How School Leaders Retain Experienced and Capable Teacher Mentors

Abstract

Purposes: Teacher turnover has especially negative effects on schools and students when experienced and capable teachers leave. This loss is significant when those teachers also serve as mentors to their less experienced colleagues. This study aimed to advance understanding about how school leaders can positively influence the retention of their school’s teacher mentors. Methodology: The framework for the study reflects a longstanding line of research on successful leadership. Using a cross-sectional research design, evidence was provided by responses to the mentor survey component of a larger four-year study examining the effects on retention decisions of a national induction programme for early-career teachers and their mentors in England. Structural equation modelling was employed to test the direct and indirect effects of school leadership and selected school conditions on mentors’ self-efficacy, well-being and job satisfaction, and ultimately retention decisions. Findings: Developing and retaining teacher mentors was associated with a suite of leadership practices which encourage collaborative cultures, provide coherent high-quality learning opportunities, and ensure what they perceive to be manageable workloads. These organizational conditions nurture the job satisfaction and self-efficacy of experienced teachers enhancing their sense of well-being at work. Implications: Results suggest four sets of guidelines for senior school leaders.

1. Purpose

Teacher turnover is a persistent, longstanding challenge in most educational jurisdictions (Sibieta, 2020; Sutcher et al., 2016). High teacher attrition rates have reached such a serious level that many governments cannot simply replace the increasing number of teachers leaving the profession (UNESCO, 2024). The purpose for this study was to better understand how those in school leadership positions, using resources in the school, can influence the retention of their school’s experienced and highly capable teachers who are serving as mentors to their less experienced colleagues.

Three reasons justified the purpose for the study. First, evidence now indicates that school leadership is a pivotal explanation for teachers’ decisions about leaving or staying in their schools (DeMatthews et al., 2022; Sipahioglu et al., 2025). However, what school leaders actually do to nurture retention remains underexamined. Second, while teacher turnover has generally negative effects on school organizations and student learning (Carver-Thomas & Darling-Hammond, 2019; Ronfeldt et al., 2013), its actual effects depend on which teachers leave. The departure of low-performing teachers provides schools with opportunities to improve the quality of their students’ learning experiences whereas the loss of highly effective teachers has mostly negative effects on those experiences (Grissom & Bartanen, 2019). Evidence about how to retain high-performing teachers is quite limited.

Moreover, research suggests that retaining experienced teachers is also a challenge for schools. Around a quarter of teachers with 8–15 years of experience and almost half of those with more than 24 years of experience reported a propensity to leave teaching in a national study on teachers’ work, lives and effectiveness in England (Day et al., 2007; Day & Gu, 2010). However, as yet, there is limited evidence on how schools can best retain their most experienced teachers. There is considerable evidence about teachers in general, as well as special education teachers (Billingsley & Bettini, 2019), teachers in hard-to-serve schools (Arthur & Bradley, 2023), second-career teachers (Hogg et al., 2023) and beginning teachers (den Brok et al., 2017). This study adds teachers considered to be “experienced and capable” by their schools and who serve as mentors to their less experienced colleagues. We refer to them subsequently as “teacher mentors” or “TMs”.

Finally, much research on mentoring has concentrated on documenting the benefits of granting experienced teachers the status and responsibilities of mentorship for their professional development and career advancement (e.g., Arnsby et al., 2025; Hobson et al., 2009; Little, 1990). Relatively less is known, however, about what school leaders do to encourage the retention of these experienced teachers serving as mentors to their colleagues.

The three research questions guiding were as follows:

• To what extent do a selected set of TMs’ personal resources (self-efficacy, job satisfaction, well-being) influence their retention decisions?
• Which organizational conditions are strongly and positively associated with those TMs’ personal resources?
• Which leadership practices are strongly and positively associated with those key organizational conditions?

2. Context

The site of this study, University College London, is one of five national providers of both the Early Career Framework programme (ECF, renamed as Early Career Teacher Entitlement from 2025) for teachers in the first two years of teaching and their mentors, and an equally ambitious set of learning experiences for the further development of specialist teachers, senior leaders, headteachers and executive leaders, the National Professional Qualifications (NPQs). The four-year longitudinal research, of which this study is one component, includes examining the effects of the ECF programme on teacher retention decisions. School leaders with teachers enrolled in the ECF were required to appoint a mentor for them from existing school staff. Once chosen, these mentors received structured additional training over a two-year programme (reduced to one year from 2024) but continued to teach in their schools with funded time provided by the Department for Education to fulfill their training and mentor duties.

While the primary goal of the ECF was to retain early-career teachers, retaining those teacher mentors and ensuring effective mentoring in schools was critical to the achievement of that longer term goal. This study was undertaken to help senior leaders identify conditions in their schools that TMs would consider favorable to their work.

3. Framework

Figure 1 summarizes the framework guiding this study. This framework serves both descriptive and explanatory purposes (Biesta et al., 2011). Descriptive purposes include selecting and organizing a sub-set of the more than 100 variables associated with teacher retention (e.g., Nguyen et al., 2020; Billingsley & Bettini, 2019). However, theories serving descriptive purposes only are not always ‘fit for purpose’ as they are limited in explaining how teacher retention decisions are made through the interaction of these variables (or factors).

Explanatory purposes are served when relationships among those variables are identified, especially when they are supported empirically. Research reviews suggest that many factors influencing teachers’ retention decisions are directly associated with features of the school organisation (e.g., Nguyen et al., 2020; See et al., 2020). Given similar student populations, in a high-retention successful school supporting the professional learning and development of teachers is found to be central to leadership strategies. These strategies are aimed at creating the social and intellectual environments which enable individuals and teams to engage with one another to improve practice (e.g., Kazemi et al., 2022; Coe et al., 2014). Building on well-established international research about the impact of successful leadership on teachers’ work and lives and student outcomes (e.g., Bellibaş et al., 2021; Day & Gurr, 2024), we developed a framework to connect variables in a relatively long, complex and largely unique chain from school leadership to teacher retention. While previous evidence has established a relationship between leadership and teacher retention, our framework explains the complex nature of that relationship.

This framework assumes that while TMs play a unique role in their schools, they are part of their school’s teaching staff and are influenced in their retention decisions by most of the same conditions as the rest of their teaching colleagues (e.g., Eleftheriadou et al., 2025; Gu et al., 2023). The framework reflects prior evidence about what motivates staff retention decisions, as well as results from our larger ECF research about what motivates ECF mentors, in particular (Leithwood et al., 2024). Structural equation modelling, a fundamentally theory-driven statistical technique to support theory development, was then used in this research to empirically test the complex relationships between variables in this framework, allowing us to encapsulate the connections between processes and outcomes of retention and formulate an empirically warranted theory of action that can be used by school leaders to improve the retention of their TMs.

The far-right box of Figure 1 includes TMs’ intentions about staying in their school, moving to a different school, or choosing to leave teaching (propensity to leave or stay). The most direct influences on these intentions are a small number of TMs’ personal resources and psychological states related to both their school and the profession of teaching including self-efficacy, feelings of satisfaction, and sense of well-being. These personal constructs are at least partly a function of a selected set of TMs’ schools’ organizational conditions, the most promising being opportunities to collaborate with other staff, opportunities for continuing their professional learning, as well as beliefs about the manageability of their workload. While senior school leaders are by no means the only influence on these organizational conditions, their influence is typically quite significant (Leithwood, 2019, Day et al., 2024). The remainder of this section provides a synthesis of the empirical evidence and reasoning which explains the key constructs specified in Figure 1 and the relationships among them.

3.1. School Leadership

While school leadership is just one of almost 100 factors associated with teacher retention by a large body of evidence, it has an outsized impact on teacher retention. A series of recent large-scale reviews of research begins to make this case. Palma-Vasquez et al. (2022), for example, report that lack of support from school leaders is a determining factor in teachers leaving the profession and moving schools. Nguyen et al. (2020) found the odds of teachers leaving schools (leavers and movers) with stronger administrative support are 0.80 times the odds of teachers leaving schools (leavers and movers) with weaker administrative support. Results related to specific groups of teachers also highlight the significance of school leadership in teachers’ ‘leaving’ decisions. Studies examining global ratings of school-based administrative support found that special educators were more likely to stay when they rated administrative support more highly (Billingsley & Bettini, 2019). According to qualitative research, the lack of administrative support was a significant school-level challenge for Latinx teachers leaving the profession (Unda, 2025). Perceived quality of school leadership was a major source of dissatisfaction for some career-change teachers (Hogg et al., 2023).

Summing up evidence about what school leaders do to influence teacher retention, Grissom and Bartanen (2019) explain that: more effective principals are better able to retain teachers because they create more positive school climates, supply teachers with greater support, provide more beneficial opportunity for professional growth, and otherwise positively shape teachers’ working conditions in ways that lead to greater job satisfaction and attachment (Ingersoll, 2001; Johnson et al., 2012; Kraft et al., 2016).

While this prior evidence indicates that school leaders have a significant influence on teachers’ retention decisions, what leaders actually do to exercise that influence remains unhelpfully vague. One of the main purposes for this study was to identify more specific leadership practice accounting for their influence on TMs’ retention decisions. To do this, the study adopted, as part of its framework, an approach to school leadership which includes a detailed set of practices that considerable prior research indicates are successful for a wide range of purposes. This approach is based on a longstanding line of research on successful leadership carried out in the UK (Day et al., 2011), Canada (Leithwood, 2012) and the US (Leithwood & Louis, 2012). Considerable evidence now indicates that this approach to leadership has demonstrable value in helping schools in quite different contexts achieve a range of improvement goals related to teaching and learning.

This approach, informed by an “integrated”1 conception of leadership, includes four “domains” or categories of leadership practices (22 in total)—Setting Directions, Building Relationships and Developing People, Redesigning the Organization, and Improving the Teaching Programme. Practices included in the Setting Directions domain have a very strong influence on how practices in the remaining domains are enacted, as would be expected. The outcome of Setting Directions practices—a widely understood and compelling sense of direction for the school’s work with students—provides the purposes and focus for leadership practices encompassed in the three other domains (Day et al., 2016). This conception of school leadership includes many of the leadership approaches and practices associated with teacher job satisfaction, well-being and retention reported in prior research (e.g., Borman & Dowling, 2008; Griffith, 2004), for example, regular and supportive communication between school leaders and their teaching colleagues. Such associations are achieved through paths of influence that are embedded in the organisational conditions created and shaped by school leadership.

3.2. Organizational Conditions

This study included three sets of organizational conditions which prior evidence suggests have significant indirect effects on teacher retention—Workload, Collaborative Cultures, and the extent to which teachers have access to Professional Learning Opportunities. School leaders have considerable potential influence on these three conditions in their schools. Workload. In the context of workload effects on job satisfaction, it is how teachers and mentors feel about or perceive their workload, as distinct from some objective measure of workload such as time spent working. The conception and measure of workload used in this study reflects this emphasis on TMs’ perspectives. These were perspectives on the adequacy of the time they had to carry out their teaching and mentoring responsibilities, the support they received from their school, and the contribution that mentoring had on their own professional development. Substantial amounts of robust evidence in the research literature also demonstrate a significant association between this conception of teachers’ workload and both their job satisfaction (e.g., Butt & Lance, 2005; Hughes, 2012; Toropova et al., 2021) and sense of well-being at work (Nwoko et al., 2023). An earlier study in our large scale four-year research programme reported strong evidence of between-school variation in TMs’ perceived workload: “In schools where leadership is effective in enabling internal and external collaboration for educational improvement and where the culture and conditions are conducive to teaching, learning and professional growth, TMs are more likely to report that they have adequate time to fulfil their mentoring role” (Gu et al., 2023). Collaborative School Cultures. These cultures are defined as the extent to which teachers and TMs are encouraged to work with one another (as in Kardos and Johnson’s (2007) “integrated professional cultures”), to engage with parents/carers in their school’s improvement efforts, and to build wider community support for those efforts. “Collaboration”, explains Admiraal and Kittelsen Røberg (2023), “may reduce feelings of isolation, and thereby reduce burnout, by increasing teacher job satisfaction, teacher confidence, and student achievement in their classes” (p. 6). Viewed as a form of participation in decision making, collaborative cultures are among those working conditions strongly related to job satisfaction (Zakariya, 2020). Borman and Dowling’s (2008) meta-analysis reported substantial reductions in turnover among teachers who participated in school-based teacher networks and had other opportunities for collaboration with their colleagues. Professional Development Opportunities. The study’s conception of such learning opportunities includes the extent to which TMs believed they had opportunities to take on new challenges, develop their own classroom teaching skills and participate in other professional learning opportunities. Unrealistic expectations for their performance by their schools, as well conflicts between their work schedules and learning opportunities are associated with reduced learning opportunities. Prior research has found a significant relationship between teacher job satisfaction and professional learning opportunities, especially for teachers who feel a strong need to improve their content and pedagogical skills (Smet, 2022). Reviews of research by both Borman and Dowling (2008) and See et al. (2020) report significantly less attrition among teachers who have and engage in professional learning opportunities.

3.3. Personal Resources as Response to Organizational Conditions

Self-efficacy, well-being, and job satisfaction are included in the framework for this study to explain the role of teachers’ cognitive and psychological responses to organisational conditions in influencing their propensity to leave or stay. Given their disciplinary origins in psychology and positive psychology in particular, these personal constructs tend to be treated as attitudinal attributes which are found to have direct associations with teachers’ retention decisions (e.g., Department for Education, 2023). However, evidence from our larger research shows that when viewed through a social cognitive lens, the meaning of these constructs and how they are formed in context are better understood as outcomes of social and cultural processes in the school organisation. This is the case because the extent to which teachers and mentors are satisfied with their jobs, confident about their teaching (i.e., efficacy), and positive about the quality of their working lives at school (i.e., well-being in school) are largely dependent on the quality of in-school professional learning cultures that are shaped by school leadership (Gu et al., 2023; Leithwood et al., 2024; Gu et al., 2025).

Moreover, the literature on teacher job satisfaction and well-being identifies two dimensions of each personal construct (Liu et al., 2023; Zakariya, 2020): one dimension related to the school context in which the teacher works, and the other dimension related to the broader teaching profession. The same teacher might have high levels of job satisfaction and well-being on one dimension but not on the other. Evidence about the domain specificity of self-efficacy (Bandura, 1997) suggests that it is likely to have these two dimensions, as well.

Job Satisfaction in school is indicated by the sense of fulfilment and gratification that TMs experience through work in their schools. This form of job satisfaction includes feelings of enjoyment at work, the likelihood that mentors would recommend their schools to others as a good place to work, and mentors’ unwillingness to change schools. Job satisfaction with teaching as a profession includes teachers believing the advantages of being a teacher clearly outweighing the disadvantages, agreement that they would choose teaching again if they had a chance to decide and not regretting their choice. Teachers’ estimates of their job satisfaction may be attributed to factors that are clearly positive, typically those associated with sources of intrinsic motivation, as well as those that are clearly negative, often those associated with sources of extrinsic motivation (Malik & Naeem, 2013; Sergiovanni, 1967).

Well-Being in school is indicated by feelings of belonging and being able to “be themselves”, as well as believing they have colleagues who care about them and treat them with respect. A positive sense of well-being in the teaching profession is indicated by teachers believing that they are good at helping students learn, have accomplished a lot as a teacher, and feel like their teaching is effective and helpful. Although often treated as distinct from well-being, teacher stress and burnout are negative expressions of well-being (e.g., Agyapong et al., 2022). Evidence indicates that these expressions of poor well-being have significant negative effects on both job satisfaction and teacher turnover (Madigan & Kim, 2021).

Longstanding evidence associates both job satisfaction and well-being with retention decisions (e.g., Butt & Lance, 2005; Doan et al., 2023; Struyven & Vanthournout, 2014). In addition, researchers have suggested that students of teachers who are satisfied with their jobs and have a strong sense of psychological well-being tend to be more successful academically than those who are emotionally exhausted or unsatisfied (Arens & Morin, 2016; Lopes & Oliveira, 2020; McInerney et al., 2018).

Self-Efficacy is typically conceptualized as a person’s anticipation of future success (Bandura, 1997). This study measured TMs’ beliefs about past successes in their role—their beliefs about their effectiveness in establishing a strong relationship with their mentees, meeting their individual needs, helping them establish good social relationships in the school, and contributing to their pedagogical skills along with wider professional skills. While these beliefs reflect TMs’ judgements of success with past experiences, they contribute to confidence in the success of their future efforts. These are among the mastery experiences which, as Bandura’s (1997) theory and considerable additional empirical evidence (e.g., Donohoo et al., 2020) indicate, have the strongest effects on teacher self-efficacy.

A positive sense of self-efficacy is significantly related to teacher job satisfaction (Kasalak & Dağyar, 2020), while negative feelings of self-efficacy contribute to teacher burnout, negative feelings of well-being and attrition (Aloe et al., 2014). Recent evidence confirms a significant relationship between teacher self-efficacy and both job satisfaction (Kasalak & Dağyar, 2020) and well-being (Zee & Koomen, 2016).

3.4. Propensity to Stay or Leave

To determine the retention decisions of TMs, our research asked them about their plans for next year: were they staying in the same school, staying in teaching but moving schools, or leaving teaching. The first two of these options allowed respondents to indicate whether they would be taking on equivalent, additional, or reduced responsibilities. Labelled “propensity to leave”, we adopted this approach to measuring retention and turnover because of evidence suggesting that propensity to leave is a strong predictor of actual turnover (Fried et al., 2008). In addition, role stress, an indicator of poor well-being, appears to be more strongly related to propensity to leave than to actual turnover (Fried et al., 2008). By extension, we argue that teachers’ sense of positive well-being in workplaces is strongly linked to their intention to stay.

4. Methods

4.1. Design

This study is part of a larger four-year, longitudinal, mixed methods study gathering quantitative and qualitative evidence on the impact of England’s ECF induction programme on participating early-career teachers’ and mentors’ work engagement, well-being, and retention trajectories over time. The cross-sectional study reported in this paper used results from the first annual survey of TMs.

4.2. Sample

TMs in this study were appointed by their senior school leaders to act as mentors of colleagues who were in the first two years of their teaching careers and enrolled in a national induction programme in England, the Early Career Framework (ECF). All things being equal, and as required by the government funding criteria, we assumed that selecting a mentor for their early-career teachers would motivate most school leaders to recruit from the pool of their most experienced teachers whose professional record and capability are highly regarded by their peers. Reflecting this assumption, evidence from a representative sample of early-career teachers in our larger study (Gu et al., 2023) reports the presence of the qualities required by the ECF policy in their mentors, including, for example, their communication skills and ability to build trusting relationships with their mentees. Identifying leaders’ mentor choices as highly experienced and capable is further supported by evidence indicating that senior school leaders are able to accurately differentiate the quality of their teachers’ performance (e.g., Grissom & Loeb, 2017; Sartain et al., 2011) relying, as in this case, mainly on their own observations and experiences with teachers in making this judgement (Goldring et al., 2015; Grissom et al., 2016; Master, 2014).

Respondents included in the study were a sample of TMs in the first year of the two-year UCL-led ECF programme. Survey invitations were sent to all TMs who were active in the programme (N = 5853) on the date of report extraction in June 2022. Online surveys were conducted on the JISC (2022) platform with each participant receiving a personal URL survey link. Between June and October 2023, 810 TMs responded, representing 14% response rate. The sample comprised 630 female participants (78%), 693 TMs were white (85%), with an age range of 23–65 (mean age = 39.9, SD = 9.23). Half (n = 407, 50%) of the TMs worked in secondary schools, with 345 (43%) in primary schools. Survey responses were coded and pseudonymized, helping to minimise common response bias effects through respondents being apprehensive to answer openly and honestly (Podsakoff et al., 2024).

There are two reasons to be confident about the external validity of this study’s findings. First, demographic characteristics of the sample are broadly representative of national figures for TMs regarding gender, ethnicity, and contract types (e.g., full-time permanent), as well as school phase (

Table 1. Survey participants and national mentor demographics.

	Mentors	National Mentors (Department for Education, 2022)
Gender	630 (78%) female	18,574 (75%) female
Ethnicity	693 (85%) white	20,281 (86%) white
Age Range	23–65
Mean (SD)	39.9 (9.23)
21–29	116 (14%)	4464 (18%)
30–39	312 (39%)	10,175 (41%)
40–49	222 (28%)	6732 (27%)
50–59	140 (17%)	2987 (12%)
60–69	17 (2%)	253 (1%)
Secondary school	407 (50%)	13,069 (52%)
Middle school	1 (<1%)
Primary school	345 (43%)	10,755 (43%)
Other	57 (7%)	1071 (4%)
Full-time permanent contract	672 (83%)	All full-time, 20,165 (81%)
Full-time fixed term/temporary contract	5 (<1%)
Part-time permanent contract	127 (16%)	All part-time, 3668 (15%)
Part-time fixed term/temporary contract	6 (<1%)

). Second, a priori power analysis indicated a sample of 112 respondents with α = 0.05 and df = 200 would have 80% power to reject a mis-specified model with RMSEA = 0.05 (Moshagen & Erdfelder, 2016); as such, our sample size was deemed sufficient. Post hoc power analysis indicated our sample of 898 TMs retained 99% power to reject a mis-specified model with RMSEA = 0.06 with α = 0.05 and df = 1929. The results of this research should therefore be highly relevant to the experiences of the national TM population.

4.3. Instrument

The instrument used to collect data from mentors included a ten-section, 66-question survey. Responses to questions in 8 of the 10 sections were rated on a scale of 1–6, with 1 representing the least positive response and 6 the most positive response, except in instances where questions were negatively valenced; responses to these questions were reverse coded.

The measurement components of each of the key constructs (e.g., school leadership, job satisfaction, self-efficacy, and well-being) are informed by the synthesis of the latest research outlined above and grounded also in the survey research in related fields led by authors in different national contexts over twenty years (e.g., Leithwood & Jantzi, 2006; Day et al., 2016; Leithwood et al., 2019). Together, they provide a secure knowledge and empirical base for the research team to develop a reliable and valid survey instrument which serves the purpose of this study (as supported by the results of the survey analysis outlined later in this paper). We conducted a pilot study with 24 ECTs to test and refine the survey instruments. Not surprisingly, results showed high levels of reliability for almost all scales (above >0.85).

The final two sections of the survey requested respondents’ intended destination for the next academic year (i.e., staying in school, moving school, leaving the profession) and demographic data including gender, ethnicity, and phase of their school. Demographics questions were categorical and included a mixture of multiple and single choice options. School context data were added to include school inspection ratings from the Office for Standards in Education, Children’s Services and Skills (Ofsted) (an indicator of a school’s performance and quality of services provision), percentage of pupils eligible for Free School Meals (a key indicator of the socioeconomic disadvantage of the school community), and geographical locations and regions (Department for Education, 2022). Inclusion of different question formats and negatively valenced survey items aimed to minimise common method bias caused by a lack of variety in scale properties (Podsakoff et al., 2024).

Measures of each of the variables in the framework, along with both unstandardized and standardized factor loadings are described in

Table 2. Final survey items, unstandardized (SE) and standardized item loadings for CFA.

Factor	Item	Unstandardised Factor Loading (SE)	Standardised Factor Loading
Learning and engagement with ECF mentoring programme	(1) To what extent have you engaged with mentor training on the ECF programme?	1	0.440
	(2) To what extent has your learning from the mentoring programme contributed to your confidence as a mentor to establish a constructive mentoring relationship?	2.712 (0.242)	0.901
	(3) To what extent has your learning from the mentoring programme contributed to your confidence as a mentor to support and challenge your ECT(s)?	2.588 (0.229)	0.952
	(4) To what extent has your learning from the mentoring programme contributed to your confidence as a mentor to provide effective developmental teaching observations and feedback?	2.438 (0.221)	0.866
In-school support and development for ECF mentors	(1) I have adequate time to carry out my role as a mentor.	1	0.877
	(2) I have adequate support from my school as a mentor.	0.682 (0.035)	0.710
	(3) The amount of time allocated to mentor-mentee sessions on ECF is appropriate to support my ECT(s).	0.947 (0.040)	0.811
	(4) Being a mentor has contributed to my own professional development.	0.382 (0.046)	0.448
Successful leadership practices (a) Setting directions	Think about the person (or people) who provide(s) THE MOST SENIOR leadership in your school (e.g., Headteacher). To what extent do you agree that they do the following regarding setting direction?
	(1) Gives staff a sense of overall purpose.	1	0.908
	(2) Demonstrates high expectations for staff’s work with learners.	0.858 (0.033)	0.881
	(3) Demonstrates high expectations for learner behaviour.	0.885 (0.042)	0.846
	(4) Demonstrates high expectations for learners’ academic achievement.	0.750 (0.046)	0.84
	(5) Demonstrates high expectations for learners’ development of good health and well-being.	0.870 (0.036)	0.867
Successful leadership practices (b) Developing teachers	Think about the person (or people) who provide(s) THE MOST SENIOR leadership in your school (e.g., Headteacher). To what extent do you agree that they do the following regarding developing teachers?
	(1) Gives us individual support to help us improve teaching practices.	1	0.876
	(2) Encourages us to consider new ideas for teaching.	0.840 (0.036)	0.854
	(3) Promotes leadership development among teachers.	0.942 (0.037)	0.86
	(4) Promotes a range of continuing professional development experiences among all staff.	0.945 (0.038)	0.865
	(5) Encourages us to think of learning beyond the academic curriculum (e.g., persona, emotional and social education, citizenship, etc.).	0.917 (0.039)	0.847
Successful leadership practices (c) Redesigning the organisation	Think about the person (or people) who provide(s) THE MOST SENIOR leadership in your school (e.g., Headteacher). To what extent do you agree that they do the following regarding redesigning the organisation?
	(1) Encourages collaborative work among staff.	1	0.817
	(2) Engages parents/carers in the school’s improvement efforts.	1.012 (0.056)	0.774
	(3) Builds community support for the school’s improvement efforts.	1.127 (0.055)	0.843
	(4) Allocates resources strategically based on learners’ needs.	1.179 (0.055)	0.879
	(5) Works in collaboration with other schools.	0.999 (0.057)	0.755
Successful leadership practices (d) Improving the teaching programme	Think about the person (or people) who provide(s) THE MOST SENIOR leadership in your school (e.g., Headteacher). To what extent do you agree that they do the following regarding managing the teaching programme?
	(1) Provides or locates resources to help us improve teaching.	1	0.886
	(2) Regularly observes classroom activities.	0.938 (0.040)	0.772
	(3) After observing classroom activities, works with teachers to improve teaching.	1.141 (0.035)	0.882
	(4) Uses coaching and mentoring to improve quality of teaching.	1.098 (0.034)	0.866
	(5) Encourages all staff to use learners’ progress data in planning for individual learners’ needs.	0.772 (0.038)	0.812
Professional growth opportunities	(1) I have many opportunities to take on new challenges.	1	0.793
	(2) I have adequate opportunities to develop my classroom teaching skills.	1.017 (0.042)	0.859
	(3) I have adequate opportunities for learning and development as a professional.	1.077 (0.044)	0.88
	(4) Opportunities for promotion within my school are adequately available to me.	1.253 (0.059)	0.726
	(5) Expectations of my performance are realistic given my role and experience.	0.855 (0.056)	0.69
	(6) Training and development rarely conflicts with my work schedule.	0.848 (0.059)	0.558
Collaborative culture	(1) Teachers in our school mostly work together to improve their practice.	1	0.809
	(2) I have good relationships with my colleagues.	0.256 (0.029)	0.428
	(3) My school has a culture of shared responsibility for school issues.	1.095 (0.054)	0.871
	(4) There is a collaborative school culture which is characterised by mutual support.	1.126 (0.058)	0.907
Mentors’ workload conditions	(1) I have sufficient access to appropriate teaching and learning materials and resources.	1	0.562
	(2) I am protected from administrative duties that interfere with my teaching.	2.069 (0.152)	0.823
	(3) I have adequate time for lesson planning and using assessment to improve learning.	2.227 (0.163)	0.909
	(4) I have adequate time to balance pastoral duties (i.e., care for pupils’ physical and emotional welfare) with teaching.	2.191 (0.158)	0.913
Self-efficacy as a mentor	(1) Overall I have been able to meet the individual needs of my early career teacher(s) (ECTs) as a mentor.	1	0.614
	(2) I have been able to establish a strong mentor-mentee relationship.	0.617 (0.065)	0.578
	(3) I have been able to effectively address my mentee’s/mentees’ learning demands in mentoring conversations.	1.080 (0.075)	0.695
	(4) My role as a mentor is meaningful to the development of my ECT’s teaching practice.	1.114 (0.082)	0.782
	(5) My mentoring has contributed to the pedagogical skills of my mentee.	1.180 (0.094)	0.764
	(6) My mentoring has contributed to the behaviour management skills of my mentee.	1.261 (0.110)	0.774
	(7) My mentoring has helped my mentee to develop their wider professional skills.	1.239 (0.103)	0.783
	(8) My mentoring has helped my mentee to build good social relationships in the school.	1.290 (0.150)	0.663
Satisfaction in teaching	(1) The advantages of being a teacher clearly outweigh the disadvantages.	1	0.821
	(2) If I could decide again, I would still choose to work as a teacher.	1.215 (0.063)	0.851
	(3) I regret that I decided to become a teacher. *	1.009 (0.061)	0.764
Job satisfaction in school	(1) I would like to change to another school if that were possible. *	1	0.69
	(2) I enjoy working at this school.	0.838 (0.058)	0.903
	(3) I would recommend my school as a good place to work.	0.977 (0.065)	0.882
Well-being in teaching	(1) I am good at helping pupils learn new things.	1	0.781
	(2) I have accomplished a lot as a teacher.	1.348 (0.124)	0.778
	(3) I feel like my teaching is effective and helpful.	1.201 (0.077)	0.859
Well-being in school	(1) I feel like I belong at this school.	1	0.837
	(2) I can really be myself at this school.	1.066	0.846
	(3) I feel like people at this school care about me.	0.984	0.824
	(4) I am treated with respect at this school.	0.977	0.86

Note: Asterisk (*) indicates the items that have been reverse coded for CFA analysis.

. While the majority of factor loadings are high, signifying a strong correlation between the observed variable and the latent factor, a small number of variables (n = 3) have lower factor loadings (specifically, 0.428, 0.440 and 0.448). These variables were not removed from the factor analysis because they were considered theoretically central to the constructs measured and difficult to replace due to the number of available items in our survey. Although higher factor loadings (i.e., above 0.5 for standardised) are preferred for construct validity, the use of standardised factor loadings above 0.4 has been considered acceptable depending on the theoretical rationale (Brown, 2015; Kline, 2016).

4.4. Analysis

Unimputed data from responses to Section 1, Section 2, Section 3, Section 4, Section 5 and Section 6 of the survey were collated and tested for construct validity using confirmatory factor analysis (see

Table 3. CFA fit indices for final scales.

Scale	X2	df	X2/df	CFI (Robust)	TLI (Robust)	RMSEA (Robust)
(i) Learning and engagement with ECF mentoring programme	0.564	2	0.282	1	0.999	0.016
(ii) In-school support and development for ECF mentors	1.926	2	0.963	1.000	1.000	0.029
(iii) Successful leadership practices 1 (a) Setting directions (b) Developing teachers (c) Redesigning the organisation (d) Improving the teaching programme	81.053	164	0.494	0.997	0.997	0.023
(iv) Professional growth opportunities	21.245	9	2.361	0.992	0.986	0.053
(v) Collaborative culture	2.039	2	1.020	1	1	0.033
(vi) Mentors’ workload conditions	1.976	2	0.988	1	1	0.033
(vii) Self-efficacy as mentor	27.641	20	1.382	0.986	0.981	0.040
(viii) Job satisfaction 2 (a) Satisfaction in teaching (b) Job satisfaction in school	12.172	8	1.522	0.994	0.989	0.040
(ix) Teacher well-being 3 (a) Well-being in school (b) Well-being in teaching	14.286	13	1.099	0.995	0.992	0.034

Note: ¹ Four-factor model of successful leadership practices. ² Two-factor model for job satisfaction, one related to general professional satisfaction and engagement, the other related to the mentor’s school context. ³ Two-factor model for well-being, one related to general professional well-being, the other related to the mentor’s school context.

). Scales were adjusted, with one item removed from engagement with and learning from the mentoring programme, two items removed from support and development of mentors, and one item removed from job satisfaction. Factor analysis supported a four-factor model for successful leadership practices (setting direction, developing teachers, redesigning the organisation, and improving the teaching programme) and a two-factor model for both job satisfaction and well-being (teaching-related and school-related). Factor scores were created from remaining items and tested for internal reliability using Cronbach’s alpha (George & Mallery, 2003). Inter-factor correlation analysis reported in

Table 4. Factor means scores, reliability, and inter-factor correlations.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
(1) Learning and engagement with ECF mentoring programme	1
(2) In-school support and development for ECF mentors	0.368 **	1
(3) Successful leadership practices–setting directions	0.102 **	0.321 **	1
(4) Successful leadership practices–developing teachers	0.144 **	0.336 **	0.754 **	1
(5) Successful leadership practices–redesigning the organisation	0.124 **	0.325 **	0.735 **	0.794 **	1
(6) Successful leadership practices–improving the teaching programme	0.142 **	0.320 **	0.696 **	0.825 **	0.781 **	1
(7) Professional growth opportunities	0.156 **	0.398 **	0.582 **	0.711 **	0.645 **	0.624 **	1
(8) Collaborative culture	0.137 **	0.303 **	0.620 **	0.659 **	0.698 **	0.636 **	0.680 **	1
(9) Mentors’ workload conditions	0.179 **	0.457 **	0.512 **	0.604 **	0.598 **	0.598 **	0.691 **	0.672 **	1
(10) Self-efficacy as a mentor	0.245 **	0.314 **	0.230 **	0.249 **	0.210 **	0.253 **	0.232 **	0.204 **	0.189 **	1
(11) Satisfaction in teaching	0.161 **	0.290 **	0.263 **	0.270 **	0.276 **	0.236 **	0.373 **	0.337 **	0.397 **	0.166 **	1
(12) Job satisfaction in school	0.081 *	0.316 **	0.572 **	0.572 **	0.532 **	0.452 **	0.587 **	0.603 **	0.493 **	0.169 **	0.437 **	1
(13) Well-being in teaching	0.111 **	0.197 **	0.352 **	0.373 **	0.366 **	0.336 **	0.402 **	0.378 **	0.350 **	0.331 **	0.358 **	0.352 **	1
(14) Well-being in school	0.115 **	0.323 **	0.544 **	0.540 **	0.535 **	0.449 **	0.667 **	0.652 **	0.555 **	0.183 **	0.429 **	0.741 **	0.524 **	1
(15) Destination	0.035	−0.025	−0.123 **	−0.142 **	−0.127 **	−0.085 *	−0.136 **	−0.061	−0.043	0.037	−0.110 **	−0.281 **	−0.012	−0.188 **	1
Mean (SD)	40.28 (10.16)	40.25 (10.17)	50.33 (0.89)	40.92 (10.08)	40.94 (0.99)	40.68 (10.13)	40.75 (0.97)	50.14 (0.85)	40.49 (10.08)	50.43 (0.57)	40.74 (10.20)	40.95 (10.11)	50.61 (0.54)	50.22 (0.91)	-
Reliability α	0.869	0.804	0.940	0.933	0.909	0.921	0.876	0.843	0.879	0.887	0.854	0.835	0.838	0.908	-

Note: ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

indicate variables significantly associated with the outcome of teacher destination.

Data were analysed using a structural equation model (SEM) in R Studio version 4.3.1 (RStudio Team, 2020) using the lavaan package (Rosseel, 2012). Pattern analysis of missing independent variable data indicated that 97% of the variables had at least one value missing, with 12% missing in total. Little’s MCAR test on the independent variables was significant (X² = 6810.387, DF = 5607, p = 0.000), indicating that data were not missing completely at random. However, there did not appear to be a high number of unfinished surveys, or high levels of missingness related to individual question items. As well, the missing data were across a range of variables predicted by other variables within the model. For these reasons missing data were taken to be plausibly MAR (Sterne et al., 2009) and multiple imputation deemed appropriate (Dettori et al., 2018).

We treated missing independent variable data using multiple imputation in SPSS version 28 (IBM, 2023). Twenty imputations were considered appropriate to the proportion of missing data (Schafer & Graham, 2002). With multiple imputation, a large number of repetitions for precise estimates is not needed. For example, ten imputations were found 95% efficient with 50% missing data; to remove noise from other statistical summaries (e.g., significance levels or probability values), 20 imputations were found effective (Schafer & Graham, 2002, p. 165).

Missing categorical outcome data (n = 12) were deleted list-wise as a minimally biased method where missing outcome data comprises less than 20% of the sample (Chen & Åstebro, 2003). This left a final sample of 798 respondents for analysis. An iterative approach was used to build the SEM structure. All variables were included in the initial structural model, based upon the study’s conceptual framework. Non-significant or confounding pathways were trimmed, and the final model retained. The retained SEM structure was applied separately to each imputed data set in R Studio and parameter estimates pooled using Rubin’s rules (Rubin, 1976). Model fit indices were considered separately for each imputed dataset using established cut-off values, with the caveat that cut-off values are frequently based upon ML estimation and not well established for DWLS estimation (Xia & Yang, 2019), thus cut-offs were treated with some caution.

5. Results

5.1. Descriptive Data Analysis

Inter-factor correlation analysis indicated that all independent variables included in the framework (Figure 1) were positively correlated with each other (Table 4). Additionally, leadership practices and professional growth opportunities, as well as job satisfaction both in teaching and in school, and well-being in school were significantly positively correlated with teacher destination: the more positive teachers were about these factors, the more likely they were to indicate they were staying in their school for the next academic year. Mean scores for all independent variables were in the range of 4.25 to 5.61 on the six-point response scale suggesting that the TMs were largely positive about their school environment, their mentoring experiences, and their personal resources and responses to school organisational conditions related to teaching and mentoring.

Not all independent variables were directly positively correlated with the outcome variable (i.e., learning and engagement with the mentor programme, support and development for mentors in school, collaborative culture, teaching working conditions, mentor self-efficacy, and well-being in teaching). Because these items were significantly positively correlated with other variables, however, they were retained in the structural equation model as mediating factors.

5.2. Structural Equation Model

A structural equation model (Figure 2) was used to test the effects on TMs’ destinations of:

• Mentoring-related factors (i.e., learning and engagement with the ECF mentoring programme, in-school support and development for ECF mentors);
• School-related factors (i.e., leadership practices, professional growth opportunities, collaborative culture, mentors’ workload conditions);
• Personal resources (i.e., mentor self-efficacy, job satisfaction in teaching, job satisfaction in school, well-being in teaching, and well-being in school).

Model fit was assessed as good across twenty imputed data sets (

Table 5. SEM fit indices for imputed data sets.

Imputation	X2	df	p	X2/df	CFI	TLI	RMSEA	SRMR
1	4158.86	1929	0	2.16	0.99	0.99	0.038	0.054
2	4141.608	1929	0	2.15	0.99	0.99	0.038	0.054
3	4155.395	1929	0	2.15	0.99	0.99	0.038	0.054
4	4122.273	1929	0	2.14	0.99	0.99	0.037	0.054
5	4175.25	1929	0	2.16	0.99	0.99	0.038	0.055
6	4182.104	1929	0	2.17	0.99	0.99	0.038	0.055
7	4184.85	1929	0	2.17	0.99	0.99	0.038	0.055
8	4161.23	1929	0	2.16	0.99	0.99	0.038	0.055
9	4158.213	1929	0	2.16	0.99	0.99	0.038	0.055
10	4124.104	1929	0	2.14	0.99	0.99	0.038	0.054
11	4152.373	1929	0	2.15	0.99	0.99	0.038	0.055
12	4142.722	1929	0	2.15	0.99	0.99	0.038	0.054
13	4150.603	1929	0	2.15	0.99	0.99	0.038	0.054
14	4117.086	1929	0	2.13	0.99	0.99	0.037	0.054
15	4153.187	1929	0	2.15	0.99	0.99	0.038	0.055
16	4190.196	1929	0	2.17	0.99	0.99	0.038	0.055
17	4171.074	1929	0	2.16	0.99	0.99	0.038	0.055
18	4167.009	1929	0	2.16	0.99	0.99	0.038	0.055
19	4165.225	1929	0	2.16	0.99	0.99	0.038	0.054
20	4093.029	1929	0	2.12	0.99	0.99	0.038	0.054

) with model fit indices of a CFI and TLI of 0.99 across all data sets, a RMSEA value of maximum value 0.038 and SRMR 0.055. Thus, all model fits were within established acceptable ranges of CFI > 0.95, TLI > 0.95 and RMSEA < 0.06; SRMR values were slightly high, but within acceptable range of <0.09 when in combination with TLI > 0.95 and RMSEA < 0.06 (Hu & Bentler, 1999).

Four groups of latent constructs were identified in the SEM (as indicated by four different shadings in Figure 2) predicting mentors’ propensity to leave or stay. Standardised path coefficients presented in Figure 2 represent the strength of relationships between the variables. To interpret how meaningful the effect sizes are in practise we used the following criteria: values below 0.3 considered as weak associations; values above 0.3 and below 0.7 considered as moderate associations; and finally values above 0.7 considered as strong associations.

As Figure 2 indicates among the four sets or domains of leadership practices, Setting Directions has a strong influence on the remaining three leadership domains providing, we assume, purposes for enacting practices in those domains. Redesigning the Organization has moderate effects on the development of Collaborative Cultures with a strong influence through Job Satisfaction to TMs’ sense of well-Being in school. Practices associated with Building Relationships and Developing People strongly influence TMs’ perceptions of their professional growth opportunities with indirect effects on their Job Satisfaction and Well-Being in school through influences on the school’s Collaborative Culture. Practices included in the domain Improving the Teaching Programme strongly influence TM’s perceptions of the quality and manageability of their Working Conditions which, in turn, has weak but significant direct influences on TMs’ Job Satisfaction.

Although considerable prior evidence suggests that self-efficacy has important effects on teachers’ attitudes and practices, its role in Figure 2 is a decidedly secondary one. Self-efficacy exercises moderately significant, supplementary influences on variables in the SEM not strongly related to the main paths of leadership influence on TMs’ retention decisions. The main influences on self-efficacy are experiences derived from the ECF programme as it is treated within the school.

6. Discussion

Based on evidence provided by a sample of more than 800 ECF teacher mentors, the purpose of this research was to identify conditions in schools that influenced the nature and value of TMs’ work and their decisions to stay in teaching, move schools, or leave the profession. Based on a considerable amount of prior evidence, the framework guiding the research assumed that school leaders were a significant influence on key conditions in the school that were likely to shape TMs’ propensity to stay or leave the school and possibly the teaching profession. The framework also assumed that the effect of these key conditions on TMs’ propensity to leave or stay would be mediated most directly by their job satisfaction and sense of well-being.

• Research Question 1: Influence of Personal Resources (Self-Efficacy, Job Satisfaction, Well-Being) on TMs’ Retention Decisions

Variation in Job Satisfaction and Well-Being explained a small but significant amount of variation in TMs’ propensity to continue in their current school, move to another school, or leave teaching. These decisions were most directly influenced by TMs’ sense of job satisfaction and well-being at school. Sense of Well-Being in the teaching profession only influenced retention decisions by influencing TMs’ sense of Well-Being at school. The most direct influence on TMs’ sense of Well-Being at school was their Job Satisfaction at school. As with Well-Being, Job Satisfaction in the teaching profession only influenced retention by influencing Job Satisfaction at school.

While the amount of variation in TMs’ propensity to stay or leave explained by the research is significant, it is nonetheless relatively small. Does that mean that the results of the study are only marginally useful? Our response to this question is almost the opposite: this result is seriously muted by the nature of our TM sample and the instruments used to measure both job satisfaction and well-being.

The TM sample was heavily dominated by those who were strongly committed to the ECF programme and their jobs as both teachers and mentors. Although this sample seems very likely to be a good reflection of the larger population of ECF mentors by several personal and school characteristics, it certainly is not a normally distributed sample by outcomes of personal resources and responses, as well as retention decisions. More than 93 percent responded to the survey questioning about their propensity to stay or leave their schools indicated they intended to stay in their current school. The portions of the survey used to measure TM job satisfaction and well-being were restricted to respondents’ levels of agreement (strongly agree to strongly disagree) about the extent to which they experienced those features known to be associated with positive attributions of job satisfaction and well-being. In the case of job satisfaction at school, at least 80% of surveyed TMs reported that the sense of fulfilment, gratification and enjoyment experienced through their work was associated with their unwillingness to change schools and the likelihood that they would recommend their schools to others as a good place to work. Such positively skewed distribution is in line with the results from OECD’s Teaching and Learning International Survey (TALIS) in both 2018 and 2024 (OECD, 2020, 2025) in which around 90% of surveyed teachers reported satisfaction with their jobs, indicating an overall resilient ‘norm’ of the state of teaching irrespective of the complex challenges inherent in the daily reality of the profession. There are, however, a large handful of other factors that evidence associations with job dissatisfaction: lack of being appreciated, poor work life balance, feeling underpaid, poor management and limited career growth opportunities, for example.

In the case of well-being at work, the research asked (strongly agree to strongly disagree) about features associated with a positive sense of well-being: TMs’ feelings of belonging and being able to “be themselves”, as well as believing they have colleagues who care about them and treat them with respect. However, the opposite of a positive sense of well-being–stress, burnout and feelings of alienation–are typically the result of conflicts with colleagues, unrealistic workloads, lack of control over one’s work and the like.

So, the most negative possible value measured in the case of job satisfaction and well-being was strong disagreement that they experienced positive features. But these are all still positive features, whereas there is a dark side to both of these variables–job dissatisfaction and the opposite of well-being (stress, burnout, alienation). Given the statistical reliability of the TM sample (as outlined above) and that both job satisfaction and well-being measures captured both the positive and negative ends of each, it seems highly likely that these variables, in combination, would explain a very large proportion of variation in TMs’ propensity to stay or leave. This conclusion receives some support from the results of a recent meta-analytic review indicating that: burnout and job satisfaction together explained 27% of the variance in teachers’ intentions to quit …[and] … burnout symptoms accounted for the majority of this explained variance. These findings suggest that burnout and job satisfaction are highly important in predicting teachers’ intentions to quit, but it appears that, although they are related, burnout [a negative form of well-being] may confer a greater risk than job satisfaction (Madigan & Kim, 2021).

• Research Question 2: Organizational Conditions Which Influence TMs’ Personal Resources

A significant part of our data analysis was aimed at better understanding the “paths” of influence connecting school leadership practices to TMs’ retention decisions through Job Satisfaction and Well-Being. As our framework indicates, prior research directed attention to three especially plausible connectors: Collaborative School Cultures, Professional Learning Opportunities and Workload Conditions.

Results of this study indicated that Collaborative School Culture had a moderately strong, direct influence on TMs’ job satisfaction at school and was influenced moderately strongly by Professional Growth Opportunities. Collaborative School Culture had no influence on other school conditions and was only influenced by Professional Learning Opportunities and Leadership Practices (described below). Professional Growth Opportunities, in contrast, had weak but significant levels of influence on TMs’ perceptions of their schools’ support for them, their sense of self-efficacy as mentors, and their satisfaction with being part of the teaching profession. TMs’ perception of their workload conditions was a third variable linking Leadership to Job Satisfaction and Well-Being with a moderate but direct influence on TMs’ Job Satisfaction at school.

• Research Question 3: Leadership Practices Positively Associated with Key Organizational Conditions

The conception of leadership guiding our study included four categories or domains of practices–Setting Directions, Building Relationships and Developing People, Redesigning the Organization, and Improving the Teaching Programme. Each of these domains were linked, uniquely, through Job Satisfaction and Well-Being, to TM retention.

Setting Directions. This leadership domain had very large effects on each of the three remaining leadership domains, a result also reported in earlier research (Leithwood et al., 2019). Practices in this domain, as a whole, create the purpose and focus for practices included in those other leadership domains. Setting Directions practices include giving staff a sense of overall purpose, helping staff clarify the reasons for implementing school improvement initiatives and providing useful assistance to staff in setting short-term goals for teaching and learning. This domain also encompasses leaders’ demonstrations of high expectations for staff’s work with students.

Direction setting practices reflect considerable evidence about “transformational” approaches to leadership, approaches which aim to elevate the values, aspirations and expectations of organizational members about what they should strive to accomplish with their students (Grant, 2012). These practices also conform to longstanding evidence about the motivational power of specific and measurable goals on organizational members, as well as their contribution to employee performance (Sun & Leithwood, 2015; Locke & Latham, 2013).

Building Relationships and Developing People. Leadership practices in this domain had a very strong influence on TMs’ perceptions of their Professional Growth Opportunities reflecting the importance of relationship building and consideration highlighted decades ago in the early Ohio State and Michigan leadership studies (e.g., Stogdill, 1948). Leadership practices in this domain include assessing staff development needs, providing support for the unique needs of individual staff, encouraging staff to consider new ideas for their work and modeling high levels of professional practice. These practices also include promoting leadership development among staff and creating relationships with staff that encourage authentic discussion of educational issues. Productive professional growth opportunities increase the contribution that individual staff make to their colleagues.

Redesigning the Organization. Practices in this leadership domain had a moderately strong influence on helping to build a Collaborative School Culture. Leadership practices in this domain include developing an atmosphere of caring and trust among staff, encouraging collaborative work among staff, and ensuring carefully coordinated participation in decisions about school improvement. This domain also includes fostering engagement of parents in the school’s improvement efforts and, more broadly, building support for the school’s improvement efforts in the wider community.

Improving the Teaching Programme. Practices in this domain strongly influenced TMs’ perceptions of their workload conditions. Practices giving rise to these perceptions included regularly observing classroom activities, working with teachers following classroom observation to help them improve their teaching, and strategically using data with teachers for this purpose. Practices also included as part of this leadership domain include leaders providing teaching and professional development resources needed for teachers to improve their teaching and encouragement to make creative uses of appropriate technologies to enhance teaching and learning. Buffering teachers from distractions to their pedagogical work is a practice also included in this domain.

7. Conclusions

This concluding section of the paper identifies three limitations of the study along with closely associated directions for future research. Also outlined are four guidelines for senior leaders about how to retain experienced and capable teachers as TMs using resources available to them within their own schools and bring about the expected learning and retention benefits for these experienced teachers and the school organisation.

7.1. Limitations and Implications for Future Research

One threat to the external validity of the study’s results is the unique nature of the sample of teacher mentors in this study. While these TMs are regarded by school leaders as experienced and capable teachers who are likely to be respected by their peers, the extent of the special nature of their additional formal mentor duties, along with the additional training provided to them, is unlike that of a sample of experienced teachers without that additional position and training. Future research using samples of experienced teachers without formal mentor responsibilities would test the external validity of our results.

A second limitation of the study was its use of “propensity” to stay or leave, that is, TMs’ anticipation of staying or leaving. While we cited evidence to justify the choice of this variable, it is clearly different from a direct measure of teachers staying or leaving. Future research using the framework for this study but substituting a measure of actual staying or leaving would be worthwhile; it would also overcome weaknesses in research design associated with single sources of evidence to measure all variables. As discussed earlier, although our sample was broadly representative of national figures for ECF mentors regarding several personal and school characteristics, it was heavily dominated by mentors who were strongly committed to the ECF programme and their jobs as both teachers and mentors. This raises concerns of possible non-response bias. Therefore, we recommend that future research attempt to capture a more varied sample of mentors’ perspectives.

Finally, follow-up qualitative research about the variables and relationships explored on this study would help elaborate and deepen the meaning of the findings from this quantitative study. Especially useful would be qualitative evidence about the relationships between Organizational Conditions and Teachers’ Personal Resources. For example, what is it about teachers’ workload that contributes to their sense of efficacy? What features of teachers’ engagement in collaborative cultures enhance their job satisfaction?

7.2. Implications for Practising School Leaders

The implications for school leaders highlighted below elaborate on each of the four domains of successful leadership included in the study’s framework. These implications or recommendations are designed to guide the work school leaders do to retain experienced teachers who may also be mentoring their colleagues, as well as to further refine their approaches to the broader challenges of school improvement.

7.3. Aim the School in Meaningful Directions

Results of the study demonstrate that Direction-Setting practices provide, among other things, the purposes shaping leadership practices in the other three leadership domains. A significant amount of prior research indicates that a work environment perceived to be “meaningful” is a powerful form of motivation, commitment and engagement (e.g., Chalofsky & Krishna, 2009); these are psychological states that go a long way toward ensuring high levels of job satisfaction and well-being at a school and, as a result, strong attachments to the school. At its core, meaningfulness depends on the purpose for one’s work, something leaders’ Directions Setting practices aim to provide. The importance of Direction Setting practices is difficult to overstate but easy to squander. The value of these practices is squandered when the processes used for determining directions is treated as a bureaucratic hurdle to be overcome as quickly and efficiently as possible and when official standards or targets for student achievement, as central as they are to a school’s survival, overwhelm the process. The outcome of Direction Setting practices should be a set of dynamic goals for the school that are: adopted by staff and student as their own personal goals; understood so deeply that staff can often determine themselves how best to achieve them or are sufficiently committed to figure that out themselves; and subject to continuous conversation about their value, relevance and how to achieve them.

Such a widely shared, inspirational vision or set of goals communicates optimism about future goals and builds staff commitment to the school (e.g., Moraal et al., 2024). School leaders, we suggest, spend whatever time it takes to ensure that the school has a widely shared, deeply understood, compelling set of directions that reflect the aspirations of the staff and community served by the school and that acknowledge, in some fashion, the legitimate priorities of policy makers.

School leaders might consider establishing a transparent process for engaging staff, as well as parents, in a step-by-step process for identifying key school goals and short-term priorities. This process is likely to work best when it includes systematic procedures for monitoring progress and revising school goals and priorities, as suggested by the evidence.

7.4. Structure the School to Encourage Engagement of All Stakeholders

Practices included in the Redesigning the Organization leadership domain had a significant relationship with collaborative school cultures. These practices have the potential to transform the school’s structure from clusters of individuals and groups each attempting to advance their understandings of the school’s goals the best way they know how, into a highly functioning “learning network”. Such a network is a stimulating and engaging structure to be part of, one not lightly abandoned, as would be the case with a decision to leave the school; this suggests that school leaders create formal structures and informal opportunities for parents/carers, teaching staff and some members of the wider school community to interact together and with one another about what the school should be accomplishing, what is working well, what needs to be improved, and who should do what.

Stakeholder engagement is subtly influenced by the trust stakeholders have in the school’s leadership. Such trust on the part of teachers, for example, is a function of a large handful of perceived leader characteristics, the most influential being leaders’ competence, consistency, openness, respect for staff and integrity (e.g., Leithwood, 2023b). Competence requires ongoing learning and all of these qualities can be refined and further developed by active reflection on the part of leaders along with opportunities for conversation with peers about how they enact those qualities.

7.5. Shape the Engagement of All Stakeholders to Foster the Development of the School’s Collective Intelligence

Practices included in the Building Relationships and Developing People leadership domain aim to enhance the capacities that school staff and other stakeholders need to make progress toward the school’s goals. We often assume that we know what those capacities are and how best to develop them. But if that were true, the amount of professional development experienced by most veteran educators would have resulted in perfect schools decades ago. The perspectives on learning evident in research on learning networks and social learning theory, however, are much different from the perspectives on learning driving a considerable amount of that professional development.

Social learning theory suggests that the collective intelligence of a learning network is a function of the volume and diversity of ideas generated by network members (Pentland, 2015). Diverse membership should be treated as a strength. A networks’ collective intelligence also depends on clear, shared goals for each stakeholder group, direct engagement in one another’s ideas and open and easy two-way communication (e.g., Leithwood, 2019; Silvia & McGuire, 2010). These conversations should be ongoing and nurtured by a strong sense that “we are all in this together”. One recent meta-analysis found that leadership practices fostering such interaction and engagement have an outsized influence on building teacher trust in parents and students (Sun et al., 2023), yet another contribution to teachers’ job satisfaction and well-being.

Learning networks build the capacity of individual members by exposing them to the practices, dispositions and ideas of others faced with similar tasks and responsibilities. A network is also a structure which is, under the right conditions, capable of stimulating potentially rich interactions among members resulting in new and creative ideas or practices not initially part of the repertoire of any individual network member. This knowledge creation goal of networks depends on participants’ willingness to collaborate together in the solving of some shared problem or the meeting of a shared purpose and to genuinely listen to the ideas of one’s network colleagues. School leaders should encourage these communication norms among staff participating in networks of colleagues.

7.6. Distribute Leadership for Pedagogical Coaching

Leadership practices included in Improving the Teaching Programme had a very strong influence on TMs’ perceptions of their working conditions. Leadership practices in this domain reflect longstanding conceptions of “instructional leadership”, especially early versions dominated almost exclusively with a focus on teachers’ pedagogy and a coach-like role for school leaders. While instructional models of school leadership have gradually expanded to encompass conditions in schools outside the classroom, evidence continues to support the value of this focus for school leaders. Indeed, some evidence indicates that the more school leaders possess relevant pedagogical content knowledge, the more they contribute to their school’s improvement progress (Quebec Fuentes & Jimerson, 2020; Stein & Nelson, 2003).

Even the most instructionally savvy senior leaders, however, have neither the time nor expertise to provide effective pedagogical coaching to their many teachers implementing an ambitious curriculum. The larger the school and the more ambitious the curriculum, the greater the importance of distributing instructional leadership practices to members of faculty, in addition to senior leaders; establishing mentor positions for new teachers reflects such distribution.

One of the often-overlooked challenges senior leaders face in distributing leadership for pedagogical coaching is to identify those members of the school that teachers themselves look to for pedagogical advice and coaching. These are often fellow teachers who provide practical advice about what works in very similar contexts to those seeking the advice. They are the school’s informal leaders who often have an outsized influence on the job satisfaction and well-being of their colleagues, as well as the school’s pedagogy. Identifying these staff members can sometimes be challenging for senior leaders, however (Spillane et al., 2010). School leaders should leverage the job satisfaction and well-being of their staff, as well as the existing pedagogical expertise in their schools, by identifying those members of staff that their colleagues look to for pedagogical advice and by making that advice more readily available to others who might benefit from it.

School leaders, in sum, retain high-quality staff members by helping to create productive work environments aimed at achieving ambitious, socially valued goals for all students. Given similar student populations, a high-retention school for experienced teachers is doing many things well, and a low-retention school not so much. This conclusion, we hasten to add, does not add to the work of already harried school leaders. The leadership practices that encourage staff retention are the same approaches that are needed for many other forms of significant school improvement.