READING BETWEEN THE LINES: THREE ESSAYS ON THE DEVELOPMENT AND IMPLEMENTATION OF MICHIGAN’S READ BY GRADE THREE LAW By Amy Cummings A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Education Policy – Doctor of Philosophy 2024 ABSTRACT Over the past two decades, U.S. states have widely adopted early literacy policies. These policies have shown short-term success in enhancing K-3 literacy skills. However, the reasons behind their widespread adoption and the factors driving their success are poorly understood. This three-paper dissertation focuses on Michigan’s Read by Grade Three Law, enacted in 2016, to delve into the adoption and implementation of these policies. The Read by Grade Three Law is an informative case study because it is one of the U.S.’s most comprehensive early literacy policies. The first paper employs interviews and policy document analysis to explore the law’s adoption and the dissemination of early literacy policies across states, highlighting the significant role of policy entrepreneurs. The second and third papers examine the policy’s implementation, specifically its family engagement requirement and districts’ methods to identify students with “reading deficiencies.” The second paper analyzes data from the Michigan Department of Education and educator surveys spanning 2019-2023, uncovering that only 20% of eligible students receive “Read at Home” family engagement plans. The findings highlight considerable differences among districts and demonstrate how educators’ understanding and perceptions are closely linked to the execution of these plans. The third paper uses superintendent survey data and state records from the 2021-22 school year to investigate how districts identify students with “reading deficiencies,” making them eligible for supports such as “Read at Home” plans. The results reveal that districts use diverse measures to identify students. These variations are related to significant disparities in identification rates, with implications for which students receive literacy support under the Read by Grade Three Law. Together, these studies illuminate the complexities of policy adoption and implementation, enhancing our understanding of early literacy policies and laying the groundwork for future research on their mechanisms of success. This dissertation is dedicated to Papa Dean. You may not be here to see me cross the finish line, but without you, I wouldn’t have crossed the starting line. iii ACKNOWLEDGEMENTS First and foremost, I am grateful to my advisor, Dr. Katharine O. Strunk. You are why I chose Michigan State University for my Ph.D. in Education Policy. Your constant support and encouragement played a crucial role in getting me here and seeing me through to the finish line. You always pushed me to excel, which has significantly enhanced the quality of this dissertation. You’ve also made me feel like a part of your family, giving me cherished memories over the last five years that I’ll hold onto forever. My journey would have been incomplete without Dr. Tasminda K. Dhaliwal. You graciously stepped in as my co-advisor and dissertation chair after Katharine left, and your collaboration during the last leg of my program has been invaluable. Thank you for the support and encouragement over the past year; I couldn’t have done it without you. I also owe a huge thanks to my other committee members, Drs. Tanya S. Wright and Scott A. Imberman. Tanya, your literacy expertise, and Scott, your guidance on quantitative methods, have both been crucial to this dissertation’s success. You helped me place these studies within the broader scope of literacy policies and learning and identify the correct methods for analysis. I also must express my deep gratitude to Dr. Rebecca Jacobsen. Your involvement in my guidance committee and contribution to my first dissertation paper significantly improved my qualitative research skills. I am profoundly grateful to the Education Policy Innovation Collaborative (EPIC) for my graduate research assistantship over the last five years. Special thanks go to Emily Mohr, Tara Kilbride, Michelle Huhn, and Jessica Sinnaeve, as well as the various members of the Read by Grade Three team and post-doctoral scholars I have collaborated with, including Qiong Zhu, John Westall, and Josep Maria Nadal Fernandez. Working at EPIC has enriched my academic iv experience and endowed me with invaluable skills that have prepared me for the next chapter in my professional journey. Next, I must extend my heartfelt thanks to my wonderful fiancée, Annie, who has been an unwavering source of support, particularly during the latter stages of my program. I will be forever grateful for your decision to move to Lansing and be with me as I complete my Ph.D. You always amaze me with your endless support for me and our little family here, and I can’t wait for this next chapter together in New York City. You are my person, and I love you. Mom, Dad, and James, you have been a constant and steadfast support throughout my entire educational journey, all the way to and through my Ph.D. program. Mom and Dad, you have always placed a high value on education, saving diligently to ensure James and I could pursue our academic goals. Thanks to your unwavering support, this commitment to education has profoundly shaped my life. You have backed every professional decision I’ve made, always believing in my ability to succeed. The trips to Lansing and my returns home to visit have created some of the fondest memories of my Ph.D. journey, providing much-needed support through the most demanding challenges, including the pandemic. Your encouragement during those low points was indispensable. Bumma, you and Papa Dean have been incredible champions of my education. Your support enabled me to move to New York City for my master’s degree, which was a pivotal step towards working in education policy in D.C. and ultimately pursuing my Ph.D. Your encouragement has been a guiding force to my current achievements. While it saddens me that Papa Dean cannot share this moment with us, my hope to make him proud remains a driving force. To the Young family—Mike, Marsha, Katie, Lesley, Tim, Andy, Michael, Alice, Emily, v and Sarah—you’ve become an extended family to me in ways I never anticipated when I started my Ph.D. program. Coming out during graduate school was unplanned, but your warm acceptance and the support you’ve extended have brought immense joy and strength to my journey these past few years. Your genuine interest and encouragement regarding my dissertation work have been incredibly meaningful, and I eagerly look forward to becoming an official member of your family later this year. My journey would have been impossible without my Lansing support system, both within and outside the program. Meg, my EPIC buddy and Qualtrics guru, it was clear from your first day at EPIC that our friendship would last a lifetime. You provided me unparalleled support— helping with brainstorming sessions, editing drafts, or simply being there for a chat or a quick fast-food run. We nearly completed all five years together, and I look forward to visiting you in the exciting places your career takes you. Caroline, my mentee, I’m so grateful you chose MSU. You’ve been incredibly supportive throughout this program, always ready for a Bachelor Monday, drinks at Horrock’s, or Fat Cat Friday. Your sense of humor and readiness for fun were essential for surviving this Ph.D. journey. You’re among the most brilliant and remarkable people I’ve met, destined for great things after this program. I can’t forget Matthew, our program’s brightest and most helpful classmate. I’ll miss our times in the EPIC lab and sharing our little corner with Caroline. And to the Drinking with Consequence trivia team—Drew, my cohort companion, Tyler, and Marge—your camaraderie has been invaluable. I also must express my gratitude to my dog, Lulu. Lulu entered my life during one of the most challenging times of this program—the onset of the pandemic at the end of my first year. Her arrival into my home brought immense joy and companionship during profound isolation. As we returned to campus and office life, Lulu was lucky to find a friend and sister in Lucy. vi Although I wish Lucy could be here to celebrate this milestone with us, I find comfort in believing she’s joyfully looking on (and probably barking) from across the rainbow bridge. Finally, in the spirit of Niecy Nash-Betts’ inspiring words during her 2024 Emmy acceptance speech, “I want to thank me.” I’m filled with pride for completing this program. There were countless times when giving up was a tempting option, when doubt crept in, making me question whether I could do this. Yet, with each setback, I found a way to stand back up, devise a new strategy, and push forward, bringing me to this moment of achievement. So, here’s to me. I am profoundly proud of what I’ve accomplished. vii PREFACE This research result used data structured and maintained by the MERI-Michigan Education Data Center (MEDC). MEDC data is modified for analysis purposes using rules governed by MEDC and are not identical to those data collected and maintained by the Michigan Department of Education (MDE) and/or Michigan’s Center for Educational Performance and Information (CEPI). Results, information, and opinions solely represent the analysis, information, and opinions of the author(s) and are not endorsed by, or reflect the views or positions of, grantors, MDE, and CEPI or any employee thereof. The research reported here was supported by the Institute of Education Sciences, U.S. Department of Education, through Grants R305B200009 and R305H190004 to Michigan State University. The opinions expressed are those of the authors and do not represent views of the Institute or the U.S. Department of Education. viii TABLE OF CONTENTS INTRODUCTION................................................................................................................ 1 PAPER 1: “A LOT OF STATES WERE DOING IT”: THE DEVELOPMENT OF MICHIGAN’S READ BY GRADE THREE LAW............................................................. PAPER 2: BEYOND THE CLASSROOM? THE IMPLEMENTATION OF FAMILY ENGAGEMENT UNDER MICHIGAN’S READ BY GRADE THREE LAW.................. 8 41 PAPER 3: WHO GETS FLAGGED? DISTRICT APPROACHES TO K-3 “READING DEFICIENCY” IDENTIFICATION UNDER MICHIGAN’S READ BY GRADE THREE LAW....................................................................................................................... 115 CONCLUSION.................................................................................................................... 162 BIBLIOGRAPHY................................................................................................................ 167 APPENDIX 1.A: INTERVIEW CODES............................................................................. 191 APPENDIX 1.B: POLICY DOCUMENT CODES............................................................. 193 APPENDIX 2.A: LIST OF DISTRICT “READ AT HOME” PLANS REVIEWED.......... 195 APPENDIX TABLES.......................................................................................................... 196 APPENDIX FIGURES......................................................................................................... 212 ix INTRODUCTION Early literacy policies have seen widespread adoption across the United States in the past two decades. By 2022, 48 states and the District of Columbia had implemented various early literacy initiatives aimed at enhancing literacy skills by the end of 3rd grade, recognized as a critical milestone for further learning and later outcomes (Cunningham & Stanovich, 1997; ExcelinEd, 2022a; Fiester & Smith, 2010; Hernandez, 2011; Sparks et al., 2014). Growing causal evidence from both state-level and national studies suggests these policies effectively enhance student achievement, particularly in the short term (Greene & Winters, 2004, 2006, 2007; Schwerdt et al., 2017; Slungaard Mumma & Winters, 2023; Westall & Cummings, 2023). However, the reasons behind the widespread adoption of early literacy policies and the mechanisms driving their apparent success still need to be better understood. Many studies focus solely on specific aspects, such as mandated retention for 3rd graders falling below a particular score on state literacy assessments, neglecting other potential policy mechanisms. Understanding the contextual factors and mechanisms underlying these policies will be crucial for interpreting existing research and guiding future investigations into their efficacy. This three-paper dissertation contributes fundamental descriptive research to the literature on early literacy policies, focusing on one of the nation’s most comprehensive initiatives: Michigan’s Read by Grade Three Law (Michigan Public Act 306, 2016). This law, enacted in 2016, encompasses a wide array of supports commonly found in early literacy policies across states. The comprehensive nature of Michigan’s policy provides fertile ground for studying the passage and implementation of early literacy policies, offering valuable insights for policymakers and practitioners. The theory of change underlying the Read by Grade Three Law is illustrated in Figure 0.1 1 and is emblematic of how these policies are designed to work in states that have adopted them (Strunk et al., 2021). The figure groups the policy’s supports into two main pathways aimed at improving K-3 literacy achievement. The first pathway, depicted along the top half of the figure, involves literacy instructional supports for educators (i.e., “Tier 1” or general education supports). This pathway aims to ensure that the literacy instruction provided to all K-3 students is high-quality and grounded in evidence-based practices. To this end, the Read by Grade Three Law requires and provides funding for the Michigan Department of Education (MDE) to develop an early literacy coaching model at the Intermediate School District (ISD) level (Michigan Public Act 306, 2016).1 These coaches offer one-on-one literacy coaching and other professional development to K-3 teachers in the ISD, focusing on evidence-based instructional practices, including the “Big 5” components of reading: phonemic awareness, phonics, fluency, vocabulary, and comprehension. Additionally, the state has invested in the development and training of the Essential Instructional Practices in Early Literacy: Grades K-3 (Michigan Association of Intermediate School Administrators General Education Leadership Network Early Literacy Task Force, 2016). These “Literacy Essentials” identify ten research-based instructional practices for K-3 teachers and serve as a foundation for professional learning and instruction across the state. The goal is for these coaching and professional development supports to improve all K-3 teachers’ literacy instruction. This, in turn, is expected to enhance overall English Language Arts (ELA) performance and long-term outcomes for all students in the state. 1 In Michigan, ISDs are educational entities that operate between MDE and local education agencies, often serving the local education agencies within a given county. Local education agencies can receive a range of services through their ISD. 2 Figure 0.1 Read by Grade Three Law Theory of Change Note. This figure is from Strunk et al. (2021). The second pathway, illustrated along the bottom half of Figure 0.1, involves monitoring, remediation, and retention (i.e., “Tier 2” and “Tier 3” increasingly intensive interventions). This pathway first relies on districts using “valid, reliable, and aligned” screening assessments to identify K-3 students needing additional literacy support.2 The law requires districts to administer these assessments at least three times a year to frequently monitor literacy proficiency in the early grades and provide early warnings about which students are behind in literacy so they can be identified for support as soon as possible (Michigan Public Act 306, 2016). Once 2 I use the term “screening assessments” here as opposed to “diagnostic assessments” as used in Figure 0.1 as screening assessments more accurately reflects the types of assessments districts use to initially identify students needing additional literacy support. 3 identified, districts must provide these students with increasingly intensive interventions to help them read at grade level, including individualized reading plans, increased time on literacy instruction, one-on-one and small group instruction, summer support, and parental involvement. If, despite these supports, a student still scores below the state-determined threshold on the 3rd- grade state ELA assessment, the law mandated retention and increased support and remediation from 2020-21 through 2022-23 (Michigan Public Act 306, 2016).3 Ultimately, these two pathways are designed to work together to improve the literacy instruction all students receive, identify students needing support early on, and provide increasingly intensive interventions as needed to ensure they read at grade level by the end of 3rd grade. Early literacy policies thus include multiple mechanisms and pathways intended to improve K-3 literacy achievement, but many studies narrowly focus on 3rd-grade retention, often neglecting other critical components. This dissertation aims to address this gap by examining other elements of these policies. I first explore the motivations behind the Read by Grade Three Law’s passage and how it came to include these components. Then, I examine the implementation of two specific elements of this theory of change: (1) the selection and use of valid, reliable, and aligned K-3 screening assessments and (2) parental involvement. The first paper, co-authored with Drs. Katharine O. Strunk and Craig De Voto and published in the Journal of Educational Change, delves into the development and passage of Michigan’s Read by Grade Three Law, shedding light on how and why states enact early literacy policies and the widespread adoption of various policy components nationwide (Cummings et al., 2023). Through interviews with state-level stakeholders and analysis of early literacy policy 3 Retention was originally scheduled to go into effect in the 2019-20 school year but was postponed due to the COVID-19 cancellation of state testing (Whitmer, 2020). In 2023, the Michigan legislature repealed the retention component of the Read by Grade Three Law, effective in the 2023-23 school year (Senate Bill 12, 2023). 4 documents, the paper uncovers the pivotal factors that drove the development and enactment of the Read by Grade Three Law. These include educational concerns such as low literacy scores, existing state-level literacy supports, and political dynamics, including an opportunity for legislative compromise. The findings also highlight the influence of policy entrepreneurs, such as ExcelinEd, a nonprofit organization chaired by former Florida Governor Jeb Bush, which advocates for retention-based early literacy policies akin to that state’s across the U.S. These findings offer valuable lessons on the intricate interplay of educational and political dynamics in the development and enactment of early literacy policies, providing crucial insights for policymakers seeking to advance literacy initiatives. The second and third papers offer in-depth, descriptive examinations of two less-explored aspects of early literacy policies. One such component, explored in the second paper and illustrated in the “Early Intervention and Support” box in Figure 0.1, is parental involvement, a key feature of early literacy policies in 27 states (ExcelinEd, 2022a). Michigan’s Read by Grade Three Law, like other states, mandates districts to provide “Read at Home” plans to families of students who have been identified as behind in literacy, including workshops for caregivers and encouragement of regular home reading (Michigan Public Act 306, 2016). While research consistently emphasizes a positive correlation between family engagement and children’s literacy development (Aikins & Barbarin, 2008; Boonk et al., 2018; Castro et al., 2015; M. A. Evans et al., 2000; Fan & Chen, 2001; Georgiou et al., 2021; Inoue et al., 2018; Lin, 2003; Ma et al., 2016; Miedel & Reynolds, 1999; Sénéchal, 2006; Voorhis et al., 2013; Wilder, 2014), the implementation of mandated family engagement in early literacy policies remains largely unexplored. This paper examines the proportion and characteristics of students provided “Read at Home” plans across Michigan districts, using administrative data from MDE and educator 5 surveys spanning 2019-2023. The analysis uncovers a stark 20% implementation rate for these plans among eligible students, with considerable variation across districts. Notably, educators’ understanding and perceptions are significantly related to plan provision, with students more likely to receive a plan when their teachers and principals understand and perceive it as effective. These findings provide valuable insights for policymakers and practitioners in Michigan and beyond aiming to integrate family engagement into their early literacy policies. Another less-examined aspect of early literacy policies, examined in the third paper and present in 38 states, pertains to identifying K-3 students with “reading deficiencies” (ExcelinEd, 2022a).4 As shown under “Monitoring, Remediation & Retention” in Figure 0.1, the Read by Grade Three Law requires districts to select and use “valid, reliable, and aligned” K-3 screening assessments to identify students needing additional literacy support. The law designates these students as having “reading deficiencies,” rendering them eligible for potentially beneficial literacy interventions, such as the “Read at Home” plans mentioned above (Michigan Public Act 306, 2016). Consequently, the methods districts employ to identify “reading deficiencies,” including the measures and specific assessments they use, have important implications for which students receive essential literacy support. Moreover, with the expanding evidence base on assessments, it is increasingly crucial to assess whether districts use measures that research has demonstrated accurately identify students needing support. This paper draws on superintendent survey data and state administrative records from the 2021-22 school year to delve into how districts identify K-3 students with “reading deficiencies” under Michigan’s Read by Grade Three Law. The findings indicate that most districts employ multiple measures, including combining various assessments and considering students’ ELA classwork performance. Notably, 4 I use the term “reading deficiency” because that is how the Read by Grade Three Law refers to students whose districts have identified them as needing additional support in literacy (Michigan Public Act 306, 2016). 6 the study emphasizes a preference for computer-adaptive tests, particularly NWEA, which research has indicated are cost-effective and accurate relative to other screening measures. Furthermore, the analysis reveals disparities in identification rates across approaches, with districts relying on a single assessment demonstrating higher identification rates, particularly among students from historically underserved backgrounds. These findings underscore districts’ diverse strategies and the importance of carefully selecting screening measures to ensure an equitable and accurate identification process. This dissertation addresses critical gaps in our understanding of early literacy policies by focusing on Michigan’s Read by Grade Three Law as a case study. Through qualitative and descriptive analyses, it sheds light on the processes involved in developing, passing, and implementing such policies. The first paper elucidates the complex interplay of educational and political factors driving the enactment of early literacy policies, offering valuable insights for policymakers nationwide. Subsequent papers delve into underexplored aspects of these policies, namely family engagement and the identification of “reading deficiencies,” revealing significant implementation challenges and disparities across districts. These findings contribute to our understanding of early literacy initiatives and provide actionable insights for policymakers and practitioners seeking to enhance student literacy outcomes. 7 “A LOT OF STATES WERE DOING IT”: THE DEVELOPMENT OF MICHIGAN’S PAPER 1: READ BY GRADE THREE LAW5 Amy Cummings, Michigan State University Katharine O. Strunk, Michigan State University Craig De Voto, University of Illinois at Chicago In recent years, a majority of states have adopted policies intended to promote early literacy, with particular attention paid to ensuring that students are reading proficiently by third grade (Council of Chief State School Officers [CCSSO], 2019). These policies share many key elements, including mandates for evidence-based literacy instruction, diagnostic and progress monitoring assessments, literacy interventions for students identified as needing additional support, and, in some cases, retention for third graders who do not meet a predetermined cut score on their state’s reading assessment. This kind of education policy transfer across states is not a unique phenomenon; researchers have documented analogous proliferations of similar policies related to charter schools, teacher evaluation, and school choice (Kraft & Gilmour, 2017; Renzulli & Roscigno, 2005; Wong & Langevin, 2007). Indeed, the spread of policy ideas across states has occurred in fields that extend beyond K-12 education, such as in health care (Carter & LaPlant, 1997), criminal justice reform (Bergin, 2010), and transportation (Moreland-Russell et al., 2013). The study of policy transfer has documented how local, state, and even national policymakers imitate others as they implement policies intended to reach some common but localized goal (Stone, 2000). 5 This manuscript has previously been published in the Journal of Educational Change (Cummings et al., 2023). Permission from the journal has been granted to include it in this dissertation. 8 Although the imitation phenomenon has been well-documented in the policy literature, there has been little attention paid to how policy transfer happens in K-12 education. This is a particularly interesting venue in which to study policy spread, as U.S. K-12 education policy is highly localized, with the majority of policy dictated by states and local school districts, school buildings, and even individual classrooms (Mitra, 2018). To better understand how education policies spread across states, we turn to the case of Michigan and early literacy policy. Michigan adopted its Read by Grade Three Law in 2016 (Michigan Public Act 306, 2016) and today is one of 19 states with similar retention-based third-grade literacy policies. Guided by Kingdon’s Multiple Streams Framework (MSF; Kingdon, 1984) and the theory of policy transfer (Dolowitz & Marsh, 1996, 2000; M. Evans & Davies, 1999; D. Marsh & Sharman, 2009; Stone, 2000), we ask, what factors influenced the development and passage of Michigan’s Read by Grade Three Law? To answer this question, we trace the policy process surrounding the Law’s conception, development, and passage, relying on data from semi-structured interviews from 24 stakeholders involved in the development of the Law, including state legislators, officials from the Michigan Department of Education (MDE), and external stakeholders (e.g., early literacy leaders, educational association leadership). In order to understand the extent to which early literacy policies spread—or transfer—between states, we also analyze policy documents from all 50 states and D.C. We find that events in the problem and political streams opened a policy window that allowed for the passage of the Read by Grade Three Law (Kingdon, 1984). Further, we find that policy entrepreneurs—individuals who are willing to invest their resources in order to get their preferred policy passed—played a key role in facilitating this process (Kingdon, 1984). After 9 Florida, one of the earliest and most influential states to implement a third-grade literacy policy, passed its policy in 2002, 18 other states adopted similar policies by 2021 thanks to the critical role of policy entrepreneurs in transferring core components of Florida’s policy across states. Our findings from this research will contribute to policymakers’ and other stakeholders’ understandings of the development and passage of third-grade literacy policies—information that will be important as these policies continue to receive national attention in both policy and research communities. Moreover, this study is one of few in the field to focus primarily on how policy entrepreneurs (see Lieberman, 2002; Mintrom & Vergari, 1998) play a critical role in joining together the multiple streams as identified by Kingdon (1984). Lastly, this study provides a nuanced view of how policy transfer (Dolowitz & Marsh, 1996, 2000) and policy entrepreneurship join together to promote the convergence of ideas and solutions to particular problems. Conceptual Framework To examine the conditions that influenced the development and passage of Michigan’s Read by Grade Three Law, we use two complementary theories: Multiple Streams Framework (MSF) (Kingdon, 1984) and policy transfer (Dolowitz & Marsh, 2000; also see Dolowitz & Marsh, 1996; Evans & Davies, 1999; Marsh & Sharman, 2009; Stone, 2000). According to MSF, policy issues emerge on government decision-making agendas through three “streams”: problem, policy, and political. In the problem stream, many conditions exist that can rise to the level of problems. Kingdon (1984) distinguishes between conditions and problems, where conditions “become defined as problems when we come to believe that we should do something about them” (p. 109). This can occur through a dramatic or obvious change in an indicator (e.g., a sudden decrease in test scores), or a focusing event such as a court case (e.g., Brown v. Board of 10 Education) or disaster (e.g., Columbine, Sandy Hook). However, a defined problem only becomes a pressing problem—one that requires deliberate political action—when it can be collectively solved (Jones et al., 2016; Petridou & Mintrom, 2020). In this way, only some conditions translate into problems, and only some problems ultimately receive political attention. The policy stream is predicated on the relative prominence of problems. Those problems deemed to be most pressing by members of the policy community garner myriad proposals (i.e., alternatives). This is particularly true when reasonable solutions already exist within the political landscape (e.g., third-grade literacy policies). Lastly, the political stream is made up of public mood, the political composition of the legislature and other governing bodies, election results, and changes in administration. If the political stream is ripe, it can interact with the other streams to form the circumstances needed to adopt policy innovations (Kingdon, 1984). When these three streams join together, they produce a “policy window” (Kingdon, 1984). This policy window can create an opportunity for policy change by allowing advocates to push their solutions forward. Notably, policy entrepreneurs, who are individuals or actors from inside (i.e., elected or appointed officials) or outside government (e.g., interest groups, research organizations), play key roles in this process. Policy entrepreneurs either wait for these streams to join together, or work to do so on their own (Doig & Hargrove, 1990). They then present a solution that addresses the pressing problem, often advocating for a specific “pet” policy or political agenda. For instance, policy entrepreneurs can shape the problem stream by interpreting and relaying for others the indicators or focusing events that cause problems to need a solution. This is done via leading by example, framing the problem to suit their needs, or communicating with others in their networks (Dewulf & Bouwen, 2012; Mintrom & Luetjens, 2017; Shpaizman et al., 2016). Kingdon (1984) further calls these collective efforts a “softening up” process, 11 helping to make meaning of pressing problems for other members of the policy community. Entrepreneurs can the gauge how receptive the policy community is to adopting their policy alternative. In turn, entrepreneurs’ alternatives tend to rise on the decision-making agenda as they galvanize policy proposals around a specific solution to a given problem (Kingdon, 1984; Mintrom & Vergari, 1998). When the administration favors their particular innovation (e.g., a retention-based third-grade literacy policy), it is likely to become law. MSF serves as a useful tool through which to examine national attention toward early literacy policy formation. For example, Young and colleagues (2010) show how political pressure from respective administrations, regional and national interest, and declining reading achievement was symbolic of the failure of public schools. This perceived failure cause governors to identify early literacy as a pressing problem in need of a solution, influencing which issues were placed on the decision-making agenda. In Wisconsin, Brown (2007) documents the development of Wisconsin’s “no social promotion” statutes that were intended to improve students’ literacy. In particular, the Wisconsin governor framed retention as the solution fix “passing students along.” The state legislature, however, resisted implementing student retention as the sole solution to the early literacy problem, instead proposing and passing a policy alternative: a multiple-indicator retention policy. Although this extant literature examining the adoption of early literacy policies through MSF helps to shed light on the rise of comprehensive literacy policies nationally, it largely neglects the role of policy entrepreneurs as critical actors in helping to shape the policies themselves. Our study aims to fill this gap. Moreover, while Kingdon's (1984) MSF helps us understand the importance of policy entrepreneurs in framing problems and particular solutions to such problems, it does not leave space for examining the particular way that early literacy policies spread across the United 12 States, generating a kind of de facto set of specific elements that were considered appropriate across individual states in their adoption of early literacy policies. To better understand how the national discourse around early literacy policies shaped individual states’ reforms, we adopt the concept of “policy transfer,” which highlights how “knowledge about policies, administrative arrangements, institutions and ideas in one political setting (past or present) [is] used in the development of politics, administrative arrangements, institutions and ideas in another political setting” (Dolowitz & Marsh, 2000, p. 344; also see Dolowitz & Marsh, 1996; Evans & Davies, 1999; Marsh & Sharman, 2009; Stone, 2000). Despite similarities to “policy diffusion” (Shipan & Volden, 2008, 2012), Marsh and Sharman (2009) argue that diffusion emphasizes organizational and structural processes in policy adoption, whereas transfer privileges a more agent-centered approach. Because we are focusing on the particular actors (i.e., entrepreneurs) that shaped Michigan’s Read by Grade Three Law (and less so on the organizational structures), policy transfer is therefore a more applicable theory. As we explain in detail below, policy entrepreneurs were highly influential agents in the passage of the Law, promoting the transfer of Florida’s early literacy policy to Michigan. But because the policy entrepreneurs in question were non-governmental agents, they could not impose their policies on a given system. Instead, they provided the rhetoric and scholarly discourse (Stone, 2000) needed to shift attitudes and legitimize their position. From this view, policy entrepreneurs promote the “voluntary” transfer (see Dolowitz & Marsh, 1996) of policies across governmental entities, particularly when their ideological disposition aligns with the prospective policy community (see Stone, 2000). This is done through several strategies, two of which we elaborate in this study: 1) emulation and 2) elite networking/interaction (see Bennett, 1991). Emulation involves borrowing ideas and adapting policy approaches, tools, or structures 13 to local conditions whereas elite networking involves agents sharing their expertise to form common patterns of understanding regarding policy. By combining the theories of MSF and policy transfer, we seek to highlight how state and national conditions resulted in the rise of early literacy as a pressing problem for Michigan. In particular, we explore how addressing early literacy in the state became an urgent problem, and how one approach gained considerable traction across states (i.e., retention-based third-grade literacy policies). Moreover, we use this framework to unpack how non-governmental actors can become critical policy entrepreneurs, interpreting problems and networking particular solutions for decisionmakers. In Michigan, we found two key policy entrepreneurs—the Great Lakes Education Project (GLEP) and ExcelinEd—influenced a majority-Republican administration to adopt a retention-based third-grade literacy policy. While some alternative and supplementary components were added by Democrats before passage, much of the Law retained the retention- based theory of change to address Michigan’s early literacy problem. Third-Grade Literacy Policies State and national efforts in recent decades have made early literacy a priority, laying the groundwork for the transference of third-grade literacy policies in several states. These efforts can be traced back to a flurry of policy recommendations in the late 1990s. In 1997, the U.S. Congress asked the National Institute of Child Health and Human Development to work with the U.S. Department of Education to establish a National Reading Panel to review the existing evidence surrounding the best ways to teach reading (Pearson et al., 2020). They found that the best approach to reading instruction included explicit instruction in phonemic awareness, systematic phonics instruction, methods to improve fluency, and ways to enhance comprehension (National Institute of Child Health and Human Development, 2019). In 1998, the 14 National Research Council published a report similarly concluding that reading ability is determined by multiple factors, including knowledge, language, and other internal processes (Pearson et al., 2020). That same year, the Department of Education implemented the Reading Excellence Act, which awarded $210 million annually from 1998 to 2000 in grants to states to improve their K-3 literacy instruction with the goal of teaching every child to read by the end of third grade (H.R. 2614, 1998). These efforts, which reflected growing concerns about flagging early literacy in the U.S., set the stage for the current wave of early literacy policies enacted nationally (National Early Literacy Panel, 2008). In 2002, the Department of Education authorized the Reading First program under No Child Left Behind, which replaced the expired Reading Excellence Act. This program similarly allocated funding to implement evidence-based reading instruction and hire literacy coaches to support K-3 reading achievement (U.S. Department of Education, 2015). These funds encouraged states to attend to early literacy, leading several states to implement their own policies alongside Reading First. On a broad level, these state policies prescribed interventions to support early literacy efforts and sometimes instituted retention policies under which third graders must score above a certain level on the state standardized literacy assessment in order to be promoted to fourth grade (CCSSO, 2019). We refer to these policies as retention- based third-grade literacy policies. Florida, which passed its Just Read, Florida! retention-based third-grade literacy policy in 2002, is largely considered the trailblazer of such policies (CCSSO, 2019). Florida’s policy includes several provisions designed to improve students’ literacy in grades K-3, including early identification of students who need additional supports, ongoing monitoring and communication with families, a range of literacy interventions, and third-grade retention for students who do not 15 meet a certain score on the state assessment. By 2021, 19 states had adopted retention-based third-grade literacy policies that contained several elements of Florida’s policy. While the evidence is clear on the importance of early literacy, it is mixed about the efficacy of retention. Research shows that students’ reading ability in the early grades— particularly by the end of third grade (e.g., Fiester & Smith, 2010)—is a strong predictor of later outcomes, including high school academic outcomes, graduation, and college attendance (Cunningham & Stanovich, 1997; Hernandez, 2011; Lesnick et al., 2010; Sparks et al., 2014). This research supports states’ efforts to create policies aimed at improving K-3 literacy outcomes. However, the sanction included in many of these policies—retention—is bolstered only by mixed evidence. Research on retention policies in Chicago, New York, and Florida that uses clearly defined retention criteria to create credible control groups has shown that these policies can improve students’ reading achievement in the short term (Greene & Winters, 2004, 2006, 2007; Jacob & Lefgren, 2004, 2009; Lorence, 2014; Lorence et al., 2002; Mariano & Martorell, 2013; Roderick et al., 2002; Roderick & Nagaoka, 2005; Schwerdt et al., 2017; Strunk et al., 2021). However, other research has shown that the positive achievement effects of these policies fade over time (Winters & Greene, 2012), or have no effect at all (Weiss et al., 2018). Despite this mixed evidence, retention-based third-grade literacy policies have continued to spread across states. In this paper, we examine how Florida’s third-grade literacy policy model transferred to other states, and in particular to Michigan. Data and Methods To help us understand what factors influenced the development and passage of Michigan’s Read by Grade Three Law, we rely on state-level stakeholder interviews (N=24) and early literacy policy documents from all 50 states and D.C. We conduct a qualitative analysis of 16 all interviews and policy documents following a coding scheme developed inductively and based in our MSF and policy transfer framework. Stakeholder Interviews We conducted semi-structured interviews in fall 2019 with 24 stakeholders involved in the development of the Law, including state legislators (n=11), MDE officials (n=5), and external stakeholders (n=8). External stakeholders included early literacy leaders (e.g., individuals working on the state’s Early Literacy Task Force), university researchers active in policy advocacy around early literacy in Michigan, leadership from state educational associations, and state employees not directly involved in education policy but who work with education data. We purposively sampled interviewees based on their involvement in the development of the Law, as our intent was not to produce a generalizable study, but rather to document how the Read by Grade Three Law specifically unfolded and why. In order to generate our sample, we began by creating a list of individuals who we knew were involved in the initial development and passage of the Law based on our knowledge of the policy landscape, reading of the legislation and associated hearings, and conversations with education policymakers. We then used snowball sampling to identify additional participants by asking interviewees whether they knew of anyone we should interview. We designed interview questions to elicit participants’ perceptions of and involvement in the development of the Read by Grade Three Law. Although we modified each interview protocol to reflect the unique role and experience of our respondents, they all followed the same general outline of topics. We first asked participants how they became involved in conversations about early literacy and in particular the Read by Grade Three Law. We then asked about their involvement in the early formation of the Law and whether and how they are currently involved 17 in its implementation. We further asked participants what they saw as the main goals of the Law, the factors that led to its formation, and their perceptions of the various interventions included in the Law (e.g., literacy coaches, retention) and how those ended up in the legislation. We concluded by asking interviewees to shared advance with policymakers in other states who may wish to create and implement similar legislation. We conducted these interviews in person or via Zoom, and each interview lasted between 45 and 60 min. We recorded the interviews, and a third party transcribed the recordings. Members of the research team subsequently vetted them for accuracy and deidentified them of any personally identifiable information. We categorized the 24 state-level stakeholders we interviewed into seven groups: representatives of educators’ associations (4 interviews, 17% of sample), early literacy leaders (5, 21%), staff from the former and current Governors’ offices (2, 8%), MDE (6, 25%), staff from other state agencies (2, 8%), and former and current legislators (6, 25%; see Table 1.1). Within the legislator group, we interviewed three Democrats and three Republicans, all but one of whom were directly involved in drafting the Read by Grade Three Law. Table 1.1 Interview Sample Characteristics Stakeholder Group Association Early Literacy Leader Governor’s Office Legislator MDE Other State Agencies Total Interviews Number 4 5 2 6 6 2 24 Percent 17% 21% 8% 25% 25% 8% 100% Given the retrospective nature of this part of the study, we necessarily interviewed 18 participants three years after the Law was passed. This may lead to concerns about the accuracy of participants’ memories. However, those we interviewed largely were still involved in the implementation of the Law, suggesting that the policy has remained top-of-mind since it was passed. They also provided a high level of detail about the policy process, giving us confidence that their accounts were accurate. Moreover, we heard similar accounts from multiple stakeholders, suggesting that their understanding of the policy process was consistent and reliable. Third-Grade Literacy Policy Documents To assess the extent to which the Read by Grade Three Law resembled other states’ third- grade literacy policies, we collected and coded all states’ early literacy policies. To compile these policies, we began by drawing from four existing datasets: (1) The National Conference of State Legislatures’ Third-Grade Reading Legislation (Weyer, 2019), (2) Education Commission of the States’ (ECS) Third-Grade Reading Policies (Workman, 2014), (3) the Council of Chief State School Officers’ Third-Grade Reading Laws: Implementation and Impact (CCSSO, 2019), and (4) ECS’ State Kindergarten-Through-Third-Grade Policies (ECS, 2018). To ensure that we captured the most recent policies for a comprehensive dataset, we supplemented this information by conducting a Google search for each state using the search phrase, “[STATE] third-grade literacy policy.” This led to the inclusion of 26 additional policy documents. When we were unclear whether a policy was related to early literacy, or whether a policy was still current, we reached out to state legislatures to confirm. Across all 50 states and D.C., we collected 167 early literacy policy documents. We excluded 26 because they were not relevant to the state’s third-grade literacy policy or were duplicative of policies found in other documents, giving us a final analytical sample of 141 19 documents. The number of documents per state ranged from zero (Montana, New Hampshire, Pennsylvania, South Dakota) to ten (Mississippi). Analytic Approach Stakeholder Interviews We analyzed all transcripts via Dedoose—a computer-assisted qualitative data analysis software—following an iterative process. First, the lead author read through all of the interview transcripts and created memos, including observations about the data to use in order to generate codes. Using these memos in combination with the existing literature, we created an initial codebook of 50 inductive and a priori codes, including definitions and examples for each code.6 We included parent codes (with child/grandchild codes nested within) for Policy Transfer (how other states’ and national policy efforts affected the development of the Law), Michigan Context (the role of the state’s unique context and circumstances), Preexisting Support for Literacy (to reflect the contribution of Michigan’s previous and ongoing efforts surrounding literacy), Research (how it was—and was not—used in the development of the Law), and Relationships between Groups (to reflect how stakeholder groups’ relationships with each other played a role). For a full list of codes, see Appendix 1.A. The lead author then discussed this codebook with a colleague who is not directly involved in, but familiar with, the project, to get feedback on its clarity and usefulness. After this, we piloted the coding scheme on one of the interview transcripts, which led to the addition of one inductive code—Sequence of Events—which we used to keep track of how interviewees 6 We organized the coding scheme into a hierarchy and employed automatic upcoding in Dedoose such that coding any child or grandchild codes would code the parent code(s) under which they were housed. An example of this is with the parent code Policy Transfer. Based on the memos we generated from an initial read of the data, we created two inductive child codes, National Literacy Policy and Other States’ Literacy Policies, to reflect the unique contribution of each. Further, we created inductive grandchild codes (e.g., under Other States’ Literacy Policies we created grandchild codes for Florida, Kentucky, Massachusetts, Minnesota, and Tennessee as these were the states interviewees mentioned in their discussion of the development of the Law). 20 described the order in which various events took place related to the development of the Law. This was added because it became clear that understanding interviewees’ recollection of the order in which events happened would be foundational to understanding the factors that led to the development of the Law. To establish reliability in the coding scheme, we conducted an interrater reliability (IRR) test between one member of our research team and the aforementioned colleague, which resulted in a Cohen’s Kappa (κ) of 0.76. Two codes were particularly problematic: Poor Literacy Performance and Urgency. After discussing these coding discrepancies with the colleague, we clarified our definitions and examples for each code to make it more explicit when to employ each. The same individuals then conducted a second IRR with a resulting κ=0.95, indicating very strong reliability in the coding scheme. The final codebook contained 51 codes, including 7 parent codes, 19 child codes, 24 grandchild codes, and 1 great-grandchild code. The lead author then coded all 24 transcripts. Another member of the research team also reviewed the interview transcripts as part of a separate analysis and identified similar themes, further supporting the reliability of the coding scheme. To analyze the interview data, we first used Dedoose to generate descriptive information related to code presence (whether a particular code was applied in a given interview), code application (how many times a code was applied in a given interview), and code co-occurrence (when two codes appeared together) overall and by subgroup. This allowed us to observe initial patterns in the data and determine what to explore further. From this initial analysis, we selected sets of coded excerpts for further analysis. We considered not only the frequency with which codes were applied, but also subgroup patterns (e.g., when codes were applied across all subgroups, or when a particular subgroup was the only group not to utilize a particular code). As 21 we read through and analyzed these selected data, we developed an analytic memo to track emergent themes, constantly comparing data (Glaser & Strauss, 1967) to that which was previously analyzed and seeking relationships between our various codes. We then read through and synthesized this memo to determine our overall findings. Throughout the findings, we incorporate direct, deidentified quotes from these interviews that are representative of larger themes. Policy Documents We also coded and analyzed the early literacy policy documents in Dedoose. We developed our coding scheme a priori based on our knowledge of third-grade literacy policies and the common elements included in them. We included parent codes for Assessment, Funding, Instruction/Tier I, Interventions, Tiers II and III, Parental Notification, Professional Development, Retention, and State Literacy Organization. These parent codes included child codes representing more detailed elements of the policy (e.g., specific interventions). In total, the codebook included 50 codes. For a full list of these codes, see Appendix 1.B. We coded each policy document for the presence of each of these 50 items, giving the document a 1 if it included the item and 0 otherwise. Because most states’ early literacy policies were comprised of more than one policy document, we then aggregated all of the policy documents from a given state to create indicators for whether the state included each item in its third-grade literacy policy. We resolved any discrepancies (e.g., if one policy document in a state indicated that retention was allowed while other indicated that it was required) by using the code from the most recent policy document. Findings In this section, we describe how events and conditions in each of the three streams (i.e., 22 problem, policy, and political) contributed to the passage of the Read by Grade Three law, and how policy transfer helps explain why Michigan’s policy—and those of many other states—so closely resembles Florida’s. We further describe how non-governmental policy entrepreneurs connected to Florida played a role in these processes, specifically through emulation and elite networking. Conditions in both the problem and political streams allowed for a policy window to open, and policy entrepreneurs had a proposal ready in the policy stream. After a period of softening up policymakers to the idea of a retention-based third-grade literacy policy, these policy entrepreneurs were able to successfully galvanize support for the passage of their preferred version of the Read by Grade Three Law in the Michigan legislature. We are also able to quantify the similarity between Michigan’s Read by Grade Three Law and Florida’s policy through our analysis of state policy documents. The high level of similarity between the two states’ policies—as well as third-grade literacy policies in other states—provides further support for the effectiveness of policy entrepreneurs in the policy transfer process. We trace these processes in Figure 1.1. 23 Figure 1.1 Read by Grade Three Law Policy Formation 24 Problem Stream The problem stream centered around Michigan’s declining literacy performance, particularly in relation to other states. Nineteen out of 24 interviewees (79%) cited this as one of the primary factors contributing to the need for a third-grade literacy policy in Michigan. As described above, Kingdon (1984) distinguishes between conditions and problems, where conditions become defines as problems through changes in indicators, focusing events, or feedback about the operation of existing programs. As shown in Figure 1.1, interviewees described a combination of these factors in elevating poor literacy performance from a condition to a pressing problem in Michigan. First, while poor literacy performance had been an ongoing condition in Michigan, as evidenced by the state’s consistently below-average fourth-grade reading scores on the National Assessment of Educational Progress (NAEP; The Nation’s Report Card, n.d.), one-third of our interviewees specifically referenced the 2015 NAEP scores as an indicator that early literacy was a problem that needed to be addressed. This was the second consecutive NAEP administration in which the state’s average fourth-grade reading score declined and was significantly lower than the national average (The Nation’s Report Card, n.d.). Indeed, problems sometimes involve comparisons, particularly when there is potential competition across governmental entities (Kingdon, 1984; D. Marsh & Sharman, 2009). Interviewees with whom we spoke often compared Michigan’s performance on NAEP to other states as a rationale for why early literacy needed to be addressed. Further, some went so far as to claim that Michigan’s performance was an urgent “crisis” in need of a fast solution. One official from the Governor’s office at the time of the Law’s passage commented: We were really alarmed by finding out that Michigan was going backwards on its NAEP 25 scores. We were really one of the few states actually going negative as you looked over the years, and that caused a lot of concern, which is what I think got this initiative started. Second, compounding Michigan’s poor performance on the NAEP, one legislator described how a court case in Highland Park (a school district in the Greater Detroit area), S.S. v. State of Michigan, acted as a focusing event that brought attention to poor literacy performance in the state. In S.S. v. State of Michigan, the American Civil Liberties Union (ACLU) of Michigan charged the state with failing to take effective measures to ensure that Highland Park students were reading at grade level after less than 10% of the district’s students in grades 3-8 scored proficient on the state assessment. Though the court ultimately ruled against the ACLU, as one legislator explained, “That event was a catalyst in the introduction of the original bill.” Lastly, legislator described Michigan’s several prior failed attempts at addressing literacy performance. Indeed, the Read by Grade Three Law was at least the state’s third attempt at making reading and literacy a priority in Michigan. In the late 1990s and early 2000s, the state implemented a plan that provided kits to parents when they left the hospital with a newborn that included books, parenting activities, and cassette tapes to promote reading. As one Republican legislator with whom we spoke said, “This was our attempt to try and get the emphasis on early childhood development.” Then, in 2005-2006, the federal government granted Michigan $80 million to improve reading. However, the same legislator felt that it was “an absolute waste of money” because “nothing ever happened from that [money].” These quotes highlight how legislators did not perceive either of these efforts as effective in improving the state’s scores and believed that they needed to approach the problem differently moving forward. This began to open a policy window for a solution like the Read by Grade Three Law. As the same legislator reinforced: 26 I think it’s the most serious thing that we’ve done. As I said, in the ‘90s, we did the reading plan for Michigan. Eighty-million dollars by the Feds in the mid 2000s to try and do—nothing. This is our, maybe not last-best hope, but it certainly, yeah, one that, maybe, you know? If it causes this much consternation, then clearly something’s working. Or, you know, it got people thinking? Political Stream Events and circumstances in the political stream further contributed to the opening of a policy window. At the time that the Read by Grade Three Law was passed, Republicans controlled both the House of Representatives and the State Senate, as well as the governor’s mansion. This created politically fertile ground for the passage of a retention-based third-grade literacy policy, as our analysis of early literacy policy documents shows that these policies have historically been the product of Republican-backed legislation. Despite political uniformity, the passage of the Law was also aided by an opportunity for political compromise. Indeed, Democratic legislators mentioned that they initially became involved in talks about third-grade literacy because they had a set of bills on educator evaluation that they wanted passed and agreed to work with the Republicans on the Read by Grade Three Law if the Republicans would reciprocate on their evaluation bills. As one Democratic legislator told us: [They] couldn’t get the votes [they] wanted, [they] needed, on [their] side of the committee to pass the third-grade reading law…Our ask was that we would help [the Republicans] on this, but we needed [their] help in passing a robust evaluation bill and process. We cut a deal…we were satisfied with the reading legislation. I think the politics may have gotten in the way of letting us get there, but the reason we were able to get [the 27 Republicans] the votes was because they were willing to give us the eval[uation] bill that we wanted. The amendments that we put in…those were really around the very specific pieces around the intervention, and the multiple methods of proving proficiency—that it wasn’t just the M-STEP. Similarly, Republicans needed votes from Democrats to pass the Read by Grade Three Law. As a result, Republicans were willing to compromise with Democrats to include elements for which Democrats advocated (i.e., alternatives), particularly “good cause exemptions” to mandatory retention for particular groups of students. Another Democratic legislator commented on this negotiation: “The exemptions were big. Exempting IEPs [students with disabilities], exempting second language learners. Those were not supported—those are not things that the advocates wanted…It’s a compromise. Legislation is making the sausage and it’s compromise.” Once these alternatives were included during caucus, the bill was more politically palatable for Democrats. So, despite some concerns with third-grade retention, many ultimately voted for the bill. The same Democratic legislator went on to say: When you look at a bill with tremendous resources and positives and assistance to those that need assistance and knowing that there were very few negatives and it most likely was not going to affect very many children, if any in the school district—in my district, it made sense to go with the bill. Ultimately, the Republican makeup of Michigan’s state government and the opportunity for legislative compromise—combined with events in the problem stream (i.e., NAEP, S.S. v. State of Michigan, negative feedback about previous efforts to improve literacy)—opened a policy window for the potential to pass a retention-based third-grade literacy policy in Michigan. 28 Policy Stream As Kingdon (1984) explains, the chance that an issue rises on the agenda increases dramatically if a solution can be attached to it. As described above, solutions are generated by policy communities—which in the case of the Read by Grade Three Law included members of various state-level stakeholder groups: association members, early literacy leaders, the Governor’s office, legislators, MDE, and other state agencies. While each of these groups agreed on the need to address early literacy in Michigan, they disagreed about the content of such legislation, particularly a retention component. As one Democratic legislator shared, “They [all] want kids to learn on track by the end of third grade. They [all] want it. [But] they disagree on how to implement it [i.e., the solution].” Nonetheless, before policymakers were even considering the Read by Grade Three Law, early literacy efforts in Michigan outside of a formal policy context were converging on elements that could be included in an eventual third-grade literacy policy. These efforts largely revolved around literacy professional development, including literacy coaching. Though literacy coaches would ultimately become part of the Read by Grade Three Law, several Intermediate School Districts (ISDs) had already hired literacy coaches before the Law was passed.7 The state also allocated ISDs $37,500 that they could apply for and match to create a salary for a literacy coach in fall 2015, a year before the Law was passed. These ISD Early Literacy Coaches spawned the creation of the Michigan Association of Intermediate School Administrators (MAISA) General Education Leadership Network’s Early Literacy Task Force (ELTF). As early literacy leaders explained to us, after the state allocated money for literacy coaches, MAISA decided to create 7 In Michigan, ISDs, which are sometimes called Regional Educational Service Agencies (RESAs), are educational entities that operate between the Michigan Department of Education and local education agencies. ISDs often serve the local education agencies within a given county. Local education agencies can receive a range of services through their ISD. 29 universal job descriptions for these coaches, as opposed to having each ISD create their own: [MAISA] convened the first meeting of the Early Literacy Task Force in December of 2015. We came together—that very first meeting was really all about creating job descriptions for the coaches that we had received funding for that came out of the governor’s Workgroup before the Read by Grade Three legislation was even passed. The ELTF originally formed to draft these job descriptions, and the organization went on to create several practice guides that served as the foundation for teacher professional development called the Literacy Essentials. Eventually, the ISD Early Literacy Coaches were folded into the Read by Grade Three Law, but their prior existence and state support laid the groundwork for including them. Indeed, 17 out of 24 interviewees (71%) recognized the role of the set of preexisting supports for literacy and took care to emphasize that they were happening before the actual passage of the Law. Early literacy efforts at MDE also predated the Read by Grade Three Law and contributed to the inclusion of various elements in the Law. MDE staff mentioned that “literacy is part of what we talk about at the Department of Education all the time. We already talk about literacy.” They described multiple grants they supported related to early literacy before the Law, including for literacy coaching, additional instructional time in literacy, and literacy assessments. Education association officials also mentioned that they had “been involved in improving literacy in Michigan for 5, 6 years” and received state grants to support teacher professional development in literacy. All of these prior efforts had become widely accepted solutions to addressing early literacy in the state and facilitated their inclusion in the Read by Grade Three Law. However, interviewees explained that the retention component of the Law was much 30 more contentious. Nonetheless, the opening of a policy window allowed key policy entrepreneurs to push forward a retention-based third-grade literacy policy. In particular, they advocated for Michigan to pass a policy similar to Florida’s. We found a clear partisan split in whether interviewees favored retention, with Republican legislators the only stakeholder group to fully support its inclusion in the Law. From their view, retention became the “teeth” that was lacking in previously failed efforts to address early literacy. As one Republican legislator with whom we spoke said, “Right now, the teeth is in the threat of retention.” They went on to say this approach was needed because “schools will not do anything unless there’s some punitive measure.” No other stakeholder group favored including retention in the Read by Grade Three Law and nearly all groups except Republican policymakers mentioned that research showed that retention is not effective in improving reading achievement in the long term, particularly in reference to Florida’s policy. On the other hand, Republicans’ reference to the efficacy of Florida’s policy focused on the short-term positive outcomes the state experienced after passing it, which they used as a rationale for passing similar legislation in Michigan. Interviewees largely attributed the Law’s passage with a retention component to the success of key policy entrepreneurs in aiding the policy transfer of such a law. As Kingdon (1984) explains, policy entrepreneurs work to “soften up” their preferred solution over time, and when an administration favors their solution, it is likely to be enacted. Interviewees traced this softening up process back to 2013 when State Representative Amanda Price (R) introduced a bill that was, according to one Democratic legislator, “basically a straight retention, no professional development, or prescripted intervention requirements or anything like that.” Though the lack of additional supports made the bill politically infeasible, this same legislator explained that Representative Price was working with ExcelinEd at the time, a group that the GLEP—a think 31 tank founded by the DeVos family—brought in. Together, ExcelinEd and GLEP would serve as key policy entrepreneurs in the passage of the Read by Grade Three Law. ExcelinEd is a nonprofit education organization based in Florida and chaired by former Republican Governor Jeb Bush, who oversaw the passage of the state’s retention-based third- grade literacy policy in 2002. Since ExcelinEd launched in 2008, the organization has been a prominent advocate of passing retention-based third-grade literacy policies across the country by providing policy and communications resources to policymakers, hosting early literacy convenings, and conducting and disseminating research on early literacy policies (ExcelinEd, 2022b). ExcelinEd even offers a ready-made policy template that provides language policymakers can use when introducing early literacy legislation (ExcelinEd, 2020). Nearly a third of interviewees across multiple stakeholder groups mentioned the organization’s influence in the development of Michigan’s Read by Grade Three Law. As one interviewee explained: I would say that at least half of [the different elements of the Law] landed in there because of…ExcelinEd, and the work they did in Florida because that was the team that was consulted, and the boilerplate language, I think, came from them. Our legislators are very involved with that group, so I know that’s where it came from. Two interviewees also viewed the legislation as being “strongly influenced” by GLEP. One Democratic legislator explained to us that GLEP was the “biggest advocate for the bill” and worked with Republican legislators to help bring ExcelinEd to Michigan. In addition, GLEP funded a great deal of the advocacy work to pass the Read by Grade Three Law. According to one external stakeholder: The [GLEP] have a lot of money, and they were able to leverage their dollars into policy…again, how do you afford to bring in legislators and advocates and others from 32 the state of Florida? You have to have money to be able to do that…They used their money towards what they believed. Conjointly, these policy entrepreneurs facilitated the policy transfer of Florida’s retention-based third-grade literacy law to Michigan. However, because ExcelinEd and GLEP are non-governmental agents (i.e., they cannot themselves pass legislation), they had to promote the “voluntary transfer” of this solution (Dolowitz & Marsh, 1996)—a challenge given that no stakeholder groups except Republican legislators supported a retention-based policy. To this end, we found that they relied on the tactics of emulation and elite networking (Bennett, 1991). Elite networking takes place when agents share their expertise to shape others’ understanding of a policy (Bennett, 1991). According to our interviewees, this process began when GLEP connected Representative Price and other Michigan legislators with ExcelinEd. One Democratic legislator told us that ExcelinEd worked with Michigan legislators on other issues, saying, “They came and talked with us many times about a lot of things, but this being one of them.” This suggests that previous relationships between the two parties may have made them a trusted partner in these new third-grade literacy policy efforts. ExcelinEd continued to engage in elite networking in a Third-Grade Reading Workgroup that former Governor Rick Snyder assembled. Governor Snyder called for the creation of the Workgroup in his 2015 State of the State Address, tasking it with making policy recommendations for how Michigan should address its poor literacy performance. The Workgroup was led by a businessperson and consisted of three Republican and three Democratic legislators—all chairs of the various education committees—as well as the State Budget Office director. Half of all interviewees across all stakeholder groups discussed the Workgroup and explained to us that early on, they identified the literacy work that ExcelinEd was doing in 33 Florida and used that as a model for their recommendations. The Workgroup members we interviewed explained that they brought in numerous individuals and groups working on early literacy “on both sides” of the issue, including MDE staff, researchers, early literacy organizations, curriculum specialists, reading interventionists, and district administrators—as well as ExcelinEd. Ultimately, one of the Workgroup’s recommendations was a “smart promotion policy” in which students were behind in reading by the end of third grade would continue to receive literacy instruction and interventions at the third-grade level while moving on to the next grade level in any subjects in which they demonstrated proficiency (Kennedy et al., 2015). In other words, they recommended a retention- based third-grade literacy policy—ExcelinEd’s preferred solution (see ExcelinEd, 2020). Interviewees explained to us that ExcelinEd remained involved throughout the legislative process after legislators formally introduced the Read by Grade Three bill. At the time of its introduction, many legislators—particularly Democrats—had still not softened up to the idea of a retention-based policy. As one Republican legislator said: I think there were a lot of legislators that were uncomfortable to hear the way [retention] was first—like a pretty draconian approach at first which is gonna capture hundreds to thousands of kids…they didn’t feel comfortable with that. But we’re able to work the system that way, people coming back to the caucus and saying, ‘I really don’t feel comfortable with this part.’ We were able to soften it up in a number of cases. The stakeholders we interviewed further explained that ExcelinEd provided testimony about the Read by Grade Three Law throughout the legislative process. Across stakeholder groups, interviewees agreed that including retention created a sense of urgency around early literacy and “provide[d] some fire under the education community.” Even Democratic legislators 34 acknowledged that sometimes policy needed to include high-stakes incentives. One Democratic legislator told us, “Unless you have a big stick, you don’t get their attention. That’s painful to say, because that’s not the way anything should work.” The Republican legislators we interviewed also mentioned that if districts ultimately decided not to retain any students, that would be okay with them. Notwithstanding, they felt that it was necessary to include retention in the Law to create a “distant threat.” As one Republican legislator said: We never in that discussion set out to say, we think that retention is a useful intervention for getting kids to read by third grade. That wasn’t really the intention. It was more about having this distant threat out there that, if you guys don’t get serious and have this conversation, this is what happens. This suggests that Democratic legislators became more open to the idea of including retention in the state’s third-grade literacy policy as ExcelinEd continued to advocate for it. However, they also explained that ExcelinEd’s outsized influence overshadowed the influence of other groups like MDE and literacy experts. As one early literacy expert reinforced: The testimony from ExcelinEd was…We need this law, because it’ll help you get better outcomes, like Florida…We had some voice but not a lot…We didn’t agree with retention…[But] different [policy] players just weren’t interested necessarily in how [MDE and others] had been approaching this, I think. We weren’t strongly at the table. Likely, these differences amongst stakeholder influences were because of former networking ties ExcelinEd built over time in Michigan, particularly with Republican policymakers. ExcelinEd also engaged in emulation, the process of borrowing ideas and adapting policies to local conditions (Bennett, 1991). Specifically, they facilitated the policy transfer of 35 Florida’s retention-based third-grade literacy policy. Michigan’s Read by Grade Three Law contains similar language and interventions to Florida’s policy, and 15 of the 24 stakeholders we interviewed (66%, across all groups) described Florida’s policy as influential on Michigan’s Read by Grade Three Law, with Florida mentioned a total of 36 times. Our analysis of state policy documents further reveals how effectively policy entrepreneurs engaged in this emulation and policy transfer process. As one stakeholder said, the Read by Grade Three Law ended up “almost a copy of the Florida law” with minor tweaks. Policy documents revealed that the two states’ third-grade literacy policies only differed on nine out of 50 areas (18%). Further, the areas on which they differed were elements that very few states included in their policies. For example, Florida’s policy includes school/district reading plans; the provision of an alternative/transitional instructional setting, online or computer-based instruction, and smaller classes for students identified as needing additional literacy supports; teacher certification requirements; and the creation of a state literacy organization. Meanwhile, Michigan’s policy includes school/district literacy leadership teams and exemptions to retention for students who are new to their school/district or based on parental appeal. Thus, Michigan’s and Florida’s policies were identical on all central components and different slightly on marginal details. We argue that these differences reflect ExcelinEd’s effectiveness in emulating Florida’s policy and transferring it and adapting it to something that could work in Michigan. However, Florida’s influence is not unique to Michigan’s Read by Grade Three Law. Forty-six states plus D.C. (92%) had at least one policy document related to third-grade literacy. Further, policies in 18 states and D.C. (37%) include a required retention component, with another 9 (18%) allowing for retention. Analogous to Florida, almost all of these policies (91%) include diagnostic and/or progress monitoring assessments to identify students who need 36 additional literacy supports and interventions for these students (i.e., Tier 2 and 3 supports; 91%). The most common interventions include additional instructional time in literacy, evidence-based literacy interventions, home reading programs that include resources for families, and summer programs—all of which were included in Florida’s law and are now included in more than half of states’ third-grade literacy policies (including Michigan’s). Additionally, 70% specify general instructional (i.e., Tier 1) requirements, most commonly evidence-based reading instruction (62%), with nearly half specifically mentioning the “big five” components of reading: phonemic awareness, phonics, fluency, vocabulary, and comprehension. Finally, 64% require that schools and districts notify families when a student is identified as being behind in reading, 55% include professional development, and 21% include literacy coaches—all of which are included in Florida’s policy. This provides evidence that ExcelinEd’s policy entrepreneurship may not be unique to Michigan and has instead extended to many states throughout the U.S. Discussion and Conclusion Evidence from our interviews shows that a policy window for the passage of the Read by Grade Three Law opened in the problem and political streams. This window was then further opened by the national climate regarding early literacy and the transference of retention-based third-grade literacy policies across states via policy entrepreneurs. These findings are supported by our analysis of state early literacy policy documents which show that elements of Florida’s policy were successfully transferred to Michigan and elsewhere. In the problem stream, Michigan’s poor literacy performance was elevated from a condition to a pressing problem by declining NAEP scores, the Highland Park court case highlighting low proficiency levels on state assessments, and negative feedback regarding previously failed attempts at addressing early literacy in the state. Events in the political stream 37 further contributed to the opening of a policy window, including a Republican-dominated state government that had proven amenable to passing early literacy policies similar to other states. In 2015, then-Governor Rick Snyder called for a Third-Grade Reading Workgroup to propose policy recommendations for addressing the state’s early literacy crisis, bringing further attention to the issue. Meanwhile, Democratic legislators saw an opportunity to compromise with Republicans on the Read by Grade Three Law in exchange for legislation on educator evaluation. When this policy window opened in the problem and political streams, policy entrepreneurs (ExcelinEd and GLEP) ensured that one solution—a retention-based third-grade literacy policy modeled after Florida’s—made it to the forefront of the legislative agenda. While various components of such a policy, particularly surrounding literacy professional development, had already been advanced by members of the broader policy community including MDE and educators’ associations, the retention component was softened up over time due to these two key policy entrepreneurs. Despite the mixed research base about the effectiveness of retention, ExcelinEd and GLEP were able to use emulation and elite networking to frame a retention-based third-grade literacy policy as successfully tested in another state—Florida—highlighting how policy can transfer from one context to another. Our analysis of state early literacy policy documents further documents the widespread transfer of several components of Florida’s policy across the 50 states and the similarity between Florida’s and Michigan’s policies. In this way, the passage of the Read by Grade Three Law was the product of developments in independent streams that were joined together by highly influential policy entrepreneurs pushing their preferred alternative out of Florida that had already successfully been transferred to several other states. 38 By tracing the evolution and development of the Read by Grade Three Law, this study highlights several key elements of the policy process, particularly how certain conditions come to be defined as a pressing problem in need of a solution. At the same time, it showcases how the theory of policy transfer can extend our understanding of Kingdon's (1984) original conception of MSF. As the political and policy contexts both nationally and locally become increasingly polarized (Fiorina & Abrams, 2008; Heltzel & Laurin, 2020; Hopkins & Sides, 2015), it is critical to understand why and how contentious reforms move through the policy process to become law. In the case of Michigan’s Read by Grade Three Law, the policy, problem, and political streams joined to surface the need for “something new” that could build on extant efforts already underway in the state. On one hand, the political realities of both parties needing the other to pass their own unrelated reforms enabled a “softening up” of the policy space such that there was an entry point for compromise. On the other hand, policy entrepreneurs played an outsized role in transferring a controversial policy that had been implemented in one state—to mixed reviews—to another. By being ready with a solution and with evidence—however mixed—from Florida, ExcelinEd and GLEP were able to successfully take key elements of the Florida policy and institute them in Michigan’s. Indeed, how the Read by Grade Three Law evolved over time has important implications for its implementation. Given several key stakeholders’ lack of buy-in to the most controversial element of the reform (i.e., retention), it should be no surprise that educators and district leaders remain critical of the overarching policy. Moreover, because Michigan could not or would not fund the non-retention components (particularly coaching) of the Law to the same level as Florida, key intermediate outcomes that should lead to the Law’s eventual success in improving early literacy in Michigan have not come to fruition (see Strunk et al., 2021, for a full 39 discussion). Thus, this study highlights that the transfer of a policy from one state to another relies on a deeper understanding of the local context and willingness from all parties to provide capacity-building and funding to ensure policy success. This may be difficult to accomplish when policy adoption relies on non-governmental policy entrepreneurs who are unfamiliar with the local state context. 40 PAPER 2: BEYOND THE CLASSROOM? THE IMPLEMENTATION OF FAMILY ENGAGEMENT UNDER MICHIGAN’S READ BY GRADE THREE LAW Amy Cummings, Michigan State University Literacy proficiency, essential for academic success, is alarmingly low across the United States. Recent data from the National Assessment of Educational Progress (NAEP), a leading indicator of educational achievement, reveals that only a third of 4th and 8th graders demonstrate proficiency in reading (NAEP, 2022a). These literacy challenges have profound implications: Students who struggle with reading early on face higher risks of dropout, mental health issues, unemployment, and incarceration (Cunningham & Stanovich, 1997; Daniel et al., 2006; Fiester, 2013; Fiester & Smith, 2010; Hernandez, 2011; Sparks et al., 2014). The COVID-19 pandemic has only intensified these challenges, with significant declines in performance among young readers. Specifically, average scores for nine-year-olds dropped five points between 2020 and 2022—the steepest drop in three decades (NAEP, 2022b). In response to these trends, policymakers are increasingly focused on improving early literacy. Recognizing that reading proficiency by the end of 3rd grade is a critical predictor of future academic success (Cunningham & Stanovich, 1997; Fiester & Smith, 2010; Hernandez, 2011; Sparks et al., 2014), 48 states and the District of Columbia have implemented policies to enhance literacy in grades K-3 (ExcelinEd, 2022a). These policies are not uniform across states, but they typically share common elements such as teacher support, ongoing progress monitoring, and evidence-based instruction and interventions (ExcelinEd, 2022a). One of the most common aspects of states’ early literacy policies is family engagement, 41 which has a recognized positive relationship with children’s literacy outcomes (Aikins & Barbarin, 2008; Boonk et al., 2018; Castro et al., 2015; Evans et al., 2000; Fan & Chen, 2001; Georgiou et al., 2021; Inoue et al., 2018; Lin, 2003; Ma et al., 2016; Miedel & Reynolds, 1999; Sénéchal, 2006; Voorhis et al., 2013; Wilder, 2014). Twenty-seven states have mandated family engagement as part of their early literacy policies, requiring “Read at Home” plans that encompass caregiver training and emphasize regular home reading for students struggling with literacy (ExcelinEd, 2022a). Despite the popularity of these plans, their implementation remains poorly understood. Michigan’s Read by Grade Three Law, which is representative of early literacy policies in other states, mandates “Read at Home” plans for 3rd graders identified with “reading deficiencies” and recommends them for identified K-2 students (Michigan Public Act 306, 2016).8 This law thus offers an informative case for studying the implementation of “Read at Home” plans, including variations in how districts apply the plans, the characteristics of students receiving them, and potential factors driving their implementation, such as educators’ perspectives and other family engagement methods employed. To systematically explore these aspects, this study employs a conceptual framework that integrates Bronfenbrenner’s (1979) ecological systems theory with various policy implementation perspectives, such as policy instruments (Bardach, 1979; Gormley, 1987; Linder & Peters, 1989; McDonnell & Elmore, 1987; Salamon, 1981, 1989; Schneider & Ingram 1990), sensemaking (Spillane, 2000; Spillane & Anderson, 2019; Spillane et al., 2002a, 2002b; Weick, 1995), filtering (Diehl & Golann, 2023), and street-level bureaucracy (Lipsky, 1980; Meyers & Nielsen, 2012). These perspectives highlight different elements of policy implementation, such as how the design of the policy itself 8 I use the term “reading deficiency” because that is how the Read by Grade Three Law refers to students whose districts have identified them as needing additional support in literacy (Michigan Public Act 306, 2016). 42 is expected to influence implementation, how educators understand and perceive policies within their own organizational context and capacities, and how these factors ultimately influence the extent to which policies are implemented on the ground. I use this framework to examine how dynamics at different layers of the educational system—from the policy landscape down to the district, school, teacher, and individual student and family levels—are related the implementation of “Read at Home” plans. This study aims to address several gaps in our understanding of “Read at Home” plans. First, analyzing how plans are implemented across districts can reveal disparities and challenges in enacting the policy. Second, investigating which students receive plans can highlight the equity implications of current implementation patterns and identify targeted demographics. Third, exploring educators’ understanding and perceptions of the plans can identify potential informational gaps and shortcomings in the policy. Finally, examining additional literacy engagement strategies can provide a fuller picture of family engagement efforts and the context in which educators implement these plans. By addressing these gaps, this paper aims to inform the development and implementation of effective family engagement strategies for students needing literacy support. Specifically, it poses the following research questions (RQs): 1. What proportion of students identified as “reading deficient” are given “Read at Home” plans by their districts? How does this vary by district demographic and geographic characteristics? 2. Within districts that provide some, but not all, “reading deficient” students with “Read at Home” plans, what factors distinguish the students who are given plans from those who are not? 3. How do educators understand and perceive “Read at Home” plans? How does this 43 vary across educators, and how are understanding and perceptions related to implementation? 4. Beyond “Read at Home” plans, in what other ways do Michigan educators report engaging with families in promoting literacy? How does this vary across educators? This study draws on administrative data from the Michigan Department of Education (MDE) and educator surveys spanning the 2019-20 to 2022-23 school years, using descriptive statistics and regression analyses to answer these RQs. The results reveal that “Read at Home” plan implementation in Michigan is notably low, with districts providing these plans to only about one-fifth of eligible K-3 students, including just 21.3% of 3rd graders who are mandated to receive them. This low implementation rate may be attributed to the fact that the Read by Grade Three Law mandates these plans without providing capacity-building supports or explicit sanctions for non-implementation, highlighting the challenges of effectively executing policies that lack the necessary resources, training, and incentives. Furthermore, while districts with higher proportions of “reading deficient” students are more likely to provide “Read at Home” plans, they often do not extend them to all eligible students. This pattern suggests potential challenges in fully serving this population. In these districts, it appears that some selection process may be occurring in determining who receives a plan, with plans more frequently allocated to students from historically underserved backgrounds—a group the literature identifies as particularly benefiting from family engagement interventions. Educators’ sensemaking about “Read at Home” plans at different levels of the educational system is also related to their implementation. Specifically, teachers’ understanding of the plans and principals’ beliefs in their effectiveness are both significantly positively related to whether a student receives a plan. Additionally, teachers show a greater propensity to engage 44 in other family literacy activities beyond “Read at Home” plans, such as providing families with literacy materials and holding regular meetings about students’ literacy progress. This underscores that sensemaking occurs within the context of implementers’ established practices and routines and suggests that these plans may not seamlessly fit into many educators’ existing family engagement strategies. These findings highlight the need for capacity-building supports to enhance the implementation of “Read at Home” plans, particularly in districts with high proportions of students identified as “reading deficient.” Providing templates for the plans and offering training for educators could improve their understanding and perceptions of the plans. Additionally, involving teachers in the development of these plans could be beneficial, as it would honor their professional judgment while helping them integrate “Read at Home” plans into their existing family engagement practices. Given that 26 other states have adopted similar initiatives, these insights can offer valuable guidance for policymakers and practitioners in Michigan and beyond to evaluate and enhance the implementation of family engagement policies. The remainder of this paper is structured as follows. The next section provides an overview of “Read at Home” plans and how they are incorporated into states’ early literacy policies, particularly Michigan’s Read by Grade Three Law. This is followed by a literature review on early literacy policies and the potential of family engagement initiatives to improve students’ literacy outcomes. I then present a conceptual framework to guide the study, followed by a description of the data and methods used to address the RQs. Finally, I present the findings, followed by discussion and conclusion sections that highlight the implications of these findings for policymakers and practitioners. 45 “Read at Home” Plans in Early Literacy Policies “Read at Home” plans are a common component of early literacy policies across the U.S. Currently, 27 states, including Michigan, incorporate these plans into their strategies to enhance early literacy (ExcelinEd, 2022a). ExcelinEd, a nonpartisan education policy organization founded by former Florida Governor Jeb Bush—who oversaw the passage of that state’s early literacy policy—advocates for and tracks the adoption of early literacy policies across states. The organization defines a “Read at Home” plan as a plan “for students identified with a reading deficiency, supplemented by a list of vetted online resource hubs for all parents to support literacy at home” (ExcelinEd, 2022a). This definition exemplifies the vague language found in states’ early literacy policies, which often specify who should receive a plan but lack detail about its content. Michigan’s Read by Grade Three Law outlines a specific process for identifying which students should be given a “Read at Home” plan, outlined in Figure 2.1 (Michigan Public Act 306, 2016). The law first requires MDE to approve a list of screening assessments, which districts select from and use to evaluate K-3 students’ literacy abilities multiple times a year. Students scoring below their district’s self-defined threshold on these assessments are identified as having a “reading deficiency,” entitling them to various supports, including “Read at Home” plans. While the law encourages districts to provide these plans to K-2 students with “reading deficiencies,” it mandates them for 3rd graders identified as such (Michigan Public Act 306, 2016). Like other states’ early literacy policies, the Read by Grade Three Law leaves the content of “Read at Home” plans up to individual districts without offering specific guidelines—only that the plans should include “parent, guardian, or care provider training workshops and regular home reading” (Michigan Public Act 306, 2016). 46 Figure 2.1 Administration of “Read at Home” Plans Under the Read by Grade Three Law Note. Developed based on requirements in Michigan Public Act 306 (2016). The Read by Grade Three Law mandates these plans without providing capacity-building supports for educators to help them create or implement the plans, such as additional funding or training. Moreover, the policy does not include any explicit sanctions for districts that fail to implement the plans, even though districts are required to report to the state whether “reading deficient” students were given a “Read at Home” plan. Additionally, MDE does not currently supply guidance or templates for the plans. While the state’s Essential Instructional Practices in Early Literacy: Grades K-3⎯which provide a foundation for literacy instruction and professional development across the state⎯include an essential practice titled “Collaboration with families, caregivers, and the community in promoting literacy” (Michigan Association of 47 Intermediate School Administrators General Education Leadership Network Early Literacy Task Force, 2016), less than 20% of teachers reported receiving professional development in this area by the 2021-22 school year (Utter et al., 2023). Consequently, although the law specifies who should receive a plan, educators receive limited guidance or support in terms of funding, resources, or training on how to create or implement these plans, and the policy itself does not include explicit sanctions for non-implementation, potentially limiting the incentives for educators to prioritize them. Examining Michigan Districts’ “Read at Home” Plans Many districts in Michigan have made their “Read at Home” plans available online, allowing for an examination of their structure and content. Although a systematic analysis of the plans is beyond the scope of the present study, I performed a targeted Internet search to gain a general sense of districts’ current “Read at Home” plans. This approach was intended to help understand the potential effectiveness of the plans and to contextualize the findings of this paper. Specifically, I conducted a Google search using the term “Read at Home Plan Michigan” and reviewed the first 30 results that included a plan (see Appendix 2.A for a list of included districts). I chose to stop at this point because the plans began to exhibit repetitive structures and content, indicating I had reached a point of saturation (Glaser & Strauss, 1967). In my review, I documented each plan’s structure and content, identifying the different sections and analyzing the language and recommendations within each section. For example, in sections recommending activities for families, I tracked whether these activities addressed the “Big 5” components of reading: phonemic awareness, phonics, fluency, vocabulary, and comprehension. I also noted whether the recommendations were grade-level specific or generalized to all K-3 families. This approach has its limitations. Since it is not a systematic review, the findings may not 48 represent all Michigan districts’ “Read at Home” plans. The group of districts whose plans were reviewed is likely biased towards districts with a stronger online presence, as they may be more likely to post their plans online. Additionally, districts that implement the plans or feel confident that their plans meet the Read by Grade Three Law’s requirements may be more inclined to post them. Google’s search process may have also biased the plans included in the review. Google presents the most relevant and useful results first, considering factors such as the intent of the search, the relevance and quality of the content, and the usability of the webpage. The search results are thus likely biased towards districts with more resources to develop high-quality, search-optimized content (Google, n.d.). Indeed, the 30 districts in Appendix 2.A reflect some differences compared to the statewide district population. They are generally higher-achieving, with 3rd-grade English Language Arts (ELA) achievement 0.04 standard deviations above the state mean and have lower proportions of non-White (28.47% vs. 36.38%) and economically disadvantaged students (52.69% vs. 64.89%). However, they have similar proportions of English learners (ELs; 6.77% vs. 5.16%), students with disabilities (21.63% vs. 20.70%), and students identified as “reading deficient” (37.97% vs. 38.25%). The districts whose plans I reviewed are much larger than the average district in terms of K-3 enrollment (1,232 vs. 514) and are more suburban (66.67% vs. 40.55%), less urban (10% vs. 20.28%), and less rural (23.33% vs. 33.09%) than districts statewide. Notably, none of the districts are charter schools, while 42.3% of districts in the state are charters. These districts also reflect some geographic variation, with many located in the densely populated southeastern region of Michigan near Detroit, several in western Michigan near Grand Rapids and along the Lake Michigan shoreline, some in the Lansing area and southwestern Michigan near Kalamazoo and Benton Harbor, and a few in the Thumb region of 49 eastern Michigan and more central parts of the state. However, none are located in Michigan’s Upper Peninsula. Although the districts whose plans I reviewed are not entirely representative of the state, the findings from this search provide valuable context about “Read at Home” plans for the present study. Without specific requirements from the Read by Grade Three Law or a template from MDE, districts’ plans resemble one another, suggesting an emulation effect across the state. The vast majority of the plans (24 of the 30) included information about the Read by Grade Three Law and began with a letter to families (21) explaining why they were receiving the plan. Nearly every plan (27) included a list of activities for families to engage in with their children to promote literacy at home, with most (23) emphasizing the “Big 5” components of reading. Other common elements included links to literacy websites providing additional resources (20), statistics about the long-term benefits of reading (16), and the promotion of reading at least 20 minutes a day (22). However, less than half of the plans (12) provided information in a language other than English or included a form for families to acknowledge receipt of the plan (5). A typical example of a “Read at Home” plan I reviewed is in Appendix Figure A2.1. The “Read at Home” plans I reviewed thus demonstrate a commitment to engaging families in their children’s literacy development by providing comprehensive strategies and resources. Every plan fulfilled the Read by Grade Three Law’s mandate of promoting regular home reading, either by encouraging families to read with their child at least 20 minutes a day, highlighting the long-term benefits of reading, or including specific activities for families to promote reading. However, these plans also highlight areas for enhancement. Although they offer a variety of activities for family literacy engagement, there is a noticeable absence of information about the caregiver training workshops mandated by the law. Additionally, most 50 plans are not accessible to families where the primary home language is not English, making it challenging to support ELs who have been identified with “reading deficiencies.” Overall, districts seem to have designed the plans for immediate distribution, creating a one-way flow of information to families. This approach may neglect the varied needs and circumstances of different students and their families. These insights prompt a closer examination of the family engagement literature to evaluate the potential effectiveness of the strategies “Read at Home” plans employ. Literature Review A substantial body of research supports incorporating family engagement into early literacy policies like Michigan’s Read by Grade Three Law. However, a closer examination reveals potential gaps between evidence-based practices for family engagement and the current requirements outlined for “Read at Home” plans. This literature review synthesizes research on key elements of effective family engagement interventions, including caregiver training, facilitating two-way communication between schools and families, and tailoring approaches to meet diverse family needs and contexts. By highlighting discrepancies between these research- backed strategies and the design of “Read at Home” plans, this section underscores the importance of the present study to identify areas for improvement. Extensive evidence consistently links family engagement to positive outcomes in children’s social, emotional, and academic development (Albright & Weissberg, 2009; Bowen & Bowen, 1998; Christenson et al., 1992, 2007; Emerson et al., 2012; Englund et al., 2004; Galindo & Sheldon, 2012; Hall, 2020; Mart et al., 2011; Mo & Singh, 2008; Patrikakou & Weissberg, 2007; Simons-Morton & Chen, 2009; Stevenson & Baker, 1987; Zellman & Waterman, 1998). Specifically for literacy development, family engagement activities like shared reading, where 51 children and caregivers read books together and discuss the stories, as well as caregiver-guided literacy instruction through activities like writing and alphabet games, are linked to stronger skills (Aikins & Barbarin, 2008; Boonk et al., 2018; Castro et al., 2015; Evans et al., 2000; Fan & Chen, 2001; Galindo & Sheldon, 2012; Georgiou et al., 2021; Inoue et al., 2018; Lin, 2003; Ma et al., 2016; Sénéchal, 2006; Voorhis et al., 2013; Wilder, 2014). Illustrating this, Sénéchal’s (2006) seminal study traced the development of literacy skills from kindergarten through 4th grade, uncovering direct links between caregiver teaching about literacy in kindergarten and crucial early literacy skills like alphabet knowledge and fluency. Furthermore, it demonstrated that storybook exposure in kindergarten predicted reading comprehension, vocabulary acquisition, and reading enjoyment in 4th grade, underscoring the importance of early literacy experiences at home on later outcomes (Sénéchal, 2006). Family engagement may be especially beneficial for students from historically underserved backgrounds, a point of particular relevance for Michigan’s Read by Grade Three Law. Under this policy, students of color, those from economically disadvantaged backgrounds, ELs, and students with disabilities are disproportionately identified as having “reading deficiencies” that qualify them for “Read at Home” plans (Strunk et al., 2022). Numerous studies, including meta-analyses, confirm significant positive associations between family engagement initiatives and academic achievement for these student groups, often highlighting shared reading as particularly beneficial (Brown et al., 2019; Fikrat-Wevers et al., 2021; He & Thompson, 2022; Jeynes, 2003, 2012, 2016, 2017; Sheridan et al., 2011). While “Read at Home” plans incorporate research-supported literacy activities like shared reading, the literature indicates that effective family engagement requires more than simply directing caregivers in what activities to do; training for caregivers on how to implement 52 these activities is crucial. The Read by Grade Three Law mandates the plans to include caregiver training workshops (Michigan Public Act 306, 2016), but the district plans I reviewed show no clear documentation of such training. This does not necessarily mean training is not happening, as districts may communicate about it through other channels. Nonetheless, an overwhelming body of research emphasizes caregiver training as vital for successful family literacy interventions (Çaliskan & Ulas, 2022; Crosby et al., 2015; Darling & Westberg, 2004; McElvany & van Steensel, 2009; Petersen-Brown et al., 2023; Reutzel et al., 2006; Sénéchal & Young, 2008; van Steensel et al., 2011). Meta-analyses confirm that training caregivers on specific literacy strategies like shared reading techniques yields significantly larger literacy gains in children compared to simply encouraging them to listen to their children read (Darling & Westberg, 2004; Sénéchal & Young, 2008). Moreover, Petersen-Brown et al. (2023) find that the length of training does not directly impact the intervention’s success, suggesting customizing training to fit caregivers’ skills and needs is more effective than adhering to a predetermined duration. Caregiver training also plays a critical role in the extent to which families actually engage in and implement the literacy activities as intended. While measuring this is challenging due to self-reporting biases and discomfort with observation (Parecki & Gear, 2013; Prins & Toso, 2008; Van Otterloo et al., 2006), evidence shows low participation when caregiver training supports are absent (Gomby et al., 1999; McElvany & van Steensel, 2009; St. Pierre et al., 2005; Wagner & Clayton, 1999). Even among participating families, many do not complete activities as intended without sufficient training (McElvany & van Steensel, 2009). Thus, while the Read by Grade Three Law requires caregiver training for “Read at Home” plans, it is unclear if districts are actually providing these opportunities, potentially hampering the plans’ 53 implementation and ability to boost literacy skills. In addition to caregiver training, research identifies facilitating two-way communication between educators and families as another crucial element of effective family engagement strategies. However, this component is not explicitly addressed in the requirements for “Read at Home” plans under the Read by Grade Three Law. Furthermore, the district plans I reviewed demonstrate a one-way, school-to-home model of engagement, with plans designed for broad dissemination to families rather than fostering interactive communication and relationship- building. A substantial body of literature consistently highlights the importance of two-way communication that allows schools and families to develop meaningful partnerships (Adams & Christenson, 2000; Clarke et al., 2009; de Carvalho, 2000; Kyzar & Jimerson, 2018; Lareau, 1989; Mapp & Kuttner, 2013). For example, Kraft and Dougherty’s (2013) randomized experiment found that teacher-initiated phone calls and messages to families, creating consistent two-way communication, significantly improved homework completion, attentiveness, and class participation by increasing parental involvement. Similarly, Pemberton and Miller (2013) documented how cultivating teacher-parent partnerships through ongoing communication transformed reading outcomes and attitudes towards family engagement in a Title I school within just two months. In contrast, more traditional family engagement approaches characterized by one-way transmission of information or resources from schools to families, without opportunities for families to provide input, can perpetuate power imbalances that undermine effectiveness (Henderson & Mapp, 2002; Ishimaru, 2017). As the “Read at Home” plans I reviewed illustrated a one-size-fits-all distribution model, this raises concerns about whether their design aligns with the research emphasis on two-way communication to allow for more personalized, authentic support between homes and schools. 54 Another critical element for effective family engagement identified in the literature is tailoring approaches to acknowledge and adapt to families’ diverse needs and preferences, which vary across racial, socioeconomic, linguistic, and geographic lines (Clarke et al., 2017; Graves Jr. & Wright, 2011; Hemmerechts et al., 2017; Ishimaru, 2017; LaRocque et al., 2011; Loera et al., 2011; Sheridan et al., 2020). For instance, Graves Jr. and Wright (2011) find that Black and Latino families tend to prefer school-based engagement like events and volunteering, contrasting with White families’ leaning towards home-based educational activities. Economically disadvantaged families may face barriers to school-based involvement like transportation issues, inflexible work schedules, or lack of childcare (Baker et al., 2016; Lechuga-Peña & Brisson, 2018), and rural families often encounter similar logistical obstacles to participation given geographic isolation (Clarke et al., 2017; Sheridan et al., 2020; Smith et al., 2011). Furthermore, language barriers can significantly hinder engagement for families where English is not the primary language spoken at home (Chrispeels & Rivero, 2001; Hornby & Lafaele, 2011; Smith et al., 2011). These differences across family contexts underscore that one-size-fits-all engagement strategies are unlikely to be equally effective. Research emphasizes the need for flexible, tailored approaches that recognize and build upon families’ existing strengths and resources, while also considering their unique challenges and circumstances (Edwards, 2016; Edwards et al., 2019, 2022; Edwards & Danridge, 2003). For example, Edwards (2016) recommends providing audiobooks for parents who find reading challenging, facilitating connections between non- English speaking families and bilingual school staff, and leveraging technology to mitigate obstacles related to timing, location, and schedules. When family knowledge is valued and they are encouraged to share insights into their children’s needs, engagement tends to be higher 55 (Doyle & Zhang, 2011). While the Read by Grade Three Law does not explicitly require this level of tailoring for “Read at Home” plans, some district plans I reviewed did provide translations into languages like Spanish. However, this represents just one aspect of tailoring for diverse needs. Many other circumstances like varying literacy levels, economic constraints, disabilities, and more would need to be considered to create truly personalized, culturally responsive, and accessible family engagement opportunities. Finally, robust implementation of effective family engagement strategies hinges on providing educators with sufficient training, resources, and capacity-building supports. However, policies often incorrectly assume educators already possess the necessary skills, knowledge, and supportive beliefs for meaningful family engagement (Mapp & Kuttner, 2013). In reality, many educators report feeling underprepared and lacking sufficient training in this area, including those in Michigan (Caspe et al., 2011; Hess & Kelly, 2005; Mapp & Kuttner, 2013; Markow & Pieters, 2009; Utter et al., 2023). Notably, the Read by Grade Three Law does not provide supplemental guidance, templates, professional development, or other capacity-building resources specifically for implementing “Read at Home” plans. This lack of support could hinder educators’ understanding of the plans and their readiness to effectively create and deliver them. In summary, while Michigan’s “Read at Home” plans incorporate some research-based strategies like promoting shared reading and caregiver instruction in literacy activities, the policy design and district plans I reviewed appear to lack other critical elements identified as best practices. These include providing training for caregivers on implementing literacy activities, facilitating authentic two-way communication between educators and families, and offering tailored engagement opportunities reflective of families’ diverse contexts. Addressing these gaps could enhance the potential for the state’s “Read at Home” plans to leverage the well- 56 documented benefits of family engagement in improving literacy outcomes. This underscores the importance of the present study’s examination to identify areas where the policy and implementation could be strengthened. Conceptual Framework Implementing “Read at Home” plans under Michigan’s Read by Grade Three Law will involve coordinated efforts spanning multiple levels of the state’s educational system⎯from the state policy landscape down to individual districts, schools, classrooms, and the students and families receiving the plans (MDE, n.d.; Michigan Achieves, n.d.; Michigan Alliance for Families, n.d.). This multi-layered approach reflects Bronfenbrenner's (1979) ecological systems theory, which posits that individual development is influenced by interconnected environmental contexts, ranging from immediate settings like families and schools to broader societal and institutional influences. The conceptual framework guiding this study, depicted in Figure 2.2, draws from Bronfenbrenner’s (1979) theory as an organizing structure to examine how factors operating at different systemic levels—the students’ family context (microsystem), classroom teachers (mesosystem), schools and districts (exosystem), and the overarching state policy environment (macrosystem)—may shape the implementation of “Read at Home” plans. Moreover, the framework integrates complementary theoretical perspectives from the field of policy implementation to provide a more comprehensive lens for understanding the intricate dynamics underlying policy enactment in real-world contexts. Specifically, the framework incorporates concepts surrounding policy instruments, which examine how the design and tools embedded within a policy are intended to translate its goals into action (Bardach, 1979; Gormley, 1987; Linder & Peters, 1989; McDonnell & Elmore, 1987; Salamon, 1981, 1989; Schneider & Ingram 57 1990). It also explores processes of sensemaking, through which implementers construct meaning and understanding of a policy based on their own experiences and organizational contexts (Spillane, 2000; Spillane & Anderson, 2019; Spillane et al., 2002a, 2002b; Weick, 1995). Additionally, the framework considers theories of filtering, whereby information about a policy gets selectively adopted, altered, or deflected as it cascades through an organizational hierarchy (Diehl & Golann, 2023). Finally, it draws upon the notion of street-level bureaucracy to understand how frontline implementers like teachers ultimately translate broad policy directives into concrete actions tailored to the realities of their local contexts and available resources (Lipsky, 1980; Meyers & Nielsen, 2012). 58 Figure 2.2 Conceptual Framework Underlying “Read at Home” Plan Implementation Note. Adapted from Bronfenbrenner (1979). 59 Together, these theoretical lenses provide a comprehensive conceptual framework for examining the complex factors—from the initial policy design down to the subjective experiences and constraints of individual implementers—that can ultimately influence whether and how “Read at Home” plans are implemented in practice. Policy Landscape At the broadest level, the policy landscape includes the Read by Grade Three Law and its specific requirements surrounding “Read at Home” plans. Policymakers employ various instruments or tools to translate a policy’s intended goals into concrete actions (Bardach, 1979; Gormley, 1987; Linder & Peters, 1989; McDonnell & Elmore, 1987; Salamon, 1981, 1989; Schneider & Ingram, 1990). These instruments are generally categorized as mandates, which require action to ensure compliance; inducements or incentives, which provide funding in exchange for specific actions; capacity-building tools, which provide funding to invest in material, intellectual, or human resources; and system-changing tools, which transfer authority to alter the delivery of public services and goods (McDonnell & Elmore, 1987). In the case of the Read by Grade Three Law, “Read at Home” plans are essentially mandated, with the policy directing districts to provide them to 3rd-grade students identified as having “reading deficiencies.” For K-2 students, the plans are recommended but not explicitly mandated. The law also broadly outlines required plan components like caregiver training and home reading promotion but lacks explicit sanctions for non-compliance (Michigan Public Act 306, 2016). However, districts must report whether “reading deficient” students were provided a plan. Mandates operate under assumptions that the required action would not consistently occur without the policy in place, that organizations will comply due to established hierarchical reporting structures, and that all implementers have equal capacity for implementation 60 (McDonnell & Elmore, 1987; Schneider & Ingram, 1990). Crucially, mandates typically provide no additional financial resources, instead imposing costs on implementers. This unfunded mandate approach characterizes Michigan’s “Read at Home” plans, as districts receive no funding, training, or capacity-building to develop and deliver the plans. All human, financial, intellectual, and relational capital demands—creating templates, purchasing materials, training caregivers, engaging families—fall on existing district resources. With only basic plan requirements outlined, districts also lack clear implementation standards or incentives beyond basic compliance (McDonnell & Elmore, 1987). This policy landscape suggests a few key implications for how “Read at Home” plans may be expected to unfold on the ground. Implementation will likely be higher for 3rd graders due to the explicit mandate for that grade level, compared to the recommendation for K-2 students. However, overall implementation levels are expected to be low due to the Read by Grade Three Law’s lack of capacity-building supports for the plans. Additionally, while self- reporting requirements introduce some accountability, the absence of explicit consequences for non-compliance may diminish perceived oversight, further constraining the implementation of “Read at Home” plans across Michigan. District Level Policy requirements then cascade down to those responsible for implementing them. In the case of “Read at Home” plans, this responsibility first falls on district leaders or superintendents (Michigan Public Act 306, 2016). Effectively managing this duty requires them to make sense of the plans and their requirements (Spillane, 2000; Spillane et al., 2002b; Weick, 1995). Sensemaking theory explains how individuals and organizations construct understandings and determine actions around ambiguous, complex situations like policy implementation 61 (Spillane et al., 2002b; Weick, 1995). For district leaders’ sensemaking of “Read at Home” plans, two key elements will be their understanding about the plans and their perceptions regarding the plans’ potential effectiveness (Matland, 1995; Spillane et al., 2002b; Weick, 1995). District leaders will first need to understand “Read at Home” plans to effectively implement them. Policy documents often contain broad and ambiguous language, requiring implementers to interpret their intended meaning (Matland, 1995; Spillane et al., 2002b; Weick, 1995). Matland’s (1995) ambiguity-conflict theory suggests policies with low ambiguity and minimal conflict between stakeholders tend to have higher implementation success, as implementers clearly grasp what actions to take and have minimal opposition or resistance to the policy. While family engagement is generally not a contentious policy topic—indicating low conflict—the Read by Grade Three Law provides minimal guidance specifying the content and structure for “Read at Home” plans, introducing ambiguity around the policy’s requirements. Clear communication and training can help reduce ambiguity and enhance implementation (Hill & Hupe, 2002; Spillane et al., 2002b). However, the lack of capacity-building supports in the Read by Grade Three Law may limit district leaders’ understanding of “Read at Home” plans, ultimately hindering their implementation (Hill, 2003; Spillane, 2000; Spillane et al., 2002b; Yanow, 1996). District leaders’ perceptions about the potential effectiveness of “Read at Home” plans will also be critical for implementation (Spillane et al., 2002b; Weick, 1995). When implementers view a policy positively and see it as aligned with existing goals and practices, they are more likely to comply and dedicate time and resources to ensuring its success (Grebing et al., 2023; Honig, 2006; Lee & Min, 2017; Sabatier & Mazmanian, 1980; Spillane et al., 2002b; Weick, 1995). Conversely, negative perceptions or skepticism can breed minimal effort, 62 lack of buy-in, and ultimately undermine implementation (McLaughlin, 1987; May & Winter, 2009). District leaders who struggle to see the value or potential impact of “Read at Home” plans may be disinclined to prioritize them. Robust training and clear communication are crucial for fostering buy-in, as they help implementers understand the policy’s goals and rationale (May & Winter, 2009; Turnbull, 2002). Without such guidance, district leaders’ negative perceptions could hamper their motivation to put in meaningful effort toward implementing “Read at Home” plans. District leaders’ sensemaking surrounding “Read at Home” plans takes on heightened importance because of how it shapes the information and vision that ultimately get filtered down to educators throughout the district (Coburn, 2005, 2006; Diehl & Golann, 2023; Honig, 2006; Marsh & Wohlstetter, 2013; Park et al., 2013; Spillane, 1998, 2000; Wong, 2019; Woulfin et al., 2016). Facing numerous policy initiatives but finite resources, district leaders must prioritize what details about “Read at Home” plans—if any—to emphasize and communicate to principals and teachers through a process called filtering (Diehl & Golann, 2023). This filtering process, also known as sensegiving (Gioia & Chittipeddi, 1991; Watson, 1994; Wong, 2019), involves district leaders selectively adopting, altering, or deflecting policies as they frame and disseminate them for educators at lower levels of the educational system (Park et al., 2013; Woulfin et al., 2016). Studies show this filtering can significantly impact how a policy manifests at lower levels. For example, Woulfin et al. (2016) find that district leaders’ filtered framing of a new educator evaluation policy in Connecticut, emphasizing accountability aspects, led principals to prioritize compliance over encouraging teacher growth. This resulted in more symbolic, surface-level policy enactment rather than full engagement with the policy’s intended goals (Woulfin et al., 2016). Similarly, how district leaders filter and frame “Read at Home” plans will shape 63 principals’ sensemaking and implementation in their schools. As part of the sensemaking and filtering process, district leaders must consider the specific organizational context in which they would implement “Read at Home” plans (Diehl & Golann, 2023; Spillane et al., 2002b). Several key factors may influence how they interpret the plans’ requirements and prioritize their implementation. For example, student needs and demographics will likely play a role. Districts with higher proportions of students identified as “reading deficient” and eligible for plans may find providing them to all students challenging. With increased eligibility often comes reduced available resources per capita, potentially undermining consistent implementation (Bovens & Hart, 1996; Schneider & Ingram, 1993). Furthermore, district diversity across racial, economic, linguistic, and disability categories may impact leaders’ perceived need and motivation to implement the plans. High diversity could spur prioritization to support historically underserved groups who research shows benefit significantly from family engagement, including students of color, economically disadvantaged students, ELs, and students with disabilities (Brown et al., 2019; Fikrat-Wevers et al., 2021; He & Thompson, 2022; Jeynes, 2003, 2012, 2016, 2017; Sheridan et al., 2011). Conversely, the complexity and resource demands associated with addressing diverse needs could deter these same districts from implementing the plans. Baseline ELA achievement levels may also factor in. Higher-achieving districts generally have more resources and well-developed infrastructures but might be less motivated to implement the plans if they perceive their current family literacy supports as successful (Archibald, 2009; Bidwell & Kasarda, 1975; Corcoran & Goertz, 1995; Darling-Hammond, 2013; Jimenez-Castellanos, 2010). In contrast, lower-achieving districts may be eager to adopt the plans as a means to improve literacy, but often struggle with resource limitations (Brown et 64 al., 2013; Debray et al., 2003; Hess & Finn, 2004; Mintrop, 2004). A district’s operational model may also play a role. Charter districts, with enhanced flexibility and autonomy compared to traditional public school (TPS) districts, could either prioritize “Read at Home” plans or shirk the Read by Grade Three Law’s requirements for them (Michigan Association of Public School Academies, 2023). This will likely depend on how aligned they see the plans with their mission. If they view the plans as aligned, they might be motivated to implement them as a strategy to recruit or retain families (Hoxby, 2003; Lake, 2008; Preston et al., 2012). Conversely, if they do not view the plans as aligned, they might be less likely to implement them, as limited oversight could lead charter districts to perceive a lower risk of being sanctioned for non-compliance with the law (Annenberg Institute for School Reform, 2014; Bulkley & Fisler, 2003; Finn et al., 2001). Finally, geographic locale can influence implementation capacity. Suburban districts, typically wealthier and with higher parental engagement, may be better equipped for implementation and face stronger community expectations to provide “Read at Home” plans (Reardon, 2016; Roscigno et al., 2006; Wepner & Gomez, 2017). Conversely, urban districts may struggle due to resource limitations, despite proximity to students and families potentially facilitating engagement (Center for Public Education, 2024; Roscigno et al., 2006). Rural districts, facing geographic isolation and fewer resources, may also face significant hurdles in engaging families effectively (Lavelley, 2018; Roscigno et al., 2006). Navigating this complex landscape, district leaders must make critical decisions about whether and how to prioritize “Read at Home” plans. Their contextualized sensemaking and filtering will fundamentally shape how the Read by Grade Three Law’s requirements for these plans take shape at subsequent implementation levels. 65 School Level The implementation of “Read at Home” plans next hinges on the school level. Schools do not simply enact district directives verbatim; principals engage in their own sensemaking and filtering process to construct meaning about policies through the lens of their school’s unique context (Coburn, 2005; Diehl & Golann, 2023; Ganon-Shilon & Schechter, 2017; Spillane et al., 2002a; Spillane & Anderson, 2019). The factors principals consider likely mirror those at the district level, such as student need and demographics, baseline ELA achievement, charter status, and locale. Principals serve as vital intermediaries between the district office and classroom implementation (Coburn, 2005; Diamond & Spillane, 2004; Diehl & Golann, 2023; Spillane & Anderson, 2019; Spillane et al., 2002a; Wong, 2019). Their proximity to students, families, and teaching staff positions them to significantly influence whether and how educators implement “Read at Home” plans. Principals’ own understanding and perceptions of the plans will shape how they frame the requirements and expectations of this policy as they filter it down to teachers. How principals frame the plans will impact whether teachers see value and legitimacy in them, which will in turn affect teachers’ ability and motivation to implement them (Diamond & Spillane, 2004; Spillane & Anderson, 2019; Spillane et al., 2002b). Research highlights that when principals pass along ambiguous, incomplete, or surface-level policy details, it can severely limit teachers’ ability to enact its intended aims (Wong, 2019). Conversely, if principals deeply understand and believe in the merits of “Read at Home” plans, they can foster robust implementation. Ultimately, principals will serve as catalysts facilitating or obstructing the implementation of the plans based on how they filter and prioritize them for teachers. 66 Teacher Level As the direct connection point to students and families, teachers will play a pivotal role in the implementation of “Read at Home” plans. While influenced by how the policy details filter down from higher levels of the educational system, teachers also undergo their own critical sensemaking process that will be key to successful implementation (Smit, 2005). Those who interact directly with policy targets, in this case the students and families receiving the plans, are commonly referred to as street-level bureaucrats (SLBs; Lipsky, 1980). In this role, teachers must translate the requirements of the policy into their daily classroom routines and practices (Cholakova & Ravasi, 2019; Coburn, 2004; Diehl & Golann, 2023; Spillane et al., 2002b). This poses challenges, as SLBs frequently work in environments such as schools, where goals are ambiguous, resources are limited, and demand exceeds supply. These conditions require SLBs to exercise discretion in how they implement policies (Lipsky, 1980). For “Read at Home” plans, teachers will need to determine the extent to which the plans can be integrated with their existing family engagement strategies and constraints (Coburn, 2004; Keiser, 2010; Lin, 2000; Meyers & Nielsen, 2012; Spillane et al., 2002b). The degree of congruence between the plans and teachers’ current norms and practices will shape their willingness and ability to implement them (Coburn, 2004; Spillane et al., 2002b). Teacher characteristics may further influence plan implementation. While limited research directly links teacher traits to family engagement outcomes, some studies suggest experienced teachers tend to resist new instructional strategies due to ingrained habits (Coburn, 2004; Ghaith & Yaghi, 1997; Spillane et al., 2002b). Conversely, highly effective teachers often exhibit more openness to innovations (Ghaith & Yaghi, 1997), which could extend to embracing “Read at Home” plans. For 3rd-grade teachers specifically, the Read by Grade Three Law’s 67 mandate that “reading deficient” students in this grade level receive plans could increase implementation compared to other grade levels. However, ingrained practices among experienced 3rd-grade teachers may still hinder adoption of the plans. Given existing practices shape sensemaking and implementation of new policies, understanding teachers’ current family literacy efforts beyond the plans will provide critical context for how the plans could be integrated or face resistance. Resource constraints pose another significant challenge for SLBs (Lipsky, 1980; Meyers & Nielsen, 2012). Time, in particular, is a major challenge for teachers, who have numerous responsibilities beyond “Read at Home” plans, including instruction, assessments, and supervisory duties. Limited time leads SLBs to prioritize and ration when implementing policies (Keiser, 1999, 2010; Keiser & Soss, 1998; Lipsky, 1980; Meyers & Nielsen, 2012; Pesso, 1978; Winter, 2001). Consequently, teachers may need to apply discretionary judgment in determining who receives a “Read at Home” plan, especially when there are many eligible students (Lipsky, 1980; Meyers & Nielsen, 2012). This discretion could manifest as positive discrimination, where SLBs assist those whom they consider most in need or deserving of service (Goodsell, 1981; Maynard-Moody & Musheno, 2003; Meyers & Nielsen, 2012). However, it could also mean providing plans to those deemed most cooperative or most likely to use them (Keiser, 1999, 2010; Keiser & Soss, 1998; Lipsky, 1980; Pesso, 1978; Winter, 2001). In the worst cases, it could result in discriminatory treatment, withholding plans from those who could most benefit because of bias or misinformation (Brodkin, 1997; Hill & Bramley, 1986; Lipsky, 1980; Meyers & Nielsen, 2012). Teachers may consider various factors when prioritizing students for “Read at Home” plans, discussed in more detail in the next section. 68 Student and Family Characteristics Finally, the framework considers how student and family characteristics might play a role in shaping who receives a “Read at Home” plan. The policy mandate element of the Read by Grade Three Law suggests that 3rd graders identified as “reading deficient” will likely be given plans more frequently than K-2 students, who are only recommended to receive them. However, when faced with resource constraints, SLBs may use discretionary judgment to determine how to disseminate “Read at Home” plans (Lipsky, 1980; Meyers & Nielsen, 2012). Teachers’ understanding of students’ perceived needs, personal biases, and situational factors could guide how they prioritize which students and families should receive plans (Keiser, 2010; Lipsky, 1980; Maynard-Moody & Musheno, 2003; Meyers & Nielsen, 2012). For instance, caregivers’ education level could factor into teachers’ decisions.9 Generally, mothers with higher education levels are more engaged in their children’s education (Baker, 2013; Curenton & Justice, 2008; Raikes et al., 2006; Weigel et al., 2006), potentially leading these families to actively seek out plans or making teachers more inclined to offer them, recognizing their involvement. Conversely, teachers might assume that students from these backgrounds already receive ample educational support at home, reducing the likelihood of providing them plans. Student demographics may also guide SLB discretion about disseminating “Read at Home” plans. Gender provides one example: Given research showing girls often exhibit higher achievement and more positive attitudes towards reading (e.g., Logan & Johnston, 2010), teachers could prioritize providing plans to boys, perceiving them as needing more encouragement for home reading. Conversely, they may direct plans to girls based on 9 Because I am not able to measure caregiver education level in the data, I am unable to include this in my analysis. 69 expectations they will use them more often. A students’ race and economic disadvantage status are also likely factors. Students of color and those from economically disadvantaged backgrounds, groups who research indicates benefit significantly from family literacy engagement (Fikrat-Wevers et al., 2021; Jeynes, 2003, 2016, 2017), may be prioritized to receive plans to leverage those potential benefits. However, this demographic being targeted could also indicate problematic biases if it stems from assumptions that these families are less engaged in literacy activities at home. Additionally, teachers may consider a student’s EL and disability status in their implementation of “Read at Home” plans. While both EL students and those with disabilities stand to substantially gain from family engagement strategies (Brown et al., 2019; He & Thompson, 2022; Sheridan et al., 2011), teachers’ perceptions of whether families can effectively access and use the plans may vary. Availability of translations, supplemental resources, and the student’s Individualized Education Program (IEP) could all factor into these judgments about providing plans to such students and families. Summary This framework delineates the complex environment surrounding the implementation of “Read at Home” plans, using Bronfenbrenner’s (1979) ecological systems theory as an organizational frame. It integrates various policy implementation perspectives, including policy instruments, sensemaking, filtering, and street-level bureaucracy, to examine how dynamics within each layer of this framework might influence the implementation of these plans. While state-level policies and practices lay the groundwork for “Read at Home” plans, their actual implementation will be shaped by factors at lower educational levels and by the characteristics of the students and families involved. The following section details the data and methods used to explore this framework, providing empirical insights into how these theoretical elements relate to 70 the implementation of “Read at Home” plans. Data and Methods Data Sources This study draws on data from multiple sources. While RQs 1 and 2 use statewide administrative data provided by MDE, RQs 3 and 4 incorporate statewide surveys of educators about their understanding and perceptions of “Read at Home” plans and the extent to which they implement other family engagement strategies. Administrative Data The administrative data span four school years, 2019-20 to 2022-23, as districts provided data on “Read at Home” plans during these years. Two key variables are students’ “reading deficiency” status and an indicator for whether they were given a “Read at Home” plan. The “reading deficiency” indicator reveals whether a student was identified as having a “reading deficiency” based on the district’s screening assessment. “Read at Home” plan eligibility is based on this status. Meanwhile, the “Read at Home” plan variable captures whether the district reported providing the student with a plan. Importantly, this does not confirm the actual receipt or use of the plan by the student or their family. Additional administrative data include student, teacher, school, and district characteristics anticipated to be pertinent based on the study’s conceptual framework. For students, these include enrolled grade level, gender, race, economic disadvantage status, EL status, and disability status. Additionally, the administrative data link students to their teachers in the 2020- 21 through 2022-23 school years, allowing an examination of the relationship between relevant teacher characteristics and whether a student was provided a “Read at Home” plan. The teacher characteristics include grade level, years of experience, effectiveness ratings, and ELA 71 endorsements—an additional measure of effectiveness, specifically in literacy. The data for schools and districts include baseline ELA achievement from the 2018-19 school year on the state’s 3rd-grade literacy assessment, enrollment size, charter status, and geographic locale. Finally, I use student administrative data to generate classroom, school, and district demographic variables. Collectively, these variables enable a comprehensive examination of variation in the implementation of “Read at Home” plans and other family engagement strategies across different levels of the educational system. Statewide Educator Surveys I also incorporate data from statewide surveys conducted annually from the 2019-20 through 2022-23 school years. These surveys targeted K-3 teachers, elementary school principals, and district superintendents and are part of the Education Policy Innovation Collaborative’s (EPIC’s) broader study of the Read by Grade Three Law (see Strunk et al., 2021, 2022). They included questions exploring educators’ sensemaking surrounding “Read at Home” plans, including their understanding and perceived effectiveness of the plans, their capacity to implement them, and the other family engagement strategies they employ beyond the plans. To develop survey questions, EPIC combined original items with adapted items from existing early literacy surveys (Baumann et al., 2000; Jacob, 2017; Marsh et al., 2008; Mesmer, 2006; RMC Research Corporation, 2019). External stakeholders, policymakers, MDE, and the Michigan Association of Intermediate School Administrators General Education Leadership Network’s Early Literacy Task Force provided feedback to refine the survey content. Educators from the target populations pilot-tested the survey and participated in cognitive interviews to further improve the clarity of the questions. Data collection occurred online during a three-month window each spring. EPIC recruited eligible participants through direct emails, website 72 promotion, social media engagement, and partnerships with the Michigan Education Association, the American Federation of Teachers, and the Michigan Association of Public School Academies.10 Sample Administrative Data Sample The administrative data sample used to address RQs 1 and 2 consists of students enrolled in grades K-3 in TPS and charter school districts between the 2019-20 and 2022-23 school years (N=1,698,409 student-year observations). To ensure the reliability and validity of the analysis, I apply several sample restrictions outlined in Appendix Table A2.1. First, I limit the sample to students in districts that received funding under Section 35a(5) of the State School Aid Act, as only these districts are required to report students’ “reading deficiency” status and the interventions provided to them, including “Read at Home” plans.11 This results in 59,991 student-year observations being dropped. Among the 97% of districts that received Section 35a(5) funding, the majority (81%, N=613) received funding consistently across all four years of the study. I restrict the sample to students in districts funded all four years (dropping 103,202 student-year observations). This ensures that students’ districts had consistent access to funding for interventions, including “Read at Home” plans, throughout the study period, reducing potential confounding effects introduced by receiving funding in some years but not others. Finally, I narrow the sample to students within these districts who have been identified with a “reading deficiency,” as districts are only required to report interventions for these students. This 10 In Michigan, charter schools are called public school academies. 11 Section 35a(5) of the State School Aid Act is otherwise known as the Additional Instructional Time Grant (MDE, 2022). Around 97% of the state’s districts receive this funding each year, totaling $19.9 million annually during the study period. 73 results in an analytical sample of 475,902 student-year observations. In the 2020-21 through 2022-23 school years, the sample is augmented to include teacher characteristics for a subset of students who could be matched to their teachers through available data. The teacher-student data links are available for 363,688 student-year observations, enabling exploration of how teacher attributes are associated with “Read at Home” plan implementation. Table 2.1 reveals differences between the analytical sample and Michigan’s broader K-3 student population. Students in the sample are much more likely to be economically disadvantaged, Black, and students with disabilities compared to those in the population. These discrepancies reflect that these students are more likely to be identified with “reading deficiencies.” Consequently, while the sample does not broadly represent the K-3 student population, it does reflect the population of interest and is aligned with the study’s aim to examine “Read at Home” plan implementation for students who qualify for these interventions. Table 2.1 Administrative Data Sample Description Economically Disadvantaged English Learner Student with Disabilities Black Hispanic Asian Other Race White Female Charter Suburb Urban Rural N Population 58.0% 8.8% 14.9% 18.5% 8.8% 3.7% 6.3% 62.7% 48.6% 12.4% 54.7% 24.7% 20.6% 1,698,409 Difference 14.1% 1.2% 7.6% 6.3% 2.0% -1.4% 0.3% -7.2% -1.9% 1.0% -3.6% 2.6% 1.1% Sample 72.1% 10.0% 22.5% 24.8% 10.8% 2.3% 6.6% 55.5% 46.7% 13.5% 51.1% 27.3% 21.7% 475,902 74 Survey Sample The survey sample comprises K-3 teachers, elementary school principals, and district superintendents who participated in EPIC’s Read by Grade Three Survey from 2019-20 to 2022- 23. Appendix Figure A2.3 provides a detailed breakdown of the response rates by year. Over these four years, the average response rates were 31.5% for teachers, 26.8% for principals, and 23.6% for superintendents. Response rates for all groups declined over time from a peak in 2019- 20 (43.4% for teachers, 44.9% for principals, and 36.4% for superintendents), with the steepest drop occurring between the 2019-20 and 2020-21 school years (reductions of 13.8, 22.3, and 10.8 percentage points, respectively). This decline is most likely attributable to the additional burdens placed on educators during the COVID-19 pandemic and to survey fatigue, as EPIC invited educators to respond to the survey for four consecutive years. The survey sample generally aligns well with the target population across the four years of the study, albeit with some discrepancies. For instance, as indicated in Appendix Table A2.2, the proportion of recently hired K-3 teachers in the sample consistently exceeds the state average. The sample also has a higher percentage of female principals and superintendents compared to statewide statistics. Despite these differences, this study presents unweighted survey results. In addition to the sample being generally well-aligned with the population, analyses using both weighted and unweighted survey responses yield similar outcomes.12 Methods RQ1: Proportion of Students Given “Read at Home” Plans To investigate the proportion of students with “reading deficiencies” to whom districts report providing “Read at Home” plans and how this varies across districts, I begin by 12 The weighted responses are adjusted based on the demographic characteristics detailed in Table 2.2 and are available upon request. 75 calculating descriptive statistics, including the average percent of “reading deficient” students given plans across all districts and by school year and grade level. I also generate histograms to visually examine variation in the proportion of students given plans across districts. These analyses reveal three distinct categories of districts based on their implementation of “Read at Home” plans: (1) districts that do not provide plans to any “reading deficient” students (the “None” group), (2) districts that provide plans to some but not all eligible students (“Some”), and (3) districts that provide plans to all eligible students (“All”). To explore the characteristics of districts within each category, I employ multinomial logistic regression given the categorical nature of the outcome variable. I estimate the following model to compare the likelihood of a district being in either the “Some” or “All” category against the “None” category: log( 𝑃(𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦𝑑𝑡,𝑘 = 𝑘|𝐃) 𝑃(𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦𝑑𝑡,𝑘 = 𝑁𝑜𝑛𝑒|𝐃) ) = 𝛽𝑘,0 + 𝐃𝑑𝑡 + 𝜏𝑡 (1) In Model (1), 𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦𝑑𝑡,𝑘 indicates the log-odds of district d falling into category 𝑘 (Some or All) in school year t, compared to the reference category (None). 𝐃𝑑𝑡 is a vector of district d characteristics in year t, including the proportion of “reading deficient,” non-White, economically disadvantaged, ELs, and students with disabilities, the district’s average scale score on the state’s 3rd-grade ELA assessment in the year before the study (2018-19), log enrollment, indicators for whether it is a charter district (compared to TPS) and located in a rural or urban area (compared to suburban). Year fixed effects 𝜏𝑡 control for year-to-year variations and account for the panel structure of the data. Standard errors are clustered at the district level, which is the entity responsible for providing “Read at Home” plans. The coefficients represent the change in the log-odds of a district being in a specific category (Some or All) versus the reference category (None) for a one-unit change in each predictor variable, holding other variables constant. To make these results more interpretable, I calculate marginal effects to show 76 how a one-unit change in each predictor is related to the predicted probability of a district falling into a specific category (None, Some, All) relative to the other two. RQ2: Factors That Distinguish Which Students Are Given Plans RQ2 delves deeper into the districts that provide some, but not all, “reading deficient” students with “Read at Home” plans and examines the factors that distinguish the students given plans from those not. Restricting the analysis to students in the Some districts introduces essential variation to analyze factors related to plan implementation. In these districts, examining the distinguishing characteristics between students who do and do not receive plans is possible because there is variation in whether a student is given a plan. Conversely, in None or All districts, there is no variation in the outcome variable; plan receipt is uniform across all students regardless of their characteristics. This prevents analysis of which factors are associated with the likelihood of receiving a plan because the outcome does not change based on individual student characteristics. While this approach narrows the analysis to a smaller subset of districts and students, it is essential for evaluating variation in implementation across student subgroups. For this analysis, I estimate a series of linear probability models (LPMs) on this subsample due to their straightforward interpretability and computational simplicity. To ensure robustness, I compare these results with logistic regression analyses, which yield estimates that are consistent in direction, magnitude, and statistical significance (available upon request). 𝑃𝑙𝑎𝑛𝑖𝑡 = 𝛽0 + 𝚾𝑖𝑡 + 𝜏𝑡 + 𝜀𝑖𝑡 (2) Model (2) includes a vector of individual student characteristics (𝚾𝑖𝑡) to examine whether 3rd graders—who are mandated to receive plans—along with female, Black, Hispanic, Asian, other race, economically disadvantaged students, ELs, and students with disabilities are more or 77 less likely to be given “Read at Home” plans than their peers.13 These student-specific predictors are supplemented with year fixed effects 𝜏𝑡 to account for year-to-year variations that could influence plan implementation. The coefficients in the model represent the difference in the probability of being given a “Read at Home” plan for students in each subgroup relative to the reference group (e.g., female relative to male). Standard errors are clustered at the district level.14 To account for other characteristics that may be related to whether a student is given a “Read at Home” plan, I introduce a series of fixed effects into various specifications of Model (2).15 These include teacher fixed effects, which account for time-invariant characteristics of teachers like their teaching style and demographic background; school fixed effects, addressing time-invariant aspects of schools such as their culture and location; and district fixed effects, which adjust for time-invariant attributes of districts, like their policies, resources, and community context. Including these fixed effects shifts the analysis to rely on within-teacher, within-school, and within-district variation rather than differences between teachers, schools, or districts, respectively. This approach is beneficial for parsing out the relationship between 13 The mean VIF for the covariates in the model is 1.16, suggesting that multicollinearity among the predictors is not a major issue. 14 Because districts are ultimately responsible for “Read at Home” plans according to the Read by Grade Three Law, clustering standard errors at this level is the most appropriate approach. It captures the policy-driven nature of the treatment and acknowledges that outcomes within districts may not be independent but rather influenced by shared district-wide policies and practices. Clustering at lower levels might not fully capture the intra- district correlation of errors; students within the same district are likely to experience similar influences related to plan implementation, leading to correlated outcomes. Not accounting for this could underestimate the standard errors, falsely inflating the statistical significance of the predictors. Indeed, moving from clustering at the district level to progressively more granular levels—from school to classroom to individual student—results in progressively smaller standard errors and, correspondingly, a greater number of predictors appearing statistically significant, even as the coefficients maintain their direction and magnitude. However, this likely reflects an overestimation of the precision of the estimates rather than a true increase in the relevance of these factors. 15 I also estimate LPMs that incorporate specific predictors at the teacher and school levels for which data are available (as opposed to fixed effects). For teachers, these include whether they are a new teacher (i.e., hired within the last five years), rated highly effective on their most recent evaluation, and have an ELA endorsement. At the school level, they include the proportion of students with “reading deficiencies,” a range of demographic factors, school size, charter status, and locale. Appendix Table A2.5 reports the results from this analysis, which show that none of the teacher predictors yield significant results, and the school predictors have minimal relationship. This suggests that factors not captured by the available administrative data are related to “Read at Home” plan implementation. As a result, I prefer the LPMs with fixed effects. 78 student characteristics and being given a “Read at Home” plan, as it holds constant the broader educational context that might also relate to plan implementation. Consequently, any observed variation in the implementation of plans is attributed to differences among students. Model (2) omits the 2019-20 school year due to the absence of teacher-student data links, which are necessary for this analysis.16 I test the robustness of these models to logistic regression (results are consistent and available upon request). RQ3: Educators’ Understanding and Perceptions of “Read at Home” Plans In RQ3, I rely on survey data to understand factors at the teacher, school, and district levels that may be related to “Read at Home” plan implementation. Specifically, I examine three items asked of teachers, principals, and superintendents in multiple years related to their sensemaking surrounding the plans. These items included their understanding of the plans (2019- 20 and 2022-23), beliefs around the effectiveness of the plans in increasing achievement (2019- 20, 2021-22, and 2022-23), and whether they have sufficient time to implement them (only asked to teachers; 2019-20 and 2020-21), each measured on a four-point Likert scale. Appendix Table A2.4 details the full text and response options for each item. To analyze educators’ responses, I combine the top two Likert-scale responses to each question and code them as a 1, indicating a positive response. I code the bottom two responses as a 0, indicating a negative response.17 I calculate descriptive statistics of educators’ responses to each item separately for each stakeholder group and school year. Chi-square tests reveal significant differences in responses across educator groups each year. To further explore these 16 Running Model (2) without teacher fixed effects and without omitting the 2019-20 school year yields robust results. This demonstrates that the exclusion of the 2019-20 school year from the analysis does not significantly alter the findings. 17 I use binary response categories instead of all the original response categories because in many cases, the percentages of educators selecting the lowest and highest categories were very small. 79 differences, I employ a series of bivariate LPMs: 𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 = 𝛽0 + 𝛽1𝐺𝑐𝑡 + 𝜀𝑐𝑡 (3) In Model (3), 𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 indicates a positive response for educator c in school year t and 𝐺𝑐𝑡 indicates a specific stakeholder group. For example, when comparing teachers and principals in 2019-20, I restrict the data to these two groups and this school year. 𝐺𝑐𝑡 equals 1 if the educator is a principal and 0 for a teacher, and a positive 𝛽1 suggests that principals are more likely to respond positively than teachers. Standard errors are again clustered at the district level because districts are responsible for providing “Read at Home” plans, leading to the possibility that educators within the same district share similar perceptions about them. I apply the Bonferroni correction to address multiple comparisons and test the robustness of the results to logistic regression (results are consistent and available upon request).18 To examine differences in perceptions across various educator characteristics, I estimate the following LPM: 𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 = 𝛽0 + 𝚻𝑐𝑡 + 𝜏𝑡 + 𝜀𝑖𝑡 (4) In Model (4), 𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 again indicates a positive response for educator c in school year t. 𝚻𝑐𝑡 is a vector of characteristics, including whether the teacher is a 3rd-grade teacher, was hired in the last five years, rated highly effective on their most recent evaluation, or has an ELA endorsement. It also includes the proportion of “reading deficient,” non-White, economically disadvantaged, ELs, and students with disabilities in the district, the district’s log enrollment, and 18 Model (3) does not include district fixed effects because the district fixed effects would essentially drop out or become redundant for the superintendent responses. This happens because there is only one superintendent per district, making it impossible to separate the effect of the district from the effect of the individual superintendent within that district. When I include district fixed effects in comparisons between teachers and principals, the results remain robust, indicating that the primary insights drawn from the analysis are not significantly influenced by unobserved district-level characteristics. This robustness suggests that variation in the perceptions of “Read at Home” plans between teachers and principals are reflective of differences in these groups’ experiences and roles, rather than being artifacts of district-specific factors. 80 indicators for whether it is a charter (compared to TPS) district and located in a rural or urban (compared to suburban) area.19 𝜏𝑡 is a vector of year fixed effects, with the years included in the model depending on the years in which the question appeared on the survey. Standard errors are clustered at the district level for the reasons outlined above. I estimate Model (4) separately for each survey item and educator group. In the principal and superintendent models, I exclude the indicators for 3rd grade, highly effective, and ELA endorsement because they are not relevant for those groups. As with the other LPMs, I test robustness to a logistic regression model (results are consistent and available upon request). Finally, I explore how educators’ sensemaking about “Read at Home” plans is related to whether a student is given a plan. This analysis again restricts the sample to students in the Some group because of the necessary variation in the outcome variable. Because the items were asked to different groups in different years, I estimate the following LPM separately for each item: 𝑃𝑙𝑎𝑛𝑖𝑡 = 𝛽0 + 𝛽1𝑇𝑒𝑎𝑐ℎ𝑒𝑟𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 + 𝛽2𝑃𝑟𝑖𝑛𝑐𝑖𝑝𝑎𝑙𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 + 𝚾𝑖𝑡 + 𝜏𝑡 + 𝜀𝑖𝑡 (5) Model (5) mirrors Model (2) from RQ2, where I estimate the likelihood of a student being given a “Read at Home” plan based on their individual characteristics and year fixed effects.20 In Model (5), I add 𝑇𝑒𝑎𝑐ℎ𝑒𝑟𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡 and 𝑃𝑟𝑖𝑛𝑐𝑖𝑝𝑎𝑙𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑐𝑡, indicator variables for whether the student had a teacher or principal who had a positive response on the item of 19 I aggregate the proportion of “reading deficient,” non-White, economically disadvantaged students, ELs, and students with disabilities, as well as log enrollment, to the district level to facilitate comparisons across educators (teachers, principals, and superintendents). This aggregation allows for a uniform measure across different levels of the educational system. However, to ensure the robustness of my findings, I also created these variables at their respective levels (e.g., classroom for the teacher-level analysis, school for principal-level analysis). The results from these alternative specifications are consistent with those reported, supporting the reliability of the conclusions drawn from the aggregated district-level analysis. 20 Model (5) does not include teacher or school fixed effects because including such fixed effects would control for all their time-invariant characteristics, potentially absorbing the very variation in understanding and perceptions I am interested in examining. Essentially, the fixed effects would account for all unobserved, individual- specific factors that do not change over time, which includes their general understanding and perceptions towards “Read at Home” plans. 81 interest. I exclude superintendents because, for the subset of students included in this analysis (i.e., those with non-missing responses from their teacher, principal, and superintendent in a given year), every superintendent had a positive response. This uniformity means there is no variation in superintendent responses within this sample, making it impossible to assess the relationship between superintendent understanding and perceptions and the implementation of “Read at Home” plans. Like the other LPMs, I test robustness to a logistic regression model (results are consistent and available upon request). RQ4: Other Family Engagement Activities The analysis for RQ4 mirrors that of RQ3, except the final step connecting educators’ survey responses to whether a student is given a “Read at Home” plan. I first descriptively analyze educators’ responses to items about their involvement in family engagement activities beyond “Read at Home” plans, including offering research-based guidance for supporting literacy at home, organizing family literacy workshops, promoting regular at-home literacy activities, distributing resources for reading practice, and discussing students’ literacy progress with families (see Appendix Table A2.4). I conduct this analysis separately for each stakeholder group and school year. I then estimate Models (3) and (4) for each item to examine differences in responses across stakeholder groups and how educator characteristics relate to their reported engagement in each activity.21 RQ1: Proportion of Students Given “Read at Home” Plans Results Descriptive analyses show that, on average, the implementation of “Read at Home” plans is low across Michigan. Districts report providing these plans to only about one-fifth of students 21 Like RQ3, I test robustness of these models to logistic regression (results are consistent and available upon request). In Model (4), superintendents are dropped from the analysis because of small sample sizes. 82 with “reading deficiencies.” Table 2.3 details these percentages, showing a modest increase from 18.7% to 22.9% over the four years of the study. It is particularly striking that 3rd graders, despite being mandated to receive “Read at Home” plans, have the lowest average rates overall (21.3%, compared to 22.4% for kindergarteners, 22.9% for 1st graders, and 21.8% for 2nd graders). Table 2.3 Percent of “Reading Deficient” Students to Whom Districts Report Providing “Read at Home” Plans, Overall and by Grade Level and School Year Overall Mean SD Kindergarten Mean SD 1st Grade Mean SD 2nd Grade Mean SD 3rd Grade 2019-20 2020-21 School Year 2021-22 2022-23 Total 18.7% (35.2%) 21.0% (36.9%) 22.7% (38.6%) 22.9% (38.7%) 21.3% (37.4%) 21.0% (37.9%) 21.4% (38.2%) 24.0% (40.2%) 23.4% (39.6%) 22.4% (39.0%) 19.9% (36.6%) 22.2% (38.4%) 25.1% (40.8%) 24.4% (40.2%) 22.9% (39.1%) 19.0% (36.2%) 21.0% (37.5%) 23.0% (39.2%) 24.3% (40.4%) 21.8% (38.4%) Mean SD 18.4% (35.7%) 21.3% (37.9%) 21.3% (38.1%) Note. This table shows the means and standard deviations for the percent of “reading deficient” students to whom districts reported providing “Read at Home” plans, overall and disaggregated by grade level and school year. The mean percentages are calculated by dividing the number of students to whom each district reported providing a “Read at Home” plan by the total number of “reading deficient” students in the district for each school year and grade level and then taking the average percentage across districts for each cell in the table. 22.6% (39.1%) 22.9% (39.3%) However, these averages mask considerable variation across districts, evidenced by the large standard deviations in Table 2.3 and illustrated through visual analysis in Figure 2.3. The histogram in Figure 2.3 illustrates districts falling into three distinct groups based on the 83 proportion of students to whom they report providing “Read at Home” plans.22 A majority, 66.65% of district-year observations, report not providing plans to any “reading deficient” students, while 26.75% provide plans to some but not all such students. A smaller fraction, 6.6%, report giving plans to all “reading deficient” students.23 Figure 2.3 Proportion of “Reading Deficient” Students Provided “Read at Home” Plans Note. This figure pools all years together. The multinomial logistic regression analysis sheds light on the characteristics of districts in each of these categories. The results in Table 2.4 reveal that the primary variable related to a 22 Figure 2.3 pools all four school years and grade levels together, but the results are similar when disaggregated by grade level and school year. 23 Throughout the four-year study, most districts (64.48%) remain in the same category (most often None). Meanwhile, 21.77% of districts change categories once, typically moving from None to Some. Additionally, 11.78% of districts change categories twice, usually shifting from Some to either None or All and then returning to Some. A small fraction, 1.96%, change categories three times, often oscillating between Some and either None or All. 84 district’s categorization is the percentage of students eligible for the plans. Districts that do not provide “Read at Home” plans to any eligible students have a significantly smaller proportion of “reading deficient” students, with a 19.8 percentage-point difference from the other two categories. These “None” districts are also significantly more likely to be charter and rural districts. Meanwhile, a higher proportion of “reading deficient” students is related to an increased likelihood of being in the “Some” group by 24.8 percentage points. These districts are also significantly less likely to be charter or rural, opposite the “None” group. Table 2.4 District Characteristics Related to “Read at Home” Plan Implementation Proportion with “Reading Deficiencies” ELA Achievement Proportion Non-White Proportion Economically Disadvantaged Proportion English Learners Proportion Students with Disabilities Log Enrollment Charter Rural Urban 2020-21 2021-22 2022-23 (2) Some 0.248*** (0.062) -0.000545 (0.002) -0.12 (0.075) -0.0349 (0.107) -0.00295 (0.106) 0.102 (0.271) 2.61E-02 (0.020) -0.121** (0.054) -0.0666* (0.038) -0.0022 (0.047) 0.00615 (0.018) 8.32E-04 (0.021) 0.014 (0.021) (3) All -0.0495 (0.036) -0.00183** (0.001) 0.0276 (0.044) -0.104** (0.052) -0.00506 (0.051) -0.154 (0.135) -0.0161* (0.010) -0.024 (0.028) -0.0176 (0.019) -0.00723 (0.025) 0.0403*** (0.015) 0.0520*** (0.015) 0.0468*** (0.015) (1) None -0.198*** (0.072) 0.00238 (0.002) 0.092 (0.088) 0.139 (0.123) 0.00801 (0.121) 0.0518 (0.310) -9.99E-03 (0.023) 0.145** (0.062) 0.0842* (0.044) 0.00943 (0.055) -0.0464*** (0.016) -0.0528*** (0.020) -0.0608*** (0.021) 85 Table 2.4 (cont’d) N District-Years in Category (1) None 1,575 (2) Some 632 (3) All 156 In the case of the “None” districts, the lower proportions of students identified as “reading deficient” suggest several possibilities. One could be that these districts, with fewer students needing “Read at Home” plans, have not prioritized the plans, or allocated sufficient time and resources to developing them. Alternatively, it might indicate knowledge gaps, as fewer “reading deficient” students may lead to less emphasis on understanding the necessary interventions for these students compared to districts with higher numbers of them. (Indeed, educators in these districts report a lower understanding of “Read at Home” plans in the survey data.) Districts in the “Some” group have a higher proportion of “reading deficient” students. This pattern could potentially relate to various factors, including possible capacity constraints, though further research would be needed to confirm this hypothesis. Nonetheless, these findings call for further exploration into the specific students given “Read at Home” plans within these districts. RQ2: Factors That Distinguish Which Students Are Given Plans The results from the LPMs in Table 2.5 provide insights into which students are more likely to be given “Read at Home” plans in districts that provide them to some but not all eligible students. The data illustrate a shift in the likelihood of being given a plan based on whether teacher, school, or district fixed effects are included in the model. In Column (1), which only includes student characteristics and year fixed effects, Asian students and students with disabilities are significantly more likely to be given plans than White students and students without disabilities. However, when teacher fixed effects are introduced in Column (2), Black 86 and economically disadvantaged students become significantly more likely to be given “Read at Home” plans, while the coefficient for Asian students is no longer significant or positive. This suggests that the significance for Asian students in the initial model may have been influenced by teacher behaviors that vary from one classroom to another. Some teachers, for instance, could be more proactive in assigning “Read at Home” plans, and if Asian students are more common in those teachers’ classes, it would look like being Asian is linked to getting a plan. But when we account for these teacher-specific factors, that link disappears, indicating that it was more about the teachers’ actions than the students’ characteristics. On the other hand, the fact that Black and economically disadvantaged students are more likely to be given plans, even after considering teacher fixed effects, suggests there is something more systematic at play for these students. It implies that beyond any individual teacher’s influence, there are consistent factors tied to being Black or economically disadvantaged that are related to an increased chance of getting a “Read at Home” plan. This may reflect a positive alignment with the literature suggesting these groups can significantly benefit from family engagement (Fikrat-Wevers et al., 2021; Jeynes, 2003, 2016, 2017). However, it could also reveal negative biases or stereotypes, with some educators possibly assuming that families in these groups are less involved with their children’s literacy or have less supportive home environments (Foster et al., 2017; Grice, 2020; Patton, 2019). This pattern persists in Column (3) with the addition of school fixed effects, suggesting that these factors operate at both the teacher and school levels. When district fixed effects are introduced in Column (4), Black students continue to be significantly more likely to be given “Read at Home” plans, underscoring a robust, district-wide pattern of targeted support. In contrast, while the coefficient for economically disadvantaged students remains similar in 87 Table 2.5 Factors Related to Whether a Student is Given a “Read at Home” Plan 3rd Grade Female Black Hispanic Asian Other Race Econ. Dis. English Learner (1) -0.00047 (0.012) 0.003 (0.004) 0.0553 (0.050) -0.00645 (0.038) 0.100* (0.055) -9.49E-05 (0.026) -0.0206 (0.025) 0.0593 (0.044) (2) -0.00262 (0.018) -9.55E-05 (0.002) 0.0116*** (0.004) -0.00592 (0.004) -0.0109 (0.009) 0.00216 (0.004) (3) 0.00199 (0.008) -0.00099 (0.002) 0.0104** (0.005) -0.00487 (0.004) -0.00965 (0.007) 0.00334 (0.005) 0.00696** 0.00795*** (0.003) -0.00498 (0.006) (0.003) -0.00497 (0.005) (6) -0.00051 (0.017) -0.0001 (0.002) (4) (5) 0.00663 -0.00519 (0.010) (0.019) 0.000409 0.000121 (0.002) (0.003) 0.0136** 0.0111*** 0.0113*** (0.004) (0.006) -0.00504 -0.0022 (0.004) (0.006) -0.00821 -0.00957 (0.008) (0.013) 0.00402 0.00754 (0.005) (0.004) 0.00717 (0.005) -0.0165* (0.010) (0.004) -0.00386 (0.004) -0.00419 (0.007) 0.00421 (0.004) (0.003) -0.00514 (0.006) (0.003) -0.00333 (0.005) (7) 0.00199 (0.008) -0.00099 (0.002) 0.0104** (0.005) -0.00487 (0.004) -0.00965 (0.007) 0.00334 (0.005) (8) -0.00429 (0.018) 0.000277 (0.002) 0.0112*** (0.004) -0.00446 (0.004) -0.00598 (0.007) 0.00471 (0.004) (0.003) -0.00497 (0.005) (0.003) -0.00333 (0.005) 0.00722** 0.00749** 0.00795*** 0.00765** Special Education 0.0271*** 0.0166*** 0.0162*** 0.0203*** 0.0163*** 0.0164*** 0.0162*** 0.0163*** Constant (0.010) 0.547*** (0.059) (0.005) 0.530*** (0.034) X (0.004) 0.523*** (0.031) (0.005) 0.515*** (0.032) Teacher FE School FE District FE 109,618 Observations Note. All models include year fixed effects. Standard errors in parentheses (*** p<0.01, ** p<0.05, * p<0.1). X X 109,618 X 109,623 X 107,484 107,432 107,430 109,623 X (0.005) 0.531*** (0.035) X X (0.005) 0.528*** (0.035) X (0.004) 0.523*** (0.031) (0.005) 0.530*** (0.035) X X X 107,481 88 magnitude, it no longer reaches statistical significance. This suggests that the relationship between economic disadvantage and receiving “Read at Home” plans may vary across districts, possibly due to differing priorities or resources. Additionally, ELs become significantly less likely to be given plans with the inclusion of district fixed effects. This could be due to the plans not being consistently available in the students’ home languages at the district level, as this is not a requirement of the Read by Grade Three Law. However, individual schools and teachers, who may have a better understanding of each family’s language background, may be better positioned to address these gaps by tailoring the plans. This could explain why the differences between ELs and non-ELs are not significant when comparisons are made within teachers and schools in Columns (2) and (3). When varying levels of fixed effects are combined in Columns (5) through (8), the results are consistent with Columns (2) and (3), where Black and economically disadvantaged students are significantly more likely to be given “Read at Home” plans than their White, non- economically disadvantaged peers. These results suggest that there may be factors at the school and teacher levels associated with plan implementation, though the specific mechanisms behind these associations warrant further investigation. Students with disabilities are consistently more likely to be given plans across all model specifications. This could be attributed to educators aligning “Read at Home” plans with IEPs, leveraging existing processes to provide students and their families with additional resources. It might also be driven by a concerted effort to meet legal obligations for these students; the prospect of legal repercussions for failing to comply with every aspect of service provision under the law could motivate a more diligent application of “Read at Home” plans for students with disabilities. Notably, 3rd graders are no more likely to be given plans than K-2 students, an 89 unexpected finding given that identified 3rd graders are required to receive plans under the law. RQ3: Educators’ Understanding and Perceptions of “Read at Home” Plans The RQ3 analyses shed light on educators’ understanding and perceptions of “Read at Home” plans, revealing the critical role of sensemaking in the implementation of these plans. The descriptive results presented in Figure 2.4 show that educators’ understanding and perceptions of “Read at Home” plans vary across groups. Panel A indicates that both superintendents and principals report understanding “Read at Home” plans better than teachers, with statistically significant differences between teachers and principals in both years (p<0.001) and between teachers and superintendents in 2022-23 (p<0.001). In the 2019-20 school year, principals reported a significantly better understanding than superintendents (p<0.01), a difference that disappeared by 2022-23 due to a significant improvement in superintendents’ understanding (p<0.001). This discrepancy in understanding among educators suggests that while administrators may have a broader perspective on “Read at Home” plans, teachers might not have an equally comprehensive understanding. These differences could be attributed to the filtering process conducted by district and school leaders, potentially limiting the information about “Read at Home” plans that reaches teachers. This points to a potential oversight in ensuring that teachers, who are crucial to effectively delivering these plans as the individuals directly interfacing with students and families, receive adequate information about them. Panel B illustrates a contrasting pattern where most teachers, despite reporting lower levels of understanding, consistently perceive “Read at Home” plans as more effective than administrators. Teachers’ perceptions of the effectiveness of these plans were significantly higher than principals’ in all three years (p<0.001) and superintendents’ in 2019-20 (p<0.01). This discrepancy might arise because teachers may perceive these plans as effective due to a 90 Figure 2.4 Educators’ Understanding and Perceptions of “Read at Home” Plans Note. Asterisks above the bars indicate significant differences between teachers and principals/superintendents each year (e.g., principals vs. teachers in 2019-20) with significance levels denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001. Significant differences between years within groups (e.g., teachers in 2019-20 vs. 2022-23) are discussed in the accompanying text. 91 50%80% 70%39%81% 82% 0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected Moderately or Very WellPanel AUnderstanding of Read at Home Plans2019 202022 2361%53% 43% 52%40% 49%57%43% 49%0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected to a Moderate or Great ExtentPanel BEffectiveness of Read at Home Plans2019 202021 222022 2322%18%0%10%20%30%40%50%60%70%80%90%100%2019 202020 21Percent Who Selected Agree or Strongly AgreePanel CSufficient Time to Implement Read at Home Plans general belief in the importance of family literacy activities, without necessarily comprehending the specific contents or mechanisms of “Read at Home” plans. Panel C reveals that many teachers feel they lack sufficient time to develop “Read at Home” plans. Only about a fifth of teachers felt they had enough time for these plans in both the 2019-20 (18%) and 2020-21 (22%) school years, which is unsurprising given time constraints are a common issue for SLBs. The proportion of teachers feeling pressed for time was significantly higher in 2020-21 than in 2019- 20 (p<0.001), a difference likely due to the heightened workload and challenges educators faced during the COVID-19 pandemic. The LPM results from Model (4) in Table 2.6 shed light on how educator characteristics and district demographics relate to their understanding and perceptions of “Read at Home” plans. Third-grade teachers, despite having similar levels of understanding as K-2 teachers, are significantly less likely to view the plans as effective or report sufficient time for implementation. This may stem from the heightened pressure on 3rd-grade teachers, whose “reading deficient” students are mandated to receive the plans. Furthermore, during the years of the study, the Read by Grade Three Law mandated retention for 3rd graders testing more than one grade level behind in reading on the state assessment, adding additional pressure on these teachers to ensure they were adequately prepared (Michigan Public Act 306, 2016). Consequently, 3rd-grade teachers may feel stretched for time as they balance covering necessary instruction with the demands of implementing “Read at Home” plans. Teachers in districts with a larger proportion of “reading deficient” students also report feeling constrained by time, possibly due to the increased workload associated with needing to develop a greater number of plans. Perhaps unsurprisingly, newer teachers show lower levels of understanding of the plans but are more optimistic about their effectiveness and ability to implement them. Meanwhile, teachers 92 who are rated highly effective or have ELA endorsements report a greater understanding of the plans but are simultaneously more skeptical about their effectiveness and their own implementation capacities. Finally, the LPM results from Model (5) in Table 2.7 indicate a strong relationship between educators’ understanding and perceptions and the likelihood of students being given “Read at Home” plans, specifically for students in districts that provide these plans to some but not all eligible students. As shown in Column (1), students are a striking 21.2 percentage points more likely to be given a plan if their teacher reports a moderate or great understanding of them. This could imply that having a deeper understanding of the plans makes teachers more likely to implement them. However, it is also possible that teachers who are implementing the plans have developed a greater understanding of them through the implementation process. Additionally, in Column (2), students whose principals regard the plans as effective for improving student achievement are 22.5 percentage points more likely to receive them. This may suggest that principal buy-in is a critical driver of the implementation of “Read at Home” plans. However, it is also possible that principals believe the plans are effective because they observe their successful implementation in their schools. In Column (3), whether teachers feel they have sufficient time to create these plans does not seem to be related to the likelihood of students receiving them, possibly reflecting the overall agreement that there is insufficient time for plan creation. Nonetheless, these findings indicate that educators’ sensemaking surrounding “Read at Home” plans is strongly related to their implementation. 93 Table 2.6 Relationship between Educator Characteristics and Perceptions of “Read at Home” Plans 3rd Grade Teacher Hired Last 5 Years Highly Effective ELA Endorsement District Prop. “Reading Def.” District Prop. Non-White District Prop. Econ. Dis. District Prop. EL District Prop. SWD Log District Enrollment Charter Rural Teacher 0.0067 (0.010) -0.0270** (0.012) 0.0515*** (0.013) 0.0429*** (0.011) 0.018 (0.039) 0.0159 (0.045) 0.00385 (0.050) 0.0127 (0.062) -0.0563 (0.175) -0.0372*** (0.013) -0.0897** (0.037) 0.0095 (0.022) Understanding Principal Supt. Teacher Effectiveness Principal Supt. Sufficient Time Teacher -0.0463*** (0.009) 0.0485*** (0.011) -0.0388*** (0.010) -0.0189* (0.010) 0.034 (0.030) -0.0528 (0.034) -0.06 (0.044) -0.0489 (0.055) -0.151 (0.147) 0.0188** (0.008) 0.116*** (0.024) -0.00129 (0.017) -0.0247 (0.034) -0.0527 (0.155) -0.015 (0.078) -0.0315 (0.086) 0.142 (0.126) -0.0549 (0.140) -0.385 (0.415) 2.87E-02 (0.023) 0.079 (0.070) 0.0846* (0.047) 0.179 (0.309) -0.499 (0.379) -0.282 (0.336) 0.678* (0.369) -0.479 (0.966) -2.74E-02 (0.073) -0.14 (0.224) -0.304 (0.230) -0.0285*** (0.008) 0.0367*** (0.010) 0.0155 (0.009) -0.0251*** (0.009) -0.0538* (0.028) -0.00759 (0.033) 8.49E-03 (0.037) 0.0332 (0.056) -0.0683 (0.131) 0.00673 (0.010) 0.0517** (0.025) 4.38E-03 (0.015) -0.0276 (0.026) -0.0925 (0.159) -0.094 (0.065) -0.0446 (0.077) -0.0763 (0.088) -0.0571 (0.132) -0.471 (0.321) -0.0316* (0.018) 0.0624 (0.064) -0.0493 (0.043) -0.095 (0.307) 0.0724 (0.336) 0.173 (0.535) 0.362 (0.651) 0.48 (1.009) 5.23E-03 (0.065) 0.0618 (0.192) 0.173 (0.202) 94 Table 2.6 (cont’d) Understanding Principal Teacher Supt. Teacher Effectiveness Principal Supt. Sufficient Time Teacher Urban Constant -0.0747 (0.048) 1.210*** (0.151) Observations 931 Note. Standard error in parentheses (*** p<0.01, ** p<0.05, * p<0.1). All models include year fixed effects. -0.00894 (0.029) 0.732*** (0.099) 10,522 0.0102 (0.019) 0.551*** -0.0688 14,086 -0.104 (0.244) 0.335 (0.590) 55 0.0578 (0.054) 0.295 -0.193 1,192 0.527** (0.203) 1.103* -0.573 65 -0.00754 (0.016) 0.183** -0.0736 11,398 95 Table 2.7 Relationship between Educators’ Perceptions and Implementation of “Read at Home” Plans (1) 0.212** (0.092) 0.0597 (0.214) Tch. Understanding Prin. Understanding Tch. Belief in Eff. Prin. Belief in Eff. Tch. Perception Sufficient Time (2) (3) 0.013 (0.062) 0.225** (0.111) Black Asian Female Hispanic 3rd Grade 0.0214 (0.083) 0.0309 (0.032) -0.101 (0.145) -0.172* (0.085) 0.0858 (0.169) -0.0985 (0.075) 0.131** (0.058) -0.0654 (0.107) 0.122*** (0.040) 828 Observations Note. Models include year fixed effects where relevant. Standard errors in parentheses (*** p<0.01, ** p<0.05, * p<0.1). (0.002) (0.073) -0.0204 (0.019) (0.031) (0.092) -0.0918 (0.078) 0.256 (0.175) -0.0358 (0.057) 0.0493 (0.039) -0.155** (0.063) 0.0398 (0.029) 2,159 -0.032 (0.060) -0.0101 (0.038) 0.017 (0.011) 0.00799 (0.079) 0.0121 (0.057) -0.0409 (0.093) 0.0185 (0.047) 0.0206 (0.040) 0.0381 (0.055) 0.0656*** (0.021) 7,398 Economically Disadvantaged Special Education English Learner Other Race RQ4: Other Family Engagement Activities RQ4 reveals that educators frequently engage in several other family literacy activities that go beyond “Read at Home” plans. While the administrative data show that districts report 96 providing just 21% of “reading deficient” students with “Read at Home” plans, survey data illustrated in Figure 2.5 highlight that educators report encouraging regular home literacy in other ways at much higher rates. Across stakeholder groups and survey years, 82% agreed to a moderate or great extent that they encourage home reading and writing activities, send home literacy resources like books and writing supplies (73%), meet with families to communicate students’ progress in literacy (69%), provide research-based guidance on how to support literacy at home (50%), and host family literacy workshops (e.g., family literacy nights; 27%). This suggests that Michigan educators are invested in fostering family engagement in literacy, perhaps in more diverse and interactive methods than required by the Read by Grade Three Law or evidenced in the district “Read at Home” plans I reviewed. Figure 2.6 presents a more nuanced picture of educators’ engagement in these activities over time and across groups, with distinct patterns emerging in conjunction with the COVID-19 pandemic. While encouraging home literacy activities (Panel A) and sending home literacy resources (Panel B) was high among all educator groups, there were notable declines during the pandemic (p<0.001 for teachers and principals between 2019-20 and 2020-21). Educators were also significantly less likely to meet with families about students’ literacy progress (Panel C; p<0.001 for teachers), provide research-based guidance on how to support literacy at home (Panel D; p<0.001 for teachers, p<0.10 for principals), and host family literacy workshops (Panel E; p<0.001 for all groups) in 2020-21 compared to 2019-20. One exception to these trends was that principals were significantly more likely (p<0.001) to meet with families to communicate students’ literacy progress in 2020-21 than they were in 2019-20, suggesting a reallocation of responsibilities during this period. 97 Figure 2.5 Frequency with Which Educators Report Engaging in Family Engagement Activities Note. The percentage for “Read at Home” plans is derived from administrative data (the average proportion of “reading deficient” students to whom districts reported providing “Read at Home” plans, pooled across grade levels and years). The other percentages are derived from educators’ survey responses about the extent to which they reported engaging in each family engagement activity, pooled across educator groups and years. 98 21%82%73%69%50%27%0%10%20%30%40%50%60%70%80%90%100% Read at Home PlansEncouraging Regular Home LiteracyActivities (e.g., Reading, Writing)Sending Home Literacy Resources(e.g., Books, Writing Supplies)Meeting to Communicate Students Progress in LiteracyProviding Research Based Guidanceon How to Support LiteracyProviding Family Literacy WorkshopsPercent Providing This Intervention (to a Moderate or Great Extent)Family Engagement Interventions Figure 2.6 Extent to Which Educators Report Engaging in Other Family Literacy Interventions 99 91%88% 85% 69%83% 88% 0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected to a Moderate or Great ExtentPanel AEncouraging Regular Home Literacy Activities2019 202020 2177%88% 86% 64%80% 91% 0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected to a Moderate or Great ExtentPanel BSending Home Literacy Resources2019 202020 2176%25% 58%76% 69%84% 91% 74%88% 94% 0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected to a Moderate or Great ExtentPanel CMeeting to Communicate Students Progress in Literacy2019 202020 212021 222022 23 Figure 2.6 (cont’d) Note. Asterisks above the respective bars indicate statistically significant differences between teachers and principals/superintendents each year (e.g., principals vs. teachers in 2019-20) with levels of significance denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001. Significant differences between years within groups (e.g., teachers in 2019-20 vs. 2022-23) are discussed in the accompanying text. On one hand, the decrease in these activities is not entirely surprising, given the significant additional demands placed on educators. The pandemic stretched educational resources thin and required teachers and administrators to adapt quickly to new modes of teaching, perhaps requiring them to prioritize immediate instructional needs over additional family engagement efforts. On the other hand, the decline is somewhat unexpected given the context of the pandemic, where many students were learning at home for extended periods. One 100 53%70% 72% 38%64% 72% 45%71% 72% 53%74% 75% 0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected to a Moderate or Great ExtentPanel DProviding Research Based Guidance on How to Support Literacy2019 202020 212021 222022 2333%51% 65% 15%30% 47% 0%10%20%30%40%50%60%70%80%90%100%TeachersPrincipalsSuperintendentsPercent Who Selected to a Moderate or Great ExtentPanel EProviding Family Literacy Workshops2019 202020 21 might anticipate educators prioritizing activities supporting home learning and early literacy, recognizing the unique opportunity to engage families in their children’s education. This suggests potential missed opportunities for strengthening the home-school connection during a time when family involvement in education was critical. These results also suggest that the pandemic’s impact may have been more pronounced on these more interactive forms of family engagement compared to the relatively stable implementation of “Read at Home” plans across the four years of the study, as described above. While “Read at Home” plans, which appear more standardized, could potentially be distributed with relative consistency, the more dynamic, interpersonal aspects of family engagement appear to have faced significant disruptions. The reliance on virtual communication platforms, social distancing measures, and the overall stress and uncertainty of the pandemic likely contributed to these challenges. Indeed, as pandemic restrictions began to ease in the 2021-22 and 2022-23 school years, there was a noticeable rebound in these more interactive family engagement practices. Finally, Table 2.8 highlights significant disparities in family engagement across different educator characteristics. Notably, 3rd-grade teachers and those in districts with high proportions of “reading deficient” students are significantly less likely to report engaging in all activities compared to their K-2 counterparts and those in districts with fewer identified students. This may stem from the pressures 3rd-grade teachers face under the Read by Grade Three Law, leading them to prioritize direct instructional time over family engagement efforts. Moreover, the heavy demands on teachers in districts with significant needs for “reading deficient” students could lead to burnout and overwhelm, further diminishing their capacity to engage in additional family literacy activities. At the same time, highly effective teachers, and to some extent ELA-endorsed 101 teachers, are significantly more likely to engage in these activities compared to their lower-rated counterparts. Relating this back to educators’ sensemaking surrounding “Read at Home” plans, it appears that highly effective teachers, who do not perceive the plans as effective, are instead directing their efforts to other activities they may either consider more effective or that are more ingrained in their everyday practices and routines. Interestingly, newer teachers, although less likely than experienced ones to send home literacy resources (perhaps because they have not yet accumulated them for their classrooms), are significantly more likely to provide research-based guidance on supporting literacy and host family literacy workshops, suggesting that they are engaging with families in alternative ways. When considering district characteristics, there are fewer consistent, significant results, suggesting that engagement in these activities may be more related to educators’ perceptions of them. Discussion This study investigated the implementation of “Read at Home” plans under Michigan’s Read by Grade Three Law, examining how these plans are implemented across districts, the characteristics of the students receiving them, and the factors related to their distribution, particularly educators’ sensemaking surrounding the plans. The findings uncovered a low statewide implementation rate, with only about one-fifth of students identified as “reading deficient” provided plans, including those in 3rd grade who are mandated by law to receive them. These low implementation levels highlight the inherent limitations of relying on policy mandates as an instrument for change. While mandates operate on the assumption that directives will be followed due to established hierarchical compliance expectations, the findings here suggest that mandates without additional supports and incentives face significant barriers to implementation. 102 Table 2.8 Relationship between Educator Characteristics and Other Family Literacy Activities Encouraging Home Lit. Activities Sending Home Lit. Resources Meeting w/ Families Research-Based Guidance Family Literacy Workshops Panel A: Teachers 3rd Grade Teacher Hired Last 5 Years Highly Effective ELA Endorsement District Prop. “Reading Def.” District Prop. Non-White District Prop. Econ. Dis. District Prop. EL District Prop. SWD Log District Enrollment Charter Rural -0.103*** (0.009) 0.00436 (0.008) 0.0268*** (0.008) 0.0267*** (0.007) -0.0410** (0.020) 0.0157 (0.024) -0.0517** (0.026) -0.0202 (0.034) 0.173* (0.101) 6.25E-03 (0.006) 0.0186 (0.017) 0.0220* (0.013) -0.122*** (0.007) 0.0125* (0.008) 0.0344*** (0.008) 0.0088 (0.008) -0.0726*** (0.024) 0.0248 (0.032) -0.0659** (0.032) 0.0685 (0.050) 0.00108 (0.102) 7.65E-03 (0.007) 0.0331* (0.020) 0.00756 (0.014) -0.0877*** (0.008) 0.0145* (0.008) 0.0336*** (0.009) 0.0244*** (0.009) -0.0757*** (0.024) -0.00321 (0.035) 0.0275 (0.034) -0.0309 (0.052) -0.0583 (0.118) 6.95E-03 (0.007) 0.0317 (0.023) 0.00672 (0.015) -0.0230** (0.011) 0.0266** (0.011) 0.015 (0.011) -0.013 (0.009) -0.00636 (0.036) 0.0201 (0.042) 0.171*** (0.046) 0.0687 (0.065) -0.0974 (0.148) 3.42E-03 (0.009) -0.00885 (0.026) 0.00253 (0.017) -0.178*** (0.011) -0.0276*** (0.010) 0.0324*** (0.009) 0.00896 (0.009) -0.0466* (0.028) -0.0235 (0.032) 0.0334 (0.035) 0.0773 (0.050) 0.0483 (0.126) -1.69E-03 (0.007) -0.0326 (0.023) 0.0182 (0.014) 103 Table 2.8 (cont’d) Urban Constant Observations Panel B: Principals Hired Last 5 Years District Prop. “Reading Def.” District Prop. Non-White District Prop. Econ. Dis. District Prop. EL District Prop. SWD Log District Enrollment Charter Rural Urban Encouraging Home Lit. Activities 8.30E-03 (0.013) 0.876*** (0.053) 11,382 -0.0332 (0.024) -0.0194 (0.055) 0.116* (0.066) -0.0949 (0.085) 0.156 (0.096) 0.394 (0.285) 3.81E-03 (0.016) 0.0353 (0.052) 0.0760** (0.037) -2.78E-02 (0.042) Sending Home Lit. Resources Meeting w/ Families Research- Based Guidance Family Literacy Workshops 5.69E-03 (0.016) 0.761*** (0.053) 19,897 -0.0673*** (0.023) -0.035 (0.054) 0.0013 (0.062) 0.0407 (0.074) 0.095 (0.092) -0.100 (0.270) 0.0261** (0.013) 0.0747 (0.047) 0.0379 (0.031) 2.18E-02 (0.033) -8.25E-03 (0.017) 0.486*** (0.061) 19,863 -0.0179 (0.027) 0.0311 (0.067) -0.0667 (0.076) 0.108 (0.096) 0.0503 (0.123) -0.264 (0.321) 1.12E-03 (0.020) 0.0289 (0.059) 0.0361 (0.039) -4.11E-02 (0.041) 1.72E-02 (0.026) 0.198*** (0.071) 11,337 -0.00267 (0.034) -0.00672 (0.098) 0.192* (0.113) 0.222 (0.145) 0.303** (0.145) -0.439 (0.430) -4.59E-03 (0.023) -0.0748 (0.076) 0.0276 (0.048) -6.58E-02 (0.063) 1.18E-02 (0.019) 0.807*** (0.058) 11,392 -0.0346 (0.025) -0.0879 (0.058) 0.0951 (0.066) -0.0516 (0.083) 0.190* (0.101) 0.384 (0.304) -1.12E-02 (0.016) 0.00397 (0.049) 0.0414 (0.032) -4.24E-02 (0.041) 104 Table 2.8 (cont’d) Constant Encouraging Home Lit. Activities Sending Home Lit. Resources Meeting w/ Families Research-Based Guidance Family Literacy Workshops 0.801*** (0.127) 979 0.917*** (0.135) 981 0.072 (0.117) 1,562 0.675*** (0.167) 1,561 0.402* (0.206) 980 Observations Note. T = Teachers, P = Principals. All models include year fixed effects. Standard errors in parentheses (*** p<0.01, ** p<0.05, * p<0.1). 105 In the case of “Read at Home” plans, the policy mandate lacks accompanying capacity- building measures such as supplemental funding, templates to guide plan development, training to build educator understanding and buy-in, or explicit sanctions to compel implementation. Without supportive resources, the personnel, material, and operational demands of developing and disseminating the plans falls on already constrained educator capacities. Moreover, the policy’s ambiguity around consequences for non-compliance may further diminish the sense of accountability and perceived implementation urgency. While self-reporting requirements are in place, the absence of explicitly outlined repercussions for failing to provide “Read at Home” plans may have signaled a lack of high-stakes oversight that could have otherwise motivated stronger implementation. Facing this dearth of capacity-building supports and lacking clear accountability pressures, educators are largely left to their own devices in determining whether and how to implement the plans. This lack of guidance and oversight might help explain such variable implementation of “Read at Home” plans across districts. Approximately two-thirds do not provide plans to any of their “reading deficient” students, while a quarter provide them to some but not all, and only around 6% provide plans to all eligible students. This variability is strongly associated with the proportion of students identified as “reading deficient;” districts with more identified students tend to offer plans to only a subset of those eligible, highlighting potential challenges in providing these supports. This finding aligns with Bovens and Hart’s (1996) observation that when many individuals are eligible for a policy, the resources available per capita decrease, constraining implementation efforts. In these districts with a high number of eligible students, teachers, acting as SLBs, exercise discretion in determining who receives a plan. The results of this study suggest that 106 these plans are being targeted to students from historically underserved backgrounds, with Black, economically disadvantaged, and students with disabilities significantly more likely to be given plans than their White, wealthier, and non-disabled peers across almost all model specifications. This indicates that teachers may exercise positive discrimination by prioritizing students whom the literature identifies as particularly benefiting from family engagement interventions. However, it could also reveal underlying biases if teachers are more likely to provide plans to these students because they perceive that their families are not already engaged in literacy activities at home. The analysis also underscores the crucial role of educators’ sensemaking in the implementation of “Read at Home” plans and highlights the potential influence of filtering. In districts where plans are provided to some but not all eligible students, principal buy-in emerges as a critical factor related to whether students are given a plan. Specifically, students are significantly more likely to be given a “Read at Home” plan when their principals perceive them as effective in improving student achievement. Additionally, students are significantly more likely to be given a plan when their teachers report a better understanding of the plans. Compared to principals and superintendents, teachers consistently report the lowest understanding of “Read at Home” plans. As information about the plans filters down from district to school to teacher, teachers may receive the least information, especially if their principal or superintendent has not prioritized the plans. This could be indicative of informational gaps surrounding “Read at Home” plans. Finally, implementing a new policy is a cognitively demanding task as implementers must integrate it into their existing routines (Coburn, 2004; Spillane et al., 2002b). Educators report engaging in a broad range of family literacy activities more frequently than they 107 implement “Read at Home” plans, including sending home literacy resources, holding meetings with families about students’ progress, and organizing family literacy events. This suggests that “Read at Home” plans may be underutilized because they do not align with educators’ existing family engagement practices. However, these results also indicate that despite low implementation of “Read at Home” plans, Michigan educators actively engage families in their children’s literacy development in many ways. Limitations While this study offers important insights about the implementation of “Read at Home” plans, it also faces several limitations. First, this study does not systematically examine the content or quality of the plans themselves. While I reviewed a non-random selection of plans districts posted online, this is unlikely to encompass the full diversity and scope of plans implemented across the state. To truly understand the potential effectiveness of these plans, a comprehensive analysis of their design, content, and alignment with best practices in literacy and family engagement is essential. Second, the administrative data on “Read at Home” plans and the survey data used in this study capture educators’ reported implementation of these plans and other family engagement activities. These sources do not consider whether students or families received or executed the plans, nor do they provide insights into their experiences or participation in other family engagement activities. This presents a substantial gap in our comprehension of families’ take-up and use of “Read at Home” plans, which will ultimately influence their effectiveness. Third, while the administrative data provide valuable insights into which students are given “Read at Home” plans, they fall short of illuminating the decision-making processes surrounding plan implementation at different levels of the educational system. Without a deeper 108 understanding of how and why certain students are selected for these plans, our grasp of educators’ sensemaking and implementation dynamics remains incomplete. Fourth, because of the available data and necessary sample restrictions, the results are only generalizable to students who have been identified with “reading deficiencies.” Therefore, this study cannot draw conclusions about how “Read at Home” plans are or would be implemented among relatively higher-performing students. Fifth, the Michigan-specific context may not mirror other states with “Read at Home” plan requirements but different demographic, geographic, or political conditions. This may impact the generalizability of the results to other states with similar early literacy policies. Finally, the study does not link “Read at Home” plans with student achievement, leaving unanswered questions about the plans’ effectiveness in enhancing literacy and the role of implementation in this effect. Directions for Future Research These limitations highlight areas for future research. First, future research could systematically examine “Read at Home” plans through a combination of qualitative research methods, including content analysis and interviews. Content analysis could provide detailed insights into the alignment of these plans with literacy best practices and family engagement strategies, while interviews with district leaders responsible for designing the plans could illuminate the decision-making processes behind their structure and content. This would offer a deeper understanding of how “Read at Home” plans are intended to function and their potential for effectively supporting literacy development. Second, future research could focus on gathering data directly from students and families to assess their receipt and use of “Read at Home” plans and participation in other family 109 engagement activities, either through surveys, interviews, or direct observation. Despite the challenges of collecting and analyzing data from this population, incorporating the perspectives of these key stakeholders could provide researchers a more comprehensive understanding of families’ implementation and perceptions of these initiatives. Third, researchers could use similar data-collection methods to understand the decision- making processes behind the distribution of “Read at Home” plans. Capturing the perspectives of those involved in their implementation, including both district and school leaders as well as teachers, could yield rich insights into the considerations influencing the implementation of these plans and which students receive them. This would not only complement the present study’s quantitative analysis but also offer a deeper understanding of the challenges, opportunities, and potential equity implications associated with implementing “Read at Home” plans. Fourth, to broaden the scope of generalizability, future studies could investigate the implementation of “Read at Home” plans and other family engagement initiatives among students across various performance levels, not just those identified with “reading deficiencies.” This could reveal insights into the applicability of family engagement strategies for a broader range of students, potentially leading to more inclusive and comprehensive approaches to improving literacy outcomes. Fifth, future research could examine the implementation of “Read at Home” plans in states with varying requirements and guidance and different demographic, geographic, or political contexts—all of which could influence the implementation of these plans. Comparing implementation strategies and outcomes across a diverse set of states could help uncover best practices and policy insights that could inform more effective family engagement initiatives and early literacy policies nationwide. 110 Finally, to evaluate the effectiveness of these plans in enhancing literacy, future studies could establish a direct link between plan implementation and student achievement outcomes. By examining the impact of “Read at Home” plans on literacy proficiency and academic performance, researchers can provide valuable insights into their overall effectiveness and inform evidence-based policy recommendations. This analysis would be particularly informative if it included a comparison with other interventions that early literacy policies require, such as additional instructional time, summer support, or one-on-one/small group tutoring. Such comparative studies could identify the most impactful strategies for improving early literacy outcomes, guiding policymakers and educators in their efforts to support student learning. Policy and Practice Recommendations This research suggests concrete steps for policymakers and practitioners to enhance the implementation of “Read at Home” plans through collaborative efforts between the state and local educators. The state has established a foundation for family engagement in literacy by mandating these plans under the Read by Grade Three Law. To build upon this foundation, the state could provide capacity-building supports for implementing “Read at Home” plans. Considering the low statewide implementation rates, such supports could prove beneficial for all districts, particularly those with high proportions of “reading deficient” students. This support could encompass funding for the plans, which might be allocated for additional personnel to oversee their creation and distribution or materials for the plans themselves. It could also involve training and communication regarding the plans, including how to create and implement them, effective methods for distributing them to students to ensure they reach families, and clarifying the criteria for determining who should receive a plan. Training and communication play crucial roles in policy implementation, as they can enhance implementers’ understanding and buy-in for 111 the policy (Hill & Hupe, 2002; May & Winter, 2009; Spillane et al., 2002b; Turnbull, 2002). Given the study’s findings that educators’ understanding and perceptions are strongly related to plan implementation, such supports could ensure that more students receive family engagement efforts as envisioned by the law. MDE could further enhance capacity-building for “Read at Home” plan implementation by developing a template that districts could use as a reference and adapt to their unique contexts. This template could draw inspiration from states like Florida, Kentucky, and Mississippi, whose education departments currently offer similar resources to districts (Florida Department of Education, n.d.; Kentucky Department of Education, n.d.; Mississippi Department of Education, n.d.). It could serve as a valuable tool to assist educators in crafting plans that not only meet the Read by Grade Three Law’s requirements but also incorporate best practices in family engagement. To maximize effectiveness, MDE could incorporate educators’ insights and experiences with family engagement into the development of this “Read at Home” plan template. Survey responses from Michigan educators indicate their dynamic engagement with families in literacy in many ways beyond “Read at Home” plans. Using this professional experience in the development of new resources could yield recommendations that not only support students identified as “reading deficient” but also resonate more deeply with educators’ established practices. Aligning new policies with existing practices fosters successful implementation by garnering buy-in and facilitating integration into daily educational routines (Bridwell-Mitchell, 2015; Spillane, 2000; Spillane et al., 2002). By involving educators in the process, MDE can ensure that “Read at Home” plans not only meet policy requirements but also benefit from the practical insights of frontline implementers. While some policymakers may consider imposing sanctions on districts that fail to 112 implement “Read at Home” plans as required by the Read by Grade Three Law, this study suggests that a more constructive approach would be to focus on fostering understanding and positive perceptions among educators. Sanctions, such as financial penalties or negative performance ratings, may create a compliance-oriented mindset that undermines the authentic engagement and tailoring needed for effective family engagement (McDonnell & Elmore, 1987; Schneider & Ingram, 1990). Moreover, sanctions could further erode educators’ perceptions of the policy, which the findings indicate are significantly related to successful implementation. Instead, investing in resources, training, and templates that support districts and educators in developing and implementing high-quality “Read at Home” plans is likely to yield more meaningful and sustainable improvements in family engagement by fostering a sense of understanding and buy-in among educators. Conclusion This study delves into Michigan’s Read by Grade Three Law and the implementation of “Read at Home” plans, offering insights into gaps and opportunities for enhancing early literacy policies through family engagement. Despite the law’s mandate to provide these plans to students identified with “reading deficiencies,” particularly 3rd graders, their actual reach and remains limited, with only about one-fifth of eligible students given plans. Variation in implementation across districts, coupled with educators’ differing understandings and perceptions of their efficacy, underscores the complex interplay between policy mandates, sensemaking processes, and implementation realities. Moving forward, there are opportunities for policymakers and practitioners to bridge the gap between policy intent and on-the-ground execution. Enhanced state-supported capacity- building initiatives for educators in creating and implementing “Read at Home” plans, alongside 113 the development of templates that not only meet the Read by Grade Three Law’s requirements but also align with best practices in family engagement, could significantly enhance the plans’ effectiveness. As this study sheds light on the challenges and potential pathways for enriching family engagement in early literacy, it underscores the importance of concerted efforts to refine and reinforce these initiatives, ensuring that all students, especially those at risk of falling behind, receive the support they need to succeed academically. 114 PAPER 3: WHO GETS FLAGGED? DISTRICT APPROACHES TO K-3 “READING DEFICIENCY” IDENTIFICATION UNDER MICHIGAN’S READ BY GRADE THREE LAW Amy Cummings, Michigan State University In response to the challenge of low literacy proficiency across the United States, 48 states and the District of Columbia have adopted policies aimed at enhancing early literacy, with a particular emphasis on achieving reading proficiency by the end of 3rd grade (ExcelinEd, 2022a). These early literacy policies commonly include various supports for teachers, progress monitoring for students, and the adoption of evidence-based instruction and interventions (ExcelinEd, 2022a). However, current research primarily examines 3rd-grade retention mandates within these policies, focusing on students who lag behind in literacy as indicated by state English Language Arts (ELA) assessments (Greene & Winters, 2004, 2006, 2007; Schwerdt et al., 2017; Slungaard Mumma & Winters, 2023; Westall & Cummings, 2023). While this body of work confirms that such policies can enhance literacy achievement in the short term, it overlooks additional policy elements that might also contribute to these improvements. A key component and critical first step, present in 38 states’ policies, involves the early identification of K-3 students facing literacy challenges (ExcelinEd, 2022a). These policies mandate that districts identify students with “reading deficiencies” through screening assessments and provide them with additional support.24 While some states require the universal 24 I use the term “reading deficiencies” because that is how Michigan’s Read by Grade Three Law and similar early literacy policies in other states refer to students who have been identified as being behind in reading (Michigan Public Act 306, 2016). 115 use of a single assessment tool, most allow districts to select from a list of approved assessments to screen for literacy difficulties among students. Research consistently emphasizes the importance of early identification of literacy challenges and the use of effective assessments in this process (Barrett et al., 2023; Michigan Assessment Consortium, 2020; National Center on Intensive Intervention, 2021; National Early Literacy Panel, 2008; Rose & Schimke, 2012; Snow et al., 1998; Torgesen, 2004; Truckenmiller et al., 2024). Screening assessments are particularly crucial for identifying “reading deficiencies” as they are designed to determine which students need additional support (e.g., Truckenmiller et al., 2024). Other assessment types, such as diagnostics for pinpointing specific literacy challenges for targeted interventions, progress monitoring for tracking students’ progress toward learning targets, and end-of-year state assessments for measuring overall literacy proficiency, can complement screening assessments (Truckenmiller et al., 2024). This paper focuses on screening assessments as the initial tools districts use to identify students with “reading deficiencies” (ExcelinEd, 2022a), aligning with the study’s aim to examine how districts flag students requiring additional literacy support. Understanding the array of screening assessments permitted under states’ early literacy policies is crucial, as educators must navigate various options to identify students in need of support. Approved screeners come in various formats, including curriculum-based measures (CBMs), which are concise assessments intended to measure students’ proficiency in specific literacy skills; computer-adaptive tests (CATs), which adapt question difficulty based on students’ responses to evaluate their skill level; and informal reading inventories (IRIs), which involve students reading passages aloud and answering comprehension questions. While CBMs and CATs generally outperform IRIs in accurately identifying students, educators often favor IRIs for their practical utility in classroom activities like forming reading groups based on 116 reading level (Barrett et al., 2023; Clemens et al., 2011; Ford & Opitz, 2008; Kilgus et al., 2014; Parker et al., 2015; Truckenmiller et al., 2024). Teachers also commonly use students’ ELA classwork to gauge literacy abilities, but these more subjective evaluations lack the precision needed for formal identification and may lead to misidentification and misaligned interventions (Kenny & Chekaluk, 1993; Kettler & Albers, 2013; Schmitterer & Brod, 2021; Virinkoski et al., 2018). Therefore, understanding educators’ choices in screening assessments is essential for ensuring early literacy policies effectively identify students who require support, allowing for the efficient allocation of resources. Our current understanding of how districts implement these various screening measures and their relationship with student “reading deficiency” identification is limited. This study addresses this gap by examining Michigan’s Read by Grade Three Law as a representative case, given that its identification requirements closely resemble those of many other states, mandating that districts select from a list of state-approved screening assessments. By doing so, this research seeks to uncover how districts carry out these mandates, specifically exploring the range of measures used and their relationships with the proportion and characteristics of students identified as “reading deficient.” To achieve this, this study focuses on two research questions (RQs): 1. What measures do districts use to identify K-3 students with “reading deficiencies,” and how does this vary across districts? 2. How do different identification approaches relate to the rate of students flagged as “reading deficient,” and do the characteristics of these students differ by approach? Using superintendent survey data and administrative records from the Michigan Department of Education (MDE) for the 2021-22 school year, this study employs descriptive 117 statistics and regression analysis to answer these RQs. The results highlight a broad spectrum of strategies for identifying K-3 students with “reading deficiencies” under the Read by Grade Three Law. Most districts use multiple measures to identify students, including combining multiple screening assessments and considering students’ performance on ELA classwork. For assessments, districts predominantly use CATs, especially NWEA, to identify “reading deficiencies.” When employing multiple assessments, districts commonly pair these CATs with CBMs, aligning with research highlighting these assessments as accurate and cost-effective screeners. Despite this, many districts continue to use IRIs or ELA classwork—some exclusively—though these methods lack robust evidence for accurately identifying students needing literacy support. While these measures may be helpful for other instructional purposes, their limited validity as screeners points to opportunities for improvement for policymakers and practitioners. Districts’ approaches to identifying “reading deficiencies” are also related to the proportion and characteristics of students identified. Districts using a single assessment tend to identify a larger percentage of students as “reading deficient,” and these students are more likely to Black, Hispanic, and economically disadvantaged. In contrast, districts employing multiple assessments or incorporating ELA classwork tend to identify fewer students and demonstrate fewer disparities among the demographics of those identified. These findings can inform policymakers and practitioners in Michigan and other states with similar policies about the trade- offs in various identification approaches. The remainder of this paper proceeds as follows. The next section examines the requirements for early literacy policies to identify students with “reading deficiencies,” emphasizing Michigan’s Read by Grade Three Law. Following this, I explore existing research 118 on practices for identifying students experiencing literacy challenges. Drawing on this research and the policy guidelines outlined in the Read by Grade Three Law, I construct a conceptual framework to guide the study’s analysis. Next, I detail the data and methods employed to address the RQs and present the study’s findings. The paper concludes by discussing the significance of these findings and offering recommendations for policymakers and practitioners focused on identifying students requiring extra support in early literacy. “Reading Deficiency” Identification Under Early Literacy Policies Identifying “reading deficiencies” among K-3 students is a crucial aspect of early literacy policies in 38 states, including Michigan (ExcelinEd, 2022a). These policies outline specific steps for identifying and supporting students with literacy challenges, although there is some variation across states. For example, Alaska mandates the universal use of a single statewide screening tool across districts, promoting consistency in the identification process (AK SB 151, 2020). Conversely, Arkansas encourages a more diversified approach, suggesting that districts employ multiple measures to identify students needing support (AR Code § 6-15-420, 2015). Similarly, Colorado requires districts to choose at least one assessment from a state-approved list while permitting the inclusion of additional assessments (CO Code § 22-7-1205). The most common approach to “reading deficiency” identification, exemplified by Michigan’s Read by Grade Three Law, requires districts to select one or more assessments from a state-approved list (ExcelinEd, 2022a; Michigan Public Act 306, 2016). Thus, Michigan serves as an informative case for examining the implementation of these policies. The law sets guidelines for districts regarding the identification of K-3 students with “reading deficiencies,” emphasizing early detection and intervention to support literacy development: (2) [B]eginning in the 2017-2018 school year, the board of a school district or board of 119 directors of a public school academy shall do all of the following to ensure that more pupils will achieve a score of at least proficient in English language arts on the grade 3 state assessment:25 (a) Select 1 valid and reliable screening, formative, and diagnostic reading assessment system from the assessment systems approved by the department under subsection (1)(a). A school district or public school academy shall use this assessment system for pupils in grades K to 3 to screen and diagnose difficulties, inform instruction and intervention needs, and assess progress toward a growth target. A school district or public school academy periodically shall assess a pupil’s progress in reading skills at least 3 times per school year in grades K to 3. The first of these assessments for a school year shall be conducted within the first 30 school days of the school year (Michigan Public Act 306, 2016). Under this policy, MDE has created a list of 35 approved screening and 29 approved diagnostic assessments from which districts can select (MDE, 2024). MDE defines an assessment system as comprising at least one screening and one diagnostic assessment (MDE, personal communication, March 4, 2024). The idea is that this assessment system can first identify students who need support through the screening assessment(s), and then identify the specific areas in which they need support through the diagnostic assessment(s). In certain instances, a single assessment may fulfill both criteria, as is the case with NWEA MAP Growth, which appears on both lists (MDE, personal communication, March 4, 2024). However, while assessments on MDE’s approved list adhere to the Read by Grade Three Law, they may not always align with research recommendations regarding their optimal use. For instance, although 25 In Michigan, charter schools are called public school academies (PSAs). 120 NWEA MAP Growth is on both approved lists, it is typically recommended for screening purposes rather than diagnostic purposes (Truckenmiller et al., 2024). This discrepancy may arise because the Read by Grade Three Law imposed significant resource investments from districts, which might have prompted MDE to create a more comprehensive list to avoid burdening districts with additional expenses while still allowing them to achieve compliance with the law. Nonetheless, this study focuses on districts’ use of the screeners on MDE’s approved list to understand their implementation of this aspect of the policy. Districts determine their own cut scores on their selected screening assessments, whereby students scoring below this threshold are flagged as having a “reading deficiency.” This identification triggers eligibility for a range of supports aimed at enhancing literacy skills, ranging from individualized reading plans and regular progress monitoring to evidence-based instruction, “Read at Home” plans to boost reading outside school, additional instructional time, and focused interventions in small groups or one-on-one (Michigan Public Act 306, 2016). Thus, the identification process is pivotal for students, providing access to interventions that could significantly improve their academic performance and future outcomes. Likewise, for districts, it entails an obligation to provide these interventions for identified students, calling for a deeper investigation into how districts implement this aspect of the policy, specifically concerning the use of measures proven to accurately identify students who require additional support. Literature Review The importance of early literacy screening in identifying students requiring additional support is well established, warranting its inclusion in policies like Michigan’s Read by Grade Three Law (Rose & Schimke, 2012). There is a strong consensus on the benefits of early identification of literacy challenges, with research indicating that students who are identified 121 early and receive targeted support have better literacy outcomes (National Early Literacy Panel, 2008; Snow et al., 1998; Torgesen, 2004). The National Early Literacy Panel (2008) emphasizes the importance of screening for and addressing early foundational skills such as phonological awareness and alphabet knowledge, as these skills are crucial predictors of future reading, writing, and spelling abilities. Similarly, Torgesen (2004) and Snow et al. (1998) highlight that targeted interventions can significantly enhance literacy outcomes by addressing challenges early on, even mitigating the expansion of achievement gaps. MDE approves 35 screening assessments under the Read by Grade Three Law for identifying students with “reading deficiencies” (MDE, 2024). These assessments vary in their suitability as screening tools and their accuracy in identifying literacy challenges. While MDE does not categorize these assessments into specific types, researchers often classify them into three distinct groups for analytical clarity: curriculum-based measures (CBMs), computer- adaptive tests (CATs), and informal reading inventories (IRIs). CBMs are standardized, brief assessments designed to evaluate students’ literacy skills such as fluency, decoding, and phonemic awareness against specific learning standards. By providing quick evaluations—typically lasting just a few minutes each week—CBMs allow educators to monitor student progress against established benchmarks, often based on national norms. CBMs have been recognized as some of the most accurate and cost-effective options for literacy screening (Barrett et al., 2023; Keller-Margulis et al., 2008; Kilgus et al., 2014; Truckenmiller et al., 2024; Wayman et al., 2007). A meta-analysis conducted by Kilgus et al. (2014) found that these assessments accurately identify between 80 and 83 out of every 100 students at risk of literacy challenges on average, indicating a high rate of true positives where students flagged as at-risk indeed show difficulties on subsequent standardized assessments. 122 Specifically, CBMs such as Acadience and FASTBridge have been independently verified to possess reliable classification accuracy and are appropriate for screening decisions in grades K-3, underscoring their potential utility in the “reading deficiency” identification process (Acadience Learning, n.d.; Illuminate Education, n.d.; Michigan Assessment Consortium, 2020; National Center on Intensive Intervention, 2021; Truckenmiller et al., 2024). However, it is important to note that these assessments often generate a range of scores and detailed data, making it crucial for districts and educators to use the appropriate score for decision-making purposes. For instance, while Acadience provides diagnostic scores in areas such as comprehension, fluency, and oral language, the composite score is the most suitable for screening students for additional literacy support (Truckenmiller et al., 2024). CATs introduce a dynamic component to screening assessments by adjusting the difficulty of questions in real-time based on the student’s previous responses. This approach tailors the assessment to the individual student’s ability level, making CATs efficient tools for screening purposes. While CATs are relatively newer than CBMs, research generally supports their effectiveness in accurately identifying students with literacy challenges (Barrett et al. 2023; Clemens et al., 2011; McBride et al., 2010; Michigan Assessment Consortium, 2020; Ochs et al., 2018; Shapiro & Gebhardt, 2012). Although a comprehensive meta-analysis on the screening accuracy of CATs is yet to be conducted, some studies suggest that CATs can be more accurate and cost-effective than both CBMs and IRIs for screening purposes (Barrett et al., 2023; Clemens et al., 2011; McBride et al., 2010; Ochs et al., 2018; Shapiro & Gebhardt, 2012). For instance, McBride et al. (2010) found that Star Reading (a CAT) was more cost-effective and provided a more accurate assessment for early literacy screening among K-2 students compared to DIBELS (a CBM) and the Texas Primary Reading Inventory (an IRI). Other CATs recognized 123 for their accuracy in early literacy screening include iReady, NWEA MAP Growth, and Lexia RAPID (Curriculum Associates, 2024; National Center on Intensive Intervention, 2021; NWEA, n.d.; Truckenmiller et al., 2024).26 However, as with CBMs, it is crucial for educators and districts to ensure they are using the appropriate score on these assessments to make informed decisions, given the range of data they provide (Truckenmiller et al., 2024). While CBMs and CATs generally offer educators an accurate assessment of which students require additional literacy support, researchers typically do not recommend using IRIs for screening purposes. IRIs, with the Fountas & Pinnell Benchmark Assessment System being one of the most well-known, involve students reading a series of leveled passages aloud and answering comprehension questions (Conradi Smith et al., 2019; Fountas & Pinnell, 2016). Teachers use these assessments to determine students’ independent reading levels by evaluating their accuracy, fluency, and comprehension. Predating both CBMs and CATs, IRIs have been popular among teachers for their straightforward scoring system, which is helpful for practical classroom activities such as creating guided reading groups (Ford & Opitz, 2008). For instance, the Fountas & Pinnell system corresponds with guided reading levels used in their Leveled Books and instructional materials (Fountas & Pinnell, 2016). However, these assessments take longer to administer than CBMs and CATs and are more prone to administrator and measurement errors. Indeed, researchers have likened the accuracy of IRIs to the randomness of flipping a coin (Parker et al., 2015). This presents a dilemma between their user-friendly nature and the higher likelihood of inaccuracies (Fuchs et al., 1982; Klingbeil et al., 2015; Nilsson, 2011; Russell, 2013; Spector, 2005; Tortorelli, 2019). Consequently, researchers recommend using IRIs for less critical decisions, such as selecting reading materials for the classroom, rather 26 As of June 30, 2023, Lexia RAPID has been discontinued (Lexia, 2023). 124 than consequential tasks like flagging students for additional support (Michigan Assessment Consortium, 2020; Spector, 2005). Beyond assessments, educators frequently rely on students’ ELA classwork performance to guide instructional decisions. While this information can be valuable in informing classroom teaching strategies, its efficacy in identifying literacy difficulties remains less established (Michigan Assessment Consortium, 2020). Teacher judgments may vary, introducing potential biases and subjective interpretations that can compromise the reliability of classwork performance for screening purposes. Issues such as unclear grading criteria and biases, including racial biases, can lead to both under-identification of students who need support and over- identification of those who are proficient (Kenny & Chekaluk, 1993; Kettler & Albers, 2013; Quinn, 2020; Schmitterer & Brod, 2021; Virinkoski et al., 2018). These challenges underscore the need for more standardized and objective assessment methods to accurately identify students with “reading deficiencies.” Still, no assessment can be 100% accurate 100% of the time, so researchers often recommend the use of multiple screeners to improve precision in identifying students needing literacy interventions (Barrett et al., 2023; Bishop & League, 2006; Clemens et al., 2011; Compton et al., 2010; Elbro & Scarborough, 2004; Felton, 1992; Kelso et al., 2020; Kettler & Albers, 2013; Michigan Assessment Consortium, 2020; Truckenmiller et al., 2024). The Michigan Assessment Consortium (2020) states, “No single assessment can be considered a definitive indicator of a student’s knowledge, skills, and interests. Multiple assessments and indicators enhance the validity and fairness of the inferences drawn by giving students various ways and opportunities to demonstrate their learning” (p. 12). Numerous studies have demonstrated that employing multiple assessments can indeed enhance accuracy in identifying 125 students at risk of literacy difficulties, particularly when combining CBMs and CATs (Barrett et al., 2023; Compton et al., 2010; Klingbeil et al., 2015, 2017; Thomas & January, 2019). However, combining CBMs or CATs with IRIs generally does not yield the same benefits (Klingbeil et al., 2015; Parker et al., 2015). While integrating multiple measures for literacy screening can enhance precision, it also introduces challenges, including resource and time demands, the cost of acquiring and implementing various assessment tools, and the need for educator training (Levin et al., 2018). Moreover, interpreting multiple scores to accurately identify students needing support increases the potential for errors. Researchers have indicated that employing more than two or three measures may offer limited additional value (Clemens et al., 2011; Truckenmiller et al., 2024; VanDerHeyden et al., 2018). This suggests that while a variety of assessments can improve the accuracy of the identification process, an excessive number could result in diminishing returns. While research specific to the identification of “reading deficiencies” under early literacy policies is limited, studies align with previous findings. For instance, in a study most relevant to the current context, Barrett et al. (2023) evaluated three screening tools under Michigan’s Read by Grade Three Law: Acadience (a CBM), Star Reading (a CAT), and the Fountas & Pinnell Benchmark Assessment System (an IRI). Their analysis of fall screening scores from 110 3rd graders, employing classification and regression tree analysis, revealed that the combination of Acadience and Star Reading achieved the highest accuracy, correctly identifying 91% of students. They also conducted a cost analysis, highlighting Acadience as the most cost-effective option at US$19.78 per accurately identified student, closely followed by Star Reading at US$24.28. The Fountas & Pinnell system was the least cost-effective at US$64.40 per student, suggesting it may not be the optimal primary identification tool for districts (Barrett et al., 2023). 126 Ultimately, research on early literacy identification underscores that educators employ diverse tools to assess students’ literacy skills. However, not all tools are equally effective for screening purposes. CBMs and CATs emerge as more accurate and cost-effective than IRIs, suggesting that districts should prioritize these assessments for identification purposes. While integrating CBMs and CATs can improve accuracy, relying on more than two or three assessments may yield marginal benefits and potentially create diminishing returns. Consequently, there is a need to examine districts’ approaches to “reading deficiency” identification to ensure alignment with research-backed methods. Conceptual Framework While the Read by Grade Three Law mandates districts to use state-approved assessments to identify K-3 students with “reading deficiencies,” implementation often varies as educators adapt policy requirements to fit their unique contexts, understandings, and practices (Bryk et al., 2015; Honig, 2006; Spillane, 2000). Given the literature reviewed above and the flexibility the law allows in choosing assessments, I expect districts to use a range of methods for identifying “reading deficiencies.” Some may strictly adhere to MDE’s approved list of screening assessments, while others might integrate students’ ELA classwork, aligning with common teacher practices. The conceptual framework depicted in Figure 3.1 and described in this section anticipates that several factors will shape districts’ identification approaches, including the Read by Grade Three Law’s requirements, features of various identification measures (e.g., accuracy, cost, time), and districts’ unique contexts. The following subsections will first explore how each of these factors is expected to be related to districts’ chosen identification approaches. Then, I outline several approaches districts may adopt based on the literature reviewed above and these 127 different factors. This section concludes by discussing the potential implications of each of these approaches for student “reading deficiency” identification. Figure 3.1 Conceptual Framework 128 Policy Requirements Because of the Read by Grade Three Law’s mandate to select at least one screening assessment from MDE’s approved list, I anticipate most districts will employ one or more of these assessments as part of the identification process. Nonetheless, the specific assessments they use, how many they integrate into their processes, and whether they supplement these assessments with other measures, such as students’ ELA classwork, will likely also depend on the features of the tools themselves and individual district contexts. Features of Identification Measures The literature reviewed above demonstrates that identification measures vary in their accuracy, cost, and implementation time. Districts face the challenge of balancing these factors when choosing measures to identify students with “reading deficiencies.” I anticipate districts will give priority to screening assessments from MDE’s approved list, not only to meet the Read by Grade Three Law’s requirements but also because these assessments are more effective in identifying students needing support compared to teacher evaluations of students’ ELA classwork (Kenny & Chekaluk, 1993; Kettler & Albers, 2013; Schmitterer & Brod, 2021; Virinkoski et al., 2018). Additionally, districts are likely to favor CBMs and CATs over IRIs due to their superior accuracy, cost-effectiveness, and quicker implementation (Barrett et al., 2023; Clemens et al., 2011; Kilgus et al., 2014; Ochs et al., 2018; Parker et al., 2015). However, some districts may still incorporate IRIs given their longstanding use among teachers, the accompanying resources such as leveled texts, and their practical utility in supporting activities like forming reading groups (Ford & Opitz, 2008). Recognizing that many teachers routinely use evaluations of students’ ELA classwork for instructional planning (Virinkoski et al., 2018), some districts might also use these evaluations to 129 identify students with “reading deficiencies.” Despite being less precise for identification and potentially prone to bias, these evaluations are already an integral part of many teachers’ practices (Kenny & Chekaluk, 1993; Kettler & Albers, 2013; Quinn, 2020; Schmitterer & Brod, 2021; Virinkoski et al., 2018). Furthermore, they offer a potentially cost-effective alternative, avoiding spending on additional assessments and training. District Context The framework also recognizes that district-specific contexts, including financial and human resources, student demographics, geographic location, and current curriculum and instructional resources, will likely shape how districts identify students with “reading deficiencies.” Financial and human resources stand to play a crucial role in determining the types and number of tools a district can use for identification. Districts with ample resources can afford a wider range of screening tools, while those with tighter budgets may opt for fewer, more cost- effective options. Despite the initial costs associated with CBMs and CATs, such as teacher training and technological infrastructure, these tools offer more value over the long term than IRIs, which are less cost-effective due to their time-consuming administration time and higher error rates (Barrett et al., 2023; Klingbeil et al., 2015; Parker et al., 2015). Teacher evaluations of students’ ELA classwork, while requiring minimal upfront financial investment, may demand considerable teacher time, which could be challenging in larger class settings where the time teachers can spend with each individual student is reduced. Consequently, districts facing financial constraints or those with larger classes are likely to favor more straightforward and efficient tools like CBMs and CATs for identifying “reading deficiencies.” In contrast, districts with greater financial flexibility and/or smaller class sizes might employ a wider array of tools, or strategies with longer administration times (e.g., IRIs, ELA classwork), given their additional 130 financial and human resources. The demographics of a district’s student population and its geographical location could also influence the choice of “reading deficiency” identification tools. Districts with diverse student populations, including varied racial and socioeconomic backgrounds or high proportions of English learners (ELs) or students with disabilities, may opt for multiple assessment tools to provide a more nuanced evaluation of students’ diverse literacy skills and needs (Michigan Assessment Consortium, 2020). Geographical context may also play a role in districts’ tool selection. In particular, rural districts, often grappling with limited technological infrastructure or internet access (Arsen et al., 2022), might lean towards more traditional assessment methods like IRIs or students’ ELA classwork due to logistical constraints in deploying technology-dependent tools. Finally, the alignment between a district’s approach to curriculum and instruction and its selection of tools for identifying students with “reading deficiencies” is crucial for creating an effective early literacy assessment system. The Michigan Assessment Consortium (2020) emphasizes that such systems should coherently integrate with a district’s curriculum and instructional strategies to ensure that assessments are directly relevant to students’ instruction. This integration makes the assessment process more meaningful and actionable for educators. Although this study lacks detailed information on districts’ curriculum and instructional practices for the school year under investigation, it is important to acknowledge that this could significantly influence districts’ decisions regarding “reading deficiency” identification tools. Identification Approaches Next, the framework outlines how policy requirements, features of identification measures, and districts’ contexts come together to shape strategies for identifying “reading 131 deficiencies.” It proposes five potential approaches districts may adopt based on these factors. The “Single Assessment” approach involves using just one assessment from MDE’s approved list. While this approach fulfills the Read by Grade Three Law’s requirements and could be resource-efficient, it is important to note that the choice of a single assessment is not necessarily indicative of limited resources. A single, high-quality assessment, especially when using a CBM or CAT, can effectively identify students needing support. The following three approaches involve using multiple tools for identifying “reading deficiencies,” which could help achieve greater accuracy or a deeper understanding of students’ literacy skills and needs. The “Multiple Assessments” approach uses more than one assessment from MDE’s approved list and may be adopted by districts with the capacity to train educators on using these tools and analyzing the resulting data. The “Hybrid Approach” combines a single assessment with students’ ELA classwork, potentially enhancing understandings of students’ literacy skills. This approach may appeal to districts seeking thorough strategies without the expense of additional assessments. The “Comprehensive Approach,” combining multiple assessments with students’ ELA classwork, requires substantial capacity to manage and interpret data from multiple sources. Districts with smaller class sizes might be more likely to adopt the Hybrid or Comprehensive approaches, as teachers can dedicate more time to evaluating each student’s ELA classwork. However, using too many measures might lead to minimal additional gains in accuracy (Clemens et al., 2011; Truckenmiller et al., 2024; VanDerHeyden et al., 2018), questioning the value of such an extensive investment. “Classwork Only” districts rely exclusively on students’ ELA classwork, which might be influenced by educational philosophies favoring teacher judgments over standardized assessments. However, evaluating students’ literacy solely through classwork demands a deep 132 understanding of each student’s skills, which can be particularly challenging and time- consuming, especially in diverse student populations with varied literacy needs. Districts with smaller class sizes or less diverse student populations might opt for this approach, as this additional capacity could permit more personalized attention. Implications for Student Identification Finally, the framework suggests that a district’s approach to identifying “reading deficiencies” will likely be related to identification rates, though the direction of these relationships is not straightforward. For instance, the Single Assessment approach might result in either over- or under-identification due to its reliance on a single evaluation method. Conversely, using multiple tools through Multiple Assessments, Hybrid, or Comprehensive Approaches could lead to more nuanced identification rates, potentially identifying additional students who require support while also reidentifying students as proficient. Additionally, identification rates under the Classwork Only approach will depend on the specific criteria used to evaluate ELA classwork, which could lead to a broad range of identification outcomes. This study aims to shed light on these intricate dynamics by exploring districts’ choices of identification strategies and how these choices relate to student “reading deficiency” identification under the Read by Grade Three Law. These approaches have direct implications for students and districts, as being identified with a “reading deficiency” qualifies students for a range of evidence-based supports from their district under the Read by Grade Three Law. Data Sources Data and Methods To explore how districts identify students with “reading deficiencies” and the relationship between these approaches and student identification, I use a combination of survey and 133 administrative data from the 2021-22 school year. I select this year because of the availability of superintendent survey data on districts’ strategies for identifying students with “reading deficiencies,” which I match with student-level administrative data indicating their “reading deficiency” status. The surveys are part of the Education Policy Innovation Collaborative’s (EPIC’s) broader study on the implementation of Michigan’s Read by Grade Three Law (see Strunk et al., 2021, 2022). Data collection occurred online during a three-month window in spring 2022. EPIC recruited eligible participants through direct emails, website promotion, social media engagement, and partnerships with Michigan education associations. Appendix Table A3.1 details the survey questions analyzed in this study, which focus on districts’ approaches to “reading deficiency” identification and the assessments they use for this purpose. Seventy-one superintendents responded to these questions, accounting for approximately 13% of districts in the state. Although the response rate is low, the 2021-22 school year context continued to challenge educational leaders due to the ongoing COVID-19 pandemic. Additionally, it is possible that the known partnership between EPIC and MDE influenced superintendents’ willingness to respond, fearing accountability for noncompliance with state policies. This could introduce bias into our sample, skewing results toward districts more confident in their compliance or less concerned about potential repercussions. To help address non-response bias, I employ propensity score weighting, described in more detail below. Nonetheless, a 13% response rate is inherently challenging, an issue revisited in the limitations section. I complement the survey data with student- and district-level administrative data provided by MDE, including students’ enrolled grade level, demographics (e.g., race, gender, economic disadvantage, EL status, and disability status), “reading deficiency” status, and 134 districts’ geographic locale and charter status. Additionally, I incorporate state financial data on district revenues to better understand the relationship between financial resources and “reading deficiency” identification approaches. I combine districts’ revenues from all sources because the descriptors of each revenue source do not clearly specify the allocation of funds for literacy- or assessment-related purposes. This approach allows for a better understanding of the overall financial context within which districts make decisions about allocating resources for “reading deficiency” identification. Finally, I use the state’s teacher-student data link to construct a measure of class size, enabling an examination of how human capital may be related to districts’ selection of identification approaches.27 Sample The analysis for RQ1 focuses on the district level, investigating how districts identify students with “reading deficiencies,” while RQ2 is at the student level, exploring the relationship between districts’ approaches and students’ likelihood of being identified. Therefore, I include sample details for both districts and students. To ensure the reliability and validity of the analysis, I employ multiple sample restrictions detailed in Appendix Table A3.2. Initially, the sample encompasses all K-3 students enrolled in Michigan traditional public schools (TPSs) or charter schools for the 2021-22 school year, totaling 421,519 students across 771 districts. I narrow the sample to include only districts receiving funding under Section 35a(5) of the State School Aid Act, as only these districts are required to report students’ “reading deficiency” status.28 This results in 407,697 students across 703 districts. I then focus on students and 27 Class sizes are determined based on available school enrollment data. I generate a unique class identifier by grouping students according to school year, school code, course identification number, course section, and teacher. Class sizes are computed by summing the number of students within each unique class group. To analyze the data at the district level, I calculate the average class size for each district. 28 Section 35a(5) of the State School Aid Act is otherwise known as the Additional Instructional Time Grant (MDE, 2022). Around 97% of the state’s districts receive this funding each year, totaling $19.9 million annually. 135 districts where superintendents took the Read by Grade Three survey, bringing the count to 41,461 students in 78 districts. However, seven superintendents who participated in the survey did not answer the questions regarding “reading deficiency” identification. I exclude these students and districts from the final analysis as their non-response could indicate various scenarios, such as an alternative identification approach, uncertainty about their district’s methods, or simply skipping past the question.29 This results in a final analytical sample comprising 38,491 students in 71 districts. Table 3.1 compares the characteristics of the analytical sample and the overall population of students (Panel A) and districts (Panel B) in Michigan. The students in the analytical sample predominantly belong to more historically advantaged groups than the broader population, with a significant underrepresentation of Black students. There is a similar trend among the districts in the sample; they generally educate students from more advantaged backgrounds. Specifically, these districts have significantly lower proportions of non-White students and are less likely to be charter or urban than the statewide district population. Given these differences between the samples and overall populations of students and districts, readers should interpret the results of this analysis with caution. To help mitigate—but not wholly resolve—these differences, I employ propensity score weights derived from the characteristics outlined in Table 3.1. I create these weights at the district level for the district-level RQ1 analysis and the student level for the 29 Appendix Table A3.3 reveals that the superintendents who did not disclose their districts’ methods for identifying students with “reading deficiencies” (i.e., did not answer the relevant survey question) differ markedly from both the analytical sample and the broader population in several key aspects. Notably, these districts have a higher percentage of students identified as “reading deficient.” Additionally, they exhibit slightly higher ELA achievement and lower proportions of non-White and economically disadvantaged students. These districts tend to be smaller in size, and none are classified as charter or urban. While not all these distinctions reach statistical significance—partly due to the small size of this group, with only seven districts—it is clear that these districts are descriptively distinct. They appear to be more historically advantaged compared to both the study’s analytical sample and the overall district population in Michigan. Although excluded from the current analysis, the unique characteristics of this group—and their high “reading deficiency” identification rates—merit further investigation in future studies. 136 student-level RQ2 analysis.30 Table 3.1 Sample Characteristics Identified w/ “Reading Def.” Economically Disadvantaged English Learners Students w/ Disabilities Black Hispanic Asian Other Race Female Prop. “Reading Def.” ELA Achievement Prop. Non-White Prop. Econ. Dis. Prop. EL Prop. SWD Enrollment Class Size Charter Rural Urban Revenue Sample Population Difference Panel A: Student Characteristics 31.41% 52.09% 7.30% 15.49% 9.23% 8.46% 5.14% 6.70% 48.48% 33.13% 57.35% 8.67% 15.21% 18.48% 8.84% 3.70% 6.41% 48.62% Panel B: District Characteristics 39.13% -0.038 28.02% 62.54% 5.42% 17.96% 542.13 22.35 12.68% 49.30% 8.45% $35.36 mil 37.49% -0.091 36.07% 65.50% 5.88% 16.38% 546.72 22.76 30.35% 40.39% 17.01% $34.91 mil -1.72% -5.26% -1.37% 0.28% -9.26%*** -0.38% 1.44% 0.29% -0.14% 1.64% 0.052 -8.05%** -2.95% -0.47% 1.58% -4.59 -0.42 -17.67%*** 8.91% -8.56%* $449,550.66 30 To create the district-level weights, I estimate the following logistic regression model: log ( 𝑝 1−𝑝 ) = 𝛽0 + 𝛽1𝜲𝑖 where p is the probability of being in the sample based on district-level characteristics 𝜲𝑖, including the proportion of students identified with a “reading deficiency,” proportion non-White, economically disadvantaged, EL, students with disabilities, log student enrollment, class size, standardized ELA achievement, log revenue, and indicators for charter, rural, or urban districts (with suburban as the reference). Based on this model, I calculate propensity scores for each district, which represent the likelihood of inclusion in the sample. I then use the inverse of these scores to generate weights: 𝑤𝑒𝑖𝑔ℎ𝑡 = , where pscore is the propensity score of the district being in the 1 𝑝𝑠𝑐𝑜𝑟𝑒 sample. For student-level weights, I apply a similar procedure, but the logistic regression model includes individual student characteristics such as “reading deficiency” status, economic disadvantage status, EL and special education status, indicators for whether the student is Black, Hispanic, Asian, another race, or female (with White and male as the references), and indicators for whether they attend a charter, urban, or rural school. In both the district- and school-level models, I cluster standard errors at the district level to account for district-level decision-making on student identification practices. 137 Methods RQ1: District Approaches to “Reading Deficiency” Identification To address RQ1, which explores the methods districts use to identify K-3 students with “reading deficiencies” and how this varies across districts, I first classify districts according to the typologies outlined in the Conceptual Framework based on their responses to the two survey items of interest. For example, suppose a superintendent indicated using only assessments in the first question and selected only a single assessment in the second question. In that case, I classify their district as “Single Assessment.” Conversely, suppose they reported using both ELA classwork and assessments in the first question and identified multiple assessments in the second. In that case, I categorize their district under the “Comprehensive Approach” and so on. I compute descriptive statistics, including the percentage of districts employing each approach and the characteristics of these districts, including their ELA achievement, proportions of non-White, economically disadvantaged, ELs, students with disabilities, enrollment, class size, revenue, charter status, and geographic locale. To assess whether the differences in these characteristics are statistically significant across approaches, I use Analysis of Variance (ANOVA) for continuous variables and chi-square tests for categorical variables.31 In districts that use assessments for identifying students with “reading deficiencies,” whether alone or in conjunction with ELA classwork, I examine the number of assessments they employ and how this varies by approach. I then delve into their specific assessment systems, overall and by approach. This includes identifying the exact names of the assessments and 31 Given the limited overall sample size and even smaller sizes of subsamples representing districts using each approach, it is not feasible to analyze the differences using a regression framework. Additionally, as this study is exploratory in nature, I focus on reporting the descriptive statistics for each category without delving into regression analysis. This approach allows for an initial understanding of the patterns and characteristics associated with each identification method, despite the constraints posed by the sample sizes. 138 categorizing them by type (i.e., CBM, CAT, IRI). RQ2: Implications for Student “Reading Deficiency” Identification To answer RQ2 and understand the relationship between districts’ approaches and the proportion and characteristics of students identified with “reading deficiencies,” I begin by analyzing descriptive statistics. I link student-level administrative data, which detail students’ “reading deficiency” status, grade level, and demographic information, with superintendent survey data by district code to determine their district’s identification approach. I calculate the overall percentage of students identified with “reading deficiencies” for each approach and the percentage of students identified by specific characteristics (i.e., grade level, gender, race, economic disadvantage, EL, and disability status). Next, I employ a linear probability model (LPM) with interaction terms to determine how a student’s likelihood of being flagged with a “reading deficiency” differs based on their demographic characteristics and the approach their district uses for identification. The model is structured as follows: 𝑅𝐷𝑖 = 𝛼 + 𝛽1𝚾𝑖 + 𝛽2𝑨𝒑𝒑𝒓𝒐𝒂𝒄𝒉𝑖 + 𝛽3(𝜲𝑖 × 𝑨𝒑𝒑𝒓𝒐𝒂𝒄𝒉𝑖) + 𝜀𝑖 (1) In Model (1), 𝑅𝐷𝑖 is a binary variable that equals 1 if student i was identified with a “reading deficiency” in the 2021-22 school year and 0 otherwise. The vector 𝜲𝑖 includes student characteristics such as grade level (1st, 2nd, and 3rd grade, with kindergarten as the reference), gender (with male as the reference), race (Black, Hispanic, Asian, and other race, with White as the reference), economic disadvantage status, EL status, and disability status.32 The vector 𝑨𝒑𝒑𝒓𝒐𝒂𝒄𝒉𝑖 includes dummy variables representing districts’ reported approaches to identifying “reading deficiencies,” including Multiple Assessments, Hybrid Approach, Comprehensive 32 The model’s average Variance Inflation Factor (VIF) is 1.22, with no individual variable exceeding a VIF of 1.46, indicating that multicollinearity does not pose a significant issue in the analysis. 139 Approach, and Classwork Only, with Single Assessment as the baseline comparison. I select Single Assessment as the baseline because it aligns with the Read by Grade Three Law’s requirement that districts use assessments for identification and represents the most straightforward, simplified approach. The model interacts the student characteristics 𝚾𝑖 with the approach indicators 𝑨𝒑𝒑𝒓𝒐𝒂𝒄𝒉𝑖 to investigate how the relationship between student demographics and “reading deficiency” identification varies across different approaches. I cluster standard errors at the district level as districts determine how to identify students. To account for comparing four different approaches against the Single Assessment baseline, I use a Bonferroni-adjusted p-value threshold. Additionally, I test the robustness of these results to logistic regression analysis (results are consistent and available upon request). After estimating Model (1), I conduct post-hoc tests to examine whether the coefficients from the LPM with interactions are significantly different from each other. Specifically, I perform F-tests on the interaction terms to jointly test the equality of the coefficients across different identification approaches. If the F-test indicates significant differences across approaches, I then conduct pairwise t-tests using the Bonferroni correction to account for multiple comparisons. This allows me to identify which specific pairs of approaches have significantly different associations with student identification rates. Finally, I calculate and report the marginal effects for each predictor within the various approaches that districts use for identifying “reading deficiencies.” These marginal effects quantify the change in the probability of a district flagging a student with a “reading deficiency” given a one-unit change in a predictor variable (e.g., being Black vs. White), holding all other factors constant. They convert the complex coefficients from interaction terms into straightforward, percentage-point changes in the probability of identification. This provides a 140 clearer and more intuitive understanding of the data and enables direct comparison across different identification approaches. Therefore, I present the marginal effects in the main results section. Appendix Table A3.4 provides the full results from the LPM with interactions. Results RQ1: District Approaches to “Reading Deficiency” Identification The analysis reveals the varying approaches districts employ to identify students with “reading deficiencies.” As shown in Table 3.2, the most common approach, used by 41.37% of districts, is the Multiple Assessments approach, which uses more than one screening tool from MDE’s approved list. The next most prevalent approach, adopted by 23.08% of districts, is the Single Assessment method that relies on just one screening tool. Together, nearly two-thirds (64.45%) of districts report using assessment tools exclusively from MDE’s approved list for identifying “reading deficiencies.” However, over a third incorporate students’ ELA classwork into their identification strategies. Specifically, 12.06% follow a Hybrid Approach that combines a single assessment with ELA classwork, while 17.93% employ a Comprehensive Approach using multiple assessments alongside classwork. Meanwhile, 5.56% of districts rely solely on ELA classwork, excluding screening assessments altogether. Table 3.2 District Characteristics by Approach N Districts Percent ELA Ach. Prop. Non- White Prop. Econ. Dis. Prop. EL Prop. SWD Single Multiple 15 23.08% -0.158 50.78% 27 41.37% -0.034 36.18% Hybrid Comprehensive Classwork 13 17.93% -0.147 29.11% 5 5.56% 0.166 21.13% 11 12.06% -0.144 24.97% Total 71 100.00% -0.085 36.09% 73.18% 62.22% 63.28% 63.35% 53.21% 64.58% 4.28% 13.19% 4.08% 17.47% 2.83% 17.90% 15.13% 15.90% 6.04% 18.96% 6.07% 16.33% 141 Table 3.2 (cont’d) Enrollment Class Size Charter Rural Urban Revenue Single Multiple 534.75 346.59 23.86 22.56 31.94% 43.16% 34.54% 41.14% 14.73% 29.44% $31.72 $23.26 mil mil Hybrid Comprehensive Classwork 715.82 470.64 23.36 21.13 25.66% 0.00% 38.61% 40.03% 7.73% 0.00% $48.74 mil $26.74 mil 498.33 23.56 0.00% 58.31% 0.00% $28.71 mil Total 514.04 23.06 27.78% 38.78% 14.27% $32.05 mil The characteristics of districts vary across these different identification approaches. Districts using the Single Assessment approach notably educate higher proportions of students from historically underserved populations compared to those adopting other methods. These districts demonstrate below-average ELA achievement (0.158 SD below the statewide mean), have the highest percentages of non-White and economically disadvantaged students (50.78% and 73.18%, respectively), and are more frequently charter (43.15%) or urban (29.44%) relative to the broader sample. Additionally, they report the lowest average revenues ($23.26 million, nearly $9 million below the sample average). These data suggest a potential association between districts facing financial constraints and opting for an approach requiring only one assessment. In contrast, districts relying on Classwork Only exhibit higher ELA achievement (0.166 SD above the statewide mean), lower percentages of non-White and economically disadvantaged students (21.14% and 53.21%, respectively), and a lack of charter or urban status (0% for both categories). Surprisingly, these districts do not have smaller class sizes than the average district in the sample, and they also report below-average revenues. This implies that the Classwork Only approach may not result from having more time or resources to evaluate each student’s ELA classwork individually. Instead, these districts may opt for this method due to serving less diverse and generally higher-achieving student populations, potentially simplifying the evaluation process. 142 Districts using the Hybrid Approach share similar demographic characteristics with the Classwork Only districts but demonstrate lower ELA achievement (0.143 SD below the statewide mean). Meanwhile, the Comprehensive and Multiple Assessment approaches broadly reflect the average district in the sample yet exhibit a few distinct differences. For instance, Multiple Assessment districts have an ELA achievement of 0.034 SD below the statewide average, whereas Comprehensive districts are 0.147 SD below. Furthermore, Comprehensive districts have a higher percentage of ELs (15.13%) compared to Multiple Assessment districts (4.08%) and have a larger average enrollment, with 715.82 students in Comprehensive districts versus 534.75 in Multiple Assessment ones. However, these differences do not reach statistical significance, likely due to the small sample sizes within each category. Turning to Table 3.3, districts that use assessments for identifying “reading deficiencies” (all except Classroom Only) typically use an average of two assessments. Single Assessment and Hybrid Approach districts use exactly one assessment by definition, while Multiple Assessment and Comprehensive Approach districts use an average of 2.81 and 2.39 assessments, respectively. This largely conforms to recommendations in the literature that suggest limiting the use of screening assessments to no more than two or three. However, 9.73% of districts employ four or five assessments, hinting at potential diminishing returns and suggesting a need for more streamlined practices in those districts. Table 3.3 District Assessment Use Mean (SD) CBM Assessments Acadience Reading AIMSWeb Single Multiple Hybrid Comprehensive 1 (0) 17.74% 10.54% 0.0% 2.81 (0.297) 75.60% 12.67% 26.12% 1 (1) 0.0% 0.0% 0.0% 2.39 (0.179) 66.06% 14.68% 13.68% Total 2.06 (0.191) 50.00% 10.91% 14.04% 143 Table 3.3 (cont’d) DIBELS1 CAT Assessments iReady NWEA Star Reading Informal Reading Inventories Single Multiple 36.51% 2.88% 100.00% 78.27% 5.40% 4.41% 81.89% 73.86% 29.46% 0.0% 60.43% 3.99% 48.10% 3.99% Hybrid Comprehensive 0.0% 100.00% 12.61% 70.93% 16.46% 0.0% 0.0% Total 24.61% 93.11% 6.52% 78.92% 16.41% 34.83% 29.43 % 1This refers specifically to DIBELS 6th and 8th editions, as these versions are included in MDE’s approved list of screening assessments. Note. The assessment systems featured in this table were reported by at least five superintendents as being used in the “reading deficiency” identification process. While the table highlights these more frequently used systems, other, less commonly used assessment systems also contribute to the overall calculations, such as the average number of assessments employed and the usage rates of CBMs, CATs, and informal reading inventories. These additional systems are included in the overall analysis but are not individually itemized in the table. 41.69% 91.69% 7.73% 83.96% 7.41% 38.89% 38.89% Fountas & Pinnell When examining the types of assessments used, CATs emerge as the most popular choice, with 93.11% of districts integrating a CAT into their identification process. Exactly half of districts use CBMs, and a little over a third incorporate IRIs. Districts also exhibit preferences for specific assessment systems. Among CBMs, DIBELS is the top choice, used by approximately one-quarter of districts.33 NWEA, a CAT, is the most popular assessment system overall, with 78.92% of districts using it for “reading deficiency” identification. The Fountas & Pinnell Benchmark Assessment System is the most commonly used IRI, adopted by 29.43% of districts. Therefore, while most districts choose assessment systems known for greater accuracy in identifying students in need, a significant number still rely on less accurate and potentially less cost-effective screeners. Districts show varying assessment preferences based on their identification approach. Those using Single Assessments overwhelmingly favor CATs, particularly NWEA, chosen by 33 In the survey, we specifically asked about the 6th and 8th editions of DIBELS, as these versions are included in MDE’s approved list of screening assessments. 144 73.86% of these districts. About 18% use CBMs, while only 4% opt for IRIs, exclusively selecting the Fountas & Pinnell Benchmark Assessment System. This shows a general preference for assessments recognized for their accuracy in identifying students. However, the use of Fountas & Pinnell by a small percentage raises concerns about the accuracy and cost-efficiency of their identification process. Districts following the Hybrid Approach solely employ CATs, preferring NWEA (70.93%), Star Reading (16.46%), and iReady (12.61%). By combining accurate assessments with evaluations of students’ ELA classwork, these districts likely aim to understand students’ literacy abilities better and address the potential gaps that a Single Assessment approach might overlook. All Multiple Assessment districts incorporate CATs, predominantly NWEA (81.89%), and often pair these with CBMs (75.60%) and/or IRIs (60.43%). Districts that blend CATs with CBMs potentially achieve greater accuracy than those that match a CAT or a CBM with an IRI. Comprehensive Approach districts follow a similar pattern, with 91.69% employing CATs, 66.06% using CBMs, and 38.89% adopting IRIs, specifically Fountas & Pinnell. However, these districts’ integration of multiple assessments with students’ ELA classwork raises concerns about the tradeoff between marginal gains in accuracy from using numerous measures and the likely drain on resources. RQ2: Implications for Student “Reading Deficiency” Identification The approaches districts use to identify students with “reading deficiencies” are also related to the proportion and characteristics of students identified. Table 3.4 presents descriptive statistics on identification rates by approach, both overall and by various student subgroups. The Single Assessment approach is associated with the highest overall identification rate, with districts using this method identifying 54.86% of students as having “reading deficiencies,” more 145 than 20 percentage points above the sample average of 34.19%. Notably, Single Assessment districts tend to have lower revenues and serve more historically underserved populations compared to other approaches, which could make it particularly challenging to provide interventions to such a high proportion of identified students. Similarly, the Classwork Only approach is related to a high identification rate of 51.4%, but these districts are generally less diverse and more historically advantaged, potentially making it less burdensome to support a large identified group. In contrast, the Multiple Assessment, Hybrid, and Comprehensive Approaches are associated with lower overall identification rates of 27.2%, 33.85%, and 26.2%, respectively. Table 3.4 “Reading Deficiency” Identification Descriptives by Approach Single Multiple Hybrid Comprehensive Classwork 5,653 54.86% 47.04% 59.26% 57.96% 55.80% 54.86% 54.87% 75.92% 62.56% 1.58% 35.45% 31.53% 67.70% 26.95% 58.97% 54.60% 54.64% 54.89% Total 38,348 4,888 N 34.19% 33.85% “Reading Def.” 26.30% 25.22% Kindergarten 1st Grade 38.31% 36.84% 2nd Grade 37.62% 35.51% 3rd Grade 35.55% 35.20% 33.27% 31.73% Female 35.06% 35.85% Male 57.49% 77.40% Black 40.76% 44.54% Hispanic 10.79% 28.84% Asian 34.73% 40.75% Other Race 27.32% 27.09% White 45.67% 47.29% Econ. Dis. 19.83% 20.44% Non-Econ. Dis. 30.70% 64.00% EL 34.51% 32.26% Non-EL 48.51% 54.88% SWD Non-SWD 31.69% 29.72% Note. The percentages show weighted averages for the percentage of identified students in each subgroup, overall and by district approach. For instance, the overall identification rate for kindergarteners is 26.3%. In districts adopting a Single Assessment approach, 47.04% are identified, whereas in districts employing a Multiple Assessment approach, 19.73% are, etc. 2,559 51.40% 25.22% 57.95% 61.57% 65.22% 49.50% 53.24% 67.74% 61.59% 63.94% 68.51% 45.70% 48.94% 53.03% 63.96% 49.95% 63.23% 49.41% 9,564 26.20% 20.74% 27.69% 26.20% 30.49% 24.71% 27.60% 42.25% 31.34% 3.15% 32.34% 26.63% 41.90% 12.72% 11.71% 28.52% 45.93% 22.80% 15,684 27.20% 19.73% 33.27% 31.66% 25.42% 26.38% 27.98% 34.72% 35.29% 10.60% 29.80% 23.89% 34.50% 16.83% 30.54% 26.95% 44.65% 23.79% 146 These findings suggest that districts relying on a single measure, whether a single assessment or ELA classwork, tend to identify a larger proportion of students with “reading deficiencies.” This could either indicate that these single-measure approaches are prone to over- identifying students, or that the higher identification rates accurately reflect greater literacy needs among the student populations in these districts. Conversely, districts employing multiple measures, such as a combination of assessments or assessments alongside ELA classwork, generally identify a smaller proportion of students. This could indicate that using multiple data points enables a more precise and accurate identification of students’ true literacy abilities and needs, or it could simply reflect that fewer students in these districts genuinely require literacy interventions. Districts’ approach choices could also reflect strategic resource allocation. Some may deliberately over-identify to provide the maximum number of students with interventions; others may under-identify to concentrate limited resources on only the most at-risk students. Crucially, these data alone cannot determine the accuracy of different approaches. Higher identification rates do not necessarily imply over-identification, just as lower rates do not inherently signify under-identification. Further research examining how these approaches align with students’ actual literacy skills will be needed to determine which methods most effectively identify those requiring additional support. Table 3.4 also reveals differences in identification rates across approaches when considering student characteristics such as grade level, race, socioeconomic status, EL status, and disability status. These disparities suggest that the approach districts use to identify “reading deficiencies” may be related to differing identification rates for various student subgroups. The marginal effects presented in Table 3.5 provide further insight into these relationships. 147 Table 3.5 Marginal Effects of “Reading Deficiency” Identification by Approach 1st Grade 2nd Grade 3rd Grade Female Black Hispanic Asian Other Race Econ. Dis. EL SWD Single 0.114* (0.066) 0.0933 (0.073) 0.0789 (0.069) -0.00726 (0.020) 0.343*** (0.095) 0.197*** (0.057) -0.224** (0.105) -0.00216 (0.077) 0.238*** (0.084) 0.0608 (0.099) 0.0564 (0.049) 5,653 Multiple 0.136*** (0.028) 0.120*** (0.034) 0.0603 (0.038) -0.00253 (0.010) 0.0678 (0.052) 0.0627*** (0.015) -0.105*** (0.025) 0.0232 (0.029) 0.149*** (0.020) -0.00164 (0.053) 0.187*** (0.031) 15,684 Hybrid 0.0750*** (0.028) 0.068 (0.048) 0.0581 (0.053) -0.0199 (0.015) 0.403*** (0.126) 0.0325 (0.031) -0.105 (0.084) 0.0924 (0.069) 0.182** (0.074) 0.269*** (0.092) 0.200*** (0.040) 4,888 Comprehensive 0.0752*** (0.024) 0.0626* (0.035) 0.111 (0.083) -0.0098 (0.007) 0.0821* (0.048) 0.0302 (0.026) -0.119** (0.057) 0.0263 (0.025) 0.245*** (0.030) -0.00537 (0.080) 0.183*** (0.019) 9,564 Classwork 0.320* (0.164) 0.354* (0.182) 0.391** (0.167) -0.0196 (0.017) 0.229*** (0.061) 0.122* (0.064) 0.159 (0.114) 0.209** (0.083) -0.046 (0.098) 0.0664 (0.060) 0.137* (0.072) 2,559 N Note. The table displays the marginal effects derived from Model (1). It highlights the statistically significant differences between specific independent variables and their reference categories within each identification approach (e.g., the difference in the likelihood of 1st graders vs. kindergarteners being identified under a Single Assessment approach). Statistically significant variances across different identification approaches are detailed in Appendix Table A3.D. The N value indicates the total number of students within districts that use each approach for identification. Robust standard errors in parenthesis. *** p <0.001 ** p < 0.01 * p <0.05. A consistent pattern emerges in Table 3.5: Black, Hispanic, and economically disadvantaged students face disproportionately higher identification rates compared to their White and non-economically disadvantaged peers under most approaches. In contrast, Asian students tend to be identified at significantly lower rates than White students. These disparities 148 are particularly pronounced under the Single Assessment and Hybrid approaches, where districts rely solely on one screening assessment, either alone or in combination with ELA classwork. Specifically, Black students are identified at strikingly higher rates: 34.3 and 40.3 percentage points above White students under these two respective approaches. Hispanic students also experience disproportionate identification, though with smaller gaps; the Single Assessment approach is related to a 19.7 percentage-point higher rate versus White students. Meanwhile, economically disadvantaged students are flagged at rates 15 to 25 percentage points higher than their non-economically disadvantaged peers across all approaches except Classwork Only. While these differences are correlational and may reflect larger genuine disparities in literacy skills among districts using certain approaches, the potential for sizable demographic gaps in identification rates—particularly with approaches relying on a single screening measure— warrants further investigation into the equity implications of each method. The patterns diverge for students with disabilities and ELs. Students with disabilities face significantly higher identification rates than their peers without disabilities across all approaches except Single Assessment. This could indicate that using multiple screening measures or incorporating ELA classwork provides a more comprehensive understanding of literacy abilities for this population. Districts may also intentionally aim to identify more students with disabilities through these multi-faceted approaches. Alternatively, larger disparities in literacy skills between students with and without disabilities could exist in districts using methods beyond just a single screener. In contrast, ELs do not consistently experience higher identification rates compared to non-ELs across most approaches. However, the Hybrid approach is a notable exception, associated with a 26.9 percentage-point higher rate for ELs. This suggests that the Hybrid method may more readily capture ELs as “reading deficient,” or that districts using this approach 149 have greater disparities between their EL and non-EL populations’ literacy skills. Grade-level trends also emerge, with older students, especially 1st graders, more likely to be identified than kindergarteners. As literacy skills become more readily apparent as students get older (Duke & Carlisle, 2010), this may enable improved identification in grade levels beyond kindergarten. The pattern could also reflect a strategic emphasis on timely intervention leading up to 3rd grade, when students could face retention under the Read by Grade Three Law during the year of the study (Michigan Public Act 306, 2016). To examine the demographic disparities in more depth, I revisit the LPM (Model 1) used to generate the marginal effects in Table 3.5 (full results in Appendix Table A3.4). This model allows for comparing each student subgroup’s identification rates under the Multiple Assessment, Hybrid, Comprehensive, and Classwork Only approaches to their respective rates under the Single Assessment reference approach. The analysis reveals non-significant interaction terms for grade levels, Asian students, ELs, students with disabilities, and economically disadvantaged students, indicating that their identification rates do not significantly differ across the alternative approaches compared to Single Assessment. However, the Black and Hispanic interaction terms show significant differences in identification rates. Specifically, Black students are less likely to be identified under the Multiple Assessment versus the Single Assessment approach, and Hispanic students face lower identification under Multiple Assessment, Hybrid, and Comprehensive approaches compared to Single Assessment. Post-hoc F-tests indicate additional differences in identification patterns across approaches for Black, Hispanic, and economically disadvantaged students beyond what the interaction terms capture. However, when conducting pairwise t-tests comparing the identification rates between each possible pair of approaches (e.g., Multiple Assessment vs. 150 Hybrid, Multiple Assessment vs. Comprehensive, etc.), none reach statistical significance after using the Bonferroni-adjusted p-value for multiple tests. These results offer several key insights. The significant F-tests suggest that, overall, there are meaningful differences in how various approaches identify Black, Hispanic, and economically disadvantaged students with “reading deficiencies.” However, the lack of significant pairwise t-tests indicates that while differences exist broadly, we cannot pinpoint which specific pairs of approaches differ significantly from each other in their identification patterns. This absence of significant pairwise differences might be due to the conservative nature of the Bonferroni correction or limited statistical power from our sample size, rather than a true lack of differences between approaches. Nonetheless, these findings hint at complex relationships between identification approaches and demographic disparities that are not fully captured by the current analysis. This underscores the need for further research with larger samples and more nuanced analytical approaches to disentangle these complex associations. Discussion Almost every state has an early literacy policy targeting K-3 student literacy improvement (ExcelinEd, 2022a). While evidence suggests these policies boost student achievement in the short term, research primarily examines 3rd-grade retention mandates without exploring other policy elements that could contribute to their effectiveness (Greene & Winters, 2004, 2006, 2007; Schwerdt et al., 2017; Slungaard Mumma & Winters, 2023; Westall & Cummings, 2023). Identifying students with “reading deficiencies” is fundamental to these policies, as it activates a suite of supports, including extra literacy instruction, targeted small- group and individual assistance, and family engagement. Thirty-seven states mandate screening assessments for this purpose (ExcelinEd, 2022a). However, there is a gap in our understanding of 151 how districts implement these requirements and which assessments they choose. This gap is significant, as ensuring that districts employ accurate and appropriate measures is crucial for identifying students who need additional support. This study presents the first known empirical evidence of the diverse methods districts use to identify students with “reading deficiencies” and how this relates to the proportion and characteristics of students identified, focusing on Michigan’s Read by Grade Three Law as a case study. Using superintendent survey data and administrative records from MDE for the 2021- 22 school year, the study uncovers a range of identification strategies, some better aligned with existing evidence than others. Most districts use multiple assessments, while about a quarter rely on a single assessment and about a third integrate students’ ELA classwork with assessments in various forms. There is a notable prevalence of CATs, especially NWEA, reflecting research that supports their relative accuracy and cost-effectiveness in student identification. Many districts combine CATs with CBMs, a combination particularly effective for enhancing identification precision. Despite this, the use of IRIs, like the Fountas & Pinnell Benchmark Assessment System, persists, even as evidence questions their effectiveness in accurately identifying students for literacy interventions (Barrett et al., 2023; Clemens et al., 2011; Kilgus et al., 2014; Ochs et al., 2018; Parker et al., 2015). The study also reveals that district strategies for identifying “reading deficiencies” are related to the proportion and characteristics of students identified. Single Assessment districts or those relying on Classwork Only tend to identify a higher proportion of students, highlighting a potential trend towards broader identification strategies or reflecting higher needs within these districts. However, the Single Assessment approach in particular, which is prevalent in districts serving high proportions of students from historically underserved populations and in districts 152 reporting lower revenues, raises concerns about the feasibility of providing effective interventions to such a large number of identified students. Districts integrating multiple measures, such as a mix of assessments and ELA classwork, typically report lower rates of “reading deficiency” identification. This trend could align with existing research suggesting that a more thorough approach can accurately assess students’ literacy needs and help prevent over- identification. Alternatively, this pattern may indicate that these districts have fewer students struggling with literacy or strategically choose to focus on a smaller group of students for intervention. Districts more frequently identify Black, Hispanic, and economically disadvantaged students with “reading deficiencies” across almost all approaches, with this trend most pronounced in districts using a single screening assessment. Conversely, districts that incorporate multiple assessments tend to identify Black and Hispanic students at significantly lower rates than the Single Assessment method. Because it is uncertain from the data available in this study whether the Single Assessment approach systematically over-identifies students from these groups or if alternative methods more accurately identify or even under-identify them, there is a reinforced need for further research to explore the accuracy and equity of various strategies by using external benchmarks, such as students’ assessment scores. This would enable researchers to identify discrepancies in how different methods align with students’ actual performance levels, offering insights into whether specific approaches indeed over- or under-identify students from particular demographic groups. Finally, it is important to note that the designation of students as “reading deficient” under the Read by Grade Three Law, while offering access to valuable literacy supports, carries the risk of stigmatizing those it aims to help, particularly if this is the terminology educators and 153 districts use to describe these students and notify their families of the designation. A substantial body of research shows that academic labels can adversely affect students’ self-esteem, academic self-concept, and achievement, with historically underserved students being particularly susceptible to over-identification (Arnold & Lassmann, 2003; Becker, 1963; Coutinho et al., 2002; Franz et al., 2023; Gold & Richards, 2012; Hattie, 2009; Ikbal et al., 2021; Jussim & Harber, 2005; Lapadat, 1998; Levin et al., 2018; Rist, 2017; Shifrer, 2016). Given that students from historically underserved groups are more frequently labeled with “reading deficiencies,” policymakers and practitioners should consider the implications of this labeling. Specifically, there is a need to evaluate whether the label itself might compromise the effectiveness of subsequent interventions due to the risk of stigmatization. Limitations This study offers important insights into the identification of students with “reading deficiencies,” yet is subject to several limitations that merit attention. One is the representation of a more advantaged sample of districts due to survey response rates. While propensity score weighting attempts to mitigate this bias and enhance alignment with the broader population of Michigan districts, this cannot fully capture the differences in how districts implement this aspect of early literacy policies. Additionally, the study’s low response rate may not adequately represent the full spectrum of identification practices across Michigan. This suggests a potential gap in understanding the varied approaches districts employ in implementing this aspect of the Read by Grade Three Law. The reliance on self-reported data further introduces the possibility of bias, as these responses may not fully encapsulate the on-the-ground realities of implementation. Furthermore, even within the same overarching identification approach category, districts may apply substantial variation in how they flag students for “reading deficiencies.” The 154 screening assessments used often provide a range of data for educators, including composite scores as well as sub-scores in specific areas of literacy. While research suggests using the composite score is optimal for identification purposes (Truckenmiller et al., 2024), this study cannot determine which specific scores districts employ to identify “reading deficiencies.” Additionally, the Read by Grade Three Law allows districts to set their own cut scores on screening assessments to identify students (Michigan Public Act 306, 2016). Consequently, even if districts are using the same assessment, they may select different cut scores, which could further affect student “reading deficiency” identification rates. As this study lacks data on the specific scores districts are using, the results cannot fully disentangle implications stemming from the broad approach category versus nuanced implementation differences under the same broad approach. This underscores the need for deeper investigation into the intricate details of how districts tailor identification processes, even within the same broader categories. Another key limitation is the inability to determine the direction of causality in the results. For example, regarding the relationship between district characteristics and their chosen approach for flagging “reading deficiencies,” the observed differences in approaches could stem from certain district contexts or circumstances leading them to select specific methods due to factors like resource availability, instructional alignment, or demographic compositions. Alternatively, districts that opt for similar approaches may simply tend to share common traits. This limitation prevents a full understanding of the underlying reasons districts gravitate towards certain methods for identifying students with “reading deficiencies.” This study also cannot establish the direction of causality between a district’s identification approach and “reading deficiency” identification rates. It is unclear whether the approach causes them to identify more or fewer students overall, or to identify certain student subgroups at higher rates than others. 155 Conversely, districts may aim to identify a certain number or percentage of students, or target specific subgroups for intervention, which could then influence their choice of identification approach. Without understanding the motivations driving districts’ approach selections, the causal mechanisms underlying the observed identification patterns remain ambiguous. Finally, this study is unable to evaluate the accuracy of different identification approaches in flagging students truly at-risk for literacy difficulties. While prior research suggests certain approaches may be better suited for screening purposes than others⎯and many Michigan districts are adopting these approaches⎯this study lacks external benchmark data on students’ actual literacy levels to compare against the “reading deficiency” designations. Without such benchmarks, it is impossible to determine which approach most reliably and equitably identifies the students genuinely in need of intervention and support. Resolving this limitation would require access to external assessment data that can validate the different identification methods. Directions for Future Research Given these limitations, several directions for future research emerge as critical to advancing our understanding of identifying students with “reading deficiencies.” First, future studies could include a broader and more diverse sample of districts, particularly those serving larger populations of historically underserved students. This would help ensure the findings represent how districts implement this aspect of early literacy policies across different educational settings. Second, future research could focus on gathering data on the specific scores and thresholds districts use to identify students with “reading deficiencies.” By surveying or interviewing districts, researchers can uncover not only whether districts are applying the 156 appropriate scores for screening decisions, but also the range of cut scores employed. This would allow for further investigation of the relationship between different approaches and student identification, revealing variations not captured in the present study. Third, future research could explore the reasons behind districts’ choice of identification approaches. While establishing a causal direction between district characteristics and identification approaches is not possible with the quantitative data available due to potential reverse causality issues, conducting qualitative studies, such as interviews with superintendents or other district-level decision-makers, could uncover valuable insights into the rationale for selecting specific strategies. These qualitative data could help establish whether certain characteristics like resources, student demographics, or institutional priorities indeed drive districts towards particular identification methods. Fourth, future studies could aim to disentangle the relationship between a district’s identification approach and the patterns of students flagged as having “reading deficiencies.” Qualitative data from district leaders could again illuminate their motivations and goals regarding identification numbers and priorities. This context could then be matched with quantitative modeling techniques to isolate the causal impact of approach choice in identification rates while controlling for potential confounding factors. Establishing this causal direction is critical for interpreting whether approach differences genuinely contribute to disproportionate identification patterns or merely reflect underlying district preferences. Fifth, identifying the most accurate method for determining “reading deficiencies” is crucial. Future studies could evaluate these identification approaches by comparing them against external benchmarks, such as standardized reading assessments that objectively measure students’ literacy skills. This analysis should include a diverse sample of districts, representing 157 the various approaches to identifying “reading deficiencies.” By collecting standardized reading scores from students across these districts, researchers can evaluate how well the different identification approaches align with students’ actual literacy skills. An approach that consistently identifies the same students as having “reading deficiencies” when compared to the external benchmark would demonstrate higher accuracy. Evaluating accuracy across various student subgroups is also important to assess equity in identification. Such research could help guide districts toward the most effective and equitable practices for identifying students in need of literacy support and intervention. Finally, investigating long-term outcomes for students identified as having “reading deficiencies” will be crucial for evaluating the impact of early literacy screening. Longitudinal studies that track students’ academic performance and well-being over time can provide valuable insights into the most beneficial identification strategies. This research could also help to identify potential unintended consequences of labeling students as “reading deficient” and how these labels affect students’ educational trajectories. Recommendations for Policy and Practice The findings from this study suggest actionable steps for policymakers and practitioners to improve the identification of students with “reading deficiencies.” It would be beneficial for MDE to refine its list of approved screening assessments, prioritizing those that are substantiated by research for accuracy. Currently, the list includes some assessments that are less effective in accurately identifying students with “reading deficiencies,” such as the Fountas & Pinnell Benchmark Assessment System (Barrett et al., 2023). If revising the list is impractical, an alternative could be to encourage districts to primarily use a CBM or CAT, which are more well- established for their accuracy than IRIs (Barrett et al., 2023; Clemens et al., 2011; Kilgus et al., 158 2014; Parker et al., 2015; Truckenmiller et al., 2024). This primary tool could be complemented by an additional assessment of the district’s choice, ideally another CBM or CAT, to further enhance precision. Such adjustments could help districts better align their practices with evidence-based approaches for identifying students requiring literacy support. The state has already made significant strides in providing resources to aid districts in their early literacy assessment practices, evidenced by comprehensive guides like the Early Literacy Assessment Systems that Support Learning and the Literacy Essentials series (Michigan Assessment Consortium, 2020; Michigan Association of Intermediate School Administrators General Education Leadership Network Early Literacy Task Force, 2022). These resources offer in-depth guidance on the administration and effective use of assessments. However, the variability in districts’ approaches to identifying “reading deficiencies” highlighted in this study indicates that further enhancements could streamline the implementation of these best practices. Enhancing training for district leaders and introducing succinct tools such as decision trees or checklists could bridge existing gaps. These resources could be built from preexisting work from researchers in the state (e.g., Truckenmiller et al., 2024) and could provide district leaders with clear, straightforward guidelines to develop and maintain assessment systems that are not only coherent and comprehensive but also align with best practices for student identification. The state also already provides valuable support for benchmark assessments, which are mandated in reading and math for K-8 students (MDE, 2023). This existing funding sets a strong precedent for extending similar financial support to screening assessments for identifying “reading deficiencies.” While some of the assessments approved by MDE for “reading deficiency” identification can also serve as benchmark assessments and thus qualify for existing funding, not all of these assessments are dual-purpose. By broadening financial assistance to 159 include both types of assessments, the state could better ensure that all districts, especially those facing financial constraints, have the flexibility to choose from a wider range of assessments. This could prevent districts from being limited to less effective or less appropriate options due to budgetary limitations, thereby supporting a more equitable and effective approach to early literacy screening. Finally, the state’s Center for Educational Performance and Information could enhance its data collection to include students’ scores from “reading deficiency” screening assessments, similar to its practices for benchmark assessments. This extension could facilitate more nuanced analyses of outcomes for students flagged as having “reading deficiencies,” analyzing their performance on these specific assessments. Collecting additional data on the cut scores districts use on these assessments could also allow for a thorough evaluation of how consistently and effectively these scores identify students who require additional support. This expanded data collection could provide critical insights into the effectiveness of early literacy identification practices across the state. Conclusion This study sheds light on Michigan districts’ varied approaches to identifying K-3 students with “reading deficiencies” under the Read by Grade Three Law. While a substantial proportion of districts rely exclusively on screening assessments, many integrate students’ ELA classwork into their identification processes. Districts’ preference for CBMs and CATs aligns with literature emphasizing their accuracy and cost-effectiveness in identifying students in need. However, many districts’ continued use of less precise measures suggests a gap between research recommendations and practice. Moreover, the study uncovers a correlation between the choice of identification approach and the proportion and demographics of students identified as “reading 160 deficient.” Districts using a Single Assessment tend to identify a higher percentage of students— and significantly higher percentages of students from historically underserved groups—than districts using other approaches. The Single Assessment approach is also more common in districts reporting lower revenues, raising additional concerns about their capacity to deliver interventions to these students. Conversely, strategies incorporating multiple assessments or ELA classwork tend to be associated with lower identification rates, especially for Black and Hispanic students. The findings underscore a need for policy adjustments and enhanced support for districts to ensure that identification practices are grounded in evidence-based strategies. By refining the list of approved assessments to include only those validated by research for their accuracy and encouraging districts to use CBMs or CATs as primary tools, the state can support a more accurate identification of students needing support. Moreover, expanding resources to assist districts in making informed decisions about screening assessments—considering factors such as accuracy, cost, and alignment with existing curriculum—could empower districts to select the most effective tools. Financial support for the acquisition of these assessment and systematic collection of screening assessment data could further enhance the equitable and effective identification of students facing literacy challenges. As Michigan and other states continue to refine their early literacy policies, integrating these insights and recommendations could significantly improve the identification process, ensuring that students receive the timely and targeted support necessary for their academic success. 161 CONCLUSION The adoption and implementation of early literacy policies across the U.S. reflect a national commitment to improving foundational literacy skills among young learners. As evidenced by the widespread enactment of such initiatives, policymakers recognize early literacy’s pivotal role in educational trajectories and life outcomes. This dissertation has contributed to the body of knowledge on early literacy policies by providing an in-depth analysis of the development and implementation of these policies, with a particular focus on Michigan’s Read by Grade Three Law. The insights from these three papers highlight several key takeaways for policymakers. First, the policy process surrounding the development of Michigan’s Read by Grade Three Law underscores the critical role of collaborative efforts among legislators, educators, policy entrepreneurs, and stakeholders in crafting early literacy policies. The policy’s successful enactment depended on several interconnected factors: acknowledging the issue of low literacy rates, an established and ongoing commitment to enhancing early literacy statewide, and a political environment open to compromise on policy content. Additionally, the influence of organizations like ExcelinEd underlines the significant role of advocacy in forming these policies, leading to a national trend of similar policy designs. Second, the examination of family engagement and screening assessments within early literacy policies highlights significant opportunities and notable challenges in their implementation. Despite the clear benefits of involving families in literacy development, the low implementation of “Read at Home” plans points to a crucial need for strategies that better engage families and educators. The state could play a pivotal role by offering capacity-building supports, such as training for educators and templates for effective “Read at Home” plans, especially in 162 districts with many students eligible for these plans. Furthermore, the varied approaches to identifying students with “reading deficiencies” emphasize the importance of districts selecting and using measures that accurately identify students who need extra support. To aid in this, the state could streamline the list of approved screening tools and refine its guidance on building robust early literacy assessment systems. While efforts to enhance these processes are in progress, some districts continue to use measures proven ineffective by research, suggesting a need for additional state support in this area. For educators and district administrators, this research underscores the critical role of building strong connections with families and using research-backed assessment methods. Improving how educators understand and perceive family engagement strategies could significantly boost the implementation of “Read at Home” plans. This might require revising the current plans to emphasize training for caregivers, bidirectional communication between educators and families, and customization to meet the diverse needs of students and their families. Moreover, it is essential to adopt screening methods that accurately pinpoint K-3 students with “reading deficiencies,” ensuring that districts identify those who most need support. Districts should thoughtfully choose assessment tools that current research supports as both accurate and cost-effective. This selection process should also consider the district’s specific circumstances, such as its resources, student demographics, and curriculum and instructional strategies, to ensure the chosen assessment system aligns well with these factors. While this dissertation has contributed to our understanding of early literacy policies, it also opens avenues for further investigation. Future studies could explore the long-term effects of “Read at Home” plans and “reading deficiency” identification on student achievement. Such research would build on existing work that assesses the general impact of early literacy policies 163 and the specific consequences of 3rd-grade retention mandates, aiming to pinpoint which components most significantly contribute to the positive outcomes observed in earlier investigations. This research will be invaluable as policymakers continue implementing and revising these policies nationwide. To truly grasp the efficacy of different policy elements, researchers will also need more comprehensive data, especially regarding students’ scores on screening assessments and standardized tests that can serve as external benchmarks. This underscores a call to action for state data agencies to broaden the scope of information collected on early literacy policies. This would support more nuanced research and provide valuable insights for policymakers striving to refine and implement more effective early literacy interventions. Additionally, there is significant potential for collecting rich qualitative data to deepen our understanding of how family engagement initiatives and identifying “reading deficiencies” unfold in practice. Exploring families’ experiences with receiving and using “Read at Home” plans can offer valuable insights into their perceptions of the effectiveness and accessibility of these strategies. Furthermore, a systematic content analysis of the plans themselves could provide critical insights into how they align with best practices in family engagement and highlight variations across districts. Equally important are the perspectives of district leaders on their approaches to devising “Read at Home” plans and determining eligibility for them, as well as their criteria for selecting “reading deficiency” screening tools and the considerations that influence these choices. Such qualitative insights can shed light on aspects of policy implementation that remain obscured in the current studies and uncover the nuances of how stakeholders enact these policies on the ground. Finally, to enhance our understanding of early literacy policies, it will be crucial to 164 broaden the scope of research to include other policy components and to examine how different states implement these policies. This dissertation, alongside EPIC’s ongoing evaluation of the Read by Grade Three Law (see Strunk et al., 2021, 2022), offers detailed insights into implementing early literacy policies in Michigan. However, given that nearly every state has enacted such policies, each with its unique components and within diverse contexts, gathering data from a broader range of states is essential for painting a comprehensive picture of early literacy policy implementation at the national level. Expanding research to explore other facets of early literacy policies is also vital. Beyond the extensively studied area of retention and this dissertation’s focus on family engagement and the identification of “reading deficiencies,” these policies often encompass additional instructional time in literacy, one-on-one and small group tutoring, summer support programs, and many other components (ExcelinEd, 2022a). Understanding the full spectrum of these varied components will require ongoing and comprehensive data collection. These efforts will be instrumental in informing policy discussions, helping to determine which elements of early literacy policies are most critical for boosting student literacy outcomes. This dissertation highlights the complexity of enacting and implementing early literacy policies, focusing on Michigan’s Read by Grade Three Law as a case study. The findings reveal the multifaceted nature of policy development and the complexities of implementing various policy mandates, including family engagement and the identification of students needing additional literacy support. By addressing these key areas, this research contributes valuable insights for policymakers, educators, and stakeholders aiming to enhance early literacy outcomes. Moving forward, it will be imperative that the lessons learned from Michigan’s experience inform the ongoing refinement and implementation of early literacy policies, ensuring 165 that they effectively support the literacy development of all students. 166 BIBLIOGRAPHY Acadience Learning. (n.d.). Acadience Reading. Acadience Learning. Retrieved February 19, 2024, from https://acadiencelearning.org/acadience-reading/ Adams, K. S., & Christenson, S. (2000). Trust and the family-school relationship examination of parent-teacher differences in elementary and secondary grades. Journal of School Psychology, 38(5), 477–497. https://doi.org/10.1016/S0022-4405(00)00048-0 Aikins, N. L., & Barbarin, O. (2008). Socioeconomic differences in reading trajectories: The contribution of family, neighborhood, and school contexts. Journal of Educational Psychology, 100(2), 235–251. https://doi.org/10.1037/0022-0663.100.2.235 Alaska Department of Education & Early Development. (n.d.). The Alaska Reads Act. Alaska Department of Education & Early Development. Retrieved July 26, 2021, from https://education.alaska.gov/akreads Albright, M. I., & Weissberg, R. P. (2009). School-family partnerships to promote social and emotional learning. In Handbook of school-family partnerships (1st ed., p. 20). Routledge. Annenberg Institute for School Reform. (2014). Public accountability for charter schools: Standards and policy recommendations for effective oversight. Annenberg Institute for School Reform. https://files.eric.ed.gov/fulltext/ED558016.pdf Archibald, S. (2009). Narrowing in on educational resources that do affect student achievement. Peabody Journal of Education, 81(4), 23–42. https://doi.org/10.1207/s15327930pje8104_2 Arkansas Comprehensive Testing, Assessment, and Accountability Program Act, AR Code § 6- 15-420 (2015). Arnold, M., & Lassmann, M. E. (2003). Overrepresentation of minority students in special education. Education, 124(2), 230–236. Arsen, D., Delpier, T., Gensterblum, A., Jacobsen, R., & Stamm, A. (2022). Educational opportunities and community development in rural Michigan: A roadmap for state policy. College of Education, Michigan State University. Baker, C. E. (2013). Fathers’ and mothers’ home literacy involvement and children’s cognitive and social emotional development: Implications for family literacy programs. Applied Developmental Science, 17(4), 184–197. https://doi.org/10.1080/10888691.2013.836034 Baker, T. L., Wise, J., Kelley, G., & Skiba, R. J. (2016). Identifying barriers: Creating solutions to improve family engagement. School Community Journal, 26(2), 161–184. Bardach, E. (1979). The implementation game. MIT Press. Barrett, C. A., Johnson, L. J., Truckenmiller, A. J., & VanDerHeyden, A. M. (2023). Comparing 167 the cost-accuracy ratios of multiple approaches to reading screening in elementary school. Remedial and Special Education. https://doi.org/10.1177/07419325231190809 Baumann, J. F., Hoffman, J. V., Duffy-Hester, A. M., & Ro, J. M. (2000). “The first r” yesterday and today: U.S. elementary reading instruction practices reported by teachers and administrators. Reading Research Quarterly, 35(3), 338–377. Becker, H. (1963). Outsiders: Studies in the sociology of deviance. Bennett, C. J. (1991). What is policy convergence and what causes it? British Journal of Political Science, 21(2), 215–233. Bergin, T. (2010). How and why do criminal justice public policies spread throughout US states? A critical review of the diffusion literature. Criminal Justice Policy Review, 22(4), 403– 421. Bidwell, C. E., & Kasarda, J. D. (1975). School district organization and student achievement. American Sociological Review, 40(1), 55–70. https://doi.org/10.2307/2094447 Bishop, A. G., & League, M. B. (2006). Identifying a multivariate screening model to predict reading difficulties at the onset of kindergarten: A longitudinal analysis. Learning Disability Quarterly, 29(4), 235–305. https://doi.org/10.2307/30035552 Boonk, L., Gijselaers, H. J. M., Ritzen, H., & Brand-Gruwel, S. (2018). A review of the relationship between parental involvement indicators and academic achievement. Educational Research Review, 24, 10–30. https://doi.org/10.1016/j.edurev.2018.02.001 Bovens, M., & ’t Hart, P. (1996). Understanding policy fiascoes (1st ed.). Routledge. Bowen, N., & Bowen, G. (1998). The effects of home micro system risk factors and school micro system protective factors on student academic performance and affective investment in schooling. Social Work in Education, 20, 219–231. Bridwell-Mitchell, E. N. (2015). Theorizing teacher agency and reform: How institutionalized instructional practices change and persist. Sociology of Education, 88(2), 140–159. https://doi.org/10.1177/0038040715575559 Brodkin, E. Z. (1997). Inside the welfare contract: Discretion and accountability in state welfare administration. Social Service Review, 71(1). Bronfenbrenner, U. (1979). Contexts of child rearing: Problems and prospects. American Psychologist, 34(10), 844–850. https://doi.org/10.1037/0003-066X.34.10.844 Brown, C. L., Schell, R., Denton, R., & Knode, E. (2019). Family literacy coaching: Partnering with parents for reading success. School Community Journal, 29(1), 63–86. Brown, C. P. (2007). Examining the streams of a retention policy to understand the politics of high-stakes reform. Education Policy Analysis Archives, 15, 9. 168 Brown, S., Riddell, S., & Duffield, J. (2013). Responding to pressures: A study of four secondary schools. In Contemporary issues in teaching and learning (p. 264). Routledge. Bryk, A. S., Gomez, L., Grunow, A., & LeMahieu, P. (2015). Learning to improve: How America’s schools can get better at getting better. Harvard Education Publishing. Bulkley, K., & Fisler, J. (2003). A decade of charter schools: From theory to practice. Educational Policy, 17(3), 291–391. https://doi.org/10.1177/0895904803017003002 Çaliskan, E. F., & Ulas, A. H. (2022). The effect of parent-involved reading activities on primary school students’ reading comprehension skills, reading motivation, and attitudes towards reading. International Journal of Elementary Education, 14(4), 509–524. Carter, L. E., & LaPlant, J. T. (1997). Diffusion of health care policy innovation in the United States. State and Local Government Review, 29(1), 17–26. Caspe, M., Lopez, M. E., Chu, A., & Weiss, H. B. (2011). Teaching to teachers: Preparing educators to engage families for student achievement [Issue Brief]. Harvard Family Research Project. Castro, M., Expósito-Casas, E., López-Martín, E., Lizasoain, L., Navarro-Asencio, E., & Gaviria, J. L. (2015). Parental involvement on student achievement: A meta-analysis. Educational Research Review, 14, 33–46. https://doi.org/10.1016/j.edurev.2015.01.002 Center for Public Education. (2024). Urban student characteristics and urban school challenges: What high-quality education for every student means. National School Boards Association. https://www.nsba.org/-/media/CPE---2024---Urban-Student-Characteristics- and-Urban-School-Challenges.pdf Cholakova, M., & Ravasi, D. (2019). Why do individuals perceive and respond to the same institutional demands differently? On the cognitive structural underpinnings of institutional complexity. In Microfoundations of institutions (Vol. 65A, pp. 99–118). Emerald Publishing Limited. Chrispeels, J. H., & Rivero, E. (2001). Engaging Latino families for student success: How parent education can reshape parents’ sense of place in the education of their children. Peabody Journal of Education, 76(2), 119–169. Christenson, S., Rounds, T., & Gorney, D. (1992). Family factors and student achievement: An avenue to increase students’ success. School Psychology Quarterly, 7(3), 178–206. https://doi.org/10.1037/h0088259 Christenson, S., Whitehouse, E. M., & VanGetson, G. R. (2007). Partnering with families to enhance students’ mental health. In Transforming school mental health services: Population-based approaches to promoting the competency and wellness of children. Corwin Press. Clarke, B. L., Koziol, N. A., & Sheridan, S. M. (2017). The effects of rurality on parents’ 169 engagement in children’s early literacy. In Rural education research in the United States. Springer. Clarke, B. L., Sheridan, S. M., & Woods, K. E. (2009). Elements of healthy family-school relationships. In Handbook of school-family partnerships (1st ed., p. 19). Routledge. Clemens, N. H., Shapiro, E. S., & Thoemmes, F. (2011). Improving the efficacy of first grade reading screening: An investigation of word identification fluency with other early literacy indicators. School Psychology Quarterly, 26(3), 231–244. https://doi.org/10.1037/a0025173 Coburn, C. E. (2004). Beyond decoupling: Rethinking the relationship between the institutional environment and the classroom. Sociology of Education, 77(3), 211–244. https://doi.org/10.1177/003804070407700302 Coburn, C. E. (2005). Shaping teacher sensemaking: School leaders and the enactment of reading policy. Educational Policy, 19(3), 476–509. https://doi.org/10.1177/0895904805276143 Coburn, C. E. (2006). Framing the problem of reading instruction: Using frame analysis to uncover the microprocesses of policy implementation. American Educational Research Journal, 43(3), 343–349. https://doi.org/10.3102/00028312043003343 Colorado Revised Statutes, § 22-7-1205. Compton, D. L., Fuchs, D., Fuchs, L. S., Bouton, B., Gilbert, J. K., Barquero, L. A., Cho, E., & Crouch, R. C. (2010). Selecting at-risk first-grade readers for early intervention: Eliminating false positives and exploring the promise of a two-stage gated screening process. Journal of Educational Psychology, 102(2), 327–340. https://doi.org/10.1037/a0018448 Conradi Smith, K. C., Parsons, A. W., Vaughn, M., & Yatzek, J. C. (2019). Elementary students’ text diets: Results from a nationwide study. Literacy Research Association. Corcoran, T., & Goertz, M. (1995). Instructional capacity and high performance schools. Educational Researcher, 24(9), 27–31. https://doi.org/10.3102/0013189X024009027 Council of Chief State School Officers. (2019). Third grade reading laws: Implementation and impact. Council of Chief State School Officers. https://ccsso.org/sites/default/files/2019- 08/CCSSO%20CEELO%20third%20grade%20reading.pdf Coutinho, M. J., Oswald, D. P., & Best, A. M. (2002). The influence of sociodemographics and gender on the disproportionate identification of minority students as having learning disabilities. Remedial and Special Education, 23(1). https://doi.org/10.1177/074193250202300107 Crosby, S. A., Rasinski, T., Padak, N., & Yildirim, K. (2015). A 3-year study of a school-based parental involvement program in early literacy. The Journal of Educational Research, 170 108, 165–172. https://doi.org/10.1080/00220671.2013.867472 Cummings, A., Strunk, K. O., & DeVoto, C. (2023). “A lot of states were doing it”: The development of Michigan’s Read by Grade Three Law. Journal of Educational Change, 24, 107–132. https://doi.org/10.1007/s10833-021-09438-y Cunningham, A. E., & Stanovich, K. E. (1997). Early reading acquisition and its relation to reading experience and ability 10 years later. Developmental Psychology, 33(6), 934– 945. Curenton, S. M., & Justice, L. M. (2008). Children’s preliteracy skills: Influence of mothers’ education and beliefs about shared-reading interactions. Early Education and Development, 19(2), 261–283. https://doi.org/10.1080/10409280801963939 Curriculum Associates. (2024). I-Ready Assessment. Curriculum Associates. https://www.curriculumassociates.com/programs/i-ready-assessment Daniel, S. S., Walsh, A. K., Goldston, D. B., Arnold, E. M., Reboussin, B. A., & Wood, F. B. (2006). Suicidality, school dropout, and reading problems among adolescents. Journal of Learning Disabilities, 39(6), 507–514. https://doi.org/10.1177/00222194060390060301 Darling, S., & Westberg, L. (2004). Parent involvement in children’s acquisition of reading. The Reading Teacher, 57(8), 774–776. Darling-Hammond, L. (2013). Inequality and school resources: What it will take to close the opportunity gap. In Closing the opportunity gap: What American must do to give every child an even chance (p. 272). Oxford University Press. De Carvalho, M. E. (2000). Rethinking family-school relations (1st ed.). Routledge. Debray, E., Parson, G., & Avila, S. (2003). Internal alignment and external pressure: High school responses in four state contexts. In The new accountability (1st ed., p. 31). Routledge. Dewulf, A., & Bouwen, R. (2012). Issue framing in conversations for change: Discursive interaction strategies for “doing differences.” The Journal of Applied Behavioral Science, 48(2), 168–193. Diamond, J. B., & Spillane, J. P. (2004). High-stakes accountability in urban elementary schools: Challenging or reproducing inequality? Teachers College Record, 106(6), 1145–1176. https://doi.org/10.1111/j.1467-9620.2004.00375.x Diehl, D. K., & Golann, J. W. (2023). An integrated framework for studying how schools respond to external pressures. Educational Researcher, 52(5), 296–305. https://doi.org/10.3102/0013189X231159599 Doig, J. W., & Hargrove, E. C. (1990). Leadership and innovation: Entrepreneurs in government. Johns Hopkins University Press. 171 Dolowitz, D., & Marsh, D. (1996). Who learns what from whom: A review of the policy transfer literature. Political Studies, XLIV, 343–357. Dolowitz, D., & Marsh, D. (2000). Learning from abroad: The role of policy transfer in contemporary policy-making. Governance: An International Journal of Policy and Administration, 13(1), 5–24. Doyle, A., & Zhang, J. (2011). Participation structure impacts on parent engagement in family literacy programs. Early Childhood Education Journal, 39, 223–233. Duke, N. K., & Carlisle, J. (2010). The development of comprehension. In Handbook of reading research (Vol. 4, pp. 199–228). Routledge. Education Commission of the States. (2018). State kindergarten-through-third-grade policies. Education Commission of the States. http://ecs.force.com/mbdata/MBQuest2rTanw?rep=KK3Q1818 Edwards, P. A. (2016). New ways to engage parents: Strategies and tools for teachers and leaders, K-12. Teachers College Press. Edwards, P. A., & Danridge, J. C. (2003). Developing collaboration with culturally diverse parents. In Collaboration for Diverse Learners (1st ed.). Routledge. Edwards, P. A., Smith, P., & McNair, J. C. (2022). Toward culturally relevant literacies with children and families of color. In International encyclopedia of education (4th ed.). https://d2l.msu.edu/d2l/le/content/1367376/viewContent/12318037/View Edwards, P. A., White, K. L., & Bruner, L. E. (2019). Adopting and adapting Michigan’s tenth essential literacy practice: Collaboration with families. Michigan Reading Journal, 51(3), 37–46. Elbro, E. R., & Scarborough, H. S. (2004). Early identification. In Handbook of Children’s Literacy. Springer. Emerson, L., Fear, J., Fox, S., & Sanders, E. (2012). Parental engagement in learning and schooling: Lessons from research. Australian Research Alliance for Children and Youth. https://www.communityhubs.org.au/wp- content/uploads/2017/10/Parental_engagement_in_learning_and_schooling_Lessons_fro m_research_BUREAU_ARACY_August_2012.pdf Englund, M. M., Luckner, A. E., Whaley, G. J. L., & Egeland, B. (2004). Children’s achievement in early elementary school: Longitudinal effects of parental involvement, expectations, and quality of assistance. Journal of Educational Psychology, 96(4), 723– 730. Evans, M. A., Shaw, D., & Bell, M. (2000). Home literacy activities and their influence on early literacy skills. Canadian Journal of Experimental Psychology, 54(2), 65–75. 172 Evans, M., & Davies, J. (1999). Understanding policy transfer: A multi-level, multi-disciplinary perspective. Public Administration, 77(2), 361–385. ExcelinEd. (2020). Comprehensive early literacy model policy. ExcelinEd. https://excelined.org/wp- content/uploads/2020/11/ExcelinEd.PolicyToolkit.EarlyLiteracy.ModelPolicy.pdf ExcelinEd. (2022a). Comprehensive early literacy policy: A state-by-state analysis of fundamental principles. ExcelinEd. https://excelined.org/wp- content/uploads/2021/10/ExcelinEd_PolicyToolkit_EarlyLiteracy_StatebyStateAnalysis_ 2021.pdf ExcelinEd. (2022b). Early literacy. ExcelinEd. https://excelined.org/policy-playbook/early- literacy/ Fan, X., & Chen, M. (2001). Parental involvement and students’ academic achievement: A meta- analysis. Educational Psychology Review, 13(1), 1–22. https://doi.org/1040- 726X/01/0300-0001$19.50/0 Felton, R. H. (1992). Early identification of children at risk for reading disabilities. Topics in Early Childhood Special Education, 12(2), 212–229. Fiester, L. (2013). Early warning confirmed: A research update on third-grade reading. The Annie E. Casey Foundation. Fiester, L., & Smith, R. (2010). Early warning! Why reading by the end of third grade matters. The Annie E. Casey Foundation. Fikrat-Wevers, S., van Steensel, R., & Arends, L. R. (2021). Effects of family literacy programs on the emergent literacy skills of children from low-SES families: A meta-analysis. Review of Educational Research, 91(4), 577–613. https://doi.org/10.3102/0034654321998075 Finn, C. E., Manno, B. V., & Vanourek, G. (2001). Charter schools in action: Renewing public education. Princeton University Press. Fiorina, M. P., & Abrams, S. J. (2008). Political polarization in the American public. Annual Review of Political Science, 11, 563–588. Florida Department of Education. (n.d.). Read-at-home plan resources. Retrieved January 31, 2024, from https://www.fldoe.org/core/fileparse.php/7539/urlt/Read-at-HomePlanRes.pdf Ford, M. P., & Opitz, M. F. (2008). A national survey of guided reading practices: What we can learn from primary teachers. Literacy Research and Instruction, 47(4), 309–331. Foster, M. D., Young, J. R., & Young, J. L. (2017). Teacher perceptions of parental involvement and the achievement of diverse learners: A meta-analysis. Journal of Ethical Educational Leadership, 4(5), 1–17. 173 Fountas, I., & Pinnell, G. S. (2016). Fountas & Pinnell Benchmark Assessment System. Heinemann. Franz, D. J., Richter, T., Lenhard, W., Marx, P., Stein, R., & Ratz, C. (2023). The influence of diagnostic labels on the evaluation of students: A multilevel meta-analysis. Educational Psychology Review, 35(1), 1–41. https://doi.org/10.1007/s10648-023-09716-6 Fuchs, L. S., Fuchs, D., & Deno, S. L. (1982). Reliability and validity of curriculum-based informal reading inventories. Reading Research Quarterly, 18(1), 6–26. https://doi.org/10.2307/747536 Galindo, C., & Sheldon, S. B. (2012). School and home connections and children’s kindergarten achievement gains: The mediating role of family involvement. Early Childhood Research Quarterly, 27(1), 90–103. https://doi.org/10.1016/j.ecresq.2011.05.004 Ganon-Shilon, S., & Schechter, C. (2017). Making sense of school leaders’ sense-making. Educational Management Administration & Leadership, 45(4), 682–698. https://doi.org/10.1177/1741143216628536 Georgiou, G. K., Inoue, T., & Parrila, R. (2021). Developmental relations between home literacy environment, reading interest, and reading skills: Evidence from a 3-year longitudinal study. Child Development, 92(5), 2053–2068. https://doi.org/10.1111/cdev.13589 Ghaith, G., & Yaghi, H. (1997). Relationships among experience, teacher efficacy, and attitudes toward the implementation of instructional innovation. Teaching and Teacher Education, 13(4), 451–458. https://doi.org/10.1016/S0742-051X(96)00045-5 Gioia, D. A., & Chittipeddi, K. (1991). Sensemaking and sensegiving in strategic change initiation. Strategic Management Journal, 12(6), 433–448. Glaser, B., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for qualitative research. Aldine De Gruyter. Gold, M. E., & Richards, H. (2012). To label or not to label: The special education question for African Americans. Educational Foundations, 26(1–2), 143–156. Gomby, D., Culross, P., & Behrmann, R. (1999). Home visiting: Recent program evaluations, analysis and recommendations. The Future of Children, 9(1), 4–26. Goodsell, C. T. (1981). Looking once again at human service bureaucracy. The Journal of Politics, 43, 763–778. https://doi.org/10.2307/2130636 Google. (n.d.). How results are automatically generated. Google Search. Retrieved May 30, 2024, from https://www.google.com/search/howsearchworks/how-search-works/ranking- results/#:~:text=Beyond%20looking%20at%20keywords%2C%20our,learned%20system s%20better%20estimate%20relevance. Gormley, W. T. (1987). Bureau-bashing: A framework for analysis. American Political Science 174 Association, Chicago, IL. Graves Jr., S. L., & Wright, L. B. (2011). Parent involvement at school entry: A national examination of group differences and achievement. School Psychology International, 32(1), 35–48. https://doi.org/10.1177/0143034310396611 Grebing, E. M., Edmunds, J. A., & Arshavsky, N. P. (2023). The relationship between buy-in and implementation: Measuring teacher buy-in to a high school reform effort. Evaluation and Program Planning, 97. https://doi.org/10.1016/j.evalprogplan.2023.102224 Greene, J. P., & Winters, M. A. (2004). An evaluation of Florida’s program to end social promotion (7; Education Working Paper). Center for Civic Innovation at the Manhattan Institute. Greene, J. P., & Winters, M. A. (2006). Getting farther ahead by staying behind: A second-year evaluation of Florida’s policy to end social promotion (49; Civic Report). Center for Civic Innovation at the Manhattan Institute. Greene, J. P., & Winters, M. A. (2007). Revisiting grade retention: An evaluation of Florida’s test-based promotion policy. Education Finance and Policy, 2(4), 319–342. Greene, J. P., & Winters, M. A. (2009). The effects of exemptions to Florida’s test-based promotion policy: Who is retained? Who benefits academically? Economics of Education Review, 28, 135–142. Grice, S. (2020). Perceptions of family engagement between African American families and schools: A review of the literature. Journal of Multicultural Affairs, 5(2). Hall, C. M. (2020). The impact of family engagement on student achievement. Networks: An Online Journal for Teacher Research, 22(2). Hattie, J. (2009). Visible learning: A synthesis of over 800 meta-analyses relating to achievement. Routledge. He, B., & Thompson, C. (2022). Family involvement and English learners’ outcomes: A synthetic analysis. International Review of Education, 68, 409–440. Heltzel, G., & Laurin, K. (2020). Polarization in America: Two possible futures. Current Opinion in Behavioral Sciences, 34, 179–184. Hemmerechts, K., Orhan, A., & Dimokritos, K. (2017). The relationship between parental literacy involvement, socio-economic status and reading literacy. Educational Review, 69(1), 85–101. https://doi.org/10.1080/00131911.2016.1164667 Henderson, A. T., & Mapp, K. L. (2002). A new wave of evidence: The impact of school, family, and community connections on student achievement. National Center for Family and Community Connections with Schools. https://sedl.org/connections/resources/evidence.pdf 175 Hernandez, D. J. (2011). Double jeopardy: How third-grade reading skills and poverty influence high school graduation. The Annie E. Casey Foundation. Hess, F. M., & Finn, C. E. (2004). Leaving no child behind? Options for kids in failing schools. Palgrave Macmillan. Hess, F. M., & Kelly, A. P. (2005). Learning to lead? What gets taught in principal preparation programs. Program on Education Policy and Governance at Harvard University. Hill, H. C. (2003). Understanding implementation: Street-level bureaucrats’ resources for reform. Journal of Public Administration Research and Theory, 13(3), 265–282. https://doi.org/10.1093/jopart/mug024 Hill, M., & Bramley, G. (1986). Analysing social policy. Blackwell. Hill, M., & Hupe, P. (2002). Implementing public policy: Governance in theory and practice. Sage Publications. Honig, M. I. (2006). New directions in education policy implementation: Confronting complexity. State University of New York Press. Hopkins, D. J., & Sides, J. (2015). Political polarization in American politics. Bloomsbury. Hornby, G., & Lafaele, R. (2011). Barriers to parental involvement in education: An explanatory model. Educational Review, 63(1), 37–52. https://doi.org/10.1080/00131911.2010.488049 Hoxby, C. M. (2003). School choice and school productivity: Could school choice be a tide that lifts all boats? In The economics of school choice. University of Chicago Press. Ikbal, M. S., Latuconsina, N. K., & Syamsinar. (2021). Analysis of the impact of labeling on students’ interest in learning physics. Journal of Physics: Conference Series. https://doi.org/10.1088/1742-6596/1760/1/012012 Illuminate Education. (n.d.). FastBridge. Illuminate Education. Retrieved February 19, 2024, from https://www.illuminateed.com/products/fastbridge/ Inoue, T., Georgiou, G. K., Parrila, R., & Kirby, J. R. (2018). Examining an extended home literacy model: The mediating roles of emergent literacy skills and reading fluency. Scientific Studies of Reading, 22(4), 273–288. https://doi.org/10.1080/10888438.2018.1435663 Ishimaru, A. M. (2017). From family engagement to equitable collaboration. Educational Policy, 33(2), 350–385. https://doi.org/10.1177/0895904817691841 Jacob, B. A. (2017). When evidence is not enough: Findings from a randomized evaluation of Evidence-Based Literacy Instruction (EBLI) (02–2015; EPI Working Papers). Gerald R. Ford School of Public Policy Education Policy Initiative, University of Michigan. 176 Jacob, B. A., & Lefgren, L. (2004). Remedial education and student achievement: A regression- discontinuity analysis. The Review of Economics and Statistics, 86(1), 226–244. Jacob, B. A., & Lefgren, L. (2009). The effect of grade retention on high school completion. American Economic Journal: Applied Economics, 1(3), 33–58. Jeynes, W. H. (2003). A meta-analysis: The effects of parental involvement on minority children’s academic achievement. Education and Urban Society, 35(2). https://doi.org/10.1177/0013124502239392 Jeynes, W. H. (2012). A meta-analysis of the efficacy of different types of parental involvement programs for urban students. Urban Education, 47(4), 706–742. https://doi.org/10.1177/0042085912445643 Jeynes, W. H. (2016). A meta-analysis: The relationship between parental involvement and African American school outcomes. Journal of Black Studies, 47(3), 195–216. https://doi.org/10.1177/0021934715623522 Jeynes, W. H. (2017). A meta-analysis: The relationship between parental involvement and Latino student outcomes. Education and Urban Society, 49(1), 4–28. https://doi.org/10.1177/0013124516630596 Jimenez-Castellanos, O. (2010). Relationship between educational resources and school achievement: A mixed method intra-district analysis. The Urban Review, 42, 351–371. Jones, M. D., Peterson, H. L., Pierce, J. J., Herweg, N., Bernal, A., Lamberta Raney, H., & Zahariadis, N. (2016). A river runs through it: A multiple streams meta-review. Policy Studies Journal, 44(1), 13–36. Jussim, L., & Harber, K. D. (2005). Teacher expectations and self-fulfilling prophecies: Knowns and unknowns, resolved and unresolved controversies. Personality and Social Psychology Review, 9(2), 131–155. https://doi.org/10.1207/s15327957pspr0902_3 Keiser, L. R. (1999). State bureaucratic discretion and the administration of social welfare programs: The case of Social Security Disability. Journal of Public Administration Research and Theory, 9(1), 87–106. https://doi.org/10.1093/oxfordjournals.jpart.a024407 Keiser, L. R. (2010). Understanding street-level bureaucrats’ decision making: Determining eligibility in the Social Security Disability program. Public Administration Review, March/April, 247–257. Keiser, L. R., & Soss, J. (1998). With good cause: Bureaucratic discretion and the politics of child support enforcement. American Journal of Political Science, 42(4), 1133–1156. https://doi.org/10.2307/2991852 Keller-Margulis, M. A., Shapiro, E. S., & Hintze, J. M. (2008). Long-term diagnostic accuracy of curriculum-based measures in reading and mathematics. School Psychology Review, 37(3), 374–390. https://doi.org/10.1080/02796015.2008.12087884 177 Kelso, K., Whitworth, A., Parsons, R., & Leitão, S. (2020). Hidden reading difficulties: Identifying children who are poor comprehenders. Learning Disability Quarterly, 45(3). https://doi.org/10.1177/0731948720961766 Kennedy, J. C., Hansen, G., Pavlov, P., Hopgood, H.-Y., Price, A., Kelly, T., Zemke, A., Roberts, J., Jameson, R., Ackerman, D., McPhee, K., & Sawher, T. (2015). Third-grade reading workgroup report to Governor Rick Snyder. Third-Grade Reading Workgroup. Kenny, D. T., & Chekaluk, E. (1993). Early reading performance: A comparison of teacher- based and test-based assessments. Journal of Learning Disabilities, 26(4), 210–279. https://doi.org/10.1177/002221949302600403 Kentucky Department of Education. (n.d.). The educator’s guide to the read at home family guide. Retrieved January 31, 2024, from https://www.education.ky.gov/curriculum/conpro/engla/Documents/Read_At_Home_Pla n_Family_Guide.pdf Kettler, R. J., & Albers, C. A. (2013). Predictive validity of curriculum-based measurement and teacher ratings of academic achievement. Journal of School Psychology, 51(4), 499–515. https://doi.org/10.1016/j.jsp.2013.02.004 Kilgus, S. P., Methe, S. A., Maggin, D. M., & Tomasula, J. L. (2014). Curriculum-based measurement of oral reading (R-CBM): A diagnostic test accuracy meta-analysis of evidence supporting use in universal screening. Journal of School Psychology, 52(4), 377–405. https://doi.org/10.1016/j.jsp.2014.06.002 Kingdon, J. W. (1984). Agendas, alternatives, and public policies. Little, Brown, and Company. Klingbeil, D. A., McComas, J. J., Burns, M. K., & Helman, L. (2015). Comparison of predictive validity and diagnostic accuracy of screening measures of reading skills. Psychology in the Schools, 52(5), 500–514. https://doi.org/10.1002/pits.21839 Klingbeil, D. A., Nelson, P. M., Van Norman, E. R., & Birr, C. (2017). Diagnostic accuracy of multivariate universal screening procedures for reading in upper elementary grades. Remedial and Special Education, 38(5), 308–320. https://doi.org/10.1177/0741932517697446 Kraft, M. A., & Dougherty, S. M. (2013). The effect of teacher-family communication on student achievement: Evidence from a randomized field experiment. Journal of Research on Educational Effectiveness, 6(3), 199–222. https://doi.org/10.1080/19345747.2012.743636 Kraft, M. A., & Gilmour, A. (2017). Revisiting The Widget Effect: Teacher evaluation reform and the distribution of teacher effectiveness. Educational Researcher, 46(5), 234–249. Kyzar, K., & Jimerson, J. B. (2018). Bridging the school-home divide in the middle grades: A process for strengthening school-family partnerships. Middle School Journal, 49(1), 13– 23. https://doi.org/10.1080/00940771.2018.1399331 178 Lake, R. J. (2008). Hopes, fears, & reality: A balanced look at American charter schools in 2008. Center on Reinventing Public Education. https://files.eric.ed.gov/fulltext/ED528668.pdf Lapadat, J. C. (1998). Implicit theories and stigmatizing labels. Journal of College Reading and Learning, 29(1), 73–81. https://doi.org/10.1080/10790195.1998.10850071 Lareau, A. (1989). Family-school relationships: A view from the classroom. Educational Policy, 3(3), 245–259. https://doi.org/10.1177/0895904889003003004 LaRocque, M., Kleiman, I., & Darling, S. M. (2011). Parental involvement: The missing link in school achievement. Preventing School Failure: Alternative Education for Children and Youth, 55(3), 115–122. https://doi.org/10.1080/10459880903472876 Lavelley, M. (2018). Out of the loop. National School Boards Association Center for Public Education. https://cdn-files.nsba.org/s3fs-public/10901- 5071_CPE_Rural_School_Report_Web_FINAL.pdf Lechuga-Peña, S., & Brisson, D. (2018). Barriers to school-based parent involvement while living in public housing: A mother’s perspective. The Qualitative Report, 23(5), 1176– 1187. Lee, S. W., & Min, S. (2017). Riding the implementation curve: Teacher buy-in and student academic growth under comprehensive school reform programs. The Elementary School Journal, 117(3). https://doi.org/10.1086/690220 Lesnick, J., Goerge, R. M., Smithgall, C., & Gwynne, J. (2010). Reading on grade level in third grade: How is it related to high school performance and college enrollment? A longitudinal analysis of third-grade students in Chicago in 1996-97 and their educational outcomes. Chapin Hall at the University of Chicago. https://www.chapinhall.org/wp- content/uploads/Reading_on_Grade_Level_111710.pdf Levin, H. M., McEwan, P. J., Belfield, C., Bowden, A. B., & Shand, R. (2018). Economic evaluation in education: Cost-effectiveness and benefit-cost analysis (3rd ed.). Sage. Lexia. (2023). 2023 discontinuation of RAPID. https://help.lexialearning.com/s/article/2023- Discontinuation-of-RAPID-27614 Lieberman, J. M. (2002). Three streams and four policy entrepreneurs converge: A policy window opens. Education and Urban Society, 34(4), 438–450. Lin, A. C. (2000). Reform in the making: The implementation of social policy in prison. Princeton University Press. Lin, Q. (2003). Parent involvement and early literacy. Harvard Family Research Project. https://archive.globalfrp.org/publications-resources/publications-series/family- involvement-research-digests/parent-involvement-and-early-literacy 179 Linder, S. H., & Peters, B. G. (1989). Instruments of government: Perceptions and context. Journal of Public Policy, 9(1), 35–58. https://doi.org/10.1017/S0143814X00007960 Lipsky, M. (1980). Street-level bureaucracy: Dilemmas of the individual in public services. Russell Sage Foundation. Loera, G., Rueda, R., & Nakamoto, J. (2011). The association between parental involvement in reading and schooling and children’s reading engagement in Latino families. Literacy Research and Instruction, 50(2), 133–155. https://doi.org/10.1080/19388071003731554 Logan, S., & Johnston, R. (2010). Investigating gender differences in reading. Educational Review, 62(2), 175–187. https://doi.org/10.1080/00131911003637006 Lorence, J. (2014). Third-grade retention and reading achievement in Texas: A nine year panel study. Social Science Research, 48, 1–19. Lorence, J., Dworkin, A. G., Toenjes, L. A., & Hill, A. N. (2002). Grade retention and social promotion in Texas, 1994-99: Academic achievement among elementary school students. Brookings Papers on Education Policy, 13–67. Ma, X., Shen, J., Krenn, H. Y., Hu, S., & Yuan, J. (2016). A meta-analysis of the relationship between learning outcomes and parental involvement during early childhood education and early elementary education. Educational Psychology Review, 28, 771–801. https://doi.org/10.1007/s10648-015-9351-1 Mapp, K. L., & Kuttner, P. J. (2013). Partners in education: A dual capacity-building framework for family-school partnerships (p. 32). SEDL, U.S. Department of Education. https://files.eric.ed.gov/fulltext/ED593896.pdf Mariano, L. T., & Martorell, P. (2013). The academic effects of summer instruction and retention in New York City. Educational Evaluation and Policy Analysis, 35(1), 96–117. Markow, D., & Pieters, A. (2009). The MetLife survey of the American teacher: Collaborating for student success [dataset]. MetLife. Marsh, D., & Sharman, J. C. (2009). Policy diffusion and policy transfer. Policy Studies, 30(3), 269–288. Marsh, J. A., McCombs, Jennifer S., Lockwood, J. R., Martorell, F., Gershwin, D., Naftel, S., Le, V.-N., Shea, M., Barney, H., & Crego, A. (2008). Supporting literacy across the Sunshine State: A study of Florida middle school reading coaches. RAND Corporation. Marsh, J. A., & Wohlstetter, P. (2013). Recent trends in intergovernmental relations: The resurgence of local actors in education policy. Educational Researcher, 42, 276–283. Mart, A., Dusenbury, L., & Weissberg, R. P. (2011). Social, emotional, and academic learning; Complementary goals for school-family partnerships. In Handbook on family and community engagement. IAP. 180 Matland, R. E. (1995). Synthesizing the implementation literature: The ambiguity-conflict model of policy implementation. Journal of Public Administration Research and Theory, 5(2), 145–174. May, P. J., & Winter, S. C. (2009). Politicians, managers, and street-Level bureaucrats: Influences on policy implementation. Journal of Public Administration Research and Theory, 19(3), 453–476. https://doi.org/10.1093/jopart/mum030 Maynard-Moody, S., & Musheno, M. (2003). Cops, teachers, counselors: Narratives of street- level judgment. University of Michigan Press. McBride, J. R., Ysseldyke, J., Milone, M., & Stickney, E. (2010). Technical adequacy and cost benefit of four measures of early literacy. Canadian Journal of School Psychology, 25(2), 189–204. https://doi.org/10.1177/0829573510363796 McDonnell, L. M., & Elmore, R. F. (1987). Getting the job done: Alternative policy instruments. Educational Evaluation and Policy Analysis, 9(2), 133–152. McElvany, N., & Van Steensel, R. (2009). Potentials and challenges of family literacy interventions: The question of implementation quality. European Educational Research Journal, 8(3), 418–433. https://doi.org/10.2304/eerj.2009.8.3.418 McLaughlin, M. W. (1987). Learning from experience: Lessons from policy implementation. Educational Evaluation and Policy Analysis, 9(2), 171–178. https://doi.org/10.3102/01623737009002171 MDE. (2022). Additional instructional time grant. Michigan Department of Education. https://www.michigan.gov/mde/0,4615,7-140-28753_74161-367703--,00.html MDE. (2023). Benchmark assessments. Michigan Department of Education. https://www.michigan.gov/mde/services/student-assessment/benchmark-assessments MDE. (2024). Early literacy assessment resources. https://www.michigan.gov/mde/services/academic-standards/literacy/early-literacy- assessment-resources MDE. (n.d.). Facts for families: What is the Read by Grade Three Law? https://www.michigan.gov/- /media/Project/Websites/mde/gsrp/parents/grade_3_reading_facts_families_final_tj.pdf?r ev=b8e0f097288f48ff8aff62734d5a7df4 Mesmer, H. A. E. (2006). Beginning reading materials: A national survey of primary teachers’ reported uses and beliefs. Journal of Literacy Research, 38(4), 389–425. Meyers, M. K., & Nielsen, V. L. (2012). Street-level bureaucrats and the implementation of public policy. In The SAGE Handbook of Public Administration. SAGE Publications Ltd. Michigan Achieves. (n.d.). Michigan’s third grade reading law. Retrieved July 27, 2023, from 181 https://michiganachieves.com/wp-content/uploads/sites/2/2019/10/Michigans-Third- Grade-Reading-Law.pdf Michigan Alliance for Families. (n.d.). Third grade reading. Michigan Alliance for Families. Retrieved July 27, 2023, from https://www.michiganallianceforfamilies.org/third-grade- reading/ Michigan Assessment Consortium. (2020). Early literacy assessment systems that support learning. https://www.michiganassessmentconsortium.org/wp- content/uploads/ELAS_Guide_2020_v5.pdf Michigan Association of Intermediate School Administrators General Education Leadership Network Early Literacy Task Force. (2016). Essential instructional practices in early literacy: Grades K to 3 (p. 6). Michigan Association of Intermediate School Administrators General Education Leadership Network Early Literacy Task Force. https://www.gomaisa.org/downloads/gelndocs/k-3_literacy_essentials.pdf Michigan Association of Public School Academies. (2023). Michigan’s charter schools. MI Charters. https://www.charterschools.org/about-michigan-charter- schools#:~:text=Charter%20schools%20are%20free%2C%20public,to%20meet%20their %20students'%20needs. Michigan Public Act 306, 98th Legislative Session MCL § 380.1280f (2016). https://www.legislature.mi.gov/documents/2015-2016/publicact/pdf/2016-PA-0306.pdf Miedel, W. T., & Reynolds, A. J. (1999). Parent involvement in early intervention for disadvantaged children: Does it matter? Journal of School Psychology, 37(4), 379–402. https://doi.org/10.1016/S0022-4405(99)00023-0 Milan Area Schools. (2017). Read at home plan for student success. https://www.milanareaschools.org/cms/lib/MI02208047/Centricity/Domain/61/MAS%20 Read-At-Home%20Plan%202017%202.pdf Mintrom, M., & Luetjens, J. (2017). Policy entrepreneurs and problem framing: The case of climate change. 35(8), 1362–1377. https://doi.org/10.1177/2399654417708440 Mintrom, M., & Vergari, S. (1998). Policy networks and innovation diffusion: The case of state education reforms. The Journal of Politics, 60(1), 126–148. https://doi.org/10.2307/2648004 Mintrop, H. (2004). High-stakes accountability, state oversight, and educational equity. Teachers College Record, 106(11), 2057–2245. https://doi.org/10.1111/j.1467-9620.2004.00430.x Mississippi Department of Education. (n.d.). Parents’ read-at-home plan for student success. Retrieved January 31, 2024, from https://www.mdek12.org/sites/default/files/Offices/MDE/OAE/OEER/Literacy/parent- read-at-home-plan_mde_lbpa%202018.pdf 182 Mitra, D. L. (2018). Educational change and the political process. Routledge. Mo, Y., & Singh, K. (2008). Parents’ relationships and involvement: Effects on students’ school engagement and performance. Research in Middle Level Education, 31(10), 1–11. https://doi.org/10.1080/19404476.2008.11462053 Moreland-Russell, S., Eyler, A. A., Barbero, C., Hipp, A., & Walsh, H. A. (2013). Diffusion of Complete Streets policies across US communities. Journal of Public Health Management and Practice, 19(3), S89–S96. NAEP. (2022a). National achievement-level results. The Nation’s Report Card. https://www.nationsreportcard.gov/reading/nation/achievement/?grade=4 NAEP. (2022b). Scores decline in NAEP reading at grades 4 and 8 compared to 2019. The Nation’s Report Card. https://www.nationsreportcard.gov/highlights/reading/2022/ National Center on Intensive Intervention. (2021). Academic screening tools chart. National Center on Intensive Intervention at the American Institutes for Research. https://charts.intensiveintervention.org/ascreening National Early Literacy Panel. (2008). Developing early literacy: Report of the National Early Literacy Panel. National Institute for Literacy. National Institute of Child Health and Human Development. (2019, June 10). National Reading Panel (historical/for reference only). Https://Www.Nichd.Nih.Gov/. https://www.nichd.nih.gov/research/supported/nrp Nilsson, N. L. (2011). A critical analysis of eight informal reading inventories. The Reading Teacher, 61(7), 526–536. https://doi.org/10.1598/RT.61.7.2 NWEA. (n.d.). MAP Growth. NWEA. Retrieved February 19, 2024, from https://www.nwea.org/map-growth/ Ochs, S., Keller-Margulis, M. A., Santi, K. L., & Jones, J. H. (2018). Long-term validity and diagnostic accuracy of a reading computer-adaptive test. Assessment for Effective Intervention, 45(3), 163–240. https://doi.org/10.1177/1534508418796271 Parecki, A. D., & Gear, S. B. (2013). Parent participation in family programs: Involvement in literacy interactions, adult and child instruction, and assessment. Dialog, 16(1), 236–252. Parker, D. C., Zaslofsky, A. F., Burns, M. K., Kanive, R., Hodgson, J., Scholin, S. E., & Klingbeil, D. A. (2015). A brief report on the diagnostic accuracy of oral reading fluency and reading inventory levels for reading failure risk among second- and third-grade students. Reading & Writing Quarterly, 31(1), 56–67. Patrikakou, E. N., & Weissberg, R. P. (2007). School-family partnerships to enhance children’s social, emotional and academic learning. In Educating people to be emotionally intelligent. Bloomsbury Publishing USA. 183 Patton, S. (2019). Parents’ and teachers’ perceptions of parental involvement [Walden University]. https://www.proquest.com/docview/2307397163?pq- origsite=gscholar&fromopenview=true&sourcetype=Dissertations%20&%20Theses Pearson, P. D., Palincsar, A. S., Biancarosa, G., & Berman, A. I. (2020). Reaping the Rewards of the Reading for Understanding Initiative. National Academy of Education. https://naeducation.org/wp-content/uploads/2020/06/Reaping-the-Rewards-of-the- Reading-for-Understanding-Initiative.pdf Pemberton, K. D., & Miller, S. (2013). Building home-school relationships to enhance reading achievement for students from families with limited financial resources. Education and Urban Society, 47(7), 743–848. https://doi.org/10.1177/0013124513508979 Pesso, T. (1978). Local welfare Offices: Managing the intake process. Public Policy, 26(2), 305– 330. Petersen-Brown, S., Hawley, E. K., Fischer, E., & Schneider, M. (2023). A review and meta- analysis of literacy interventions implemented by caregivers at home. Preventing School Failure: Alternative Education for Children and Youth. https://doi.org/10.1080/1045988X.2022.2158161 Petridou, E., & Mintrom, M. (2020). A research agenda for the study of policy entrepreneurs. Policy Studies Journal. https://doi.org/10.1111/psj.12405 Preston, C., Goldring, E., Berends, M., & Cannata, M. (2012). School innovation in district context: Comparing traditional public schools and charter schools. Economics of Education Review, 31(2), 318–330. Prins, E., & Toso, B. W. (2008). Defining and measuring parenting for educational success: A critical discourse analysis of the Parent Education Profile. American Educational Research Journal, 45, 555–596. https://doi.org/10.3102/0002831208316205 Quinn, D. M. (2020). Experimental evidence on teachers’ racial bias in student evaluation: The role of grading scales. Educational Evaluation and Policy Analysis, 42(3), 375–392. https://doi.org/10.3102/0162373720932188 Raikes, H., Pan, B. A., Luze, G. J., Tamis-LeMonda, C. S., Brooks-Gunn, J., Constantine, J., Tarullo, L., Raikes, H. A., & Rodriguez, E. T. (2006). Mother–child bookreading in low‐ income families: Correlates and outcomes during the first three years of life. Child Development, 77(4), 924–953. https://doi.org/10.1111/j.1467-8624.2006.00911.x Reading Excellence Act, H.R. 2614, 105th Congress (1998). https://www.congress.gov/congressional- report/105th-congress/senate-report/208/1 Reardon, S. F. (2016). School segregation and racial academic achievement gaps. RSF: The Russell Sage Foundation Journal of the Social Sciences, 2(5), 34–57. https://doi.org/10.7758/RSF.2016.2.5.03 184 Renzulli, L. A., & Roscigno, V. J. (2005). Charter school policy, implementation, and diffusion across the United States. Sociology of Education, 78(4), 344–365. Reutzel, D. R., Fawson, P. C., & Smith, J. A. (2006). Words to Go!: Evaluating a first-grade parent involvement program for “making” words at home. Reading Research and Instruction, 45(2), 119–159. Rist, R. C. (2017). On understanding the processes of schooling: The contributions of labeling theory. In Exploring Education (5th Edition). Routledge. RMC Research Corporation. (2019). Michigan’s Read by Grade Three Law: An inside look. RMC Research Corporation. Roderick, M., Jacob, B. A., & Bryk, A. S. (2002). The impact of high-stakes testing in Chicago on student achievement in promotional gate grades. Educational Evaluation and Policy Analysis, 42(4), 333–357. Roderick, M., & Nagaoka, J. (2005). Retention under Chicago’s high-stakes testing program: Helpful, harmful, or harmless? Educational Evaluation and Policy Analysis, 27(4), 309– 340. Roscigno, V. J., Tomaskovic-Devey, D., & Crowley, M. (2006). Education and the inequalities of place. Social Forces, 84(4), 2121–2145. https://doi.org/10.1353/sof.2006.0108 Rose, S., & Schimke, K. (2012). Third grade literacy policies: Identification, intervention, retention. Education Commission of the States. Russell, C. (2013). Four decades of informal reading inventories. Review of Higher Education & Self Learning, 6(22), 1–21. Sabatier, P., & Mazmanian, D. (1980). The implementation of public policy: A framework of analysis. Policy Studies Journal, 8(4), 538–560. https://doi.org/10.1111/j.1541- 0072.1980.tb01266.x Salamon, L. M. (1981). Rethinking public management: Third-party government and the changing forms of government action. Public Policy, 29(3), 255–275. Salamon, L. M. (1989). Beyond privatization: The tools of government action. The Urban Institute. Schmitterer, A. M. A., & Brod, G. (2021). Which data do elementary school teachers use to determine reading difficulties in their students? Journal of Learning Disabilities, 54(5). https://doi.org/10.1177/0022219420981990 Schneider, A., & Ingram, H. (1990). Behavioral assumptions of policy tools. The Journal of Politics, 52(2). Schneider, A., & Ingram, H. (1993). Social construction of target populations: Implications for 185 politics and policy. American Political Science Review, 87(2), 334–347. Schwerdt, G., West, M. R., & Winters, M. A. (2017). The effects of test-based retention on student outcomes over time: Regression discontinuity evidence from Florida. Journal of Public Economics, 152, 154–169. Senate Bill 12, 0012, 102nd Legislative Session, SB-0012 (2023). http://www.legislature.mi.gov/documents/2023-2024/billintroduced/Senate/pdf/2023- SIB-0012.pdf Sénéchal, M. (2006). Testing the home literacy model: Parent involvement in kindergarten is differentially related to grade 4 reading comprehension, fluency, spelling, and reading for pleasure. Scientific Studies of Reading, 10(1), 59–87. https://doi.org/10.1207/s1532799xssr1001_4 Sénéchal, M., & Young, L. (2008). The effect of family literacy interventions on children’s acquisition of reading from kindergarten to grade 3: A meta-analytic review. Review of Educational Research, 78(4), 880–907. Shapiro, E. S., & Gebhardt, S. N. (2012). Comparing computer-adaptive and curriculum-based measurement methods of assessment. School Psychology Review, 41(3), 295–305. Sheridan, S. M., Knoche, L. L., Kupzyk, K. A., Edwards, C. P., & Marvin, C. A. (2011). A randomized trial examining the effects of parent engagement on early language and literacy: The Getting Ready intervention. Journal of School Psychology, 49(3), 361–383. https://doi.org/10.1016/j.jsp.2011.03.001 Sheridan, S. M., Koziol, N., Witte, A. L., Iruka, I., & Knoche, L. L. (2020). Longitudinal and geographic trends in family engagement during the pre-kindergarten to kindergarten transition. Early Childhood Education Journal, 48, 365–377. Shifrer, D. (2016). Stigma and stratification limiting the math course progression of adolescents labeled with a learning disability. Learning and Instruction, 42, 47–57. https://doi.org/10.1016/j. learninstruc.2015.12.001 Shipan, C. R., & Volden, C. (2008). The mechanisms of policy diffusion. American Journal of Political Sciences, 52(4), 840–857. Shipan, C. R., & Volden, C. (2012). Policy diffusion: Seven lessons for scholars and practitioners. Public Administration Review, 72(6), 788–796. Shpaizman, I., Swed, O., & Pedahzur, A. (2016). Policy change inch by inch: Policy entrepreneurs in the Holy Basin of Jerusalem. Public Administration, 94(4), 1042–1058. Simons-Morton, B., & Chen, R. (2009). Peer and parent influences on school engagement among early adolescents. Youth & Society, 41(1), 3–25. https://doi.org/10.1177/0044118X09334861 186 Slungaard Mumma, K., & Winters, M. A. (2023). The effect of retention under Mississippi’s test- based promotion policy (Working Paper 2023–1). Wheelock Educational Policy Center. https://wheelockpolicycenter.org/high-quality-education/ms-read-by-grade-three/ Smit, B. (2005). Teachers, local knowledge, and policy implementation: A qualitative policy- practice inquiry. Education and Urban Society, 37(3), 292–306. https://doi.org/10.1177/0013124505275426 Smith, J., Wohlstetter, P., Kuzin, C. A., & De Pedro, K. (2011). Parent involvement in urban charter schools: New strategies for increasing participation. School Community Journal, 21(1), 71–94. Snow, C. E., Burns, M. S., & Griffin, P. (1998). Preventing reading difficulties in young children. National Academy Press. Sparks, R. L., Patton, J., & Murdoch, A. (2014). Early reading success and its relationship to reading achievement and reading volume: Replication of “10 years later.” Read Writ, 27, 189–211. Spector, J. E. (2005). How reliable are informal reading inventories? Psychology in the Schools, 42(6), 593–603. https://doi.org/10.1002/pits.20104 Spillane, J. P. (1998). A cognitive perspective on the LEA’s role in implementing instructional policy: Accounting for local variability. Educational Administration Quarterly, 34, 31– 57. Spillane, J. P. (2000). Cognition and policy implementation: District policymakers and the reform of mathematics education. Cognition and Instruction, 18(2), 141–179. https://doi.org/10.1207/S1532690XCI1802_01 Spillane, J. P., & Anderson, L. (2019). Negotiating policy meanings in school administrative practice: Practice, professionalism, and high-stakes accountability in a shifting policy environment. In Innovations in Educational Change (pp. 121–145). Spillane, J. P., Diamond, J. B., Burch, P., Hallett, T., Jita, L., & Zoltners, J. (2002a). Managing in the middle: School leaders and the enactment of accountability policy. Educational Policy, 16(5), 731–762. https://doi.org/10.1177/089590402237311 Spillane, J. P., Reiser, B. J., & Reimer, T. (2002b). Policy implementation and cognition: Reframing and refocusing implementation research. Review of Educational Research, 72(3), 387–431. https://doi.org/10.3102/00346543072003387 St. Pierre, R. G., Ricciuti, A., & Rimdzius, T. (2005). Effects of a family literacy program on low-literate children and their parents: Findings from an evaluation of the Even Start family literacy program. Developmental Psychology, 41(6), 953–970. https://doi.org/10.1037/0012-1649.41.6.953 Stevenson, D. L., & Baker, D. P. (1987). The family-school relation and the child’s school 187 performance. Child Development, 58(5), 1348–1357. https://doi.org/10.2307/1130626 Stone, D. (2000). Non-governmental policy transfer: The strategies of independent policy institutes. Governance: An International Journal of Policy and Administration, 13(1), 45–62. Strunk, K. O., Wright, T. S., Kilbride, T., Zhu, Q., Cummings, A., West, J., Turner, M., & De Voto, C. (2021). Michigan’s Read by Grade Three Law: Year one report. Education Policy Innovation Collaborative. https://epicedpolicy.org/wp- content/uploads/2021/03/Year_One_RBG3_Report.pdf Strunk, K. O., Wright, T. S., Westall, J., Zhu, Q., Kilbride, T., Cummings, A., Utter, A., & Mavrogordato, M. (2022). Michigan’s Read by Grade Three Law: Year two report. Education Policy Innovation Collaborative. https://epicedpolicy.org/wp- content/uploads/2022/02/RBG3_Rpt_Yr2_Feb2022.pdf The Nation’s Report Card. (n.d.). NAEP data explorer. Retrieved January 26, 2022, from https://www.nationsreportcard.gov/ndecore/xplore/NDE Thomas, A. S., & January, S.-A. A. (2019). Evaluating the criterion validity and classification accuracy of universal screening measures in reading. Assessment for Effective Intervention, 46(2). Torgesen, J. K. (2004). Lessons learned from research on interventions for students who have difficulty learning to read. In The voice of evidence in reading research (pp. 355–382). Paul H. Brookes Publishing Co. Tortorelli, L. S. (2019). Reading rate in informational text: Norms and implications for theory and practice in the primary grades. Reading Psychology, 40(3), 293–324. Truckenmiller, A. J., Cho, E., Bourgeois, S., & Friedman, E. (2024). Uses and misuses of commercial reading assessment: An applied framework for decision making in grades K through 6. The Reading Teacher, 77(5), 609–623. https://doi.org/10.1002/trtr.2274 Turnbull, B. (2002). Teacher participation and buy-in: Implications for school reform initiatives. Learning Environments Research, 5, 235–252. U.S. Department of Education. (2015, December 7). Reading First [Program Home Page]. U.S. Department of Education. https://www2.ed.gov/programs/readingfirst/index.html Utter, A., Cortez Hendricks, L., Westall, J., Wright, T. S., & Strunk, K. O. (2023). Understanding opportunities for K-3 teachers’ professional development in literacy [EPIC Research Report]. Education Policy Innovation Collaborative. https://epicedpolicy.org/understanding-opportunities-for-k-3-teachers-professional- development-in-literacy/ Van Otterloo, S. G., Van der Leji, & Veldkamp, E. (2006). Treatment integrity in a home-based pre-reading intervention programme. Dyslexia, 12, 155–176. 188 https://doi.org/10.1002/dys.311 van Steensel, R., McElvany, N., Kurvers, J., & Herppich, S. (2011). How effective are family literacy programs? Results of a meta-analysis. Review of Educational Research, 81(1), 69–96. VanDerHeyden, A. M., Burns, M. K., & Bonifay, W. (2018). Is more screening better? The relationship between frequent screening, accurate decisions, and reading proficiency. School Psychology Review, 47(1), 62–82. https://doi.org/10.17105/SPR-2017-0017.V47- 1 Virinkoski, R., Lerkkanen, M.-K., Holopainen, L., Eklund, K., & Aro, M. (2018). Teachers’ ability to identify children at early risk for reading difficulties in grade 1. Early Childhood Education Journal, 46, 497–509. Voorhis, F. L., Maier, Epstein, J. L., & Lloyd, C. M. (2013). The impact of family involvement on the education of children ages 3 to 8: A focus on literacy and math achievement outcomes and social-emotional skills. MDRC. https://files.eric.ed.gov/fulltext/ED545474.pdf Wagner, M., & Clayton, S. (1999). The parents as teachers program: Results from two demonstrations. The Future of Children, 9(1), 91–115. Watson, T. J. (1994). In search of management—Culture, chaos and control in managerial work. Routledge. Wayman, M., Wallace, T., Wiley, H. I., Tichá, R., & Espin, C. A. (2007). Literature synthesis on curriculum-based measurement in reading. The Journal of Special Education, 41(2), 85– 120. Weick, K. E. (1995). Sensemaking in organizations. SAGE Publications, Inc. Weigel, D. J., Martin, S. S., Bennett, K. K., & Kymberley K. Bennett. (2006). Contributions of the home literacy environment to preschool‐aged children’s emerging literacy and language skills. Early Child Development and Care, 176, 357–378. https://doi.org/10.1080/03004430500063747 Weiss, S., Stallings, D. T., & Porter, S. (2018). Is Read to Achieve making the grade? An assessment of North Carolina’s elementary reading proficiency initiative. North Carolina State University College of Education. Wepner, S. B., & Gomez, D. W. (2017). Challenges facing suburban schools: Promising responses to changing student populations. Rowman & Littlefield. Westall, J., & Cummings, A. (2023). The effects of early literacy policies on student achievement. https://www.johnwwestall.com/_files/ugd/788e40_44567ef0956145649cff24c9b61efa6d. pdf 189 Weyer, M. (2019). Third-grade reading legislation. National Conference of State Legislatures. http://www.ncsl.org/research/education/third-grade-readinglegislation.aspx Whitmer, G. (2020, April 2). Executive Order 2020-35: Provision of K-12 education during the remainder of the 2019-20 school year. Michigan.Gov. https://www.michigan.gov/whitmer/news/state-orders-and- directives/2020/04/02/executive-order-2020-35 Wilder, S. (2014). Effects of parental involvement on academic achievement: A meta-synthesis. Educational Review, 66(3), 377–397. https://doi.org/10.1080/00131911.2013.780009 Winter, S. C. (2001). Reconsidering street-Level bureaucracy theory: From identifying to explaining coping behavior. Association of Policy Analysis and Management, Washington, D.C. Winters, M. A., & Greene, J. P. (2012). The medium-run effects of Florida’s test-based promotion policy. Education Finance and Policy, 7(3), 305–330. Wong, L. S. (2019). Administrators’ unintentional sensegiving and system reform outcomes. Education Policy Analysis Archives, 27(3). https://doi.org/10.14507/epaa.27.3854 Wong, K. E., & Langevin, W. E. (2007). Policy expansion of school choice in the American states. Peabody Journal of Education, 82(2/3), 440–472. Workman, E. (2014). Third-grade reading policies. Education Commission of the States. Woulfin, S. L., Donaldson, M. L., & Gonzales, R. (2016). District leaders’ framing of educator evaluation policy. Educational Administration Quarterly, 52(1), 110–143. https://doi.org/10.1177/0013161X15616661 Yanow, D. (1996). How does a policy mean? Georgetown University Press. Young, T. V., Shepley, T. V., & Song, M. (2010). Understanding agenda setting in state educational policy: An application of Kingdon’s multiple streams model to the formation of state reading policy. Education Policy Analysis Archives, 18, 15. Zellman, G. L., & Waterman, J. M. (1998). Understanding the impact of parent school involvement on children’s educational outcomes. The Journal of Education Research, 91(6), 370–380. https://doi.org/10.1080/00220679809597566 190 APPENDIX 1.A: INTERVIEW CODES • Policy Transfer o National Literacy Policy ▪ ALEC ▪ Common Core ▪ Education Reform Community ▪ ExcelinEd ▪ National Reading Panel ▪ Reading First o Other States’ Literacy Policies ▪ Florida ▪ Kentucky ▪ Massachusetts ▪ Minnesota ▪ Tennessee • Michigan Context o Economic Factors o Educational Factors ▪ ACLU Lawsuit ▪ Poor Literacy Performance • Urgency ▪ Prior Lack of Success with Early Literacy Efforts o Human Capital ▪ Workforce o Influential Groups ▪ GLEP ▪ Lobbyists ▪ Unions o Local Control o Political Factors ▪ Elected School Board ▪ Lack of Collaboration ▪ Lack of Involvement of Researchers ▪ Lack of Stick-With-It-Ness ▪ Party Politics ▪ Term Limits • Preexisting Support for Literacy o Coaching o Early Literacy Task Force o Funding/Grants o Gov. Snyder’s Third-Grade Reading Workgroup o Literacy Essentials o PreK-12 Literacy Commission o Previous Third-Grade Reading Bills • Relationships between Groups 191 o Collaborative o Tenuous • Research o When Research Was Used o When Research Wasn’t Used • Sequence of Events 192 APPENDIX 1.B: POLICY DOCUMENT CODES • Assessments o Diagnostic Assessments o Progress Monitoring Assessments • Funding • Instruction/Tier I o Evidence-Based Reading Instruction ▪ Big 5 o School/District Literacy Leadership Teams o School/District Reading Plans • Intervention/Tiers II and III o Additional Instructional Time in Literacy o Additional Interventions for Retained Students o Additional Opportunities for Error Correction and Feedback o Additional Opportunities for Guided Practice o Alternative/Transitional Instructional Setting o Assignment to Highly Effective Teacher o Evidence-Based Literacy Interventions o Frequent Progress Monitoring o Home Reading Program ▪ Resources for Families o Individual Reading Improvement Plans o Online or Computer-Based Instruction o Out-of-School-Time Interventions o Reading Specialist o Small Group/1:1 Interventions o Smaller Classes o Summer School • Parental Notification of Reading Deficiency • Professional Development o Literacy Coaches o Professional Development in Literacy o Teacher Certification Requirements • Retention o Allows Retention o Alternative Ways to Demonstrate Proficiency ▪ Alternative Assessment ▪ Course Grades ▪ Portfolio ▪ Proficiency in Other Subjects ▪ Summer School Attendance o Exemptions ▪ Administrator Appeal ▪ English Learners ▪ New to School/District 193 ▪ Parent Appeal ▪ Previously Retained ▪ Students with Disabilities ▪ Teacher Appeal o Requires Retention • State Literacy Organization 194 APPENDIX 2.A: LIST OF DISTRICT “READ AT HOME” PLANS REVIEWED 1. Hartland Consolidated Schools (Howell, MI) 2. Imlay City Schools (Imlay City, MI) 3. Forest Hills Public Schools (Grand Rapids, MI) 4. Dearborn Public Schools (Dearborn, MI) 5. Huron Intermediate School District (Bad Axe, MI) 6. Blissfield Community Schools (Blissfield, MI) 7. Howell Public Schools (Howell, MI) 8. Whiteford Agricultural School District (Ottawa Lake, MI) 9. Warren Consolidated Schools (Warren, MI) 10. Madison School District (Adrian, MI) 11. Anchor Bay School District (Casco, MI) 12. Allendale Public Schools (Allendale, MI) 13. Clarkston Community Schools (Clarkston, MI) 14. Berkley Schools (Oak Park, MI) 15. Kentwood Public Schools (Kentwood, MI) 16. Mason Public Schools (Mason, MI) 17. Eastpointe Community Schools (Eastpointe, MI) 18. River Rouge School District (River Rouge, MI) 19. Coloma Community School District (Coloma, MI) 20. Eaton Regional Education Service Agency (Charlotte, MI) 21. Eau Claire Public Schools (Eau Claire, MI) 22. Central Montcalm Public Schools (Sheridan, MI) 23. Muskegon Public Schools (Muskegon, MI) 24. Cedar Springs Public Schools (Cedar Springs, MI) 25. Gull Lake Community Schools (Richland, MI) 26. Parchment School District (Kalamazoo, MI) 27. Caledonia Community Schools (Caledonia, MI) 28. St. Clair County Regional Education Service Agency (Marysville, MI) 29. Zeeland Public Schools (Zeeland, MI) 30. Milan Area Schools (Milan, MI) 195 APPENDIX TABLES Table A2.1 Administrative Data Sample Restrictions 2020- 21 2021- 22 2022- 23 Total Student- 2019- 20 Year Obs. 437,356 415,849 421,519 423,685 1,698,409 427,954 402,689 407,697 401,078 1,639,418 396,854 376,272 380,405 382,685 1,536,216 K-3 Students 35a(5) Districts 35a(5) in All Years “Reading Deficient” 109,854 115,517 122,863 127,668 475,902 114,906 122,192 126,590 363,688 TSDL Available Note. The “K-3 Students” row contains the total population of K-3 students enrolled in TPS and charter schools in Michigan. The “35a(5) Districts” row restricts the sample to students in districts that received these funds from the State School Aid Act, and the “35a(5) in All Years” row restricts it to students whose districts received these funds in all four years of the study. The “Reading Deficient” row restricts the sample to students identified with “reading deficiencies.” Total District-Year Obs. 779 754 613 611 609 N/A 196 Table A2.2 Survey Sample Descriptives 2019-20 Female Hired Last 5 Years Black Hispanic Asian Other Race White ELA Endorsement Urban Rural Suburb Charter 2020-21 Female Hired Last 5 Years Black Hispanic Asian Other Race White ELA Endorsement Urban Sample Teachers Pop. Diff. Sample Principals Pop. 95.2% 40.4% 6.6% 1.3% 0.7% 1.4% 90.2% 38.7% 24.4% 23.8% 51.9% 10.9% 93.9% 41.0% 4.7% 1.4% 0.9% 1.6% 91.6% 36.4% 21.6% 95.1% 33.6% 3.7% 1.3% 0.7% 1.2% 93.2% 40.1% 20.4% 22.1% 57.4% 10.7% 94.5% 35.5% 6.5% 1.4% 0.7% 1.3% 90.3% 38.8% 24.7% 60.0% 30.5% 13.7% 1.3% 0.4% 1.0% 83.6% 32.5% 26.4% 21.4% 52.2% 6.6% 61.5% 30.9% 14.5% 1.3% 0.4% 1.0% 82.8% 32.6% 25.9% 0.1% 6.8% 2.9% 0.0% 0.0% 0.2% -3.1% -1.4% 3.9% 1.6% -5.6% 0.2% -0.6% 5.5% -1.8% 0.1% 0.2% 0.3% 1.3% -2.3% -3.1% 68.4% 36.8% 11.1% 1.2% 0.3% 1.6% 86.4% 35.7% 23.0% 26.8% 50.2% 14.8% 69.3% 37.4% 13.4% 1.0% 0.0% 0.5% 85.1% 35.1% 22.8% 197 Diff. 8.4% 6.3% -2.6% -0.1% -0.2% 0.6% 2.8% 3.1% -3.4% 5.4% -2.0% 8.2% 7.8% 6.5% -1.1% -0.3% -0.4% -0.6% 2.3% 2.5% -3.1% Superintendents Pop. Sample Diff. 34.9% 39.8% 5.2% 1.6% 0.5% 1.0% 91.7% 29.5% 10.6% 41.3% 48.1% 18.8% 38.3% 42.0% 4.3% 2.5% 0.0% 1.9% 91.9% 22.4% 13.7% 26.9% 36.2% 6.6% 0.6% 0.2% 0.9% 91.8% 21.0% 12.0% 40.6% 47.4% 11.9% 30.2% 38.1% 6.8% 1.0% 0.3% 1.1% 91.1% 21.5% 12.0% 8.0% 3.5% -1.4% 1.0% 0.3% 0.1% -0.2% 8.4% -1.4% 0.7% 0.7% 6.8% 8.1% 3.8% -2.5% 1.5% -0.3% 0.7% 0.8% 0.9% 1.6% Table A2.2 (cont’d) Rural Suburb Charter 2021-22 Female Hired Last 5 Years Black Hispanic Asian Other Race White ELA Endorsement Urban Rural Suburb Charter 2022-23 Female Hired Last 5 Years Black Hispanic Asian Other Race White ELA Endorsement Urban Sample 26.5% 51.9% 11.7% Teachers Pop. 21.5% 53.8% 11.0% 94.0% 30.2% 4.0% 1.4% 0.8% 10.6% 92.3% 35.4% 20.8% 24.6% 54.6% 10.5% 94.0% 30.2% 4.0% 1.4% 0.8% 10.6% 92.3% 35.4% 20.8% 93.2% 23.4% 6.5% 1.4% 0.8% 1.3% 90.1% 38.0% 24.3% 21.9% 53.8% 11.1% 93.2% 23.4% 6.5% 1.4% 0.8% 1.3% 90.1% 38.0% 24.3% Diff. 5.0% -1.9% 0.7% 0.8% 6.8% -2.5% -0.1% 0.0% 9.3% 2.2% -2.6% -3.6% 2.7% 0.9% -0.6% 0.8% 6.8% -2.5% -0.1% 0.0% 9.3% 2.2% -2.6% -3.6% Sample 26.5% 50.7% 18.5% Principals Pop. 22.7% 51.4% 10.9% 61.8% 18.5% 14.4% 1.2% 0.8% 1.7% 82.5% 34.0% 25.9% 22.4% 51.7% 9.8% 61.8% 18.5% 14.4% 1.2% 0.8% 1.7% 82.5% 34.0% 25.9% 71.0% 23.7% 12.0% 0.3% 0.3% 4.2% 86.3% 37.6% 24.1% 26.5% 49.3% 15.2% 71.0% 23.7% 12.0% 0.3% 0.3% 4.2% 86.3% 37.6% 24.1% 198 Diff. 3.7% -0.6% 7.6% 9.2% 5.2% -2.4% -0.9% -0.5% 2.5% 3.8% 3.6% -1.8% 4.2% -2.4% 5.3% 9.2% 5.2% -2.4% -0.9% -0.5% 2.5% 3.8% 3.6% -1.8% Superintendents Pop. 41.2% 46.8% 15.0% Sample 44.7% 41.6% 20.4% Diff. 3.5% -5.1% 5.3% 42.2% 18.1% 3.6% 1.2% 0.0% 8.0% 91.6% 27.7% 13.4% 43.9% 42.7% 14.9% 42.2% 18.1% 3.6% 1.2% 0.0% 8.0% 91.6% 27.7% 13.4% 30.1% 21.2% 7.2% 1.4% 0.6% 2.0% 89.8% 22.8% 14.4% 38.7% 46.9% 12.9% 30.1% 21.2% 7.2% 1.4% 0.6% 2.0% 89.8% 22.8% 14.4% 12.0% -3.1% -3.6% -0.2% -0.6% 6.1% 1.8% 4.9% -1.0% 5.2% -4.2% 2.1% 12.0% -3.1% -3.6% -0.2% -0.6% 6.1% 1.8% 4.9% -1.0% Table A2.2 (cont’d) Rural Suburb Charter Sample 24.6% 54.6% 10.5% Teachers Pop. 21.9% 53.8% 11.1% Diff. 2.7% 0.9% -0.6% Sample 26.5% 49.3% 15.2% Principals Pop. 22.4% 51.7% 9.8% Diff. 4.2% -2.4% 5.3% Superintendents Pop. 38.7% 46.9% 12.9% Sample 43.9% 42.7% 14.9% Diff. 5.2% -4.2% 2.1% 199 Table A2.3 Survey Response Rates 2019-20 Sample Pop. 7,110 Teachers Principals 745 Superintendents 192 16,401 43.4% 44.9% 1,659 36.4% 528 RR RR 2020-21 Pop. 19,633 29.6% 1,844 22.6% 25.6% 632 RR 2021-22 Pop. 20,057 26.9% 1,824 21.7% 15.1% 590 Sample 5,392 395 89 Sample 5,811 417 162 RR 2022-23 Pop. 18,981 26.0% 1,740 17.8% 17.2% 505 Sample 4,942 310 87 200 Table A2.4 List of Survey Items Used in RQ3 and RQ4 Analysis RQ Item Text 3 How well do you understand the following aspects of the Read by Grade Three Law? • This question asked about several aspects of the law, one of which was “the provision of ‘Read at Home’ plans (defined in the law as training workshops and regular home reading) to families of students who have been identified as having a ‘reading deficiency.’” Please indicate the extent to which you believe each of the following elements of the Read by Grade Three Law will be effective in increasing student achievement. • This question asked about several aspects of the law, which of which was “Read at Home” plans. Educator Group (Years) Teachers, Principals, Superintendents (2019-20, 2022-3) Response Options Not at all, Slightly, Moderately, Very well Teachers, Principals, Superintendents (2019-20, 2021-22, 2022-23) Not at all, To a small extent, To a moderate extent, To a great extent, I don’t know.1 To what extent do you agree with the following statements? Teachers (2019-20, 2020-21) Strongly disagree, Disagree, Agree, Strongly agree • This question contained several statements, one of which was, “I have sufficient time to create ‘Read at Home’ plans for students who are identified as having a ‘reading deficiency.’” 4 To what extent do you engage with families in each of the following ways? • Providing research-based guidance on how families can support literacy development • Providing family literacy workshops (e.g., family literacy night) 201 Not at all, To a small extent, To a moderate extent, To a great extent Teachers, Principals, Superintendents (2019-20, 2020-21, 2021-22, 2022-23) Teachers, Principals, Superintendents (2019-20, 2020-21) Response Options Table A2.4 (cont’d) RQ Item Text • Encouraging families to engage in literacy activities with their children at home on a regular basis • Sending home literacy resources that students can practice reading at home (e.g., books, letter name games, writing supplies) • Meeting with families to communicate students’ progress in literacy Educator Group (Years) Teachers, Principals, Superintendents (2019-20, 2020-21) Teachers, Principals, Superintendents (2019-20, 2020-21) Teachers and Principals (2019-20, 2020-21, 2021-22, 2022-23), Superintendents (2021-22, 2022-23) 1I exclude “I don’t know” responses because a) it is not clear whether this indicates a positive or negative response and b) it suggests the educator may not know what a “Read at Home” plan is. 202 Table A2.5 Linear Probability Models with Teacher and School Characteristics 3rd Grade Female Black Hispanic Asian Other Race Economically Disadvantaged English Learner (1) -0.00047 (0.012) 0.003 (0.004) 0.0553 (0.050) -0.00645 (0.038) 0.100* (0.055) -9.49E-05 (0.026) -0.0206 (0.025) 0.0593 (0.044) (2) -0.00455 (0.013) 0.003 (0.004) 0.0581 (0.049) -0.0143 (0.038) 0.108* (0.055) 0.00404 (0.026) -0.0229 (0.023) 0.0699 (0.043) (3) 0.0139 (0.012) 0.00395 (0.004) 0.00871 (0.023) -0.0075 (0.029) -0.0115 (0.038) -0.0275* (0.016) 0.0111* (0.006) 0.0142 (0.016) (4) 0.00764 (0.013) 0.00385 (0.004) 0.00912 (0.022) -0.0142 (0.028) -0.0148 (0.036) -0.0241 (0.016) 0.0114* (0.007) 0.0182 (0.016) Special Education 0.0271*** 0.0287*** 0.0240*** 0.0252*** Teacher Hired Last 5 Years Teacher Highly Effective Teacher ELA Endorsement School Proportion “Reading Deficient” (0.010) (0.010) 0.005 (0.021) -0.036 (0.053) -0.007 (0.017) School ELA Achievement School Proportion Non-White School Proportion Econ. Dis. School Proportion EL School Proportion SWD (0.007) -0.322* (0.177) (0.004) (0.003) 0.089 (0.147) (0.208) (0.188) 0.110 (0.195) 0.092 (0.452) (0.007) 0.00616 (0.020) -0.0474 (0.049) -0.0146 (0.016) -0.317* (0.173) -0.00395 (0.003) 0.0685 (0.155) -0.212 (0.183) 0.141 (0.195) 0.0628 (0.450) 203 Table A2.5 (cont’d) School Log Enrollment Charter Rural Urban Constant (1) (2) (3) 0.0817* (0.047) 0.001 (0.086) 0.021 (0.051) 0.212*** (0.079) 5.983 (4.348) 95,791 (4) 0.068 (0.047) -0.0145 (0.089) 0.0229 (0.051) 0.238*** (0.081) 5.464 (4.425) 86,071 Observations Note. Standard errors in parentheses (*** p<0.01, ** p<0.05, * p<0.1). All models include year fixed effects. 0.547*** (0.059) 109,623 0.568*** (0.059) 98,342 204 Table A3.1 Survey Items Used in the Study Question N Districts 86 The Read by Grade Three Law requires districts to identify students who have a “reading deficiency” if they “score below grade level or are determined at risk of reading failure based on a screening assessment, diagnostic assessment, standardized summative assessment, or progress monitoring assessment.” How is your district defining a “reading deficiency”? Please mark all that apply. What is your district using as its diagnostic assessment(s) per the Read by Grade Three Law? Please mark all that apply. Note. The second survey question asked superintendents to list their “diagnostic assessments,” although the intent was to inquire about screening assessments. The response options listed MDE’s approved screening assessments, and this question immediately followed the one about identification methods, which likely clarified our focus on screening tools. However, the potential misuse of terminology might have led some superintendents to report their diagnostic assessments, or possibly all assessments used in their early literacy assessment system, encompassing both screening and diagnostic tools. 72 205 Table A3.2 Sample Restrictions K-3 Students in TPS & Charter Schools District Received 35a(5) Funding Survey Data Available Superintendent Responded to Item of Interest N Students 421,519 407,697 41,461 38,491 N Districts 771 703 78 71 206 Table A3.3 Characteristics of Districts Not Specifying a “Reading Deficiency” Identification Approach Prop. “Rdg. Def.” ELA Ach. Prop. Non-White Prop. Econ. Dis. Prop. EL Prop. SWD Enrollment Class Size Charter Rural Urban Revenue NR 55.59% 0.071 19.68% 61.70% 3.28% 16.66% 424.29 23.74 0.00% 57.14% 0.00% $26.03 mil Sample 39.13% -0.038 28.02% 62.54% 5.42% 17.96% 542.13 22.450 12.68% 49.30% 8.45% $34.52 mil Pop. 37.49% -0.091 36.07% 65.50% 5.88% 16.38% 546.72 1.24 30.35% 40.39% 17.01% $34.91 mil NR vs. Sample 16.46% 0.110 -8.34% -0.84% -2.14% -1.29% -117.84 23.74 -12.68%** 7.85% -8.45** -$8.49 mil NR vs. Pop. 18.10% 0.162 -16.39%* -3.80% -2.60% 0.29% -122.43 22.76 -30.35%** 16.75% -17.01** -$8.88 mil Note. NR = Non Response 207 Table A3.4 Relationship between Student Demographics, Identification Approaches, and “Reading Deficiency” Status (1) -0.0708 (0.097) -0.00831 (0.113) -0.0643 (0.099) 0.0766 (0.106) 0.114 (0.066) 0.0215 (0.072) -0.0393 (0.072) -0.0391 (0.071) 0.206 (0.177) 0.0933 (0.073) 0.0269 (0.081) -0.0252 (0.087) -0.0306 (0.081) 0.261 (0.196) 0.0789 (0.069) -0.0186 (0.079) Multiple Assessments Hybrid Approach Comprehensive Approach Classwork Only 1st Grade 1st Grade x Multiple Assessments 1st Grade x Hybrid Approach 1st Grade x Comprehensive Approach 1st Grade x Classwork Only 2nd Grade 2nd Grade x Multiple Assessments 2nd Grade x Hybrid Approach 2nd Grade x Comprehensive Approach 2nd Grade x Classwork Only 3rd Grade 3rd Grade x Multiple Assessments 208 Table A3.4 (cont’d) 3rd Grade x Hybrid Approach 3rd Grade x Comprehensive Approach 3rd Grade x Classwork Only Female Female x Multiple Assessments Female x Hybrid Approach Female x Comprehensive Approach Female x Classwork Only Black Black x Multiple Assessments Black x Hybrid Approach Black x Comprehensive Approach Black x Classwork Only Hispanic Hispanic x Multiple Assessments Hispanic x Hybrid Approach Hispanic x Comprehensive Approach Hispanic x Classwork Only 209 (1) -0.0208 (0.087) 0.0317 (0.108) 0.312 (0.180) -0.00726 (0.020) 0.00473 (0.022) -0.0127 (0.025) -0.00254 (0.021) -0.0123 (0.026) 0.343*** (0.095) -0.276* (0.108) 0.0593 (0.158) -0.261 (0.107) -0.114 (0.113) 0.197*** (0.057) -0.135** (0.059) -0.165* (0.065) -0.167** (0.063) -0.0749 (0.086) Table A3.4 (cont’d) Asian Asian x Multiple Assessments Asian x Hybrid Approach Asian x Comprehensive Approach Asian x Classwork Only Other Race Other Race x Multiple Assessments Other Race x Hybrid Approach Other Race x Comprehensive Approach Other Race x Classwork Only Economically Disadvantaged Economically Disadvantaged x Multiple Assessments Economically Disadvantaged x Hybrid Approach Economically Disadvantaged x Comprehensive Approach Economically Disadvantaged x Classwork Only English Learner English Learner x Multiple Assessments English Learner x Hybrid Approach 210 (1) -0.224 (0.105) 0.119 (0.108) 0.118 (0.135) 0.105 (0.120) 0.383 (0.155) -0.00216 (0.077) 0.0253 (0.082) 0.0945 (0.103) 0.0285 (0.081) 0.212 (0.113) 0.238** (0.084) -0.0895 (0.086) -0.056 (0.111) 0.00707 (0.089) -0.284 (0.129) 0.0608 (0.099) -0.0625 (0.112) 0.208 (0.135) Table A3.4 (cont’d) English Learner x Comprehensive Approach English Learner x Classwork Only Student w/ Disabilities Student w/ Disabilities x Multiple Assessments Student w/ Disabilities x Hybrid Approach Student w/ Disabilities x Comprehensive Approach Student w/ Disabilities x Classwork Only Constant (1) -0.0662 (0.127) 0.00562 (0.116) 0.0564 (0.049) 0.131 (0.058) 0.143 (0.064) 0.126 (0.053) 0.0804 (0.087) 0.132 (0.094) 38,348 Observations Note. This table shows the results from Model (1). Robust standard errors in parenthesis. *** p <0.001 ** p < 0.01 * p <0.0125. I use 0.0125 as the Bonferroni-adjusted p-value threshold for statistical significance to adjust for making four comparisons (i.e., 0.05/4=0.0125). 211 APPENDIX FIGURES Figure A2.1 Example “Read at Home” Plan 212 Figure A2.1 (cont’d) 213 Figure A2.1 (cont’d) 214 Figure A2.1 (cont’d) 215 Figure A2.1 (cont’d) 216 Figure A2.1 (cont’d) 217 Figure A2.1 (cont’d) Note. This “Read at Home” plan is from Milan Area Schools (2017). 218