The explanation for the fadeout in test scores may well be that Success for All does a good job of teaching phonics/decoding but a poor job of building the academic knowledge and vocabulary that becomes increasingly important to reading proficiency as grade levels go up. I wrote about that possibility on my own Substack a while ago ( https://nataliewexler.substack.com/p/to-gauge-success-we-need-to-look ).
At the time, I wasn't sure whether Success for All actually built knowledge systematically of not. But after my post came out, I was contacted by someone from Success for All and I was able to ascertain that the curriculum is not organized around topics -- just skills, including comprehension skills.
Maybe Steubenville kids are acquiring knowledge in social studies and science, but most schools spend way more time on math and especially reading than on the "content areas." And it sounds like the kids in Steubenville aren't acquiring much knowledge from their reading curriculum. If their parents have limited education themselves, they probably aren't acquiring much academic knowledge and vocabulary at home either. So when they get to middle school and certainly high school, where texts start assuming a lot more knowledge (and familiarity with complex syntax), they're likely to hit a wall.
There is a tendency among districts to focus on contextual differences and think they can only learn from others exactly like them. Steubenville, a small city district in a community that lost a lot of industrial jobs, doesn't seem that generalizable. Only 3% of school districts fit a similar profile. Doesn't make it right, but that's how it is.
I'm doing some of the legwork you call for, writing a case study on Worcester County, a high-performing, high-poverty school district in rural Maryland. I hadn't planned on submitting it to a peer-reviewed journal-- it's a big headache-- but you make a strong case for the need. The rigorous follow-up could be done with the right funding: survey research that equally samples sustained excellence and districts spinning their wheels.
You write "sainthood and fraud aren't the only possibilities." Washington, DC is a reminder that they can co-exist! John Merrow found real cheating going on in DC, but the Rhee/Henderson leadership also generated real growth on NAEP. 4th grade reading scores grew more than a full grade level from 2009 to 2017.
For every Finland, we have a Singapore, an Ireland, an Ontario, even a Long Beach and Boston with real gains for a time. Seems to me benchmarking is valuable even if some success is a mirage and other efforts fade or stall out.
(1) “We would need to believe that a cheating conspiracy can be maintained across multiple buildings without any educators, students, or parents crying foul. Not easy.” <—-Hello Atlanta Public Schools, 2009. (2) A cohort of students that does well in early grades with diminishing results in middle school would honestly be what I would expect from a Success For All district that extends it through middle school. But I guess I have a challenge for you to find a district that maintains early success through grade 8, in general, because it is rare. Asking for a friend who is a newly elected school board member.
I love this Tim--yes, we should use outliers to form hypotheses, not jump to prescriptions. And specifically, analyzing and pursuing ideas about what's replicable. After all, in research the question isn't "What works?" but rather "What works, for whom, and under which conditions?" I'm in favor of finding places with fairly normal conditions who still find ways to be outliers. My nominee for your list? Worcester County, MD. I've really enjoyed working with them in the past and they are the real deal!
I see other trying to make sense of the reason for that reading drop off but we have to call this what it is. This has been a known thing for Steubenville for such a long time. I left an internship in 2011 because teachers were telling me they were pressured to do this. I was a naive intern so I reported it not realizing it was a district wide issue. Let me tell you, I was treated HORRIBLY. After I left, I reported it to the State Support Team and their administrators said they knew but couldn’t do anything. I don’t think I’ve ever felt more vindicated when I pulled up the list today and saw that ALL SIX of Steubenville City’s schools were found to have anomalies on Ohio state tests. Wells Academy specifically got recognized repeatedly by the governor for getting super high scores. For anyone who is familiar with statistics and psychometrics, it was always clear that those scores just were not possible. Just to give you an idea of how bad Steubenville’s cheating patterns must be—Cleveland Municipal is the second largest school district in Ohio with 68 schools. Only 3 of their schools were flagged. Steubenville has 6 schools and all 6 were flagged.
The piece of this story I keep coming back to isn't actually the testing anomaly. It's this:
Why does Steubenville's extraordinary elementary reading performance erode so dramatically by eighth grade?
That question deserves at least as much curiosity as the district's remarkable third-grade scores.
I say that both as someone who works in learning and as a parent who has spent years trying to understand what happens when a child who is clearly intelligent—and who appears to be reading successfully—doesn't develop the underlying machinery needed to make increasingly complex reading automatic.
We have lived a version of this.
My child had years of schooling without sufficient explicit exposure to the orthographic skills she needed. She could read. She could comprehend. She could think deeply. She could compensate.
And for a long time, those things can look like the same thing as having a fully developed reading system.
They aren't.
We have had to fight to put that instruction back in place—to explicitly address the relationships among sounds, spelling patterns, morphology, word structure, and the automatic recognition of increasingly complex words.
That experience has changed the question I ask when I see reading data like Steubenville's.
Daly reports that 99% of Steubenville third graders were proficient in reading in 2025, while only 49.5% of eighth graders were. At the same time, eighth-grade math remained extraordinarily strong. That discrepancy is fascinating.
Perhaps the investigation will uncover testing problems. Perhaps it won't.
But even if every elementary score is legitimate, I don't think we should treat the eighth-grade decline as merely an odd footnote to an otherwise remarkable success story.
It may tell us something important about reading instruction itself.
Third-grade reading and eighth-grade reading are not simply easier and harder versions of the same task.
As students move through school, the linguistic load changes. Words become longer and morphologically complex. Vocabulary becomes increasingly specialized. Sentences become syntactically denser. Background knowledge matters more. Students encounter words such as industrialization, constitutional, photosynthesis, socioeconomic, and metamorphosis rather than the vocabulary of early elementary texts.
At that point, reading fluency depends on an increasingly sophisticated interaction among decoding, orthographic knowledge, morphology, vocabulary, syntax, background knowledge, working memory, attention, and automaticity.
A student can compensate for weaknesses in some of those systems remarkably well when texts are simple.
Especially a bright child.
Context helps. Pictures help. Vocabulary helps. Reasoning helps. Memory helps. A child can predict a word from the sentence or recognize enough of it to keep moving. Strong comprehension can mask inefficient word reading for a surprisingly long time.
Then the text changes.
There are fewer contextual supports. The vocabulary becomes unfamiliar. Words contain multiple meaningful units. Sentences carry several ideas at once. Students aren't merely reading stories anymore; they're reading science, history, mathematics, primary sources, technical explanations, and abstract arguments.
Eventually compensation becomes expensive.
The child hasn't necessarily become a worse reader. The environment has finally exceeded the compensatory system's capacity.
That's one reason I would be very cautious about looking at exceptional third-grade reading proficiency and concluding that we have discovered an exceptional K–12 reading system.
We may have discovered an exceptional early-literacy system.
Those are not necessarily the same thing.
And that is where I think Daly's larger argument becomes even more important.
He argues that if Steubenville were a corporation outperforming its competitors for decades, researchers and consultants would have descended on it to figure out what it was doing. Instead, remarkably little rigorous investigation has been done.
I agree—but I would expand the investigation.
Don't just ask, "What is Steubenville doing that produces these extraordinary third-grade scores?"
Ask:
What happens to those readers afterward?
Follow the children longitudinally.
Look beneath proficiency.
Look at decoding and word recognition. Look at spelling. Look at orthographic mapping and automaticity. Look at morphological knowledge. Look at vocabulary and background knowledge. Look at oral versus silent reading. Look at comprehension when the subject matter is familiar and when it isn't. Look at reading rate and accuracy when students encounter unfamiliar multisyllabic words.
And perhaps most importantly, determine whether the children who were "proficient" in third grade developed the underlying reading architecture that lets them keep becoming more sophisticated readers without relying more and more on compensation.
Because proficiency is a threshold, it isn't a diagnosis of how a child reads.
That's the lesson my own family has learned the hard way.
When a capable child crosses the proficiency line, the system tends to stop asking questions. The score says the child can read. The child may even have excellent comprehension. Meanwhile, weaknesses in orthographic knowledge or automatic word recognition can remain largely invisible.
By middle school, those weaknesses aren't invisible anymore.
And once that happens, families can find themselves in the bizarre position of fighting to obtain foundational literacy instruction for a child everyone agrees is intelligent and who everyone has spent years describing as a successful reader.
That should make us reconsider what we mean by "success."
Daly writes that successful school systems should be studied with the intensity that businesses devote to unusually successful companies. Absolutely. But longitudinal erosion deserves the same curiosity as exceptional performance.
In fact, it may tell us more.
If Steubenville's early reading results withstand scrutiny, I don't want to know only why its third graders perform so well.
I want to know why so many of those third-grade readers apparently aren't still performing that way five years later.
What did they learn?
What didn't they learn?
What did the assessments measure?
What were students able to compensate for at eight or nine years old that they could no longer compensate for at thirteen or fourteen?
And what happened instructionally between those two points?
Those aren't questions that diminish Steubenville's accomplishments.
They're exactly the questions we should ask if we're serious about learning from them.
The goal shouldn't be to find a reading program that produces extraordinary third-grade scores.
It should be to understand how we create readers whose skills keep deepening as the intellectual demands of reading increase.
As a parent, that's the outcome I care about.
And I suspect that's where some of the most important information in the Steubenville story may ultimately be hiding.
I'm curious about Roma ISD in Texas. It's in the poorest county in Texas, with lots of emerging bilingual students and low education rates among parents, but the elementary schools get extremely high reading scores
No one needs to adjust NAEP reading scores for demographics Dan. It lets states like Texas--with high fiscal capacity but below average school funding-- off the hook. Mississippi, by contrast, would love to have Texas's capacity but doesn't, and maxes out what it can because governors going back to Haley Barbour see a strong connection between education and economic growth.
Hi David. Not sure what you are saying. If Mississippi does not adjust scores for demographics, like poverty, the adjusted scores go down. They would no longer be number one for math and reading. I'm not defending Texas by any means. Just saying that a closer look at the Mississippi data reveals less than impressive results. Where are the reading gains for 8th grade and 11th grade (ACT)? Educational miracles never outlast data.
The explanation for the fadeout in test scores may well be that Success for All does a good job of teaching phonics/decoding but a poor job of building the academic knowledge and vocabulary that becomes increasingly important to reading proficiency as grade levels go up. I wrote about that possibility on my own Substack a while ago ( https://nataliewexler.substack.com/p/to-gauge-success-we-need-to-look ).
At the time, I wasn't sure whether Success for All actually built knowledge systematically of not. But after my post came out, I was contacted by someone from Success for All and I was able to ascertain that the curriculum is not organized around topics -- just skills, including comprehension skills.
Maybe Steubenville kids are acquiring knowledge in social studies and science, but most schools spend way more time on math and especially reading than on the "content areas." And it sounds like the kids in Steubenville aren't acquiring much knowledge from their reading curriculum. If their parents have limited education themselves, they probably aren't acquiring much academic knowledge and vocabulary at home either. So when they get to middle school and certainly high school, where texts start assuming a lot more knowledge (and familiarity with complex syntax), they're likely to hit a wall.
There is a tendency among districts to focus on contextual differences and think they can only learn from others exactly like them. Steubenville, a small city district in a community that lost a lot of industrial jobs, doesn't seem that generalizable. Only 3% of school districts fit a similar profile. Doesn't make it right, but that's how it is.
I'm doing some of the legwork you call for, writing a case study on Worcester County, a high-performing, high-poverty school district in rural Maryland. I hadn't planned on submitting it to a peer-reviewed journal-- it's a big headache-- but you make a strong case for the need. The rigorous follow-up could be done with the right funding: survey research that equally samples sustained excellence and districts spinning their wheels.
You write "sainthood and fraud aren't the only possibilities." Washington, DC is a reminder that they can co-exist! John Merrow found real cheating going on in DC, but the Rhee/Henderson leadership also generated real growth on NAEP. 4th grade reading scores grew more than a full grade level from 2009 to 2017.
For every Finland, we have a Singapore, an Ireland, an Ontario, even a Long Beach and Boston with real gains for a time. Seems to me benchmarking is valuable even if some success is a mirage and other efforts fade or stall out.
I was thinking of you when I read this David! I just nominated Worcester too :) Well said.
Please submit!
(1) “We would need to believe that a cheating conspiracy can be maintained across multiple buildings without any educators, students, or parents crying foul. Not easy.” <—-Hello Atlanta Public Schools, 2009. (2) A cohort of students that does well in early grades with diminishing results in middle school would honestly be what I would expect from a Success For All district that extends it through middle school. But I guess I have a challenge for you to find a district that maintains early success through grade 8, in general, because it is rare. Asking for a friend who is a newly elected school board member.
I love this Tim--yes, we should use outliers to form hypotheses, not jump to prescriptions. And specifically, analyzing and pursuing ideas about what's replicable. After all, in research the question isn't "What works?" but rather "What works, for whom, and under which conditions?" I'm in favor of finding places with fairly normal conditions who still find ways to be outliers. My nominee for your list? Worcester County, MD. I've really enjoyed working with them in the past and they are the real deal!
I see other trying to make sense of the reason for that reading drop off but we have to call this what it is. This has been a known thing for Steubenville for such a long time. I left an internship in 2011 because teachers were telling me they were pressured to do this. I was a naive intern so I reported it not realizing it was a district wide issue. Let me tell you, I was treated HORRIBLY. After I left, I reported it to the State Support Team and their administrators said they knew but couldn’t do anything. I don’t think I’ve ever felt more vindicated when I pulled up the list today and saw that ALL SIX of Steubenville City’s schools were found to have anomalies on Ohio state tests. Wells Academy specifically got recognized repeatedly by the governor for getting super high scores. For anyone who is familiar with statistics and psychometrics, it was always clear that those scores just were not possible. Just to give you an idea of how bad Steubenville’s cheating patterns must be—Cleveland Municipal is the second largest school district in Ohio with 68 schools. Only 3 of their schools were flagged. Steubenville has 6 schools and all 6 were flagged.
The piece of this story I keep coming back to isn't actually the testing anomaly. It's this:
Why does Steubenville's extraordinary elementary reading performance erode so dramatically by eighth grade?
That question deserves at least as much curiosity as the district's remarkable third-grade scores.
I say that both as someone who works in learning and as a parent who has spent years trying to understand what happens when a child who is clearly intelligent—and who appears to be reading successfully—doesn't develop the underlying machinery needed to make increasingly complex reading automatic.
We have lived a version of this.
My child had years of schooling without sufficient explicit exposure to the orthographic skills she needed. She could read. She could comprehend. She could think deeply. She could compensate.
And for a long time, those things can look like the same thing as having a fully developed reading system.
They aren't.
We have had to fight to put that instruction back in place—to explicitly address the relationships among sounds, spelling patterns, morphology, word structure, and the automatic recognition of increasingly complex words.
That experience has changed the question I ask when I see reading data like Steubenville's.
Daly reports that 99% of Steubenville third graders were proficient in reading in 2025, while only 49.5% of eighth graders were. At the same time, eighth-grade math remained extraordinarily strong. That discrepancy is fascinating.
Perhaps the investigation will uncover testing problems. Perhaps it won't.
But even if every elementary score is legitimate, I don't think we should treat the eighth-grade decline as merely an odd footnote to an otherwise remarkable success story.
It may tell us something important about reading instruction itself.
Third-grade reading and eighth-grade reading are not simply easier and harder versions of the same task.
As students move through school, the linguistic load changes. Words become longer and morphologically complex. Vocabulary becomes increasingly specialized. Sentences become syntactically denser. Background knowledge matters more. Students encounter words such as industrialization, constitutional, photosynthesis, socioeconomic, and metamorphosis rather than the vocabulary of early elementary texts.
At that point, reading fluency depends on an increasingly sophisticated interaction among decoding, orthographic knowledge, morphology, vocabulary, syntax, background knowledge, working memory, attention, and automaticity.
A student can compensate for weaknesses in some of those systems remarkably well when texts are simple.
Especially a bright child.
Context helps. Pictures help. Vocabulary helps. Reasoning helps. Memory helps. A child can predict a word from the sentence or recognize enough of it to keep moving. Strong comprehension can mask inefficient word reading for a surprisingly long time.
Then the text changes.
There are fewer contextual supports. The vocabulary becomes unfamiliar. Words contain multiple meaningful units. Sentences carry several ideas at once. Students aren't merely reading stories anymore; they're reading science, history, mathematics, primary sources, technical explanations, and abstract arguments.
Eventually compensation becomes expensive.
The child hasn't necessarily become a worse reader. The environment has finally exceeded the compensatory system's capacity.
That's one reason I would be very cautious about looking at exceptional third-grade reading proficiency and concluding that we have discovered an exceptional K–12 reading system.
We may have discovered an exceptional early-literacy system.
Those are not necessarily the same thing.
And that is where I think Daly's larger argument becomes even more important.
He argues that if Steubenville were a corporation outperforming its competitors for decades, researchers and consultants would have descended on it to figure out what it was doing. Instead, remarkably little rigorous investigation has been done.
I agree—but I would expand the investigation.
Don't just ask, "What is Steubenville doing that produces these extraordinary third-grade scores?"
Ask:
What happens to those readers afterward?
Follow the children longitudinally.
Look beneath proficiency.
Look at decoding and word recognition. Look at spelling. Look at orthographic mapping and automaticity. Look at morphological knowledge. Look at vocabulary and background knowledge. Look at oral versus silent reading. Look at comprehension when the subject matter is familiar and when it isn't. Look at reading rate and accuracy when students encounter unfamiliar multisyllabic words.
And perhaps most importantly, determine whether the children who were "proficient" in third grade developed the underlying reading architecture that lets them keep becoming more sophisticated readers without relying more and more on compensation.
Because proficiency is a threshold, it isn't a diagnosis of how a child reads.
That's the lesson my own family has learned the hard way.
When a capable child crosses the proficiency line, the system tends to stop asking questions. The score says the child can read. The child may even have excellent comprehension. Meanwhile, weaknesses in orthographic knowledge or automatic word recognition can remain largely invisible.
By middle school, those weaknesses aren't invisible anymore.
And once that happens, families can find themselves in the bizarre position of fighting to obtain foundational literacy instruction for a child everyone agrees is intelligent and who everyone has spent years describing as a successful reader.
That should make us reconsider what we mean by "success."
Daly writes that successful school systems should be studied with the intensity that businesses devote to unusually successful companies. Absolutely. But longitudinal erosion deserves the same curiosity as exceptional performance.
In fact, it may tell us more.
If Steubenville's early reading results withstand scrutiny, I don't want to know only why its third graders perform so well.
I want to know why so many of those third-grade readers apparently aren't still performing that way five years later.
What did they learn?
What didn't they learn?
What did the assessments measure?
What were students able to compensate for at eight or nine years old that they could no longer compensate for at thirteen or fourteen?
And what happened instructionally between those two points?
Those aren't questions that diminish Steubenville's accomplishments.
They're exactly the questions we should ask if we're serious about learning from them.
The goal shouldn't be to find a reading program that produces extraordinary third-grade scores.
It should be to understand how we create readers whose skills keep deepening as the intellectual demands of reading increase.
As a parent, that's the outcome I care about.
And I suspect that's where some of the most important information in the Steubenville story may ultimately be hiding.
I'm curious about Roma ISD in Texas. It's in the poorest county in Texas, with lots of emerging bilingual students and low education rates among parents, but the elementary schools get extremely high reading scores
https://www.ets.org/Media/Research/pdf/RM-24-06.pdf
To me it is not surprising when educational miracles are exposed. Rather, it is surprising that so many people initially accept them. https://www.skeptic.com/article/mississippi-miracle-smoke-and-mirrors-vs-reality/
No one needs to adjust NAEP reading scores for demographics Dan. It lets states like Texas--with high fiscal capacity but below average school funding-- off the hook. Mississippi, by contrast, would love to have Texas's capacity but doesn't, and maxes out what it can because governors going back to Haley Barbour see a strong connection between education and economic growth.
Mississippi has legitimately raised 4th grade reading achievement by a full grade level in the last decade (from 209 to 219, 2013-2024). They've always had high retention numbers too. Here's a more robust look at that data, written by a Californian with no skin in the game: https://fordhaminstitute.org/national/commentary/mississippi-didnt-cheat-its-reading-gains-are-real
Hi David. Not sure what you are saying. If Mississippi does not adjust scores for demographics, like poverty, the adjusted scores go down. They would no longer be number one for math and reading. I'm not defending Texas by any means. Just saying that a closer look at the Mississippi data reveals less than impressive results. Where are the reading gains for 8th grade and 11th grade (ACT)? Educational miracles never outlast data.