How Teachers’ Knowledge of Curriculum Supports Partnering with Students in Their Science Learning

ABSTRACT A key goal of science learning today is to support students in posing and answering questions about phenomena that matter to them while establishing a need to grapple with disciplinary ideas and practices reflected in standards. To be successful in this endeavor, teachers need to learn to partner with students on the direction of their learning while remaining attentive to goals for learning reflected in standards. In this paper, we explore the idea that educators’ developing knowledge of curricular purposes and structures can help support their enactment of instructional materials that help them balance these two goals. Our study draws on 36 teachers’ experiences of learning to teach with instructional materials aligned to the Next Generation Science Standards as part of a field test that included multiple cycles of professional development workshops and enactments. Our findings illustrate how teachers’ knowledge of curriculum can help them make sense of their own growth in teaching.

For more than two decades, science education researchers have sought to map and understand the kinds of knowledge teachers need to teach science effectively and equitably.Much of this scholarship has focused on teachers' pedagogical content knowledge, that is, on their knowledge of strategies needed to help students learn specific content and to address students' difficulties with particular content (Gess-Newsome & Lederman, 1999;Hume et al., 2019).Other lines of scholarship have explored how such knowledge can be developed, either through professional development activities (e.g., Roth et al., 2011) or through embedding supports for knowledge development within instructional materials (e.g., Davis et al., 2017;Marco-Bujosa et al., 2017).Within these lines of inquiry, however, there has been comparatively little scholarship on teachers' knowledge of curriculum, a key dimension of knowledge for teaching identified by Shulman (1987) and later by Magnusson et al. (1999) in science.Such knowledge pertains to the purposes of learning that curriculum seeks to support, as well as to the structures of materials intended to support meeting those learning goals.
Addressing gaps in our understanding of teachers' knowledge of curriculum is important because curriculum plays a central role in improving teaching and learning outcomes.Curriculum materials provide models teachers can use to change their own practice to CONTACT William R. Penuel william.penuel@colorado.eduInstitute of Cognitive Science, University of Colorado Boulder, 1777 Exposition Drive, Boulder, CO 80302, USA align with goals for science learning promoted in those materials (e.g., Davis et al., 2017;Penuel & Gallagher, 2009).In addition, experimental studies have repeatedly shown the promise of high-quality materials for improving student outcomes in science (e.g., Krajcik et al., 2021;Penuel et al., 2015;Schneider et al., 2022).
Another reason why the study of teachers' knowledge of curriculum is important is that the kinds of curriculum materials that align with contemporary standards in science are very different from ones with which teachers may be familiar.Whereas most textbooks and curricula present science as a body of knowledge to be learned, the Framework for K-12 Science Education (National Research Council, 2012) and subsequent reports (National Academies of Sciences Engineering andMedicine, 2019, 2022) call for materials that engage students in framing and pursuing answers to questions about phenomena and problems using science and engineering practices.Further, guidance to curriculum developers writing to the Next Generation Science Standards (NGSS; NGSS Lead States, 2013), which were informed by the Framework's vision, indicates that student questions should help create an "explicit need" for science learning and "have frequent opportunities to feel as if they are driving the learning sequence through their questions and emerging understanding" (NextGenScience, 2021, p. 7).Such curricula demand that teachers do more than drive the learning themselves: they need to partner with students in their knowledge building, giving students a strong say in the questions they pursue and how they pursue them.
In this paper, we examine the knowledge teachers develop of curriculum that supports them in partnering with students to build knowledge collectively in science around students' questions about a given phenomenon or design challenge.We focus specifically on what curriculum knowledge teachers developed and on how such knowledge supported their enactment of and reflection on new instructional materials aligned to the NGSS.Drawing on a purposive sample of 36 teachers who participated in a large-scale field test of curriculum materials designed to align with the NGSS, we analyze the successes and challenges of teachers in using materials to organize instruction around student questions over a three-year period.We found that teachers in the sample did discern key purposes and structures of materials and reported that they supported their implementation, but they also cited the opportunity to gain more experience by repeatedly implementing units with the same instructional model as a key resource for implementation.

Context for the study
Forty-five states and the District of Columbia have adopted standards based on A Framework for K-12 Science Education (NRC, 2012) since 2013.The vision laid out in the Framework for science education is particularly ambitious, in that it aims to ensure all students gain a deeper sense of science as both a body of knowledge and evolving set of practices.
Preparing educators to support this goal has proven a challenge for many reasons.The Framework's vision for teaching is very different from typical practice, especially when it comes to supporting students in engaging in science practices such as asking questions and defining problems (Jimenez-Alexandre et al., 2000;NASEM, 2015;Windschitl & Stroupe, 2017).Challenges teachers face include learning how to elicit and make use of student questions (Harris et al., 2012), how to support students in engaging in academically productive discussions (Michaels & O'Connor, 2015), how to draw on students' own ideas and experiences as resources for learning and position students as knowers and doers of science (Atwater, 2000;Moje et al., 2004;Stroupe, 2014), and how to manage time within the classroom to meet learning goals while also following students' questions, especially when they deviate from teachers' learning plans and canonical science (Miller et al., 2018).
Our study took place within this policy context and focused on a large-scale effort to develop and field test a set of open access instructional materials designed for middle grades science called OpenSciEd.A multi-institutional developers consortium led both the development of OpenSciEd and the field test.The field test involved more than 300 teachers across 10 states, self-selected based on their interest in the initiative and represented by state agency leaders in science (Edelson et al., 2021).The effort provided a useful context for studying the growth of teachers' knowledge of curriculum in ways that supported them in partnering with students in the direction of their learning.

Conceptualizing teacher knowledge of curriculum
To conceptualize teacher knowledge of curriculum, we draw on Magnusson et al.'s (1999) definition, which includes two parts aligned to Shulman's (1987) original conception.One component in this framework pertains to knowledge of curricular purposes, while the other pertains to knowledge of specific curricular programs and materials.Below, we elaborate on how we defined these forms of knowledge in the context of our study and how they overlap with and diverge from earlier conceptions.

Knowledge of curricular purposes
Knowledge of curricular purposes refers to "knowledge of educational ends, purposes, and values, and their philosophical and historical grounds" (Shulman, 1987, p. 8).In our conceptualization, this entails understanding not only the educational standards materials are intended to address as Magnusson et al. (1999) argued was important, but also the wider vision for teaching and learning that materials are intended to help realize.In science today, developing understanding of the standards means gaining a sense of how student understanding is expected to develop over the course of a grade band as well as a unit, as articulated in both standards and curriculum documents (Krajcik et al., 2014).It also entails coming to understand science-as-practice as a vision of proficiency in science, which is supported by organizing curricula around building toward explanatory models of phenomena and solutions to design challenges, rather than on developing inquiry skills separate from science knowledge (Furtak & Penuel, 2019;Reiser, Michaels et al., 2017).

Knowledge of curricular structures
Our definition of the structures of curriculum materials includes both how they are organized into sequences of lessons and units and the key elements of materials' design that teachers need to apprehend to use them with integrity.Our definition privileges knowledge of the instructional model underlying a given set of materials, that is, goals, principles of organization, and activity structures that may be realized across multiple units or sets of materials.In doing so, our definition emphasizes depth of knowledge of specific materials, rather than broad knowledge of structures and activities from a range of curricula, as earlier definitions of curriculum knowledge did (Magnusson et al., 1999;Shulman, 1987).
We emphasize the importance of teachers' knowledge of instructional models because effective use of curriculum materials requires teachers to maintain integrity to the goals and structures of their underlying instructional models.Maintaining integrity to an instructional model can promote both effective science learning and equity, when compared to more commonplace patterns of organizing teaching through a series of activities that present the same content but not in a coherent manner (Wilson et al., 2010).At the same time, asking teachers to implement materials with fidelity without also helping them understand the underlying instructional models used to develop them is unlikely to yield improvements to teaching and learning (Penuel et al., 2011).Knowledge of instructional models underlying a particular set of materials can help teachers make decisions in planning and enacting materials that maintain integrity to their goals while also being responsive to students (Bybee et al., 2006).
An example of an instructional model that is intended to guide the development of NGSS-aligned curriculum materials is the storyline model (Reiser, Novak et al., 2017).As other instructional models in science do (e.g., the 5E model; Bybee et al., 2006), the storyline instructional model has interlocking parts that are intended to work together and that are grounded in specific ideas about how students learn science.For example, units are organized around a potential trajectory of student investigation of an anchoring phenomenon or problem, and lessons are sequenced in a way that is intended to make sense from the student point of view.In addition, lessons are organized around a set of interconnected teaching routines (DeBarger et al., 2010), each of which is intended to help students build understandings of science ideas, practices, and crosscutting concepts identified in the targeted standards, or performance expectations.To support students in building knowledge together, there is guidance for teachers in the use of "talk moves" (Michaels & O'Connor, 2011) that can facilitate academically productive discussions that elicit initial ideas, build understanding, and work toward consensus.The model is grounded in ideas from both constructivist and sociocultural learning theories; at its core, it assumes that students learn science by engaging in activities of "figuring out" key ideas through engaging in approximations of professional science and engineering practices, rather than "learning about" them through lecture or laboratory exercises where students follow a prescribed set of steps to arrive at a pre-defined conclusion (Schwarz et al., 2017).
The storyline model provides a useful context for the study of teacher knowledge of curriculum for three key reasons.First, it is one of just a few instructional models designed explicitly to support the vision of the Framework and of the NGSS.Second, the instructional model attempts to provide teachers with specific guidance for handling a key dilemma teachers face in how to address targeted standards while being guided by students' own ideas and questions.Third, both storyline instructional materials and associated professional development make the instructional model explicit to teachers.It is intended that teachers understand the model, to help guide their teaching in the classroom.

The current study
In this study, we sought to answer three broad questions about what teachers learned as they planned instruction, enacted curricular materials, and participated in professional development related to storyline units in middle school science: (1) Which curricular goals did teachers discern and become more confident in supporting over time?(2) Which aspects of the instructional model became easier for teachers to use in classrooms, and to what did they attribute the shifts?(3) Which aspects of the instructional model remained challenging for teachers to use?
We focus here on confidence as an indicator of growth, and we took knowledge of curriculum to be indicated when teachers referenced specific curricular purposes and structures as supporting their growth in implementing storyline units.While not a direct measure of their knowledge (i.e., a test), our structured interview protocol elicited teachers' knowledge-in-use to plan and enact instruction, which has been argued to be an essential focus for studies of teacher knowledge in science (Chan et al., 2019).
Our study focuses primarily on educators' own perspectives on their growth in knowledge.We highlight excerpts from interviews where educators attributed their growth to one or more sources-to professional learning, to supports in the materials, or to experience.Thus, our study design did not allow us to estimate the causal impact of each of these potential sources of educator growth.We recognize this is a limitation of the study; however, teachers' own desire for and perceptions of growth are important drivers of their efforts to improve their practice.
This research was approved by the human subjects review board at the University of Texas at Austin and all protocols were followed for obtaining active consent of human participants.

Participants
As part of our study design, we invited a subset of 36 field test teachers to participate in interviews, the analysis of which is the focus of this paper.They were selected for this study because they had been asked to reflect on what was getting easier over time about teaching with storylines, what remained challenging, and what supported their learning while teaching with storyline units for two or more years between August 2018 and June 2021.Table 1 below describes the characteristics of the teachers in the sample, showing a roughly even spread across grade levels taught, the majority self-designating as female (n = 26) and white (n = 27), with a few as Native Hawai'ian an or Pacific Islanders (n = 4).The sample included teachers with a range of experience in teaching science (M = 12.5 years teaching, Range: 0-29).
Importantly, none of the teachers had prior familiarity with the curriculum, as these materials were being developed and field tested for the first time by this set of teachers.Only five teachers reported using any kind of inquiry-based materials, and only one had used materials that resembled anything like those being field tested.
During the pandemic, many teachers continued to teach units in fall 2020 and spring 2021, but in different modalities.About one-quarter taught in person in fall 2020, another quarter taught in a remote instruction situation, and the remaining taught units in both situations.In spring 2021, roughly three in five teachers taught in person, and two in five taught units in remote learning situations.
Teachers provided us with data on their student demographics at the classroom level, and these appear below in Table 2.The plurality of students was white (41%), and over one third were Hispanic/Latine (33%).A little over one in ten students was in special education (11%), and 13% were gifted.About one in five (18%) were emergent multilingual learners.

Study context: field test for OpenSciEd middle school
As noted above, this study took place as part of a large-scale field test of an open education resource called OpenSciEd and involved roughly 340 teachers.Here, we describe in greater detail the key curricular purposes and structures of OpenSciEd units and their associated professional development workshops.

Curricular purposes of OpenSciEd units
Each OpenSciEd unit is intended to address a set of related performance expectations of the NGSS.Each unit is organized around an anchoring phenomenon or design challenge, with the broader purpose of supporting students in developing and using disciplinary core ideas, science and engineering practices, and crosscutting concepts to explain the phenomenon or solve the design challenge.In this respect, these materials are like other materials that aim to address new science standards.What makes OpenSciEd materials distinctive is that they follow a storyline approach, in which standards are addressed in a way that is driven by students' own interests and questions (Reiser & Novak, 2017).Storyline curricula are designed to be coherent from the student perspective, that is, the sequence of learning activities should "make sense" to a student developing toward proficiency with targeted performance expectations because students are taking up and addressing questions they have identified about phenomena or design challenges (Reiser, Novak et al., 2017).
To teach storyline units such as those in OpenSciEd requires an understanding of the value of coherence from the student perspective, as well as how lessons build upon one another.Professional development activities directly support the development of this understanding through two kinds of activities.First, in the introductory workshop, teachers explore what it means for a curriculum to be coherent from the student point of view through a classroom vignette and explore a contrasting vignette where the reasons for engaging in a particular activity are not apparent.Second, in each workshop where a new unit is introduced to teachers, teachers "re-build the storyline;" that is, they review the individual lessons of the unit to identify what students are expected to figure out and what new student questions are expected to be generated by those lessons.They then assemble these publicly and discuss how the lessons are expected to fit together from the student point of view.

Curricular structures of OpenSciEd units
Teaching routines.There are five different teaching routines that are used within OpenSciEd and introduced to teachers within professional development workshops (Table 3).Each routine addresses one of the key goals or purposes associated with supporting student sensemaking within storyline units.
Each of the routines in Table 2 serves a distinctive purpose in supporting students making progress toward developing understanding of the targeted three-dimensional learning standards in a way that is driven by student questions.The Anchoring Phenomenon Routine does so by guiding students toward questions that create a need to develop understanding of the science ideas targeted in the unit.The Navigation Routine and Problematizing Routine both help students to organize the ideas they are developing and identify gaps in knowledge that can be addressed by pursuing new questions that arise from their investigation.The Investigation Routine helps students develop a grasp of scientific and engineering practices, by engaging students in seeing just how using the practices together can help them build knowledge.The Putting Pieces Together routine helps students apply and generalize knowledge they have built through studying the anchoring phenomenon and investigative phenomena along the way.
Importantly, the routines are featured centrally in OpenSciEd professional development workshops.For example, each workshop begins by engaging teachers in the Anchoring Phenomenon Routine as if they were students experiencing the routine, in "student hat" (Lowell & McNeill, 2020).Typically, each professional development workshop also entails experiencing and reflecting upon a lesson early in the unit, where teachers can learn how to orchestrate the Navigation Routine and see the transition from that routine into the Investigation Routine.Teachers also review students' consensus models that are part of the Putting Pieces Together routine.
Supporting academically productive talk.Whole class discussions are a key curricular feature of OpenSciEd units, and they are supported by three different "discussion types" called out for teachers in the materials (Lowell et al., 2021).Some discussion types that appear in Table 4 below appear more commonly at the beginning of lessons and units (e.g., Initial Ideas), while others are more common toward the conclusion of lessons and units (e.g., Consensus), because their purposes are to help students organize and extend knowledge they have developed.As with the routines, each discussion type serves a distinctive purpose in supporting student sensemaking, as depicted in the table.Further, the curriculum materials include specific prompts teachers can use with lessons, so the discussions are fully integrated with content.
The OpenSciEd professional development also offers teachers opportunities to gain understanding and practice with these different types of discussion.In the introductory workshop, teachers are introduced to the idea of "talk moves as tools" and provided with a list of moves and their purposes.In addition, teachers have opportunities to participate in these different kinds of discussions while in "student hat."The curriculum provides a discussion planning tool, and in some workshops, teachers are given the chance to rehearse (Kazemi et al., 2016) a discussion with their colleagues.

Data sources and methods
The data come from interviews conducted with teachers after the last two semesters of unit enactment, which took place during the pandemic, in fall 2020 and spring 2021, and from surveys conducted with this same group of teachers at two time points, one early in their implementation of units, and one later.Data from interviews and surveys were collected by the second and fifth authors, as part of the larger program evaluation to inform revisions to the curriculum and to inform and build capacity for future implementation.Each participating state recruited teachers and was responsible for all decisions around compensation.
The choice and structure of compensation differed by state; however, teachers who received compensation did not receive anything extra for participating in an interview.
Interviews with teachers were 25-40 minutes in length, took place online and were recorded using Zoom, and the audio files were transcribed for analysis.The interviews were intentionally kept relatively short to make participation less burdensome for the teachers in the study.Thus, while some probing of answers occurred, the interviews were not designed to be in-depth.The semi-structured interview protocols included a standard set of questions about teachers' experience with each unit they enacted, how well-prepared they felt by the professional development workshops, and their perceptions of student engagement and sensemaking.After the first two years of the project, additional questions were included in the protocol, which became the focus for this paper: We asked teachers how they would describe their journey of learning to teach with these units, what was getting easier about teaching through storylines, and what remained challenging about this approach.Teachers' responses to these questions served as the primary dataset for this analysis.
The surveys we used in the analysis were administered digitally via Qualtrics between summer 2018 and winter 2020 when teachers first joined the field test and with each professional development workshop in which they participated.For 33 of the 36 teachers in the sample, we draw upon two survey timepoints to provide baseline and follow-up measures.For most teachers in our sample, the baseline survey is from summer 2018 (n = 24) and the follow-up survey from winter 2020 (n = 31).A few teachers responded to their first (n = 9) or last (n = 2) survey at a different time, and so we used those surveys as the baseline or follow-up, respectively.For our analyses, we focus on a subset of five survey items that asked teachers about their confidence with using the various instructional routines introduced during the professional development sessions.These items followed a five-point Likert-scale response format, with 1 representing "very unsure" and 5 representing "very confident."We used these data to allow for comparisons across teachers with respect to their sense of how well they were able to adapt teaching to their students in ways intended in the instructional units and to examine how teachers changed over time with respect to their confidence with particular routines.
Because some data analyzed in this paper were collected during the first year of the COVID-19 pandemic.We saw the impact of this major disruption most clearly in interviews with teachers, where some teachers were adapting to remotely teaching and others were in-person but with necessary but newly distracting public health precautions in place.Because of this, teachers' responses to questions about what was challenging in the shift to storyline teaching often began with talk of COVID-related constraints.We chose not to focus on these challenges in this paper in addressing our research questions; however, it was evident from teachers that the year was particularly challenging for both students and teachers.In addition, one pattern from student exit ticket data (not used in this study) stood out, namely a decrease in how much students said they contributed out loud to knowledge building activities in the classroom, a fact to which we return in the discussion that we conjecture could have influenced our findings.Surprisingly, this decrease did not depend on mode of instruction: it was true for students in both remote and face-to-face learning.

Approach to analysis
In developing our analysis, we privileged interview data in developing findings, using quantitative data from surveys to situate responses to interview questions about teachers' confidence and experiences of implementing materials within the broader sample.This approach is appropriate, given our interest in understanding the ways that knowledge of curriculum figured in teachers' growth and challenges in implementation.
Interview data were analyzed through a multi-step process in which the researchers reviewed the transcript excerpts and generated a set of inductively derived codes.This first set of descriptive codes sought to capture teachers' own ways of naming what they gained facility with over time and what remained as ongoing challenges.These codes included a wide range of aspects of teaching that teachers pointed to, including: pacing, launch of the unit, classroom discussion, the storyline approach itself, specific curricular tools such as the progress tracker, and the driving question board, and more.These inductive codes were then categorized based on knowledge of the curricular design (e.g., topical coding such as the Anchoring Phenomenon Routine and Navigation were categorized as routines of the units).In addition, references to components of the curriculum codes also included attention to pacing and timing, the storyline pedagogical model, and attention to student engagement.Thus, while the initial inductive coding focused broadly on teachers' framing of what they had learned and were learning in teaching science with storylines, curriculum resources (tools) and routines surfaced as key aspects in their reflections.Authors 2 and 3 coded the data corpus and then looked within coded data for variation or nuance related to two main areas: shifts in their overall goals for teaching science and distinct aspects of teaching with the storyline units.Each main theme was then interpreted considering the key forms of curriculum knowledge articulated within our conceptual framework.The survey analysis was conducted by the third and fourth authors and focused on the items about teacher confidence with OpenSciEd instructional routines and practices.For each teacher in the sample, an analyst identified the first and last set of responses available, to create a matched dataset of baseline and follow-up responses.The researcher then conducted descriptive statistics, significance testing (i.e., to test the significance of the change from baseline to follow-up using a paired t-test), and estimated effect sizes (Cohen's d) on each item of interest.
To organize the findings sections, we relied on themes developed from coding of interviews, rather than the survey data.We did so, because interviews provided us with a richer dataset for understanding teachers' developing knowledge of curriculum-in-use than did surveys.Where survey findings are relevant to the theme, we present patterns from teacher responses to support claims related to teacher change.We report effect sizes from statistical analyses in the text, as well as in a table in an Appendix to the paper.

Findings
In their interviews, teachers appropriated the goal of supporting coherence from the student perspective.They also became more confident in their ability to enact routines to help launch units, to use practices to figure out pieces of the science ideas, and to connect lessons to what has gone before and what is coming up in future lessons.In addition, teachers reported gaining more confidence in orchestrating academically productive talk using talk moves.At the same time, using pacing guidance and supporting students in doing their own sensemaking remained a challenge for teachers, indicating difficulty in maintaining integrity to the model while also addressing the requirement that they provide opportunities for students to develop understanding of all standards for their grade level or band.

Learning with respect to curricular purposes
Nearly three-quarters (n = 26; 72%) of teachers described shifts in their teaching that are linked to different goals for storyline curriculum units: supporting coherence from the student perspective, positioning students as knowledge builders, and building knowledge collectively.As one teacher put it, the curricular resources "helped me change from teaching [from] the PowerPoint . . .and 'here's the content that you need to know' . . .and then 'do this cookbook lab'" (23NS21).Another said that the associated professional development and storyline curriculum materials together "bring the cohesiveness I have always wanted and tried to make happen" but could not previously (04CS21).
Other teachers described the professional learning and teaching with the units as "empowering" (15NS21) for the students in their classrooms and helping them learn how to partner with their students in developing knowledge.Teachers talked about their students "really feel[ing] like scientists when they get to do the work and you place a value on their data" (10ER21), and that this approach "really does take me off of the stage, and 'I'm a learner with all of you guys,' and I'm the facilitator, so I'm facilitating this discussion.I don't have the answers, and so the power of sitting face to face in a circle is just so valuable" (06CS21).
Still others commented that they appreciated how storylines supported students working together to figure out science ideas by pursuing answers to questions together.As one teacher put it: I think the power of the storyline approach is that it's really giving the kids a shared experience.We're all having a shared experience in the beginning and we're all experiencing this phenomenon that none of us really can figure out.Everybody can share something they notice, everybody can share something they wonder.(07CS21) It is worth calling out that "noticing and wondering" is a regularly used curricular structure within OpenSciEd units, particularly when students are introduced to a new anchoring phenomenon or investigative phenomenon.
This teacher also appreciated what she saw as a focus on equity, which supported students' interest in the activities of the curriculum: I think really establishing that culture and the norms that every voice is valued, then every single student has to share a question, so we do that.I think that really works to reach all students.I think the storyline approach really helps the students to be able to see how one thing is building on the other.We're trying to figure out our questions . . .I think the kids really feel invested in what we're doing.(07CS21)

Learning with respect to curricular structures
Teachers became more confident in using some of the key routines of storyline curricula, but using discussion types remained a challenge for some, as did following the intended pacing of units, as elaborated below.

Growing facility with the anchoring phenomenon routine
Many teachers (n = 12, 33%) referenced the building of a Driving Question Board as part of the Anchoring Phenomenon Routine at the beginning of a unit as becoming easier for them to do over time.A Driving Question Board is a public representation of students' questions, generated after students have explored an anchoring phenomenon, and attempted to make sense of it with initial ideas and identifying related phenomena, which could facilitate sensemaking through analogizing.It is a tool that is used in other models of projectbased science (Weizman et al., 2008), and so it is familiar to some teachers who have engaged in project-based science in the past.Remarkably, the portion of teachers reporting that they felt "very confident" getting students to ask questions at the beginning of the unit to guide the lessons that follow, jumped from 30% at baseline to 65% after having taught with OpenSciEd.Of the 33 teachers in the study that completed surveys, 17 or 51% of teachers' confidence increased, while 14 stayed the same and 2 decreased in confidence.
Even teachers familiar with driving questions found that the Driving Question Board, in combination with the Anchoring Phenomenon Routine, helped organize the process of generating questions.As one teacher said: [Even] before OpenSciEd . . .my units always start with questions.The driving question board, though, I think is an amazing way to organize it.It just really made that process even more efficient and meaningful, so I really love that driving question board.
The opportunity to participate in the Anchoring Routine in "student hat" was one support that helped teachers appreciate the power of the routine for supporting student engagement.As one teacher described: [Before that PL workshop] I don't know if I'd ever really thought about how sound traveled . . .I'm saying, 'Wait,' just like as a student, I got so into it, I'm thinking, 'Wait, is that what makes this happen?What about this?' They're like, 'That's great.Put your question up.' I'm like, 'I will, and I put that question up, and I'm seriously, authentically engaged as a learner in a way that I don't feel most of the time . . .But this reached somewhere deep in me.Then I said, 'If I feel like that, how do I get my kids to feel like that?' (18NS21) Through engaging in this routine as a student, teachers were also gaining experience with the purposes and value of the routines.

Mixed experiences with revisiting student questions in the navigation routines
Many teachers also increased their confidence around helping students identify the next steps for an investigation, rather than telling students what to do-a goal directly supported by the Navigation Routine.At the outset, only 28% of teachers reported feeling "very comfortable" with this practice, but after working with OpenSciEd, that number jumped to 48% (with many more teachers reporting they felt "somewhat comfortable" with this task).From the beginning of the study to its end, 14 or 44% of teachers' confidence in successfully implementing this practice increased, while 10 stayed the same, and 8 teachers' confidence decreased.These suggest that while there was growth in confidence overall in the cohort, there was some variability among teachers.
One of the ways that teachers whose confidence increased was to regularly return to the Driving Question Board.In interviews, some teachers reported that this move got easier for them over time.One teacher said that initially revising and adding to students' questions every few lessons "was awkward for me at first and then it was very beneficial" (05CS21).Another teacher said that with experience, they got better at "remembering to go back to the Driving Question Board, really trying to ground things [in students' questions]," not just during the start of a unit, but in later lessons as part of building coherence with and for students (08Gen21).A third teacher commented that they see the value of "really taking everything back to that Driving Question Board a lot of the time through the unit-I think those routines are really ingrained and obvious [for me] now" (11NH21).
Consistent with the finding that several teachers' confidence in implementing this practice either stayed the same or decreased, two teachers said that returning to students' questions from the Driving Question Board remained a challenge for them.As one teacher said, I still have a hard time when I do the driving question board, like coming back to it and remembering to always come back to those pieces, all of the different posters and things like that, just constantly bringing those back in and utilizing them (13ER21).
For these teachers, the variety of artifacts created as part of the routines were harder to manage and keep in the foreground of their teaching.

Supporting engagement in using science practices through multiple routines
Teachers also increased their confidence around engaging students in the core practices in the Framework and NGSS.While 30% of teachers reported feeling "very confident" helping students to use science and engineering practices to figure out pieces of core ideas at baseline, that number jumped to 58% after teaching with the curriculum (with many more teachers reporting they felt "somewhat comfortable" with this task).This was a statistically significant increase over time (t (32) = 2.18, p = .037,d = .715;See, Table A1).Over the duration of the study, 14 (42%) teachers' confidence increased, 14 teachers' confidence stayed the same, and 5 reported less confidence in this aspect of teaching.This goal is explicitly supported by the investigation routine, a routine used to organize a large proportion of lessons in OpenSciEd units.
One of the specific core practices in the Framework and NGSS is developing and using models, and within storylines, multiple routines support the practice.Students develop initial models in the Anchoring Phenomenon Routine, develop models as part of the Investigation Routine, and revise and critique models in the Putting Pieces Together Routine.
Developing greater facility with helping students develop and revise models was something about one-fifth of teachers reported in interviews (n = 7, 19%).Similarly in surveys, the portion of teachers reporting that they felt "very confident" pushing students to go deeper to revise their explanatory models more than doubled, from 24% at baseline to 52% (t(32) = 2.81, p = .008,d = .742)after teaching with OpenSciEd (with many more teachers reporting that they felt "somewhat confident" with this practice).Teachers noticed that the fact that the practice was used across routines in a unit gave them a window into student learning, particularly students' growing grasp of the practice of developing and using models.As one teacher put it: It was really fascinating with the models to see how helpful it was.That shift [of] having them draw those models, then seeing how they progress -and it's always challenging for them to do their first models for a unit and then seeing how those progress.That's a really cool thing as a teacher, but also, I think as a student, they get to see their progress as well.(17NS21) Notably, students' struggles with models-both at the beginning of units and at the outset of the school year-made engaging students in the practice challenging.Part of what facilitated teachers' learning was not the routines per se, but students' own growing comfort with the practice of modeling.As one teacher put it: The modeling piece was really hard for me at the beginning.I'm starting to understand not just how to do it-the models themselves make sense-but teaching kids who have no experience with making models, how to build models, is really different.As they get better at it and I get better at it, the modeling is getting easier.(08Gen21) Other teachers who pointed to growth in their facility with engaging students in modeling commented similarly that using the practice gave a variety of students access to making sense of the science, and that revising models over time was a way for students to reflect on and document their own and each other's thinking.

Supporting academically productive talk through discussion
Many teachers also reported growing in their confidence around helping students to make sense of scientific ideas.Remarkably, the portion of teachers who reported feeling "very confident" about helping students to put pieces together of disciplinary core ideas and crosscutting concepts jumped from 15% to 58% as teachers had time to implement OpenSciEd (again, with many more teachers reporting that they felt "somewhat comfortable" with this work).This difference was statistically significant (32) = 2.33, p = .026,d = .895).Over the duration of the study, 15 (45%) of teachers reported increased confidence in this aspect of teaching with storylines, while 14 teachers' confidence stayed the same, and 3 teachers reported less confidence.This goal is explicitly supported within the Putting Pieces Together Routine.
One of the ways that teachers said that they helped students to make connections and grasp key ideas was through discussion.Twenty-two percent (n = 8) of teachers interviewed described themselves as getting better at facilitating discussions as part of the units.There were multiple reasons teachers gave for this, including getting students to contribute and participate in the science ideas and teachers' own familiarity with the "arc" of the storyline itself, rather than the specific discussion types.One teacher said, "I think setting up discussions is getting easier over time, and I think part of that is my comfort with figuring out where students are going, and pulling out the right questions, and just my experience there" (06Eco20).As with implementing other programs, repeated practice helped too-one teacher says what helps is "just experience.The second time you teach it, you're better.The third time you teach it, you're better."This teacher described taking detailed notes on lessons to help her remember to ask specific questions, and that with repeated practice and time, "I know that will just become more innate" (12ER21).
As with engaging students in science and engineering practices, teachers attributed their improvement with orchestrating discussions to students' growing familiarity with engaging in academically productive talk.Several teachers talked about how their students' contributions improved alongside their own question-asking and facilitation, as they engaged in the discussion types over time.Some explicitly referenced routines as a support for getting better at orchestrating discussions.For example, one teacher referenced a routine called the "scientist's circle," in which students come together to build understandings together through discussion: I've been able to narrow down my procedures for things like [the] scientist's circle.The first year was really, really crazy and my discussions weren't that good with my kids, and they weren't asking good questions, and so I really learned how to get them there to ask better questions and to have better discussions.That's definitely been a big change throughout the three years that I've taught them.(09ER21) Not all teachers became more confident in their ability to engage students in rich discussion.Some teachers continued to struggle with discussions throughout the field test period, and their struggles were a source of frustration, given the importance of discussion within the storyline curriculum model.As one teacher put it, the struggle pertained directly to the challenge of letting students struggle with the science ideas without simply providing them with answers: For me, it's the discussion piece [that is still difficult].Having really good meaningful discussions with my kids . . .they struggle with thinking scientifically and having those discussions.Getting them to have really meaningful discussions and not jumping in as a teacher.(09ER21) In addition, some teachers commented on how aspects of facilitating discussions such as "wait time" became even more difficult during the period of remote learning due to COVID-19.One said: I think for me, the challenge was the online platform.If were in the classroom, I think it would have been so much smoother.But wait time, it's always the wait time, waiting for the kids to let them percolate and let them come up with it . . . .It's just wait time, getting them to respond.(08ER21) A curriculum that relies heavily on student discussion and sense-making places high demands on teachers to facilitate academically productive talk, particularly in remote learning.

Struggling to follow pacing guidance
One of the challenges all teachers face is how to prepare students for all standards on which students will be assessed within the time frame of a school year.And while OpenSciEd units are designed with the intention that teachers can implement units for a given grade level within a typical district's number of days for academic instruction, several teachers (n = 7, 20%) struggled throughout the field test with pacing and length of units as prescribed within the materials.These teachers expressed concern about the length of lessons or units and felt the need to prioritize certain lessons or activities to make them fit within the time allotted by the school or district.One said, I think that sometimes I feel like I need to just speed up a little bit and back and maybe not give them as much time of like, 'Okay, you need to figure this out,' but really push them along because I found that sometimes [the unit] can drag on for a while.(20NS21) Other teachers worried about whether it would be feasible to teach the full set of units tested in the field trial in one school year: Next year, trying to get through all six units, yeah, I don't know how that's going to happen.But I think it's important to try to do that because it does really all connect.I guess the hard part is, what are the essential pieces and what pieces can I just [say], 'Okay, kids, we're not going to go deep into that because we're going to just keep moving?' (06CS21) Taken together, these sets of concerns related to pacing point to an ongoing challenge for teachers in navigating the tension between following students' ideas and fitting teaching the units in with the constraints of the school calendar.

Discussion
In this study, we explored how teachers' growing knowledge of purposes and structures of storyline units supported their enacting of OpenSciEd field test units.The goals and interlocking components of the storyline instructional model, we found, were both evident to teachers and served as resources to them in planning and enacting instruction.Notably, some components became more salient to them over time, while using activity structures and tools for supporting academically productive talk remained a challenge.
Our findings about what became easier for teachers and what remained hard for them are largely consistent with findings both from other studies of OpenSciEd teachers (e.g., Cherbow & McNeill, 2022;Lowell & McNeill, 2022) and past scholarship on teachers' challenges in enacting instruction where teachers partner with students in knowledge building and where academically productive talk is a necessary component of learning (e.g., Clarke et al., 2016).For many teachers, the proposition that students' ideas can serve as a basis for organizing a learning sequence is unfamiliar.Further, having build science knowledge together is challenging for students and teachers alike.In part, the practice is challenging because it requires teachers to make judgments on the fly about what and whose student ideas to work with as productive resources for sensemaking (Alzen et al., 2022;Stroupe, 2014).No doubt, the experience of the pandemic made this challenge even harder, as evidence from student exit tickets and some interviews indicate.In storylines, teachers must manage another tension, namely, how to avoid a situation where students have pseudoagency (Miller et al., 2018), that is, the appearance that they are guiding the direction of the unit when in fact the teacher alone is directing the trajectory of lessons.Teachers can manage this tension, but only if they periodically pursue student questions that had not been anticipated by storyline designers (Cherbow & McNeill, 2022).
For these teachers, a silver lining was that because they had opportunities to teach multiple storyline units, they were able to learn not just through materials and professional development workshops, but also through experience and reflecting on experience.For these teachers, key elements of the instructional model functioned as tools for making sense of their enactments.Many attributed their growth to their own and students' growing experience with storyline units.The fact that there were common routines, patterns of talk, and arcs of units-all elements of the instructional unit-is what enabled teachers to develop useful knowledge of these curricular structures.
In this respect, our study illustrates empirically what many teacher educators know to be the case and scholars (e.g., Kazemi & Hubbard, 2008) have proposed-that the development of teachers' knowledge unfolds over time and across multiple contexts.Much professional development research focuses on the impacts of programs intended to support a single cycle of curriculum enactment on teacher knowledge and practice (Fishman et al., 2013;Penuel et al., 2009).It is relatively rare for studies to examine multi-year, multi-cycle professional learning where there are opportunities for educators to bring back what they learned in practice to professional learning and have repeated opportunities to develop knowledge of curriculum.Those that do have shown some promise in changing practice and enhancing student outcomes, in part because longer exposure may "develop conceptual and practical tool commitment" among educators (Longhurst et al., 2017, p. 440).This study supports this conjecture in part, by illustrating the ways that repeated practice with units that follow a common underlying structure can support teachers in gaining confidence with making shifts to practice toward partnering with students in the direction of their own science learning.Importantly, professional learning experiences made explicit that structure and gave repeated opportunities to deepen their understandings of those structures over time, an approach that in the past has been shown to be important for improving teaching practice and student outcomes (Penuel et al., 2011).

Conclusion
In this paper, we explored how teachers' growing knowledge of curriculum supported them in enacting units that were designed to align with the Framework for K-12 Science Education's (National Research Council, 2012) practice-focused vision for science learning.We focused on a relatively understudied aspect of knowledge for teaching-curriculum knowledge-to better understand its role in supporting enactment.Our study provides some evidence, moreover, for our initial conjecture that such knowledge can help teachers enact materials that are different from those they have in the past.We cannot say for certain whether in their enactments teachers maintained integrity to the curricular goals and structures of OpenSciEd; however, we can say that their knowledge of curriculum was central in helping them make sense of their journey of learning to teach with storylines.
We see this study as an exploratory one into science teachers' curriculum knowledge and its role in planning, enacting, and reflecting on teaching and learning.Future research on teachers' knowledge of other instructional models, as well as studies that explore how teachers use their knowledge to compare different curricula when selecting them, will help to build our understanding of the role of curriculum knowledge in science.Such studies are likely to help us advance knowledge of the larger role that curriculum can play in future reform efforts and in supporting curriculum implementation in a wide range of contexts.
Despite its exploratory nature, the study does provide promising evidence about the value of sustained professional learning that highlights how instructional routines and discussion can support student sensemaking in science.That evidence points specifically to the utility of repeated cycles of professional learning followed by enactment of units, a type of professional development that is relatively uncommon in science.The field trial demonstrated that it is feasible to offer such professional development to teachers, and future studies should examine how many cycles are necessary to develop teacher knowledge of curriculum and to improve teaching and learning outcomes.

Table 1 .
Characteristics of teachers in the sample.

Table 2 .
Demographics of the student sample.

Table 4 .
Three discussion types of OpenSciEd.