Complexities and Contradictions in Policy Implementation: Lived Experiences of Three School Principals in South Africa

In this article, we discuss and highlight the complexities and contradictions that emerged from our analysis of data generated from three school principals from a rural community in South Africa. School principals play an intermediary role in the policy-practice sphere. Their role puts them in a difficult position, where they have multiple accountabilities. Using a template analytical framework, the findings suggest that their intermediary role in policy implementation is fraught with complexities, challenges, and contradictions. The findings further indicate that faced with conflicting demands of the government and that of their local contextual realities, principals make choices about which policy to implement according to government’s expectations and which ones not to. We conclude that an alternative understanding of policy implementation as a form of negotiation and compromise-seeking processes among stakeholders at the school level is desirable.


Introduction and Background
Through this article, we seek to highlight lived experiences of three secondary school principals about complexities and contradictions that pervade the discourse and policy implementation in South Africa. The discourse on complexities and contradictions in policy implementation is drawn largely from the South African context and also the international sphere. What informed our decision to draw from these contexts is the view that complexities and contradictions in policy implementation are a worldwide phenomenon (see Clegg, 2010;Fullan, 2011;Honig & Hatch, 2004;Tabak, 2006), and South Africa is not immune to this phenomenon. Complexities and contradictions we are referring to emerge from our analysis of stories narrated by the three participating school principals. This analysis suggests that the government's expectations of principals in terms of their role in policy implementation is fraught with contradictions. For instance, while the South African government has given schools and their governance structures (School Governing Bodies [SGBs]) powers to make policies, it simultaneously expects school principals to act like conveyor belts of its policies. We, therefore, argue that perhaps such contradictions emanate from complexities surrounding policy discourse. These contradictions could arise from both anecdotes and literature that suggests that the government in South Africa, through its Department of Basic Education (DBE), sometimes expects outcomes that are beyond the principals' abilities and mandates.
The discussion we present is based on an empirical study that involved three secondary school principals in a rural community south of Durban, South Africa. In highlighting these complexities and contradictions, we draw ideas from the experiences of these principals and from their perceptions of policy. Based on their policy conception, we sought to understand how they implemented policies in their respective schools. Drawing from literature, we know the South African government's position with regard to what policy is, but we do not know how principals understand it and how they react to it. We consider their understanding to be important as it may explain why they do the things in the way they do. We believe that such understandings may contribute to the theoretical understanding of policy implementation within the ever-growing field of policy studies, nationally and internationally.

Research Question
The main question that underpinned the study is: • • Research Question 1: What are the lived experiences of the three school principals regarding their role in policy implementation?
This question emanates from the policy background we present in this section. In South Africa, the DBE is responsible for secondary schools, under which the participating schools fall, and it entrusts principals with the responsibility of policy implementation at the school level. Principals are accounting officers in their schools (Mestry, 2013;Xaba & Ngubane, 2010), and that, ipso facto, makes them significant policy actors as well (Preston, Jakubiec, & Kooymans, 2013). Their position as principals also puts them in an advantaged position to influence the manner in which policy is interpreted and translated into action (Bayeni, 2016). Because principals are a point of access into the schools, they have inherent power to control and direct the manner in which various policies move from government to the schools and teachers for implementation (Bayeni, 2012). We argue in this article that such power may also pose a threat to the DBE in terms of its expectation that the policies that are implemented at school level may not mirror its policy intentions (Smit, 2005) but those of the principals. However, such an expectation by the DBE ignores the view that policy messages enshrined in the policy text are not always translated into actions in the schools as implementation sites (Bayeni, 2016;Coburn, 2006;Thorius & Maxcy, 2014). Actually, policy-practice at the different implementation sites is diverse. This is due to the fact that to respond to diversity and local demands, principals as policy implementers have to engage in policy interpretation and mediation before implementing it (Braun, Maguire, & Ball, 2010;Levinson, Sutton, & Winstead, 2009;Thorius & Maxcy, 2014). Such a view recognizes that by virtue of their position, principals, as policy implementers, are powerful actors in the policy process (Hodgson, Edwards, & Gregerson, 2007). This suggests that the move from policy formulation to implementation is neither linear nor unproblematic. Therefore, to expect that policy will be implemented as contemplated at the policy formulation level may be unreasonable. Nevertheless, complexities we are highlighting in this article seem to show that there is a disconnect between policy makers and policy implementers.
In the context of South Africa, particularly since she attained democracy in 1994, there has been a plethora of education policies that were formulated, aimed at addressing, inter alia, apartheid inequalities. Overall, such policies were implemented to varying degrees of success (Heimans, 2012;Palmer & Rangel, 2011;Teelken, 2008;White-Smith & White, 2009). In the same vein, there is unanimity among scholars in South Africa that policy intentions are not mirrored in policy-practice (Lungu, 2001;Smit, 2005). The disconnect between education policy as text and policy as practice is not a new phenomenon, but it has been acknowledged by several scholars in various parts of the world (Honig, 2006;Levinson et al., 2009;McLaughlin, 2006;Spillane, Gomez, & Mesler, 2009). Studies conducted locally and internationally suggest that the translation of policy into practice poses a major challenge to successful implementation (Honig, 2006;McLaughlin, 2006;Spillane et al., 2009). In our endeavor to conceptualize the policy formulationimplementation phenomenon, we draw from various conceptions of policy and other contextual realities in South Africa. Starting with conception of policy from the government's perspective, there are three dimensions to it. First, policy is understood as a tool to designate the different behaviors of some actors in dealing with a problem or a matter of concern (Anderson, 2003). Second, policy can be conceived as a statement of intent (Ball, 2007). Third, policy can also be viewed as "whatever governments choose to do or not to do" (Anderson, 2003, p. 21). The three dimensions suggest that the government expects certain behaviors of people at local levels. While the DBE regards principals as accountable officers at school level, it also acknowledges that there are challenges regarding the skills base of principals. It is unclear as to whether the skills deficit is related to the manner in which principals conceptualize policy or their role in it. Nonetheless, the government believes that it has to act and rectify the situation.
To address these challenges, the DBE employed three strategies. The first strategy entailed the DBE organizing short-term development workshops for principals when new policies have been formulated. With regard to the second strategy, the DBE designed a formal leadership and management qualification that was of a 2-year duration. For instance, in 2007, the DBE designed and piloted the Advanced Certificate in Education (ACE) for school principals as an example of the 2-year qualification we are referring to here. That 2-year program focused on conceptual and practical issues relating to school leadership and management in the South African context. Finally, the DBE designed a Handbook that guides principals about how to interact with policy issues. The DBE's belief in using the three strategies was that after acquiring the requisite skills, principals would be able to implement its policies as intended. However, empirical evidence (Dieltiens, 2005;Grant-Lewis, Naidoo & Weber, 2000;Naidu, Joubert, Mestry, Mosoge, & Ngcobo, 2008) suggests that the opposite is true, owing to complexities involved in policy implementation. For instance, some sizable number of principals continue to fail to implement policy with fidelity. For instance, what some principals actually do in practice continues to contradict the DBE's expectations. This suggests that principals mediate and selectively choose some aspects of policy and implement them by focusing on what works for them at the school level (Bayeni, 2016;Bhengu, 2012). This yields more complexities and contradictions. In short, it has become unclear as to what constitutes effective policy implementation. For instance, is it about the extent to which principals comply with the expectations of policy formulators at the national level of government? Is it about the extent to which principals, as implementers, select certain parts of policy and implement them to the exclusion of those parts they deem to be unworkable? Or, is it about how principals take conscious decisions to adjust the policy script to suit their needs at the school level? These questions necessitate further scrutiny into policy implementation, thus exposing those contradictions between the DBE's expectations and those practices that principals found working for them and their schools.

Policy Landscape: South African and International Perspectives
In the past two decades, there has been a growing discourse among policy scholars in South Africa that highlights a failure of policy implementation (De Clercq, 2001;Ntombela, 2006;Smit, 2005). When a democratic government, under the leadership of Nelson Mandela, came into being in 1994, the then Department of Education (DOE) instituted a plethora of education policy initiatives. Given the country's political history of colonialism and apartheid, almost all policies developed during that period aimed at transforming leadership, governance, and management practices at the school level (Bhengu, 2005), to align them with the democratic dispensation. Another policy perspective from some recent publications suggests that policies are not meant for implementation but for enactment (Braun, Ball, & Maguire, 2011;Braun et al., 2010;Mulcahy, 2015;Singh, Thomas, & Harris, 2013). The arguments that these scholars advance is that citizens had to get a sense that the country had broken ranks with the past and had begun to embark on a new route underpinned by values of freedom, democracy, redress, and respect for human rights including their right to participate on issues affecting their lives. Our review of literature elicits three dimensions that dominate policy-practice discourse. First, there is the notion of policy as an authority prescript; second, policy is viewed as a prescriptive process; and third, policy is viewed as an interactive discursive practice. We present the three dimensions below.

Policy as an Authority Prescript
Policy as an authority prescript and as a prescription are closely related but separate; hence, we discuss them separately. The notion of regarding policy as an authority prescript presents policy as a linear, straightforward process with clear definite outcomes. The government expects such outcomes to be achieved irrespective of local dimensions and complexities. The authority prescriptive dimension of policy seems to ignore the view that local conditions where policies are implemented have their own complexities and challenges that have to be considered if successful implementation has to occur. For instance, Levinson et al. (2009) argue that technical rational analysis of policy views policy as extending the interests of those who wield power (Park, Daly, & Guerra, 2012). That is why those who are in power are inclined to influence policy direction. This approach focuses on what Levinson et al. (2009) call the modalities of domination within the policy-practice terrain. Such a view has its own weaknesses. For instance, although powerful groups such as policy makers may decide on the content of a policy text, they may not successfully dictate what happens at the local level because they fall outside of the circumscribed contours where policy is implemented. In the school context where policy is translated into practice, there are interacting human and nonhuman factors that shape policypractice (Honig, 2006;McLaughlin, 2006;Spillane et al., 2009). Policy makers cannot control these factors, at least in any direct manner.
Furthermore, in the school environment, there is unequal share of power between principals, Heads of Department, and teachers. Here, principals tend to wield more power compared with teaching staff within the schools. This is more likely to happen because closely coupled with the principal's responsibility is the approval or disapproval of the teachers' behaviors and the day-to-day operations in the school. Endowed with such power and the considerable allegiance they command from teachers, principals can control, change and direct, or ignore policies from the DBE (Bayeni, 2012;Smit, 2005). This scenario suggests the presence of complexities about power (perceived or real) that policy makers believe they have while ignoring the fact that principals as local actors are also endowed with power that they can use to favor their own interests and/or that of the local communities. These are some of the contradictions of policy.

Policy as a Prescriptive Process
Although policy as an authority prescript may be the policy maker's desirable thought that acknowledges the power of the policy maker, the implementation of such policy as a prescriptive process proves to be challenging and complex (Braun et al., 2010). In the debate of a prescriptive process of policy, the policy as text is assumed to have the ability to constrain behavioral practices of policy implementers in ways that conform to the dictates of policy formulators (Coburn, 2006). Of course, we argue in this article that such a view is flawed, particularly if we consider the dynamics of policy-practice complexities. A similar misconception by policy makers is that of assuming that policy process is one directional and straightforward. We argue elsewhere in this article that the policy process involves a to and fro, bidirectional and sometimes cyclic movement. Therefore, the notion of policy as a prescriptive process at ground level may remain wishful thinking. Rather, it is an interactive, everchanging process that is always in the "state of becoming" (Webb & Gulson, 2012).

Policy as an Interactive Discursive Practice
Policy research acknowledges several views that suggest that policy does not get transmitted smoothly from the DBE to the schools (Apple, 2002;Morris & Scott, 2003;Young & Lewis, 2015). In fact, policy implementation is not a straightforward affair (Berkhout & Wielemans, 1999;Young & Lewis, 2015) but is contested, challenged, and reinterpreted on the basis of the context and embedded experiences. In the context of education, policy passes through a number of various levels in which there are people who hold different view of policy from those who make it. At each level, policy is highly mediated by the politicking and contextual dynamics. Subjected to such dynamics, policy is constantly "re-contextualised" so that its message fits "the discursive practice" (Apple, 2004, p. 4) at play.
Various scholars (Dunn, 1981;Majone, 1989) regard the complexity of policy implementation as normatively simple just because it is found and is applied in several disciplines, although it is conceptualized differently in different contexts. A conceptual contradiction exists here in that due to its ubiquitous nature, there is no consensus on what it means and describes (Anderson, 2003;Ball, 2007;Dye, 2005;Wedel, Shore, Feldman, & Lathrop, 2005). Against this background, it is critical to note that policy is not a linear process (Coburn, 2006). Instead, it is a product of a mediated and a negotiated interactive process, among the significant elements such as policy makers, the context, and policy implementers, all of whom do not see policy in the same light and hence co-influence one another. Therefore, it is important to understand the dynamic and contested nature of policy operations at implementation sites. Principals being the heads of the schools, it is of particular importance to understand what they do when interacting with policies, which leads to a change of policy intentions as contemplated by DBE (policy makers). It is important to develop theoretical explanations of how policy created by the DBE is mediated and/or even changed by the principals as ground-level bureaucrats in the implementation process.

Theoretical Framework Underpinning This Study
To frame this study, we draw from two theoretical constructs, namely, technical rational model (TRM) and sense-making theory (SMT). These constructs complement each other in terms of providing insights about understanding policy implementation. When used in combination, these theoretical constructs give a balanced perspective about what happens, both at policy formulation and at implementation spheres. In short, TRM, on one hand, explains how policy makers formulate policy, while the SMT, on the other hand, enables us to understand how and why policy is constructed and mediated as it is being implemented. Preceding discussions have shown that policy makers have their own expectations and assumingly wield power to direct what happens at the implementation level. However, school principals, too, wield power at the school level as accountable officers, and are thus custodians of practice. Because of that reality, it is critical that we understand both perspectives to policy implementation.
Policy makers usually operate with the assumptions that policies they make are instrumental in addressing many problems at the implementation level by changing or controlling the behavior of the concerned citizens (Apple, 2004;Collinson, 2005;Levinson et al., 2009). Furthermore, policy makers believe that they are better positioned to diagnose the problem using what terms prognostic framing. Imbedded in these assumptions is the belief that policy will move in a smooth, orderly, and linear way from formulation to implementation areas. We argue that such a notion of policy-practice as a straightforward mapping process is a myth given the existing empirical evidence that negates this notion (Gilmore & Murphy, 1991;Kari, 1998). We further argue that it is short-sightedness to hold the view that policy implementers should be regarded as cogs of the government machinery. Such a view is contradictory in the context of decentralization and autonomy of local structures as is the case in South Africa.
SMT encapsulates a dominant view that accentuates continuous adaptation and reconstruction of policy as it moves from formulation to implementation spheres (Gilmore & Murphy, 1991). SMT provides theoretical insights about the process by which people create an understanding of policy so that they can act in an informed manner (Savolainen, 1992). Furthermore, SMT can be understood as an activity of getting an understanding of something and follow it with an attribution of meaning to it (Kari, 1998). There is a notion that, normally, individuals can proceed smoothly with their own activities as long as there is no disruption. However, as soon as disruption occurs, which Dervin (1983) refers to as a discontinuity, individuals engage in maneuvering as an attempt to deal with discontinuity. By so doing, they are able to bridge the gap so that they can continue with the activities. Bridging the gap entails exploring other means of dealing with the challenges in ways that enable them to survive in the environment (Dervin, 1983). Therefore, in the context of the school, principals have to bridge the gap between policy makers' expectations and what principals need to do to survive in the local environment. Principals are confronted with many challenges in the form of new policy dictates or discontinuity (Dervin, 1983), and in their response, they bridge the gap by reconstructing policy messages in their schools. Actually, what principals do can be seen as contributing to the understanding of the relationship between policy as espoused by policy formulators and their policy implementation within their schools. In addition, SMT enables principals to engage with policy messages in ways that encourage them to challenge their own assumptions and continue to improve their practice over time (Coburn, 2001).

Method
This is a qualitative case study of three principals who were purposively selected based on the reputation of their schools for good learner academic achievement. The study is located within an interpretivism paradigm. Such a paradigm is appropriate for an inquiry where views and experiences are interpreted from the perspectives of those who are studied (Cohen, Manion, & Morrison, 2011). We adopted an ethnographic methodology because of its appropriateness for research that seeks to understand socio-cultural lives of the people (LeCompte & Schensul, 2010). In this study, we sought to understand how school principals interacted with policy in the process of implementing them. To elicit indepth understanding of how the three principals interacted with policy process required that we had an extended stay in the community (Wolcott, 1995). This was based on our attempt, not only to understand how principals played their role in the policy implementation process but also to understand how their practices are consistent with their understanding. To enhance opportunities of soliciting a comprehensive picture about their lived experiences, we had a prolonged stay in the field interacting with them in their natural setting (McDonald, 2005).
Like many qualitative studies, we employed multiple data-generation techniques, and these are semi-structured individual interviews, personal observations, documents' reviews and shadowing (Schensul & LeCompte, 2013). We chose semi-structured interviews for their flexibility and for enabling participants to freely express their feelings and meaning (Miles & Huberman, 1994). We supplemented semi-structured interviews with observations made prior to and during the principals' shadowing process. We shadowed principals from the moment they began their working day until they departed for home. We spent an average of 15 hr observing principals as they carried out their daily routines. According to many sociologists and anthropologists (Czarniawska-Joerges & Liber, 2007;Marcus & Michael, 1986;McDonald, 2005), shadowing includes techniques such as participant or nonparticipant observations. Shadowing is useful in revealing subtleties of perspectives and purposes shaping those actions in the real-time context of an organization (McDonald, 2005).
The following section outlines the research activity on the research sites. The introduction of a study and establishing rapport with potential participants is critical for the credibility of the results of a study such as this one (Schensul & LeCompte, 2013). After we had introduced the study, its purpose, and all protocols had been clarified, we began our formal discussions about principals' lived experiences regarding their interactions with policy implementation processes. Protocols included seeking permission for interviews and observations, and guaranteeing confidentiality and anonymity to the participants (Schensul & LeCompte, 2013). Issues of autonomy and rights to participate and to withdraw at any time were clarified during this phase (McDonald, 2005). The data generation process took the form of series of conversations that started with the introduction of the study, followed by an explanation of the protocols. Formal semi-structured interviews followed in which we solicited principals' experiences of mediating and interacting with policy. These discussions were followed-up with personal observations and school documents review. Documents review entailed looking at enrolments and schedules of examination results for the past 6 years. We also looked at government's policy documents kept in the schools, and the files containing school-generated policy documents. It was important for us to know whether schools kept the records of policy documents since such policy documents give direction to the principal. We were also interested in understanding the extent to which the content of school-generated policies were consistent with government's policy documents. Each principal was shadowed. In the process of shadowing, we would follow the principals wherever they went. We would take down notes regarding the manner they communicated with the staff. We also noted the process followed in decision making.
On average, each interview lasted between 45 min and 1 hr. Before the interviews commenced, we assured participants of complete confidentiality. We also ensured that before we began discussions, we obtained permission from each participant to audio-record them. All audio interviews were transcribed verbatim before we did the analysis. In analyzing data of this study, we used the template analysis method (Randall, Cox, & Griffiths, 2007, p. 1190. This method can be described as "occupying a position between content analysis where codes are all predetermined, and grounded theory where there is no a priori definition of codes." Template analysis allows for both deductive and inductive elements. In terms of this analysis method, we provided a set of a priori codes, which we then expanded on as additional themes emerged from the data. We then summarized and organized the themes in a meaningful way (Crabtree & Miller, 1992).

Findings and Discussion
We gave the three participants the following names: Nkebeni, Masogesi, and Nathi. Their schools were renowned for producing good National Senior Certificate results for a decade. One of the assumptions about their respective schools' success was that their policy implementation was effective. Effective implementation was strongly associated with the principals' leadership and ability to translate policy into practice. Because of these assumptions, we thought it useful to provide an overview of their profiles. In that way, we ensured that the findings are not separated from the person on whom the findings are based. With this in mind, we provide a profile of each principal before discussing the findings.
Nkebeni, the principal of School A, is a well-experienced leader. He has 11 years of experience as principal, 4 as deputy principal, and 19 as an educator. His school is surrounded by an informal settlement. Residents from these settlements were alleged to have vandalized the school by removing doors and windows from schools to use in building their own shack dwellings. On his appointment as principal, Nkebeni did not only confront these shack dwellers about vandalism they had allegedly committed, but at the same time, he encouraged parents to become an integral part of the school by rendering their services. Nkebeni turned the school around. Grade 12 results have improved from an average of 40% pass rate to that of 95%.
Masogesi is the principal of School B. He had only 2 years' experience as a principal. Being a white principal in a multiracial school, with 90 staff members and more than 95% African learners, he had learned to contend with a new, small governing body, which consisted of African parents. Having learners coming from affluent backgrounds with a good command of English and also able to afford high school fees, the principal did not have to make any considerable financial adjustment. With adequate resources, the principal was able to procure most of the necessary resources including employing additional staff to relieve the workload of teachers. He tended to challenge most of government policies that were handed down to the school. Masogesi argued strongly that government policies were recorded in school documents and strictly followed. However, the actual policypractice suggested a different reality. The average pass in the National Senior Certificate stood between 85% and 100%.
Nathi is the principal of School C and has a teaching experience of 35 years. He started at this school as a teacher and was promoted to the position of Head of Department (HOD) and, thereafter, to principal of the school. The school is located in an area where traditional faction fights were rife. Learners belonged to either of the fighting factions. Sometimes, such fights would spill over to the school premises. He managed to bring order and peace in the school by bringing the fighting groups to the negotiations table. The dispute was resolved, and reconciliation process followed. School C was once regarded highly by the community and the DBE because of its good results and discipline. At the time of his appointment, the school had lost its good image caused partly by traditional faction fights. Consequently, the pass rate had plummeted to between 30% and 50%. Besides resolving faction fights, he introduced a culture of teaching and learning, which included the introduction of study periods before and after school. Since his arrival, results consistently oscillated between 80% and 100%. This made his school one of the best in the country.
From the data analysis, three main themes emerged regarding how and why principals implemented policies the way they did. These themes are as follows: • • Principals' understanding of their intermediary role in the policy-practice. • • Defective policy implementation monitoring encourages hollow compliance. • • Power struggle between the DBE and civil society stakeholders.

a. Principals' understanding of their intermediary role in the policy-practice sphere
We started by asking each of the three principals what they understood their roles to be in the policy process. The data elicited suggest unanimity of views in that they regarded themselves as strategically positioned to act as intermediaries between policy makers and policy end users. Actually, they stand between the government and local stakeholders.
In that regard, Honig (2004) argues that principals as intermediaries occupy a contested space of policy making and policy implementation. Our aim was to understand principals' lived experiences, especially considering that policy implementation is not easy and is fraught with complexities. For example, as accountable officers at school level, principals, on one hand, are expected to carry out the mandate of the DBE. In some ways, they are equal to the teachers in terms of them being employees of the DBE, while on the other hand, principals have to serve teachers and other stakeholders at the local level. In that way, they are accountable to them as well, and they have to ensure that they keep them satisfied. That is one way that complexities surrounding the principal's job manifest themselves. Complexities surrounding multiple accountabilities for principals contribute to different ways in which they play their intermediary role and how they conceptualize policy. Returning to the first issue (principals' intermediary role), principals needed to have the ability to navigate and mediate the DBE's mandates and teachers' demands in a manner that satisfied both parties. That may be the reason why despite their declared common understanding of their intermediary role, the manner in which they played such a role was different. This is how Masogesi viewed his role as intermediary: Serving two masters; the DBE and the teachers at the same time is not easy. It is difficult to satisfy both. Therefore, we decide that we stay in the middle, partly satisfying the DBE and the teachers. As we do that, we make changes from what DOE tells us to do in the form of policy. Teachers feel DBE policies are demanding too much of their time beyond their job description.
The views expressed above were also shared by Nkebeni, when he highlighted the futility of expecting policy to be applied as it is, without making any adjustments. This is what he said: The government policies are stupid because they don't consider the difficult realities of managing schools. Sitting at the centre between school practices and DBE policies, I find it difficult to follow policy as it is. I have to make adjustments so that I can make peace with school realities.
There are policy areas where the policy is not being followed as expected. One of the reasons for this is that there is silence regarding some of the critical issues particularly in the learner discipline policy. For instance, in the discipline policy, there is no mention of cleaning the school premises as part of punishment for minor misconduct. Therefore, when it comes to disciplining learners, Nkebeni made it clear that for minor case of misconduct, he "implements" the policy his own way. This is what he had to say: For cases of minor misconduct, I punish learners by giving them light manual work like cleaning the yard. For serious misdemeanours I call their parents and make them aware of the problem; they sign against the recorded misconduct, and I further inform them that if the learner repeated the same misconduct, he or she will be expelled. What can I do? This is my way of applying policy.
Another example of ignoring national policy can be seen where corporal punishment continues to be meted out to the learners despite it having been declared illegal 20 years into the democratic dispensation. Masogesi, the principal of School B, argued that alternatives to corporal punishment did not work for him. This is what he had to say: Alternatives to Corporal Punishment Policy requires teachers to punish learners by, for instance, detaining them. Teachers complain that if they detain learners, they are detained as well because they have to supervise and monitor learners during detention. For minor offences, I do not detain learners as policy requires, I just give them few lashes and that is done.
The mention of lashes clearly indicates that this principal uses corporal punishment. The views expressed by Nkebeni and Masogesi, respectively, were shared by Nathi, the principal of School C. Nathi said that he implemented the drug policy his own way. He argued that the expulsion of learners, as an approach of dealing with misdemeanors, was almost impossible. However, evidence from document reviews indicates that expulsion was one of the measures that the principals used. Justifying the use of expulsion of learners, Nathi echoes, But for serious offences, I get rid of the child [expel] in a smart way. Let me take one example of a learner who brought drugs into the school. This is a very serious offence. Therefore, I called his parents into the school. When they came, I politely advised them to take their child away from the school to another school. I did not record any misbehavior. For a learner to change a school voluntarily without a bad record in his or her file, instead of expulsion, will mean [that] the child will get easily admitted into another school without bad records.
The second view, as advanced by Masogesi, regards policy as just a guide to practice, and purports that it should respond to local needs of schools. Therefore, as part of principals' intermediary role, we also sought to understand how they conceptualized policy. Two dominant views emerged from the analysis. One view regarded policy as prescription and, thus, is sacrosanct. Masogesi argued that "principals as intermediaries are just conveyor belts who should not question anything." Such a view suggests that principals are not supposed to change anything when the policy gets into the school. However, what we have noted from the interviews, the reviews of documents, and observations is that principals made a major shift from the original policy. This, in fact, contradicts the notion of portraying principals as conveyor belt.
Drawing on the theory of educational policy implementation, it can be argued that at different tiers of the educational system, policy is mediated and engaged in by context-embedded actors (Hamann & Lane, 2004;Heimans, 2012;Honig, 2001;Levinson et al., 2009), for example, school principals. Mediation entails the intertwined processes of negotiation, interpretation and sense making of policy (Coburn, 2003;Hamann & Lane, 2004). Honig (2001, p. 428) similarly argues that "when faced with policy messages [from outside], individuals engage in constructivist process." As this happens, principals adapt policy to their contexts. It transpired in the study that principals, on receiving policy, mediated and adapted it to suit their contexts.
In shadowing principals, we ascertained that principals played an intermediary role between the DBE and the teachers. This became clear when they communicated with DBE officials telephonically and also when they addressed teachers. For instance, while shadowing Nkebeni, a DBE official called, reminding the principal to submit some documents to the DBE offices by the end of the business day. Nkebeni instantly made a promise to do that. He made the promise without consultation with the teachers who were expected to do the spade work of collating and making them ready for submission to the DBE. Immediately after the call with the DBE official, he called a meeting of the School Management Team, instructing the HODs to inform teachers in their departments to fill in the required documents. The promise he made to the DBE and instruction he gave the School Management Team displayed the intermediary role where the principal was simultaneously satisfying the DBE needs and instructing his management team to get the teachers to act on the DBE's requirements. Such actions indicate the accounting responsibilities of principals. Clearly, he or she is at the center of the two establishments, that is, the DBE and the school.
Another incident is where Masogesi received a call from the DBE office asking him to urgently report to its offices. The principal left everything he was doing and dashed to the DBE offices. A further incident is where Nathi acted on behalf of DBE by calming the teacher who was angry with the DBE for failing to renew his contract. In this regard, Nathi had to present himself as the DBE official and talked with the teacher convincingly as a DBE official ensuring his contract had received urgent attention and would be sorted out as soon as possible. The teacher was happy when a falsified district official reassured him that he would ensure that the contract is renewed.

b. Defective policy implementation monitoring encourages hollow compliance
Data elicited from the interviews indicated that the mentoring of policy implementation was not effective. According to the participants, it was more important to just fill in forms for compliance. Their view was that when they submit forms to the DOE, circuit managers (CMs) are satisfied that schools complied with their instruction. CMs were less interested in knowing whether policies were actually implemented or not. These officials hardly visited schools to provide support and monitor policy implementation. The lack of monitoring highlights a contradiction in terms of what policy implementation entails. Dominant narratives from the principals suggested that CMs were more interested in ensuring that schools submitted reports rather than in knowing how schools accounted for what they did. This is what Nkebeni had to say in that regard: The officials are more concerned with the paper work. They barely visit the schools, and when they do, they bring their ticking book. They want us to submit records of what the school is doing. And they just tick in their books without checking the validity of the entries made.
Masogesi echoed similar sentiment when he said, The principals in our districts do what they believe will help their schools achieve good results. Department's officials do not come to the schools because they are afraid of the teacher unions. So, as principals, we do what we think will benefit our schools.
It is evident that there is lack of monitoring and that may have contributed to what we may call policy filtering where schools make choices about which policies to observe and which ones not to. The lack of monitoring has contributed to a situation where principals do not even see the need to keep records of what they do at school. Nathi expressed this view this way: Not all things can be recorded. Why must we record everything, what for and for whom? Nobody checks these things.
We confirmed his utterances through reviews of policy documents kept in the schools. Such reviews suggested a discrepancy between what principals claimed to be doing and what they recorded. For instance, we found that what was documented in the school policy corresponded with findings from the interviews. We then realized that mere reliance on school policy records to gain insights about the schools' operations was inadequate. We could have been lulled into believing that schools strictly adhered to the DBE policies. The views that Nathi expressed demonstrated a worrying contradiction in terms of multiple accountability imperatives. He seemed to be fixated on only accounting to CMs and on keeping records as compliance measure. When shadowing principals, we noted that there were documents on their table from the DBE. Some had forms to be completed and returned. One of them was the Integrated Quality Management Systems policy, and it required principals to fill in forms by inserting evaluation scores on teachers' performance. That could only happen after teachers had been observed and assessed while teaching. However, that had not happened with these principals. Nevertheless, they just filled in performance scores without observing the teachers. When we asked about that, the principal confidently explained that CMs did not monitor compliance with policy. The principal also highlighted a heavy workload, saying that he did not have time to do justice to all paperwork requirements.

c. Power struggle between the DBE and civil society stakeholders
The advent of a democratic dispensation in South Africa came with new realities and dynamics, some of which had not been anticipated. For instance, the devolution of some powers regarding decision making on governance issues to school level is one of the gains for local communities. For the first time, communities were legally enabled, through the South African Schools Act, No. 84 of 1996 (Republic of South Africa [RSA], 1996), to make decisions on substantive issues such as admission of learners and choice of teachers they prefer to be employed in their schools. However, this study has highlighted some complexities and contradictions in the implementation of this Act. While in some instances, it emerged that principals looked up to officials of the department to exercise close monitoring activities, there were also areas where the local community stakeholders claimed and defended their powers against what they perceived to be DBE interference. This created tensions between these centers of power, and struggle for power persisted. The struggle for power ensued where interests and preferences of the school communities were apparently challenged by the DBE. One such instance was where community structures took the provincial DOE to court over what they perceived to be illegal admission of learners. Some schools found the DBE's admissions policy to be unworkable, and they instituted their own versions of this policy. In many instances, the courts have ruled in favor of the SGBs' decisions. This is indicative of the view that power struggles between the DBE and civil societal organizations exist. School principals are caught in the cross fire, and in some instances, they are part of the disputes as they form part of SGB, while they remain the employees of the provincial DOE.
In the context of this study, Nathi argued that SGBs of many affluent schools, invariably, former Model C schools, were affiliates of the School Governing Body Foundation (SGBF). The SGBF, through its strong legal and financial base, was able to contest the provincial and national DBE policies. In cases where the principals were still unsatisfied with the outcome, the SGBF would, on the principals' behalf, challenge the DBE in court. In many instances, the SGBF won such cases. Such successful judgments emboldened principals while it made the DBE extra cautious when attempting to force its policies onto affluent schools affiliated to the SGBF. Such experiences encouraged principals to disregard DBE policy. In some ways, SGBF gave principals a mandate by approving or disapproving the implementation of DBE policy. Taking advantage of SGBF power, some school principals ignored or made adjustments to the DBE policies.
In case of any department policy that is hard to implement, I report it to the SGBF. Together with other principals who are members of SGBF, we would ask the Foundation to take the issue up with the DBE. If no agreement is reached, Foundation takes legal steps. (Nathi) Sentiments expressed above by Nathi were also shared by Masogesi, the principal of School B, when he highlighted some of the policies whose implementation ran contrary to the DBE's prescriptions. He said, The KwaZulu-Natal Department of Education made a decision that admission of learners to public schools would no longer be based on feeder schools model. It was proclaimed that learners must be admitted on the first come first serve basis irrespective of their place of residence. Some principals were aggrieved by this law and took the issue to court through SGBF. Because of SGBF action, the MEC for Education in KwaZulu-Natal issued a circular to all schools withdrawing the earlier decision. The circular stated that schools should continue to admit learners based on their school policies.
What is described above demonstrates the power that structures in society have and that they can use their legal and financial strength to challenge even the government's department and win cases. Through such cases involving the SGBF, it has become clear that deciding how policy is implemented cannot and does not rest with policy makers only, but other players such as school principals and School Management Teams have capacities to project their power to protect their interest in this highly contested terrain. Such powers have a strong influence in the way that principals implemented policies. Policy implementation process at school level is fraught with such competing discursive practices since policies from outside into the school terrain are colored with DBE agenda with little consideration of possibility of changes (Liaisidou, 2011) from the schools. As DBE policies were being implemented, a new frame of policy responsive to the school contextual demands emerged, which overshadowed or superseded DBE policy. During the shadowing process, we asked principals about specific policies that gave them sleepless nights. All participants mentioned the learner admissions policy and the Annual National Assessment (ANA) policy. We have mentioned before that it seems as if the principals defeated the DBE in the court of law through the SGBF. With regard to ANA policy, principals defeated DBE through collaboration between themselves and teacher unions. Such defeats forced the DBE to discontinue ANA policy altogether. The defeats presented here demonstrate the high level of complexity and contestation in the policyimplementation process.

Conclusion
One of the key conclusions to make is that principals in this study proclaimed policy to be prescription and sacrosanct, and therefore, they must be observed. However, they did not display leadership practices that were consistent with their proclaimed beliefs about policy implementation. We noted, for instance, that all three principals, irrespective of their statements about policy implementation, implemented policies their own way, which in some instances contradicted policy intentions. It was intriguing to note that the actual principals' practice of policy was not recorded in any school documents. Table 1 summarizes the policy situation in the three schools. It captures what we found when we reviewed files kept in the schools, contrasting their content with what principals had told us in relation to the availability of government's policy documents, the availability of school-based policy documents and leadership practices that prevailed in the schools.
However, Nathi did not admit that he changed policy. Instead, he presented himself as a passive policy carrier who transmitted the DBE policy without changing it. One may argue that such principal's presentation was a tactful account that conveniently depicted him as a law-abiding employee, by implementing DBE policy as it was expected of him. Contrary to his verbal presentations, it emerged that the very principal changed policy when he implemented it. Although he verbally portrayed himself as passive policy carrier, in practice, he acted as active policy adapter.
The focus in this study was on policy implementation through the lens of school principals. The manner in which the school principals changed DBE policy can be conceived, at a superficial level, as inability of implementing policy. However, at the abstract level, it can be associated with policy enactment. Policy enactment refers to "an understanding that policies are interpreted and translated by diverse policy actors as they engage in making meaning of official texts for specific contexts and practices" (Singh et al., 2013 p. 466). The findings showed that all principals engaged in a "creative process of making sense of policy and putting that sense into action" (Mulcahy, 2015, p. 507).

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.