Preservice Teachers’ Adaptations to Tensions Associated with the edTPA During Its Early Implementation in New York and Washington States

The edTPA is a teaching performance assessment (TPA) that the states of New York and Washington implemented as a licensure requirement in 2013. While TPAs are not new modes of assessment, New York and Washington are the first states to use the edTPA specifically as a compulsory, high-stakes policy lever in an effort to strengthen the quality and accountability of teachers and teacher educators. This study examines 24 New York and Washington teaching candidates’ experiences with the edTPA during its first year of consequential use for state certification. The data, drawn from qualitative interviews that were part of a larger mixed-methods study, reveal that preservice teachers had to mediate several tensions associated with the edTPA’s dual role as a formative assessment tool and a licensure mechanism. In this paper, we identify those tensions, describe candidates’ efforts to mediate them, and discuss the extent to which that Education Policy Analysis Archives Vol. 23 No.103 2 mediation process may or may not contribute to the improvement of teachers’ practices. Given the edTPA’s positioning in a policy context – specifically, the potential for the assessment’s locus of control, high stakes, and opaque rating process to distort the procedures it is intended to measure – the paper concludes with recommendations for teacher education programs aimed at capitalizing on the edTPA’s benefits and mitigating its unproductive tensions.

mediation process may or may not contribute to the improvement of teachers' practices. Given the edTPA's positioning in a policy context -specifically, the potential for the assessment's locus of control, high stakes, and opaque rating process to distort the procedures it is intended to measurethe paper concludes with recommendations for teacher education programs aimed at capitalizing on the edTPA's benefits and mitigating its unproductive tensions. Keywords: edTPA; preservice teacher education; teaching performance assessment; teacher certification; teacher accountability; educational reform.

Introduction: Defining a Problem Space Associated with High-Stakes Teacher
Performance Assessment 1 It is not particularly controversial to suggest that systematically documenting novice teachers' planning, instruction, and assessment practices can help cultivate meaningful learning among those teachers and pinpoint strengths and opportunities for improvement in teacher education programs (Darling-Hammond & Snyder, 2000;Peck & McDonald, 2013;Wei & Pecheone, 2010). Far more controversial is the suggestion that performance assessments of teaching should drive the design and evaluation of teacher education programs and determine whether or not new teachers are fit for state licensure (Cochran-Smith, Piazza, & Power, 2013;Rennert-Ariev, 2008). In 2013, New York and Washington became the first states to mandate by law that their candidates successfully complete the edTPA -an assessment designed by the Stanford Center for Learning, Assessment, and Equity (SCALE), and grounded in performance measures already used by the state of California and the National Board for Professional Teaching Standards -for initial certification. This movement toward using teacher performance assessments (TPAs) as high-stakes tests, which is destined to expand as more states adopt them in an effort to raise expectations for teaching and teacher education, introduces a number of tensions for those impacted by the assessment process and test results.
In this article, we explore how the edTPA's positioning as a state certification test generated particular tensions for preservice teachers during the assessment's first year of implementation in New York and Washington. These tensions, located in our data and identified as this paper's core category of analysis, fall within three general subcategories: (1) support tensions, or ambiguities about what modes of assistance are appropriate and allowable in light of the assessment's characterization as both formative and summative; (2) representation tensions, or uncertainties about how best to demonstrate complex, contextualized, continuous teaching practice via the edTPA's performance tasks; and (3) agency tensions, or difficulties in negotiating external factors that influenced candidates' teaching circumstances and practices. Via semi-structured interviews with edTPA-taking candidates in New York and Washington, we address two central questions. First, what did candidates disclose as tensions that factored into the process of completing their edTPA portfolios; and second, how did candidates mediate and mitigate those tensions?
We argue that, because becoming a teacher in New York and Washington is contingent upon successful completion of the edTPA, the process inevitably demands successful negotiation of the tensions generated by the exam's high-stakes status, some of which scantly link to effective teaching practice. Our results contribute to a critically important discussion on the implications of positioning TPAs as state-level accountability mechanisms, a trend that seems likely to grow. Specifically, we build on an emerging base of literature that explores how the policy context of highstakes TPAs impacts the learning experiences and teaching practices of the candidates whose performance they intend to measure (e.g., Margolis & Doring, 2013;Okhremtchouk, Newell, & Rosa, 2013). In our discussion of this study's results, we distinguish between productive and unproductive tensions associated with high-stakes TPAs and present several considerations for teacher educators who aim to reduce effects that undermine preservice teachers' learning and practice, particularly during early adoption of TPAs as accountability tools (Stillman, 2011).

Background: The Evolution and Consequences of High-Stakes Teacher
Performance Assessment

TPAs and Their Uses as Policy Levers
Teacher performance assessments are not new phenomena. Porter, Youngs, and Odden (2001) explain that the National Board for Professional Teaching Standards (NBPTS) began using a portfolio of practice for its certification process in the mid-1990s. Three central components have marked the NBPTS portfolio since its inception: video recordings of teaching practice, student work samples, and written commentaries designed to reflect the contextualized nature of those artifacts (Sato, Wei, & Darling-Hammond, 2008). In 1998, TPAs entered the state policy sphere when the California legislature passed a law requiring universities to use either the state-sponsored California Teacher Performance Assessment (CalTPA) or an approved alternative for preservice teacher credentialing. Because of concerns about CalTPA's generic design, several universities formed a consortium to create an assessment portfolio that adapted to different subject area pedagogies and integrated multiple artifacts through a cohesive "teaching event" (Pecheone & Chung, 2006, p. 34). That assessment -the Performance Assessment for California Teachers (PACT), which first was consequential in 2002 and remains in use today -was a precursor to and model for the edTPA. We would consider both the PACT in California and edTPA in New York and Washington to be highstakes, given their roles in determining teaching candidates' certification status.
Two core assumptions underlie TPAs' positioning as policy levers for strengthening the field of teaching. One is that valid and dynamic assessments can help to elevate instructional quality; and the other is that TPAs do a better job in that regard than other modes of assessment (Wilson, Hallam, Pecheone, & Moss, 2014). On the first point, Darling-Hammond (2010), Peck, Singer-Gabella, Sloan, andLin (2014), and Wei and Pecheone (2010) all suggest that TPAs can serve as tools to name and clarify standards of effectiveness and expertise in teaching, to scaffold learning for those being evaluated, and to target resources needed to strengthen specific professional competencies and practices. Chung (2008) and Bunch, Aguirre, and Tellez (2009) substantiated these suggestions via studies of the PACT and its impacts. They claim that the PACT presses candidates to acutely define and evaluate the various learning needs of their students and to use evidence from those evaluations to reflect on and improve their teaching. In another study of beginning teachers' experiences with PACT, Okhremtchouk et al. (2009) found that most of their participants perceived the processes of video recording and writing about their work to strengthen their understandings of learning and teaching.
There are several caveats to the aforementioned assumptions. First, while high-stakes assessments influence teachers' practices when they are used as policy levers, the extent to which they fundamentally change those practices is uncertain, since they interact with a number of individual and social factors that also steer what teachers do (Cohen, 2011;Grant, 2001). Second, the potential for a performance assessment to strengthen teaching rests largely on its intents and applications. That is, it matters to those who administer, complete, and interpret TPAs whether they are: (1) positioned as externally rated high-stakes tests, designed to hold universities and their candidates accountable for achievement; or (2) adopted by teacher education institutions as locally controlled, formative tools for supporting candidates' growth and identifying opportunities for program improvement (Cochran-Smith et al., 2013). Third, because teaching quality is defined divergently, establishing a valid assessment system that meets divergent criteria is replete with conceptual dilemmas and technical complexities (Berliner, 2005;Cohen, 2010;Sato, 2014). Sato (2014) explains that the edTPA's construct of teaching quality is research-based and consistent with proficiencies and practices that are delineated in the Interstate Teacher Assessment and Support Consortium (InTASC) Standards, yet questions about the assessment's face validity remain. These questions include whether the intrusions of video recording and associated permissions threaten the authenticity of candidates' teaching; whether extensive curricular controls or other attributes of stifling school climates disadvantage candidates in those situations; and whether a snapshot of three to five lessons, 20 minutes of video, and three student assessment results credibly captures the comprehensiveness and complexity of a beginning teacher's capabilities.

Tensions Associated with TPAs' Positioning as Policy Levers
Speculation on the edTPA's potential impact runs the gamut from its celebration as a barlike exam that could unite states and teachers around common professional visions (Mehta & Doctor, 2013) to concerns that it could have little effect on the practices of teachers and teacher educators aside from creating additional obstacles to joining the profession (Wei & Pecheone, 2010). Several studies demonstrate tensions associated with the discrepancy between high-stakes TPAs as tools for strengthening the field and as accounting mechanisms to be gamed in the process of entering the field. For the purposes of this paper, we define tensions as competing priorities or demands that strain or generate conflict among those called to address them. Lit and Lotan (2013) explored the process of integrating the PACT into teacher education programming, revealing a conflict between "the formative nature of the work of educators and the summative imperatives of a high-stakes assessment" (p. 59). The following point demonstrates one impact of this conflict on teacher educators and candidates: In regards to the [TPA], we request that our colleagues not provide professional judgment and feedback in most cases. . . . Simply put, a high-stakes assessment, at times, requires us to ignore or undermine "teachable moments" in the educational experience of our candidates. (Lit &Lotan, 2013, pp. 63-64) Peck, Gallucci, andSloan (2010) similarly note that focusing on policy compliance can unsettle the ways in which teacher education faculty support candidates and peers support each other in the TPA completion process. These kinds of support tensions, which lie at the intersection of high-stakes TPAs' formative and summative intents, are important for stakeholders to acknowledge and negotiate, given how pivotal effective, timely scaffolding and feedback are to candidates' perceptions of and efficacy with TPAs (Darling-Hammond & Snyder, 2000;Okhremtchouk et al., 2009;Pecheone & Chung, 2006).
Another source of conflict involves representing teaching as complex, responsive, and evolving, despite TPAs' prioritization of some competencies, practices, and perspectives over others. Sandholtz and Shea (2012) studied how university supervisors' predictions of candidates' PACT scores harmonized with actual test performance and found inconsistencies grounded in different representations of teaching and modes of assessment. Specifically, supervisors gave stronger consideration to affective qualities, like candidates' relationships with and fairness toward students, and based their evaluations more on observations of dynamic, ongoing classroom activity than on writing competency, a key criterion of PACT performance. Sato (2014) explains that such representation tensions played strongly into the National Association of Multicultural Education's (NAME's) rejection of edTPA as a licensure exam and the American Association of Colleges of Teacher Education's (AACTE's) response to that rejection. When NAME (2014) suggested that the edTPA potentially inhibits preservice teachers' curricular and instructional flexibility, AACTE repudiated the notion of standardization while simultaneously expressing support for the field's "efforts to identify the core practices we know impact student learning in PK-12 classrooms and a common set of metrics for evaluating that practice" (AACTE, 2014, para. 3).
A third source of concern about using TPAs as high-stakes policy levers are what we call agency tensions, or the extent to which preservice teachers have control over circumstances that affect their assessment performance. Okhremtchouk et al. (2009) found that constraints in some school placement sites, including cooperating teachers' lack of knowledge about the PACT or disagreement with its tenets, hindered candidates' completion efforts. Similarly, Margolis and Doring (2013), who conducted a study of the edTPA's pilot implementation at one Washington university in 2011, question the fairness of high-stakes assessments of practice when student teachers' degrees of control over their work conditions can vary drastically from one school setting to another. Further, they indicate that candidates may find few models in their placement schools of the kind of teaching endorsed by the edTPA, which presents tensions related to both agency and representation.

Design and Implementation of the edTPA in Context
The edTPA is a performance assessment of teaching that attends to three core elements of practice -planning, instruction, and assessing student learning -with strong emphases on: (1) drawing from students' prior knowledge and experience as instructional assets; (2) representing the subject matter in ways that meet diverse students' needs; (3) analyzing classroom interactions and student work; and (4) using the results of those analyses to inform ongoing practice. In an effort to capture the complexities thereof, the Stanford Center for Assessment, Learning, and Equity (SCALE) designed the edTPA to link written commentaries and artifacts associated with the aforementioned core elements. First, the planning task includes a series of three to five lesson plans and a commentary describing their aims and rationales. Second, the instruction task includes up to 20 minutes of raw classroom video and a commentary explaining the teacher's efforts to promote a positive learning environment and deepen student learning. Finally, the assessment task includes work samples from three focus students, with feedback from the teacher, and a commentary demonstrating how the teacher interpreted those assessment results in the interest of strengthening student learning and pedagogical decision making.
Presently, the edTPA fulfills multiple assessment roles across the United States, proliferating in state policy spheres via licensure requirements and on local levels within teacher education programs. For example, Wisconsin will be the next state to follow New York's and Washington's leads in adopting the edTPA as a mandatory, high-stakes, state-level certification test (AACTE, 2015). By contrast, the Tennessee State Board of Education authorized the edTPA as an alternative to the Praxis Series in 2013, upon the encouragement of several piloting institutions of higher education, including Vanderbilt University and the University of Memphis. Similarly, the California Commission on Teacher Credentialing granted state-approved teacher education programs permission to begin using the edTPA as an alternative to CalTPA and PACT in the 2015-2016 academic year. In more than a dozen other states, where no policy exists for consequential use of the edTPA, one or more institutions of higher education have adopted it as a program-level assessment, in accordance with accreditation standards and institutional improvement goals. Taking into consideration the edTPA's substantive depth and complexity, its myriad functions and modes of implementation across the United States, and disparate positions on its suitability as a lever for change in teaching and teacher education, it is unsurprising that interpretations of the assessment's value and significance vary considerably by the circumstances of its implementation.

Theoretical Framework: Preservice Teachers' Mediation of Policy and Practice
Within an Accountability Context Wei and Pecheone (2010) propose a useful framework for conceptualizing the roles of TPAs as large-scale assessments of teaching practice. Their framework encompasses three perspectives: (1) a design perspective, which concentrates on how particular records of practice, like lesson plans, student work samples, videos of teaching, and commentaries on candidates' pedagogical rationales, approximate authentic teaching; (2) a sociocultural perspective, which attends to how TPAs, as sources of evidence, can be used to affect change within the activity system of teacher education; and (3) a policy context perspective, which emphasizes how TPAs advance particular political ends and reward and punish certain stakeholders when used as accountability mechanisms. Wei and Pecheone explain that it may not be possible to disentangle these perspectives, a position with which we agree. In other words, while some researchers and teacher educators posit TPAs as valid measures of practice and meaningful opportunities for growth from a design perspective, their employment as accountability levers within a policy context can recast performance as a technical matter of deciphering rules and responding to prompts, maximizing scoring potential, and avoiding failure. This argument recalls Campbell's (1976) aphorism that the more far-reaching an evaluation tool is for decision making, the more subject it will be to corruption pressures, and the more likely it will be to distort the processes it is intended to monitor. We find this a meaningful way to frame the tensions that center this paper.
We also acknowledge that authority flows multidirectionally (though certainly unevenly) among stakeholders in policy situations. Specifically, while top-down controls impact the aims and practices of teachers within their purview, teachers can act upon those controls as engaged mediators (Cohen & Hill, 2001), affecting how they are interpreted and implemented (Lacey, 2012;Roth & Lee, 2007;Stillman, 2011). As states intensify their efforts to steer relationships among teachers and students via high-stakes testing and certification and evaluation policies, it is important to examine the ways in which teachers exercise agency within that milieu (Leonardatos & Zahedi, 2014). In the context of TPAs, for instance, Caughlan and Jiang (2014) position teachers as brokers who use artifacts of practice to justify the importance of pedagogical autonomy and flexibility to the process of meeting performance standards. Granted, this degree of agency does not rise to the level of modifying assessment guidelines to reduce their constraints on practice; and because the edTPA and its use as a high-stakes certification test are so new, no research yet examines the consequences for candidates of strategically interpreting test criteria to meet divergent pedagogical values, goals, and practices. Nonetheless, the assumption that preservice teachers exert some control over how they interact with the edTPA shapes our discussion of their responses to the assessment's tensions.

Method of Study Sampling
This article is based on a study of teaching candidates' experiences with and perspectives on edTPA during its first year of consequential use in New York and Washington. We used a sequential mixed-method approach (Creswell, 2014), initiated through surveys of 104 candidates across nine teacher education institutions in both states. From a smaller pool of survey respondents who indicated a willingness to discuss their experiences in depth, we selected and interviewed 24 candidates, representing six teacher education institutions in New York and one in Washington. The two institutions that were not represented in our interview sample produced only two survey respondents apiece. In both cases, participants either chose not to participate in the interviews or did not respond to our invitations to schedule them. We did not engage in member-checking during this study, though we regularly asked respondents for points of clarification on ambiguous remarks within the semi-structured interviews. The logistical challenges of affirming our results with the 24 respondents outweighed the potential value of assessing the adequacy of our data. Further, because our intent was to capture participants' perceptions of and experiences with edTPA in the midst of their work on it, the passage of time between data collection and any member-checking we might have done likely would have impacted their interpretations of the evidence we collected (Angen, 2000;Sandelowski, 1993).
We administered the 40-item survey from June through October 2014, focusing on edTPA completers who recently had graduated from or were close to finishing their teacher education programs. To do this, we established contacts with faculty or staff representatives at each participating university. Those representatives, in turn, forwarded the study information, human subject protocol language, and survey link directly to eligible candidates. In an effort to diversify our participant pool, we took a purposive sampling approach, choosing institutions based on four criteria: (1) the sizes of their teacher education programs, both large and small; (2) the degree status of their candidates, both undergraduate-and graduate-level; (3) their funding status, both public and private; and (4) their geographic locations. The three private institutions were located in mid-sized cities, with school districts that have high free and reduced meals (FARMs) rates and serve predominantly students of color. The six public colleges and universities were situated in a range of demographic contexts, with two in large cities and the rest in small to mid-sized cities and rural environments. One public and one private institution housed graduate schools of education; the other seven admitted undergraduates to their teacher certification programs.
At the end of the survey, we asked candidates if they would be willing to participate in an indepth interview about their experiences with the edTPA, and 71 responded affirmatively. Approximately one quarter of those respondents later declined to take part in the interview or did not return e-mails soliciting their participation. We began semi-structured interviews in June 2014 and concluded them in early September 2014, when we felt we had achieved a diversified and representative sample of sufficient size.
Participants' subject areas and grade levels served as key sampling criteria, as we sought to gather data from a range of candidates across the K-12 spectrum. Of the 24 interview participants, 19 were from New York State and five were from Washington. Eight pursued certification in elementary grades (K-6); 12 pursued certification in middle-level and secondary grades (5-12), with field specializations in science (3), English language arts (3), social studies (2), mathematics (2), world languages (1), and agriculture (1); and four pursued K-12 certifications in music education (2), teaching English to speakers of other languages (ESOL) (1), and general special education (1). Several candidates sought additional licenses or endorsements in the areas of special education, literacy, and ESOL, though they completed their edTPA portfolios in their primary certification areas.

Data Collection
For this paper, we draw exclusively from our study's interview data, with survey results synthesized and discussed in a separate report (Meuwissen, Choppin, Shang-Butler, & Cloonan, 2015). We grounded our interview protocol in extant literature on teacher performance assessments, and in accordance with our sequential mixed-method approach, we refined the protocol using results from the survey. For instance, we noticed that a sizeable number of survey respondents perceived that the edTPA was not a fair assessment of their teaching practices. Consequently, we framed interview questions to elicit a definition of fair assessment and an explanation of what factors they would consider when determining whether an assessment of teaching practice is fair or unfair. Similarly, many candidates reported in the survey that they had not been well supported in the process of selecting artifacts to include in their portfolios. Consequently, we asked interview participants to explain what it means to be well supported and to specifically indicate what kinds of supports they received, from whom and in what ways they received them, and to what ends those supports were helpful or not.
The interviews focused on four themes: (1) teaching candidates' knowledge of the edTPA's purposes, contents, and contexts; (2) candidates' perceptions of the edTPA's positioning in their teacher education programs; (3) candidates' viewpoints about the assessment's fairness, credibility, and process of completion; and (4) candidates' experiences constructing their edTPA portfolios and submitting them to their state education departments. Alongside 19 open-ended questions and accompanying probes, which focused on how candidates' teacher education and student teaching experiences aligned with the edTPA, how they fulfilled the edTPA's requirements, and what they learned throughout the process, we asked respondents to read aloud and interpret what they considered to be particularly noteworthy segments of their edTPA commentaries. These exegeses and their follow-up questions revealed candidates' beliefs about the edTPA's effectiveness as an approximation of teaching, their interpretations of the rubric prompts and criteria, and their assumptions about how raters evaluate edTPA portfolios.
The interviews lasted from 35 to 69 minutes, with a mean of 46 minutes. We conducted 21 interviews by phone and three in person. In each case, we used a digital audio recorder to capture the conversation between researcher and participant. The researchers transcribed the first four audio files -with a professional transcriptionist completing the rest -to develop an early understanding of emerging themes and to fine-tune the protocol for the remaining interviews. For example, one question in the initial protocol asked candidates to choose segments from each of the three commentaries -planning, instruction, and assessment -to read aloud as exemplary representations of their teaching and explain why they perceived those segments to be particularly strong representations. We amended this question to allow candidates to choose one compelling passage from any commentary, as three proved time consuming and challenging for some respondents to produce.

Data Analysis
We used quantitative tools to analyze the survey results and a qualitative approach to interpret our interview data, regularly checking across sources for concurrent themes (Creswell, 2014). Once all of the interviews were transcribed, we divided them into stanzas, which typically consisted of question-answer exchanges and relevant follow-up questions, so that any codes applied to the stanza captured the full exchange (Saldaña, 2013). Our initial codebook contained anticipated themes based on the content and categorization of questions in the semi-structured interview protocol. Via inductive analysis, we added new codes to the book and recoded existing data as necessary, in response to the emergence of new themes. For example, the code edTPA_Student_Teaching_Tensions emerged over several transcripts to describe the ways in which fulfilling edTPA requirements generated conflicts or pressures within the student teaching experience. The authors and two research assistants coded transcripts by hand, using spreadsheets to organize codes as present or absent in each stanza. All members of the research team coded the first four transcripts and reconciled disagreements collectively. After that, we coded the remaining transcripts in pairs.
We maintained inter-rater agreement scores using Fleiss's Kappa, which takes into consideration random agreement that might occur between coders (Landis & Koch, 1977). The Fleiss Kappa statistic for paired coding ranged from 0.45 to 0.83, with scores in the 0.4 to 0.6 range indicating moderate agreement and scores higher than 0.6 indicating substantial to near-perfect agreement. Again, paired coders resolved discrepancies within each transcript by consensus discussion. Both of these analytical processes -first, establishing an agreement and reconciliation protocol as a whole group, and then using that protocol throughout pairwise coding -strengthened the study's interpretive validity, as the range of Kappa scores demonstrates (Onwuegbuzie & Teddlie, 2003).
Finally, we used NVivo to manage the coded data and create sets of reports for each code, from which the tensions theme that anchors this paper emerged. From there, we generated explanations of specific tensions and candidates' ways of mediating them, tracing them back to the original interview data to verify those explanations and double-check for credible alternatives (Miles & Huberman, 1994). We used two criteria to select quotations from the data to include in our report. First, were they consequential to identifying, explicating, and substantiating specific claims related to the three tensions at hand; and second, did they reflect patterns of response among multiple participants? Via those criteria, the first author and two research assistants independently passed through the data once again to excerpt coherent stanzas of appropriate length, and then triangulated their selections to single out specific quotations for inclusion.

Findings
The high-stakes, summative nature of the edTPA generated several tensions for candidates seeking state certification in New York and Washington. It is expected that a fundamentally new performance assessment program will prompt some turbulence and adjustment among stakeholders, particularly at the beginning of its implementation. The results of our study demonstrate that many of these tensions, and the process of adapting to them, had more to do with managing the technical and administrative demands of the assessment than authenticating or improving teaching practice. Herein, we define three kinds of tension found in our data -support tensions, representation tensions, and agency tensions -and explain how edTPA-completing candidates mediated those tensions in their efforts to successfully complete the assessment for certification.

Candidates' Framing of Support Tensions
We define support tensions as ambiguities about how teacher educators and cooperating teachers can and should help candidates, and how candidates can and should help each other, throughout the edTPA completion process. The strongest exhibition of this tension was uncertainty about precisely what kinds and degrees of support are allowed by rule, as the following juxtaposition of interview data and policy statement demonstrates: [Our program faculty] told us to read the [edTPA] handbook, just read the handbook. You can only read the handbook so many times… And you can't really ask your supervisor for help, so if you need clarification on a task, you don't really have that support because they're not supposed to help you. It's a test for us. (Elementary teaching candidate, June 15, 2014) Myth: Faculty cannot assist candidates to prepare for edTPA. Fact: The actual policy is just the opposite; faculty are encouraged and expected to provide formative support to candidates… Faculty can provide candidates with support documents, handbooks, samples of previously completed edTPA materials, and lesson planning templates that help them understand rubrics and other materials (SCALE, 2014a, p. 3). The candidate's explanation that edTPA-takers at his institution largely were left to interpret the test's criteria and build their portfolios independently seems oppositional to the formative potential of performance assessment and the social circumstances of learning to teach. Yet within the context of edTPA's positioning as a summative, high-stakes test, performance assessment inevitably must take the form of individual demonstrations within a rigorous and consistent assessment system, rather than a collaborative, instructive, dialogic process. Or must it, according to SCALE's "myth versus fact" contention, which appeared in a document designed to dispel rumors and doubts about the assessment?
To confuse matters further, there was a mid-year policy shift on this issue. At first, it was acceptable to have "discussions with candidates aimed at improving teaching competence aligned with program values and edTPA rubric constructs" but unacceptable for teacher educators to offer "leading comments, about the clinical observations, aimed at helping a candidate pass edTPA" (SCALE, 2013, p. 2). Further, teacher educators initially could "[explain] rubric constructs or rubric language [but not use edTPA rubrics] to provide formal feedback . . . on drafts" or mock-ups of edTPA tasks (SCALE, 2013, p. 3). A set of revised guidelines, published in April 2014, altered the language and standards of support significantly. Thereafter, it would be acceptable to: (1) "[engage] candidates in formative experiences aligned with edTPA (e.g., assignments analyzing their instruction, developing curriculum units, or assessing student work);" (2) [explain] scoring rubrics, and [use] these rubrics in formative exercises or assignments;" and (3) "[ask] probing questions about candidates' draft edTPA responses or video recordings, without providing direct edits" (SCALE, 2014b, p. 2) of candidates' work products. All that remained on the list of unacceptable supports were editing a candidate's materials, suggesting specific responses to prompts or artifacts to submit, and posting candidates' edTPA products to publicly accessible websites.
Candidates also described support tensions at the school placement level, with two primary themes emerging. First, cooperating teachers had little knowledge of edTPA's purposes, processes, and consequences, which meant that largely, they also were unfamiliar with specific methods of supporting candidates throughout edTPA completion. Second, different interpretations of the edTPA across candidates' support networks, coupled with the aforementioned confusion about permissible assistance, sometimes gave rise to inconsistent support. Inevitably, preservice teachers had to mediate those inconsistencies amidst their internship placements. Participants described their challenges as follows: When Regarding the first comment, while most cooperating teachers tried to facilitate candidates' work on the edTPA, they knew little about the assessment and lacked systematic, proactive support strategies. Only five candidates indicated that their mentors had heard of the assessment before their placements, and about half of those who explicitly discussed their cooperating teachers' support efforts characterized those efforts as irresolute (e.g., "She wasn't opposed to what I had to do") or ad-libbed (e.g., "My teacher said, 'I don't really know what you're doing, but you can just tell me what you need, and I'll make it happen'").
The second comment demonstrates that the edTPA factored decisively into how university personnel and cooperating teachers negotiated support for candidates. In fact, three candidates indicated that their university-and school-based mentors disagreed, at least somewhat, about how they should approach the assessment. One explained that her cooperating teacher interpreted the edTPA to closely resemble the statewide teacher evaluation model and, thus, attempted to steer the candidate toward demonstrating and writing about outcomes associated with that model. However, the candidate's supervisor suggested instead that she align her edTPA portfolio with universitysupported practices and assessment products, which varied from the state evaluation model and allegedly were more representative of what the edTPA demanded.

Candidates' Mediation of Support Tensions
The primary mode by which candidates mediated these support tensions was social networking. Almost entirely, candidates' networks consisted of preservice teachers at their universities, all simultaneously working on edTPA portfolios. In some cases, these networks formed within content areas; in others, they formed across them, either because there were few colleagues in the same certification area with whom to collaborate, or because candidates met each other through program-wide edTPA workshops and decided to continue their collaborations beyond them. Via their networks, candidates engaged in various edTPA-preparatory activities, including: (1) parsing and discussing their interpretations of handbook language; (2) searching for edTPA resources online and sharing those perceived to be helpful; (3) trading and critiquing drafts of edTPA commentaries; and (4) discussing video clip production strategies.
Two notable themes emerged via participants' descriptions of these networking activities. First, candidates sometimes described them surreptitiously, as the following examples demonstrate: We were all very into this mindset that we couldn't talk to anyone [about our edTPA products]. So we started a secret Facebook group where we would post things on the wall about how to do certain things. . . . That was my primary source of help the semester I completed the edTPA. (ELA teaching candidate; August 25, 2014) We had a student-run edTPA help group that was secret and private; it was everybody at my school who was working on edTPA. People would post things like, 'What do they mean by contextual understanding?' and then everyone would kind of give their two cents. . . . We agreed to proofread for each other. (Elementary teaching candidate, June 18, 2014) Comments like these suggest that candidates perceived at least some of their social networking to be beyond the scope of what was permissible, according to the edTPA's support restrictions. Second, while most respondents who discussed these networks characterized them as instrumental, several also suggested that they were limited by their members' lack of experience or expertise. One, for instance, described an edTPA-focused social media group as "25 students, all my age, who were like the blind leading the blind," while another explained that participants in her network sometimes were frustrated by disagreements about handbook language and rubric expectations: "Nobody agreed, really. . . . Everybody gave their opinions and what they thought, but nobody really had answers." On the matter of seeking resources online, one candidate explained that she and her colleagues looked for examples of video clips that might be considered appropriate for the edTPA and congruent with its criteria. She recounted an incident in which they located samples of teaching practice on YouTube, then strategized their own video recordings using those samples as technical models: The [edTPA videos] were just up there on YouTube. I think some of them were even labeled with edTPA; which, now that I think about it, seems very incongruent with their confidentiality policy. But there they were! (ESOL teaching candidate, June 11, 2014). Another respondent explained that a dearth of model edTPA artifacts led him to upload his own materials, minus the video recordings ("because of confidentiality issues"), to a website of his own design, "so that my colleagues and other people down the line can get a sense of what a completed edTPA looks like." He added that, because he had passed the assessment comfortably, his portfolio should serve as an exemplar for current and future candidates in his program. Further, he uploaded a 58-slide presentation to that website, walking edTPA-takers through some "quick tips for success," as well as detailed interpretations of and decisions associated with each of the assessment's prompts.
Fewer than half of the respondents discussed specific strategies for mediating support tensions at the school placement level. Among those who did, the predominant approach was to overtly position edTPA as an authority mechanism that singularly determined whether or not they would be certified to teach. For example, one candidate explained her enactment of a teaching practice that was uncommon in her placement classroom as follows: "My cooperating teacher didn't want me to do it. But I said, 'Well, here's what they need me to do [for edTPA],' and he said, 'Okay.'" Other candidates were more diplomatic, emphasizing how stressful the edTPA process was and appealing to cooperating teachers' empathic regard for their circumstances. Said one, "I got free rein to do whatever I needed to do. . . . My cooperating teachers were really great at listening to me about some of the struggles I would have." Regarding cooperating teachers' unfamiliarity with the edTPA and lack of support strategies, candidates assumed a significant share of day-to-day responsibility for explaining the assessment's goals and procedures and suggesting ways in which their school-based mentors could help them. Suggestions included modifying course schedules to accommodate particular modes of instruction and assessment at opportune times and assisting with video recording. In at least three cases, however, cooperating teachers' assistance with the recording process generated additional challenges, as one participant describes here: I told [my cooperating teacher] he really needed to focus on me and what I was doing with the students, and how I was interacting with them. And he would just . . . videotape the entire classroom and look on different students even if I wasn't working with them. So there was definitely a disconnect in what we filmed and what things I needed to do to address the edTPA. (Social studies teaching candidate, June 20, 2014) While this candidate attempted to convey how important it was for her instruction to be filmed, her cooperating teacher seemed to think that it also would be beneficial to capture how her students worked independently, as a testament to her planning and classroom climate. She addressed this dilemma by including the cooperating teacher's recording in her edTPA portfolio, and then fashioning the instruction commentary to link particular instructional rationales and techniques to learning effects that were evident in the video clips.

Candidates' Framing of Representation Tensions
Another kind of tension related to representing and demonstrating teaching practice via the edTPA, given the high-stakes nature of the exam and the complexity and contextualized nature of teaching and learning. These representation tensions were evident in the ways candidates appraised and described the relationship among the edTPA, their practices, and teaching in general, with two key positions emerging. On one hand, teaching is a continuous process built over time on complex human interactions, some of which are as much about understanding and strengthening positive interpersonal relationships with learners as they are about powerful subject-matter instruction. The edTPA's representation of teaching by way of lesson plans, short video clips, and three student work samples excludes many of these important interactions. On the other hand, the edTPA's core elements of planning to meet diverse students' needs, teaching for critical engagement with the subject matter, and assessing to generate student feedback and strengthen instructional decision making are central to the practice of teaching. And although the edTPA portfolio cannot fully represent the intricacies of these practices over time, the assessment is a means to sample and demonstrate them in a bounded way.
Most of the respondents evinced some fusion of both positions. Specifically, 16 of 24 candidates indicated that the competencies demanded by the edTPA credibly represent what it means to teach. Yet a majority of those candidates also expressed concern over whether or not evaluators would consider their particular samples of practice to exhibit the right competencies in the right magnitudes. Some tried to pack all of the instructional proficiencies into their two brief video clips, while others resigned themselves to the presumed impossibility of doing so. Further, several suggested that successful enactment of the assessment's core elements corresponds with a foundation of interactions and relationships that cannot be depicted prominently in the edTPA: edTPA put me on the spot to explain exactly why I did this and that, and here's why I would do it again or not . . . but it didn't highlight the relationships I developed. . . . My program focused a lot on nurturing the child as a whole, and the edTPA didn't focus on that part of teaching. Candidates also spoke of the pressure to portray teaching practice in an orderly fashion. To paraphrase several respondents, although teachers may strive to control classroom interactions and learning outcomes via planning, instruction, and assessment, inevitably teaching -and especially student teaching -involves unanticipated circumstances and divergent responses to them. In accordance with its high-stakes status, edTPA's representation of teaching as orderly and convergent discouraged some candidates from including practical challenges in their portfolios and made them anxious that isolated incidents could derail their edTPA portfolios. For instance, one music teaching candidate explained that a student's prolonged absence from small-group lessons prompted her to feign continuity across the edTPA unit in a way that did not manifest in practice: The first day I was all set to do edTPA; I had practiced with three kids, and I was ready to do the video, and one wasn't there. The second lesson, he was there, but after that, I had one more lesson to go and he wasn't there again. So we held off and I taught them something totally different. And I said, 'Okay, we'll just do this lesson next week.' . . . So now these kids are two weeks behind… but it was staged in a way where we were like, 'Here's your next lesson!' (Music teaching candidate, June 18, 2014) In another case, an elementary teaching candidate explained that, during a lesson recorded for her edTPA portfolio, she spent several minutes working individually with a student who was "lying on the rug and misbehaving" while her cooperating teacher assumed instructional responsibilities. When asked how she represented that incident in her edTPA commentaries, she noted, "Um, I didn't. I don't know how you would capture that, you know, based on what's valued from reading the rubrics." Finally, three candidates observed that the assessment's prompts and rubrics convey slightly different representations of teaching, with the latter including more specific, more delimiting language than the former. For example, one candidate called the rubric language "so clear and precise" while explaining simultaneously that the prompts contained "a few just confusing pieces of language that might be clarified." Another saw a sharper distinction between the prompts and the rubric criteria, claiming that she generally ignored the less precise questions in favor of specific tasks delineated by the rubrics. It should be noted that none of the candidates who identified this tension within the edTPA's language had seen examples of completed edTPA products that might have assisted them in connecting the assessment's evaluative vocabulary with demonstrations of practice.

Candidates' Mediation of Representation Tensions
Participants indicated that they mediated representation tensions in two key ways. First, they sought to demonstrate a consistent and continuous trajectory of practice across their edTPA portfolios, even when their actual teaching and classroom interactions were more tumultuous. Second, they followed the edTPA handbook directions closely and incorporated phrases from the rubrics into their commentaries in a persuasive effort to demonstrate their competencies, even when they felt uncertain about them. Thus, we would argue that the effectiveness of candidates' efforts to address representation tensions seemed to hinge on one especially important proficiency: strategic writing. This is because the places in which candidates could most effectively articulate an orderly trajectory across the portfolio and interpret their practices via the edTPA rubrics were the planning, instruction, and assessment commentaries.
One writing strategy that candidates employed was to set the edTPA rubrics beside their commentaries and artifacts and analyze each of them in checklist-like fashion, to ensure that their products overtly represented each of the required competencies. They did this work both individually and collegially, sometimes within university seminars or the informal social networks we described earlier. As one reported: Something I found really helpful when I was writing was going rubric by rubric together; we looked at the differences between a [score of] one and a [score of] five on each rubric, and then we wrote down key ideas that would help us [meet the higher-level criteria]. (Special education candidate, October 3, 2014) Another approach involved using the rubric language strategically to define candidates' practices as effective and guide evaluators in their readings of the edTPA portfolios. One respondent explained this process as follows: We were told to use the language from the [prompts and rubrics] in our statements. . . . So I used a lot of key words, which I thought would bring the grader to look at my work and say, 'Okay, here, she's using the language'." . . . Our practicum teacher advised us to do that because she's a grader [of edTPA portfolios], so I know that's one of the things that pops off the page for them. So even though I learned a lot from this process, it felt more like a game, like me saying, 'Hey I'm just going to put this language in there . . . so you can find it and move on instead of looking and really trying hard.' (ELA teaching candidate, September 23, 2014) Finally, six participants acknowledged that they created commentary content, student assessment feedback, or a combination thereof for the primary purpose of addressing the edTPA rubric criteria, even though those artifacts did not align well with their typical classroom practices.
Some candidates explained that, despite these efforts, they struggled to mitigate the representation tensions associated with completing the edTPA. When describing their difficulties, they used phrases like "if my portfolio is graded according to the rubrics" and "I don't know how the raters interpreted my assessment evidence," which demonstrate the obscurity of the evaluation process and imply a perception that raters also have to contend with the same representation tensions that they do. Put differently, those who crafted their edTPA products in anticipation of how raters might evaluate them also engaged in some speculation on how the rating process works. For example, one candidate presumed that evaluators triangulate all of the edTPA artifacts to develop a robust portrait of teaching, while another supposed that they quickly search the portfolio for key words and phrases, spending as little time as possible with the lesson plans, videos, and student work samples in order to move the process along and meet quotas.

Candidates' Framing and Mediation of Agency Tensions
The third kind of tension we found in candidates' conversations about the edTPA related to controlling and acting upon external factors that impacted their teaching circumstances. These tensions derived from an assumption, built into the edTPA, that the assessment's conceptual framework and core teaching tasks are embedded in placement school and classroom cultures when, in many cases, candidates perceived a mismatch in that regard. Consequently, their efforts to complete edTPA tasks according to the rubric criteria sometimes were constrained by resistance among stakeholders at their placement sites and challenges associated with shifting classroom practices to meet the demands of the assessment.
For example, one math teaching candidate explained that she would be more apt to construct an inquiry-based unit, with tools and practices closely aligned to the edTPA, "if I had my own classroom. . . . But the approach my CTs used was very standardized, and I got the message that 'we've been doing this for years and this is what we do.'" She went on to explain that the practices reinforced in her teacher education program harmonized well with the constructs of math learning and teaching embedded in the edTPA, but not with those exhibited in her placement school. Consequently, agency tensions manifested as a struggle for power to enact alternative modes of teaching in her internship placement -modes that were high-stakes for the candidate, as she needed to demonstrate proficiency with them in order to become state certified. Another example of that power struggle is evident in the comment below: I would say that my teacher kind of . . . just gave me a topic and said okay do this. It was hard, because it took me a while to sit down and plan, since I was like, 'Well, how am I going to do what the teacher needs me to do, and do what the edTPA needs me to do?' (Special education candidate, October 3, 2014) A third candidate drew from similar experiences to argue that the edTPA is an unfair initial certification test, and that it might instead be used to maintain ongoing certification, once teachers have more autonomy over planning, instruction, and classroom community-building.
A second kind of agency tension centered on school-institutional rules and protocols associated with collecting student data for the edTPA. Although this was a minor issue overall, it was a source of confusion and stress among several candidates whose placement school mentors and leaders were unfamiliar with the exam and had no plan for integrating it into school operations. One candidate explained her experience as follows: When I tried to do [my edTPA during my first placement], they delayed saying whether I could videotape or not. So I think on my last day, the principal came in and asked me if I was going to videotape; and I said, 'Umm, it's my last day,' and they were like, 'Okay, good.' So I never got a clear 'no,' but they were trying to discourage me from videotaping. (Science teaching candidate, July 11, 2014) Two others participants shared concerns that they might accidentally include students in their portfolios whose guardians had not granted permission for their children to appear in edTPA video clips. This had implications for classroom management, one noted, because "younger kids don't sit still," and thus it would be difficult for her to keep them out of the camera frame, despite her best efforts.
To conclude our findings, we report on candidates' efforts to mediate these agency tensions; but we do so briefly because participants delineated no particular strategies beyond those already discussed in previous sections of this paper. In situations in which candidates sought to implement instruction for the edTPA that digressed from cooperating teachers' interests or common school practices, they appealed to the test's authority as a certification requirement, solicited the arbitration of university supervisors and faculty mentors, or just endured the complications associated with meeting divergent expectations. Further, they used the edTPA commentaries to explain the contextual circumstances that led to their planning, instruction, and assessment decisions, hoping that test evaluators would take those circumstances into consideration during the rating process. For the most part, however, candidates who struggled with agency tensions simply lamented their impacts on student teaching, as

Discussion
According to state records, 81% of New York's edTPA-taking candidates passed the assessment during its first year of consequential use, as did 98% of Washington's candidates. 2 One interpretation of these results is that the edTPA's integration with teacher education programming and state certification policy has been relatively effective in its early stages, and that teaching candidates and teacher education programs generally "were ready for the edTPA . . . despite inevitable confusion surrounding the implementation of a new exam" (Robinson & LaCelle-Peterson, 2014). While this may be a reasonable conclusion, our research suggests an alternative focal point: test-readiness and implementation must be situated within a policy context in which the edTPA currently is used in New York and Washington as a high-stakes gatekeeping mechanism to determine whether or not candidates will be licensed to teach.
Returning to our theoretical framework, our findings demonstrate that this policy situationone in which the stakes associated with the edTPA assign significance to particular pedagogical practices and outcomes -impacts the social circumstances of learning to teach and candidates' efforts to represent their teaching. Wei and Pecheone (2010) suggest that, in such a context, it could be challenging for teacher education programs to advance veritably broad views of teaching that accommodate the agency needed to engage in contextualized problem solving and classroom change, as well as the kinds of authentic social relationships among novice teachers and mentors that support those change efforts.
In light of this challenge, Stillman (2011) argues that teacher learning in an accountability context ought to involve critical and systematic analysis of the tensions associated with dissonant state-, district-, and classroom-level pressures that play into teachers' work. While most of our participants identified and mediated some fusion of tensions associated with edTPA completion, we have little evidence to suggest that they investigated them systematically, in collaboration with university-and school-based mentors and peers, through a lens of critiquing accountability mechanisms' roles in and effects on teachers' professional lives. Rather, candidates' interactions with the edTPA focused on representing their teaching and the process of learning to teach as cleanly as possible, in the interest of passing the exam and achieving state certification.
Although all 24 of our study participants accomplished that goal, their pathways to success were marked with support tensions, representation tensions, and agency tensions that required mediation alongside efforts to strengthen their practice in the field. We would argue that, within the policy context described above, these tensions emerged via the edTPA's dual positioning as: (1) an accountability mechanism summatively used to gatekeep beginning teachers' entry into the profession; and (2) a formative tool for learning by gathering, synthesizing, and reflecting on evidence of teaching practice. On the question of whether summative and formative assessment goals can coexist, Brookhart (2010) argues that they can, as long as: (1) the timing and consequences of the assessment are well communicated and appropriate; and (2) all of the assessment's elements, including performance tasks, interpretive language, and feedback, are criterion-referenced and grounded in the purposes of encouraging learning and informing decision making. Bennett (2011) adds a point of caution that any assessment characterized as formative should disclose its inferential assumptions and processes to those being assessed and allow for extensive support and dialogue among test users and test takers, if all are to benefit from its outcomes.
With these premises in mind, our results suggest that the edTPA's use as a state certification test has underscored the assessment's summative function and repressed its formative potential. While the edTPA's prompts and rubric criteria capture crucial and complex elements of teaching, the exam's pass/fail status and its sparse feedback in the form of numerical ratings provide little useful information to edTPA takers for improving their practice. Further, many participants in this study demonstrated confusion over permissible support roles within their teacher education networks and no knowledge of external portfolio evaluators' inferential assumptions and processes, which factored markedly into the positions of those who perceived the edTPA to be unfair. In other words, the edTPA's mode of implementation in New York and Washington deviates from Brookhart's (2010) and Bennett's (2011) principles of conjoined summative and formative assessment, contributing instead to the support, representation, and agency tensions identified in this paper.

Were the Tensions Associated with the edTPA Productive for Candidates?
We agree, with qualifications, that there should be restrictions to entering the teaching profession, that TPAs can serve productive purposes, and that candidates should be held responsible for high-quality planning, instruction, and assessment. However, we point out that some of the tensions that candidates faced and mediated during early implementation of a compulsory edTPA in New York and Washington States were not necessarily productive toward the ends of improving teaching and student learning -one of the central purposes of TPAs articulated in the research literature (Bunch et al., 2009;Chung, 2008;Darling-Hammond, Newton, & Wei, 2013;Peck et al., 2014;Wei & Pecheone, 2010). We draw from Stillman's (2011) conception of productive tensions as situations in which teachers are able to learn from challenging or adverse conditions, given opportunities to grapple with, adapt to, and strengthen their practice through them. Our results demonstrate how externally controlled conditions of candidates' field experiences, including curricular and instructional constraints at the school level and specific pedagogical expectations within the edTPA, converged to challenge their capacities. Yet candidates in New York and Washington inevitably must mediate those challenges successfully, as there are no feasible alternatives to the edTPA that lead to teaching licensure in those states. In that regard, the edTPA may offer a cloudier picture of candidates' potential as teachers than of their abilities to navigate the complications associated with a high-stakes performance assessment.
We found two mediation strategies to be productive for candidates, with caveats. The first was social networking as a means of strengthening their knowledge of and performance on the test. Via these networks, preservice teachers worked to clarify language associated with learning and teaching in their disciplines, as articulated in the edTPA; they shared planning, instruction, and assessment strategies that aligned with the edTPA rubric criteria; they discussed challenges to successful completion of the assessment; they peer-reviewed each other's writing; and they provided emotional and intellectual support throughout a process marked by vexations and uncertainties. Ironically, as some candidates suggested and the 2013 edTPA support guidelines implied, these kinds of formative feedback-generating interactions also tested the limits of acceptable support and excluded teacher educators and supervisors, whose targeted guidance might have strengthened candidates' learning experiences and teaching practice (Okhremtchouk et al., 2013;Pecheone & Chung, 2006).
The second productive mediation strategy was positioning the writing process, which directed candidates to rationalize their curricular and instructional choices and support them with evidence, as an intentionally interpretive act. Consistent with results reported by Bunch et al. (2009), some participants in this study affirmed that writing reflectively and analytically about their teaching helped them to advocate for particular instructional aims, connect their teaching methods and students' learning outcomes to those aims, and discuss the implications of their teaching for different kinds of learners, including students with disabilities and English language learners. In other words, the edTPA's written components provided them with opportunities to broker discretion in and clarify their pedagogical decision making (Caughlan & Jiang, 2014). For others, however, the representation tensions inherent in the writing process, contextualized by the assessment's high stakes, led to a kind of gaming imperative, with candidates focusing predominantly on what kinds of artifacts and narrative content evaluators might rate favorably. While it is important that teachers learn to reflect on, explain, and reconsider their teaching through written self-assessment, we associate the gaming imperative with an externally controlled, non-transparent, high-stakes evaluation process and question its legitimacy for strengthening teachers' practices and their students' learning.
Finally, we discovered several efforts to mediate the edTPA's tensions that seem unlikely to generate any improvements in candidates' teaching and their students' learning. These efforts included translating the edTPA's rules and processes for cooperating teachers and arbitrating discrepant demands from school-and university-based mentors; positioning the assessment as an authority mechanism in an effort to steer curricular and instructional priorities in their cooperating teacher's classrooms; and modifying the classroom environment in unusual ways to make the edTPA portfolio appear chronologically cohesive, or to address video recording restrictions. We found no evidence in this or previous studies that these procedures, which originate in the edTPA's highstakes status, strengthen teaching candidates' professional growth. Instead, they struck us as attempts to control for constraints and variables in the assessment system, with the core goal of avoiding test failure. The presence of these unproductive tensions corroborates Wei and Pecheone's (2010) argument that edTPA's situation within a policy context -one in which the assessment serves political ends as an accountability lever -inevitably affects the fidelity of its approximation of teaching and its potential as a tool for strengthening the quality of K-12 teaching and teacher education.

Conclusion: Implications for Teaching Quality and Teacher Education
Like its predecessors -the PACT and NBPTS portfolio -the edTPA's purpose is to improve teaching quality by assessing and evaluating, in a robust and valid way, not just what teachers know about learning and teaching, but how they enact their practice and use evidence thereof to impact student learning outcomes. This is a noble and well-warranted intent, and one that we believe represents a positive direction for the field of teacher preparation. But unlike the PACT, which began in schools of education as a tool for strengthening programs while also meeting the terms of a California state law requiring candidates to pass a performance assessment of teaching upon program completion; and unlike the NBPTS portfolio, which is voluntary and reserved for experienced teachers with far more agency in their professional situations than preservice candidates; the edTPA in New York and Washington rests within a policy structure that seems designed to shift the locus of control over defining and evaluating quality teaching toward the state.
While several scholars cautiously suggest that such policy could encourage positive changes within teacher education programs under the right conditions (Lit & Lotan, 2013;Margolis & Doring, 2013;Peck et al., 2010;Peck & McDonald, 2013), our research demonstrates the added potential for candidates, and perhaps teacher educators, to focus defensively on managing social and instrumental factors that interact with the assessment rather than improving practice through it. Over time, it will be important to investigate questions that this study, conducted early in the edTPA's use as a state licensure mechanism, does not address. For instance, how will the edTPA's normalization within the teacher education and certification landscape affect the discourse on teacher quality and the practices associated with learning to teach? What attributes of teaching will be emphasized; and how will teacher education programs and school-based mentors represent and support those attributes, as today's test takers become tomorrow's cooperating teachers?
Turning back to the flows of authority we identified in our theoretical framework, teacher education institutions in New York, Washington, and other states in which the edTPA is adopted as a high-stakes certification assessment have several questions to consider in choosing how to position the assessment within their programs. First, how might programs adapt to align with the edTPA's core pedagogical underpinnings, yet still maintain their unique identities and expand candidates' views of teaching to encompass important practices like building social relationships with students and community members, mediating policy influences on their teaching, and leading school change initiatives with colleagues? Our results demonstrate that placing heavy stock in the edTPA has the potential to forsake important professional roles and dispositions for "tested" instructional tasks (Zeichner, 2012). Second, how might institutions use evidence from their candidates' edTPA performance to affect programmatic changes while simultaneously acknowledging and working to mitigate the tensions and validity threats qualifying that evidence? Our study of candidates' experiences with the edTPA in its first year of implementation in New York and Washington offers an impression of the kinds of tensions and threats teacher education programs should consider addressing.
Finally, and more generally, how can teacher educators use the edTPA experience concomitantly as a tool for strengthening teaching and an opportunity to help candidates develop their proficiencies as engaged mediators of policy and practice (Cohen & Hill, 2001)? It is apparent that the edTPA's location in a policy context generates conflict between the formative value and summative imperative of teacher performance assessments. Like other standardized tests used to leverage high-stakes decisions, the edTPA is subject to corruption pressures that enable distortion of the very practices it is intended to evaluate (Campbell, 1976). Given the results of our study, we recommend that teacher educators not only support the competencies and practices evaluated by the edTPA, but also illuminate and analyze the impacts of the assessment in a policy context and reinforce what seems to matter most to successful edTPA completion: positive support relationships; the ability to adapt to multiple -sometimes divergent -expectations and demands; and effective synthesis of pedagogical aims, practices, and evidence into arguments about their consequences for student learning. Doing so, and doing so constructively, seems pivotal to capitalizing on the productive qualities of the edTPA, mitigating the potentially deleterious effects of the assessment's unproductive tensions, and strengthening beginning teachers' roles as engaged mediators of policy and practice beyond the edTPA experience.

About the Authors
Kevin W. Meuwissen, Ph.D. University of Rochester kmeuwissen@warner.rochester.edu Kevin Meuwissen is an Assistant Professor of Social Studies Education and directs the social studies teacher preparation program at the University of Rochester. His research interests include the political socialization of preservice and early-career social studies teachers, the intersections of policy and teaching practice in teacher education and professional development, and adolescent learners' political thinking. He has published in Theory and Research in Social Education and The Social Studies, served as a pedagogical consultant and program evaluator for several U.S. Department of Education-funded Teaching American History Programs, and delivered testimony on teacher certification policy to the New York State Legislature.