Making All Children Count : Teach For All and the Universalizing Appeal of Data

In this paper, we argue that in order to bind Teach For All’s universal/izing statement of problems and solutions to the specificities and the special conditions of member programs’ local contexts, what is needed is a shared set of discursive practices, a way of bringing together the commonalities found in each country while separating the noise of particular politics and histories. That common set of discursive practices is shaped around the notion of data. This paper is structured as follows: First, we contextualize Teach for All by (briefly) juxtaposing the universal and specific elements of the network, including the organization’s mission, target population, its recruits (and recruiting tactics), vision, and its production of a particular kind of teacher. Then, we present the two competing, yet complementary, logics of data that are at play in Teach for All – the use of data itself and the notion of data speak – along with their epaa aape Education Policy Analysis Archives Vol. 23 No. 48 2 underlying assumptions. We conclude by questioning the logics of this set of discursive practices, and outline our skepticism regarding how data is mobilized to produce particular subjectivities and objects.

specificity lies a collective claim of universal problems, with universal solutions, backed by a universal notion of justice.In the words of Teach For All's CEO Wendy Kopp: "If our problems are the same, then it gives me hope that the solutions must be sharable" (Cited in Murlidah, 2013).The solutions put forth by Wendy Kopp and Teach For All are clear-cut and center around the organization's firm assertion that the teacher is the most important variable for improving education and changing peoples' lives and destinies (Dee & Wykoff, 2013;Hess, 2006;Kopp, 2003).By providing quality teachers in a way that is engaging and cost-effective, Teach For All positions itself as revolutionary, subversive of the status quo, and all the while, commonsensical.
However, less clear-cut are the shared problems Teach For All aims to solve.Despite sweeping claims that schools (globally and in the US) are failing society and especially under-represented groups, the actual substance of the shared problems remains ambiguous.Given the shaky foundations of a problem posed as universal in settings as diverse as South Africa, Pakistan, Germany, and Israel, some questions emerged for us: what is the system of thought that allows for a problem to be posed as universal, yet have specific appeal in such different locations?What is the language in which this problem is expressed?What kinds of teachers do these discursive practices contribute to producing?
In this paper, we argue that in order to bind the specificities and the special conditions of each context to the universal/izing statement of problems and solutions that Teach For All proposes, what is needed is a shared set of discursive practices, 2 a way of bringing together the commonalities found in each country while separating the noise of particular politics and histories.That common set of discursive practices is shaped around the notion of data.Inspired by Gitelman's (2013) work in media history, we define data as the "units or morsels of information that in aggregate form the bedrock of modern policy decisions by government and nongovernmental authorities" (p. 1).We do not believe that data speak for themselves, or that the utterances of data are clear and unambiguous.
This paper is structured as follows: First, after a brief description of our methodological approach, we contextualize Teach For All by juxtaposing the universal and specific elements of the network -the organization's mission, target population, its recruits (and recruiting tactics), vision, and its production of a particular kind of teacher (Friedrich, 2014).Then, we present the two competing, yet complementary logics of data that are at play in Teach For All -the notion of data speak and the use of data itself -along with their underlying assumptions.We conclude by questioning the logics of this set of discursive practices, and outline some implications regarding how data are mobilized to produce particular subjectivities and objects.

Methodological Approach
In an effort to begin answering the questions posed above, we turned to the global social media site Twitter.As real time data aggregator, Twitter acts as a digital archive, capturing the opinions and commentary of its 234 million active users. 3Among these users are the Teach For All Network, its thirty-four partner organizations, and a growing global teaching corps.We examined their accounts to better understand how the Teach For All message of one problem, one solution is discussed, taken up, and disseminated.Specifically, we collected all of the tweets from the accounts 2 "Discursive practices are characterized by the delimitation of a field of objects, the definition of a legitimate perspective for the agent of knowledge, and the fixing of norms for the elaboration of concepts and theories.Thus, each discursive practice implies a play of prescriptions that designate its exclusions and choices."(Foucault, 1977, p. 199) 3 https://about.twitter.com/company of the Teach For All Network, Wendy Kopp, all Spanish and English-speaking member organizations, and the CEOs of these organizations for the month of October 2013, coinciding with the annual Teach For All Global Conference. 4 In all, we examined 26 Twitter accounts and approximately 1,500 tweets and retweets.We combed through all the documents and files the tweets linked to, as well as the programs' websites.We then coded for emerging patterns, and using Nvivo10, we focused on the ways in which notions of data were mobilized as legitimating tools for posing specific problems.
Given the stance we take on data throughout this article, we do not consider the tweets and other documents as inherently truth-telling, thus the categories do not emerge from the data themselves, but from our interactions with them.We ourselves mobilize data to tell a story, to explore an issue that concerns us in relation to the ways in which educational change is framed and discussed.These discussions extend beyond the Teach For All network and have implications for the field of education writ large.

Teach For All and the Universal(izing) Language of Data
In order to understand how this discourse of data works, it is important to contextualize the conflicting and complementary ways in which Teach For All presents itself.Teach For All's programs target specifically underrepresented populations, in low-resourced schools, with high teacher turnover, and low academic achievement.The challenges facing these students, Teach For All argues, are systemic and pervasive, and include everything from economic inequalities, to racism and sexism, to language differences.In India, for example, educational inequality is linked to English language skills and widespread poverty.For those yet committed to the program's message, Teach for India retweets, "Why am I doing what I am doing when I know 800 million in this country can't afford an ice cream?" (Teach for India, 2013).Equally remote figures are mobilized across Teach For All in order to establish an urgent need for global change.For instance, Teach for Canada, one of the newest partner organizations, plucks statistics about indigenous populations to paint a dire image of the country's educational landscape.Decontextualized statistics like "Only 3% of First Nation students (on reserve) will graduate from university" (Teach For Canada, 2013) appear daily across the program's Twitter feed.The implied solution: Teach for Canada.Teach For All proposes to contribute to solving these issues with a way to cut through these barriers to educational justice: placing high-quality teachers in the schools that need them the most.
Teach For All, with its message of increasingly globalized educational crises, envisions a network of partner organizations as the most efficient and effective way to address what has become a common concern.While this network is promoted as a way to more easily share Teach For All's best practices and develop a sense of community between corps members, much of the rhetoric mirrors talk around the economies to come, whether they are called the Networked Society or the Data/Knowledge Economy (see for instance Castells, 2011).Tweets across the Teach For All network depict a rapidly changing and increasingly connected world, one that not only risks leaving underprivileged kids behind, but also eludes many in the traditional business sector.With its network of member programs and their multi-national sponsoring corporations -Credit Suisse, Deutsche Post DHL, Google, The Bezos Family Foundation, and Visa, to name a few -Teach For All invites recruits to join the ranks of the forward-looking global citizens.This not only means offering mini-courses on leadership skills and establishing connections with executives at the world's largest firms, but also connecting Teach For All's data-driven teaching strategies with the entrepreneurial mindset required to make it in the "real world."In some ways, their approach seems to be working: Teach For All corps members are recruited from fields outside of education, and -at least according to program affiliates -they go on to land high-profile jobs and are behind some of the world's leading educational technology start-ups (Kopp, 2014).These include Khan Academy, Be a Smarter Cookie, Edthena, and Kickboard, what Wendy Kopp (2014) describes as an "app that's as easy to use as Facebook and lets educators track and share data about their students" (para.12).The primacy of data across Teach For All's internal operations and marketing campaigns forms a feedback loop where global promises to transform educational data hinge upon data-dependent conceptions of teaching and learning.With the upsurge of international data gathering and Teach For All's deepening infrastructure, corps members are positioned at the forefront of the educational data market.
The development of these skills will not only help corps members in their future careers, but also develop them into specific kinds of teachers.This is a teacher who consciously and methodically makes a difference, accepts no excuses, and has the data to prove it.With the kind of tenacity our rapidly changing global economy demands, the corps member takes on a challenge before which so many of her predecessors have failed: combatting the odds stacked against disadvantaged children.In order to do so, the corps member sacrifices, for two years, more certain and material rewards, embarking in a crusade for justice.She or he does not do so alone, but is supported by a community of like-minded individuals whom s/he should rely on, without losing faith in the power of one (see Ahmann, this volume).This teacher should gather and analyze data, using this activity to improve practice, scaffolded by proven methods for capturing the minds of students.The teachers' main focus should be student learning, as demonstrated by empirical preand post-intervention data.It is only through such evidence that teachers can know for certain whether or not students have mastered a specific skill.Student data that emerge within the classroom are made increasingly legitimate and scalable when couched within Teach For All's broader rhetoric of universal educational inequality, such as its frequent references to glaring PISA scores.
The use of PISA serves to exemplify the ways in which the special conditions of each context are bound to the universal/izing statement of problems and solutions through the common language of data, bringing together the commonalities found in each country without regard to local idiosyncrasies (Popkewitz, 2011).Reading the world through data allows for visions of both scalability as well as the highly personalized and adaptive teaching necessary for succeeding in the 21st economy.
The notion of data is not that of mere numbers 5 .Data present us with the operationalization of human and non-human interactions into ways of measuring and informing interventions and policies.In this sense, the datum itself is an intervention, a transformation of sorts.Data transform not only the phenomena they are trying to measure, but the agents involved as well.For instance, the call for more data from inside the classroom requires educational artifacts, such as an exam, to be transformed into a homogeneous measuring tool that can then be transposed onto a cell on a table that "says" something about what is and is not working in that interaction, an approach that requires the teacher to become a scanner.That is, since data is not merely there for the taking, the teacher must scan the classroom for possible data points to be captured and interpreted as data.In other words, data require our participation; data need us (Gitelman, 2013).Therefore, the very interaction between teacher, student and curriculum is transformed into something to be read -into something that is either intelligible as a source of information for policy-making, or as something not translatable into data and thus discarded.
In our current Era of Big Data (Gitelman, 2013), the push for the production and use of "big data"6 in education can be understood as emerging from a confluence of discourses and technologies that transcend the field (MIT Technology Review, 2013).Without going into much detail, we first argue that this push would be unthinkable outside the framework provided by the economization of the social sphere that founds neoliberalism (Ball, 2009;Foucault, 2008;Taubman, 2009).When educational policy and pedagogical relations are read through an economic lens, they are disassembled into individuals who make rational choices guided by their own self-interest (which, as Friedman and Friedman (1990) themselves point out, is not necessarily monetary).In order to make the best possible choice, be it a choice between what sort of teacher education is preferable for a particular population, what method is best for teaching fractions, or what school to send our children to, information becomes paramount.Social justice, within this framework, is about providing all individuals with the most transparent, unbiased and clear data available so that they can in turn make the best possible choices for themselves, letting meritocracy determine the fate of rational individuals.
This move towards a neoliberal governmentality has been influenced by what Camicia and Franklin (2011) term a neoliberal cosmopolitan discourse that emphasizes a global community that is best related by market rationality, as well as a growing trust in numbers, to use the title of Porter's (1996) influential work.According to Camicia and Franklin (2011) "students and workers are most efficiently related through technologies of standardization, surveillance, and accountability" (p.314).Porter (1996) discusses quantification as a technology of distance, which minimizes the need for personal trust and intimate knowledge, aiming at producing knowledge that is independent of the subjectivities producing it.Two aspects of Porter's (1996) analysis are especially relevant for the call for more data mobilized by Teach For All, which we will outline hereafter.First, Porter (1996) analyzes the invention of phenomena by the knowledge about it, providing the historical examples of crime rates or unemployment rates, to which we could add the at-risk student or the lowperforming school.The fact that these phenomena are "invented" does not mean that these categories are weak.On the contrary, once they become official, they become increasingly real, with real effects on people's lives: "Public statistics are able to describe social reality partly because they help to define it" (Porter, 1996, p. 43).The second aspect is the unique role that quantification has had in the American context, where this technology of distance becomes deeply intertwined with the ideals of a democratic society.Whereas subjective judgment and expert discretion were seen as prerogatives of the elites allowed to make those judgments, quantification led to objective knowledge seen as public and universal.It made possible the construction of rules and procedures as ways to neutralize individual agency by making anyone capable of learning said rules without the need to understand the phenomena: "The ideal is a withdrawal of human agency, to avoid the responsibility created by active intervention.Subjectivity creates responsibility" (Porter, 1996, p. 196; for the relationship between responsibility, subjectivity and justice, see also Derrida, 1990).
The third strand that contributes to the push for the use of big data in education -and this is both a cause and effect of the push for data -consists of the technological advances made in the artifacts available for collecting, processing and communicating big data in the last two decades.This is a case in which fields of knowledge have created the technologies that produce the kind of data that have themselves invented new fields of knowledge.In other words, and taking Porter's (1996) idea of quantification inventing new objects much further, what we have here is the emergence of technologies that create new ways of studying, which create new objects and technologies, in a feedback loop that fills university departments and libraries (for a historical analysis of this phenomenon, see Kittler, 1999).In many ways, it is this very newness -and the explicit erasure of historical reference -that makes these technologies and their corresponding ways of knowing so valuable.
One phenomenon that is central to all of this is the emergence of a new privileged agent for social change.The distrust in large bureaucracies -both from within and from without the Statehas combined with a concern about growing inequalities and a desire for immediate remedial action to channel the age-old discourse about change from below (from the "grass-roots") into the active production of a new subjectivity: that of the social entrepreneur (Ball, 2012;Friedrich, 2014).The social entrepreneur channels his/her willpower, capital (social, cultural, and economic), and unique disposition into innovative and risky approaches to solving issues; they experiment, then scale up whatever works.As much as the Teach for All organizations work to recruit teachers, they also frequently remind us how few recruits make the cut, for example by linking to an article that states, "A dozen Teach for All partner organizations (including ones in India, Israel, Argentina, Australia, and Spain) accept 10% or less of their applicants" (Sellers, 2013).What is not required of these social entrepreneurs is expertise in the field of their intervention -they are prized as moveable pieces within the field of social change.Take for instance a tweet in which Teach for Bangladesh mistakenly links to a critical Politico article on Teach for All with the misleading hook, "Have a read on how TFA is ensuring that the "movement" is in fact across a variety of sectors" (2013, October 26) or when Teach for Pakistan hyperbolically tweets, "Thomas Friedman talks about #TFAllGC2013 where he said being with 32 CEOs in @TeachForAll was like being w 32 Malalas" (2013, October 29)."Being a CEO" is "being a CEO" is "being a CEO," regardless of the focus of the organization.This is where the universal/izing appeal of quantification and the push for big data come in, for -with the right disposition and sufficient capital -data will tell anyone what is needed, what to do about it, and how to do it.Data speak for themselves, as long as the right individuals have the will to listen.

The Two Competing, Yet Complementary Logics of Data
The notion that data speak for themselves should not lead one to believe that the utterances of data are clear and unambiguous.In fact, we would argue that the ways in which data are being mobilized in/by Teach For All respond to two competing, yet complementary, logics.
The notion of data speak.The overarching logic could be described as a call for data per se, that is, as a continuous need to produce and process data about anything and everything in order to figure out problems without defining them in advance.We term this logic data speak, and in many ways, it gains credence in tandem with the pervasiveness of the use of data, a point we will return to later.We argue that Teach For All uses data speak to bind educational problems across network organizations.The programs have to buy into the common problem before they can buy into the solution.Data are used as a rhetorical tool -not only does it look and feel like a truer, superior logic; it also establishes a kind of camaraderie among those who speak it.Organizations share alarming figures of local educational disparities and retweet the difficulties facing other member programs.Global challenges beyond the reach of TFA's educational reform are also shared.For instance, Teach for Canada tweets "Nearly 43 per cent of the indigenous population in Manitoba has yet to finish high school" shortly before retweeeting, "The good news: the number of child labourers has fallen by a third since 2000.#notochildlabour" (2013, October 18).Across Teach For All's Twitter feeds, the camaraderie and the shared logic are what unify the common problems.The assumptions embedded in this logic have powerful effects in the day-to-day workings of different organizations and agents.Some of these underlying assumptions are that: Data represent objectivity and rigor of analysis and valuation.Implicit in the presentation of these numbers is the possibility of raw data, information that is neutral, objective, and devoid of politics or bias.The data replace the phenomena at stake with a higher degree of "reality", or at least transparency.In regards to data about student achievement, Landahl and Landahl (2013) argue that, The new criterion-referenced grading system and the international assessments have made failure visible.They have created data that speaks the language of crisis.Paradoxically, they have also fed dreams of high performance.[...] In this sense, the gap between visions and perceived reality has dramatically increased."(p.73) We would like to go one step further, by implying that data do not only make failure visible, they make a particular notion of failure possible.In tweets like Teach for India's "Why am I doing what I am doing when I know 800 million in this country can't afford an ice cream?" (2013, October 17) and Teach for Canada's "Why does 10 juice boxes cost $30 in my community?"(2013, October 5), educational problems -in particular those attached to the quality of the teacher -are expanded to now include things like access to a (Western notion of) childhood.
Data represent the possibility for present or future large-scale intervention of human interaction and stand as a proxy for development.Data, by themselves, appear to establish a distinction between those phenomena that can -either now or at a later time -be acted upon at a large scale and those that cannot.In other words, data allow for visions of scalability in a highly personalized and adaptive, hyper-connected global context.In this view, collection and processing of data are necessary, independent of problem identification, because it is the possibility for change that "matters."Data have been figured as a "gold mine" or "the new oil of the Internet and the new currency of the digital world" (Gitelman, 2013, p. 123).Without data, no reform is possible, even if the direction of said reform is not clear (yet).Hence, "data speculation," or amassing data so as to produce patterns -as opposed to having an idea for which one needs to collect supporting data -is paramount (Gitelman, 2013).According to Gitelman (2013), "data is the material for informational patterns still to come, its value unknown or uncertain until it is converted into the currency of information" (p.123).School systems -and one could extend this to other systems as well -that do not produce data about themselves are seen as lacking or deficient.In a continuum of development then, systems that produce a "surplus" of data, beyond if/how they are used, are conceived as more developed, because of the potential they present for reform.Model countries, regions, organizations, etc. are such because of the kind of data that they make available about themselves, and calls for action about systems in crisis are calls for the production of good data.This is what Evan Kornbluh (2013) from Teach For China refers to when, drawing from Friedman's speech at the Teach For All Global Conference in 2013, he mentions the economic impact of globalization and the information revolution, which "bear enormous relevance to the work of Teach For All organizations to prepare under-served students around the world to compete and thrive in this transforming economy."What counts here is not what the content of the data is interpreted as saying, but the existence of data by themselves as an indicator of progress, or at least as its condition of possibility.
It is no surprise then that an organization whose image relies on a certain kind of information would also prioritize that kind of information in the classroom, by emphasizing what Teach For All board member and OECD Head of Education Andreas Schleicher refers to as non-routine cognitive and analytic skills, skills he has also called "the currency of 21 st century economies" (Schleicher,n.d.,para. 2).Schleicher distinguishes these skills from traditional ways of learning that emphasize content knowledge, arguing that non-routine skills include the ability to process and analyze information, and generate new information (Schleicher, 2012).It is worth noting that in his Teach For All 2013 Global Conference presentation, Schleicher (2013) never names these nonroutine skills; instead, he tells the room of educators and partner CEOs "of course you can figure out what those skills are, but then you have to teach and learn them."Measuring these skills seems as important, if not more, than the skills themselves, as forty minutes of his forty-five minute speech on "What Makes a Great School System" were devoted to presenting Pisa results.It's fitting then that Schleicher (year?) closed his Teach For All lecture with his catchphrase, "Without data, you are just another person with an opinion." The quantity and quality of data available can be used as a proxy for the will to reform.What stands in the way of the production of data is not mainly technological advances or technical difficulties, but the will to change the status quo.The lack of available data can be attributed to the obstacles posed by those who do not want to or see the need to reform, thus it becomes a matter of convincing decision-makers to adopt a new way of being in the world by privileging data production and dissemination as a condition of possibility for common sense reform (Hess, 2006).
Shanghai, the site of 2013's Teach For All Global Conference, is here the model region.According to a New York Times article written by Thomas Friedman (2013): In 2003, Shanghai had a very 'average' school system, said Andreas Schleicher, who runs the PISA exams.'A decade later, it's leading the world and has dramatically decreased variability between schools.' He, too, attributes this to the fact that, while in America a majority of a teacher's time in school is spent teaching, in China's best schools, a big chunk is spent learning from peers and personal development.As a result, he said, in places like Shanghai, 'the system is good at attracting average people and getting enormous productivity out of them,' while also, 'getting the best teachers in front of the most difficult classrooms.'If only every other country showed the will to reform that Shanghai did…".(para.10) Returning to Teach For All, these three assumptions are present in the ways in which the organization mobilizes the call for more data production by the different constituencies it is trying to affect.One example of this is the wide circulation among the Teach For All network via Twitter of international comparative reports (produced or compiled by NGOs such as UNICEF) indicating the countries with the worst teacher absenteeism problems, lowest performance, etc., without there being any correlation between these countries and the locations of the Teach For All programs7 .While the pertinence of these reports is of little consequence, the tone of their data matters.Alarming figures of educational inequity -whether internationally between countries, or locally between boys and girls, Whites and racial minorities, suburban and urban students -are tweeted daily by Teach For All and participating programs, establishing a virtual landscape of educational crisis.For supporters to understand "the challenge" and imagine "the solution," the important thing is for there to be data, not necessarily to read what the data are saying.This is evident in this lengthy quotation from a Fortune Magazine article about Wendy Kopp: As passionate as Kopp is about Teach For All, she doesn't set targets for numbers of countries served; her mission simply is 'to see rising education levels and narrowing educational disparities around the world.' [She recalls an initial McKinsey analysis that suggested 60 countries in which Teach For All could be viable.]'A year in, half the countries we were in were not even on that list,' she says.The social entrepreneurs, more than country attributes, affect the programs' viability.She says, 'When that person is determined enough to make this work, it works.'Spoken from experience.(Sellers, 2013, par. 10) McKinsey -a founding partner of Teach For All, and former employer of Teach For All's cofounder and Teach First founder and CEO Brett Wigdortz -is mentioned to grant legitimacy to the data indicating the countries in which Teach For All would be viable, yet the fact that half the countries Teach For All is in right now are not on that list also indicates that what the data say is actually almost irrelevant.Will is necessary for data, but will also trumps data.
The use of data.Embedded within our notion of data speak is the use of data, or what one traditionally thinks of when reports and charts are circulated among members of an organization.The idea is that this specific data set serves as a diagnostic of a particular situation, and contains within it the seeds for the solution of a problem in need of addressing.Data are collected intentionally, following what is considered by the researchers as the most appropriate method for that situation, and is then read through the lens of its usefulness for solving specific issues.
One example of the use of data is the emphasis each Teach For All program puts in training its corps members to collect data through tests about the performance level of their students in the subjects being taught.Corps members learn how to translate tests into data, how to track data, how to create interventions to affect the data, and finally how to assess their own interventions in terms of changes (or lack thereof) in data.In its 2011 annual report, Teach For China outlined the importance of this work.The authors note that Teach for China was "working to develop a rigorous system for measuring Fellow and student progress" and that "these student success metrics and detailed data on student performance [...] will allow Fellows and their support staff to set ambitious but attainable goals and refine their work over time" (p.33).The primacy of rigorous, detailed data points to a particular logic in which abstracted representations of learning and development are made more utilizable for change than immediate, everyday interactions with students.
A perhaps more sophisticated "use of data" by Teach For All lies in its recruitment efforts.Teach For America and Teach For All use the company SalesForce to listen to what others are saying about the organization, to tailor their recruiting accordingly, and to generate/reinforce certain values and ideas amongst their corps members and recruits.SalesForce provides its clients with the tools for "turning customers into advocates", by "listen[ing] to your customers and engag[ing] in real time, earning their trust at every touchpoint" (SalesForce, 2014) 8 .Through advanced, multi-tiered data gathering and analysis tools such as the ones provided by SalesForce, Teach For All learns what kinds of figures sound good to recruits and corps members.In other words, otherwise seemingly useless facts and figures are found to be productive in garnering the support and camaraderie of Teach For All followers and in maintaining a strong company image.The influential website Information Management named Ellen Shephard, Teach For America's CIO and SVP, one of 2011's 25 top information managers.That year, Teach For America and Teach for All's data architecture was completely rebuilt to coordinate roles and data.
We hire very smart people but we were siloed and transacting all over the organization.Our program is executed in a very cohesive manner among many client teams who had no way to effectively share information, no way to effectively have 8 https://www.youtube.com/watch?v=vMpqQjvzzVk their business processes linked.So for starters I became Darth Vader of Excel and Access databases.People know we are building what has to be a very integrated experience that is effective, efficient and cost effective as we grow.(Information Management, 2011) The logic of the use of data, then, is the one that calls for the production/collection of specific data, their processing, and their employment towards decision-making to address problems that were determined before the data were collected.This logic could be simplified by representing it with the following graph.9The affective use of data.A common mistake made by people working within qualitative methodologies is to think that quantitative studies are necessarily impersonal, devoid of an affective dimension (colloquially referred to as "cold numbers").As mentioned earlier, Porter (1996) describes how quantification is seen as a "technology of distance," which has historically allowed social researchers and scientists to learn about specific populations without having to get to know them.While this may be historically accurate, contemporary technologies of quantification and data management have complicated the relationship between data and affect.Childress (2014), in discussing the increasing tendency at personalizing technology and advertising, quotes SalesForce's CMO Michael Lazelow as stating: "[Consumers] are searching for emotional connections to companies, not just transactional relationships… The value of targeted marketing isn't just from the marketing message itself, it's from the intelligence and optimization" -data -"with each interaction that builds a one-to-one relationship with each customer to increase brand loyalty and drives sales."(para. 27) According to Higgins, Hess, Weiner, and Robison (2011) -the first author being a former business professor turned education professor at Harvard -"stubborn achievement gaps, increased competition among school providers, and a heightened focus on performance have created an appetite for creative problem solving and scalable, transformational initiatives" (para.6) that appeal to the kinds of high-achieving, non-traditional teachers Teach For America recruits.To this end, the kinds of change Teach For All envisions matter less than the scope of these changes.Higgins et al. (2011) argue that while there is ongoing debate about the effectiveness and tenure of Teach For America members, it is more important that the organization recruits young leaders who have the superior skill set and the drive required of education entrepreneurs, a field Higgins et al. (2011) believe is of growing importance.Their rhetoric of "stubborn achievement gaps" (2011), which persist in the US between boys and girls, white students and black and Latino students, rich students and poor students, has travelled globally, shifting in accordance with the needs of the local programs.While we are not suggesting that education systems are equitable, Teach For All's ability to carefully cull data that frame educational crises in places like Argentina, Estonia, and Sweden not only drives teacher recruitment and program growth, it establishes the need for future, globally networked education entrepreneurs."Numbers, categories, data, statistics are central to this technology, are central to achieving cybernetic input-output equations demanded by the contemporary neoliberal state.They also sit against the structure of feeling where meta-narratives are at least dying" (Lingard, 2011, p. 365).In other words, the logic of the use of data is not to be posited in opposition to the affective dimension of policy intervention.On the contrary, they feed off of each other.

Data and the Production of a Particular Kind of Teacher
The logic of the use of data is central to the network's understanding of effective teaching.One example of this can be seen in the teacher assessment model shared by the programs within the network, termed Teaching as Leadership [TaL] (Teaching as Leadership, n.d -a): At Teach For America, the Teaching As Leadership rubric is an evolving centerpiece of our training and support systems.It is designed and used as a professional development tool, helping new teachers to identify their strengths and weaknesses and to accelerate their improvement for the sake of students' learning.(p. 1) This model is "outcome oriented," meaning that "a particular strategy is only as valuable as its influence on students' academic progress, and teachers should experience greater success with their students as they improve their proficiency" (Teaching as Leadership,p.1).The people assessing the corps members in action use a comprehensive 16-page rubric that not only categorizes the teacher into pre-novice, novice, beginning proficiency, advanced proficiency and exemplary, but also tracks the teacher's development by translating observations into data that can be communicated, shared, compared, and extrapolated across contexts.Teachers are both data sets and data producers, as one of the aspects that the TaL rubric evaluates is the ways in which teachers collect, track and use data about their students' performance.In other words, the language of data use is what makes TaL, TaL.Data about corps members are used to terminate contracts, push for stronger performance, or reward excellence.Data are also used to revise the rubric itself and collect more data.The profiles highlighted under case studies in TaL's home page merge the personal (energetic teachers interacting with individual children, usually from disadvantaged backgrounds) with concrete data indicating, without doubt, the teacher's success in increasing the students' academic achievements (Teaching as Leadership, n.d.-b).As mentioned above, the affective dimension plays a central role in making these numbers not only intelligible, but also embraceable by any reader (For more on the narrative powering Teach For All, see Ahmann, this volume).
This use of data with the purpose of developing effective teachers encounters the logic of the data speak, by contrasting the ways in which Teach For All's teachers teach, assess and react to the ways in which other teachers work, data usage being the distinguishing factor.For instance, TaL cautions against "lack of good data (aka having to rely on guesses and hunches)", what it refers to as a "pitfall" of teaching, and advises corps members to ask themselves: "Am I efficient when I gather and enter data?If not, what are the areas of inefficiency?What would I need to change to accomplish more in less time?" (CIE: Continuously Increase Your Effectiveness, n.d.).Under this logic, data -and their usage -are good per se, beyond specific results.The very fact that Teach For All's teachers focus on producing and using data makes them already better teachers, before they are even assessed on whether their data usage is increasing student achievement (even if the particular notion of student achievement mobilized by the organization is inseparable from specific data).What matters most is that they are not left behind in the wake of our increasingly networked, datadriven world.Teach For All teachers are drivers of change, revolutionaries riding on the promise of data's authority.

Conclusion: What Are We, Without Data?
Andreas Schleicher seems to be fond of the popular refrain: "Without data, you are just another person with an opinion" 10 (see for example his presentation of the 2009 PISA results: Schleicher, 2010).We hesitate to embrace this premise on two accounts.First, as we have argued in this article, data in themselves do not possess a unique truth value that is separate from the ways in which it is deployed as part of a specific, historical regime of truth; that is, as part of "the epistemological field that allows for the production of what counts for knowledge at any given moment, and which accords salience to particular categories, divisions, classifications, relations and identities" (Poovey, cited in Rose, 1999, p. 29).In other words, a datum itself has no epistemological or ontological grounding (Gitelman, 2013).Data are a historically produced set of objects that act and are acted upon as if in themselves, independently from the subjects and processes, they said something about the world.We claim here that the appearance of data as a salient phenomenon in education reform most definitely says something, but that something is not inherent to the "content" of the data.
Second, and related to the previous point, the phrase implies that the two opposing ways of providing grounding to an argument -with data or with opinion -will produce, on the one hand, a unifying, unambiguous, evidence-based solution, and on the other, just another unfounded, multiple perspective.Since truth is in data, possession of data leads to enlightenment, as solutions are already embedded in the data, if they are not the data themselves.Opinions, on the other hand, are merely subjective, worthless for thinking about addressing the imminent crises.In this article, by contesting the notion of an unambiguous relation between data and truth, we therefore challenge the idea of solutions being already there in data, in any form of relation, as well as the understanding of (objective) data as a separate category from (subjective) opinion.Data and opinion do differ in the ways in which they are mobilized to tell truth, which does not imply different proximities to an external truth.
Throughout the paper we have positioned Teach For All as an exemplar organization in terms of its utilization not only of data, but of data speak.We have argued that, while complementary, there is a tension between the organization's efforts to use data to drive decision-making, and the ways in which they call for data themselves as a solution to the problems afflicting the field of education.Data stand in for rigor, objectivity, scalability, and, perhaps most importantly, the will to reform.The tension emerges from a difference between treating data as a means and as an end, and the kinds of subjectivities and discursive practices that are produced by these two understandings.For instance, the first logic attracts (and produces) the teacher who follows a particular rationality that treats data as undisputable truth, while the second one requires the teacher to become a scanner, and teaching to become the act of collecting data.The only decision here is the decision to gather information, to filter what counts as data (i.e. as learning), as the data themselves are the solution to the problem (i.e., the lack of data/learning).
Data speak also exerts an affective pull that engages an audience in thinking about education in particular ways, and in framing the issues about teaching, learning, achievement and reform as a matter of common sense with which any rational individual would connect.Data are not only a way of looking at information, but they serve as a universalizing language that anyone anywhere can and should care about.Across the Teach For All network, data are what gives credence to a common battle against educational injustice.This way of presenting data obscures the very notion that, following the title of Gitelman's (2013) book, raw data is an oxymoron.
It is important to note here that we are not arguing against the use of data.That would be foolish for many reasons, as data can also be understood as the ways in which we all process and attempt to make sense of information.What we are arguing for is understanding the foundations of data to be contingent, to be part of a historically produced and situated system of thought (Foucault, 1977;Poovey, 1998) in which notions such as objectivity, rigor, and scalability form a grid that allow for data, as objects, to exist.Data cannot be understood outside this grid.And the ways in which reform efforts such as Teach For All frame the issues of contemporary education as universal, and the solutions as knowable, have embedded in them the dangers of eliminating or at least marginalizing other ways of asking questions and attempting to address them, ways that fall outside the new common sense.

Figure 1 .
Figure 1.The logic of the use of data.