Teaching Practices for Scientific Argumentation in K–12 Classrooms

Abstract

Frameworks of core practice in science teaching name practices either generally enough to span the disciplinary practices of the field or specifically toward explanation and modeling, leaving open what teaching requires when the practice at hand is argument. I elaborate those frameworks by specifying seven foundational teaching practices for argumentation that are scientific in nature. The practices are creating and leveraging instances of uncertainty, framing joint enterprise, making argumentation public, introducing and mediating argumentation tools, establishing and refining shared criteria for what counts as an acceptable argument, developing and maintaining norms for critique and disagreement, and noticing and responding to student reasoning and needs. What distinguishes them is the shift from building an account of a phenomenon to adjudicating among competing accounts, which asks a class to work out what counts as adequate reasoning rather than only to apply standards it has been given. Because the sciences reason in more than one style, and each carries its own epistemic criteria, the criteria a class develops are style-dependent rather than general. Equity is constitutive of the practices rather than a parallel commitment. I close by considering enactment under institutional constraint and implications for teacher educators, researchers, and curriculum designers.

1. Introduction

Argumentation, which is often defined as the process of proposing, supporting, critiquing, and challenging claims about how or why the world works (Erduran & Jiménez-Aleixandre, 2008; Henderson et al., 2018), is a fundamental knowledge-building practice of science (Duschl & Osborne, 2002; Ford, 2012; Osborne, 2010) and a central focus of many contemporary frameworks for K–12 science education (Australian Government Department of Education, n.d.; Cetin et al., 2016; National Research Council, 2012; White et al., 2023 for examples). It is important to note at the outset, however, that argumentation is not unique to science. People argue in courts, in legislatures, in families, and in the ordinary business of settling what to believe, and students arrive in science classrooms already practiced in several of these forms. What distinguishes argumentation that is scientific in nature is not the moves involved, since proposing, supporting, critiquing, and challenging are common to all of them, but what a community treats as sufficient to settle the matter. In scientific argumentation the grounds that make evidence acceptable are settled within a style of reasoning, and disagreements are in principle resolvable by appeal to those grounds even when they remain unresolved in fact. This is what makes the criteria a class develops both learnable and revisable, and it is what the practices I propose are specified toward. I use the shorthand scientific argumentation to describe this practice.

It is important, at this point, to distinguish between scientific and socioscientific forms of argumentation, because this is where the difference is most consequential and most often missed. Arguments about whether to permit a land use, mandate a vaccine, or site a facility draw on scientific evidence and on styles of reasoning that the sciences have developed, and the empirical work such arguments require is continuous with the work described here. What is not continuous is the adjudication. The question at issue in a socioscientific argument includes what ought to be done and at whose cost, and no style of scientific reasoning settles that, so two people applying the same evidence, but different criteria, can reach valid but opposed conclusions (Nielsen, 2013; Sadler & Zeidler, 2005). A class arguing about the mechanism of a phenomenon can agree to use criteria that eventually settle the matter, and a class arguing about what to do cannot, because the criteria that would settle it are not disciplinary criteria at all. I take the seven practices to be relevant to socioscientific argumentation and not sufficient for it, and I confine my claims here to argumentation whose object is an account of the material world.

Researchers have, over the past two decades, produced a substantial body of work on how students participate in scientific argumentation (Berland, 2011; H.-T. Chen et al., 2016; Jiménez-Aleixandre et al., 2000; Kulatunga et al., 2013; Sampson & Clark, 2011) and what students learn from and about it when they have repeated opportunities to engage in scientific argumentation (Asterhan & Schwarz, 2007; Boran & Bag, 2016; Y.-C. Chen et al., 2016a; Demircioglu & Ucar, 2015; B. Hand et al., 2021; Sampson et al., 2013; Walker & Sampson, 2013). Researchers have also designed, tested and refined new curricula, instructional models, and technology-enhanced environments that are designed to make argumentation a fundamental knowledge-building practice in science classrooms (Belland et al., 2008; Berland & McNeill, 2010; Cavagnetto, 2010; Clark & Sampson, 2006; B. Hand, 2007; McNeill, 2009; Sampson et al., 2011). The field, as a result, knows a great deal about how to design materials and tools that promote and support student engagement in scientific argumentation.

The development of these materials and tools is, however, a necessary but not sufficient condition for creating learning experiences that are conducive to learning from and about scientific argumentation in ways that are equitable for all students. Teachers are not interchangeable conduits for these curricula, instructional models, and technology-enhanced tools; they shape what scientific argumentation looks like inside their classrooms through the choices they make about which materials to use and when (McNeill, 2009; McNeill et al., 2017, 2018, 2026), how to respond to what students say and do (Berland et al., 2020; Y. Chen et al., 2019; Y.-C. Chen, 2025; Hammer et al., 2012; McNeill & Pimentel, 2010; Robertson et al., 2016), and which voices and ideas to take up as resources for the work (Calabrese Barton & Tan, 2020; Rosebery et al., 2010; Warren et al., 2001). If all students are to learn how to participate in argumentation in ways that are consistent with the norms and epistemological commitments of the sciences, while drawing on rather than dismissing the ideas and ways of knowing that students bring with them, then teachers will need to learn a set of foundational teaching practices that they can use across classrooms, curricula, and content areas.

The science education community has begun to take up this challenge as part of the broader core teaching practices movement (Grossman et al., 2009; McDonald et al., 2013). Researchers have not only proposed sets of core or high-leverage teaching practices for science (Kloser, 2014; Luehmann et al., 2024; Windschitl et al., 2018; Windschitl & Calabrese Barton, 2016) but also have begun to specify how these practices can be organized into coherent programs of teacher preparation (Gray, 2026a). These frameworks have advanced the field considerably by establishing a shared language of practice and by demonstrating that ambitious and equitable instruction is learnable by novices (Kang & Windschitl, 2018; Stroupe et al., 2020; Thompson et al., 2019). These frameworks, however, describe the work of science teaching as a whole, and the practices they name stand in different relations to any one disciplinary practice. Some are pitched at a level of generality that holds across the disciplinary practices of the field (see Kloser, 2014). Others are oriented toward particular kinds of disciplinary work, most often the construction of explanations and the development and use of models (see Windschitl et al., 2012). What such frameworks leave open is how these forms of work need to be enacted when students are engaged in a specific practice, such as scientific argumentation.

That question is worth answering for three reasons. The first is disciplinary. Scientific argumentation is the means by which the products of the other practices of science are held accountable, since a model, an explanation, or an investigative method becomes knowledge only once a community has examined the case made for it (Ford, 2008a; Kuhn, 1993; Osborne, 2014a, 2014b). The second is empirical. The variation described above locates the problem squarely in the work of teaching rather than in the design of materials. The third is practical. When the teaching practices that scientific argumentation requires are left unspecified, argumentation is liable to be taken up as one more piece of content to be covered, with creating products, such as claim–evidence–reasoning worksheets, the focus of classroom activity rather than as a way of participating in the discipline (Allen & Rogers, 2015; McNeill et al., 2017; Nussbaum et al., 2024). Teacher educators, meanwhile, are left without a shared focus for designing learning experiences that prepare future teachers to support argumentation in equitable and disciplinary-authentic ways (Altun & Ozsevgec, 2025; Aydeniz & Ozdilek, 2016; Boyer, 2016; Erduran et al., 2006; Mesci et al., 2025), and those who lead professional learning for practicing teachers are left without a productive object for sustained collaborative work (Garet et al., 2001; Osborne et al., 2013; Penuel et al., 2007; Thompson et al., 2019).

In this article, I propose a set of seven foundational teaching practices associated with promoting and supporting scientific argumentation in K–12 classrooms. The practices are creating and leveraging instances of uncertainty, framing joint enterprise, making argumentation public, introducing and mediating argumentation tools, establishing and refining shared criteria for what counts as an acceptable argument, developing and maintaining norms for critique and disagreement, and noticing and responding to student reasoning and needs. My aim is not to displace the existing core practice frameworks or to propose a competing set. It is, rather, to add a specification layer to them, elaborating what the practices they name require when the disciplinary work in play is argumentation. I argue that these seven practices, taken together, constitute a coherent and learnable set of forms of work that pre-service and in-service teachers can use across different curricula and instructional approaches, and that they offer a productive focus for both initial teacher preparation and long-term professional learning.

I proceed as follows. I first outline the theoretical framework that shapes how I understand teaching practices and the process of learning to support argumentation. I then review three strands of empirical work on argumentation in science education, with particular attention to what is known about teacher moves and to what existing frameworks of core practice specify about argumentation and what they leave to be elaborated. I next describe how I developed the framework, including how I assembled the literature, how I generated and tested candidate practices, and the criteria I used to evaluate them. I then turn to the practices themselves, defining each, explaining why it matters for student learning, and grounding it in the relevant empirical literature. I then consider how the practices relate to one another, how each can be distinguished from enactments that resemble it, how the set stands in relation to existing frameworks of core practice, and what is at stake for equity in each. Because the practices are enacted by teachers who do not set the terms of their own work, I then consider what their enactment looks like under institutional constraint. I close by discussing what the seven practices contribute and what they leave unresolved, acknowledging the limitations of my proposal, and drawing out implications for teacher education, for ongoing professional learning, and for future research.

2. Theoretical Framework

I approach the question of how to create space for scientific argumentation in classrooms through two interrelated theoretical lenses, which together shape what I mean by teaching practices and how I think teachers and students learn through and from them. The first lens is a view of teaching as situated and relational work. From this perspective, teaching is not the delivery of content nor the application of generic pedagogical techniques but rather a coordinated activity in which teachers, students, materials, and ideas are brought into relation with one another in ways that shape what counts as legitimate participation, what is recognized as knowing, and whose contributions are taken up as resources for collective sensemaking (Engle, 2012; Engle et al., 2014; Greeno, 1998). Teaching practices, from this view, are best understood as recurring patterns of interaction that organize this coordinated activity, that are anchored in disciplinary commitments and learning goals, and that are enacted in the moment in response to students and to the unfolding situation (Grossman et al., 2009; Lampert, 2010; Loewenberg Ball & Forzani, 2009; McDonald et al., 2013). This framing is consistent with how scholars in the core practices tradition describe teaching as a practice that is at once principled and adaptive (McDonald et al., 2013) and as a form of work that develops over time through deliberate engagement in the activity itself (E. A. Davis et al., 2017; Lampert et al., 2009; Stroupe, 2014; Stroupe et al., 2022).

The term practice, however, carries more than one meaning in this literature and the two meanings do different work. In the situated tradition associated with Lave and Wenger (1991) and Wenger (1998) practices are the shared activities that constitute a community, held together by the community’s negotiated sense of what it is doing and what its members owe one another. Practices in this sense belong to a community and are not specified from outside it. In the core teaching practices tradition, by contrast, a practice is a nameable, recurring, and learnable unit of professional work that can be represented, decomposed, and rehearsed with novices. I use the term in the second sense when I write about teaching and in the first sense when I write about students learning to argue. I take from the situated tradition its account of how a community’s activity is held together, and I draw on Wenger’s construct of joint enterprise when I describe the first of the seven practices. I do not take its account of learning as movement from peripheral toward fuller membership, which would cast school science as a destination that students travel toward rather than one way of knowing among those they can come to command. The seven practices I propose are therefore not an attempt to prescribe the culture of a professional community. They are an attempt to name forms of work that are already visible across the empirical literature on scientific argumentation, at a grain size that makes them available as objects of both pre-service and in-service teacher learning.

The second lens is a sociocultural view of learning scientific argumentation as participation in a form of disciplinary discourse (Brown et al., 1989; P. Kelly, 2006; Wertsch, 1991). From this perspective, students learn scientific argumentation not by being told what an argument is and then being asked to produce one, but by participating in extended episodes of proposing, supporting, critiquing, and challenging claims in the company of others who share a common goal (Berland et al., 2020; Ford, 2008b, 2012). Such episodes depend on the presence of genuine uncertainty, because a claim that all participants already accept gives students nothing to argue about and reduces argumentation to the retrieval of an expected answer (Y.-C. Chen, 2020; Y.-C. Chen & Qiao, 2020; Manz, 2015; Manz & Suárez, 2018). This view situates argumentation within a broader tradition of dialogic pedagogy in which knowledge is treated as something built and tested in talk rather than transmitted through it (Y. Chen & Techawitthayachinda, 2021; Y.-C. Chen, 2020, 2025; Watkins & Manz, 2022). Learning, in turn, involves shifts in how students take up the discourse, tools, and norms of the discipline (Cavagnetto et al., 2010; Y.-C. Chen et al., 2016a; G. Kelly & Chen, 1999; Windschitl, 2019) as well as shifts in how students see themselves as people who can contribute to such activity (Engle & Conant, 2002; V. Hand & Gresalfi, 2015; Tan & Calabrese-Barton, 2008). It is important to note that participation in scientific argumentation is not separable from questions of equity. Whose claims are heard, whose evidence is treated as relevant, and whose ways of thinking are recognized as valuable are all shaped by the histories that students and teachers bring with them into the classroom (Bang & Medin, 2010; Rosebery et al., 2010; Warren et al., 2001). A theoretical framework that takes participation seriously must therefore also attend to the ways that classrooms can either reproduce or interrupt patterns of marginalization (Calabrese Barton & Tan, 2020; Ha & Kim, 2022; Kang & Zinger, 2019).

What students are learning to participate in, however, is not a single generic mode of scientific argumentation. Kind and Osborne (2017), drawing on Crombie’s (1994) cognitive history of science, argue that scientific reasoning is better understood through six distinct styles, each of which brings into being its own ontic, procedural, and epistemic entities. Students may be asked to engage in argument from evidence, but the forms that argument takes and the criteria used to justify its conclusions depend on the domain in question and on the style of reasoning in play. These styles emerged as contingent products of historical circumstances rather than as expressions of universal rules of reasoning, and following Hacking (2012), Kind and Osborne argue that these styles answer to no standard of truth and reason higher than their own. They have become authoritative because they have been successful, which is not the same as claiming they are the only way of reasoning about the material world. From this perspective, I do not think an argument should be viewed as a structure with reasoning as one of its parts. An argument is a way of reasoning about something. Reasoning is what the whole argument does, and what an argument contains is a claim, evidence, and a justification, by which I mean the disciplinary grounds that make evidence of this kind acceptable as support for a claim of this kind. All three take their shape from the style of reasoning in play, because what makes an analysis of data adequate and what makes disciplinary grounds sufficient are settled within a style rather than across the sciences as a whole. It is important to note, however, that Kind and Osborne are themselves skeptical of the turn toward practices in science education, on the grounds that practices name the actions scientists take in the moment but do not, in themselves, specify what those actions are trying to achieve. This point is well taken. A focus on scientific practices requires more than a description of what scientists do as they generate new knowledge. It also requires the disciplinary grounding that styles of reasoning supply, because without that grounding a teacher has no principled answer to the question of what a class’s criteria are criteria for, or why they came to be valued in a particular field.

Two things follow from a focus on styles of reasoning in science, and they pull in the same direction. The first is disciplinary. Of the six styles, what schools commonly present as “the scientific method” draws on only two, and students therefore receive a correspondingly narrow account of the nature of scientific reasoning (Kind & Osborne, 2017). The second is a matter of justice. Gray (2026b) argues that this narrowing produces two hierarchies sustained by a single logic, one positioning experimental sciences above historical sciences and one positioning Western science above Indigenous knowledge traditions, since retrodiction, abduction, consilience, and narrative reasoning are treated as speculative wherever they appear. Curriculum standards and standardized assessments have reinforced the narrowing, because one method is easier to specify and to test than a plurality of them (Gray, 2026b; Rudolph, 2005, 2023).

I therefore treat the styles of reasoning that school science uses as one community’s way of knowing and doing rather than as the standard against which others are measured. Kind and Osborne (2017) themselves acknowledge, in a footnote to their account of the six styles of reasoning, that what they present is an argument for the nature of science in the Western tradition that forms the basis of school curricula across most of the world, and that there are good arguments that it fails to represent how other cultures reason about the material world (p. 12). Those ways of knowing are worth learning and learning them well means coming to understand the rules of the game as that community plays it. Other communities, reasoning under different historical conditions, have arrived elsewhere with rigor of their own (Bang & Medin, 2010; Gray, 2026b; Medin & Bang, 2014; Warren et al., 2020). Students are themselves members of several communities at once, and what instruction in scientific argumentation can offer them is not conversion from one reasoning style or way of thinking to another but the capacity to recognize which rules are in play, to reason well within them, and to move among them as needed. Making that choice visible and acceptable, rather than presenting the criteria of school science as simply how thinking is done, is what distinguishes an equity focus within these practices from an equity commitment running alongside them.

This view also shapes how I understand what students bring. I use the term argumentation repertoire, adapting Gutiérrez and Rogoff’s (2003) account of repertoires of practice, to describe the range of ways of proposing, supporting, critiquing, and challenging that a person has available from participation in the various communities in which they take part at a given time and in a given context (Rosebery et al., 2010; Warren et al., 2020). An argumentation repertoire is not a partial version of a scientific one, and coming to argue in ways that are scientific is not the correction or completion of what a student already does (Bang & Medin, 2010). Styles of scientific reasoning are cultural tools, and coming to know them means having more ways of making sense available and coming to judge which are apt for a given community, context, and question. A student who reasons probabilistically about risk in one setting while coming to reason experimentally in another has gained something to move among rather than had something corrected. On this view, equitable instruction in scientific argumentation widens the range of what a student can draw on and sharpens the judgment about when to draw on each and why, rather than substituting one repertoire for another or privileging one while marginalizing others.

These two lenses, taken together, lead me to think about teaching practices as doing two kinds of work at once. The first is the work of setting up opportunities to learn. Teachers shape the conditions under which argumentation can take place by selecting tasks and phenomena that contain genuine intellectual openings, by introducing tools and representations that make the moves of the practice more visible to students, and by establishing norms for how students talk, listen, and contribute to the collective work (Berland et al., 2016; Engle & Conant, 2002; Manz, 2015; W. Sandoval & Reiser, 2004; Windschitl et al., 2008). The second is the work of being responsive in the moment to what students are doing and saying as they take up those opportunities (Berland et al., 2020; Hammer & Elby, 2003; W. Sandoval, 2005). Scientific argumentation is, by its nature, an activity whose course is not fully predictable; students propose unanticipated ideas, raise disagreements that point toward productive but unplanned avenues of inquiry, and bring experiences and ways of thinking that the planned activity did not foresee (Y. Chen et al., 2019; Y.-C. Chen, 2020, 2025; Manz & Suárez, 2018). Responsiveness of this kind begins with noticing (Krist et al., 2023; Mercier & Hinman, 2025), since a teacher cannot take up an opening she has not registered as an opening, and I therefore treat noticing as a constitutive part of the seventh practice rather than as a general disposition that operates outside the practices.

This dual structure gives me an analytic criterion for what makes a teaching practice foundational, which is that the practice enables a teacher to create opportunities to learn and to be responsive to what students do and say as they participate in that opportunity. A practice that only sets up opportunities will support argumentation only when student participation aligns with what the teacher expected, and a practice that only enables in-the-moment responsiveness will produce isolated episodes rather than sustained disciplinary engagement. The seven practices outlined in this article were therefore selected, in part, because they do both kinds of work.

3. Literature Review

The literature on scientific argumentation has grown substantially over the past two decades, and I focus here on three strands of work that bear most directly on my argument. The first strand examines what students learn from and about scientific argumentation when they have repeated opportunities to engage in the practice. The second strand examines the tools and instructional materials that have been developed to support it. The third strand examines what teachers do to promote and support argument when using those tools and materials. I then turn to existing frameworks of core practice in science teaching, which take up the work of teaching at a grain size that the literature on argumentation has left comparatively unspecified, in order to establish what they do and do not settle about the teaching of scientific argumentation.

Studies in the first strand have shown that students who participate in extended argumentation experiences tend to develop stronger conceptual understanding of the content under investigation (Asterhan & Schwarz, 2007; Y.-C. Chen et al., 2016a; Zohar & Nemet, 2002), more sophisticated views of the nature of scientific knowledge (Erenler et al., 2025; Khishfe, 2014; Mesci et al., 2025), and improved abilities to construct and critique arguments using evidence (Çetin & Eymur, 2017; Y.-C. Chen et al., 2016a; McNeill, 2009). Taken together, these studies establish the value of scientific argumentation as a goal of science instruction. What they leave open is how classrooms come to support it.

Studies in the second strand have developed and examined a wide range of resources that teachers can draw on to support scientific argumentation in K–12 classrooms. These resources include full curricula such as IQWST (Shwartz et al., 2008) and OpenSciEd (Edelson et al., 2021), instructional models such as Argument-Driven Inquiry (Sampson et al., 2011) or the Science Writing Heuristic (B. Hand, 2007), and various technology-enhanced learning environments (Clark et al., 2007). Researchers have also created and tested more targeted resources that teachers can incorporate into an instructional sequence such as writing prompts (Belland et al., 2008; McNeill & Krajcik, 2009), talk supports (Chin & Osborne, 2010; Christodoulou & Osborne, 2014) and tools that foster computer-mediated interactions that support specific types of reasoning or discourse (Bell, 2013; Bell & Linn, 2000; Clark & Sampson, 2006; Fan et al., 2020). Reviews of this body of work suggest that well-designed resources can create a context that can foster participation in scientific argumentation (Cavagnetto, 2010; Fakhriyah et al., 2021; Weiss et al., 2022; Zhou, 2024).

A separate set of studies in this strand, however, also documents what happens when the same materials or tools are enacted differently. Teachers can use the same instructional materials or a specific tool in ways that either open up or shut down student thinking (Berland, 2011) and students can complete tool-mediated tasks without engaging in the disciplinary work the tool was designed to support (Berland & Reiser, 2011; McNeill et al., 2006). What this work establishes is that instructional materials and tools are necessary but not sufficient. What a resource affords depends on how it is mediated, which puts the teacher rather than the resource at the center of the question.

Studies in the third strand have focused directly on teachers and on the nature of the moves they make. McNeill and colleagues have documented the discursive moves teachers use to elicit, press on, and connect student claims (McNeill et al., 2018; McNeill & Knight, 2013), the patterns of talk that distinguish argument-rich from argument-poor classrooms (McNeill & Pimentel, 2010; Pimentel & McNeill, 2013), and the beliefs and orientations that shape what teachers attend to and how they respond (González-Howard & McNeill, 2019). Berland and colleagues have examined how teachers structure participation so that students come to treat scientific argumentation as a meaningful collective activity rather than a procedural task, with particular attention to the conditions under which disagreement is taken up as an intellectual opportunity (Berland et al., 2016; Berland & Hammer, 2012). Osborne and colleagues have catalogued moves that appear to support quality argumentation and have proposed instructional sequences that scaffold teachers and students into the practice of arguing from evidence (Osborne et al., 2004; Osborne et al., 2016). Chen and colleagues have specified what teachers do to help students manage epistemic uncertainty during episodes of scientific argumentation, particularly through their use of questions and their appeals to authority and accountability (Y. Chen & Techawitthayachinda, 2021; Y.-C. Chen, 2020, 2025; Y.-C. Chen & Qiao, 2020). What stands out across this work is that the significance of a given move depends on the conditions under which it is made, since the same press for reasoning can open inquiry in one classroom and function as a demand for the expected answer in another.

This contingency bears on what it is reasonable to ask of this literature. Because the work a move does depends on what preceded it and on whose idea is being taken up, an account that maps particular moves onto particular student outcomes is the wrong thing to seek. What this strand leaves open is instead a matter of grain size. The work has proceeded largely at two levels, that of the individual move, such as pressing students to say why they think what they think or revoicing (McNeill & Pimentel, 2010; Pimentel & McNeill, 2013), and that of the orientations and beliefs that shape a teacher’s practice as a whole (McNeill et al., 2017; Simon et al., 2006). The intermediate level, at which a move is understood as one way of accomplishing a recurring form of work that a teacher can name, learn, and refine, remains comparatively unspecified within the literature on argumentation. Instructional models such as Argument-Driven Inquiry (Sampson et al., 2011), Model-Based Inquiry (Windschitl et al., 2008), and the Science Writing Heuristic (B. Hand, 2007) do specify teacher actions at something close to this level, but they do so within the architecture of a particular model rather than as practices that travel across contexts.

That work of specification has been undertaken within the literature on core practices in science teaching. Windschitl et al. (2012) proposed four core practices anchored in a model-based inquiry framework, among them pressing students for evidence-based explanations, and their accompanying analysis of how priming tools and face-to-face tools mediate enactment remains influential (see also Windschitl et al., 2018). Kloser (2014), using a Delphi expert panel methodology, identified a set of nine core science teaching practices, including facilitating classroom discourse, eliciting and using student thinking, and providing feedback, and his analysis usefully surfaced the persistent variability in the field’s language for practice. Gray (2026a) extended this work to the program level, organizing novice learning across grain sizes that run from moment-to-moment moves through unit-level design and embedding equity commitments across all of them. These frameworks are not all pitched at the same grain size. Windschitl et al.’s practices are the largest, spanning coordinated sets of interactions across a unit, and Kloser’s are smaller. Gray’s framework does not sit at a single point, since it names levels running from unit-level instructional sequences down to the high-leverage practices and finer strategies through which that work gets enacted.

The three frameworks, however, offer different kinds of guidance to a teacher whose students are engaged in the science and engineering practice of argumentation, and the difference matters for what follows. Kloser’s (2014) teaching practices are named at a level of generality that holds across the disciplinary practices of the field, which is what gives the set its value as a shared language for teacher education and also what leaves unspecified what changes when students are asked to critique one another’s ideas rather than to contribute ideas to a common pool. Windschitl et al. (2012), by contrast, do specify, but they specify toward explanation and modeling, so that argument appears as a means through which an explanation is strengthened rather than as disciplinary work supported in its own right. Gray’s (2026a) framework exhibits both relations at once, since its high-leverage practices are named as generally as Kloser’s while its instructional sequences build directly on Windschitl et al.’s. These absences are not oversights. All three frameworks were designed to describe science teaching as a whole, and arguing from evidence is one science and engineering practice among those the field names (see National Research Council, 2012). The elaboration these existing frameworks invite, and that I attempt in what follows, is a specification of what their practices require when the disciplinary work in play is argumentation. That specification is the contribution of the seven practices I propose. Before describing them, I make explicit how I arrived at them and the criteria I used to evaluate them.

4. Development of the Framework

4.1. Synthesis of the Literature

The seven practices are the product of a narrative synthesis of the empirical and conceptual literature on scientific argumentation in K–12 classrooms, informed by my own program of research on the design and study of argumentation-focused instruction. I did not conduct a systematic review, and the reason why is worth stating. A systematic review is well suited to questions about the effect of an intervention on an outcome, where the constructs of interest are named consistently enough across studies for a search protocol to find them. The object I was after is different. A recurring form of teaching work is not something most authors in this literature set out to name, and the field has no shared vocabulary for teacher action, so a protocol built on search terms would return studies that label similar moves in incompatible ways while missing studies in which the relevant work is visible in the analysis but nowhere in the title, abstract, or keywords. Identifying forms of work therefore requires reading across studies whose authors were not attempting to specify practices at all, which is what a narrative synthesis permits, and a systematic protocol does not.

The second source is my own work. I have spent two decades designing, enacting, studying, and revising argumentation-focused instruction with teachers, and that experience shaped what I recognized in the literature and what I judged to be difficult, learnable, or beside the point. It is important to note that this is a source of insight and a source of bias at once. It gave me an account of where teachers struggle that no reading of published studies would supply, and it also gave me commitments that could tilt the set toward the work that one instructional model makes salient. I have tried to constrain that tilt in two ways. I required, as a baseline condition, that each practice be grounded in an empirical finding about scientific argumentation in K–12 classrooms rather than in conjecture or in my own experience. I then described each practice as a form of work rather than as a feature of any curriculum, instructional model or technology-enhanced learning environment, testing each against the question of whether a teacher working from unrelated materials could recognize and enact it.

4.2. Assembling the Literature

I began by collecting the empirical and conceptual scholarship that takes the teaching of argumentation as its focus, meaning work in which what teachers do to promote and support argumentation is the object of study rather than the background condition for studying something else. This body of work is small enough to be read closely and large enough to support the kind of synthesis I attempted. I then extended the corpus through backward and forward citation chaining from those studies, which brought in work on student participation, on the design of tools and materials, and on classroom discourse more broadly, where teacher action is often visible in the data or the analysis even when it is not the stated focus. I then searched deliberately for scholarship on topics that the resulting corpus underrepresented, including argumentation in multilingual classrooms, epistemic injustice, styles of scientific reasoning, and the professional learning of practicing teachers.

I included empirical studies of argumentation in K–12 science settings in which teacher action was visible in the data or the analysis, along with conceptual scholarship on argumentation, core practice, and epistemic justice that bears on how such action should be understood. I imposed no date restriction, though the bulk of the corpus was published after 2000. I excluded studies of argumentation outside science except where they informed the construct distinctions I draw above, and I excluded studies of undergraduate instruction except where they bear directly on the preparation of teachers. The corpus is predominantly Anglophone and North American or European, a limitation I return to at the close of the article.

4.3. Generating and Testing Candidate Practices

Generating candidates proceeded in four passes. I first read across the corpus for descriptions of what teachers did, recording each as a short phrase and setting aside how the authors themselves labeled it. I then grouped phrases that appeared to accomplish the same work even when the vocabulary differed, so that pressing a student to say more about an anomalous result and holding a competing account open long enough for a class to compare it with another were treated as instances of the same underlying work. I next named each grouping at the intermediate grain size described below, which required deciding in each case whether a grouping was a single form of work or several. I finally tested each candidate against the analytic criterion set out in the theoretical framework, which is that a foundational practice must both create opportunities for argumentation and give the teacher something to draw on when students take those opportunities up in unanticipated ways, and then against the three criteria that follow.

Several decisions at the third pass account for the shape of the set. I separated the work of establishing criteria for an acceptable argument from the work of establishing norms for critique, because one governs what counts as adequate reasoning and the other governs how disagreement is conducted, and classrooms routinely have one without the other. I kept noticing and responding as a practice in its own right rather than distributing it across the other six, because a teacher can enact the remaining practices as a sequence of planned conditions without ever registering what students do with them, and that failure is common enough in the literature to warrant naming. I combined the design of tasks containing genuine intellectual openings with the in-the-moment work of sustaining those openings, because the literature indicates that either without the other reliably collapses into the retrieval of an expected answer.

4.4. Criteria Used to Evaluate Candidate Practices

Three criteria were applied to each candidate that survived the passes described above. As a baseline condition, I required that each practice be grounded in empirical findings about scientific argumentation in K–12 classrooms. Beyond that baseline, I asked of each candidate whether it could support a focus on equity, whether it was doable for teachers at a productive grain size, and whether it was usable across instructional materials and curricular contexts. I make these criteria explicit next so that readers can evaluate the practices against the standards I used to arrive at them.

4.4.1. The Practice Has the Potential to Support a Focus on Equity

My first criterion is that a practice should have the potential to promote and support a focus on equity. I treat equity as a criterion rather than as a separate strand of work because equity is constitutive of these practices rather than a commitment running alongside them, and the claim is narrower and more specific than it may first appear. In a practice whose object is producing an account of a phenomenon, questions about whose participation is supported and whose knowledge is recognized can be asked about the practice from outside it. One can ask whether participation in a modeling lesson was equitably distributed without that question bearing on what a model is or what makes one adequate. Argumentation does not permit the same separation. Its object is not only an account but the standard by which accounts are judged, and a standard for what counts as adequate reasoning is simultaneously a determination of whose ways of justifying a claim will register as justification at all. The disciplinary question and the equity question are, in this case, the same question asked in different ways. Equity, as a result, cannot be named as a practice of its own. A practice named separately is a practice that can be omitted, and locating the question of whose knowledge counts outside the work of establishing criteria licenses exactly the enactment this argument denies. I take up what that failure produces, and what it is called, in the description of Practice 5 (see page 19).

Windschitl and Calabrese Barton (2016) describe rigor as a characteristic of the interaction between learners and those supporting them, codetermined by the standards particular to a task, the support the teacher offers, and the intellectual activity learners engage in. I extend that account to equity in the case of argumentation. A class that raises the intellectual demand without widening the range of reasoning that can meet it has not produced rigor, and a class that widens participation while leaving the standard unexamined has not produced equity. The nature of scientific argumentation makes this interdependence unusually visible, because the questions raised earlier about whose contributions are recognized bear directly on what a class comes to treat as adequate reasoning.

To specify what I mean by a focus on equity, I draw on Philip and Azevedo’s (2017) analysis of how equity has been framed in science education. They identify four framings, which I refer to as equity foci on the grounds that the term more directly captures what their framework asks teachers and researchers to attend to inside the classroom. These are access, meaning whether all students reach high-quality science instruction; identity, meaning whether students from all backgrounds can see themselves as people who contribute to science; disciplinary engagement, meaning whether students’ ideas are taken seriously as contributions to the work of science rather than as material to be replaced by canonical knowledge; and justice, meaning whether school science connects to the communities it is meant to serve. Philip and Azevedo argue that the field has concentrated on the first two. I aim at all four, with the greatest weight on disciplinary engagement, which argumentation is best positioned to support, and with more contingent openings toward justice. It is important to note, however, that these foci are more useful for locating a commitment to a goal than for guiding the work it takes to achieve that goal, so in describing the practices I attend to what a teacher does rather than to which focus is in play.

A practice meets this criterion, in operational terms, when it creates openings for teachers to recognize and build on the argumentation repertoires that students bring rather than to bypass them (González et al., 2005; Gutiérrez & Rogoff, 2003; Medin & Bang, 2014; Rosebery et al., 2010). Argumentation requires participants to draw on what they know and have experienced in the moment, and when teachers do not work actively to surface that knowledge, argumentation defaults to a narrow band of officially sanctioned ideas, ways of knowing, and ways of participating in a practice (Bang et al., 2013; Calabrese Barton & Tan, 2020; Kang & Zinger, 2019). This commitment to equity, as a result, must run through all seven practices.

4.4.2. The Practice Is Doable at a Productive Grain Size

My second criterion is that a practice should represent something doable for pre-service and in-service teachers, which depends heavily on the grain size at which it is specified. Pre-service teachers are often presented with images of ambitious instruction that feel inaccessible given the constraints of the classrooms they enter (E. A. Davis et al., 2017; Kang & Windschitl, 2018) and in-service teachers are understandably reluctant to adopt approaches that appear to require wholesale change in how they teach (K. S. Davis, 2003; Garet et al., 2001; Johnson, 2006). I therefore sought practices that could be named, modeled, rehearsed, and analyzed, and that could be taken up incrementally (Grossman et al., 2009; Kazemi et al., 2009; Lampert et al., 2013) without requiring wholesale transformation of an existing context or way of teaching.

An example makes the choice concrete. Consider the work of helping a class develop standards for argument quality. At the grain size Windschitl et al. (2012) use, this work is not separately visible, since it is one aspect of pressing students for evidence-based explanations, a practice that unfolds over the course of an entire unit. At the strategy grain size Gray (2026a) describes, it appears as a specific move, such as generating a “gotta-have-it list” with student input that establishes what an acceptable argument must include. At the grain size I use, it is the recurring work of building, applying, and revising a shared account of what makes an argument acceptable in (and outside of) science, which is accomplished through many such moves and across many lessons. Work at this level must accommodate the fact that what counts as an acceptable argument varies across contexts, domains, and types of inquiry, so that criteria adequate to an experimental investigation will not be adequate to a historical one. It also must include the noticing that allows students to see that variation as a difference in the goals and conventions of different communities rather than as a ranking.

The seven practices I propose therefore sit at an intermediate level, larger than the strategies Gray distinguishes and smaller than the core practices Windschitl et al. name, and roughly where Kloser’s set and Gray’s high-leverage practices are also pitched. What distinguishes them at this level is not the grain size but the disciplinary work they require. I chose this level because a practice named here is broad enough to be enacted differently across grade levels and content areas, yet specific enough that a teacher educator can say what it looks like, what it sounds like, and how to tell whether it is going well or not. A novice can enact it, receive feedback, and improve (Lampert, 2010), and an experienced teacher can use it as a focus for sustained collaborative work (Thompson et al., 2019). A larger grain size obscures what the teacher actually does, and a smaller one is difficult to enact apart from the larger pattern of work in which it is embedded.

4.4.3. The Practice Is Usable Across Contexts

My third criterion is that a practice should be usable across the range of curricula, instructional models, and policy environments in which teachers work. A set of practices that requires a particular program to be in place would have limited reach. The seven practices are therefore described as forms of work rather than as specific activities. A teacher using an existing model-based inquiry (Windschitl et al., 2008) or storyline (Reiser et al., 2021) unit, facilitating an Argument-Driven Inquiry lesson (Sampson et al., 2011), or providing students with a technology-enhanced tool designed to support student-to-student interactions during an activity can each engage in the work of making argumentation public, though the routines and resources they draw on will differ. This is not a claim that all instructional materials, instructional models, and technology-enhanced tools are equally hospitable to argumentation. It is a claim that the work I describe is foundational in a way that is not tied to any single resource or approach.

4.5. Candidates Considered and Set Aside

Seven candidate practices did not survive this process, and

Table 1. Candidate practices considered and set aside.

Candidate	Why It Was Considered	Why It Was Set Aside
Eliciting students’ initial ideas	Widely documented and consequential for participation	Not specific to argumentation, and distributed across Practices 1, 2, and 7
Teaching the structure of an argument explicitly	Appears in most argumentation- focused curricula	A move within Practice 4, and one that produces the near-miss when enacted without the other practices in place
Pressing students for evidence-based reasoning	The single most documented teacher move in this literature	Sits at the move grain size rather than the practice grain size, and is one way of enacting Practices 3, 5, and 7
Assessing and providing feedback on written arguments	Consequential for what students conclude the work is for	Largely Practice 5 applied to individual products after the fact, and therefore not doing the second kind of work required
Organizing and managing small-group collaboration	Necessary for most argumentation to occur at all	General to collaborative instruction and already named in existing frameworks (Kloser, 2014)
Connecting science to students’ lives and communities	Central to equitable participation	Treated as an aspect of Practice 2 and as constitutive of all seven rather than separate, since naming it as its own practice would license enacting the others without it
Building consensus and bringing arguments to closure	Classes must eventually settle questions	Set aside because premature closure is what Practice 1 holds off, and because closure managed by the teacher is the mechanism through which epistemic assimilation occurs (Stroupe et al., 2025)

records them along with the reason in each case. I include them because the set I propose is a configuration rather than a discovery, and a reader is better positioned to evaluate it knowing what was left out and why. I do not claim that this is the only defensible configuration. I claim that it is the configuration that survives the criteria I have made explicit, that each practice in it does both kinds of work the theoretical framework requires, and that the set is small enough to serve as a curricular spine for teacher preparation while remaining specific enough to say what a teacher is doing and how to tell whether it is going well. A reader who is interested in drawing the boundaries differently can use this information to make an informed decision, which is also a contribution to the field.

5. The Seven Foundational Teaching Practices

Scientific argumentation, as I have noted, is a process in which students propose, support, critique, and challenge claims about how or why the world works. There are many ways for teachers to support students in this work, and the seven practices below are not intended to exhaust them. They are, rather, the practices that met the three criteria I described above. For each, I define the practice, explain why it matters for student learning, and close by considering how it carries implications for equity. After describing all seven, I take up how they relate to one another, how each can be distinguished from enactments that resemble it, how the set stands in relation to existing frameworks of core practice, and what is at stake for equity across the set.

5.1. Practice 1: Creating and Leveraging Instances of Uncertainty

Creating and leveraging instances of uncertainty refers to the work of designing tasks that contain genuine intellectual openings and of sustaining and using those openings once they are on the table. It includes the work teachers do in advance, when they select or design tasks whose available evidence supports more than one defensible account, and the work they do once uncertainty is in play, when they hold off on adjudicating long enough for competing accounts to be developed and compared (Y. Chen et al., 2019; Y.-C. Chen & Qiao, 2020; Ha et al., 2024; Manz, 2015; Watkins & Manz, 2022). Recognizing that an uncertainty has arisen, particularly when it arises unbidden from something a student says, is closely related work, but it belongs to a different practice. I treat it under Practice 7 (see page 22), where I also take up how the two practices depend on one another.

This practice matters because argumentation is a response to uncertainty (Ford, 2012; Manz, 2015; Osborne, 2010). When the answer to a question is already known to all participants, there is nothing to argue about, and what passes for argumentation becomes an exercise in retrieving an expected answer rather than weighing alternatives. Studies of classroom argumentation indicate that productive episodes are frequently anchored in moments when students find that they disagree, that their data conflict with their predictions, or that two accounts might both fit what they have seen (Berland et al., 2016; Y. Chen et al., 2019; Engle & Conant, 2002). Openings of a second kind concern the investigation itself. Whether a comparison was fair, whether an instrument measured what the class intended it to measure, and whether a procedure could have produced the pattern in the data are questions students can take up productively, and they are questions scientists argue about routinely (Ford, 2008a; Kind & Osborne, 2017). Openings of a third kind concern sufficiency. How many trials are enough, how much agreement across groups is enough, and how far a class should trust a pattern observed once are not questions the data answer on their own, and a class that never confronts them learns that evidence either exists or does not rather than that its adequacy is something to be argued for. Teachers who sustain any of these moments support more substantive thinking than teachers who hurry past them toward a settled conclusion (Y. Chen & Techawitthayachinda, 2021; Y.-C. Chen, 2025; Ha et al., 2024; McNeill & Pimentel, 2010). Where this practice is well developed, disagreement and confusion are read as openings for learning rather than as obstacles to be cleared away, and the basis on which a question is settled shifts from the teacher’s authority to grounds the class can examine and contest.

Uncertainty takes different forms depending on the style of reasoning a task calls for, and a teacher designing for uncertainty is therefore designing for a particular kind of it (Kind & Osborne, 2017). In an investigation organized around experimental evaluation, the open question is often whether a difference can be attributed to what was manipulated or to variation the design did not control. In one organized around categorization and classification, it is whether the features used to sort cases are the ones that matter. In one organized around historical or evolutionary reasoning, it is which of several accounts best explains traces that cannot be reproduced. Probabilistic reasoning is a case of its own, since uncertainty there is not an obstacle to be resolved but the object of the argument itself. Uncertainty that remains generic, in the sense that the class simply does not yet know the answer, tends not to generate argument, because students have not had a chance to decide what would count as making progress on it.

The openings a teacher creates are not neutral with respect to who can enter them. A phenomenon is genuinely uncertain only for students who know enough about it to have grounds for doubt, and one drawn from a narrow band of experience will leave many students without an opening (Bang & Medin, 2010; Rosebery et al., 2010). It is therefore important to also draw on uncertainties that students themselves notice in their own lives and communities (Warren et al., 2020). Sustaining uncertainty carries a related risk, since the pressure to resolve typically comes from the students who already hold the expected answer, and yielding to it forecloses the work for everyone else.

5.2. Practice 2: Framing Joint Enterprise

Framing joint enterprise refers to the work of proposing to students, and negotiating with them, what the class is collectively trying to figure out, how the class will decide when it has figured it out, and why both of those questions matter. It presupposes that something is genuinely at issue, which is what Practice 1 supplies, and it does the work of converting that opening into a problem the class owns. An uncertainty the teacher has engineered but that students experience as the teacher’s puzzle produces compliance rather than argumentation, and the difference lies in whether the class has taken the question up as its own. Framing joint enterprise is therefore the practice through which a teacher establishes a shared problem space, names the phenomenon under investigation as something the class is working on together, and positions individual contributions as contributions to a collective effort rather than as performances for the teacher. I take the term joint enterprise from Wenger (1998), where it names the negotiated purpose that holds a community’s activity together and the mutual accountability that purpose generates, including a community’s sense of what has to be justified and what is good enough. Because that purpose is negotiated rather than assigned, the teacher’s work here is to open and sustain the negotiation rather than to settle it. Framing joint enterprise is not a preliminary step that a teacher completes at the start of a unit and then sets aside. It is recurring work that the teacher returns to as the class encounters new findings, new questions, and new disagreements (Engle & Conant, 2002).

The enterprise a teacher frames for argumentation has a second object alongside the phenomenon, which is the question of what will count as knowing. Is this analysis of the data appropriate to the question being asked? Is the interpretation of that analysis warranted? What makes evidence of this kind acceptable as support for a claim of this kind, and on whose authority? A teacher who frames the enterprise as explaining the phenomenon alone leaves these questions to be settled by default, and the default is usually the teacher’s approval. A teacher who frames them as part of the collective work opens them to examination, which is what allows the style of reasoning in play to become visible as a choice rather than as the nature of things. This is framing work rather than criteria work. Practice 5 (see page 19) takes up the recurring business of building, applying, and revising criteria; what Practice 2 does is establish that the criteria are part of what the class is figuring out.

This practice matters because the social organization of argumentation depends on participants seeing themselves as engaged in shared intellectual work (Berland & Hammer, 2012; Engle, 2012; González-Howard & McNeill, 2019). When students do not understand what the class is trying to figure out, or when they perceive the work as the teacher’s rather than their own, the conditions that make argumentation meaningful break down (Berland et al., 2020; Ford, 2012). Students who do not see a problem as theirs are unlikely to take risks in proposing ideas, to invest in critiquing the ideas of others, or to hold themselves accountable to disciplinary standards of evidence. Studies of argument-rich classrooms indicate that teachers who consistently frame the work as a collective endeavor support more sustained engagement and more equitable participation than teachers who frame it as a series of teacher-directed tasks (Berland & Reiser, 2011; Jiménez-Aleixandre et al., 2000; McNeill et al., 2017).

Who is included in the collective a teacher invokes, whose questions are taken up as worth collective attention, and whose ways of formulating a problem are recognized as legitimate starting points all shape who is positioned as a contributor (Bang & Medin, 2010; Calabrese Barton & Tan, 2020; Philip & Azevedo, 2017). The same holds for what the class is told will settle a disagreement. A teacher who frames experimental results as the only evidence that counts forecloses forms of reasoning central to the historical sciences and to knowledge traditions outside school science alike, and does so before any student has had occasion to offer one. Teachers, therefore, must attend to not only whether students see the question the class is taking up as meaningful but also to whether the terms on which the class will settle it leave room for the ways of reasoning students bring as well as the styles of reasoning that are privileged in fields of science that are often ignored in science classrooms.

5.3. Practice 3: Making Argumentation Public

Making argumentation public refers to the work of bringing student claims, the evidence offered for them, and the disciplinary grounds given for treating that evidence as acceptable support into shared view, so that the class can examine the relationships among them, weigh competing accounts, and revise arguments in response to critique. It includes representations such as whiteboards, posters, and shared documents, as well as the discourse moves teachers make when they ask students to say not only what they think but why, and how they would justify their position if pressed. The tools that support this work are the subject of Practice 4 (see page 18), and the two are closely coupled in enactment even though they name different forms of work.

What differs from the public records that support sensemaking is what goes on the record. A record that supports argumentation must show competing claims, the evidence brought to bear on each, the grounds offered for treating that evidence as acceptable, and the points at which the accounts diverge (Duschl, 2007; Jiménez-Aleixandre, 2007; Osborne et al., 2004). It also must make visible the basis on which the class is judging, since the analysis a group performed, the interpretation it drew from that analysis, and the standard it applied are themselves open to challenge. A record that shows what students concluded without showing how they arrived there, and by what standard, gives the class an account to accept rather than an argument to weigh.

What needs to be brought into shared view also depends on the style of reasoning in play, because the points at which an argument becomes contestable differ from one style to another (Kind & Osborne, 2017). A representation supporting an argument built through experimental evaluation needs to make the structure of the comparison visible, since that is where the argument is most open to challenge. One supporting an argument built through categorization and classification needs to make the sorting criteria visible. One supporting an argument built through historical or evolutionary reasoning needs to set competing accounts alongside the traces each must explain. A representation that displays claims and evidence without exposing the point of contest lets a class look at an argument without giving it anything to argue about.

This practice matters because argumentation depends on the availability of competing claims and the grounds offered for them (Clark & Sampson, 2006; Duschl & Osborne, 2002; Osborne, 2010). When arguments remain tacit, or when only the claim is public while the evidence and justification stay implicit, the class loses the ability to examine competing accounts and to develop a shared sense of disciplinary standards (Duschl, 2007, 2008; Jiménez-Aleixandre & Crujeiras, 2017; G. Kelly & Licona, 2017). Studies of argument-rich classrooms have shown that public representations capturing argument structure, together with discourse moves that surface justification, support more sophisticated argumentation and more equitable participation than instruction relying on private teacher–student exchanges or on representations that capture only conclusions (Y.-C. Chen et al., 2016b; Grooms et al., 2018; Walker et al., 2019; Walker & Sampson, 2013). The shift does more than expand what is externalized. It changes what the class treats as a contribution, because it holds students accountable not only for what they think but for the grounds they can offer.

Making thinking public raises the cost of contributing, and that cost is not evenly distributed. Students whose ways of speaking align with the register of school science risk less by being visible than students who do not (González-Howard & McNeill, 2016; Licona & Kelly, 2020). The public record carries a second risk, since an idea recorded in the teacher’s phrasing can circulate under a student’s name in a form its author would not recognize, which converts a contribution into something done to a student rather than by one. It is therefore important for teachers to not only attend to whether arguments are visible but also attend to whose evidence the record admits, whose names stay attached to the claims on it, and whether the conventions governing what goes on the record are themselves available for examination.

5.4. Practice 4: Introducing and Mediating Argumentation Tools

Introducing and mediating argumentation tools refers to the work of selecting tools that can scaffold argumentation, introducing them so that their purpose and use are visible, modeling their use in the course of actual disciplinary work, and continuing to mediate that use over time so that the tools support rather than substitute for the substantive work of arguing. I name this practice in terms of mediation rather than provision because the literature is clear that supplying tools is not what makes them productive. What makes them productive is the ongoing work a teacher does to keep their use tied to the disciplinary purposes they were designed to serve (González-Howard & McNeill, 2016; McNeill & Krajcik, 2009; Thompson et al., 2019; Windschitl et al., 2012). This practice is closely coupled with Practice 3 (see page 17), since many of these tools exist precisely to make argumentation public, but the work of mediating a tool is distinct from the work of bringing arguments into shared view.

The tools in scope include representational tools such as graphic organizers, argument structure frameworks, and other representations (Lombardi et al., 2018; McNeill et al., 2006; McNeill & Knight, 2013; Simon et al., 2006; Walker et al., 2019), discursive tools such as science circles, talk moves, and sentence starters (Alzen et al., 2025; González-Howard & McNeill, 2020; McNeill et al., 2026), content resources such as data, models, and readings (Kang et al., 2016; Reiser, 2018; W. Sandoval & Reiser, 2004), and the public records teachers and students build together to track important ideas, findings from investigations, and unresolved questions across a unit (Gray, 2026a; Windschitl et al., 2018). What unites them is that each is meant to make some aspect of argumentation more visible or more amenable to collective work than it would otherwise be.

This practice matters because argumentation makes considerable demands on students, particularly when they are new to a particular style of argument (Grooms et al., 2018; Jiménez-Aleixandre et al., 2000; Sampson et al., 2011). Well-designed and well-used tools make the moves of the practice visible and make the structure of an argument available for inspection (Belland et al., 2008; Sampson et al., 2011; Sampson & Walker, 2012). It is important to note, however, that the same tool can support sophisticated argumentation in one classroom and a procedural completion task in another, and the difference appears to lie in how the teacher introduces, models, and mediates it over time (González-Howard & McNeill, 2019; McNeill et al., 2017, 2018; McNeill & Knight, 2013). For example, providing students with an argument graphic organizer to be filled in and submitted for a grade at the end of a lesson produces one kind of student work, but using the same tool as a way to put the grounds behind a claim in front of the class during a lesson in order to facilitate critique and revision produces another. Absent that mediating work, tools tend to be absorbed into existing classroom routines, and argumentation becomes an exercise in filling in boxes and completing assignments (Campbell et al., 2019; Jiménez-Aleixandre et al., 2000; McNeill et al., 2017).

Part of the mediating work this practice names is supplying what the tool leaves out. The widely used argumentation templates are style-agnostic by design, which is what makes them portable across content but also what makes them liable to become form-filling. A claim, evidence, and reasoning frame does not tell a student that the contestable point in an experimental argument lies in the structure of the comparison, or that in a classification argument it lies in the criteria used to sort, or that in a historical argument it lies in whether independent lines of evidence converge on the same account (Kind & Osborne, 2017). A teacher who introduces the template alongside the question of where this particular argument is most open to challenge is doing something categorically different from a teacher who distributes it as a form.

It is important to note here that instructional tools carry assumptions about the linguistic and cultural resources students bring, and studies of multilingual classrooms have shown that written argument structures can either provide a shared frame for participation or become an additional barrier to it depending on how the teacher mediates their use (González-Howard & McNeill, 2016; Infante & Licona, 2021; Licona & Kelly, 2020, 2022). The mediation is therefore where the equity question lives rather than in the tool, which is also why a tool that helps one class can be inert in another. Teachers, as a result, must not only attend to which tools they choose to bring into their classroom, which ones they leave out, and which ones need to be modified before they are put into use but also how they choose to mediate the use of that tool.

5.5. Practice 5: Establishing and Refining Shared Criteria for What Counts as an Acceptable Argument

Establishing and refining shared criteria refers to the work of developing, with students, a shared account of what the discipline treats as an acceptable argument, what makes evidence relevant, and what makes a claim worth taking seriously. The phrasing matters. Criteria of this kind are not properties of an argument considered in isolation. They are standards a community has come to treat as authoritative, and part of the work is helping students see them as such rather than as arbitrary rules (Ford, 2008a, 2012; Kind & Osborne, 2017; Medin & Bang, 2014). The work recurs at the start of an investigation when criteria are introduced, throughout it as they are tested against student work, and across investigations as they are refined in light of what the class has learned about the standards of the discipline.

This is the one practice in the set with no counterpart at the practice level in the frameworks discussed earlier, and the nearest analogues presuppose criteria rather than establish them. Providing feedback (Kloser, 2014) requires a standard against which work is judged but does not name the work of arriving at one. Teaching toward a clear learning goal (Gray, 2026a) locates the goal with the teacher. Pressing students for evidence-based explanations (Windschitl et al., 2012) presupposes a standard for what makes an explanation evidence-based rather than naming the work of arriving at one. In a classroom organized around building an account of a phenomenon, that presupposition is defensible, since the discipline’s standards can be carried by the teacher and applied to student work. Argumentation cannot proceed on that arrangement. A class weighing competing accounts needs a shared basis for preferring one, and a basis supplied by the teacher returns the class to teacher approval by another route. The standard itself therefore must become an object of collective work, which is what distinguishes this practice from the rest of the set.

This is also the point at which the disciplinary overview I drew on in the framework does its work, because it is what tells a teacher what a class’s criteria are criteria for. These standards are not uniform across the sciences. Kind and Osborne (2017) argue that scientific reasoning comprises six distinct styles, each carrying its own procedural and epistemic constructs, so that what makes an argument acceptable depends on the style of reasoning associated with it (see

Table 2. Style-dependence of criteria for an acceptable argument.

Style of Reasoning	What the Argument Establishes	What the Criteria Attend to
Mathematical deduction	That a relationship holds necessarily, given the representation adopted.	Whether the representation captures the relevant quantities and whether the derivation follows validly.
Experimental evaluation	That a manipulated factor accounts for an observed difference.	Whether the comparison isolates the factor claimed and whether the difference exceeds what ordinary variation would produce.
Hypothetical modeling	That a proposed mechanism accounts for a phenomenon.	Whether the mechanism explains what is observed and generates expectations that can be checked against further evidence.
Categorization and classification	That the entities under study are of distinct kinds.	Whether the features used to sort are applied consistently and whether the resulting categories do explanatory work.
Probabilistic reasoning	That a pattern in a population supports a prediction within stated bounds.	Whether the sample supports the inference and whether uncertainty is represented rather than suppressed.
Historical and evolutionary reasoning	That a past event or sequence best explains the traces available in the present.	Whether the account is the best available explanation of those traces and whether independent lines of evidence converge on it.

). An argument built through experimental evaluation, for example, is judged partly on whether the comparison was controlled well enough for the difference observed to be attributable to the manipulation. An argument built through categorization and classification, in contrast, is judged on whether the criteria used to sort cases are consistent and consequential and an argument built through historical and evolutionary reasoning is judged on whether the account offered is the best available explanation of traces that cannot be reproduced. A class that has developed criteria for one style has not thereby developed criteria for another, which is part of why this practice recurs rather than resolving.

This practice matters because argumentation depends on shared standards for judging competing ideas (Christodoulou & Osborne, 2014; Duschl, 2007, 2008; González-Howard & McNeill, 2020). When students do not know what makes one argument more acceptable than another, or when the only standard they recognize is teacher approval, argumentation collapses into either the production of expected answers or the trading of personal opinions (Berland & Reiser, 2011; Jiménez-Aleixandre et al., 2000; Stroupe, 2014). Students in classrooms where criteria are made explicit, used as a basis for evaluation, and refined over time tend to produce more sophisticated arguments and to develop a more nuanced sense of how scientific knowledge comes to be accepted (Y.-C. Chen & Qiao, 2020; Jiménez-Aleixandre & Crujeiras, 2017; Jin & Kim, 2021; Nussbaum et al., 2008). It is important to note that criteria are not best handled as a fixed list handed to students, since criteria built up through use are more likely to be meaningful and more likely to be applied consistently in self- and peer-evaluation (Berland et al., 2020; González-Howard & McNeill, 2020; Stroupe, 2014).

This practice also carries a hazard that the others do not, and I want to name it directly. Stroupe et al. (2025) argue that when those with authority establish the standards and the measures, mandate the correct way of knowing, and treat departures from it as error, the result is a form of testimonial injustice they call epistemic assimilation, in which responsibility for a perceived shortfall is placed on the learner. A teacher who presents criteria for an acceptable argument as settled disciplinary standards, and who then measures students against them, does the thing they describe. The risk is not incidental to this practice. It is what the practice becomes when the first word in its name is dropped and the criteria are supplied rather than established.

Two things follow for how the practice is enacted. The first concerns the standing of the criteria themselves. Because styles of reasoning are historically contingent rather than universal, the criteria attached to them are open to examination, and part of the work is helping students see why a given standard came to be treated as authoritative and what it includes and excludes (Ford, 2008a; Philip & Azevedo, 2017). A class that can ask why controlled comparison earns the credibility it does is in a different relation to the criterion than a class that has been handed it. The second concerns what the criteria treat as acceptable justification. Criteria recognizing only canonical forms of evidence or treating the conventions of mainstream academic argument as the only legitimate way of establishing that something is so produce contributory injustice, in which a student’s justification fails to register not because it is weak but because the criteria have no place for it (Bang et al., 2013; Stroupe, 2022; Stroupe et al., 2025; Warren et al., 2020). Whether a class can put its criteria in question depends in turn on the noticing taken up in Practice 7 (see page 22), since a standard can be revised only once someone has registered that it does not fit.

5.6. Practice 6: Developing and Maintaining Norms for Critique and Disagreement

Developing and maintaining norms for critique and disagreement refers to the work of establishing, with students, shared expectations for how they will propose, support, critique, and challenge one another’s ideas, together with the ongoing work of holding the class accountable to those expectations. It includes articulating norms explicitly, modeling what they look like in the course of disciplinary work, and the moment-to-moment work of responding when norms are upheld or breached (Alzen et al., 2025; Engle et al., 2014; González-Howard & McNeill, 2020; Jang & Hand, 2017; Sampson et al., 2013). I name this practice in terms of critique and disagreement rather than communication and participation more generally because governing how disagreement is conducted is distinct from organizing participation in general.

This practice matters because scientific argumentation is a social practice, and the social conditions under which it takes place shape what is possible within it (Duschl, 2008; Ha & Kim, 2022; Kuhn et al., 2013; Yun & Kim, 2015). Classrooms in which disagreement reads as personal conflict, in which students hesitate to challenge one another for fear of damaging relationships, or in which a few students do all the critiquing tend not to support substantive argumentation (Sampson & Clark, 2011; Southerland et al., 2005). Studies of argument-rich classrooms indicate that productive argumentation depends on norms that explicitly authorize challenge, that frame critique as a contribution, and that distribute the work of disagreement across the class rather than concentrating it in a few students (Alzen et al., 2025; Berland & Hammer, 2012; Cavagnetto et al., 2010; González-Howard & McNeill, 2019). It is important to note that norms of this kind cannot be posted on a wall and assumed to take hold. They are developed, modeled, and maintained over time, and while the work of maintaining them is shared, the teacher carries the primary responsibility for naming and modeling them as needed.

Without explicit attention from the teacher, the work of critique tends to be taken up by students already comfortable challenging others in public (Engle et al., 2014; Kang et al., 2016). A related problem concerns credibility rather than participation. A class does not extend equal weight to every challenge, and the challenges that carry are not always the strongest ones (Stroupe et al., 2025). A teacher must also attend to the register of the critique. Direct, individual, oppositional disagreement is one convention among several, and treating it as the default aligns the practice with the argumentation repertoires some students bring to the classroom while running against those of others (Bang & Medin, 2010; Lee et al., 2013; Warren et al., 2001). Therefore, the pedagogical question is not which register is correct but how the class comes to recognize several as available and to judge which one is most suitable for a given exchange. Finally, and perhaps most importantly, teachers must attend to whose ideas get critiqued. A classroom in which canonical claims are treated as beyond challenge while students’ contributions face sustained scrutiny reproduces a structure of authority that argumentation might otherwise help the class examine (Philip & Azevedo, 2017).

5.7. Practice 7: Noticing and Responding to Student Reasoning and Needs

Noticing and responding to student reasoning and needs refers to the work of attending closely to what students say, write, and do during argumentation and using what is noticed to shape what happens next. The practice has two moments that are analytically separable even though they run together in enactment. Noticing is the work of registering the substance of a contribution, including the openings a contribution creates. Responding is the work of building the next instructional move from it. I use the term responsiveness for the pair because it carries specific meaning in the science education literature (Berland et al., 2020; Hammer et al., 2012; D. M. Levin et al., 2009; D. Levin et al., 2012). Responsiveness in this sense is not a general disposition to listen well. It is an instructional stance in which students’ ways of thinking, questions, and puzzles are treated as the material from which the next move is made.

Among the things a teacher notices, uncertainty deserves particular attention, because it is what connects this practice to Practice 1 (see page 15). A teacher can design a task rich in intellectual openings and still lose these opportunities for argumentation, since uncertainty frequently arrives at unexpected times, in an offhand remark, an anomalous result, or a disagreement between two students who have not realized they disagree. Recognizing such a moment as an opening rather than as a digression is noticing work, and it is a precondition for the leveraging that Practice 1 (see page 15) describes. The two practices are therefore mutually dependent. Designing for uncertainty accomplishes little if its arrival goes unregistered, and noticing an opening accomplishes little if the teacher has no repertoire for sustaining it.

This practice matters because two converging lines of work establish that what teachers attend to shapes what they take up. Studies of teacher noticing have shown that attention directed primarily at whether students have reached the canonical answer tends to pass over productive lines of thought that diverge from the expected path (Y.-C. Chen et al., 2017; Colley & Windschitl, 2016). What a teacher does next depends on how a contribution has been interpreted rather than on what was said, so the same student utterance can occasion a press for grounds, a redirection, or nothing at all. Studies of follow-up moves suggest that what teachers say after students contribute drives the depth of subsequent argumentation more than the questions they pose initially (Carpenter et al., 2020; Pimentel & McNeill, 2013). A teacher responsive in this practice-specific sense is not tracking what students think in the moment so much as how the class is arguing, where the argumentation is gaining traction, and where it is stalling or narrowing. The moves that follow, such as pressing for the grounds behind a claim, surfacing a competing interpretation, slowing the conversation to examine a single piece of evidence, or naming an emerging disagreement so the class can take it up, are made in service of the argumentation rather than in service of a predetermined endpoint (Y.-C. Chen, 2020; González-Howard & McNeill, 2019).

Noticing is patterned by expectation, which makes it the point at which the other practices most often fail without anyone observing that they have. Reasoning that does not take an anticipated form is liable to register as confusion or as off-task talk rather than as a contribution, and a teacher who does not hear it as reasoning has no occasion to respond to it as such (Krist et al., 2023; Robertson et al., 2016; Warren et al., 2020). This is why I treat noticing as constitutive of the practice rather than as a disposition operating outside it.

5.8. How the Practices Relate to One Another

Having described each practice on its own, I want to return to how they work together, since a set of seven independently described forms of work is not yet a framework. Figure 1 shows how the seven practices stand in relation to one another. The practices are not a sequence, and Figure 1 is not a flowchart. They are related in three ways, and naming those relations makes clear why I treat them as seven forms of work rather than three or twelve.

The first relation is generative. Practice 1 supplies what the other practices operate on. A class with nothing genuinely at issue has no rival accounts to make public, to represent with a tool, to judge against criteria, or to disagree about, and argumentation without uncertainty reduces to the retrieval of an expected answer. The remaining practices can be enacted faithfully in such a classroom without producing argumentation at all, which is why Figure 1 places this practice outside the frame rather than within it. An intellectual opening is a condition of the work rather than a feature of how the work is organized, and a teacher who has not created one has nothing for the other practices to do.

The second relation is enabling. Practice 2 turns an opening into collective work. An uncertainty that a teacher has created but that students do not experience as theirs to settle produces compliance rather than argumentation, and asking students to make thinking public, to weigh ideas against criteria, or to respond to one another rarely generates substantive engagement when they do not see the underlying problem as their own. What Practice 2 frames, moreover, has two objects rather than one. The class is figuring out something about the world, and it is also figuring out what will count as knowing it, and the second of these is what the practices inside the frame elaborate. This is why the frame in Figure 1 encloses the four that follow. Practices 3 through 6 are not additional conditions on the work but the forms that work takes once a class has taken an uncertainty up as its own.

The third relation is division of labor among the four practices inside the frame. Practices 3 and 4 make rival accounts available for collective examination, one by putting claims, evidence, justifications, and points of divergence where the class can see and revise them, the other by mediating the representations through which students construct and examine those accounts. Practices 5 and 6 supply the two standards a class needs, and the distinction between them is worth holding onto. Practice 5 governs what counts as adequate reasoning, and Practice 6 governs how disagreement is conducted. Classrooms routinely have one without the other. A class can be scrupulously civil while having no shared account of what makes one argument stronger than another, and a class can hold sophisticated criteria that no one dares apply to a classmate’s work.

Practice 7 stands outside this ordering because it is not a further condition or a further form of work but the responsiveness on which the adaptive half of each of the others depends. A teacher cannot take up an opening that he or she has not registered as an opening. This dependency is strongest for Practice 5, since a standard can be revised only once someone has noticed that it does not fit the case in front of the class, which is why the criteria work and the noticing work cannot be separated in enactment even though they can be named apart. Practice 6 carries a dependency of a different kind, since a class whose norms authorize challenging claims but not standards has no way to put its own criteria in question.

It is important to note that the practices overlap in enactment, and I do not treat this as a defect to be engineered away. A single teacher move can enact several at once. A question that surfaces a disagreement between two groups can make thinking public, sustain an uncertainty, and constitute a response to something the teacher has just noticed. The practices are analytically distinct forms of work rather than a partition of classroom time, and what justifies naming them separately is that they fail independently. A teacher can frame a genuine enterprise and then close the uncertainty too quickly, can sustain uncertainty and then leave the competing accounts in students’ heads rather than before the class, and can make argumentation public while leaving the standards for judging it in her own hands. Each of these is a different failure requiring a different response, which is what a framework at this grain size is for.

5.9. Near-Misses in Enactment

Because these practices are described at a level of generality that invites readers to recognize them in what they already do,

Table 3. The seven practices, with enactments that instantiate them and enactments that resemble them.

Practice	Enactment That Instantiates the Practice	Enactment That Resembles It
1. Creating and leveraging instances of uncertainty	The task is built so that the available evidence supports more than one defensible account, and the teacher holds off on adjudicating long enough for competing accounts to be developed and compared.	The teacher poses a question with a known answer and treats the interval before students produce it as uncertainty, closing the discussion as soon as the expected answer appears.
2. Framing joint enterprise	The class works on a question it is trying to settle together and on how it will know when the question is settled, and the teacher returns to both often so that students can say what is at stake and why it is not yet resolved.	The teacher announces an objective and tells students they will argue like scientists, and the activity ends once the class reaches the correct answer.
3. Making argumentation public	Claims, the evidence offered for them, the grounds for treating that evidence as acceptable, and the points at which accounts diverge are recorded where the class can see them, return to, and revise.	Groups present finished arguments in turn to the teacher while classmates wait for their own turn, with no mechanism by which one group’s thinking bears on another’s.
4. Introducing and mediating argumentation tools	The teacher introduces a tool by modeling the thinking it is meant to make visible, then returns to it to ask whether it is helping, adapting or setting it aside when it is not.	The tool is distributed as a worksheet to be completed and collected, and quality is judged by whether the sections are filled in rather than by whether the argument holds.
5. Establishing and refining shared criteria for what counts as an acceptable argument	The class builds a written account of what the discipline treats as an acceptable argument, applies it to arguments students have produced, and revises it when students meet cases it does not handle.	The teacher supplies a rubric at the start of the unit, refers to it when grading, and leaves it unchanged regardless of the arguments students produce.
6. Developing and maintaining norms for critique and disagreement	Disagreement is treated as ordinary and is directed at ideas, and the teacher intervenes on how a critique is made rather than on whether one is made at all.	Norms are posted as rules for respectful talk, and disagreement is discouraged in practice because the teacher redirects whenever an exchange becomes uncomfortable.
7. Noticing and responding to student reasoning and needs	The teacher follows what students are saying closely enough to recognize an unanticipated idea or an emerging disagreement as an opening, and changes what happens next in response.	The teacher circulates, acknowledges contributions warmly, and then proceeds with the planned sequence regardless of what was said.

sets each practice alongside what I call a near-miss, meaning an enactment that carries the surface features of the practice without doing its work. These near-misses are not hypothetical. Each corresponds to a pattern documented in the empirical literature, including the completion of tool-mediated tasks without the disciplinary work the tool was meant to support (Berland & Reiser, 2011; W. A. Sandoval & Millwood, 2005), the use of a single tool in ways that either open up or close down thinking (McNeill & Knight, 2013), and the rapid closing of uncertainty in pursuit of a settled conclusion (Berland & Hammer, 2012; Pimentel & McNeill, 2013).

5.10. Relationships to Existing Frameworks

Table 4. The seven practices in relation to existing frameworks of core practice in science teaching.

Practice	Kloser (2014)	Windschitl et al. (2012)	Gray (2026a)	What Specification for Argumentation Adds
1. Creating and leveraging instances of uncertainty	Linking science concepts to phenomena; engaging students in investigations	Constructing the big idea	Instructional sequences	A phenomenon is selected so that the available evidence supports more than one defensible account, and the opening is sustained rather than resolved, because the object is a contest among accounts rather than engagement with a phenomenon
2. Framing joint enterprise	Building classroom community	Constructing the big idea	Positioning students as sensemakers; teaching toward a clear learning goal	The enterprise has a second object alongside the phenomenon, which is what will count as knowing, and it is negotiated with students rather than assumed
3. Making argumentation public	Facilitating classroom discourse	Helping students make sense of material activity	Constructing and organizing public records; representing student reasoning	What is recorded includes the justifications offered and the points at which accounts diverge, and the record stays open for revision rather than serving as a display of results
4. Introducing and mediating argumentation tools	Constructing and interpreting models	No counterpart, though the analysis of priming and face-to-face tools addresses this work	Strategies	The tool is introduced by modeling the thinking it is meant to make visible and is adapted or set aside when it stops helping, and its use is judged by whether the argument holds rather than by whether it was completed
5. Establishing and refining shared criteria for what counts as an acceptable argument	No counterpart	No counterpart	A gotta-have-it list, at the level of strategies	The entire practice. Criteria are built through use rather than supplied, are specific to the style of reasoning in play, and are revised when the class meets cases they do not handle
6. Developing and maintaining norms for critique and disagreement	Building classroom community; facilitating classroom discourse	No counterpart	Orienting students to one another and to the discipline	Disagreement is treated as ordinary rather than as a breakdown, and the teacher intervenes on how a critique is made rather than on whether one is made at all
7. Noticing and responding to student reasoning and needs	Eliciting, assessing, and using student thinking about science; providing feedback	Eliciting students’ ideas to adapt instruction	Eliciting and responding to student ideas; making sense of students’ participation to inform instruction	What must be noticed is different. The openings are an emerging disagreement, an unanticipated justification, and a contribution the class’s criteria have no place for

Note. The three frameworks are not pitched at a single grain size, and Gray’s does not sit at one point. Entries in the Gray column name high-leverage practices unless otherwise indicated. Where the corresponding work appears at another level of that framework, the level is named instead, which is the case for instructional sequences in row 1 and strategies in rows 4 and 5. Entries reading no counterpart indicate that the framework names no form of work at any level that does what the practice does, rather than that the work would be foreign to it.

sets the seven practices alongside the frameworks they elaborate. Reading across a row shows what an existing framework already names and what remains to be specified once the disciplinary work in play is argumentation. Reading down the fourth column shows the shape of the specification as a whole, which is the shift from producing an account to adjudicating among competing ones and the second object that shift introduces.

5.11. Equity Within the Framework

Table 5. How equity is at stake within each practice.

Practice	Foci Most Engaged	What Is at Risk in This Practice	What an Equity-Oriented Enactment Attends to
1. Creating and leveraging instances of uncertainty	Disciplinary engagement, identity	An opening is genuine only for students who have grounds to be uncertain about it, so a phenomenon drawn from a narrow band of experience distributes uncertainty unevenly before any student has spoken	Whether the phenomenon gives all students grounds for doubt, and whether uncertainty is being sustained for the class or only for those who do not already hold the canonical answer
2. Framing joint enterprise	Identity, justice, disciplinary engagement	Who is included in the collective the teacher invokes, whose formulation of the problem counts as a legitimate starting point, and what the class is told will settle a disagreement	Whether students can see themselves in the question, and whether the terms on which the class will settle it leaves room for the ways of reasoning students bring
3. Making argumentation public	Access, disciplinary engagement	Visibility raises the cost of contributing, and that cost falls unevenly on students whose ways of speaking do not match the register of school science; a public record can also fix an idea in terms its author would not recognize	Whether contributions are represented in terms their authors endorse, and who is bearing the risk of being visible
4. Introducing and mediating argumentation tools	Access, disciplinary engagement	Tools carry assumptions about the linguistic and cultural resources students bring, and the same written structure can supply a shared frame or impose an additional barrier depending on how it is mediated	Whether the tool is making reasoning visible or substituting a format for it, and for which students each is happening
5. Establishing and refining shared criteria	Disciplinary engagement, justice	Criteria that register only canonical forms of evidence produce contributory injustice, in which a justification fails to count not because it is weak but because the criteria have no place for it	Whether the class can put its criteria in question when it meets a contribution they do not accommodate, rather than only applying them
6. Developing and maintaining norms for critique and disagreement	Access, identity	Norms framed as rules for respectful talk suppress disagreement, and whether a critique is heard as critique rather than as rudeness or as deference depends on whose manner of disagreeing the class recognizes	Whether the norms govern how a critique is made rather than whether one is made, and whose way of registering disagreement is legible to the class
7. Noticing and responding to student reasoning and needs	All four	Teachers notice what they expect, so reasoning that does not take an anticipated form goes unregistered, which is the mechanism through which the other six practices fail without anyone observing that they have	Whether unanticipated reasoning is registered as reasoning rather than as confusion or as off-task talk

gathers in one place what has been said about equity across the seven descriptions, so that the commitment running through them is visible as a whole. Reading across a row shows which of Philip and Azevedo’s (2017) foci a practice most engages, what is specifically at risk in that practice, and what an equity-oriented enactment attends to. Reading down the third column shows that the risks are not variations on a single concern that could be lifted out and handled separately. What is at stake when a teacher selects a phenomenon differs from what is at stake when a class applies its criteria, which differs again from what is at stake in what a teacher hears as reasoning. Each risk is internal to the work of that particular practice, which is the case for treating equity as constitutive rather than as an eighth practice, made here in summary form after the descriptions have supplied the detail.

5.12. Enacting the Practices Under Institutional Constraint

The seven practices as I have described them could be read as assuming conditions that few teachers enjoy, and I want to address that reading before describing what enactment looks like when those conditions are absent. The practices are described at their fullest realization because that is what a specification is for, and describing a practice at its ceiling can make it sound like a practice that requires one. None of the seven, however, asks more of a teacher’s autonomy, time, or materials than the frameworks they elaborate already ask. Pressing students for evidence-based explanations (Windschitl et al., 2012) and facilitating classroom discourse (Kloser, 2014) are demanding under a prescriptive pacing guide in exactly the way that making argumentation public and maintaining norms for critique are demanding. What the specification adds is not difficulty but detail about what the difficulty consists of.

It would be a mistake, then, to conclude that the more general frameworks fare better under constraint. They do not fare better, they are silent, and silence is not robustness. A framework that names facilitating classroom discourse without specifying what a class does when its criteria are mandated does not thereby equip a teacher for that situation. It leaves him or her to improvise a response to a problem the framework has not named. Naming the problem is what allows a response to be taught, rehearsed, and evaluated, and I take the account below to be an argument for specification rather than a concession against it.

Standards documents, adopted curricula, pacing guides, and assessments specify what students will be held to, and those specifications frequently privilege one style of reasoning over others, most often by treating controlled experimentation as the form that rigorous science takes (Gray, 2026b; Rudolph, 2005). A teacher may recognize that an argument built through retrodiction or through consilience across independent lines of evidence is rigorous on its own terms and still be required to prepare students for an assessment that does not. Practice 5 asks a class to build, apply, and revise a shared account of what makes an argument acceptable, and a mandated rubric arrives with that account already settled and beyond revision. I want to name this issue rather than write around it, because navigating such requirements is part of the work rather than a condition to be satisfied before the work begins. Three responses are available, and they differ in how much room a teacher needs.

The first is available to every teacher regardless of constraint. A teacher can be explicit with students that the standard being applied is a requirement of a particular institution rather than a fact about how good thinking works. Saying that the state assessment will ask for a claim, evidence, and reasoning in a particular form, and that this is one community’s way of settling what counts rather than the only defensible one, costs nothing in coverage and changes what students learn about the standard’s authority. This is Practice 2 enacted under constraint, since what a teacher frames as the enterprise includes what the class is told will settle a disagreement, and a requirement named as a requirement remains available for examination in a way that a requirement presented as the nature of science does not. The second response asks for more room. A teacher can meet the requirement while widening the range of argument around it, treating the mandated form as one style the class will become fluent in while making space for others in the arguments the class builds together.

Practices 1, 3, and 4 do much of this work, since the choice of phenomenon, the design of public records, and the mediation of a required template largely determine which styles of reasoning the class has occasion to use. The third response asks for the most room and will not always be available. A teacher can make the requirement itself an object the class examines, asking why this form of argument came to be the one assessed, what it captures, and what it cannot capture. Practice 6 is a precondition here, since the norms a class holds determine whether a standard can be challenged at all.

It is important to note that the first response is not a lesser version of the third. Making a choice visible as a choice is what allows a class to hold a standard and its authority apart, and without that separation the second and third responses have nothing to work on. It is also the response least contingent on a teacher’s institutional position, which matters because the teachers with the least room to widen or examine a mandated standard are often those working under the most prescriptive conditions. The practices are not, on this view, contingent on a permissive setting. What varies with the setting is how far a teacher can carry them.

Each practice, correspondingly, has a version available under conditions that permit little else. A teacher who cannot select her own phenomena can still decline to resolve the uncertainty a required phenomenon contains until competing accounts have been compared. A teacher working from a scripted sequence can still record where two groups’ accounts diverge rather than only where they converge. A teacher issued a mandated template can still ask what disciplinary grounds would make the evidence in it acceptable, which satisfies the form while restoring what it omits. These are not diminished enactments to be apologized for. They are the practices operating at the floor rather than the ceiling, and a teacher who does only these things is doing the work.

6. Discussion and Limitations

I have proposed seven teaching practices for supporting argumentation that is scientific in nature, offered as an elaboration of the core practice frameworks that describe science teaching more generally rather than as a replacement for them. Each practice does two kinds of work, shaping the conditions under which substantive scientific argumentation can take place and giving the teacher something to draw on when it does not unfold as planned. Rather than restate the practices here, I want to consider what they add to the core practices conversation, what they leave unresolved, and where the argument I have made is vulnerable.

Windschitl et al. (2012), Kloser (2014), and Gray (2026a) have each proposed sets of core practices for science teaching, and each has been generative. They stand in different relations to argumentation, as I argued earlier, but they share a common orientation toward the work of arriving at an account of a phenomenon. That orientation is what the elaboration I have attempted brings into view, because argumentation asks a class to do something further. A class that is adjudicating among competing accounts needs not only to produce reasoning but to work out what will count as adequate reasoning, and that second object is what the seven practices are specified toward. Two consequences follow that the general frameworks have no occasion to foreground. The first is Practice 5, which has no clear analogue in those sets and yet appears across the argumentation literature as among the most consistent conditions distinguishing argumentation that moves beyond superficial exchange from argumentation that does not (Christodoulou & Osborne, 2014; González-Howard & McNeill, 2020; McNeill et al., 2017; Sampson & Clark, 2011). The second is the style-dependence of the criteria that Practice 5 addresses. Criteria are specific to the style of reasoning in play (Kind & Osborne, 2017), so criteria adequate to an experimental argument are not thereby adequate to an argument built from traces of the past. The question of what a criterion is a criterion for does not arise until the disciplinary work is specified, which is why a framework pitched at science teaching in general leaves it unasked.

It is important to note what this positioning does and does not imply. It does not imply that every practice named in the science education standards requires its own set of teaching practices, which would produce an unusable proliferation. It implies, rather, that specification is worth undertaking where a practice carries enough disciplinary particularity that the general formulations underdetermine what a teacher should do. Argumentation meets that test, in my judgment, because it is the practice through which the products of the others are examined and held to account (Duschl & Osborne, 2002; Ford, 2008a, 2012; Osborne, 2014a). Whether modeling or investigation carries enough disciplinary particularity is a question I have not tried to settle here, and the answer may well differ between them.

The teaching practices described here also leave a good deal unresolved. The largest of these I took up in the preceding section, where I argued that the teaching practices describe forms of work available to teachers without altering the institutional conditions under which teachers work. That the first response there is available to nearly every teacher is worth something, but it is a modest something, and a framework that depends on individual teachers naming a constraint they cannot change is not a substitute for changing it.

Beyond that, the argument carries four limitations I want to name directly. First, for the reasons given in the account of how the framework was developed, a different reader of the same literature might arrive at a different set or at the same practices articulated differently. Delphi studies of expert consensus (Kloser, 2014) and systematic reviews are alternative routes to the same destination, and I would welcome work that subjects this set to either. Second, the grain size at which I have pitched the practices is a choice with costs. A finer articulation would capture more of the specific moves teachers make at the cost of parsimony, and a coarser one would be easier to hold in mind at the cost of analytic traction. Third, a reader looking for sustained engagement with a single justice framework will find the present treatment thinner than a dedicated article would offer. The literature I have drawn on is also predominantly Anglophone and North American or European, which shapes what has counted as evidence for the practices and leaves out traditions of scholarship that would likely alter both the set and its emphases. Fourth, the article is conceptual, and the practices have not been tested in the sense that empirical studies might test them. The contribution is a framework that subsequent empirical work can take up, and that work remains to be done.

7. Implications

The most direct implications are for teacher educators who prepare pre-service science teachers and who design professional learning experiences for those already in classrooms. The seven practices offer a way of organizing argumentation-focused preparation that is more parsimonious than the empirical literature and more specific than frameworks written for science teaching as a whole. A program taking them as a curricular spine would need to do three kinds of work with each. It would need to help teachers develop a conceptual understanding of what the practice involves and why it matters. It would need to provide structured opportunities to enact the practice in low-stakes settings such as rehearsals, microteaching, or simulated discussions before enacting it with students. And it would need to support teachers in analyzing records of their own enactments, attending both to the conditions they set up and to how they responded once students took up what was offered. This is consistent with the practice-based teacher education tradition on which I drew earlier (Lampert, 2010; McDonald et al., 2013; Stroupe et al., 2020), and what the seven practices contribute to that tradition is a specification of what a teacher’s work becomes when a class is not only building an account but adjudicating among competing ones.

Two further implications for teacher educators follow from the account of criteria I have given. The first is that novices need experience with argument in more than one style of reasoning. A preparation program that rehearses argumentation only in the context of controlled investigation will produce teachers fluent in one set of criteria and unprepared for the retrodictive and consilience-based arguments that geology, evolutionary biology, and climate science require, which limits what those teachers can convey about how the sciences reason and narrows whose ways of reasoning their classrooms can recognize (Gray, 2026b; Kind & Osborne, 2017). The second is that preparation should include the work of enacting these practices under constraint. Novices will encounter mandated criteria and assessments that privilege one style and treating that encounter as a departure from good practice rather than as part of it leaves them without a way to act.

The implications for researchers are different in character, since the practices are a conceptual contribution and have not been studied as a set. I do not think the productive question is whether these are the correct seven. A set of practices that held across all contexts is unlikely to exist, and asking whether a given list is right invites exactly the kind of settlement that the contextual variability of this work should discourage. The more useful questions concern function. How does the enactment of each practice vary with the style of reasoning an investigation calls for, with the content at hand, with the students in the room, and with the instructional model in use? Which of the practices are hardest for novices to develop, and does facility with some carry others along? What does it take to enact them in settings where mandated criteria run against the teacher’s judgment? My hope is that the practices function as a shared object around which studies of teacher learning, classroom enactment, and student argumentation can accumulate rather than each study constructing its own framework. Work in the core practices tradition suggests that a shared set, even a contested one, can do that kind of cumulative work for a field (Lampert, 2010; McDonald et al., 2013; Stroupe et al., 2020).

For curriculum designers, the implications are more modest, since the practices are not a curriculum and do not specify what tasks or instructional sequences should look like. They do suggest design priorities. Because each practice involves both setting up conditions and responding to what students do with them, the design question is not which practices a curriculum should support but which aspects of them. Curricula can supply phenomena that contain genuine intellectual openings, representations that expose the point at which an argument becomes contestable, and starting criteria that a class can then revise. They cannot supply noticing, and curricula that script every teacher question and follow-up tend to crowd out the responsiveness that argumentation depends on. Curriculum design also determines, largely by what phenomena it selects, which styles of reasoning students encounter across a year, which makes it one of the more consequential levers on whether students meet science as a plural set of ways of reasoning or as a single method.

My hope is that the seven practices prove useful to teacher educators looking for a way to organize argumentation-focused preparation, to researchers looking to build cumulative work on teacher learning and classroom enactment, and to curriculum designers looking for a vocabulary for the balance between structure and responsiveness. Whether they support equitable argumentation is not settled by the list itself. It depends on whether the choices embedded in them are made visible, and that is work that must be done in classrooms rather than in articles. I expect the practices will look different in five years than they do now, and a framework of this kind should invite exactly that.