Middle leadership in international schools: evidence from Malaysia

ABSTRACT This paper examines middle leadership of the heads of English, maths and science departments in four international secondary schools in Malaysia. It focuses on their roles, responsibilities, role relationships, instructional engagement and leadership involvement within the theoretical framework of instructional, distributed and teacher leadership. The study is a qualitative multi-method case study, involving observation, documentary analysis, and semi-structured interviews with 12 heads of department, 36 teachers, and four principals. With respect to the middle leaders’ roles and responsibilities, the findings illustrate cross-school and in-school differential developments, with contextual factors contributing to uneasy role relationships. The results also show that, while broad-based leadership opportunities are limited for the middle leaders and teachers, the most powerful and common feature of all these international schools is the centrality of teaching and learning.


Introduction
International schools have become key players on the social and economic landscape of education in the twenty-first century, with a turnover of revenue on a multi-billion scale (MacDonald 2006). International School Consultancy (ISC) reports the total number of English-medium schools to be over eight thousand, serving over 4.6 million students (www.isc-r.com). Malaysia is one of many countries contributing to the development of international schools.

Purpose of the paper
This paper presents and discusses research on middle leadership in four selected Malaysian secondary international schools within the framework of instructional, distributed and teacher leadership. The paper addresses the following research questions: (1) What are the roles, responsibilities and role relationships of middle leaders in the selected international schools? (2) How, and to what extent, are middle leaders involved in the leadership of the selected international schools? (3) How, and to what extent, are the leadership practices undertaken by the middle leaders linked to teaching and learning in the selected international schools? (4) How, and to what extent, can the practices of middle leaders in the selected international schools be understood through distributed and/or teacher leadership?
Following Wise (2001), we define middle leaders as practitioners who are heads of department (HoDs), manage the academic curriculum and lead a number of teachers. In this enquiry, the leadership practice of English, maths and science HoDs in four selected international secondary schools were explored.
The first research question is largely descriptive, and aims to examine the roles and responsibilities commonly undertaken by the HoDs. It also seeks to explore the quality of the relationships between the HoDs and their role sets, most notably the teachers and the principals. Research question two addresses the extent to which the HoDs are involved in the leadership of their schools. Research questions three and four aim to understand middle leadership practice against the backdrop of instructional, distributed and teacher leadership theories.

Literature review
Partial privatisation of education worldwide has revealed the limited capacity of solo leadership (e.g. Crawford 2012), leading to the revitalisation of the leadership role of the middle leaders. A review of the existing literature suggests five major themes pertinent to this role, distinguished below: . Roles . Responsibilities . Role relationships . Instructional engagement, and . Leadership involvement

Roles
In the mid-1990s, Ofsted (1996, the British government's inspection unit in England and Wales, defined middle managers as practitioners who are both teachers and managers of other teachers, and, in the capacity as heads of department (HoDs), they are engaged in managing the academic curriculum (Wise 2001). Wong, Wong, and Peng (2010, 63) define middle leaders in Hong Kong as teachers with formal administrative responsibilities, and in Australia, Gurr and Drysdale (2012, 57) define them as leaders with 'significant responsibility'.
Over time, the HoDs have been recognised to be 'key figures' (Busher and Harris 1999, 307), and in England, they are described as leaders who are capable of making a significant impact on the performance of departments (Earley and Fletcher-Campbell 1989). In China, the middle leaders are 'experienced teachers' who enjoy a respectable position with long-term professional commitment to one school (Tam 2010, 374).
Organisationally, HoDs reside at the centre. In Jarvis's (2008, 27) enquiry in England, and Fitzgerald's (2009, 58) study in New Zealand, the HoDs describe themselves as a 'conduit' (also see Glover et al. 1998). This 'middle' position highlights the importance of role sets, and their potential influence on middle leaders. Wise and Bush (1999) suggest several role sets, notably teachers, senior leaders, parents and governors. Wise (2001) adds that, of all these, teachers tend to have the strongest influence on middle leaders (58.8 per cent), compared to 21.7 per cent of influence from their senior leaders (Wise and Bush 1999). Tam (2010, 383) in China points out the 'challenging' nature of the HoDs' role. This partly arises from the department members' uncertainty surrounding the nature and the scope of an HoD's role (Jarvis 2008), affecting departmental performance (Earley and Fletcher-Campbell 1989).

Responsibilities
There are four parameters to be considered about the HoDs' responsibilities; the nature, the scope, the priorities, and the perspectives of the senior leaders vis-àvis the middle leaders.
Following Wise and Bush (1999), the nature of HoDs' responsibilities can be divided into four categories of academic, administrative, managerial and educative. Of these, all the responsibilities under 'managerial tasks' are highly relevant to this paper, and are shown in Table 1.
There are some concerns about the HoDs' scope of tasks, and these include additional roles and additional responsibilities. Earley and Fletcher-Campbell (1989) found that most of the middle leaders, alongside being homeroom teachers and pastoral heads, undertook responsibilities that extended to the whole school. More than half (51 per cent) of the HoDs in Wise and Bennett's (2003) enquiry in England, and 20 per cent of the HoDs in Weller's (2001) study in the US, report additional responsibilities. Fitzgerald (2009), in New Zealand, andGhamrawi (2010) in Lebanon, report a similar trend.
As a survival strategy, task priority may be introduced. This is evident in the rankings the HoDs provided for tasks presented to them by Wise and Bush (1999), as shown in Table 2, which displays the top four priorities.
Considering a low mean as indicative of a high priority, the table illustrates that, for the researched HoDs, the teaching role is the most important responsibility, highlighted in two intercontinental definitions in Europe (e.g. Ofsted 1996 in England) and in Asia (e.g. Wong, Wong, and Peng 2010 in Hong Kong).
In their systematic review, Bennett et al. (2007) identified two paradoxical pairs that can affect the perspectives of senior managers vis-à-vis middle leaders. One of these is whole-school emphasis vs. departmental loyalty, which involves the tendency of the HoDs to ally themselves with their departmental members. This influences the HoDs' relationships with their schoolbased role sets, senior leaders and teachers. Brown and Rutherford (1998, 86) identified that part of the tensions surrounding middle leaders emanated from their reluctance to involve themselves in whole-school matters.
Role relationships Tam (2010, 383) describes an HoD's role in China as 'challenging'. In England, Glover et al. (1998, 287) announce that the role of a middle leader is 'fraught with difficulty', with considerable tensions surrounding 'the nature and expectations of middle leadership' (Bennett et al. 2007, 462). In an early study, Lambert (1975) suggested the following typology, indicating the level of agreement between the headteachers and the HoDs with regard to the role functions of the latter cohort (see Table 3).
As the table illustrates, only the 'expressive' quartiles attract the lowest percentage of agreement, and between these two, 'expressive academic' receives the least consistency, and this is the area, which according to Lambert (1975, 37), is 'likely to be the source of possible role-conflict'.

Instructional engagement
HoDs are part-teachers and part-leaders, and when 'outside' the classroom, they remain connected with the classroom through several means, most notably monitoring (see Blasé and Blasé 2002;Southworth 2002).
Monitoring is a managerial responsibility (Wise and Bush 1999; see Table 1 above), and an important component of instructional leadership. There are two types of monitoring; formal and informal. Formal monitoring is commonly linked to appraisal, which occurs on a regular basis, by appointment, and by mutual agreement (Turner 2000). Informal monitoring, according to Glover  (1998), lacks rigour and may lead to insufficiently useful outcomes. Wise (2001, 338) deems it appropriate to approach informal monitoring as complementary to formal supervision. Alternatively, as Bennett (1995) states, some HoDs prefer to check exercise books. There are some persisting concerns over the effective conduct of monitoring (e.g. Ofsted 1996 and 2015). First, monitoring may affect the relationships between the HoDs and the teachers (Bullock 1988;Earley and Fletcher-Campbell 1989;Wise 2001). In some cases it is described as 'risky' (Hannay and Denby 1994), with the observers regarded as 'spies' (Metcalfe and Russell 1997). In Lebanon, Ghamrawi (2010) reports similar concerns. Second, monitoring, as an expected 'quality assurance process' may contrast with collegiality, as a process that emphasises 'mutual learning' (Bennett et al. 2007, 462). However, the most important, and persistent, barrier to monitoring remains to be shortage of time (Earley and Fletcher-Campbell 1989;Adey and Jones 1997;Brown and Rutherford 1998;Wise and Bush 1999;Wise and Bennett 2003;Busher 2005;Gurr and Drysdale 2012).

Leadership involvement
Middle leadership is the embodiment of distributed leadership (DL), and by extension, teacher leadership (TL). Harris (2008) speaks favourably about the outcomes of a leadership style that is inclusive and participative. A question often asked in the leadership and management literature concerns the nature of DL. We argue that DL is equivalent to the autonomy for middle leaders to take important decisions, and to the authority to implement those decisions without resorting to senior leadership for permission; any subsequent sharing of decisions is pursued for the purpose of whole-school coordination.
There is growing evidence about the benefits of shared decision-making. In Hong Kong, Tam (2010) attributes the achievements of Michael, a successful HoD, to his inclusive approach of collective decision-making. Similarly, Earley and Fletcher-Campbell (1989) and Harris (2008) in England, Turner (2000) in Wales, Hannay and Ross (1999) in Canada, and Chang (2011) in Taiwan, speak positively about the benefits of participative leadership.
However, Adey's (2000, 427) study in England shows that 47.9 per cent of the participants rated their influence on whole-school policy as average, and 31.6 per cent claimed that they had 'little or no influence'. These findings resonate with Buckby's (1997) study in England, andMercer andRi's (2006) enquiry in China, where the middle leaders were demanding an increase in their leadership involvement. These findings, however, contrast with Brown and Rutherford's (1998) recognition of leadership apathy as a cause of conflict.
Similarly, in their large-scale study on teacher leadership, Xie and Shen (2013, 342) conclude that teachers in the researched public schools in the US describe their leadership involvement as 'still mainly confined to the boundary of the traditional areas of classroom'.

Middle leadership in Malaysia
There is only limited evidence on middle leadership in Malaysia. The most directly relevant research is Javadi's (2014) enquiry into middle leadership in an international school in southern Malaysia. According to Javadi, the HoDs hold diverse roles, with dominant managerial responsibilities. He reports tension in the relationships between the HoDs and the senior managers, where most criticisms are directed at lack of autonomy. The HoDs are reluctant to conduct lesson observations, and resort to alternative methods such as checking worksheets. Lack of time continues to hamper the work of the HoDs at this school. According to Javadi, the majority of the HoDs do not perceive themselves as leaders of their departments. None of the middle leaders had received any formal training, relying, instead, on learning on the job. Evidence from this school in Malaysia is largely consistent with the international literature on middle leadership practice (e.g. Earley and Fletcher-Campbell 1989;Bennett 1995;Glover et al. 1998;Wise and Bush 1999;Adey 2000;and Wise and Bennett 2003

Methodology and methods
The research methodology employed in this study was qualitative in nature, and interpretivist in orientation, with an emphasis on seeking practitioners' views on the practice of middle leadership. A case study approach was deemed appropriate to permit in-depth access to the various areas of interest to this study. A salient feature of case studies is the possibility of employing multiple methods, most notably interviews, observations and documentary analysis.
An important issue in qualitative research is credibility, a.k.a. internal validity. This technique is used to increase the likelihood of the collection of healthy data (Flick 2009, 257). A major strategy to increase credibility of research is through triangulation (Bryman 2008), which operates on the logic of juxtaposing sources of information to ascertain their accuracy (Bush 2002). The possibility of using multiple methods in case study approach (e.g. Robson 2002) leads to methodological triangulation, which enables the possibility of comparing the accessed sources of information (Bush 2012).

Interviews
Interviews are the most common method of data collection (King 2004), and provide access to the phenomena, as conceived by humans (Arksey and Knight 1999). Semi-structured interviews were used to glean individual perceptions on a matter (Flick 2011). Follow-up questions and probes, an essential feature of semi-structured interviews, were also used to develop an in-depth understanding of an issue (Coleman 2012).
An interview guide was designed for each role set, containing four sections, linked to the research questions; roles and role relationships, leadership involvement, instructional engagement, as well as opinions and feelings about the school and departmental leadership. The interviews lasted between 45 minutes and one hour, and informed consent was sought and, the participants agreed to record the interviews.

Observation
Unlike interviews, observations provide powerful evidence of what people do. Of the several types, the 'observer-as-participant' (e.g. Punch 2009) was used to permit 'superficial' access (Waddington 2004, 154), complemented by in-depth interviewing (see above). Observations were used to understand the extent to which interactions were professional, attention was focused on teaching and learning, and leadership was decentralised. An observation schedule was created with blocks of 15 minutes (see Denscombe 2010, 200) for the activities in the staffrooms and in staff and/or departmental meetings.

Documentary analysis
Documents usefully provide access to the underlying sophisticated world of organisations (Bryman 2008). The most common documents available in schools, and used in this enquiry, are 'policies' (Fitzgerald 2012), e.g. role descriptions, or 'records of meetings' (Denscombe 2010), e.g. minutes, as well as staff handbooks.
Data from the documents, the observations and the interviews enabled a comprehensive picture of middle leadership practice in the four selected international secondary schools in Malaysia.

Sampling
Sampling for this study took place at two levels; sites and participants.

Sites
A search in School Advisor (www.schooladvisor.my) generated 21 schools in Kuala Lumpur. To avoid sampling error, which occurs as a result of discrepancies between the population and the sample size (Bryman 2008), stratified sampling was employed. This technique works upon the logic of the higher the homogeneity of a sample size, the more the accuracy of findings (see Kumar 2011), and this was achieved through the following criteria: . Secondary school . Curriculum type, i.e. the British IGCSE Curriculum . Predominantly Malaysian staff Schools that failed to meet these requirements were eliminated, yielding a stratum of eight schools (n = 8). However, the aim was for four, a number considered to be feasible in terms of time and research design. To achieve this, scrutinising the web pages of these eight schools was chosen as the primary point of contact. Of these, two sites were eliminated, as they were in the process of recruiting new principals, yielding a stratum of six schools (n = 6). This figure was subsequently reduced to four as, of the six potential sites, only four consented to this research, hence n = 4.

Participants
It was crucial to contact practitioners who would be in possession of and willing to share information pertinent to their profession (Kumar 2011). This non-probability strategy is called purposive sampling, which was used to determine target departments as well as participants.
Traditionally, maths, science and languages -English in this studyhave been central to schooling, and their results alone are used by the Programme for International Student Assessment (PISA) (OECD 2017). To contact the HoDs of these departments, the principles of 'relevance' and 'knowledge' were considered, which are both central to purposive sampling (Denscombe 2010). Furthermore, the authors agreed that a sample size of three teachers in each department would provide a comprehensive understanding of middle leadership practice. However, recalling the fluidity of the staff at international schools, unless the targeted teachers had been in post for some time, such an understanding would not be fully achievable. Therefore, snowball sampling was used, the starting point of which is from 'a known network' (Dattalo 2008, 6), which in this case, was the HoDs who would recommend their long-serving colleagues. However, this strategy was unnecessary in schools B, C and D as the number of participants required precisely matched the number of teachers available.
In total, 12 HoDs, 36 teachers and four principals (n = 52) agreed to participate in this research, enabling respondent triangulation. This strategy is used to address the issue of credibility (internal validity), as noted above, through eliciting information from multiple stakeholders.

Data analysis
In this enquiry, data collection and analysis received contemporaneous attention (e.g. Hesse-Biber and Leavy 2011). Data were reduced (see Miles and Huberman 1994), using two key techniques; coding (Cassell and Symon 2004) and memoing (Bryman 2008). In this study, the codes and memos used were consistent with the five main themes of middle leadership identified in this study; roles, responsibilities, role relationships, instructional engagement and leadership involvement.
Departments were the primary units of analysis, and received substantial analytical treatment. This approach enabled comparison of middle leadership practice in individual departments within each school, and across the four schools.

Findings and discussion
This section presents an integrated thematic discussion of the findings of the case-study international schools with respect to roles, responsibilities, role relationships, instructional engagement and leadership involvement.

Roles and responsibilities
The examination of roles in the four schools involved documentary analysis and interviews. These data point to three important conclusions. First, while middle leadership is perceived to be well-established at schools A, D, and C, it is loosely defined and poorly understood at school B, as described by the English HoD: Initially I was supposed to be the A-level coordinator, but the person who was the English HoD left, and it just slid over to me. It was never defined to me; it was just a given title. I have no job description. I have no idea what the role involves.
This finding lends support to the conception of differential cross-school development of departments. Second, within a single school, departments may display varying stages of development. For example, the English department at school A, and the science departments at schools D and C, were perceived to be faring better than the other departments in the same school. In school B, where middle leadership is ill-defined, the English department was found to be the least effective, as evident in this teacher's remark: 'I like the school because no one comes and tells me what to do. I can do whatever I want. There is freedom in this school'.
Finally, the consistent grand picture from the cross-case data analyses indicates that middle leaders' role conceptions, within and across the four schools, tend to be arbitrarily defined and subjectively interpreted, leading to differential, and inconsistent, middle leadership practice. The documents in the four schools articulate the formal organisational expectations of the HoDs. They require that the HoDs be 'excellent teachers' (school A) with 'relevant expertise' (school B) and 'full weekly teaching load' (school C) who 'take responsibility for teaching and learning' in their schools (school D) (see Wise 2001). These phrases serve to highlight the importance of teaching and learning for these schools, and echo the priority of teaching for Wise and Bush's (1999) researched HoDs (see Table 2). Emphasis on teaching also highlights the dominance of managerial responsibilities (see Table 1), as reflected in this comment by the maths HoD at school C: I have to go and make sure that the lessons are conducted well; students [and]  These circumstances lead to the conclusion that 'middle management' continues to frame the roles and responsibilities of the participating HoDs, undermining the notion of 'middle leadership' in the title of this paper (see Buckby 1997;Adey 2000;Mercer and Ri 2006).
Hierarchically, HoDs are situated in the 'middle'. In all the schools, a middle leader is described as a 'facilitator', 'messenger', 'spokesperson', 'go-between' (school A), 'inspector' (school B), 'intermediary', 'coordinator', 'facilitator' (school C), 'organiser', 'supervisor', 'middle person', 'representative', and 'bridge' (school D). According to the participants, a leader in this position is expected to 'fight', 'take care', 'support', 'link', and 'pass messages' between the 'top' and the 'bottom'. Although different participants chose to select distinct terms to describe their HoDs, they all allude to the 'conduit' role of the middle leaders, re-echoing earlier claims (e.g. Glover et al. 1998;Jarvis 2008;Fitzgerald 2009 The distinction between the formal role articulations in the staff handbooks, and the role comprehension by individuals, uttered in the interviews, especially at school C, led to the emergence of the 'visibility' and 'invisibility' dichotomy. 'Visibility' can be used to describe the immediate services which teachers receive from their HoDs. Conversely, 'invisibility' can be used to describe the services which teachers do not receive from their HoDs when they are expecting them. These situations are most likely to occur when, even if available, the official staff handbooks are rarely consulted, as was the case with all the participants in these four schools.
Middle leaders' roles and responsibilities at schools D and B are special, and hence, require further attention. In the relatively new school D, middle leadership activities are strongly influenced by the 'tyranny' of the heavy timetable, restricting leadership activities other than teaching: … very heavy timetable that we've all got. I've got only two free periods a week where none of us are teaching … there's no space anywhere to think properly. (The English HoD at school D) This may account for the dominance of the teaching role for the HoDs in this school (see Wise and Bush 1999, Table 2 above). In school B, however, middle leadership activities are overshadowed by the noticeable presence of the principal, leading to a loose understanding of middle leadership. The only reference in the staff handbook consists of a few lines. What unites, or otherwise, the HoDs with the principal is unclear, but what is certain is that it is not their role. This makes it difficult to discuss middle leadership in terms of departments because such a concept barely exists in the minds of the participants at this school. These circumstances have forced them to rely on the principal for accessing 'accurate' information: Almost all the decisions are approved by the principal. So, I can't be simply giving instructions because I have bigger people than me here, so I consult them before I give orders to [the teachers]. (The science HoD at school B) The variations in the middle leadership practice, and the differences in understanding the role are not limited to the school level; they also extend to the department and individual levels, leading to extensive uncertainty (e.g. Earley and Fletcher-Campbell 1989;Jarvis 2008), and abundant personal judgements.

Role relationships
Factors influencing the relationships between the HoDs and their school-based stakeholders are varied and distinct, and these include concerns over attitude (school A), principal's dominance (school B), unstable senior leadership (school C) and tenure (school D), confirming claims by Glover et al. (1998), and Bennett et al. (2007), that the middle leadership role is replete with tension.
The picture that emerges from the data suggests that concerns about the middle leaders' behaviour and attitude outnumber those expressed about their knowledge and expertise. Half of the 12 HoDs are confidently admired for their knowledge and expertise. In contrast, most of them, that is the nine HoDs at schools A, B and C, are blamed, although variously, for their behaviour and attitudes: Some of them [teachers] I find quite pushy in a sense that, when I say let's do this, then someone has a suggestion to do something else. (The HoD at school A) I may disagree with a lot of things she says, [but] whenever we raise concerns she doesn't really listen to us; I feel I can't work. (An English teacher at school A) This conception is consistent with Lambert's (1975, 37) finding that the 'expressive-academic' section of his typology is an area which is 'likely to be the source of possible role-conflict'.
At school D, the HoDs are the only leaders who are consistently praised. This is not because they are brilliant leaders, or exercise leadership in its most inclusive fashion, but because, being in a new school, they have not yet had the opportunity to abandon the 'safe' domain of the classroom and department to experience the 'less comfortable' domain of active and broad-based consultative leadership, which is characteristically complex and challenging: So far it's good. I mean we do go out, and have some drinks, and at school, [the HoD] is professional … she's very supportive. (An English teacher at school D) In school B, the quality of the relationships has been affected by overdependence on the principal. Given the reported and observed tense relationships between some of the participants, as well as the other members of staff, and in the absence of a departmental identity, the quality of the participants' relationships with the principal has gained considerable significance: I think they [the teachers] cannot immediately go to the principal and talk to her. You know, as a teacher, you cannot knock on her door and immediately talk to her. (The maths HoD at school B) This may explain that why, despite tense relationships, all the participants expressed satisfaction with the principal, although less so with the way the school was managed.
A certain degree of 'subdued' resentment could be discerned among the participants in school Ca sort of unexpressed 'concord' that had convinced them to take refuge in their isolation. Considering the leadership history of school C, with eight principals since the late 1990s, the participants' behaviour might be regarded as a legitimate response to this highly volatile leadership trajectory. The participating teachers are not at odds with their HoDs; they and their HoDs are all victims of unstable leadership. According to the interview data, the average tenure of the HoDs is about eight years. Thus, what these three departments represent is a relatively stable middle leadership, as opposed to a history of unstable senior leadership, which continues to be a source of concern at this school: The owners of the school like to micromanage, and he [the head of secondary] has been here for a short time. I don't think they trust him yet. (An English teacher at school C) If the directors and the owners let us run the school, we'll flourish. If they put restrictions on us, and keep interfering, then we'll go backwards again to where it was. (The head of secondary at school C) The quality of the relationships in school D is a function of tenure. The science HoD has the longest length of servicesince 2014 and for two years. Although the maths HoD has also been in this school for two years, she was only promoted to this position in 2015, when both the head of secondary (HoS) and the English HoD were recruited. Interview and observation data found both the science HoD and the HoS on favourable terms with each other: The science HoD has done a fantastic job. She is leading [the department] in the right direction; definitely she's experienced [and] knowledgeable, and she's got great vision. (The head of secondary at school D) This positive relationship can be understood in the context of the support and advice the former is able to provide to the latter due to her familiarity with the school's structure and processes from the beginning. This was evident from the frequent visits the HoS made to the science lab. Although both the English and the maths HoDs joined the leadership team at the same time, the advantage of the latter, a Malaysian leader, for the HoS is her familiarity with the local context: I think the head of secondary is very supportive … I have a very good relationship with him. So far everything is good because there is somebody who is willing to listen. (The maths HoD at school D) I think there are still bits of communication that we haven't been clear with her [the English HoD], and she's not sure about all the aspects of her role. That's understandable because she's been here for a few months. (The head of secondary at school D) In broad terms, the evidence about conflicts between the HoDs and the teachers, as well as the HoDs and the principals, contrasts with Wise and Bush's (1999) findings about departmental loyalty. These authors report that 58.8 per cent of the HoDs in their research chose to ally themselves with their departmental staff, compared to 21.7 per cent of influence they tend to accord their senior leaders. The evidence from this study does not seem to reflect this. On the contrary, there seems to be more agreement between the HoDs and their senior leaders than between the HoDs and the teachers: The explanation for this discrepancy may be that, unlike Wise and Bush's enquiry in public schools, the research in this paper takes place in private schools, at the centre of which lie accountability, fees and parents. These may have convinced the HoDs at all the schools to align themselves with the senior management team and not so much with the teachers (see the discussion below for more on this point).

Instructional engagement
Of the themes guiding instructional leadership, the evidence focuses mainly on monitoring (e.g. Blasé and Blasé 2002;Southworth 2002). When compared to the existing empirical literature, the general picture about lesson observations provides mixed messages. In contrast to the concerns expressed by Ofsted (1996Ofsted ( , 2015, Bullock (1988), Earley and Fletcher-Campbell (1989), Hannay and Denby (1994), Metcalfe and Russell (1997), Wise (2001), Bennett et al. (2007) and Ghamrawi (2010), the notion of lesson observation in these schools is found to be the least contentious area, but with variations in terms of development. According to the documentary analysis, supported by the interviews, monitoring in schools A and D has received sufficient attention, but this is largely under-developed at schools B and C: [The HoD is expected to] appraise staff as required by the school policy and use the process to develop the personal and professional effectiveness of teachers. (School A's staff handbook, section 6: managing performance) [The HoD is expected] to assist in the observation of subject teachers to identify strengths and weaknesses in the teaching of the subjects he/she is in charge of. (School D's staff handbook, clause 2: observe teachers) Moreover, contrary to the findings by the afore-mentioned authors, the participating teachers tend to accept their HoDs' monitoring roles with little resistance, and tend to view them as a common feature of their job descriptions. A science teacher at school A has this to say about the inadequacy of onelesson observations: I do not agree with this because there are some teachers who prepare one lesson for the observation. I think that every lesson you should be able to observe, and the head [of department] should actually observe more lessons to get an overall view. It's not really fair. This remark reinforces the claim above that the HoDs tend to align their leadership activities with organisational requirements rather than departmental fervour. Significantly, this finding also coheres with Wise and Bush's (1999, 192) understanding to hold 'external pressure' responsible for their researched HoDs' increasing attention to monitoring (see Table 2), exerted in England through Ofsted inspections, and in the case of this study in Malaysia, by parents as fee-paying clientele. Parents' opinions are taken very seriously across the four case studies so much so that school D's head of secondary felt compelled to conduct lesson observations in his new school: Yes, I've done that on a number of occasions. It's important because I feel there are some teachers who have limited experience … Therefore, it has an impact on learning … on progress [and] on what students go home and communicate to parents.
Arguably, the most powerful feature of school A lies in its supervision of teaching and learning. It conducts formal lesson observations, based on mutual agreement, and linked to performance management, as well as informal drop-ins, known as 'learning walks', on an ongoing basis, and throughout the academic year, which echo the systems and mechanisms described by Glover et al. (1998) and Turner (2000): I've done two formal observations of every member of the department this year … we also have a programme this year called 'learning walks', so a member of the senior management team plus a head of faculty would normally at least once a week go on multiple learning walks which would involve dropping in on anywhere between four, six or seven lessons in a day for only 5-10 minutes to just make a note of what was going on and then feed back to the teacher what was in there of good practice (The English HoD at school A) The learning walks model pursues two goals. First, it creates opportunities for inter-departmental observation and learning, enabling, for example, a science teacher to visit a language classroom, and vice versa. Second, it could potentially serve as an antidote to 'showcase' lessons, as it is exercised on the basis of surprise visits. A 'showcase' lesson is one that is exclusively planned to satisfy an observer, and may depart, to a small or large extent, from the norm: 'Some teachers know that, when I'm coming, they will put up a show' (the science HoD at school A). The supervision processes are uniformly understood by the participants and routinely exercised by all the HoDs. The staff handbook contains several detailed sections that clearly articulate the instructional expectations and performance indicators. This contrasts with the absence of a distinct document on professional behaviour, and serves to cohere with the view above that instruction (subject knowledge) is taken more seriously than behaviour (attitude). Moreover, alternative methods, such as checking exercise books, are also used in school A (e.g. Bennett 1995): The Head of Department will be able to implement and supervise an effective system to check student's written work by way of checking books and challenge teachers about any errors or inconsistencies in marking. (School A's staff handbook, section 3: assessment and evaluation) Similarly, school D tends to place great emphasis on instruction so that all the other aspects of school leadership are either aligned with this, or have been affected by it. The formal and informal monitoring models in school D resonate considerably with the systems in place in school A, albeit at a lower degree of rigour due to the young age of this school, and its heavy timetable: 'I do one formal observation myself; I do half term, which they get feedback from. I also do drop-ins for 10 minutes of their lessons' (The science HoD at school D).
Monitoring of teaching and learning at school B also receives attention, but this is limited compared to schools A and D, where there are some criticisms. Commenting on the 'surprise' nature of classroom visits at school B, a science teacher remarked that 'it would have been a little more polite to have let me know earlier' rather than 20 minutes before the start of his lesson. Despite some observed irregularities, such as high teacher absenteeismseven in one instanceleaving some classes unattended and indicating poor relief management, school B seems to be conscious of the importance of quality instruction, and thus, pays attention to formal lesson observations. One distinctive feature of monitoring in this school is its 'dual-observer' system, where two observers, one essentially being the subject expert, visit a lesson and provide feedback. Although the overall observation timescale is announced, the actual classroom visits remain unannounced, with a view to capturing authentic content delivery, and discouraging showcase lessons: Because sometimes when we tell the teachers that we're going to class, they will prepare. We want to see what actually they are doing in normal classes. (The science HoD at school B) Questions about formal monitoring at school C received mixed responses, which led the researcher to seek information from various role sets. Nevertheless, the interview data suggest a similar model to that of school A. However, at the time of this research, the school was engaged in seeking accreditation from an international body, which had caused a moratorium on lesson observations. Also, alternative methods, such as checking exercises books, receive encouragement in the staff handbooks (see Bennett 1995).
As the data above indicate, all the case-study schools conduct formal lesson observations, while schools A and D also undertake informal monitoring. Both Southworth (2002), and Blasé and Blasé (2002), highlight the benefits of informal interactions in improving the quality of teaching, and both these authors are silent on the issue of formal observations. Despite this, Glover et al. (1998) are sceptical about the effectiveness of informal monitoring, and Wise (2001) suggests a combination of both formal and informal observation methods, as in schools A and D. Moreover, as stated by Bennett (1995) in England, and Javadi (2014) in Malaysia, there are indications that in schools A, C and D, the HoDs rely on alternative methods, such as checking exercise books, to monitor instructional quality.
Time constraints are frequently cited as a barrier to the monitoring role of middle leaders (Earley and Fletcher-Campbell 1989;Adey and Jones 1997;Brown and Rutherford 1998;Wise and Bush 1999;Wise and Bennett 2003;Busher 2005;Gurr and Drysdale 2012). The findings from the four case-study schools suggest that middle leaders' teaching role continues to dominate their job scope, albeit to varying degrees. This is consistent with Wise and Bush's (1999) evidence about the very limited leadership time allocations for the HoDs. The most affected school is D, with HoDs who have trouble finding a balance between their heavy teaching load and their formal and informal monitoring responsibilities. The data indicate that impediments to quality control are not necessarily limited to school-based factors. School C is a good example of the impact that an external intervention, the accreditation project, can have on the internal functions of the HoDs.
In general, however, monitoring is the most powerful feature of these schools, but it constitutes only one component of instructional leadership. Evidence of other elements of instructional leadership, for example, modelling and mentoring, is very limited at these four case-study schools.

Leadership involvement
As noted earlier, middle leadership is the embodiment of distributed leadership, and by extension, teacher leadership. Conceptions about the HoDs', and the teachers', leadership roles can be analysed by making a distinction between 'serious' and 'less serious' issues.
During the field work in all the four schools, it became clear that the HoDs' autonomy may be applied differentially for 'serious' and 'less serious' categories. In broad terms, 'less serious' matters are linked to teaching and learning, and confined to the classroom domain. In contrast, 'serious' issues pertain to policies, and operate beyond the boundary of the classroom or the department. For the former category, the HoDs claimed, and were reported to have, a greater measure of autonomy, compared to the latter category, for which opportunities were said to be limited: Only if it is a very drastic change … she [the HoD] still has to ask the principal … it depends how serious or how much it concerns the parents or the whole school. (An English teacher at school A) We can take a decision for small things, but if we have a bigger problem, we should consult [the principal]. (The maths HoD at school B) If it's something very simple that does not involve costs to the management or the directors or the owners of the school, then I think that is fine. So, that depends on the sort of issues or things on the table. (The science HoD at school C) Like for the activities, [e.g. book week and trips], we can actually decide and implement them, but things like the budget … we have to get the principal's approval. (An English teacher at school D) Evidence from school A suggests that three elements inform the extent to which a matter is considered serious or otherwise; task type, location, and people. For example, if a specific task should take place outside the classroom, and it involves parents, it should then be a serious matter, for which broadbased leadership opportunity is limited. On the contrary, if a particular task should occur inside the classroom, which involves students, it should then be a less serious matter, for which there are more opportunities for participative leadership. In short, there is more possibility for distributed and teacher leadership within the domain of the classroom, and much less so outside this domain, confirming evidence from Buckby (1997) and Adey (2000) in England, Mercer and Ri (2006) in China, and Javadi (2014) in Malaysia, that middle leaders continue to have limited opportunities for a wider leadership role. Similar distinctions began to emerge at school B, despite the concentration of leadership in the figure of the principal. Wider leadership was also limited at school C, where there had been multiple leadership successions, and at school D, where it was constrained by a heavy teaching load. Distributed leadership and, by extension, teacher leadership entail two important dimensions; the power to make decisions and the authority to implement those decisions without the need to seek permission from senior leaders. However, the HoDs, especially at schools A and D, preferred to consult their superiors, and offered several reasons for this choice. One reason, given by the English HoD at school A concerns accountability, which caused her to feel compelled to avoid 'creating problems for the school' or 'creating any major logistical challenges'. Another reason implies a 'face-saving' attempt to preclude embarrassing situations. Both the English HoDs at schools A and D said that they chose to seek their superiors' views in order not to find out later that they were not allowed to proceed with what they had finalised with their teachers: The final reservation contains cultural overtones. For example, the maths HoD at school A felt that it would be disrespectful if he tried to bypass the assistant HoS to reach the HoS: I think, as a professional, there should be a proper channel. For example, if my teachers wanted to implement or suggest something, I think they ought to let me know first … at least I'm aware. I think if you jump the channel, it's not very respectful. Similar views were expressed by teachers, particularly in schools A and D. The rationale behind this preference can broadly be divided into two categories; technical and cultural. Technical consultation takes place with the HoDs, and/ or senior teachers, for their ideas, experience, and knowledge; one goal of this, in the words of a teacher at school A, is for standardisation purposes. Culturally speaking, some other teachers consider consultation with the HoDs 'out of respect' (school A), and 'the right way to do [things]' (school D).
The picture that emerges from these four schools points to restrictive conceptions and barriers for the HoDs and the teachers to exercise leadership in the manner that characterises distributed and teacher leadership in their most collegiate form. This evidence, from Malaysia, confirms research in the USA where teachers, and by extension HoDs, describe their leadership involvement as 'still mainly confined to the boundary of the traditional areas of classroom' (Xie and Shen 2013, 342).

Conclusion
This paper reports findings of a six-month enquiry into middle leadership practice at four selected international secondary schools in Malaysia. Studies pertinent to middle leadership and international schools are growing. However, in Malaysia, the evidential base is extremely limited. Therefore, this study is the first major examination of middle leadership in Malaysian international schools, and claims substantial contributions.
The evidence from these four case study schools undermines the use of 'leadership' to refer to the activities undertaken by the participating HoDs. Alternatively, and despite calls for 'leadership', as early as in the 1980s (e.g. Earley and Fletcher-Campbell 1989), 'management' seems to be a more appropriate term. The instructional responsibilities of the middle managers in this study tend to take top priority. Their role is understood in fragments, leading to the emergence of the 'visibility/ invisibility' dichotomy. Lack of clarity of roles and responsibilities, especially at schools A and D, has resulted in the development of arbitrary relationships, to which considerations such as the dominance of the principal at school B, and leadership instability at school C, can be added. Leadership participation across the four schools was found to be a function of three parameters; task type, location and people, leading to the introduction of the 'serious/less serious' dichotomy, indicating that the most serious cross-school decisions fall within the remit of the senior leader. However, instruction, and the monitoring aspect of instructional leadership, continue to be the most powerful features of these case study schools.
There are several lessons to draw from these findings. First, despite expectations about the diversity of international schools, the evidence suggests somewhat uniform practices. For example, in all these schools, roles lack clarity, instructional responsibilities dominate, and complaints about professional attitudes prevail. Although instructional quality is valued, through formal and/or informal monitoring mechanisms, broad-based leadership remains limited for middle managers and teachers. Second, the great majority of previous studies on middle leadership and management were conducted in public schools, not in the international school sector. The findings of this enquiry are largely consistent with the international literature, reinforcing this assumption that bearing 'international' in the names of schools does not necessarily mean a significant difference. Conversely, there are great similarities between the middle leadership activities undertaken in international schools and with those in their public counterparts. The final lesson to be learned from this study resonates with Ribbins's (2007, 27) view that 'we should not assume that, just because time has changed, things are necessarily significantly different'.