I taught fourth and fifth grade for several years at a charter school in DC. Then I moved into a central office role, where I designed and led professional learning for literacy teachers across the district. I also worked with school leaders to coach this same group of teachers in between professional learning days. Each summer, I’d analyze the end-of-year data and create a professional learning plan for the year. I’d ask myself, “What does the data tell me students need? What do teachers need to know and do to support their students in these areas?” I’d build a plan to support teachers as they supported students in the areas they needed most.
Our district-wide professional learning cadence was pretty standard: a few days before the start of the school year and one day about every other month during the school year. Usually, I’d choose a book that aligned with the shift we needed to see in classrooms, read it, and design the year’s learning around it. And because I was also their coach, what we covered in a session didn't end when the session did.
One year we worked through Making Thinking Visible, and how to build thinking routines into each part of the literacy block. Teachers would experience the routines themselves in our sessions. Sometimes I'd use a really juicy excerpt from the book, and other times artwork — something that would challenge their thinking so they'd have to actually apply the routine. We'd debrief how the routines changed our understanding. Then, each teacher made a plan to try one or two of the routines in a specific part of the literacy block the following week. When I visited their classrooms over the next two months, I'd look for evidence they were using them, and during planning time we'd talk about how to tweak a routine for a specific lesson. On the next professional learning day, we'd spend a few minutes hearing what worked and what didn't, and teachers would offer each other suggestions.
Year after year, even after school closures during COVID, our students made growth in reading. Part of the reason for our success was that we were on the same page. I was in the teachers’ classrooms every day. I was planning with them; observing lessons and providing feedback; and we were reviewing their students’ exit tickets or assessment data and creating plans to target specific skills in small groups. I wasn’t standing on the sidelines observing the work. I was doing the work alongside the teacher.
So when I stood up on professional learning days with a plan for what we needed to do to support students, the teachers came alongside me. And that’s how teacher practice shifts. Through relationships. You earn the opportunity to deliver a professional learning session by demonstrating that you can and will do the work.
And now as a parent with two kids in public school, I have another view of the classroom. This time, I’m outside looking in. I know what it takes to shift how reading gets taught, because I've done it. What I didn't expect was how little of that I could see from the outside, as a parent. So I went looking for what my state had actually done and what every other state tried too.
Pre-K Funding
Children arrive at kindergarten with wildly different preparation. Some students can read words with three sounds, and some students are working on identifying the letters in their name. Students who are behind on their first day of school can spend years trying to catch up. One study concluded that the chance of a struggling first-grade reader still struggling in fourth grade was close to nine in ten.¹In theory, pre-K can close that gap before it opens. If students learn the prerequisite skills they need before they get to kindergarten, more students could start kindergarten on track. It’s a proactive approach that requires a lot of infrastructure (and money) to implement successfully. Nationally, pre-K reaches about a third of four-year-olds.² Whether your family is in that third depends less on whether your child needs it than on where you live and what you can pay.
A fully funded approach means a parent can enroll their child in public school at four instead of five. Behind that, the state is paying school systems to find classroom space, buy curriculum, hire and train teachers, and monitor how children are doing.
This was my experience in Washington, DC with my daughter. DC is one of the few places that provides public pre-K for three-year-olds.³ When my daughter was three, we entered her in the pre-K lottery, and she got a spot at the public school near our house. If we had not gotten a spot through the lottery, we would have paid for another year for day care. But we were guaranteed a spot at our neighborhood school the following year, at four.
Florida runs a partially funded approach. Instead of funding public school programs, the state subsidizes tuition at privately owned centers. When my son entered the four-year-old class at his day care, we were told we could sign him up for Voluntary Pre-K (VPK) by filling out a form and save $30 a week on tuition during the school year. Florida claims to have universal pre-K, but the state only funds 540 instructional hours for the whole school year, or about 3 hours a day.⁴ Most early childcare providers will not accommodate this schedule, and parents are left to pay the difference or opt out entirely.
And that's before you get to the states where the answer is no. Forty-four states and DC fund a program that meets the standard researchers use. Six don't: Idaho, Indiana, Montana, New Hampshire, South Dakota, and Wyoming. Montana and Wyoming run early childhood programs that do similar work but fall outside the definition. Indiana funds pre-K only for children whose parents are working or in school, so the researchers who track this profile it separately and leave it out of the national numbers.²
Pre-K can make a real difference by the time a child reaches kindergarten. But it only works if parents can actually use it.
Professional Learning for Teachers
States also passed laws requiring school systems to train their teachers and leaders in the Science of Reading, a vast body of research about how the brain learns to read. Most of these laws included funding for training. In an eighteen-month stretch, 36 states spent more than $660 million training teachers already in classrooms.⁵Literacy reform laws were necessary because teachers, along with the curricula that school districts purchased, were using debunked methods for teaching students how to read. Methods like three-cueing, or teaching students to guess words based on the pictures or context, instead of teaching students how to sound out the words. Researchers have found, over and over, that all students benefit from instruction in foundational skills that is explicit, systematic, and sequential. Not guessing. Teach students sound and spelling patterns so that they can read any word. For decades, a lot of students weren't taught how.
Those three words can decide whether your child learns to read. Explicit means the teacher provides clear explanations and examples instead of letting the students figure it out. Systematic means there are routines with specific steps for teaching each concept. Sequential means the concepts progress from simple to more advanced so that nothing depends on a skill a student hasn't been taught yet.
Florida's laws are among the most complete in the country. State law requires districts to train reading coaches, classroom teachers, and administrators, and specifies that training in foundational skills must be based on the Science of Reading, with phonics as the primary strategy for teaching word reading. Teacher preparation programs are held to the same standard and are barred from teaching the three-cueing method. Anyone delivering the most intensive reading intervention, or teaching a third grader who's been held back, has to be reading endorsed or reading certified.⁶
And it still didn't reach everyone. Even with more than half of states requiring literacy training and funding it, NCTQ estimates over half a million teachers may not have gotten it.⁷
And here's what made this so easy to miss: many of the states’ standards did require students to show mastery of phonics concepts. What they did not say was how to teach it. The first-grade Common Core standard asks students to “know and apply grade-level phonics and word analysis skills in decoding words.” This describes an outcome instead of a method. A standard like this does not require instruction to be explicit, systematic, or sequential. A teacher could teach to this standard using a balanced literacy approach, meaning they teach the phonics concepts as they “come up” in literature instead of systematically in a sequence where concepts build on each other. With this approach, if the concept never “comes up,” it doesn’t get taught. The reading laws states passed requiring instruction aligned to the Science of Reading were written to close this gap.⁸
So yes, teachers share some of this responsibility. And, while using methods aligned to the Science of Reading will certainly move the needle for student reading proficiency, it’s not enough. In fact, it lays decades’ worth of neglect at the feet of teachers and asks them to fix it with inadequate resources and embarrassingly low compensation. Do more with less for even less. And none of it — the materials, the training, the support for using both — was ever the teacher's decision to make.
So did any of it work?
States passed the laws and spent the money, but reading kept going the wrong direction.The timing matters. Reading scores stopped improving in 2013, seven years before the pandemic, and before most of these laws were written. In the two years right before schools closed, reading was already falling as fast as it would fall during the closures. Eighth-grade reading is now at its lowest point since 1990, and fourth grade is back where it was before 2003.⁹
Math recovered. By 2024, it was improving at the rate it managed before any of this started. Reading kept falling. Between 2022 and 2025 the average school district went backward by about 15% of a year's reading growth, and twenty-seven states fell further behind than they already were.⁹
That’s the bad news. But there’s good news, too.Researchers at Harvard and Stanford compared state reading gains against how many of eighteen tracked literacy policies each state had actually implemented as of January 2024. Every state that improved between 2022 and 2025 had implemented more than seven. No state that had implemented fewer than seven improved at all.¹⁰
So the count of reforms mattered up to a point, and then it stopped mattering. Passing more laws did not produce more improvement.¹⁰
Some states implemented multiple pieces of these reforms and still haven't turned around. The researchers point to Florida, Arizona, and Nebraska as examples. Their conclusion is that evidence-based reading reform may be necessary but insufficient.¹⁰
Florida is on that list.
Florida started its statewide reading initiative in 2001, before almost anyone else. Since then it's mandated Science of Reading training, required aligned curriculum, held back third graders who can't read, and partially funded VPK for every four-year-old.⁶
For a long time, it paid off. Florida's fourth-grade reading scores climbed from 1998 onward, and Florida still does better than most of the country.
But between 2022 and 2024, the share of Florida fourth graders reading at a basic level or above dropped from 71% to 62%. Nearly four in ten are now below basic, meaning they can't reliably understand what's on the page in front of them.¹¹
My daughter attended the first part of kindergarten at a private school with a reputation for rigorous academics. And no, it was not the school where students were getting explicit phonics instruction, whatever I'd been told on the tour. She spent the second half of the year at a highly ranked public school. Honestly, I was disappointed with both.
At the private school, my daughter’s teacher was using a balanced literacy approach to foundational skills instruction, often called Word Work or analogy phonics. This approach provides students with a list of words that have the same ending, sometimes called word families (e.g., sat, bat, cat, mat). Students are taught to “decode” the words by looking at the end of the word first (–at). Then, they guess the first sound. Sometimes this approach includes pictures that go along with the words. About a month into school, we received a list of over 100 words with a note from her kindergarten teacher that we needed to learn them at home. My daughter had already learned how to read in pre-K at our neighborhood school in DC, so she continued to sound out the words instead of memorizing them or guessing.
Her kindergarten teacher at the public school sent home leveled readers each week for students to read. She started my daughter out on Level A books, which were inappropriate for several reasons. My daughter was reading well beyond “Level A.” These books have predictable and repetitive sentence patterns. A Level A book reads something like this: I see a hat. (with a picture of a hat); I see a shoe. (with a picture of a shoe); I see a dog. (with a picture of a dog). You get it. These books have no plot, and they often do not make any sense. Leveled readers are also a cornerstone of the balanced literacy approach and not aligned to the Science of Reading.
How is it that after two decades of reading reform, in a state with some of the most complete reading laws in the country, two kindergarten teachers at two different but well-resourced schools are still using a debunked approach to teaching reading?
What was missing?
States legislated hard on what teachers should know. They spent far less effort on the curriculum materials and coaching support for teachers. When researchers at the Albert Shanker Institute coded the reading laws states passed, they found that curriculum and coaching got noticeably less attention than anything else. Professional development was discussed extensively in 32 states' laws. Teacher preparation, 25. Curriculum, 16. Coaching, 11.¹² Only nine states' laws send districts to an approved curriculum list. Another fourteen or so provide lists as recommendations, sometimes mandates, but the lists are not part of the reading laws. Which leaves about half the country where nothing in the law tells a district that the curriculum has to match the research.¹²So, teachers across the country are in a real predicament. Let’s say their district purchased a balanced literacy curriculum, like Lucy Calkins’s Units of Study, prior to the Science of Reading laws and mandated training. The district provided the teachers with training in the Science of Reading but did not purchase a new curriculum. Teachers are left to decide if they will continue to use the familiar balanced literacy materials or recreate a curriculum from scratch aligned to an approach they just learned about. If a hospital system sent all of its doctors to a conference to learn about a highly effective treatment and then chose not to provide the equipment or medicine required to implement it, we’d have a lot of questions for the hospital, not the doctors.
It’s not enough to tell teachers they need to stop doing what they’ve been doing and do something different. States and school districts also need to provide teachers and students with high-quality curriculum materials, training on how to use them, and coaching to support implementation. Without these critical pieces, teachers are made to feel solely responsible for fixing the whole problem.¹³
Coaching got even less attention. Naming a coach in a law is not the same as funding one.¹² In the research, coaching means something specific: an instructional expert working one-on-one with a teacher, consistently, over months, inside that teacher's own classroom, on specific skills.¹⁴
And there's strong evidence that it works. Researchers pooled sixty studies of coaching, most of them randomized trials, and most of them about reading. They found that coaching changed how teachers taught. And it showed up in how much children learned: students gained about as much as they do when a new teacher grows into an experienced one, which normally takes five to ten years.¹⁴
They also found that coaching worked better alongside training than on its own, and the researchers think the order matters. Teachers who learned about a topic or strategy in a workshop first, and then had a coach in their room to help them apply it, saw roughly double the results of teachers who only got a coach.¹⁴ Training and coaching were never meant to be separate line items, but states treated them like they were.
The conditions for coaching also matter. The researchers found that coaching worked best in small programs. Once a program grew past about a hundred teachers, the results fell to roughly a third of what the small ones achieved. The reason isn't mysterious: small programs used a few carefully chosen coaches, working with teachers who wanted to be there. Large programs had to hire a coaching corps and send them everywhere.¹⁴
I was on the small end of that. I led professional learning and coached about forty teachers a year in DC, few enough that I could be in their rooms often enough to know how each of them taught.
The largest study in that whole meta-analysis was Florida. More than 2,300 reading coaches, working across the state. Over the four years researchers tracked it, the average effect on reading was close to zero, detectable in only two of those years, and tiny even then. Nobody can say exactly which part broke, whether it was making it mandatory or the impossibility of finding 2,300 good coaches. But something did.¹⁴One more thing from that Harvard and Stanford report. When the researchers say what states should be measuring next, they name two things: whether student achievement improves when a school changes its curriculum, and whether it improves when a school gets a literacy coach. They want the federal government to pay for evaluating them one at a time instead of the whole package at once.¹⁰
Curriculum and coaching. Three sets of researchers got there separately — one reading state test scores for 35 million children, one reading the text of the laws, one reading sixty coaching studies. All three land on the same two things, and those are the two that got the least attention when states wrote their reading laws.A teacher who has been trained in the Science of Reading, handed a curriculum that contradicts it, and given no coaching to reconcile the two is not a teacher who has been given a reform. She's been given a contradiction and told to sort it out during her planning period.
My daughter still goes to the public school. My son just started kindergarten there. I’m going to approach the shift I want to see in classrooms like I always have. By doing the work. Earning trust before making suggestions.
So, I joined the PTSA. I’m going to attend the monthly School Advisory Council meetings. I’m adding my name to volunteer in the classroom when I can. Each time they see my friendly face lending a helping hand will be another deposit of trust. I’ll keep you posted on how it goes.
Next week: when your child is behind in reading, the first thing most of us do is look for a tutor. Reading intervention is something different, and it's usually what a struggling reader actually needs.
Endnotes
- Juel, C. (1988). "Learning to Read and Write: A Longitudinal Study of 54 Children From First Through Fourth Grades." Journal of Educational Psychology, 80(4), 437-447. Juel found a .88 probability that a poor reader at the end of first grade would still be a poor reader at the end of fourth. Spira et al. (2005) found a comparable 70% probability for first graders below the 30th percentile. This finding has been challenged — Phillips, Norris, Osmond & Maynard (2002), following 187 children in rural Canada, found reading categories more flexible than Juel's sample suggested. The consistent conclusion across both bodies of work is that struggling early readers rarely catch up without intervention.
- National Institute for Early Education Research, State of Preschool 2025 Yearbook, covering the 2024-25 school year. State-funded preschool enrolled 37% of four-year-olds nationally. State spending reached nearly $14.4 billion across 44 states and the District of Columbia. Six states did not fund a program meeting NIEER’s definition: Idaho, Indiana, Montana, New Hampshire, South Dakota, and Wyoming. Indiana’s On My Way Pre-K is profiled in the Yearbook but excluded from national totals and rankings because eligibility carries a parental work or school requirement; Virginia’s Mixed Delivery and CCSP programs are excluded on the same grounds, though Virginia’s Preschool Initiative does meet the definition. NIEER also notes that Montana and Wyoming operate early childhood programs providing services similar to state-funded preschool, so the six are states without a program meeting NIEER’s criteria rather than states with no early childhood funding of any kind. Definition criteria are set out in the Yearbook’s methodology section.
- NIEER reports DC enrolling 82% of three-year-olds, by far the highest in the nation; nationally, 9% of three-year-olds are enrolled in state-funded preschool.
- Fla. Stat. §1002.55(2) and §1002.63(2) require a school-year VPK program to comprise 540 instructional hours (private providers and public schools, respectively); §1002.61(2) sets the summer program at 300 hours. FLDOE notes that providers may structure daily and weekly hours as they choose so long as the annual total is met, so 540 hours across a roughly 180-day calendar averages about three hours a day rather than being fixed at three hours by law. Legislation to raise the school-year requirement to 1,440 hours (SB 70 / HB 191, 2025) died in committee.
- Holston, S. (2024). State of the States 2024: Five Policy Actions to Strengthen Implementation of the Science of Reading. National Council on Teacher Quality, January 2024. NCTQ reports that 36 states invested over $660 million in literacy training for teachers already in the classroom within the preceding eighteen months; the figure is the organization’s own, stated on its report page, rather than a secondary paraphrase. Author attribution follows the Fordham Institute’s citation of the report rather than a byline in the report itself, and no page number has been verified against the full PDF. Much state literacy-training spending in this window drew on federal ESSER relief funds, which have since ended; the share is not itemized here. NCTQ conducts advocacy alongside its research.
- Fla. Stat. §1012.98 (district training requirements); §1004.04 (teacher preparation program standards, including the prohibition on three-cueing); §1011.62 and §1008.25 (reading endorsement requirements for Tier 3 intervention and for teachers of retained third graders). On curriculum: §1001.215 directs the Just Read, Florida! Office to work with the Florida Center for Reading Research to identify scientifically researched and evidence-based reading instructional and intervention programs incorporating explicit, systematic, and sequential approaches to phonemic awareness, phonics, vocabulary, fluency, and text comprehension, and incorporating decodable or phonetic text strategies. §1006.28(2)(c) requires each district superintendent to certify annually to the Commissioner of Education that core reading and reading intervention materials used in kindergarten through grade 5 meet those requirements, and that materials are being implemented as designed. Fla. Admin. Code R. 6A-6.053 requires district reading plans to provide explicit and systematic core instruction (Tier 1) and explicit, systematic intervention at Tiers 2 and 3. Subsection numbering within §1001.215 has changed across statutory revisions; cite the current year’s text.
- NCTQ, State of the States 2024: Five Policy Actions to Strengthen Implementation of the Science of Reading, January 2024. The figure appears as a section heading: thirty states require science-of-reading professional learning for elementary teachers and thirty-six invested over $660 million in the previous eighteen months, yet "over half a million teachers may be left without it." NCTQ presents this as an estimate rather than a headcount and does not publish its derivation in the executive summary.
- National Governors Association Center for Best Practices & Council of Chief State School Officers (2010). Common Core State Standards for English Language Arts, Reading: Foundational Skills, Grade 1. The standard quoted above is RF.1.3; the same anchor standard appears at kindergarten and grades 2 through 5. Its sub-standards specify skills students should acquire — decoding regularly spelled one-syllable words, knowing spelling-sound correspondences for common consonant digraphs — but do not specify how instruction should be delivered.
- Dewey, D.C., Fahle, E.M., Kane, T.J., Reardon, S.F., & Staiger, D.O. (2026). From Learning Recession to Learning Recovery: Understanding the Sources of U.S. K-12 Improvement. Education Scorecard, May 2026; see note 10 for the collaboration and data source. The report dates the onset of the “learning recession” to 2013, when math and reading improvement stalled and average scores began to decline, and finds the average annual loss in reading immediately before the pandemic (2017–19) was as large as during it (2019–22). It reports Grade 8 NAEP reading at its lowest point since 1990 and Grade 4 at pre-2003 levels. Math’s annual rate of improvement returned to pre-2013 levels in 2022–24 while reading continued to decline through 2024, with 2025 showing the beginnings of a turnaround. Between 2022 and 2025 the average district improved by .01 grade equivalents in math and lost .15 grade equivalents in reading, and twenty-seven states declined further in reading. Achievement is measured in grade-level equivalents, where one grade level is roughly the average improvement students make from one grade to the next between fourth and eighth grade; the “15% of a year’s reading growth” phrasing above converts the .15 figure and is not the authors’ own wording. The 2025 estimates cover only the 40 states where tests and proficiency definitions were unchanged from 2024 and adequate district-level data were available; the authors note the spring 2025 state administrations fell between the 2024 and 2026 NAEP administrations and that 2025 estimates will be updated when the 2026 NAEP is released.
- Dewey, D.C., Fahle, E.M., Kane, T.J., Reardon, S.F., & Staiger, D.O. (2026). From Learning Recession to Learning Recovery: Understanding the Sources of U.S. K-12 Improvement. Education Scorecard, a collaboration of the Center for Education Policy Research at Harvard University, the Educational Opportunity Project at Stanford University, and faculty at Dartmouth College. The analysis links state test results for roughly 35 million students in grades 3-8 to the NAEP scale via the Stanford Education Data Archive. Seven states plus the District of Columbia improved in reading between 2022 and 2025 (Louisiana, Maryland, Tennessee, Kentucky, Indiana, Mississippi, Minnesota, and D.C.); 27 states declined further. The eighteen policy elements are tracked by the Foundation for Excellence in Education (ExcelinEd); the researchers used ExcelinEd’s January 2024 state reports, counting elements in full or partial implementation. The ten states below seven elements were South Dakota, Hawaii, Rhode Island, New Jersey, New Hampshire, Georgia, Wisconsin, California, Washington, and Massachusetts. Two caveats from the report: it finds no strong correlation between the magnitude of improvement and the number of policy elements adopted, so the finding is a threshold rather than a dose response; and the 2025 analysis covers only the 40 states with comparable data, excluding Alaska, Colorado, Illinois, Kansas, Maine, Montana, New Mexico, New York, North Dakota, Oklahoma, Oregon, and Vermont. The researchers caution that this is far from conclusive evidence that adopting science-of-reading policies will produce improvement, concluding that a comprehensive literacy strategy may be a necessary condition for improvement while adoption alone is not sufficient. In their recommendations and conclusion the authors call for a national effort to identify which of these reforms is most effective, proposing that states use their own data systems to track whether achievement growth improves when schools change curricula or receive a literacy coach, that the federal government fund evaluations of specific policy aspects such as literacy coaches and third-grade retention, and that payments to literacy curriculum providers be made contingent on students’ achievement growth.
- NCES, 2024 Reading State Snapshot Report: Florida, Grade 4, Public Schools. Florida's average score of 218 was higher than the national public-school average of 214; 62% performed at or above NAEP Basic. Florida's 2024 results were significantly lower than 2022 (average score 225, 71% at or above Basic) and significantly higher than 1998 (206, 53%).
- Neuman, S. B., Quintero, E., & Reist, K. (2023). Reading Reform Across America: A Survey of State Legislation. Albert Shanker Institute, July 2023, pp. 14 and 16 (Table 4). The authors coded 223 bills enacted in 45 states and DC between 2019 and 2022 across more than 40 dimensions; inter-rater reliability averaged Cohen’s Kappa 0.75. Table 4 reports the number of states whose laws discuss each support extensively: professional development 32, teacher preparation 25, curriculum 16, principal leadership 17. Nine states’ laws directly refer districts to approved curriculum lists; ten states’ laws provide no curriculum information at all. Coaching is described in the legislation of 20 states and more extensively in 11; these coaching counts appear in the report’s discussion rather than in Table 4, which covers teacher preparation, professional development, curriculum, and principal leadership only. Three limitations the authors state: the analysis covers legislation only, and they note an American Public Media analysis finding some 14 additional states maintain curriculum lists not referenced in the laws coded; the study does not track funding provisions or supplementary guidance accompanying legislation; and five states (Hawaii, West Virginia, New Hampshire, New Jersey, New York) enacted no reading legislation during the study period and are excluded from the counts. The Albert Shanker Institute is endowed by the American Federation of Teachers; AFT president Randi Weingarten is also the institute’s president and co-signed the report’s foreword. Research funded in part by the Ibis Group.
- Albert Shanker Institute (2023). The report concludes that this legislation relies largely on increasing teachers’ knowledge through preparation and professional development, while placing far less attention on the other infrastructural features needed to ensure teachers can teach in accord with the science, and warns that without an interconnected, coherent infrastructure, teachers will be “in the hot seat, potentially made to feel solely responsible” for the entire improvement process.
- Kraft, M.A., Blazar, D., & Hogan, D. (2018). “The Effect of Teacher Coaching on Instruction and Achievement: A Meta-Analysis of the Causal Evidence”. Review of Educational Research, 88(4), 547-588. Sixty studies using causal designs (56 randomized controlled trials), yielding 186 effect sizes for instruction across 43 studies and 113 for achievement across 31. Pooled effects: 0.49 SD on instructional practice and 0.18 SD on student achievement (Table 2). The authors define coaching as individualized, intensive, sustained, context-specific, and focused on discrete skills. On pairing: the meta-regression in Table 4 finds that combining coaching with group training is associated with effects 0.31 SD larger on instruction and 0.12 SD larger on achievement, against intercepts of 0.24 SD and 0.09 SD for coaching without it — the basis for the “roughly double” comparison above, which is arithmetic from those figures rather than a number the authors state. The authors stress these regressions are descriptive rather than causal, and that limited statistical power prevents ruling out smaller relationships; in the fully specified model the instruction coefficient holds at 0.20 SD while the achievement coefficient falls to 0.09 SD and is no longer statistically significant. Forty-eight of the 60 studies paired coaching with group training, so the coaching-alone comparison group is small. On scale (Table 6): programs with fewer than 100 teachers averaged 0.63 SD on instruction and 0.28 SD on achievement, against 0.34 SD and 0.10 SD for programs of 100 or more, both differences significant at .05. Adjusting for possible publication bias by trim-and-fill lowers the pooled estimates to 0.34 SD and 0.14 SD. The statewide Florida program is Lockwood, J.R., McCombs, J.S., & Marsh, J. (2010), “Linking Reading Coaches and Student Achievement: Evidence from Florida Middle Schools,” Educational Evaluation and Policy Analysis 32(3), 372-388 — the largest study in the sample, covering more than 2,300 reading coaches; Kraft et al. report effects on reading achievement significant in only two of four years, with average effect sizes between 0.01 and 0.03 SD, and note they cannot determine whether this reflects the program’s mandatory nature, the challenge of staffing a coaching corps that size, or other factors. Nearly all content-specific studies concerned reading; pooled effects for math (0.04 SD) and science (0.35 SD) were not statistically significant and rest on too few studies to support inference. The authors also report, as explicitly exploratory findings, a weighted correlation of .37 between instruction and achievement effects and an estimate that a 1 SD change in instruction corresponds to a 0.21 SD change in achievement, both at p = .16. Page numbers refer to tables rather than pages, as these are stable across the published and pre-pagination versions.