Improving accountability in education: the importance of structured democratic voice

Accountability, a cornerstone of contemporary education policy, is increasingly characterized by external monitoring and an emphasis on outcomes or results. Largely absent in discussions of accountability are the voices of stakeholders who work, learn, and teach in schools and other educational institutions. This article highlights the critical importance of structured democratic voice. This process involves enabling diverse education stakeholders to make use of organized opportunities to articulate their views, especially in discussions of planning and evaluation, in ways that their concerns are heard and valued. Using illustrative examples at the regional, national, and local level, this article discusses how structured democratic voice can help re-imagine approaches to accountability while strengthening the enabling environment, increasing trust in the system, and improving policy ownership. Notwithstanding the challenges of building local capacity and sustaining political commitment, this article highlights the important consequences of implementing a process of structured democratic voice: in particular, sustaining educational reforms over time and meeting ambitious collective goals in education.


Introduction
Accountability is a cornerstone of contemporary education policy. Some consider it a 'virtuous practice' (MacIntyre 1984); others view it as indistinguishable from 'good governance' and transparency (Bovens 2006;Dubnick 2014;Gorur 2017;Stensaker and Harvey 2011). Widespread interest in accountability is reflected in research, where the use of the term increased tenfold in studies published between 1965 and 2000 (Dubnick 2014). Accountability is also a buzzword in global education policy. For example, the Education 2030 Framework for Action, passed by UNESCO member states in November 2015, refers to accountability-in the forms of 'accountability,' 'accountable,' or 'to account for'more than 20 times (UNESCO 2016a). As Volante (2007) remarks: "[by] the early twenty first century it was clear that the application of accountability systems was one of the most powerful trends in education policy" (p. 7).
The emergence and spread of accountability are attributable, in part, to five political and social trends (UNESCO 2017). These global trends-namely, the massification, marketization, decentralization, standardization, and increased documentation of education-reflect the increased importance of education in society. The massification of education 1 3 is particularly important. As more and more students attend school, questions arise as to who is accountable for ensuring the quality of education provided. Primary school enrolment expanded notably in the latter half of the twentieth century (Dorius 2012), and grew even faster in the early 2000s with the impact of the Millennium Development Goals and the Education for All Goals. And while the global primary school age population increased by approximately 2% from 2000 to 2015, the numbers of out of school primary age children declined from 100.1 million to 61.0 million during the same period, resulting in an additional 52.5 million primary age children enrolled in school (Heubler 2008;UNESCO 2016b). Yet with more children in primary school, international donors are directing fewer resources to this schooling level, and aid to basic education fell steadily from its peak in 2011 (GCE 2015;UNESCO 2016c). Concurrently, domestic resources for basic education largely remained stagnant (Steer and Smith 2015). With resources decreasing and countries seeking greater value from existing allocations, many argued that accountability systems could improve resource effectiveness (Murgatroyd and Sahlberg 2016).
In this article, we argue that most accountability reforms in education do not achieve their intended impact and that a critical condition for strengthening accountability in education involves providing different actors with an opportunity to articulate and represent their views as the accountability process enfolds. Drawing on literature about the importance of democratic voice and public participation, we refer to this process as structured democratic voice. Structured democratic voice involves organizing opportunities for diverse education stakeholders to articulate their views when policies are being planned and evaluated in ways that stakeholder concerns are heard and valued. Through the provision of voice to citizens or their representatives, accountability approaches can be re-imagined and the policy enabling environment strengthened, resulting in increased policy ownership and more effective implementation. We start by examining two key characteristics of prevailing accountability systems: their reliance on external implementation and their focus on outcomes. This is followed by discussions of the link between structured democratic voice and strengthened accountability, with illustrative examples at the regional, national, and local level. Themes in the current accountability literature are then revisited in light of these examples. Specifically, we discuss how the inclusion of structured democratic voice can help overcome shortcomings in dominant accountability approaches. The final section highlights important implications of the arguments and evidence reviewed.

The global push to make education institutions accountable
In many respects, this is the 'age of accountability' since accountability permeates all social services, including education (Hopmann 2008). Power (2000), in laying out the 'audit society,' suggests that the explosion of audits, one indicator of increased accountability, was, in part, caused by a greater demand from citizens and tax payers for accountability and transparency of service providers. In practice this was reflected in the belief that "the public has a right to expect that its resources are being used responsibly and that the public institutions are accountable for caretaking the public trust" (Supovitz 2009, p. 215). Crisis narratives around education and learning, which have gained momentum in recent years, have undermined public trust and apportioned blame to schools and teachers. For instance, perceived poor performance on national and international assessments led the Ministry of Education in Turkey to blame teachers for their inability to implement the new curriculum (Gür et al. 2012) and a government led task force in Pakistan to identify high teacher salaries as the crux of the problem (The Pakistan Education Task Force 2011). Additionally, sounding the alarm about an education emergency can push governments to act, often reinforcing desired policy. In Spain, for example, poor PISA results were used to further strengthen the government's position on early school tracking (Bonal and Tarabini 2013).
Increased references to the 'global learning crisis,' in particular, suggest that earlier commitments to universalizing access to education are no longer valid or even misguided. National and international policy makers should focus on ensuring good quality education, as measured by increased student learning (Ansell 2015). According to this prevailing narrative, national education systems must allocate more resources to measuring and monitoring learning and underscore ineffective practices among education personnel, often portrayed as lazy or unmotivated (Alhamdan et al. 2014).
The elaboration of accountability mechanisms is a typical response to this narrative and involves two major aspects: first, the external monitoring of schools and teachers and second, a greater emphasis on outcomes or results. The growth of externally applied accountability is associated with the rise of managerialism, initially part of the neo-liberal reforms of the 1980s (Hursh 2005). Managerialism supports efforts by external evaluators to increase cost-cutting and define clear standards. It is assumed that those providing services-teachers and, to a lesser extent, school leaders-need to be managed and to "demonstrate publicly that they fulfil accountability requirements" (Larsen 2005, p. 300). Managerialism today, sometimes called new managerialism (Biesta 2004), is deeply rooted in an ethos of efficiency and performance. New public management, which highlights the adoption of private management practices in the public sphere, creates a quasi-market in education, reinforcing the authority of managers over service providers and embracing a strong customer orientation (Simkins 2000). Drawing on neo-liberal economic theories, it sees education interactions in transactional terms, and focuses on the benefits of marketization and privatization, in addition to increased performance measures and accountability (Tolofari 2005).
The application of accountability mechanisms, buttressed by managerialism, has eroded trust in the teaching profession in many countries (Fitzgerald 2008). In Japan, for example, externally imposed accountability and the introduction of national testing in 2007 forced many teachers to reconsider their role and question the goals of teaching (Katsuno 2012). In South Korea, following the 2008 election of President Myung-Bak, a new education policy implemented in a top-down manner with little input from teachers and students resulted in considerable pushback (Kang 2012). This included protests by the Korean Teachers and Education Workers Union: Governmental policy makers wanted to show people the government's big efforts to bring innovative change among teachers, but it has resulted in no positive fruit. Rather, those policies have played a role in breaking a cooperative culture among teachers (Comment by KTU, cited in Symeonidis 2015).
The proposed Korean policy to hold teachers accountable for their student test scores was similar to efforts in the United States under No Child Left Behind, in effect between 2002(Chung and Chea 2016Smith 2014). It required tests for all students in the third and eighth grades, as well as one grade in upper secondary education. Results were aggregated at the school level and made public with federal funding contingent on school performance (UNESCO 2017). Contested by many in the education community, it was replaced by the Every Student Succeeds Act, which marginally reduces the amount of testing but largely maintains the push for test-based accountability. A similar withdrawal occurred in South Korea after new leadership was in place and a report identified South Korean students as among the unhappiest students in OECD countries (Chung 2017;Chung and Hong 2015).
Accountability also entails a shift from a focus on educational inputs and enabling conditions to one on outputs, outcomes, and results (Perie et al. 2007;Sahlberg 2009), in which external actors place responsibility for improved outcomes on schools or teachers (Perie et al. 2007;Smith 2014). With the unprecedented spread of international, regional, and national assessments (Smith 2014;Kamens and Benavot 2011), student test scores have become the preferred way to report on education outcomes, thereby establishing and legitimizing a 'global testing culture' (Smith 2016a). Using information from standardized assessments, authorities develop mechanisms for sanctioning or rewarding teachers and schools (Sahlberg 2010;Verger and Parcerisa 2017).
Monitoring learning through test scores and using results to drive policy reform are also practices supported by international organizations such as the World Bank, the OECD and, to a lesser extent, UNESCO (Benavot and Smith 2019). For instance, an analysis of terms associated with a testing culture or accountability found that they increased fivefold from one World Bank education sector strategy to the next (Joshi and Smith 2012). World Bank projects in education increasingly included financing for learning assessments between 1998 and 2009 (Benavot and Smith 2019). Among the most common policy solutions in the Bank's knowledge products is performance-based pay for teachers (Fontdevila and Verger 2015). The prominent role played by test scores to assess quality creates a reality in which "testing has become synonymous with accountability, which becomes synonymous with education quality" (Smith 2016b, p. 7).

Structuring opportunities for all education stakeholders to be heard
Largely absent in discussions of quality or accountability are the voices and views of those who work, learn, and teach in schools. Few spaces are available for teachers, local leaders, and students to articulate their concerns about policy initiatives or planned reforms. Structured democratic voice, as defined in this article, involves extending the work on democratic voice and public participation to education. MacDonald's (1976) seminal work on democratic evaluation differentiated democratic from bureaucratic and autocratic evaluation, in part by the redistribution of power in decision making and instillation of mutual accountability in democratic evaluations. Drawing on this and related work, structured democratic voice focuses on the provision of organized opportunities for educators to have "a voice in decision-making" (Alsbury and Whitaker 2007, p. 165). This process seeks to include already established voices as well as those not regularly heard (Brown et al. 2015).
We suggest that structured democratic voice in education is most effective when multiple stakeholders-including parents, teachers, students, and other community members-are able to articulate their views on policy planning and evaluation in ways that their concerns are heard and valued. This mirrors the key principle in participatory democracy that "all parties should be included in reaching governing decisions" (Ryan 2004, p. 449). Structured democratic voice can 1 3 be direct democratic voice-for instance, when citizens are invited to directly participate in policy planning through formal consultations or conferences-or it can be representative democratic voice, when annual review processes include representatives who articulate the views of different stakeholder groups. Structured democratic voice differs from less structured or unstructured processes such as parent surveys or public protests, which tend to be more temporary and reactionary approaches, and thus less sustainable.
Other terms in the education and accountability literature are related to structured democratic voice. Smith and Rowland (2014) distinguish between tangential voice and infused voice. Tangential voice provides parents a one-time opportunity to voice their opinion to education decision makers who hold complete authority. Infused voice involves input from parents who are able to "permeate every level of the educational process" (p. 101). Fox (2015) differentiates between tactical and strategic approaches to social accountability. Tactical approaches represent a single limited approach while strategic ones involve "citizen voice coordinated with government reforms that bolster public sector responsibility" (p. 352). The latter best resembles a diagonal accountability approach. A hybrid of vertical and horizontal accountability, the diagonal approach involves "direct citizen engagement with state institutions" (Fox 2015, p. 347). Thus, structured democratic voice is most closely related to infused voice and strategic diagonal accountability since it involves a coordinated approach that includes stakeholders in decision making at multiple points in the policy development and evaluation process.
Structured democratic voice seems especially pertinent if the task is to achieve ambitious national or international goals in education, since the responsibility for their achievement does not rest with a single individual or institution. Regardless of the overarching aim-whether to ensure inclusive, equitable, and quality education; or develop employability skills; or create globally competent and culturally aware citizens-education is best understood as a shared responsibility (UNESCO 2017). Indeed, many actors and institutions play a critical role in achieving a broad educational vision. It is only through collective engagement that educational aims and policy goals can be achieved. Individuals cannot and should not be held solely accountable for such shared responsibilities. Providing space for structured democratic voice highlights this shared responsibility and the interconnectedness of actors. It also recognizes that "accountability actions are part of a broader and longer process of engagement between actors and the state" (Dewachter et al. 2018, p. 168) and helps avoid incoherency across responsibilities while promoting ownership in the larger education goals (Fancy and Razzaq 2017).
Trust is vital in this process. It can influence the accountability approach chosen and its effectiveness. Trust in education can be seen as trust in the person, trust in the profession, and trust in the process (UNESCO 2017). In education, trust in the person is generally widespread-parents are often supportive of their child's teacher(s). Trust in the profession is more problematic and varies across countries. Just because parents (and others) value their local teachers does not mean teachers, as a profession, are viewed as capable and trustworthy. Finally, among those being held to account, there must be trust that the process is fair and just. When people perceive the process as unjust or corrupt, commitment to intended policy is adversely affected. Structured democratic voice can improve trust in the profession by recognizing the value of educators and by including them in the decision-making process. Structured democratic voice can also improve trust in the process if teachers and other educators feel their views and concerns have been heard. Collaborative processes, where all parties listen to each other, "can both build trust and create incentives for more voice" (Fox 2015, p. 356).
One avenue to increase trust involves the creation of shared aims and purposes. For example, higher education roundtables in the United States, supported by The Pew Charitable Trusts, brought government, business, civil society, and education leaders together to discuss the aims and future agenda of higher education across the country (Burke 2005;Nichols 1995). In North Dakota, this led to a flexible accountability program with broad support among leaders in government, business, and higher education and an agreement to continue roundtable discussions to constructively address potentially conflicting views (North Dakota University System 2003). In Finland, education reforms in the late 1970s, which resulted from extensive political consultations, led to greater local school autonomy. By the 1990s, a trustbased school culture had taken root and teachers and principals were given opportunities to participate in curriculum development and policy reform (Sahlberg 2007). The public lent support to the teaching profession and gave teachers the autonomy and discretion to deliver good quality education for the next generation. This abiding trust in teachers continues to be a crucial trademark of the Finnish education model (Auren and Joshi 2016).
By contrast, limited trust in the profession or the process can undercut teacher engagement in decision making. In South Africa, lack of trust in teachers may help explain their limited involvement in the policy-making process after apartheid. Despite the transition to a more participatory government, teachers were often left out of education reform planning (Govender 2008). This may be due to frustration with a profession that benefited from political placements in top positions (Zengele 2013) and teacher unions that split over whether they wanted to maintain the inequitable practices of the past or push the progressive movement of the government (Govender 2015). In the end, teachers were not included in the decision to implement outcomes-based accountability (Jansen 2002). In other studies, externally imposed and sanction-oriented accountability systems foster situations where educators feel threatened by external actors making decisions beyond their control, thereby enhancing risk aversion and undermining trust (Sahlberg 2009;Stensaker and Harvey 2011). Such situations can lead to a vicious cycle where low levels of trust result in more external, sanction-based accountability, which in turn further undermines trust.

Where there is trust, accountability improves
Trust is thus a kind of lubricant, a needed condition, in contexts where accountability operates. As seen in Fig. 1, efforts of education actors are influenced by the amount of trust present, as well as the economic, political, and social context. The same accountability approach is unlikely to work well in all contexts or at all times. Country history and current circumstances should be considered when considering different accountability approaches. Within the context of SDG 4, the challenge is getting various actors-government officials, teachers, school leaders, and others-to work together toward a shared set of targets. The interdependence of actors is illustrated in intersecting cogs or gears.
UNESCO (2017) identified four aspects of the enabling environment, represented by oil drops in Fig. 1. When the enabling environment goes unfulfilled, individual efforts are stymied. The enabling environment consists of resources, capacity, motivation, and information. Regardless of the ingenuity of the accountability approach, if actors lack key conditions, no approach will be effective. For example, for teachers to be held accountable for the quality of their instruction (an individual responsibility), certain conditions are crucial. Teachers must have • pedagogical capacity (have they received the training and support to teach this group of students in this subject?), • instructional resources (do they have all the necessary teaching materials and physical resources necessary to fully implement their pedagogical approach?), • sustained motivation (do they feel their work is valued and appreciated?), and • access to clear information (do they know what their responsibilities are and how to give an account for their actions?).
Beyond strengthening trust in the process and teaching profession, structured democratic voice is likely to increase actor motivation and commitment. When actors can contribute to the accountability policies put in place, they are more likely to embrace them (Salmi 2009). Participation in such a process increases actors' sense of ownership encourages them to view policies as legitimate and the evaluation of their efforts as fair.
Having a voice in policy development also contributes to the information component of the enabling environment. Direct participation in policy creation allows actors to raise questions and concerns. It helps reduce confusion over individual responsibilities while minimizing overlapping or duplicate work activities. With a fuller understanding of their and others' responsibilities, individuals are more likely to accurately report on the tasks they perform, and feel fairly treated (Cerna 2014).

Structuring democratic voice in practice
While instances of structured democratic voice in education are relatively rare, this section outlines several notable examples from which lessons can be drawn. These insights are especially pertinent in East Asia and other regions where citizens increasingly clamor for greater authority in decision making (Welzel and Dalton 2017). To begin, international organizations have increasingly encouraged democratic participation in education planning and evaluation. For example, UNESCO's International Institute of Educational Planning (IIEP) suggests that government and non-government stakeholders be represented in joint steering committees with formal power to appraise and approve education sector plans (IIEP 2010). The Global Partnership for Education (GPE) also supports civil society engagement through the creation of local education groups-in 2014, 35 national coalitions reported engagement in such groupsand, more importantly, through the establishment of Joint Sector Reviews (JSRs).
Over the past two decades JSRs have been used in the health and education sectors, aiming to promote country ownership and mutual accountability. Ideally, JSRs are meant to bring together representatives of government agencies, donors, civil society, and other relevant stakeholders, on a yearly basis, to discuss and evaluate sector-wide progress, culminating in an annual report. This version of structured democratic voice is not without challenges. A recent GPE-commissioned evaluation of 39 JSRs held between July 2014 and December 2015 highlights the fact that stakeholder participation is uneven (Martinez et al. 2017). The GPE report indicates that the Ministry of Finance was only present in about 53% of the reviews. Parents' associations were present in only one-third of the reviews; teachers' unions in less than half of JSRs.
Another challenge is converting the presence of different actors into exchanges in which all voices are heard and valued. In practice JSRs tend to be donor-driven initiatives, and discussions often limited to the effectiveness of donor investment or donor reports to the government (Holvoet and Inberg 2009). Other actors, including civil society representatives, tend to take a back seat, viewing their role as observers rather than participants. The lack of active participation of all parties and an over emphasis on donor issues undermines the JSRs aim of fostering mutual accountability. The strong donor orientation may also limit a more comprehensive review of the education plan or system. Of the 39 plans evaluated in the 2014-2015 period, only one-third of the JSRs adequately followed up on activities from the previous year (Martinez et al. 2017). If the national ownership of JSRs is to improve they need less focus on donor concerns and more on planning and implementation issues.
Some countries in Latin America have provided opportunities for structured democratic voice. Examples include Colombia's review of its National Development Plan (OECD 2016) and Mexico's independent election administration (Avritzer 2002). In Argentina, following a 2006 education law, The Confederation of Education Workers was formally included in the National Education Quality Council, thereby participating in efforts to plan policy and evaluate progress. In Uruguay, teachers overcame an externally imposed policy by mobilizing teacher unions to place representatives on the central education board (Gindin and Finger 2013). Unfortunately, teacher union participation in education decision making is unusual-not only in Latin America, but in other regions. A recent survey of 70 unions across more than 50 countries found that approximately three in five were never or rarely consulted on issues related to instructional materials or pedagogical practices (Symeonidis 2015).
Brazil represents a well-established example of structured democratic voice in education planning. The 1996 education law called on the ministry to quickly formulate a decennial education plan (Bodião 2016;de Andrade Tosta and Coutinho 2016). Beginning in 1997, the National Education Council, established by the National Congress on Education, opened up a series of consultations with civil society organizations, professional associations, and specialists who helped develop national guidelines for education (Bodião 2016: Federal Republic of Brazil 2014. During this period, participants from unions, academic institutions, scientific and student organizations, as well as organized social movements, came together to establish the National Forum in Defense of Public Schools. This Forum led a democratic process in which the proposed education plan was drafted, and then eventually submitted to the government. The submitted plan was endorsed by over 70 parliamentarians in early 1998 and was in place from 2001 to 2010 (Bodião 2016).
Drawing on insight from this process, Brazil's ministry of education took the lead in developing subsequent plans, including the creation of municipal, state, and national conferences (Bodião 2016). These conferences encouraged dialogue between civil society organizations and the government. More than 3 million people were involved in the entire process, including over 450,000 delegates. Central to these consultations was the Brazilian Campaign for the Right to Education, established in 1999, which currently includes over 200 groups, operating in 22 of 27 Brazilian states.
In 2010, the National Education Conference in Brazil developed final amendments to the reference document circulated by the ministry and submitted the proposed plan to the House of Representatives (Bodião 2016). The national conference, however, lacked actual authority allowing the House of Representatives to disregard some proposals and include others not discussed in the conference at all. The 4 years between submission of the plan and its sign off by the President were marked by disagreements around public financing, specifically whether 7% or 10% of GDP should be allocated to education. In the year immediately following the plan's submission, civil society groups took advantage of the call for amendments and submitted more than 2900 for consideration. Groups were further brought together through a media campaign titled PNE for Real, where members gave interviews, wrote articles, and held media events. This pressure was outlined in a technical note, 'Why 7% of GDP for education is not enough: Calculation of investments need for the new PNE to ensure a minimum quality standard.' The note concluded that an additional 5.4% of GDP was necessary, well above the government's estimate of 1.9% (Bodião 2016). The civil society campaign was partially successful as target 20 of the 2014/2023 PNE identified 10% of GDP as the investment goal by the end of the plan (Federal Republic of Brazil 2014).
In Africa, communities in Ethiopia and Malawi provide noteworthy examples of a partnership model, which builds government capacity while including community members in the problem identification and evaluation process. Supported by a civil society organization (CSO), the model is centered around a school performance appraisal meeting, which works within the government's monitoring schedule to improve the capacity of those already tasked with data collection through targeted training. At such meetings, local authorities are trained on an integrated information system that provides input into electronic school report cards. The report cards then act as the primary prompt during the school performance appraisal meeting and are presented visually to ensure participation from those with lower literacy skills. The discussion culminates in a school action plan outlining how each actor can contribute to reaching the jointly identified goals (Visser et al. 2017).
In Wolaita Zone, Ethiopia, the final product included a Gender Action Plan with clearly delegated responsibilities for all, including the Girls' Club Coordinator. Evaluations of the project found positive gains on girls' attendance and their attitudes toward education (Link Community Development 2017). An external, post-project evaluation found significant gains in achievement with decreased gender disparity. Additionally, parents were more likely to encourage their daughters to attend and teachers improved their gendersensitive instruction over the project (Visser et al. 2017). Importantly, this was not a one-time CSO intervention. The sustainability of the approach is demonstrated in the continuation of school performance appraisal meetings in many participating schools. Such partnership arrangements, however, do not make then immune to fluctuations in funding.
In Ethiopia, lack of financial resources was identified as the greatest threat to its longevity (Visser et al. 2017).

Obstacles to effective structured democratic voice
The three examples above reflect the challenges of implementing structured democratic voice in practice. As seen in Fig. 1, accountability does not exist in isolation. Instead, mechanisms interact with, and are influenced by, the social, economic, and political environment. This suggests asymmetry in application where no approach is expected to be universally effective. A recent review by Pagatpatan and Ward (2017), focusing on decision making in health policy, highlights possible challenges when considering structured democratic voice. In their evaluation of 77 articles, they identify four factors that underlie the effectiveness of public participation: political commitment, partnership synergy, inclusiveness, and deliberativeness. Political commitment or political will is necessary in partnering with the public. It involves officials "ceding genuine voice to the stakeholders" (Pagatpatan and Ward 2017, p. 522) within an environment in which different viewpoints can be discussed openly. Additionally, resources are provided to help finance and sustain the accountability efforts while improving local capacity to ensure capable and informed democratic participation. As seen above, political will can be partial or insufficient. In Brazil, voice without authority was provided for the national conference, undermining the effectiveness of structured democratic voice. While in Ethiopia, the government benefited from collaborative efforts to improve the education information system but seemed unwilling to provide the financial support needed to ensure their long-term sustainability.
Authoritarian governments are less likely to cede decision-making authority in education. While this has historically been the case in many East Asian countries (e.g., Vietnam, Lao PDR, China), some have suggested a form of 'soft authoritarianism' where limited space for public intervention is provided in targeted areas. In a region prioritizing economic growth over political freedom, public participation in policies related to economic prosperity is more likely (Chou and Huque 2016). Importantly, Chou and Huque (2016) suggest that increased use of structured democratic voice will help generate greater political will for such opportunities in the future.
Partnership synergy also influences the success of structured democratic voice. Synergy involves the ability of different groups to work together to reach a consensus. This can be difficult with a diverse group of stakeholders and requires a long-term relationship to build trust (Pagatpatan and Ward 2017). Recognizing group plurality while highlighting shared identity can help facilitate partnership synergy. Open discussions on locally relevant issues, in which individuals voice their concerns and propose ways forward, can help facilitate this. For example, efforts to eliminate child marriage in Malawi were enhanced by meetings that allowed all community members to describe the factors contributing to the issue, before collectively outlining an action plan (UNESCO 2018).
Inclusiveness and deliberativeness are about who is included in the process and the quality of the dialogue. Without trust, inclusive participation and quality dialogue are less likely (Pagatpatan and Ward 2017). As noted in the JSR examples above, representation, even of key actors, can be uneven. Excluded groups or individuals may disengage if they believe their voices go unheard or make little difference (Madero and Morris 2016). Power dynamics can also play a role. For instance, the historical dominance of top-down decision making and economic reliance on donor financing likely contributed to the limited engagement with civil society in JSRs. Marginalizing some voices over others can lower the diversity and quality of the conversation and threaten the sustainability of the accountability mechanism.
Mexico City, which seeks to become the most environmentally sustainable city in Latin America, adopted a 15-year Green Plan during the Marcelo Ebrand Administration (2006Administration ( -2012 and promoted the role of sustainable democratic voice. Public participation was structured into a pre-launch consultation and a post-launch oversight board, which monitored, evaluated, and suggested improvements to the program. Representativeness was a concern throughout. In the initial consultation only 5% of the city's citizens participated. Board members raised concerns that NGOs, included to represent certain marginalized groups, were actually acting in their own self-interests. Furthermore, information about the plan and board's decisions was only available on a website, although less than 30% of the population had internet access in 2010. Deliberativeness was also a concern as inaccurate or unreadable documentation undermined the quality of the board discussions. Lack of representation, dissatisfaction with pre-meeting reports and within meeting dialogue, as well as lack of political commitment beyond the Ministry of the Environment contributed to high board member turnover. As individuals became disillusioned with the process, a change in administration led to the demise of the program (Madero and Morris 2016).

Dominant models of accountability and their critics
Notwithstanding the challenges involved, two important implications emerge from instances of structuring democratic discussions in education policy and planning. First, to reach comprehensive education goals, such as increased or more equitable learning outcomes, there needs to be coherence across actors and policies. This requires input from diverse actors at different levels in meaningful ways. When stakeholders are included in the creation and implementation of shared aims, they see them as pertinent to their community. Furthermore, including individuals who have clearly identified responsibilities in teaching and learning is integral to converting policy intentions into practice and subsequent evaluation. Second, the inclusion of structured democratic voice in policy and planning processes reduces duplication of efforts, which can lead to fractured accountability. When an individual is held accountable for multiple tasks, pulling them in opposing directions, and diminishing their ability to fully complete any single responsibility, fractured accountability is like to occur.
To what extent have the dominant models of accountability in education incorporated these ideas and principles? Arguably, the most influential framework is the one advanced in the World Bank's 2004 World Development Report (WDR). The 2004 WDR emphasized the importance of public services responding to the demands of local end users, advocated for greater decentralization and controland more accountability (World Bank 2003). The WDR accountability framework details relationships between clients, providers, and policy makers (Lateef 2016), and consists of two paths to accountability: the long route and the short route. The long route includes citizens influencing policymakers who would then have oversight authority over service providers. Providers would give an account of the action, not directly to the clients but to the policy makers. The World Bank made clear that this was the less desirable option. The large bureaucracy, and embedded politics, was expected to ground progress to a halt, making the government largely unresponsive to citizen demands (Edwards 2012). Even in a well-functioning system, the long route was seen as weak (Lateef 2016).
The short route was considered preferable as it gave direct authority to clients, allowing them voice and choice in providers (Lateef 2016). A key underlying assumption of the short route was that "transparency (access to information) combined with participation would lead to more accountability, which in turn would improve service delivery" (Dewachter et al. 2018, p. 159). With information in hand, clients would be conscientious consumers, rationally analyzing the possibilities and switching providers when the analysis deems appropriate. The expectation that clients engage as consumers was confirmed 10 years after the release of the 2004 WDR when the report's lead authors made clear the only option available in the short route, in their opinion, was market-based accountability (Lateef 2016). This clarification was not surprising and largely mirrors the shift in accountability relationships from a political relationship-where citizens and the government work together for the common good-to an economic relationship-where taxpayers are consumers of services (Biesta 2004). In the latter relationship, parents and teachers/schools are often pitted against each other, diminishing the potential for collaborative dialogue that would identify and remedy shared concerns.
A presumed advantage of the short route was its ability to overcome the principal-agent problem (Lateef 2016). This problem involves the ability of the client or principal (in education often the parent) to have their needs met by the agent (in education, teachers are usually considered the agent). The agent is hired by the principal but misaligned interests and aims often lead to problems in efficiency (Smith 2017). According to the short route, teachers (the agent) should improve information available to parents (the principal), allowing for an evaluation of interests and potential sanction (perhaps in changing schools) in the event of misalignment. The expected result is the agents will correct their behavior to maintain their relationship with the principals. Local control, illustrated by the short route, has become a mantra in the accountability literature. Short route practices supported by the World Bank have influenced a wide array of international organizations including the Brookings Institution, GPE, UNESCO, and Save the Children (Edwards 2016).
Critiques of the 2004 WDR have questioned the dichotomous presentation of seemingly mutually exclusive pathways. The short route and the long route to accountability are intertwined and should not be viewed in isolation (Dewachter et al. 2018). The short route "needs the 'long route' of responsive elected authorities to work" (Fox 2015, p. 347). And, indeed, "accountability approaches effective in some contexts and some aspects of education may be detrimental in others" (UNESCO 2017, p. 7). Thus, adopting appropriate complimentary approaches leads to the best results (Dewachter et al. 2018).
By highlighting the importance of structured democratic voice, the shortcomings of both the short and long route become clear. As Dewachter et al. (2018) point out, the short route considers a limited range of actors and these actors are expected to work individually and independently from one another, leading the approach to discount the power of larger civil society. The assumed independence of individual actors creates a situation in which accountable parties are pulled in different directions, diminishing the time and quality of their efforts and leading to fractured accountability. In reviewing the principal-agent problem what was originally seen as a two-party relationship has been stretched to include multiple principals (Fox 2015). In education, the assumption that these multiple principals represent a homogeneous set of interests (Fox 2015), and express similar demands from educators, is violated. The multiple principals demand different tasks from the responsible party (Lateef 2016), reducing the potential gains from accountability.
The long route to accountability advances a limited notion of who constitutes the government and, therefore, who has authority under this approach. The long route clearly separates the governed (clients) from those that govern (policymakers), with the latter holding full oversight authority. The long route fails to consider "other public 'checks and balances' institutions, such as legislatures, the judicial system, audit institutions, ombudsman agencies, or public information access reforms" (Fox 2015, p. 347). Common examples of accountability through the long route-namely, elections and protests (Lateef 2016)-are actually temporary in nature or limited to one-time, tangential voice (Smith and Rowland 2014). With structured democratic voice, the government cannot be clearly separated from the governed because the people are the government. With stakeholder voices included in policy planning and evaluation, they have authority to help reconsider and reform an education system. Structured democratic voice is, therefore, best represented in a revised long-route approach, involving three elements (Dewachter et al. 2018). The traditional view of the long route is laid out as political accountability, where authority remains in the hands of policymakers. Added to this are citizen-led and civil society-led social accountability. Social accountability, although typically associated with the short route, is placed here as citizens are invited into governance space. Citizen-led social accountability decreases the distance between the governed and government, providing for direct interaction between citizens and policymakers, mediated by a civil society organization. In both types of social accountability, citizens retain some decision-making authority. The former could be accomplished through a form of structured direct democratic voice, while the latter would likely involve a form of structured representative democratic voice.

Concluding remarks
The introduction of externally driven, results-oriented accountability in education has fostered unwanted competition and undermined trust. It has done little to raise learning levels or reduce disparities and equity gaps (Lingard et al. 2017;Smith 2016c;UNESCO 2017). Notwithstanding the challenges of creating and implementing internally focused accountability systems, based on shared responsibility and actively engaged stakeholders, such systems are more likely to contribute to real gains in education, including progress on ambitious education targets (Fullan 2011). This paper argues that structured democratic voice is a key component of effective accountability. When multi-party deliberations are applied to education, their impact is palpable: better aligned and consensual aims, less reliance on temporary responses by strongly vested actors, and accountability which is less fractured and more coherent. Creating platforms for democratic voice in education planning and policy makes accountability practices less susceptible to external shocks and political whims.
Structured democratic voice is nurtured in organized spaces where actors with intersecting interests in education overlap (see Fig. 2). These spaces enable participants to channel their input and responses into all aspects of policy making and policy evaluation. Such a process of constructive deliberation differs from the dominant short-and long-route approaches to accountability, in which rigid boundaries among the three groups are maintained. Structured democratic voice, when properly applied, creates benefits for all involved. Parents, community members and other stakeholders, located outside of the school, are more confident that their voices are being respected and heard, which increases their trust in the process. Teachers and school leaders benefit since the enabling environment to fulfill their responsibilities improves. They have more detailed information as to their individual responsibilities and are more motivated to address the learning needs of students. Additionally, by participating in decision-making processes, teachers find that trust in the teaching profession improves-educators are seen as knowledgeable and committed. Trust in the process also improves since participation leads to a more fair and just education system. Finally, the government benefits through increased capacity as they are exposed to a variety of perspectives and learn from those on the front line of education, which ultimately leads to efficiency gains.
While research is needed to identify which conditions best support the inclusion of structured democratic voice into an accountability system, several obstacles are clear. First, capacity among all stakeholders is a concern and can lead to situations where, despite the involvement of multiple actors, participants may not contribute equally or effectively (Martinez et al. 2017). Second, governments may be disinterested in incorporating non-official voices in the decisionmaking process. Many are not convinced as to the benefits of sustained multi-stakeholder engagement.
A long-term commitment to a rigorous participatory process may dismay governments (IIEP 2010). To push through a new plan or policy through quickly, governments may prefer to limit decision making to more controlled consultations, few in number. Such actions do little to build capacity within government or, importantly, engender ownership or commitment from committed actors responsible for delivering education (Williams and Cummings 2005).
Donors certainly have a complex role to play in this process. On one hand, they are deeply convinced of the value of accountability, and often support leading accountability models, which are embedded in agreements with governments. On the other hand, they can dominate instances of structured democratic voice, pushing out weaker, less articulate stakeholders who may feel intimidated in such spaces or whose inputs are viewed as less "evidence based" and anecdotal. Donors need to consider how best to foster truly collaborative, multi-stakeholder arrangements.
Structuring democratic voice in education is a process worth pursuing and sustaining. It replaces tokenistic pledges from government that allow one-time inputs into decision making, and promotes a sustained presence across generations and changes in political leadership. The collaborative potential of structured democratic voice improves the chances that ambitious national and international goals of education can be met.