al. (2017: 22) point out, ‘it is more difficult to be effective at complex tasks when the task or context is unfamiliar’. Similarly, Allen and Sims (2018) argue that teachers constantly make professional judgements based on a mental database of past examples. While Allen and Sims (2018) primarily focus on repeated experiences teaching specific year groups or content, high turnover also means that teachers are less likely to hold examples of previous interactions with a specific student in their ‘mental database’.
Coleman (1988) highlights the importance of information flows in facilitating cooperation, and Ashcroft et al. (2016) argue that improved knowledge sharing through management information systems can improve ‘informational continuity’. Importantly, in the medical context described by Haggerty et al. (2003: 1220), ‘informational continuity’ does not only involve information about medical conditions, but also ‘patient’s preferences, values, and context’. It is therefore notable that in schools, management information systems typically hold information about student attainment and progress, but not these ‘softer’ forms of information.
According to Ronfeldt et al. (2013), institutional knowledge is often a casualty of teacher turnover, and this can undermine student achievement. There is a particular risk that high turnover might erode cultural understandings of ‘the way we do things around here’, a building block of school culture (Deal and Kennedy, 1983; Menzies et al., 2010). Given this, Ashcroft et al. (2016) emphasise the importance of culture building and institutional memory, citing the example of a school in which the headteacher brings all staff together at the end of the year to look back at the past and reflect on the future. Staff are asked to bring an idea with them for something they want to commemorate, and these memories are compiled on a large physical timeline. The timeline grows over time, and is reviewed each year ‘so that current members can walk along it as they tell the stories and remember together the stories of those now gone’ (Ashcroft et al., 2016: 59). High-turnover environments may make rituals such as these, which curate institutional knowledge and culture, more important.
Collaboration and collegiality
Ronfeldt et al. (2013) suggest that staff collaboration and collegiality can suffer as a result of the ‘disruptive effects’ of turnover, undermining student achievement. This hypothesis finds support in a later study showing that students attain more highly when teachers collaborate (Ronfeldt et al., 2015). Allen and Sims (2018) also emphasise the importance of ‘informal CPD’ (continuing professional development), and the exchange of information that takes place during teachers’ social interactions, such as end-of-the-week social events at the pub. There is a risk that turnover might disrupt these social rituals, but it is worth noting that there is no consensus in the literature regarding the relationship between turnover and collaboration or collegiality (Sims, 2021).
Teacher allocation: an underappreciated decision
As noted earlier, teacher turnover is not only driven by teacher attrition; between-, and within-school churn also play a role (Atteberry et al., 2017; Sims, 2020; Worth and De Lazzari, 2017). Patterns of teacher allocation, sometimes referred to as ‘assignment’ (Atteberry et al., 2017), can add to or diminish turnover, since allocation (both as a result of teacher preferences and of management decisions about timetabling) determines who students are taught by.
Although students who are disadvantaged (due to low prior attainment, socio-economic deprivation or membership of a minority ethnic group) might benefit the most from continuity, previous studies suggest that these students experience the most churn, and are most likely to be taught by new teachers. This is partly because they are more likely to attend schools where turnover is high, and partly because even within the same school, they are more often allocated to newer teachers (Atteberry et al., 2017; Ronfeldt et al., 2013). If ‘newness’ and disruption to continuity of care help to explain the negative effects of turnover (as I have argued above, and as Atteberry et al.  have done previously), then inequitable allocation patterns may have a disproportionate impact on these students.
The following sections therefore focus on allocation patterns, and how these might impact on different groups of students. I begin by synthesising existing research and new, large-scale survey data to explore how teachers’ preferences and allocation patterns might impact on the degree of continuity experienced by different groups of students. I then explore an approach to teacher allocation known as ‘looping’, which offers one way of increasing continuity.
Teacher allocations are not always made with students’ best interests in mind, but more effective schools appear to allocate teachers more equitably (Loeb et al., 2012). Research conducted in the largest school district in Florida suggests that influence over the timetable results in less experienced teachers being disproportionately matched to disadvantaged students in terms of ethnicity, socio-economic background and prior attainment (Grissom et al., 2015). Kalogrides and Loeb (2013) find that sorting effects (whereby minority ethnic and low socio-economic status students are more likely be taught by novice teachers) are primarily driven by prior attainment.
At present, there is a paucity of research on trends in teacher allocation within English schools, partly because English data sets do not normally allow student data to be matched to individual class teachers. This leads to an important gap in the evidence base, since allocation patterns have the potential to differ considerably from those reported in US studies. For example, Grissom et al. (2015: 7) conducted their research in a district where a collective bargaining agreement gave teachers a formal right to express preferences on class allocation, and where headteachers were required to take ‘into account seniority and employee preferences’. No such expectation exists in England. Furthermore, grade retention or repetition, whereby students repeat a year of their schooling, is fairly common – although controversial – in the US, but very rare in England (EEF, n.d.; Warren et al., 2014). This may affect the frequency with which students are taught by the same teacher in multiple years. This is particularly relevant to the discussion in the ‘Looping’ section. The data presented below therefore represent a tentative initial insight into possible trends and the drivers behind these, even though the data are drawn from a convenience sample (albeit a very large one).
Data from the teacher survey service Teacher Tapp show that more than two-thirds of teachers in England have a preference for teaching classes with higher attaining students, and almost one in five ‘strongly agree’ that they prefer teaching these classes (Allen and McInerney, 2019; see Figure 1). Meanwhile, more than half of teachers have some influence over their timetable – or which students they teach (see Figure 2). Taken together, these findings suggest a possible mechanism that might drive a flight away from low-attaining classes (Connolly et al., 2019), resulting in these students being more exposed to churn, and more frequently being taught by new teachers (who have had less opportunity to express a timetabling preference). Given that students from lower socio-economic backgrounds are concentrated in low-attaining classes, this trend would also be likely to have a disproportionate impact on them, such that English trends may mirror those reported in the US studies cited above.
Senior teachers in England appear to have more influence over allocation than their more junior colleagues, despite the lack of a formal right to influence allocation. However, their degree of influence over the timetable is linked to seniority rather than to years of experience.
Although the data reviewed above are only indicative, they provide grounds for believing that trends observed in the US – whereby established teachers’ preferences result in them moving away from less ‘desirable’ classes – may also apply in England. This risks having an inequitable impact on low-attaining classes (and the socio-economically disadvantaged students who are concentrated within them), resulting in a reduction in continuity of care. It should be noted, however, that, anecdotally, some senior teachers report seeing it as their responsibility to teach more ‘challenging’ classes, and this may counteract the posited trend.
‘Looping’ is a term used to describe students being allocated to the same teacher more than one year in a row – thereby reducing the degree of within-school churn that students and teachers experience, and thus increasing continuity of care. The practice is fairly rare in most Western school systems, although it is fairly common in Finland and has been documented in Germany, the United States, Sweden, Israel and Italy (Tourigny et al., 2020). Looping has been extensively studied in the US, but the strategy is under-researched in England. This is unfortunate, given that the absence of the type of regulations affording senior teachers more influence over class allocations (Grissom et al., 2015), and schools’ high degree of autonomy, mean that school leaders would have considerable freedom to implement the policy, if it were deemed to be beneficial. Moreover, new survey data from Teacher Tapp show that many teachers would be open to the approach, particularly if research showed it to be beneficial (see Figure 3 and Figure 4).
Given these findings, I now provide a brief summary of key studies on the impact of looping, including how these link to the view, proposed in this article, that turnover, continuity of care, and relationships and behaviour are closely linked.
In a small-scale qualitative study in a childcare setting, Hegde and Cassidy (2004) link looping to continuity of care. They find that teachers working with looped groups of students reported that the approach led to more trusting and secure relationships. These findings are consistent with a larger-scale study of students in Grades 2–5 in Florida, in which surveyed teachers and principals reported that looping led to improved relationships and increased the instructional time available, as well as the effectiveness of classroom instruction (Cistone and Shneyderman, 2004). The study also compared academic outcomes between a looped sample of students and a matched group, finding small improvements in reading, mathematics, progression and attendance. However, a quasi-experimental study of looping among Grade 4 students in a French Canadian school did not find a consistently positive impact on relationships, although it did find improvements in mathematics, writing and reading (Tourigny et al., 2020).
In the ‘Trust’ section, I argued that turnover may contribute to behavioural problems by eroding trust. Studies have also linked looping – and the continuity it brings – to improvements in behaviour. Albornoz et al. (2023) report improvements in behaviour among looped Grade 8 students in Chile, as well as improvements in attendance, progression and teacher expectations. A study of looping in high schools in Tennessee finds improvements in behaviour, alongside increased attendance and test scores (Wedenoja et al., 2020). Improvements in attainment linked to looping have been found in studies of mathematics attainment in Grades 3–5 in North Carolina (Hill and Jones, 2018), and in mathematics and English language arts in Indiana (Hwang et al., 2021).
It is worth noting that the effect sizes (based on different measures) reported in the above studies were generally small. However, the practice appears to have heightened benefits for disadvantaged (or ‘historically underserved’) students (Hwang et al., 2021; Wedenoja et al., 2020). This makes looping a potentially promising approach for schools seeking to reduce disparities in educational attainment, particularly given that the costs associated with allocation strategies are likely to be low, and that school leaders have considerable autonomy in the area (Ost and Schiman, 2015).
However, parents and students may have concerns about looping, if they fear that it will result in students remaining with a disliked teacher, or one whom they consider to be ineffective. Yet the least effective teachers appear to improve when working with a familiar class, with one study suggesting that looping could have an even greater impact on the effectiveness of less-skilled teachers (Hill and Jones, 2018). Moreover, Wedenoja et al. (2020) argue that teachers and students who do not initially form a good relationship might adjust to each other as they grow more familiar. This would be unsurprising, given the arguments made in the section ‘The impact of teacher turnover’. Nonetheless, teachers may be keen to move on from classes they find difficult, and it is notable that in the Teacher Tapp survey reported above, only around a quarter of secondary school teachers would want to continue teaching 75 per cent or more of their current students, even if research showed this to be beneficial. There is also a risk that overriding teacher preferences could cause resentment among teachers; if this were to drive teachers out of the profession, the strategy would be counterproductive. Yet the popularity of looping with teachers in Florida (Cistone and Shneyderman, 2004) offers some reassurance that it would not necessarily result in increased teacher attrition.
Another potential concern is that looping could undermine the benefits of ‘grade-specific experience’; if looping requires teachers to move up with a year group, they will inevitably have to teach new content, and this can generate additional workload and drive teachers out of the profession (Allen and Sims, 2018; Ost and Schiman, 2015). However, reassignment to an adjacent grade (as is the case with looping) may be less problematic than non-adjacent switches, and looping teachers may not experience the same challenges that are associated with other grade reassignments (Blazar, 2015).
Discussion and future work
Teacher turnover in all its forms deprives students of continuity of care, with deleterious effects. I have argued that this effect is likely to result from: reduced trust, which undermines collective action; loss of student-centric and institutional knowledge; and reduced professional collaboration.
Given that students’ sense of safety and predictability can be derailed when teachers’ departures undermine relationships, schools operating in high-turnover environments should pursue strategies that promote the development of trust, and which encourage cooperation. These might involve avoiding single-person dependencies by ensuring that students have strong relationships with several key adults, extra-curricular activities such as residential trips that accelerate the emergence of strong relational bonds and intentionally curating institutional memory and informal cultural practices. Formal and informal opportunities to exchange knowledge might also nurture esprit de corps and counteract the tendency of high turnover to undermine collaboration and collegiality.
The importance of consistency and routines is often emphasised in government guidance (Bennett, 2017), and I have argued that one of the reasons that this is important is that it can create predictability at an institutional level by maintaining ‘management continuity’ (Haggerty et al., 2003). Given that new staff can take time to learn ‘how things are done around here’ (Deal and Kennedy, 1983; Menzies et al., 2010), schools need to prioritise careful induction and mentoring of all new staff (rather than just trainee or newly qualified staff).
Teacher allocation strategies can increase or decrease teacher churn, which, in turn, determines students’ exposure to turnover. Schools should monitor how teacher preferences are aggregated into potentially harmful and inequitable allocation patterns, and take a research-informed approach to timetabling. Encouragingly, teacher allocation is an area over which school leaders in England have considerable autonomy, and there is a growing body of research around the benefits of different strategies. It also appears that teachers’ allocation preferences are malleable in response to evidence, although, as noted in the introduction, it is possible that teachers signing up to the survey app are more interested in, and responsive to, research and evidence than the average teacher. Although its measured effects tend to be small, looping provides one possible allocation strategy which might reduce churn. However, the potential for unintended consequences (particularly for teacher retention), and the tradeoffs with grade-specific experience, should be carefully weighed up.
Further research is needed on teacher allocation, to help school leaders choose between different strategies. Ideally, future studies would explore the potential for hybrid strategies that combine the benefits of student- and content-specific experience and studies, paying careful attention to potential ramifications for teacher satisfaction and retention.