The implications of school improvement and school effectiveness research for primary school principals in Ethiopia

In the last 20 years the Ethiopian education system has rapidly expanded, leading to a 500% increase in primary school enrolment. The Ministry of Education (MoE) has sought to address a perceived decline in educational quality through nationally-mandated programmes for school improvement and teachers’ professional development. Such programmes – imported on the advice of the United States Agency for International Development (USAID), Department for International Development (DfID) and other development partners – are implemented “mechanistically”, without adaptation for societal or organisational cultural contexts. This article reviews empirical research from the last decade, including “grey” literature, in the school improvement (SI) and school effectiveness research (SER) traditions to draw implications for primary school principals in Ethiopia.


Introduction
In the last two decades the Ethiopian education system has undergone massive expansion. The number of young people participating in primary schooling has grown from three million in the early 1990s to 17 million currently (MoE 2012a). In this period Ethiopia has gone from having one of the lowest education participation rates in the world (ODI 2011), to enrolling 85% of young people aged 7-14 (ODI 2011), and is judged to be "on track" for achieving the Millennium Development Goal of universal primary education by 2015 (MoFED 2012). Expansion has been achieved through increased state investment (Dom 2009), community contributions to the construction and maintenance of rural schools (MoE 2005;Jeilu 2009), and direct support from foreign donor countries amounting to roughly 35% of the education budget (ICAI 2012).
The rapid expansion has been accompanied by concerns for the quality of primary schooling (MoE 2008;ICAI 2012), which the Ethiopian government has sought to address through policies for decentralisation, community participation, teacher and school leader professional development, and textbook production (MoE 2008(MoE , 2010. As part of these reforms, tools for schools' self-evaluation and development planning have been introduced, drawing from the school improvement (SI) research tradition (Hopkins 2001). School effectiveness research (SER) has been undertaken as part of the Department for International Development (DfID)-sponsored Young Lives project, and the United States Agency for International Development (USAID)-funded Education Quality Improvement Programme (EQUIP). Until now, no attempt has been made to synthesise these findings or draw out implications for school leaders. This article investigates recent (2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) empirical research from these two research traditions (SI and SER) to identify implications for school principals and "blind spots" in current research. This article addresses the questions: What are the major findings of national and international studies in Ethiopia within the SI and SER traditions? What are the implications of this research for school principals?

Context
While the resources available to any education system are always restricted to some degree (Glover and Levačić 2007), this is especially the case in Ethiopia, which ranks 174/187 on the Human Development Index (UNESCO 2010a). It is one of the least urbanised countries in the world, with 84% of the population living in rural areas where 40% of households have access to safe drinking water and 5% have electricity (CSA 2012). However, stability and development in recent years have led to steady improvements in the standard of living across the country: infant mortality has decreased by 39% in the last 15 years (CSA 2012), and life expectancy has risen from 50 to 62six years above the sub-Saharan average (World Bank 2013).
Following the overthrow of the Derg military dictatorship in 1991, the state was reconfigured as a Federal Democratic Republic, with significant authority devolved to nine ethnically-based regions and two city administrations. In this federal system, the Ministry of Education (MoE) sets national policies and curricula which the Regional Education Bureaux (REBs) are responsible for adapting and administering; and directives and funds flow down to the Woreda (district) Education Office (WEO), which is charged with allocating budgets and supervising the quality of education in schools (UNESCO 2010b). Primary schooling covers the first eight years of general education (Grades 1-8, ages 7-14), and a selection of quality indicators for this phase is presented in Table 1.
These figures represent national averages; there are regional variations, and the extent of within-region variation is not captured in these statistics. However, these figures indicate some major challenges. Just over half of the students who enrol in primary education complete the eight-year programme, and a quarter drop out after the first year, meaning that roughly three million young people of school age are not participating in formal education (MoE 2012b). The national policy of "automatic promotion" in Grades 1-3 (MoE 2005) is not implemented consistently, with roughly 10% of students repeating each grade. Drop out and repetition have been attributed to young people's engagement in work and the inadequacy of schools' infrastructure and learning resources (MoE 2012b). The lack of sanitation facilities is a barrier for all, but female students in particular (MoE 2010), and this is especially a problem in rural areas (MoE 2010), where a large part of responsibility for the construction and maintenance of schools is borne by local communities (Jeilu 2009;Yamada 2013). The government expresses concern (MoE 2008) at a perceived deterioration in the quality of schooling, as evidenced by an overall decline in Grades 4 and 8 attainment in National Learning Assessments (NLAs) between 2000-2007; although Dom (2009) points out that such a dip is not surprising given the inclusion of many previously-excluded young people from deprived backgrounds. Charged with meeting these challenges is the principal. In the past, there were no formal qualification requirements for principals, who were elected to the post by colleagues (Abebayehu 2005). The government (MoE 2008) intends that all primary school principals receive diploma-level training in educational planning and management, however the most recent sector development plan (MoE 2010) set no targets for achieving this. In official publications, principals are referred to as "leaders" and "managers" interchangeably (e.g. MoE 2008). Dimmock interprets leadership as: a higher order set of abilities such as goal-setting, visioning, and motivating, whilst management is viewed as a lower order group of activities concerned with the maintenance of performance through supervision, co-ordination, and control. (Dimmock 2002, 33) In Ethiopia, the principal is best understood as a local manager of an administrative unit, but the current challenges, and recent policy initiatives, call for principals to take a leadership role in schoolsa point in tension with the bureaucratic structure of the education system, as will be revealed in this article.

What does SI tell us about schooling in Ethiopia?
The SI tradition emerged in industrialised countries as a decentralised approach to school reform within nationally-set policy and accountability structures (Bryk et al. 2010). SI is concerned with how schools can effect beneficial change for students in terms of learning (Bryk et al. 2010) and the quality of experience (Rudduck, Chaplain, and Wallace 1996), and has been defined as "the process of enhancing the way the school organises, promotes and supports learning" (MacBeath and Mortimore 2001, 37). Improving schools' internal conditions requires the involvement of all levels of the school community (Rudduck, Chaplain, and Wallace 1996;Hopkins 2001), demanding leadership of change for organisational and professional learning (Pedder and MacBeath 2008;Bryk et al. 2010). Key elements of SI are: Self-evaluationwith inputs from all levels of the school community for the purpose of identifying barriers to learning (MacBeath 2010); Development planningwith broad participation in decision-making, to foster ownership (Bryk et al. 2010) and ensure impact across all levels of the school (Hopkins 2001); Continuous professional development (CPD)which emphasises that schools are "places of learning" for staff as well as students (Mitchell 2013).
Each of these elements has been a feature of MoE policy in the last decade, starting with the national programme for teachers ' CPD (MoE 2004), and followed by the school improvement programme (SIP) (MoE 2007) which fits Harris' (2000) conception of a "mechanistic", top-down programme. The SIP introduces a school selfevaluation framework covering four domains: (1) Teaching and learning (2) Student environment (3) Leadership and management (4) Community involvement Each domain comprises 10 or more indicators by which schools are required to evaluate their performance. Some of these indicators are reasonable ("The effectiveness of teaching and learning in all curricular activities is reviewed regularly", [MoE 2007, 25]) while others are inappropriate ("The school website, where available, is current, interesting and interactive", [MoE 2007, 40]) and demonstrate the framework's foreign provenance.
Like other recent initiatives, SIP was introduced on the advice of development partners such as Voluntary Service Overseas (VSO) (Pillay 2010), and reflects both the SI tradition's focus on learners, self-evaluation and development planning (Hopkins 2001), and the World Bank's push for financial decentralisation (Garcia and Rajkumar 2008). Implicit in SIP is a repositioning of the principal from an administrator to a leader of the school. Whereas principals were previously responsible for administration, and upwardly accountable within the civil service system, SIP charges them with decentralised planning based on community consultation and locally identified development priorities. SIP is predicated upon principles of Western educational theory (e.g. democracy, reducing power differentials and collaboration), and has not been adapted for Ethiopian societal or organisational cultural contexts. The following sections consider empirical research findings relating to the conditions for SI in Ethiopia.

Self-evaluation and development planning
As indicated earlier, the positioning of the school as a self-managing unit is central to SI initiatives. Workneh (2012) explains that this ideal has only recently become a feature of central initiatives through policies such as the SIP, and is not yet embedded in practice. His qualitative study investigated participation in school management and decision-making using interviews with teachers (43) and directors (15) at 15 schools in Amhara, Tigray, Oromia, SNNPR and Addis Ababa. He found that school-level stakeholders showed variable willingness and ability to participate in self-evaluation and development planning for SIP domains 1, 2 and 3. A significant barrier to school-based management occurred at the district level, since WEOs passed down central and regional directives and were also responsible for equipping and staffing schools and training teachers. Only limited funds were passed on to the school for maintenance and equipment (domain 2), and WEOs retained responsibility for procuring material resources; complex purchasing processes resulting in frequent blockages of essential learning materials. Workneh (2012) found that WEOs resisted schools' attempts to contextually adapt the SIP, hindering local improvement efforts.
While the MoE (2008) resolved that school grants should be sent directly to schools to facilitate local decision-making and planning, the principals in Workneh's (2012) study had not received these grants, and many felt that WEOs would take responsibility for them, were such funds to become available. Regarding teacher recruitment and in-service training, principals were reliant on decisions made by the WEO, and some teachers expressed the belief that such decisions were influenced by "social connections, personal affiliation and political attachment" (Workneh 2012, 14). Schools were unable to hold WEOs to account because, as one teacher explained: "the [principal] in this school does not want to complain to [the] WEO probably because he wants to be liked by the officers in the woreda" (Workneh 2012, 15). This study suggests a paradox in the system: WEO officials are the superordinates of principals within the civil service structure, yet the principal is charged with bottom-up accountability to local communities. Despite these obstacles to school-based management, Workneh (2012) is optimistic about the progress that has been made in this area, especially regarding community involvement in decisionmaking (domain 4).
Community participation is further explored in Jeilu's (2009) UNESCO study, which surveyed directors (58) and parents (91) at 64 schools in Amhara, Oromia, SNNPR and Addis Ababa. Most respondents (83%) reported that their school was supported by the community through cash contributions for books, furniture and maintenance, which the author reports offered communities "a sense of ownership over their schools" (Jeilu 2009, 55). However, 74% indicated that the community did not support the school by engaging in consultation, as required for selfevaluation and development planning. Swift-Morgan (2006) investigated the meaning of "community participation" in eight rural communities in SNNPR using interviews with teachers, principals and parents. She found that, through the Parent Teacher Association (PTA), parents did participate in decision-making, regarding issues such as school repairs and negotiating fees (domain 2); but they were rarely involved in planning for improvements. Even though the PTA was ostensibly charged with evaluating teachers' performance (domain 1), Swift-Morgan's (2006) respondents felt that the community had no viable role to play in this areaprimarily because parents were "uneducated", as one teacher explained: the giving of this authority is very bad because most of the PTA members are uneducated and very poor. The educated cannot be evaluated by the uneducated. (Swift-Morgan 2006, 355) Parents seemed to agree with this position, regarding the classroom as "beyond the purview of their knowledge and responsibilities" (Swift-Morgan 2006, 355). Nevertheless, many expressed a desire for greater links between the school and the community, with more meetings between staff, the PTA and other community members, and a role in financial decisions relating to improvement projects (domain 4).
Other studies discuss the role of the kebele (village association) in supporting schools. A kebele representative normally sits on the PTA, often chairing it. The kebele can use "persuasive and coercive methods" (Workneh 2012, 12) to encourage parents to send their children to school, sometimes imposing fines on parents who withdraw their children. In his case study of a woreda in Tigray, Micheal (2012) found that kebele representatives advised and oversaw the work of principals in some schools (domains 3 and 4), for example, requesting monthly reports on dropout statistics, and encouraging the appointment of female students to class monitor positions.
In summary, WEOs can act as a structural barrier to principals' and community members' agencyespecially regarding teacher recruitment, training, and material and financial resources. However, principals can play a role in encouraging community members to participate in infrastructural development and promoting student retention. Equally, there is some evidence of community members evaluating the work of principals.

CPD
As discussed earlier, CPD is a core element of SI. A recent empirical study (Fekede, Fiorucci, and Catarci 2013) used interviews with 18 teachers from three schools to explore teachers' perspectives on the nationally-mandated CPD programme. The authors found that teachers were extremelyeven passionatelydemotivated by working conditions, low salaries (also Sarton, Lalla-Maharaj, and Parsons 2009) and poor leadership. Teachers particularly criticised the "political appointment" of principals, who the respondents found to be unqualified and incapable. However, although the stated purpose of this study was to "explore and describe the nature of teachers' professional development … from teachers' perspectives" (Fekede, Fiorucci, and Catarci 2013, 5), this crucial task remains undone. The article is limited to teachers' views on the coercive, non-consultative introduction of the national CPD programme, and their grievances with working conditions. Further research on the programme would be useful, including data from in-school CPD sessions and an exploration of links between CPD and classroom practice.
What does SER tell us about schooling in Ethiopia? SER is concerned with identifying the variable effectiveness of schools within a sample, and the school factors responsible for the variation (e.g. Mortimore et al. 1988). SER makes use of input-process-output models, where inputs are student characteristics (e.g. gender, prior attainment) and outputs are academic attainments; an "effective school" is "one in which students progress more than might be expected on the basis of their intake characteristics" (Sammons et al. 1997, 160). By comparing more and less effective schools, researchers have identified "characteristics of effectiveness", which in the West include a participative approach to leadership, student responsibility and parental involvement (Sammons et al. 1995). A key rationale for SER, then, is that such knowledge can be used by teachers, school leaders and policy-makers for school and system-wide improvement efforts (MacGilchrist et al. 2004). While the differential effectiveness of schools in postindustrial societies has been calculated in the range 5-15% (MacBeath and Mortimore 2001), in developing countries the range is likely to be far greater (Harber and Davies 1997). In Ethiopia the urban/rural wealth divide is one reason for this (MoFED 2012); past studies have found that urban schools employ more teachers and have higher non-salary expenditure per student (World Bank 2005).
Quantitative studies of students' academic outcomes have been conducted through DfID-and USAID-sponsored bilateral projects (see Table 2).
All three studies focused on the early grades (1-3). The USAID studies measured reading ability only, while the Young Lives study additionally measured writing and numeracy; and all three collected academic outcome data through researcher-administered tests. DeStefano and Elaheebocus (2010) acknowledge that students' unfamiliarity with such tests may have adversely affected their scores, and an additional risk is that such tests fail to adequately reflect the taught curriculum (Rutter et al. 1979).

Correlates of attainment
Tassew, Jones, and Bekele (2005) investigated the relationship between students' academic attainment and characteristics at the level of the individual (e.g. gender, age), household (e.g. parental education, wealth) and school (proximity). The study offers macro-level insights into the outcomes of schooling for different categories of student, but is not SER per se since data are not aggregated at the school level, so we do not know how different students fare within individual schools. The team found that both household wealth and parental education were positively associated with student retention and progress through school, and poverty was found to be an all-round barrier to achievement in reading, writing and numeracy (see Table 3).
In the poorest households, 73% of surveyed students aged seven to eight could not write at all, compared to only 16% from the wealthier households. The distance between home and school was also negatively associated with students' academic outcomes. In follow-up discussions in the five research sites, the researchers (Tassew, Jones, and Bekele 2005) identified the following factors as undermining student and parent confidence in schools, potentially leading to drop-out at a later stage: lack of textbooks; excessively large class sizes; lack of separate latrines for male and female students; discrimination against poorer students; low teacher motivation; and violence from teachers and other students. Unfortunately, a similar list of factors promoting student retention was not provided.
Piper's (2010) USAID study measured the reading competency of over 13,000 students in 338 schools in eight regions. Based on students' ability to read aloud a graded 60-word passage in the regional language, Piper (2010) concluded that the percentage of Grade 2 students who were non-readers was 18% in Harari, 10% in Addis Ababa, and over 25% in SNNPR, Oromia, Tigray and Amhara regions. In urban areas girls outperformed boys, while in rural areas the reverse was true. This was primarily a study of student attainment rather than conventional SER; however, attention was given to the relationship between student attainment and some school-level factors, including: the availability of textbooks, the language of instruction and urban/rural location. Using multiple regression analysis Piper (2010) identified relationships between students' reading fluencymeasured in words per minute (wpm) loss or gainand factors at the individual, family and school level (see Table 4).
As in Tassew, Jones, and Bekele (2005) report, academic performance was closely associated with household characteristics: students living in urban areas with affluent, educated parents had the greatest reading fluency. At the school level, the provision of textbooks was positively associated with fluency, while grade repetition (in contravention of national policy guidelines in Grades 1-3 -MoE 2005) had a negative association of equal magnitude.

Variable school effectiveness
In rural Oromia, DeStefano and Elaheebocus (2010) investigated the comparative effectiveness of 24 primary schoolsnine government, and 15 so-called "community" schools, which were established by local communities with financial and training support from Save the Children. Over 90% of Ethiopia's 30,000 primary schools are government schools (MoE 2012b), including those receiving support from local communities and non-governmental organisations (NGOs) (MoE 2005; Jeilu 2009; UNESCO 2010c). In DeStefano and Elaheebocus' (2010) study, the "community schools" were designated such because they appointed teachers who did not have the minimum qualifications for working in the government sector; Save the Children sponsored these community school teachers through the certification process, and once all were qualified, the schools received government status.
Assessing the reading ability of 20 Grade 3 students at each of 24 schools, the team concluded that "every school in the sample could be called underperforming" (DeStefano and Elaheebocus 2010, 27). They found that the average reading fluency varied widely across schools, with community school students averagely reading at a rate of 19.0 wpm, compared to 14.2 wpm in government schools (see Table 5).
While community schools outperformed government schools overall within the sample, three schools of this category were also amongst the lowest scoring, including one where 75% of Grade 3 students were non-readers. That said, 20 is a small sample for judging the performance of schools whose enrolment ranges from 179 to 718 students, and in the absence of base-line measurements we cannot assess schools' relative effectiveness at promoting student attainment (e.g. Sammons et al. 1997).
One factor positively associated with student attainment was the length of time schools had been receiving NGO support for salaries, materials and equipment; however, no association was found between teacher training and attainment. The concept of opportunity to learn (OTL)an index of measures including instructional time, school closure and teacher/student absencehas become increasingly popular in the context of UK and US interventions in developing countries (e.g. Moore 2010; ICAI 2012). In DeStefano and Elaheebocus' (2010) study, the between-school variation in OTL is extremely interesting: school closure, and teacher/student absence reduced instructional time by up to 43% in government and 29% in community schools. Systematic lesson observations further indicated that students were "off-task" 64% of the time; and of especial interest is the finding that, despite the wide cross-school variation in OTL and student attainment, no strong relationship was found between these two variables. The authors claim: we may have stumbled upon a useful insight into the relationship between schools, instruction, and learning: There is no relationship. If we do not observe the teaching of reading, why should we expect to find a relationship between teaching and reading ability? (DeStefano and Elaheebocus 2010, 29) The claim that there is no relationship between schools, instruction and learning is not supported by the evidence of the study; on the contrary, Table 5 indicates that average reading fluency in the top-performing school is eight times that of the lowest-performing school, suggesting a strong relationship between student attainment and school attendance. Despite a limited focus on teaching practices (based on foreign-developed lesson observation criteria) and a single measure of attainment (with the limitations previously discussed), the authors identify "teaching deficiencies" (DeStefano and Elaheebocus 2010, 28), noting: Lessons were hauntingly similar across grades, subjects, and schools. A single approach to teaching seemed to predominate: demonstration at the blackboard followed by seatwork and copying.
Foreign researchers' characterisation of Ethiopian teaching methods as "haunting" is less than helpful; and from a technical point of view, it is not clear what they mean by "demonstration"does this include explanation, examples, modelling? The meaning of "seatwork" is also unspecifiedmight this not describe the larger part of classroom activities anywhere in the world? While systematic observation using foreign codings may reveal general patterns in teaching practice, it cannot provide the kinds of insight necessary for positive change in schools. By contrast, Poluha's (2004) ethnographic study of a school in Addis Ababa provides useful contextual details. Like DeStefano and Elaheebocus (2010) she found that teachers spent up to half the lesson copying notes onto the board, however they did so because there was a shortage of textbooks, and students needed the notes to pass their examinations. Poluha (2004) also found wide variation in the style and quality of teachers' explanations, and the extent to which teachers encouraged whole-class participation important nuances within the "single approach" identified by DeStefano and Elaheebocus (2010). Finally, while household characteristics are strongly associated with attainment (Tassew, Jones, and Bekele 2005;Piper 2010) these data were not considered as contributory factors in student performance, and so the claim that Ethiopian teaching practices are "deficient" (DeStefano and Elaheebocus 2010) is neither methodologically defensible, nor receptive to the multiple legitimate ways that teachers may support students' learning or the socio-cultural values implicated in different pedagogies (Guthrie 2011;Tabulawa 2013). This is not to discount the important contribution made by DeStefano and Elaheebocus' (2010) study in identifying the scope for improving teacher and student attendance. In summary, the relationships between students' backgrounds, early grade literacy and numeracy attainment and certain school characteristics have been investigated (Tassew, Jones, and Bekele 2005; Piper 2010; DeStefano and Elaheebocus 2010); outcomes in other subjects, later grades, values, engagement or social skills have not been researched. Within the school, the provision of textbooks was associated with increased reading fluency (see Table 4), but other factors such as the use of time and teacher training were not found to have a significant effect on student attainment (DeStefano and Elaheebocus 2010), although these claims can only be made tentatively due to the sample size in this study.

Implications
At first glance, the evidence reviewed earlier presents a fatalistic picture for school principals; while the SIP holds them responsible for leading school improvement, they lack the necessary autonomy to manage budgets, procure textbooks or determine important issues such as teacher recruitment and training. The WEO acts as a barrier to school-level decision-making, undermining local development efforts through partial, un-transparent judgements, and resisting contextual adaptations of national policies (Workneh 2012). Access to textbooks significantly increases student attainment (Piper 2010), yet as we have seen, principals do not control the procurement of these vital learning resources, and there is no evidence of them successfully holding WEOs to account for under-staffing or resourcing. Another fatalistic aspect of schooling is the power of home factors on student attainment; those born into educated urban families are the highest achievers (Piper 2010), while poverty has a negative effect on all aspects of literacy and numeracy development (Tassew, Jones, and Bekele 2005). In light of these issues, what can school leaders do to improve students' learning?
Fortunately, the research reviewed here suggests several paths by which principals may exert a positive influence. Firstly, they may improve student retention and attainment by implementing a policy of automatic promotion in the early grades, as recommended by the MoE (2005). Students who repeat a grade or miss school fall behind their peers (Piper 2010). Poor attendance (UNESCO 2010c), participation in work (MoE 2012b) and low test scores (Poluha 2004) all contribute to grade repetition. In the context of Ghana, Bosu et al. (2011) report on the action research of a principal who realised that the achievement of some Grade 6 boys was being negatively affected by regular absence on market days; she raised the issue with the students and their parents and organised catch-up classes for work they had missed, with the result that the students' attendance and attainment greatly improved, and all completed their primary schooling. While similar studies on school leadership in Ethiopia are lacking (Workneh 2012), evidence from this review suggests that attendance may be improved with support from the PTA, kebele and other local stakeholders (Micheal 2012;Workneh 2012;Yamada 2013); the principals' role here is to create opportunities for dialogue in which community members' contributions are valued (Swift-Morgan 2006).
There is a general lack of research into teaching and learning in Ethiopia, domain 1 of the SIP. While some studies present a deficit view of teaching methods, others (e.g. Poluha 2004) acknowledge strengths and variations in classroom practice. Evidence from post-industrial contexts (e.g. Bryk et al. 2010) highlights the importance of instructional leadership for improving student attainment: "the more leaders focus their relationships, their work, and their learning on the core business of teaching and learning, the greater their influence on student outcomes" (Robinson et al. 2008, 636). In Ethiopia scant attention has been paid to the work of principals. Workneh (2012) quotes a teacher as saying: The headteacher works very closely with the teachers, and the teachers work very smoothly with students. Teachers report any problem to the headteacher, and we discuss and solve the problems. We work as a team and in unity. The headteacher's office also closely follows up our activities. (Workneh 2012, 16) It would be useful to consider such accounts alongside observational data of school processes.
Despite the gaps identified, the current review of the literature suggests three principles which may underpin leadership for school improvement: Focusing on learningincreasing students' opportunities to learn by maximising attendance, access to textbooks and time available for learning; making learning the focus of professional capacity-building efforts (e.g. CPD). Promoting participation and respect for allfacilitating community-wide engagement with school evaluation and decision-making; countering violence and disengagement throughout the school and wider community. Harnessing resources to improve learning conditionsdeveloping resources and infrastructure with support from the community; holding WEOs accountable for providing textbooks and other learning resources.
I use the word "leadership" rather than "management", since rather than a managerial "maintenance of performance through supervision" (Dimmock 2002, 33), these principles entail a transformative relationship between principal and school community.
Having identified three pragmatic areas for principals' attention, it remains to be said that the positive potential of WEOs to support school improvement is, as yet, un-researched. What is the comparative attainment of students in different woredas? How do WEOs support school improvement efforts, and with what success? "Success" is an important word here, sadly missing from much research of schooling in Ethiopia. While improvement is a worthy aim for any school or system, local models of improvement are needed to show the way forward for schools and their communities.