Identifying the theoretical foundations of coaching as a form of ongoing professional development in early childhood education: a meta-narrative review

ABSTRACT

Despite evidence of the effectiveness of coaching as a form of ongoing professional development for educators, including in early childhood education, the field lacks an understanding of what the key components of coaching are and how and why they ‘work’. An improved theoretical basis for coaching would assist in identifying these underlying mechanisms of change. However, the field has been largely undertheorised, and an awareness of which theories have been applied in the field is lacking. This study entailed a meta-narrative review of the literature to identify which theories have informed the study of ongoing professional development in education, with a focus on coaching for early childhood educators. We identified 11 foundational papers addressing ongoing professional development in education that proposed or applied theories, and 69 papers on coaching for early childhood educators with a theoretical focus published between 2002 and 2022. Among both groups of papers, there were 47 different main theories applied or proposed. The findings indicate that coaching and other forms of ongoing professional development in education remain undertheorised and theoretical consensus is lacking. The findings suggest that further empirical testing of theories is required before any path towards building a theoretical consensus will become clearer.

KEYWORDS:

• Coaching
• meta-narrative review
• early childhood education
• mentoring
• continuing professional development
• theory

Introduction

Coaching as a form of ongoing professional development has been shown to be among the most effective forms of professional development for improving teaching practice and supporting children’s learning and development, including in early childhood education and care (ECEC) (Markussen-Brown et al. 2017, Egert et al. 2018, Kraft et al. 2018, Brunsek et al. 2020, Yang et al. 2022). We define coaching as the provision of individualised, collaborative and ongoing professional development for an educator that facilitates their growth or learning of job-related skills and practices in order to support children’s learning and development. We define ongoing professional development to include facilitated in-service professional development opportunities that provide sustained support to educators in their workplaces such as coaching, mentoring or consultation.

Egert et al. (2018) found that coaching was almost three times more effective than other forms of professional development at improving classroom quality in ECEC, including other forms of ongoing professional development. Further, Kraft et al. (2018) found that coaching can have a significant effect on both teaching quality and children’s achievement, including in prekindergarten and Brunsek et al. (2020) found that professional development for educators, including in ECEC, was more likely to have a positive impact on children’s development when a coaching component was included due to the ability of coaching to support educators to provide high-quality instruction.

Improving the quality of instruction in ECEC is crucial, given that high-quality ECEC has been shown to support positive learning and development outcomes for children in the short and long term (Fukkink and Lont 2007, Neuman and Dickinson 2010, Melhuish et al. 2015, Goldfeld et al. 2016). This understanding, coupled with evidence about the effectiveness of coaching, means that coaching has become increasingly popular in ECEC as a way to improve teaching practice and support children’s learning and development (Schachter 2015, Lochmiller 2021).

Despite evidence of the effectiveness of coaching across numerous, often small-scale, studies, research has found that effective coaching, like other forms of professional development, is difficult to replicate in other contexts or take to ‘scale’ (Fukkink and Lont 2007, Schachter 2015, Kennedy 2016, Kraft et al. 2018). For example, Kraft et al. (2018) found that the greater the number of services and settings in which coaching was implemented, the less effective it was. This could be for several reasons, such as because the fidelity of an intervention is more difficult to maintain on a wider scale or because the original programme was not effective or well-suited to the participants or their context (Borko 2004, Postholm 2012, Hill et al. 2013, Goldsmith et al. 2014, Mendive et al. 2015, Schachter et al. 2018, Markowitz and Seyarto 2023). However, it could also be because the studies of coaching on a wider scale lacked rigour or used unsuitable methods for measuring outcomes (Hill et al. 2013). Nonetheless, the lack of success in harnessing the potential of coaching at a broader level indicates that the field lacks understanding of what the key components of coaching are and how and why they ‘work’ and, as such, which aspects to emphasise when taking coaching to scale. Given the promise of coaching, this is a problem worth investigating further.

It is difficult to reach an understanding of how and why coaching ‘works’ when coaching varies so greatly. Previous studies have identified which strategies and approaches have been commonly used within coaching, but which combination of these strategies are used, and in what manner, differs greatly (Artman-Meeker et al. 2015, Schachter 2015, Elek and Page 2019). In addition, despite indications from the empirical literature that the individualised, ongoing, responsive and in situ nature of coaching might be critical to its success (Elek and Page 2019), there is certainly no consensus on this issue. This lack of consensus is shared by the field of professional development more broadly. For instance, Desimone (2009) contended that a consensus existed about the critical features of professional development, many of which she noted were associated with coaching (Desimone and Pak 2017). However, others have found that research has not been able to predict learning outcomes based on these features (Opfer and Pedder 2011, Sims and Fletcher-Wood 2021). This inability to understand what is driving the success of coaching and other forms of professional development leaves decision-makers faced with a bewildering array of professional development and coaching approaches to choose from.

An improved and well-recognised theoretical basis for understanding coaching would assist us to identify the mechanisms of change within coaching. Mechanisms refer to the ways in which components or features of a programme bring about change, on their own, together or in a series of steps (Pawson and Tilley 2004).

There has been over a decade of calls for an enhanced theoretical understanding of professional development in education (Sheridan et al. 2009, Dickinson et al. 2011, Hamre et al. 2012, Snyder et al. 2012). However, several authors have observed that professional development in education has remained undertheorised (Kennedy 2014, Schachter 2015, Boylan et al. 2018), including coaching specifically (Yang et al. 2022). This limits our ability to deeply understand educators’ learning, with implications for policy and practice in education (Kennedy 2014). When theories or models have been proposed, they have been found to be diverse and contradictory and to undertheorise the change processes which take place (Boylan et al. 2018).

A sound theoretical foundation could be lacking because the body of literature has been dominated by empirical studies (Borko 2004, Kraft et al. 2018), perhaps driven by accountability frameworks in education (Opfer and Pedder 2011). Several authors have noted that the theoretical perspectives that underpin research about professional development in education are not often made explicit (Isner et al. 2011, Opfer and Pedder 2011, Hamre et al. 2012), particularly in empirical literature. For example, in Schachter’s (2015) review of the empirical literature on professional development interventions in ECEC, most of which included coaching, only 15% of studies drew on references to explain their conceptualisation of professional development. Despite the relatively large body of empirical literature, however, not all theories have arisen from empirical findings, and few theories have been tested empirically (Sheridan et al. 2009, Schachter 2015).

Although theories are not absent from the body of literature, there is no clear picture of the range of theories used. For example, while Schachter (2015) identified some studies that referred to theory to inform their understanding of professional development, she did not note which theories they referred to. As such, we have a limited basis for understanding the theoretical lenses through which researchers in the field of coaching in education and in ECEC have interpreted their findings about what might be driving the success of coaching. This deficiency limits the ability of researchers to identify what might be the mechanisms of change at play in coaching and other forms of ongoing professional development and thus where further efforts to analyse such programmes should be directed.

The current study

To address this gap, we undertook a review of the literature to identify which theories have informed the study of coaching and other forms of ongoing professional development in education, with a focus on coaching in ECEC. The primary research question was: Which theories have been used to understand coaching and other forms of ongoing professional development for educators in early childhood education and care?

We defined a theory as ‘a set of concepts and the proposed relationship between them’ (Maxwell 2005, p. 42) which is designed to explain some aspect of how the world ‘works’ and can include explanations presented as conceptual models or frameworks, for example (Sandelowski 1993, Maxwell 2005).

To answer our research question, we first sought to identify which theories were employed or proposed within the foundational literature on ongoing professional development in education more broadly. We did so to understand the wider body of research in which studies of coaching sit. Our definition of foundational literature was informed by Greenhalgh et al. (2005) which refers to papers that are: e.g.

(1) part of a recognised research tradition that draws on an existing body of knowledge and seeks to advance it;(2) makes an original and scholarly contribution; and,(3) has subsequently been cited as influential by competent researchers. (Greenhalgh et al. 2005, p. 421)

We went on to explore which theories were applied within the literature on coaching in ECEC specifically, with ECEC referring to the provision of services to support the learning and development of children aged from birth to school age. As such, supporting research questions included:

1. Which theories are prevalent in foundational literature on coaching and other ongoing professional development for educators?

2. Which theories have been commonly applied to studies of coaching and other ongoing forms of professional development for educators in early childhood education and care?

Our definition of coaching, as stated at the outset of this article, was deliberately broad as there is no agreed definition of coaching, and we wished to cast our net widely for the purposes of this review. For the same reason, we included coaching, mentoring and other similar forms of ongoing professional development within the scope of this stage of our study. We did so because while we acknowledge that other forms of ongoing professional development are seen by some as distinct from coaching (e.g. Fletcher 2012, Sherbourne 2016), they are often similarly defined (e.g. Lofthouse et al. 2010, Edwards-Groves 2014). For example, many authors have noted there is significant overlap between how coaching and mentoring are defined and enacted, such as being collaborative and responsive, led by a more experienced other, involving reflection, and focussed on improving educators’ knowledge or practice (e.g. Crafton and Kaiser 2011, Fletcher 2012, van Nieuwerburgh 2012, Ambrosetti et al. 2014, Rogers et al. 2020, Wong et al. 2020, Yang et al. 2022).

Materials and methods

We used an approach known as a ‘meta-narrative review’ to conduct our review of the literature (Greenhalgh et al. 2005, Wong et al. 2013). Meta-narratives can be understood as research traditions or conceptualisations within a field which have unfolded over time, in which the subject is viewed through a particular ‘lens’ or paradigm by a particular group of researchers (Greenhalgh et al. 2005, Wong et al. 2013).

A meta-narrative review is a relatively new, ‘semi-systematic’ method for undertaking a literature review (Wong et al. 2013, Snyder 2019). It is designed to map the range of ways a research area has been approached, conceptualised or theorised in order to identify the mechanisms by which programmes achieve their goals and, ultimately, to inform policy or create a research agenda (Greenhalgh et al. 2005, Wong et al. 2013, Snyder 2019). Meta-narrative reviews share some features in common with other semi-systematic review methods, such as integrative reviews. Both these review types can be described as configurative reviews, designed to explore and interpret research in a field and generate new ways of understanding, and can be contrasted with aggregative reviews such as systematic reviews and meta-analyses (Gough et al. 2012, Sandelowski et al. 2012). However, a meta-narrative review tends to take a broader view of a field than an integrative review and seeks to map the ways others have approached or theorised an issue rather than to build a new theoretical model (Snyder 2019). A meta-narrative review is suited to fields in which definitions are ‘fuzzy’, in which different research and practice traditions overlap and which have been conceptualised differently by different researchers over time (Greenhalgh et al. 2005). The method was thus suited to this review, as approaches to coaching and other forms of ongoing professional development in ECEC vary widely and are underpinned by a variety of aims, principles and definitions.

After first loosely ‘mapping’ the research terrain to identify the key research traditions, a meta-narrative review involves identifying foundational articles within each tradition. This is followed by a series of overlapping steps in which the literature is systematically searched and screened to identify a body of literature sufficient to enable an analysis of the conceptual understandings within each research tradition, after which the literature is mapped and synthesised. Finally, recommendations are made regarding policy implications or a research agenda (Greenhalgh et al. 2005).

A meta-narrative review should be adapted to suit the research agenda and context (Greenhalgh et al. 2005). The phases of our review are captured in Figure 1. Our adaptations primarily entailed adjusting the planning phase to focus more on identifying the research ‘problem’, embarking on the appraisal stage prior to mapping, and drawing from Miles et al. (2014)’s method of using matrices for analysis and mapping. This meant that mapping was followed immediately by the synthesis phase. Our process was a reflexive, iterative one in which, for example, mapping processes identified additional variables of interest, entailing the revisiting of earlier steps.

Figure 1. Phases of the review

Procedures

Problem identification

The problem identification phase entailed defining the objectives and focus of the review and defining the variables of interest to guide the inclusion and exclusion criteria. After an iterative process, and for pragmatic reasons, we determined that the ultimate focus of our review would be on mapping the different ways coaching has been theorised within the field of education, with a focus on the field of early childhood education.

To achieve this, we set out to locate relevant literature in two categories, from broader to more specific: foundational papers on ongoing professional development in education (Group A) followed by papers on coaching and similar forms of ongoing professional development in ECEC (Group B). Papers in both groups were required to present, propose or be explicitly informed by a theory. Unlike Greenhalgh et al. (2005), we did not limit studies in the latter group to empirical studies. This allowed for the inclusion of literature within a range of traditions and research methodologies, from theoretical to empirical and both qualitative and quantitative (Whittemore and Knafl 2005).

Search and appraisal phases

To define the search strategy for our review, we identified the method of intervention (coaching and other forms of ongoing professional development) as the primary variable of interest. The population was limited to teachers or educators. Slightly different search and appraisal processes were required for Group A and Group B, as follows.

Inclusion and exclusion criteria

The inclusion criteria for both groups are outlined in Table 1. All papers were limited to papers that pertained to coaching or other forms of ongoing professional development for teachers and educators and to papers published in English. Papers were not required to be peer reviewed and could be in the form of journal articles, reports, book chapters or dissertations, for example. In line with our definition of theory, papers were included only if they identified or presented a theory to explain how and why coaching or professional development ‘works’. The theory was required to include not just a description of the features or processes of coaching but also an explanation of the relationship between these features or processes. This could include discussions about the mechanisms of change which take place when coaching is enacted, such as proposed causal pathways or drivers of change, and could take the form of a visual representation of components and the relationship between them.

Table 1. Inclusion criteria

Inclusion criteria for both Groups	Additional inclusion criterion for Group A
- Concerns coaching, mentoring or other forms of ongoing professional development - In an educational context - Provided for teachers or educators, including pre-service - Explicitly informed by, presents, proposes or applies a theory - Published in English	- Highly cited (average >100 times per annum)
	Additional inclusion criteria for Group B
	- Set in ECEC - Published between 2002 and 2022

For Group A, we considered papers to be foundational if they were cited an average of 100 times per year or more and indexed in academic databases. Papers in Group B were required to pertain to coaching or similar forms of ongoing professional development in ECEC, such as mentoring, and for practical reasons were limited to papers published between 2002 and 2022.

Search and appraisal phases

The three steps for searching the literature in meta-narrative reviews described by Greenhalgh et al. (2005) become progressively more systematic, with each step informing the next. The steps can be summarised as: (a) initially search by intuition, browsing, etc. to map the diversity of perspectives and approaches; (b) search for foundational papers by examining references of references; and (c) a more comprehensive search, including of electronic databases (Greenhalgh et al. 2005). Our search strategies for each group involved similar steps, as outlined below. We appraised the relevance of each study to the research question (Greenhalgh et al. 2005) via the application of the inclusion criteria during the search and screening processes.

To identify foundational papers (Group A), we undertook the following steps: (1) intuitive and manual ‘browsing’; (2) database searches; and (3) an ‘ancestral’ search for references.

To undertake the first step, we drew on our personal reference databases and knowledge of the field of research. We sought to include authors with a range of perspectives and include grey literature. For authors whose work was well-known to us, we selected the most highly cited of their works for consideration. We identified 19 papers for screening using this method.

We then undertook a search of academic databases which enabled sorting by the number of citations: Scopus, Web of Science, Dimensions and Google Scholar.¹ We selected deliberately broad search terms to allow for the identification of papers using a range of methods, e.g. empirical or theoretical papers. We used search terms such as: (coach* OR mentor*) AND (educat* OR teach* OR school*). No date range was set, and publications were limited to those published in English. Specific search terms and delimiters were tailored to each database and research tradition as detailed in the supplementary material. In line with the iterative search method supported by this method, after reviewing key terms of relevant studies, we decided to repeat our searches of the above databases using terms such as ‘continuing professional development’ AND (educat* OR teach* OR school*) and the same limiters and adjustments for each database.² After completing each search, we sorted the results by number of citations and selected the most highly cited records from each database for screening.³ We identified 775 studies for screening using these methods.

After identifying papers using these two methods, we screened all papers using Covidence software. We first screened the titles and abstracts with reference to the generic inclusion criteria. Then, we reviewed the full text of each remaining study with reference to the generic inclusion criteria and the criterion of being highly cited. Google Scholar was used to identify the number of citations.

We then undertook an ‘ancestral’ search to identify papers commonly cited in the papers included after the initial screening of full text. We defined commonly cited in this instance to be papers cited by more than one foundational study we had included so far. We identified papers cited by more than one of these papers using two methods. First, we used the artificial intelligence (AI) software ResearchRabbit.ai and its function to identify ‘earlier works’. Second, we located references within the included papers using the Scopus database and ConnectedPapers, another AI tool. We then pooled the ancestral papers sourced using both methods and identified duplicate records, thereby identifying any paper cited by more than one of the original foundational papers. This process identified 126 papers. We imported each of these into Covidence and subjected these to the same screening as all other papers.

The final number of papers included in Group A was 11. The final result of the screening process for Group A is depicted in Figure 2, and the source of each included study is included in Table 2.

Figure 2. PRISMA for Group A search and screen.

Table 2. Characteristics of group A papers

Author/s	Date	Source	Peer reviewed	Citations per year (average)	Country	Focus	Educational level^	Article type	Theory/theories applied
Borko	2004	Ancestral	Yes	508	USA	PD general	School	Conceptual	Situated learning theory
Clarke and Hollingsworth	2002	Hand search	Yes	150	Australia	PD general	School	Conceptual; empirical	Study-designed (influenced by situated learning theory)
Darling-Hammond et al.	2017	Hand search	Yes	503	USA	PD general	School (secondary)	Secondary	Adult learning theory
Desimone	2009	Ancestral	Yes	419	USA	PD general	School	Conceptual	Study-designed
Guskey	2002	Hand search; ancestral	Yes	245	USA	PD general	School	Conceptual	Study-designed (influenced by James-Lange theory)
Joyce and Showers	2002	Hand search	No	326	USA	Coaching; PD general	School	Conceptual	Organisational and change theorists; conceptual systems theory; self-concept theory
Kennedy	2016	Database	Yes	197	USA	PD general	School (empirical and secondary)	Secondary	Study-designed
Kraft et al.	2018	Hand search; database	Yes	198	USA	Coaching	School (primary and secondary); ECEC	Secondary	Study-designed
Opfer and Pedder	2011	Ancestral	Yes	175	UK	PD general	School (primary and secondary)	Conceptual; secondary	Complexity theory
Timperley et al.	2007	Hand search	No	149	New Zealand	PD general	School	Secondary	Study-designed
Yoon et al.	2007	Hand search; ancestral	Yes	104	USA	PD general	School (primary)	Secondary	Study-designed

^ Level of schooling specified if stated; PD = professional development

To identify papers in Group B, we repeated similar steps. First, we used our existing reference library and prior knowledge to identify potential papers for inclusion. We identified 35 studies for screening in this way.

Then, we searched the electronic databases: Scopus, Web of Science, Dimensions, Academic Search Complete, ERIC (EBSCO), PyscINFO and Proquest Education. We used a wider selection of databases from those we used for Group A as we were not concerned with being able to sort records by the number of citations. We used similar search terms as for Group A but with the addition of search terms to limit the search to papers in ECEC and papers with a conceptual focus, such as (theor* OR framework* OR model*) AND (kindergarten* OR pre-kindergarten* OR preschool* OR childcare OR ‘early childhood education’). The search was limited to papers from 2002 to 2022. Search terms and combinations and delimiters were tailored to each database and are detailed in the supplementary material. We identified 1496 studies using this method.

We then screened the titles and abstracts of each with reference to the inclusion criteria, followed by the full text. During this stage, 13 additional records were identified for screening from reference lists. After screening was complete, 69 studies were included in Group B.

The results of the screening process can be found in Figure 3, and the source of each included paper is noted in Table 3.

Figure 3. PRISMA for Group B search and screen.

Table 3. Characteristics of group B papers

Author/s^	Date	Source	Peer reviewed	Country	Focus	Article type	Main theory/theories applied
Amireh	2016	Database	Dissertation	USA	Coaching	Empirical	Adult learning
Arteaga et al.	2019	Database	Yes	USA	Coaching	Empirical	Study-designed
Balduzzi and Lazzari	2015	Database	Yes	Italy	Mentoring	Conceptual; empirical	Reflective practice
Bawani and Mphahlele	2021	Database	Yes	Other: South Africa	Mentoring	Empirical	Social cognitive; communities of practice
Beaunae	2009	Database	Dissertation	USA	Mentoring	Empirical	Study-designed
Ben-Harush and Orland-Barak	2019	Hand search; database	Yes	Other: Israel	Coaching	Empirical	Cultural historical activity theory
Byrne and Norman	2020	Hand search	Yes	UK	Other: continuing PD	Empirical	Adult learning; transformative learning
Caroll-Lind et al.	2016	Database	Yes	New Zealand	Coaching; Mentoring	Empirical	Cultural historical activity theory
Cheng	2014	Database	No	Taiwan	Mentoring; other: community of practice	Empirical	Situated learning
Clough	2015	Hand search	Dissertation	USA	Coaching	Empirical	Study-designed
Cramer	2019	Database	Yes	USA	Coaching	Empirical	Other: social network theory
Crawford et al.	2017	Database	Yes	USA	Coaching	Empirical	Adult learning
Dunst et al.	2013	Database	Yes	USA	Coaching	Conceptual	Implementation science
Filliettaz	2014	Database	No	Other: Switzerland	Mentoring	Empirical	Other: interactional participation
Garte and Kronen	2020	Database	Yes	USA	Mentoring	Empirical	Cultural historical activity theory
Hadley et al.	2015	Database	Yes	Australia	Other: continuous professional learning	Conceptual	Other: contextual theory of leadership
Harris and Kaya	2015	Other: iterative	No	New Zealand	Mentoring	Conceptual	Other: feminist relational-cultural theory
Hine	2019	Database	No	USA	Coaching	Conceptual	Study-designed; transformative learning
Hobor	2014	Hand search; database	Dissertation	USA	Coaching	Empirical	Socio-cultural; transformative learning
Holland	2021	Database	Dissertation	USA	Coaching	Empirical	Implementation science
Howard	2014	Database	No	USA	Coaching	Empirical	Study-designed
Jamil	2012	Database	No	USA	Coaching	Empirical	Self-efficacy
Jenkins et al.	2015	Other: iterative	No	New Zealand	Mentoring	Conceptual	Study-designed
Kraft, Blazar, Hogan	2018	Hand search; database	Yes	USA	Coaching	Secondary	Study-designed
LaParo and King	2019	Database	No	USA	Coaching; PD general	Conceptual	Ecological
LaParo and Siskind	2022	Database	Yes	USA	Mentoring	Empirical	Ecological; other: systems theory
Lazzari et al.	2013	Database	Yes	Italy	Other: continuing PD with pedagogic guidance	Secondary; conceptual	Other: Reggio Emilia
Lee et al.	2019	Other: iterative	No	Other: South Korea	Coaching	Conceptual	Other: motivational interviewing
Lindberg	2010	Database	No	Other: Finland	Mentoring; PD general	Conceptual	Reflective practice
LoCasale-Crouch, et al.	2016	Database	Yes	USA	Coaching	Empirical	Implementation science
Lockhart-Findling	2016	Database	Dissertation	USA	Mentoring	Empirical	Guskey’s model
Mann et al.	2015	Database	Yes	USA	Coaching	Empirical	Social cognitive; self-determination
Martin	2020	Other	No	USA	Mentoring	Empirical	Transformative learning
Meadan	2017	Database	Yes	USA	Coaching	Secondary; conceptual	Study-designed
Miguel	2021	Database	No	USA	Coaching	Conceptual	Adult learning; Guskey’s model; self-efficacy
Morris et al.	2014	Database	No	USA	Coaching	Empirical	Study-designed
Navarro	2017	Database	Dissertation	USA	Mentoring	Empirical	Study-designed
Nolan and Guo	2019	Hand search	No	Australia	Mentoring	Empirical	Transformative learning
Nolan and Molla	2018	Hand search; database	Yes	Australia	Mentoring	Empirical	Habitus and reflexivity; transformative learning
Nolan and Molla	2018	Database	Yes	Australia	Mentoring	Empirical	Habitus and social practice
Nolan and Molla	2018	Database	Yes	Australia	Mentoring	Conceptual; empirical	Habitus and social practice
Peterson	2012	Database	Yes	USA	Mentoring	Empirical	Transtheoretical model
Philippou et al.	2015	Database	Yes	Other: Cyprus	Other: continuing PD	Empirical	Communities of practice; other: action research
Pianta	2021	Database	Yes	USA	Coaching	Empirical	Ecological
Quinones et al.	2020	Database	Yes	Australia	Mentoring	Empirical	Cultural historical activity theory
Ramey and Ramey	2008	Other: iterative	No	USA	Coaching	Conceptual	Study-designed
Rivera	2020	Database	Dissertation	USA	Mentoring	Empirical	Other: theory of mentor preparation
Roberts et al.	2014	Database	Yes	USA	Coaching	Empirical	Ecological
Rogers et al.	2020	Hand search	Yes	UK	PD general; coaching	Secondary	Study-designed
Rouse	2015	Other	No	Australia	Mentoring	Conceptual	Reflective practice
Royster	2020	Database	Dissertation	USA	Coaching	Empirical	Other: curricular noticing framework
Seplocha	2018	Database	No	USA	Coaching	Conceptual	Adult learning
Shannon et al.	2015	Hand search; database	Yes	USA	Coaching	Empirical	Study-designed
Shearsby	2015	Other: iterative	No	New Zealand	Mentoring	Conceptual	Socio-cultural
Sherbourne	2016	Hand search; Database	Dissertation	USA	Coaching	Empirical	Socio-cultural
Shiu	2002	Database	Other: Conference presentation	Taiwan	Mentoring	Empirical	Socio-cultural
Sibley et al.	2010	Other: iterative	No	USA	Mentoring	Conceptual; empirical	Transtheoretical model
Solarski	2021	Database	Dissertation	USA	Coaching	Empirical	Other: whole teacher approach
Steed and Durand	2013	Database	Yes	USA	Coaching	Empirical	Self-efficacy
Thapa and Akpovo	2020	Database	Yes	Other: USA/Nepal	Mentoring	Empirical	Other: framework of cultural humility
von Suchodoletz et al.	2018	Database	Yes	Other: UAE/USA	Coaching	Empirical	Self-efficacy
Waaland	2016	Database	Yes	Norway	Mentoring; other: educational leadership	Empirical	Autonomy
Waaland	2021	Database	Yes	Norway	Mentoring	Empirical	Job characteristics theory
Waaland	2014	Database	Yes	Norway	Mentoring	Empirical	Job characteristics theory
Walsh et al.	2021	Database	Yes	USA	Coaching	Empirical	Transformative learning
Weiland et al.	2018	Hand search; database	Yes	USA	Coaching; other: PD general	Secondary; conceptual	Adult learning
Whatman	2016	Database	No	Australia	Mentoring	Empirical	Cultural historical activity theory
Wyatt et al.	2017	Database	Yes	USA	Coaching	Empirical	Socio-cultural; other: bridging multiple worlds framework
Yang et al.	2022	Database	Yes	Other: China (Hong Kong)	Coaching	Secondary	Situated learning

^ Full references are provided in the supplementary material. PD = professional development

Mapping, synthesis and recommendations phases

We used matrices to support mapping processes, as described by Miles et al. (2014). Matrices are generally tables with labelled rows and columns connecting concepts and issues of interest. They are not simply tabulations of data but an analytic process supporting synthesis and interpretation, including the identification of patterns or themes (Miles et al. 2014).

The first step in mapping involved identifying which theories were most often mentioned in the included studies. These theoretical groupings then formed the units of analysis for subsequent mapping and synthesis (Greenhalgh et al. 2005). Additional mapping supported the elicitation of patterns. For the purposes of the discussion in this paper, we looked for patterns by mapping papers and the theories they used and the features of the papers, such as date, type of paper, country of lead author, sector (ECEC or other) and focus of the paper (e.g. coaching or mentoring). We highlighted relevant sections in papers using qualitative data analysis software and summarised the highlighted sections to help populate the cells in some matrices. Mapping in a meta-narrative review can take place alongside several steps, informs each subsequent step, and can entail returning to previous steps (Greenhalgh et al. 2005). As such, if gaps were identified or questions raised, we returned to earlier phases as required and used additional data to support further mapping.

In the synthesis phase, we reviewed all the data and patterns elicited during the mapping stage and presented the key findings to answer the research question and sub-questions. In the discussion section, we report on this phase by discussing the patterns we identified and the strengths and limitations of the findings, and consider commonalities, trends, contradictions and gaps. Finally, we draw out implications for practice and research arising from the ‘recommendations’ phase of the meta-narrative review.

Results

Group A: foundational studies

Of the 11 documents we identified as foundational, there were nine peer-reviewed papers and two items which were not peer reviewed: one report and one book, both of which we hereafter refer to as ‘papers’. As depicted in Table 2, papers in Group A were published between 2002 and 2018. Eight were from authors in the USA, one from Australia, one from New Zealand and one from the UK. These included six papers which were conceptual or opinion papers and six which were secondary studies (literature reviews or meta-analyses), with one of these studies having both a conceptual and a review focus (Opfer and Pedder 2011). The authors of one paper included empirical data to support their theory (Clarke and Hollingsworth 2002). Two papers focussed on coaching and nine on ongoing professional development more broadly. All papers except one were focussed on professional development for school teachers only; one included coaching for teachers and early childhood educators in their meta-analysis (Kraft et al. 2018). Papers in Group A were all indexed in academic databases and were cited between 104 and 508 times on average per year (between 702 and 6608 times in total).

From a review of papers in Group A, we inferred a range of possible reasons why they were highly cited. Several may have been cited because they presented evidence for the effectiveness of continuing professional development or coaching specifically (Joyce and Showers 2002, Darling-Hammond et al. 2017, Kraft et al. 2018). Others may have been cited because they presented evidence for, or justification of, features common to ongoing professional development. The features emphasised included: the importance of opportunities for teachers to practise new skills and observe their effects (Clarke and Hollingsworth 2002, Guskey 2002, Joyce and Showers 2002, Borko 2004, Timperley et al. 2007, Kennedy 2016, Darling-Hammond et al. 2017, Kraft et al. 2018); the effectiveness of professional development with a longer duration or intensity (Joyce and Showers 2002, Borko 2004, Timperley et al. 2007, Yoon et al. 2007, Darling-Hammond et al. 2017); and the importance of responding to individual teacher’s motivations, characteristics or prior knowledge (Borko 2004, Desimone 2009, Opfer and Pedder 2011, Kennedy 2016).

A diversity of theories informed papers in Group A. Four papers referred to pre-existing theories as informing their work. Borko (2004) applied situated learning theory in her paper, citing Lave and Wenger (1991 cited in Borko 2004) and Greeno (2003 cited in Borko 2004). Darling-Hammond et al. (2017) mentioned adult learning theory as underpinning their work, citing Trotter (2006 cited in Darling-Hammond et al. 2017). Opfer and Pedder (2011) referred to complexity theory (e.g. Davis and Sumara 2006 cited in Opfer and Pedder 2011) and applied this theory to their analysis of ongoing professional development. Joyce and Showers (2002) described their work as influenced mainly by organisational change theory (e.g. Fullan 1991, 2001, Huberman 1992, both cited in Joyce and Showers 2002). Joyce and Showers (2002) also noted the influence on their work of conceptual systems theory (Harvey et al. 1961, Hunt 1971, both cited in Joyce and Showers 2002) and self-concept theory (Maslow 1962 cited in Joyce and Showers 2002).

The remaining seven papers presented their own novel, study-designed theory in the form of a theory of change, a theory of action or similar. Among authors of study-designed theories, the authors of two papers referred to extant theories which had influenced their theory. Guskey (2002) noted that his model of professional learning was influenced by psychological theories such as the James-Lange theory, as proposed by James et al. (1890 cited by Guskey 2002). Clarke and Hollingsworth (2002) identified situated learning theory as underpinning their theory (Greeno 1997, Cobb and Bowers 1999, Clarke 2001, all cited in Clarke and Hollingsworth 2002).

Study-designed theories presented in Group A were generally presented in diagrammatic form, often with key features connected with arrows depicting the relationships between them. The relationships between features in these theories fell into three main groups: linear relationships (e.g. Guskey 2002, Kraft et al. 2018), in which the features are seen to connect in a unidirectional causal pathway (Boylan et al. 2018), non-recursive relationships, in which the features are seen to influence one another without a clear unidirectional sequence (e.g. Yoon et al. 2007, Desimone 2009) and cyclical relationships, in which multiple ‘entry points’ and causal pathways are seen as possible and learning is seen as continuous (e.g. Clarke and Hollingsworth 2002).

Across all papers in Group A, the theories focus on different aspects of professional development, although often in combination. For example, several theories emphasise the subject matter of the professional development (Joyce and Showers 2002, Timperley et al. 2007, Desimone 2009, Darling-Hammond et al. 2017, Kraft et al. 2018) and the processes of the professional development (Joyce and Showers 2002, Desimone 2009, Darling-Hammond et al. 2017). Others focus on the importance of the school or organisational context (Joyce and Showers 2002, Borko 2004, Timperley et al. 2007, Opfer and Pedder 2011) or individual teacher characteristics and motivation (Clarke and Hollingsworth 2002, Borko 2004, Opfer and Pedder 2011, Kennedy 2016), and so on. Even theories with apparent similarities differ in important ways. For example, Guskey (2002) and Desimone (2009) both proposed linear theories with similar components, but proposed different chains of causality, positioning the components in a different order.

Group B: studies of coaching and similar forms of ongoing professional development in ECEC

We identified 69 documents for inclusion in Group B (Table 3). Of these, 37 were peer-reviewed articles, 20 were not peer reviewed (including books or book chapters), 11 were dissertations and one was a conference paper. All are referred to hereafter as ‘papers’.

More studies in Group B were published in the most recent 10 years (2013–2022) (n = 61) than the first 11 years (2002–2012) (n = 8), with almost half published in the most recent five years (2018–2022) (n = 30). More than half of the papers in Group B were led by authors based in the USA (n = 39). Of the others: eight were from Australia; four from New Zealand; three from Norway; two each from the UK, Italy and Taiwan; and nine from other countries. Thirty-six included papers that had coaching as a focus, 26 focussed on mentoring and four had another focus (e.g. ongoing professional development). Seven focussed on a combination of two or more of these. A focus on coaching, rather than on other forms of ongoing professional development, became more common over time. Forty-seven included papers were purely empirical papers, 13 were purely conceptual papers and three were papers focussed only on secondary data such as literature reviews. Six were a combination of two or more of these types.

Papers were included in Group B if a theory was a focus of their description or analysis. As in Group A, the theories applied varied greatly among papers in Group B. More than one-third of papers (n = 26) mentioned or applied theories not mentioned or applied by any other paper in Group B. This included 13 papers which presented their own novel, study-designed theory. Adult learning theories were most often mentioned or applied as the main theories (n = 16). The most common of the adult learning theories was Mezirow’s transformative learning theory (Mezirow 2003) (n = 7). Adult learning theories were followed in prevalence by Bandura’s social-cognitive theory or theory of self-efficacy (n = 6), cultural historical activity theory and socio-cultural theories or social constructivism as described by Vygotsky (1978) (both n = 5).

Papers with any empirical component were the majority of papers in Group B (n = 50). These papers applied the widest range of theories (n = 30), with adult learning theories (including transformative learning theory), cultural historical activity theory, social cognitive theory and socio-cultural theory being the only theories applied four times or more. Seven study-designed theories were proposed among this group of papers. Papers with any conceptual aspect in Group B (n = 19) applied 17 theories in total. These were most often adult learning theories or study-designed theories. Transformative learning theory was not one of the adult learning theories used by conceptual papers. Literature reviews or secondary studies in Group B were few (n = 6), and half of these papers applied study-designed theories as their main theories. There was overlap between empirical, conceptual and review papers, with some papers having both a conceptual and empirical aspect, for example.

Papers with any focus on coaching mentioned a different range of theories than those with a focus on mentoring, despite an overlap between these two groups of papers. Papers in Group B on coaching (n = 36) most often applied adult learning theories (including transformative learning theory) and social cognitive theory, these being the only theories applied more than four times. Ten study-designed theories were proposed. Papers with any focus on mentoring (n = 30) applied a still wider range of theories, with no theory mentioned more than four times each; the only theory mentioned four times among papers on mentoring was cultural historical activity theory.

Both groups

Forty-seven different theories were applied as the main theory by the combined total of 80 papers in Group A and Group B, when transformative learning theory is included as an adult learning theory. When papers from both groups are considered together (Table 4), we see again that papers most often applied theories that had not been used by any other included papers: 34 theories were applied as the main theory by only one paper each. This included six study-designed theories proposed by papers in Group A, meaning that these six study-designed theories within foundational papers, while highly cited in the wider body of literature, were not generally applied as the main theories in Group B. The only exception was Guskey’s (2002) model, which was proposed in Group A and used as the main theory by two papers in Group B (Lockhart-Findling 2016, Miguel 2021). Clarke and Hollingsworth’s (2002) and Desimone’s (2009) models were both mentioned once each in Group B, but not as main theories. Other theories presented in foundational papers used as the main theory by papers in Group B were adult learning theory (n = 9 in Group B) and situated learning theory (n = 2 in Group B).

Table 4. Prevalence of theories in Group A and Group B.

Theory	Group A	Group B	All studies
Adult learning theories (total)	1	16	17
Transformative learning theory	0	7	7
Other adult learning theories	1	9	10
Social cognitive theory/self-efficacy	0	6	6
Cultural historical activity theory	0	5	5
Socio-cultural	0	5	5
Ecological	0	4	4
Habitus/reflexivity/social practice	0	3	3
Implementation science	0	3	3
Guskey	1	2	3
Situated/situative learning theory	1	2	3
Transtheoretical model of behaviour change	0	2	2
Communities of practice	0	2	2
Job characteristics theory	0	2	2
Self-determination theory/autonomy	0	2	2
Complexity theory	1	0	1
Organisational and change theories; conceptual systems theory; self-concept theory^	1	0	1
Clarke and Hollingsworth	1	0	1
Desimone	1	0	1
Kennedy	1	0	1
Kraft et al.	1	0	1
Timperley et al.	1	0	1
Yoon et al.	1	0	1
Other theories in Group B (total)^^	N/A	26	26
Pre-existing theories applied once each in Group B		13	13
Study-designed theories applied once each in Group B		13	13

^ This group of theories is listed together as these were the main theories cited in the paper by Joyce and Showers (2002) in Group A. ^^ This includes theories used in only one paper each, excluding theories presented in foundational studies, which appear in separate rows.

When the papers in both groups were considered together, just 13 theories were mentioned by more than one included paper. Of these, adult learning theories were the most common, including transformative learning theory (applied by more than one-fifth of the papers; n = 17). No other theories were applied by more than six papers each.

Discussion

This study aimed to identify theories that have informed the study of coaching in ECEC. It did so by first identifying which theories have been used in foundational literature on ongoing professional development in education and then identifying which theories have been commonly applied within the literature on coaching and other similar forms of ongoing professional development in ECEC specifically. In doing so, we aimed to provide information to guide the design of coaching programmes, to support reflection and refinement of coaching practice and to frame areas for further research about the mechanisms of change which drive the effectiveness of coaching programmes.

Despite numerous calls to understand the processes and mechanisms that underpin the success of professional development and coaching (Sheridan et al. 2009, Dickinson et al. 2011, Hamre et al. 2012, Snyder et al. 2012, Schachter 2015), we found a field that remains undertheorised. In Group A, in which we sought to identify foundational studies on ongoing professional development in education, we identified 132 out of 678 papers (19.4%), which warranted full-text review by virtue of having a focus on theory. Only 11 of these we considered foundational by virtue of being also highly cited (1.6%). In Group B, from among the 945 studies screened, we identified only 69 papers (7.3%) on coaching or similar forms of ongoing professional development in ECEC in the last 20 years that were explicitly informed by, applied or proposed a theory, and these were from papers identified using search strategies intended to find papers with a theoretical focus. Our study also found that research on coaching and similar forms of ongoing professional development in ECEC remains largely empirical, with 50 of the 69 papers in Group B having an empirical component (72.5%). Our findings thus echo Schachter’s (2015) finding that empirical studies do not tend to refer to the literature to inform their conceptualisation of professional development.

We did not identify a theoretical consensus in the field. While Desimone (2009) noted there is no clear guidance about what it would take to claim that consensus has been reached in research, our findings make it plain that we are not near that point. Just as coaching approaches are diverse, we found a field in which theories were many and varied – we found 47 theories applied to professional development and coaching from among 80 papers. Further, the theories varied regarding which aspects of professional development they emphasised and how the relationship between components of the theory were depicted.

Nor did we find a field in which researchers have consistently tested or built on theories proposed by others. There was, however, evidence of attempts by authors of foundational papers to build a consensus around a theory of professional development and the resulting learning and growth among educators or children. As well as those papers in Group A who drew on extant theories, several authors in Group A made a deliberate effort to build on the theoretical work of other foundational papers. For instance, among the earlier foundational papers, Clarke and Hollingsworth (2002) presented a model which they argued could incorporate other theories, such as Guskey’s original model (Guskey 1986, cited in Clarke and Hollingsworth 2002). In 2009, Desimone (2009) claimed that there was sufficient evidence to support a consensus around the critical features of professional development and a proposed model which she argued captured this consensus. She drew on the work of Borko and Guskey extensively in this paper, including their papers in Group A, and referenced Clarke and Hollingsworth’s (2002) model. In 2011, Opfer and Pedder (2011) cited Clarke and Hollingsworth (2002) and Borko (2004), agreeing that previous models of professional development in education were too simplistic. Opfer and Pedder (2011) also discussed limitations of the models presented by Guskey (2002) and Desimone (2009) and refuted Desimone’s claims about a consensus around critical features of professional development. They also cited other work by authors of papers in Group A. In 2016, Kennedy (2016) echoed calls for a better theoretical understanding to underpin coaching. She drew on the findings of several other studies we identified as foundational (Borko 2004, Timperley et al. 2007, Yoon et al. 2007, Desimone 2009, Opfer and Pedder 2011). However, Kennedy (2016) noted that several of these papers emphasised components or ‘design features’ of professional development (Timperley et al. 2007, Yoon et al. 2007, Desimone 2009), not all of which were supported by the findings of her study.

These efforts to build a consensus around theory were not carried forward into the literature on coaching and similar forms of ongoing professional development in ECEC. Papers in Group B did not draw from theories used or proposed in the foundational papers in Group A, with the exception of 13 papers which referenced either adult learning theory, Guskey’s model or situated learning theory. As Opfer and Pedder (2011) noted, the various ‘strands’ of research in the field of professional development in education have tended to remain separate.

The findings of our study confirm that coaching in education is a field that has been differently conceptualised by different researchers and was thus suited to examination via a meta-narrative review (Wong et al. 2013). However, the diversity we uncovered was great, and the lack of referencing of foundational papers by papers on ongoing professional development in ECEC was striking. Hence, we were unable to identify ‘meta-narratives’ within the field, in which a research tradition can be seen to have unfolded over time. Further, we found little evidence of researchers looking to research from other fields to understand the mechanisms of change within coaching and other forms of ongoing professional development. As noted in Tables 2 and 3, only a few looked to fields such as organisational change, implementation science or the psychology of behaviour change or motivation, for example. As such, we found a field in which calls for an increased theoretical focus and momentum towards a theoretical consensus remain current.

Limitations

The limitations of this study pertain mainly to its scope. For practical reasons, we limited the scope of our search to the field of education. This might have limited our ability to identify meta-narratives. Similarly, we limited our search in Group B to ECEC rather than to the field of education more broadly, as initial test searches with this wider scope yielded an impractically large number of studies for screening. We also did not undertake a comprehensive search of the ‘grey’ literature, relying on intuitive and ancestral searching to locate literature not listed in academic databases, as recommended by Greenhalgh et al. (2005).

A further limitation relates to reliability. To enhance reliability, our screening processes could have benefited from being undertaken by more than one independent reviewer. However, this process is not required for a meta-narrative review given its iterative nature. Further, calculating the number of citations, a key inclusion criterion, did not entail subjectivity and thus would not have been affected by dual review. To support rigour in our screening processes in place of dual review, uncertainties about screening decisions at full-text stage were discussed among all authors before a final decision was made.

Implications

Our study did not identify a theory of coaching around which a consensus has been built in the field of ECEC. However, the findings of our study have implications for both practice and research.

Implications for practice

For those planning or delivering coaching, drawing on theory can help plan, refine or reflect on practice. This study identified a range of theories practitioners could draw on in this regard. Given the diversity of theories identified, there is a temptation to suggest that practitioners should draw from whichever theory appears to best align with their circumstances, resources or pre-existing plans. One might also be tempted to recommend that practitioners choose from the theories presented in the 11 foundational papers or from the 13 theories mentioned more than once across the groups. Boylan et al. (2018) suggest a series of questions that could support an informed selection of an appropriate theory for research or practice. However, not all theories are of equal explanatory value, and no theories identified in this study have been tested empirically on coaching undertaken in a range of circumstances. As such, further work needs to be done to identify which theories may be most useful to inform coaching practice, raising implications for further research.

Implications for research

We found a field that remains undertheorised, implying that further theory-building is required. Given the wide range of theories identified, the most pressing need is not to develop new theories. Instead, a research priority should be to build a consensus around theoretical issues.

As a first step towards this goal, we suggest that empirical research should test the explanatory power of existing theories. However, as Opfer and Pedder (2011) argued, testing a simplistic understanding of causal pathways is unlikely to identify mechanisms of change that could apply in a range of circumstances. As such, research should focus not on understanding if a coaching programme ‘works’, but on what the underlying mechanisms are that bring about change, in what context and for whom (Pawson and Tilley 2004).

Further, it is unlikely that one theory will have universal applicability, given the diversity of approaches, participants and contexts in which coaching takes place (Boylan et al. 2018). It may be that multiple theories are needed to allow for coaching to be understood as the embodiment of multiple pathways or mechanisms of change within various contexts (Pawson and Tilley 2004, Maxwell and Mittapalli 2010). As such, and before new theories are proposed, a next step could be seeking to make connections between theories shown to have empirical support (Boeler 2002 cited in Desimone 2009). As more theories and mechanisms are tested, the path towards building consensus will become clearer.

Conclusion

While coaching has been shown to be among the most effective forms of professional development for educators, coaching practice has been found to be extremely diverse and difficult to replicate or take to scale. This may be because the field lacks understanding of the mechanisms of change which drive the success of coaching. Our study suggests that coaching in ECEC remains undertheorised, and theoretical consensus is lacking regarding coaching and other forms of ongoing professional development. We found a wide range of theories applied and proposed and limited evidence of efforts to build on previous theoretical work. By identifying foundational literature in the field of ongoing professional development in education, our study helps to prepare the ground for further work in building theory and evaluating coaching in ECEC. Given the significance of high-quality ECEC in supporting young children’s learning and development, an improved understanding of the mechanisms of change that underpin the success of coaching is essential to harnessing the potential of coaching to improve teaching practice and support positive learning and development outcomes for children.

Disclosure statement

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

Additional information

Funding

The work was supported by the Australian Government Research Training Program Scholarship.

Notes

1. We searched Google Scholar using Harzing’s open-source software which allows sorting by number of citations.

2. We used the term ‘continuing’ rather ‘ongoing’ to refer to ongoing professional development in our search terms as this was better aligned with the terms used in academic databases.

3. For the first searches using coach* OR mentor*, we extracted the first 100 records after sorting by citation number. For searches using ‘continuing professional development’ we extracted the top 25–70 records as additional records were not highly cited.