Communities of Practice and PISA for Schools : Comparative Learning or a Mode of Educational Governance ?

This paper examines the Organization for Economic Cooperation and Development’s (OECD) PISA for Schools, a new variant of the Programme for International Student Assessment (PISA) that compares school-level performance on reading, math and science with international schooling systems (e.g., Shanghai-China, Finland). Specifically, I focus here on a professional learning community – the Global Learning Network (GLN) – of U.S. schools and districts that have voluntarily participated in PISA for Schools, and how this, arguably, helps to normatively determine ‘what works’ in education. Drawing suggestively across diverse thinking around contemporary modes of governance, and emerging topological spaces and relations associated with globalization, and informed by interviews with 33 policy actors across the PISA for Schools policy cycle, my analyses suggest that GLN allows the OECD to discursively and normatively constrain how ‘world-class’ schools and systems, and their policies and practices, are defined. Education Policy Analysis Archives Vol. 25 No. 92 SPECIAL ISSUE 2 However, and in light of the productive capacities of power relations, I also argue that GLN provides opportunities for local educators and leaders to undertake meaningful collaboration and sharing, and to find policy spaces outside of those defined by more performative discursive framings of school accountability. To this end, I explore how GLN may help to foster alternative policy spaces from which educators can ‘talk back’ to national and state authorities, and potentially promote more ‘authentic’ understandings of, and possibilities for, schooling accountability.


Introduction
In contemporary times, data-based comparisons between schools and/or schooling systems have arguably become something of a fetish, an overarching schooling meta-policy affecting curriculum, pedagogy, and the experiences of teachers and students alike (Lingard, Martino, & Rezai-Rashti, 2013).Far from merely providing a benign means to learn from others, comparisons of schooling policy, practice and performance are instead now central to contemporary modes of educational governance, in which schooling is subjected to the concomitant glare of both national and global eyes (Nóvoa & Yariv-Mashal, 2003).Moreover, such performance comparisons are increasingly embedded in putatively high(er)-and low(er)-stakes regimes of teacher and schooling accountability.These have manifested both within national systems -for instance, Value-added measures, or VAMs, in the USA (see Amrein-Beardsley, 2014;Holloway, 2017); the National Assessment Program -Literacy and Numeracy, or NAPLAN, in Australia (see Lingard, Thompson, & Sellar, 2015) -and between national (and some subnational) systems; for instance, the Programme for International Student Assessment, or PISA (see Lewis & Lingard, 2015).
It is this current milieu of schools, schooling systems, educators and policymakers looking around to others -to the national, the international and the global -that prompts the research presented in this paper.Specifically, I focus here on the development of a professional learning community -the Global Learning Network (GLN) -that has arisen in response to a new school-level instrument for international benchmarking and policy advice: the Organization for Economic Cooperation and Development's (OECD) PISA-based Test for Schools (hereafter 'PISA for Schools').Unlike the more renowned 'main PISA' undertaken by national (and some subnational) schooling systems every three years, PISA for Schools instead measures the performance of 15-year old students at individual schools in reading, mathematics and science, and compares this against the schooling systems assessed by main PISA. 1 This positions schools within a globally commensurate space of measurement and comparison (Lingard & Rawolle, 2011), encouraging local educators to engage with, and learn from, the policies and practices advocated by 'high performing' schooling 1 At present, PISA for Schools is limited to schools in the USA, UK (excluding Scotland) and Spain, as well as a more restricted implementation in the cities of Moscow and Brunei.Before a school can request to participate in PISA for Schools, national education authorities -via their representatives to the PISA Governing Board at the OECD -must first grant permission for the test to be made available in their country (see Lewis, 2017b).systems as determined by PISA (e.g., Shanghai-China, Singapore, Finland), as well as the policy learning of the OECD itself.These school-level performance data and examples of 'best practice' are made available to participating schools via a 160-page report.With the development of PISA for Schools, and the emergent relevance of the OECD within local schooling spaces, we can see evidence of what Sellar and Lingard (2014) describe as the enhanced scale, scope and explanatory power of the Organization's education policy work.
PISA for Schools is also unique insofar as schools or districts voluntarily purchase the testing and policy services of the OECD by way of a private accredited provider, usually an edu-business, rather than a school's participation being organized (or even mandated) by the education authorities of nation-states.Within broader moves towards a global governance of education (Meyer & Benavot, 2013), as well as the growing significance of comparison in contemporary modes of governance (Nóvoa & Yariv-Mashal, 2003), this distinctive function of PISA for Schools opens up new local schooling spaces to global policy actors like the OECD, while, at the same time, minimizing the potential influence of the nation-state to mediate such international forms of evidence.Indeed, such discourses, processes and relations have been recognized to influence how schooling is both understood and practiced across national and, increasingly, subnational schooling spaces, enabling global actors (such as the OECD) to overcome physical distance and 'reach into' (Allen & Cochrane, 2010) decidedly more local policy spaces in order to influence policy processes within (see Lewis & Hardy, 2017;Lewis, Sellar, & Lingard, 2016).It should be noted, however, that the voluntary, userpays model associated with PISA for Schools has resulted in the majority of participating U.S. schools to date being located in large, affluent suburban districts, with these communities often also positioned as 'high performing' by existing accountability measures, such as mandated state-level testing. 2n order to accommodate such changing empirical realities, and informed by the theoretical 'toolbox' approach of Ball (1993), I draw here upon an eclectic variety of intellectual resources to help understand these new spaces and relations of global educational governance, particularly around emerging relational -or 'topological' -spatialities and rationalities associated with processes of globalization (Allen, 2011;Allen & Cochrane, 2010;Amin, 2002;Lury, Parisi, & Terranova, 2012).This reflects the inclusion of horizontal relationships and new non-governmental actors into the work of the State across the education policy cycle (see Au & Ferrare, 2015;Hursh, 2016;Verger, Lubienski, & Steiner-Khamsi, 2016), even while these new horizontal relationships exist in tandem with older vertical relationships and modes of organization through which the State now functions. 3oreover, such topological thinking helps us to understand spaces determined less by physical location (e.g., the nation-state; the state) and more by the myriad relations, both material and discursive, that 'flow' between individuals and organizations in the development and enactment of education policy.Informed by the global-local connectivity that such flows enable, I focus here on how PISA for Schools makes possible new relations, spatialities and modes of governance between the OECD and participating U.S. schools and districts, particularly with respect to how such relations can influence local understandings of schooling.In the context of the OECD promoting global 'best practice' to local schooling sites, this requires us to consider how such policy learning is encouraged throughout the epistemic communities and policy networks constituted by PISA for Schools, such as GLN.In what follows, I examine how PISA for Schools facilitates the creation of professional learning communities between schools and districts, the OECD and several key partner organizations in the USA, and how these, arguably, help to normatively determine (and even constrain) 'what works' in education (see Auld & Morris, 2016;Dudley-Marling, 2011;Lewis, 2017a).My particular focus is GLN, a professional learning community that PISA for Schools participants can voluntarily enter into, which has been developed and supported by the U.S. not-for-profit America Achieves, in conjunction with: the OECD; U.S. philanthropic foundations, including Bloomberg Philanthropies, the Kern Family Foundation, and the William and Flora Hewlett Foundation; and the US school advocacy network EdLeader21.This collaboration includes U.S. schools and districts that have participated in PISA for Schools, and claims 356 U.S. member schools in 149 districts across 30 states (America Achieves, 2014c).My analyses suggest that GLN allows the OECD, through PISA for Schools, to discursively and normatively constrain how world-class schools and systems, and their policies and practices, are defined, and thus how individual teachers and school leaders should use these insights to inform local reform efforts.This results in local possibilities for thinking and practicing schooling at the school and district level becoming rendered almost wholly through the prism of PISA, or what has elsewhere been described as 'PISA lenses' (Carvalho, & Costa, 2015).
However, and in light of the productive capacities of power relations, such professional learning communities should not merely be construed as an extension of the OECD's policy influence and governance.They also, arguably, can provide opportunities for local educators and leaders to undertake meaningful collaboration and sharing, and to find policy spaces outside of those defined by more performative discursive framings of school accountability.As Foucault (1978) usefully notes, 'relations of power are not in super-structural positions, with merely a role of prohibition or accompaniment; they have a directly productive role, wherever they come into play' (p.94; emphasis added).Thus, a more nuanced understanding and critique is required to appreciate how PISA for Schools is neither solely positive nor negative in outcome or intent, reflecting a situation that is decidedly more both/and than either/or.
To this end, I also explore how GLN may help foster alternative policy spaces from which educators can 'talk back' to national and state authorities, and potentially promote more authentic understandings of, and possibilities for, schooling accountability.Where educationally productive, the opportunity for including and valuing local input into policies, pedagogies and curricula is perhaps indicative of so-called 'rich accountabilities' (Lingard, Martino, Rezai-Rashti, & Sellar, 2016).This is where the totalizing influence of data-driven, 'top-down' modes of schooling accountability is tempered somewhat by more individual, 'bottom-up' ways of understanding practice and student outcomes, and accompanied by local determinations of how such data might be used and valued.

Research Methods
The research here is informed by data collected over a 14-month period, between October 2013 and January 2015, during which time I examined the development and administration of PISA for Schools by the OECD and its partner organizations, and its subsequent implementation in three school districts within the U.S. states of New York, Virginia, and Texas.Much of the following analyses draw upon semi-structured interviews undertaken with 33 key policy actors from organizations involved across the PISA for Schools policy cycle, with a particular focus on those who were active in the U.S. deployment of the program.These included: the OECD Directorate for Education and Skills; the PISA Governing Board (PGB), a body composed of government representatives from PISA-participating countries that sets and oversees the policy objectives of PISA, including PISA for Schools; U.S.-based philanthropic foundations that funded the development and upkeep of PISA for Schools; U.S. not-for-profit and private organizations (America Achieves, EdLeader21) that helped administer PISA for Schools in the USA, including responsibilities around school recruitment, promotion and collaborative learning; the edu-business (CTB/McGraw-Hill) acting as the accredited provider of PISA for Schools in the USA during the period 2013-2015; and leaders from three U.S. school districts in New York, Virginia, and Texas that participated in PISA for Schools testing.All interviews were confidential and approximately one hour in duration, and the research was conducted in keeping with ethics requirements of the university institutional review board.
Reflecting the 'topological' spatialities of contemporary policymaking and the geographically dispersed nature of the PISA for Schools policy network, these interviews were conducted across a diverse set of locations and using a variety of methods, including both face-to-face and electronicallymediated formats (e.g., Skype, FaceTime).Most of the interviewees were purposefully recruited after extensive Internet searches to identify key stakeholders in the development and administration of PISA for Schools, so they could in turn provide significant insights into a newly developed, and, at the time, narrowly implemented, assessment program.However, and demonstrating the highly relational (and more often collegial) nature of the PISA for Schools network, additional conversations with research participants were often serendipitously arranged with the voluntary assistance of, and invitation by, prior interviewees.
Complementing these interviews is the analysis of relevant print documents, audiovisual materials and websites from organizations involved in the development and implementation of PISA for Schools.These documents include: school-level reports received by PISA for Schools participants in the USA; other OECD documents and technical reports related to PISA for Schools, and the Organization's broader education policy work; promotional and administrative materials associated with the U.S. implementation of the assessment; and various school and district-based reports and stakeholder communications.Of particular focus is the report received by participating schools, which outlines school-level performance on the PISA for Schools test and examples of 'best practice' from international schooling systems, and which also provides links to other OECD publications for the purpose of informing local schooling reform.These technical documents help to nuance the data collected during research interviews.Collectively, the various data analyzed here help reveal the diverse, and often conflicting, perspectives of the organizations and individuals associated with PISA for Schools, while also highlighting the frequent disparities between official institutional reports and the potentially more candid talk of policymakers themselves.

'They are the Experts': Governing Teacher Learning Through GLN
Creating professional learning communities around PISA for Schools, as typified by GLN, is far from an exercise of secondary importance to the OECD, and such developments were positioned as central to the long-term purpose of the assessment.Indeed, and invoking the topological spatiality and relationality that links together otherwise disparate schools and policy actors, a senior OECD official proclaimed the Organization's desire to 'join up' participating schools, and create a global platform to share effective policies and practices through PISA for Schools: I think the next important step is to bring the schools together, to create a platform where schools that have done it can actually share their experiences with other schools [and] can learn from other schools all around the world… And enabling schools to sort of join up -to speak with each other and learn from each other -is really, I think, the true objective, the long-term objective, of this exercise.(Emphasis added) Such inclinations for international networks of PISA for Schools participants were not only present amongst the OECD but also, importantly, its partner organizations.In fact, some of the most enthusiastic advocates outside of the OECD were the U.S. philanthropic foundations that initially funded the development of PISA for Schools and, through America Achieves, the GLN professional learning network.This sentiment was clearly reflected in the comments of a U.S. philanthropic executive: I think from the beginning, that's been a concern: How do we network the schools virtually, physically, regionally?How do we get them together so they can learn from the high performers but also, possibly, have a connection with an international school that's much more effective?(Emphasis added) Regardless of whether such comments originated from actors at the OECD or the supporting partner organizations, it is interesting to note how these opportunities to 'learn from other schools all around the world' was still very much predicated on comparative PISA performance and the marking of difference.As such, and reinforcing the normalizing role of such testing regimes and comparative data to render education largely through 'PISA lenses' (Carvalho & Costa, 2015), 'very different results' at the local level were the basis by which schools should look around for, and share, examples of 'what works'.However, this process is underscored by the problematic notion that schooling system performance on PISA meaningfully represents their effectiveness, or that performance can be causally attributed to certain 'high performing' schooling policies and practices, which can themselves be readily transferred between different schooling contexts to produce similar outcomes (see Gorur, 2016;Gorur & Wu, 2015).Notwithstanding these difficulties, the U.S. not-for-profit America Achieves, an organization funded by many of the same philanthropies supporting the development and enactment of PISA for Schools in the USA, created the GLN professional learning community in April 2013, soon after the U.S. release of PISA for Schools.As an America Achieves executive noted, the justification for GLN was to 'raise the bar' for U.S. education through PISA for Schools by 'facilitat[ing] their access to the best policies and practices from around the world' (America Achieves, 2014b, p. 5).This 'access' was to be enabled through a variety of scheduled 'virtual' and 'physical' convenings produced and coordinated by America Achieves, in which local educators and policymakers can share their experiences around PISA for Schools, and discuss the OECD's proffered policy advice.Significantly, these professional conversations are to be driven almost exclusively by PISA for Schools, with GLN opportunities for data-driven school and district-level professional development including:  Access to a community of school and district leaders dedicated to comparing their performance to, and learning from, world-class systems;  Access to video case studies documenting best practices in world-leading schools;  Virtual convenings that highlight concrete actions schools have taken in response to assessment results and the practices that have contributed to world-leading performance at schools;  Regional meetings that bring together school and district leaders across school systems to deepen understanding of results, share best practices and identify actions to be taken in response to results to improve student outcomes; [and]  An annual conference where hundreds of leaders gather to deepen their understanding of their results, learn about best practices, make connections with peers involved in this work and gather information to identify actions to take to improve student outcomes.(America Achieves, 2017, no page number) What is most apparent here is the absence of any meaningful problematization around the implied causality between 'best practice' and performance ('best practices in world-leading schools'; 'practices that have contributed to world-leading performance'), which largely ignores the inherent dangers in making unsubstantiated causal links between performance and in-school policy settings (see Alexander, 2012;Gorur, 2016;Meyer & Schiller, 2013).Adding further to the resonances between the OECD and GLN is that these 'best practices' are sourced exclusively from schooling systems that demonstrate 'consistently high results' on the main PISA test (OECD, 2012).As such, GLN provides the means, in effect and intent, to give further 'institutional force' (Ball & Junemann, 2012) to the policy utterances of the OECD, thereby promoting 'best practice' as something that can be readily transferred between international systems and schools, irrespective of local context.This rationale also invokes the idea of governing by examples (Simons, 2015), where specific 'soft' forms of educational evidence ('best practice') are used to help steer local possibilities for action and reform, but where improving 'hard' numbers and data provides the impetus for employing such practices.
The influence wielded by the OECD, in terms of deciding which policies and practices should be emphasised within GLN, extends far beyond providing examples of 'best practice' within the PISA for Schools report.Despite the OECD being one node, albeit a significant one, within the PISA for Schools policy network, it still arguably maintains considerable control over the learning activities of GLN.Indeed, when I questioned an America Achieves manager on the OECD's influence over GLN activities, they candidly revealed that the objective for America Achieves is to 'align with what the OECD thinks is right': Absolutely.I will be candid and say it's not official.Andreas [Schleicher] has not said to us, 'We need to approve everything you do'.It's more that they [the OECD] are the experts on this [and] they created the assessment.And so before we do any virtual convening that touches upon analyzing test results, before we do any regional meetings, they always review the materials.Between our virtual convenings and regional meetings in-person, we've had about 15 different 'interactions' [as of January 2015], and the OECD has been present for most of them and reviewed materials for all of them, so they're a really critical partner in this work.And I think everything we do, we want to make sure that it aligns with what they think is right.(Emphasis added) The purposeful seeking of the OECD's support for all learning activities undertaken in GLN provides a clear demonstration of 'soft power' (Nye Jr, 2004) in educational governance, with America Achieves encouraged to promote the broader education policies of the OECD through 'techniques of persuasion, attraction and seduction' (Williamson, 2016, p. 133).While there is no compulsion for the OECD to pre-approve the learning materials and convenings used in GLN ('it's not official'), the willing submission of America Achieves to this process of 'expert' review ('they always review the materials') implies that the OECD retains a critical, and arguably dominant, influence in this work amongst the partner organizations that facilitate the development and delivery of PISA for Schools (see also Lewis, 2017b).This soft steering implies that there is no substantive challenge to the OECD's rendering of 'what works', thereby potentially positioning GLN more as a mouthpiece to normatively promote the policies and discourses of the OECD, and less as a truly critical partner for meaningful professional development.
Beyond 'unofficially approving' the convening materials created by America Achieves, the OECD's 'soft' steering of GLN also includes promoting its own research publications to school and district leaders, creating 'recommended reading' lists for local educators and policymakers.Although the 160-page PISA for Schools report received by participating schools already makes extensive use of OECD research and publications (see Lewis, 2017a), this self-referencing is promoted further still by GLN encouraging local educators and policymakers to examine OECD texts as part of their ongoing professional development.An America Achieves manager spoke openly of such partiality in GLN, and the clear extent to which OECD publications were ultimately preferenced over other possible renderings of evidence and schooling reform: We have a monthly newsletter in which we always highlight new articles or publications, and it's typically an OECD resource in there.We also, when we do different sessions, will reference different OECD documents...One of the things that we shared that got a lot of clicks were those fantastic videos that the OECD did in partnership with Pearson ['Strong performers and successful reformers'].(Emphasis added) Thus, even as school and district leaders freely meet under the aegis of GLN to share their insights from PISA for Schools, there are unmistakable discursive limits to what can be said and who can say it, outside of which lie non-OECD and non-PISA utterances.This largely restricts the 'conditions of possibility' (Foucault, 1994), or the 'possibilities of policy' (Ball & Junemann, 2012), for how local schooling reform can be conceptualised, and enacted, to that 'authorized' by the OECD.In this constraining of local actions, we can see clear resonances with what Ball (1993) describes as a 'moving discursive frame' (p.15) of policy, in which possible ways of understanding and practicing education are limited to those made available by the language of the dominant (read: PISA) discourse.As the America Achieves interviewee noted above, there is little apparent concern for how the discursive permeates and 'performs' practice, and here we can see how epistemological modes of governing are central to the OECD's global governance of education (see Sellar & Lingard, 2014;Woodward, 2009).
Fostering a Data-Driven Teacher Disposition: Creating 'PISA-Based People' Over and above advocating a particular discursive construction of 'performance' (i.e., that which can be measured by PISA) and 'best practice (i.e., that which leads to improvements in measurable performance), a central role of GLN is to produce a particular data-driven disposition amongst the local educators using PISA for Schools.Indeed, there was a palpable sense amongst the participating teachers and administrators that they considered themselves to be more progressive, globally minded and forward-looking than their non-PISA counterparts.One U.S. principal even went so far to suggest that such attributes were indicators for 'PISA-based people', and that these data-driven individuals were the most desirable (and employable) teachers: The other thing I'd like to say is that for me, in terms of hiring and sitting down and choosing staff, I want to hire teachers that get this [PISA for Schools] … It is the future of education and I want it to actually make their minds kind of click too, and I sort of feel like they can lead the kids in this learning by sort of being 'PISA-based people' themselves.
(Emphasis added) This ably demonstrates how PISA for Schools, and performance data more generally, is influential in processes of teacher subjectivation and the constitution of schooling practices more broadly (Ball, 2003;Comber, 2012;Hardy & Lewis, 2017;Thrupp, 2013), creating educators that can be largely described as 'numbers people' (Hardy, 2015).Even though PISA for Schools is notionally 'low-stakes', in the sense that it is not explicitly linked to evaluations of teacher tenure or individual teacher performance, it is interesting to note how some school leaders clearly might choose to use favorable teacher dispositions towards international testing and evidence when hiring ('I want to hire teachers that get this').In this sense, the lower stakes nature of PISA for Schools is perhaps elided by the dominance (and prevalence) of the data-driven and data-responsive logics that underpin such comparative assessments.
In fact, the GLN's school recruitment website explicitly suggests that PISA for Schools participation is a useful way to identify U.S. educators who are seriously committed to learning from and implementing international 'best practice', and to mark those who, by default, are not: Members of the Global Learning Network are educators and district leaders who take the OECD Test for Schools to demonstrate a willingness to embrace the truth about how well they are preparing their students for our changing world.[They also] commit to learning best practices from others -nationally and globally -and implementing those practices to attain high levels of student achievement.(America Achieves, 2014a, no page number; emphasis added) There are several interesting (and questionable) values being promoted here, foremost of which is that PISA for Schools is positioned again as somehow providing the definitive version ('the truth') of a school's performance, which diminishes other possible ways of articulating local forms of educational accountability.In the context of the teaching subject, it also depicts PISA for Schools participants as those who are willing to leave the 'darkness' for the 'light' of PISA learning, as though a desire to compare performance and implement 'best practices from others', and especially those of 'high performing' international schooling systems, is a prerequisite for being an effective educator.Such constructions of the teaching subject create a normative binary, juxtaposing those teachers (and schools) who shun the international with the (supposedly) progressive, effective and outward-looking 'PISA-based people' who don't.Indeed, this is a teaching subject largely intent on discovering 'what works' that is free from any contextual considerations, despite the repeated exhortations of comparative education scholars to avoid such unsubstantiated casual inferences (see Biesta, 2007;Gorur & Wu, 2015;Simola et al., 2013;Steiner-Khamsi, 2010;Tan, 2012).

'This is About Collaboration': Towards a More Educative Appropriation of PISA for Schools?
Of course, this reading of PISA for Schools and GLN as merely facilitating the OECD's global governance of education is one that largely overlooks local attempts to use these professional development opportunities for more broadly educative purposes.Potentially, a more instructive disposition amongst teachers and district leaders might instead represent what Hardy (2014) describes as a 'logic of appropriation'.Here, tools of educational accountability (such as PISA for Schools) are actively engaged with to facilitate meaningful student learning and professional development, rather than teachers being seduced by (or succumbing to) the performative demands of such technologies.By highlighting this appropriation, I am not purposefully ignoring the clear discursive constraints imposed by the OECD, its partner organizations and learning communities like GLN upon how schooling is locally understood and practiced.However, the interviews I conducted with school and district personnel would suggest that local U.S. educators can (and often do) use PISA for Schools to encourage alternative policy spaces -perhaps reminiscent of Foucault's (2007) 'cracks' and 'fissures' -in which productive professional dialogue, and more 'authentic' modes of accountability, might be possible outside of dominant discursive regimes.Despite the apathy with which many U.S. educators regularly engage with test-driven accountabilities, many district and school-level personnel were enthusiastic about participating in PISA for Schools, especially when it was considered alongside existing modes of national and statelevel student testing.Given that PISA for Schools has no mandated public release of student results and no official sanctions for schools (or teachers) that fail to reach performance benchmarks, many educators felt that the 'low-stakes' nature of the assessment could more readily drive instructional conversations that transcended mere test scores.In turn, they believed this would avoid the 'name and shame' reaction that many existing high-stakes tests often elicit, especially when these are combined with the highly publicized release of school performance data (Keddie, 2013;Smith & Fey, 2000).
For instance, a U.S. superintendent explained how she/he felt PISA for Schools data could help foster the transition from focusing on local test scores to 'rich discussions' about student (and teacher) learning: One of the things that PISA for Schools helps reinforce is … [moving] away from the 'I got you!' mentality to 'How can we learn?' That's one of the most powerful transitions that [are] a result of a high accountability system.The first phase of it is always, in every country of the world it seems to be, 'Oh my God, who's that?' and 'I got you', and 'Why are you at such a low performing school?'When you can get past that -that's when you get into very rich discussions about schooling and learning, and not about test scores.(Emphasis added) The benefits of this (potentially) more educative approach were also noted by a U.S. assistant superintendent, as PISA for Schools data could not be used by educational authorities -be they district, state or national -to hold individual students and teachers to account.As such, the absence of sanctions and other punitive measures (staff replacement, school closure) meant that conversations were more inclined to be about instruction ('content, curriculum, level of rigor'), rather than accountability ('we got you!'): It's not high-stakes for the kids.So then we can talk about content, curriculum, level of rigor and that type of thing, as opposed to, 'Are you an exemplary school or an unacceptable school?'And 'we got you!' on this, and now you have all these interventions and these sanctions, and that kind of stuff… I'm hoping we can keep it lower-stakes so it can feed more the instructional conversations, as opposed to the accountability conversations.(Emphasis added) This prospect is a welcome contrast to the dominant 'Anglo-American approach' to top-down, testdriven accountabilities (Lingard & Lewis, 2016), typified by the census testing of all students that attained especial prominence in the USA under the No Child Left Behind (NCLB) and Race To The Top (RttT) initiatives.In fact, it is clear from the statements above ('I got you'; 'why are you at such a low-performing school?';'are you an unacceptable school?') that the prevailing U.S. focus on 'highstakes' performance data has produced considerable performative demands, and affective impacts, on local educators (Berliner, 2013;Hursh, 2013;Lipman, 2013).Consequently, the 'lower-stakes' nature of PISA for Schools, in both intent and perception, may encourage local educators and policymakers to see the creation of school accountability data as merely the start of a professional dialogue, rather than a more performative end.In this sense, PISA for Schools might possibly be considered more an assessment as and for learning, rather than being an assessment of learning that merely measures performance outcomes, and hence somewhat more educational than accountable in nature.While the types of conversations enabled by PISA for Schools may well be inclined to follow the discursive limitations of the OECD, such as the evident focus on the identification of generic 'best practice' and 'data-driven continuous improvement', there is nonetheless a clear possibility for more productive 'teacher talk' to emerge from the Global Learning Network.However, and even with this more nuanced approach to school performance accountability, the limiting assumption that testing, ipso facto, improves learning and performance remains a significant position in U.S. education discourses, especially amongst national policymakers vis-à-vis the prevalence of testing in government policies like RttT.This is despite evidence suggesting that such an emphasis merely improves student test-taking and strengthens teaching to the test (Berliner, 2013;Hursh, 2013).
Neither should we dismiss out of hand the admittedly voluntary nature of PISA for Schools, with many local educators considering it a valuable tool for promoting professional collaboration and learning, within both 'sanctioned' learning communities like GLN and other more ad hoc constructions.Indeed, a U.S. superintendent whose schools participated in the pilot study enthused that: … PISA for Schools is by far the best vehicle that I know of to really prompt professional conversations between districts… PISA for Schools is at this stage a very voluntary thing.I mean, no state, no organization has twisted the arms of any group of people that I know about and said, 'Thou shalt take PISA for Schools!' The people who have done it are people who have sought it out, probably believe that it's valuable and want to learn from it.(Emphasis original) Such ringing endorsements from an experienced and presumably dedicated educator cannot, nor should not, be easily discounted, especially if we are to value the professional and contextual knowledge that educators clearly possess, and also their ability to employ this knowledge in the best interests of their students and schools.Although I have highlighted elsewhere the potential dangers to curriculum and pedagogy if PISA for Schools is used to normatively borrow global 'best practices' (Lewis, 2017a;Lewis et al., 2016), benefits may still emerge simply by virtue of local educators using the test -in conjunction with other forms of evidence -for diagnostic purposes, and for sharing their experiences and insights.This professional learning would occur between sites of similar or different socio-cultural backgrounds, but in a way that recognizes, and seeks to accommodate, the nature of these differences and how they necessarily influence educational practice.Such a vernacular approach aligns with what Kemmis and colleagues (2014) describe as 'site-based education development', in which education, and educational practices, are developed in response to the local needs and circumstances of each specific schooling site, rather than being downwardly imposed from government authorities or, for that matter, intergovernmental organizations like the OECD.We can perhaps see early attempts at such 'site-based' peer-to-peer dialogue and professional development through the various activities of GLN, including 'regional convenings' -physical, faceto-face meetings of school and district staff from the same geographic region or state -and the nation-wide 'virtual convenings' of PISA for Schools participants in the USA.For instance, the agenda for the June 10, 2015, virtual convening of GLN reveals that such peer-to-peer sharing and dialogue between individual member schools is a central activity of the community: Representatives from Kohler High School (Kohler School District, Wisconsin) and Berkmar High School (Gwinnett County Public Schools, Georgia) will highlight the practices that they feel contributed to the strong achievement in science at their respective institutions, in addition to sharing the steps that they are taking to continue improving student outcomes.(America Achieves, 2015, no page number; emphasis added) Although somewhat problematic for its assumed causality between certain practices and student performance, which in turn invokes a 'treacherous leap' from data to policy (Gorur & Wu, 2015), this sharing nevertheless has the potential to benefit schools and educators, especially when contrasted with using the more abstract system-level results from main PISA as the basis for informing local practice.Furthermore, there appears to be at least some acknowledgement here that local context does matter, insofar as U.S. schools are learning from other 'similar' U.S. schools and not contextually distinct international schooling systems (e.g., Shanghai-China, Finland).
In spite of the continuing focus on 'high performers' and 'best practice', there is at least a sense here that comparisons with more contextually-similar school peers may well produce more meaningful opportunities for policy learning (see Meyer & Schiller, 2013).This may also go some way to overcoming the reductive effects of PISA for Schools, and the elision of local historical and socio-economic factors.Thus, I would argue that there are conceivable benefits from local educators sharing their experiences and insights within professional networks in ways that help to acknowledge the importance of local context, even if these conversations are largely being held in, or at least instigated by, the language of PISA and its reductive focus on reading, math and science.

Teachers 'Talking Back': Using PISA for Schools to Challenge Schooling Authorities
Perhaps most importantly for driving more enduring (and authentic) processes of policy change, PISA for Schools potentially empowers local educators to 'speak back' to their state and federal authorities.This represents a topological 'reaching out' (Allen & Cochrane, 2010) by schools and districts into the political centers of State power to influence events occurring within their own local schooling spaces.In response to the more common imposition of 'top-down' accountabilities and policy discourses from governmental (or non-governmental) policy actors, PISA for Schools may instead enable local communities and educators to petition their respective education authorities to enact progressive change in more 'bottom-up' ways.This opportunity to 'talk back' was clearly not lost on one particular U.S. superintendent: If there was PISA for Schools data out there, how will a state authority resist pressure from local communities who say, 'I'm looking at my PISA for Schools data and I'm looking at somebody else's.Why are you inhibiting certain conversations?Why do we have to wait for you to make a decision at the state level before we proceed with certain policies?'I think it will empower local organizations and local communities that want to take power to force questions with high-level authorities.(Emphasis added) Here we can see the intentional 'reaching out' by schools and districts via PISA for Schools to the OECD, and other schooling sites through GLN, to make new local actions possible, including by potentially influencing the educational conduct and policies of national (or state) schooling authorities.Notwithstanding the geographical distance between the OECD and other schooling systems, the relational (or 'topological') proximity afforded by PISA for Schools may help local educators and policymakers to draw upon comparisons with elsewhere to drive purposeful conversations with their own education authorities.Of course, it must be noted that this petitioning of national and state authorities assumes that these authorities are indeed still relevant to the school.Greater moves towards Charter schools in the USA, in the context of increasing neoliberal and libertarian agendas (Kretchmar, Sondel, & Ferrare, 2014;Lubienski, Brewer, & La Londe, 2016;Tanner, 2013), and the very presence of the PISA for Schools test itself, would appear to suggest that schools (and teachers), rather than systems, are increasingly positioned as the most important educational unit.This means that national and state-level authorities may, in the context of this enhanced global-local nexus, have a diminished ability to intervene in all schools, with this logic presuming that schools and teachers are, by extension, solely responsible for the performance of their students, outside of broader socioeconomic and cultural influences.While these developments may enhance the abilities of some schools to exercise greater local agency, this also has obvious implications for the ability of systems to work in a redistributive way to address issues of equity, access and resourcing.These moves towards greater local autonomy would thus seem to benefit already more advantaged and affluent school communities -which arguably reflect the majority of U.S. schools and districts that have to date participated in PISA for Schools -while possibly being to the detriment of others, an issue further compounded by PISA for Schools being a voluntary, user-pays assessment.
Also relevant in this context are the increasing moves in the USA for local forms of resistance to the 'top-down' imposition of 'high stakes' student testing.These have recently included a parental 'opt-out' of tests in New York State that is supported by national teachers' unions (e.g., The American Federation of Teachers) and investigative journalists (Hursh, 2016;Singer, 2015), and teacher boycotts in Seattle following the introduction of standardized Measure of Academic Progress testing (Zeichner, 2013).Invoking this resistance to the imposition of 'topdown' testing, an educational consultant revealed how school and district leaders saw PISA for Schools as an opportunity to replace existing, and (in their eyes) flawed, educational accountability measures with something allegedly 'superior': In Virginia and New York, the superintendents are gung-ho about going to their state's Department of Education and saying, 'Please, dear God, allow us to substitute PISA for Schools for the idiotic, inane, weak state-level high school [state] standards assessment.You want accountability -fine!We'll provide you the accountability data you want for our schooling but what we'll provide you [with] is PISA for Schools, which would be vastly superior'.These local educators have readily positioned PISA for Schools as a welcome alternative to existing testing regimes and, relatedly, the State's dominance of schooling policy, providing the impetus for a possible 'exit' (Hirschman, 1970) for schools from other (apparently) more performative assessments, modes of accountability and relationships with central schooling authorities.It is also interesting to note that it is not so much the demand for accountability data per se that is being criticized, so much as it is the type of accountability data ('idiotic, inane, weak state-level high school standards assessments') that the State is demanding.
While this ostensibly sees schools and districts confronting one set of dominant power relations (i.e., state and national authorities) with another (i.e., the OECD), and despite these discussions being conducted in the 'language of PISA', the value of empowering local educators and communities to exercise their agency -possibly their 'exit' but particularly their 'voice' (Hirschman, 1970) -should not be underestimated.This is particularly so if it encourages a greater 'data legibility' among teachers, parents and citizens, which helps to build democratic enfranchisement and involvement in debates around how education could be alternatively rendered (see Biesta, 2004;Henig, 2013).Ranson (2003) notes how such pluralistic discursive exchanges (and even contestation) are integral, and not somehow peripheral, to the constitution of a public domain like education, as legitimacy can only ever be truly achieved through extensive democratic deliberation.In this way, PISA for Schools may well help some local educators provide different ways of being accountable and doing accountability.

Discussion and Conclusion
There are many inherently value-laden questions around the purpose of education that require deep contemplation, rather than the 'knee-jerk' reactions that often accompany the publication of performance league tables, and especially when these comparisons are based on narrow and decontextualized conceptions of 'what counts'.As a complex social practice, education needs to be treated complexly, and not in a way that reduces this complexity for the sake of facilitating comparison, policy and governance over the education of students.There is also the need to complement such educational accountability with meaningful two-way horizontal relationships between schools and their communities, thus creating a multilateral 'answerability' (Ranson, 2003) and level of expectation that holds all stakeholders -parents, teachers, local communities and schooling systems -'to account'.Such a reconstitution of educational accountability would give schools (and their communities) the capacity to demand of their systems and policymakers that schools (and their communities) are provided with the necessary resources, both human and otherwise, to achieve what is demanded of them.
What I would argue here then is that outright opposition to accountability per se is clearly not a justifiable position, nor one I would advocate.Rather, we need a more progressive reconceptualization of accountability -such as the 'rich accountabilities' model proposed by Lingard and colleagues (2016) -that acknowledges the broader societal purposes for schooling, and which holds systems, schools and communities accountable in 'bottom-up' and 'top-down' ways.While it is perhaps too early in the lifecycle of PISA for Schools to see whether it can actually enable such a reconceptualization, and hence my opinions here are largely speculative at this point, this is arguably a key way in which such comparative assessments might exercise a more educative, and less performative, effect on local schooling practice.This is not to say that PISA for Schools -and the focus on individual school-level performance rather than sub/national system-level performanceovercomes all, let alone some, of the many and varied problems that have been associated with main PISA, or that providing schools with an opportunity to be represented as yet another data point is something that we should advocate in itself.However, if PISA is to be held aloft as one of the numbers that most count, and if schools and educators are to be held to account as the result of such data, then data that is (more) relevant to and acknowledging of local context can be seen as, at the very least, somewhat of an improvement.If we are ever to realize something beyond the present (and exceedingly dominant) articulations of top-down, test-based educational accountabilities, then providing data that is less high-stakes for teachers and students, less abstracted from the classroom, and (potentially) more enabling of productive professional learning is a good first, albeit preliminary, step in the right direction.
Rutkowski and colleagues (2014) have posed the question, Should individual schools participate in PISA, although this was admittedly underscored by a search for 'lower-cost alternatives' to schooling accountabilities in the context of 'shrinking state budgets ' (p. 71).To this pressing question, I would also ask how and to what end should schools seek to position themselves within the global commensurate space of measurement constituted through PISA.If PISA for Schools merely serves as a vehicle for 'high performing' and privileged institutions to promote themselves as being somehow 'better than Finland', in order to aid their already considerable advantage, or to normatively borrow practices from radically different schooling systems or societies (see Lewis, 2017a), then clearly this is problematic.However, the comments of educators and policymakers above would suggest that PISA for Schools and GLN, at the very least, does indeed help to promote professional collaboration, and a form of 'PISA learning' that acknowledges the importance of context to rethinking how schooling can be locally thought and practiced.
It must be emphasized, however, that this potential does not efface the many normative 'PISA effects' regularly visited upon schooling policy and practice by such large-scale comparative assessments (Breakspear, 2012;Grek, 2009;Lewis & Lingard, 2015).Nor does it suggest that, on balance, the drawbacks of PISA for Schools are exceeded by its potential benefits.On the basis of my analyses, it is conceivable that the many such opportunities for professional learning will be subordinated to the OECD's discursive construction of education performance, where 'low' comparative performance serves merely as the catalyst for the decontextualized, and problematic, adoption of 'best practice' from 'high performing' schooling systems (e.g., Shanghai-China, Singapore, Finland).Neither should one assume that the replacement of the State's top-down educational authority with the horizontal one of the OECD, or professional learning communities like GLN, automatically resolves all potential imbalances of power around the thinking and practice of education.Foucault (1983) presciently suggested in relation to alternative solutions 'not that everything is bad, but that everything is dangerous' (p.231), with the implication that each possible rendering of education policymaking and governance, be it 'top-down' or 'bottom-up', will necessarily bring into existence its own unique set of challenges and problematics.
Whatever else can be said, my analyses highlight the possibility of PISA for Schools, and associated bodies like GLN, being 'productive' -a term deployed here knowingly (Foucault, 1978) and giving rise, potentially, to more meaningful professional collaboration and dialogue between local educators, policymakers and communities.In light of the potential benefits outlined by the educators above, it is perhaps more profitable then to seek a 'productive balance' between the normative and performative demands of enumerative assessments like PISA for Schools, and their other, more tentative, educative opportunities.education policy analysis archives editorial board