I’m pretty confident the knee jerk skepticism about the “Mississippi Miracle” is the result of the immense psychic injury a lot of blue staters suffered from having to entertain the idea that a bunch of dumb hicks are doing something better than they are.
I frequently get this vibe, yes. There's this overt disgust at having to take Mississippi seriously and people are insulted at even comparing their school outcomes to Mississippi's, let alone to that comparison being favorable to Mississippi.
I think you might have this one backward. People rooting for this Mississippi Miracle to be valid are going to find themselves embarrassed. Have you considered that you are not looking at this from a neutral position?
I have looked at this exhaustively from tons of different angles, reading everything I can find arguing both for and against there being a real effect here, and looking myself at tons of NAEP data for evidence of selection of the population taking the test. If there's a rebuttal that explains the data, I will share it immediately, because I care about getting the right answer here. This rebuttal is really, really bad, and the people who published it should be embarrassed.
If the Mississippi Miracle is real, that’s good. Hoping it’s true is objectively the morally correct position to have. You understand that right? It means we have successfully identified a replicable educational intervention. No one who is excited about that prospect will ever have cause to be embarrassed.
On the other hand, anyone who’s rooting against the success of the state with the blackest population in the country should pause for a moment and rethink their priorities.
I had a college professor I know retort: "yeah, but they still have to live in Mississippi", which I think is exactly the kind of smug cope that's letting red states eat our lunch.
Viewing where to live through a blue state/red state lens is silly. You’re telling me you’d rather live in Spokane than Austin because of how everyone but your immediate neighbors votes?
I don't think that much thought went into it. I think it's just a way of avoiding having to think about how maybe maybe your brand of performative progressivism isn't the End of History, which would be uncomfortable.
That’s what I like about living in Madison. It’s a nice blue city in a state purple enough that there’s no danger of state policy going off the deep end to the left.
I also think that at least some of the psychic injury is the result of phonics being an integral part of the reforms. Lots of progressives really hate phonics.
At this point like 10% of my media consumption is Kelsey Piper pointing out that Mississippi did actually meaningfully improve education outcomes and Andy Masley pointing out that data centers don't actually use very much water. I'm awed at their persistence in the face of such poor epistemics on the part of their opponents and annoyed that this much effort is required to even put up a decent flight against pervasive motivated reasoning.
I genuinely dont understand the idea that retention amounts to selection bias in any way that a normal person would understand that concept.
Also it bears noting that if Mississippi’s policy is anything like Florida the retention level is basically they need to only be one year behind. My threadbare level 2s last year were basically proficient second grade students.
The few 1s I’ve had who can’t pass in summer school are literally in 3rd grade with early first grade or late k proficiency. They don’t know all the letters and sounds for them or have single digit percentile working memory or something else. What do standards even mean if we say they should pass like that.
I have a strong prior for claimed gains in education being artifactual but in this case, it does seem like Mississippi genuinely improved.
It can't be selection bias because the retention policy would merely shift the lower-performing students over by a few years max and the gains have been occurring for many years, and before the retention policy ever went into effect. It can't simply be that you're giving the poor students extra time or that you've optimized retention for the test because the gains have been occurring even in the students who don't get retained. You're not seeing any noticable jumps in the data, as would be the case if there were some huge change in methodology one year.
I don't know what explains all of the stats except that Mississippi made their students better readers. Maybe there's some weird combination of statistical artifacts going on making it seem this way when it isn't so but the simplest explanation at present is that the gains are real.
Hi Kelsey, I really enjoy your articles on child education. Please keep writing them. I'm not sure why people are trying to debunk the Mississippi Miracle theory. It seems pretty obvious that teaching kids the mechanics of phonically reading would be the best method to teach them. Further holding children back, especially in early grades, isn't a punishment; it's an opportunity for them to catch up.
I might have missed this in your reporting, but is there data on how much better children do on reading tests after being held back?
Do you or any other readers have suggestions on how parents can start teaching a 3 almost 4-year-old to start using phonics to read? Are there iPad apps I could use with him, or physical books/games?
Teach Your Child To Read in 100 Lessons is pretty good! We've tried a bunch of software programs (Starfall, Reading.com, Mentava) and some of them are fine but none of them are vastly better than any other. The core thing you want to do is just teach them the most common sounds made by each letter (watch a couple videos on how to do this, it's more helpful to kids to say 'b-' 'p-' than 'buh' and 'puh'. Once they've mastered that you want to model the skill of blending sounds together, and every app has a different approach to teaching this so you can just bounce around and see which one resonates with your kid.
I taught both my kids using the "teach your child to read in 100 easy lessons".
I agree it's great, and think every school should use it. Maybe 15 or 20 mins each day was all it took.
My daughter was reading Harry Potter by 3rd grade and was a high school level by 5th grade frequently reading the same books I read (I filter out anything with any adult content).
My son struggled a bit more (from stubbornness and not wanting to do the work). But we were persistent and now he's turning into a super reader as well.
I mostly agree that you have the stronger argument here but I think there are still real concerns.
1. I think the strongest argument against the ideas in your article is the strong pretrend. If Mississippi scores were going up before the reforms you describe, and then the reforms didn't change the trajectory, then it's hard (not impossible but hard) to attribute the results to the reforms.
2. Today, retention can absolutely lead to moving out of the data set in many contexts, given the significant expansion of non-public schools.
3. You are skeptical that kids passing the test might be retained for other reasons under the previous policy, but there are lots of possibilities -- retaining students for behavior issues, or for struggles with other aspects of the curriculum. Focusing retention on kids that fail the reading test could absolutely cause selection effects.
1). MS got significant mileage out of their NCLB reforms. 2013 was not the first step they took towards improving instruction, and I think was substantially an effort to standardize on stuff that was going well in 180 trial schools and that they were now confident enough in to take statewide. One reformer said in 2024 that 2013 "“t[ook] the Barksdale model that we had been vetting in those 180 schools, [and] took it to scale.”
2) Of course you should check private enrollment. Private school enrollment in MS fell over the time period in question.
3) For this to have the observed effect on the upper deciles they'd have had to be retaining students basically uniformly across reading ability levels. I absolutely don't buy this.
1. I think "Mississippi has a 20+ year history of successful incremental education reform" is a plausible read of the data but that's somewhat different from your original thesis in your article.
2/3. My point is that you made some weak claims in this specific post on these two points about retention -- it doesn't always turn kids into subsequent year test takers, and it could have changed composition even if the numbers stayed similar.
More generally, the other decile results suggest that Mississippi is doing something right, but I think it's more puzzling than you are giving it credit for. Both the general trends in education reform, and the specific reforms you discussed, should be primarily impactful for below-median students. After all, one of the stylized facts about the phonics debate is that the strongest kids learn reading easily under most methods, but better instruction is really needed for kids who struggle. Similarly retention should obviously not impact 90th percentile students. So I think it's clear there is something interesting going on here but I think there's less clarity yet on what.
One factor to note is that we're discussing three different tests.
- There's the federally administered NAEP that tries to construct a nationally representative sample. That's the one journalists and policymakers look at when they call miracles. (Apparently miracles only occur in 4th grade ELA NAEPs and never at any other age or in math?) Some subset of Mississippi students takes this test every couple of years.
- There's the Mississippi Academic Assessment Program (MAAP) for English Language Arts (ELA) test for grade 3. This is the standardized test Mississippi developed in-house and is the summative exam in April of the 3rd grade year that determines whether a kid is retained or promoted. Every single public school 3rd grader takes this test or the alternate (see below). Mississippi's initial report indicates that 22.7% of 3rd graders did not meet the requirement for promotion in 2025. [1] This is much bigger than figures reported in media. Kids who are retained receive a bunch of additional literacy interventions during the year they repeat. However, *retained students do not retake the MAAP ELA test for grade 3.*
- Which brings me to the third test that needs to be mentioned here, the MAAP-A. That extra A stands for "Alternate". Students who are retained take the MAAP-A, a different test aligned to standards originally designed for students with special needs. You can review the difference between the typical standards and the alternate standards to which the MAAP-A is aligned to get a sense of whether the alternate test is easier or not. [2] Simply put, retained kids are required to meet a lower standard for promotion the second or third time around.
I feel like this weakens the power of the retention itself but gives explanatory power to the treatments given to retained kids. It does seem like spending the time and effort to provide additional targeted services is worth it. The retention serves the purpose of identifying and making space for those services to occur (instead of remediation in 4th or whatever).
What I don't understand is how 1 in 5 third graders apparently fail to meet the legally required standard for promotion, but Kelsey and others report somewhere between 7 and 10% of kids are being retained in a given year. We seem to be missing ~10% of the kids who failed to meet the standard. Now, you'll notice in my link that this is the "Initial Summative Assessment Results" from May 2025 so maybe there is some work done over the summer to give those kids another chance? Waivers? I don't know. Looking at the detailed spreadsheet MSDOE provides doesn't make it any clearer because it says 75.6% earned a level 3 or higher, the threshold required for promotion. This means 24.4% of students scored below the threshold to be promoted. [3]
The more I look, the more confused I am. Does it mean there's fraud or that the miracle isn't real? No, it doesn't. I don't want to come off like a skeptic because the NAEP scores really are improving and I've adjusted my perspective in part because of Kelsey's great reporting. I just don't get how Mississippi is actually doing retention or where that missing portion of students is going.
The upshot is that 104 students failed, but only 19 are being retained for next year. Of the remaining, 80 got an exemption and 5 enrolled somewhere else (the latter is why I think Kelsey should be more concerned about enrollment changes). Exemptions are mostly for being an ELL or special education student, or for having been retained already.
Makes me wonder if we should even call these retention policies. Maybe another name is better. It's really just about identifying students for special services. I kind of hate how much attention is given to retention what that seems like the least important component.
You would still have to contend with why everyone else went in the other direction. Such that, sure, you could make an argument that MS isn't really doing anything amazingly well. But it is not that uncommon for someone that is simply showing up and doing valid work will excel when everyone else is making mistakes.
I think the evidence that Mississippi is doing _something_ well is very strong. I think the evidence that we know what in particular that is is quite a bit weaker.
Right, but isn't the main argument here that it is all the things? I'd even support adding that it is specifically not doing the things that were done in other places that is most important.
Notably, this argument doesn't say that retention can't be a factor. It probably is. The argument is that it isn't the only factor, though. Right?
The general claim is that Mississippi shows that we can do better teaching kids to read, and that other places should therefore do better by adopting those practices. But if we don't know why Mississippi is doing better it's hard to adopt the relevant practices. Similarly we have always knows that some teachers are more effective, but just knowing that doesn't help anything because "get good" isn't a reform.
Right, so this is rebutted by finding places that tried to replicate and failed. It is not advanced one way or the other by just insisting it isn't real.
Put a different way, I get the skepticism. But specific criticisms have to also be treated as a hypothesis to test. And if you can't back it up with a successful prediction, then what do you have?
If all you have is "I have my doubts." I mean, fine? But where do you take that?
Was wondering could schools have improved the retention policy annually (better identification?) but I don't know how many years you can do that effectively.
Great work you are doing here Kelsey, easily worth the cost of my subscription.
I do want to emphasize one other point. Having a clear retention policy almost certainly is effecting other students as well not just the ones held back. Knowing there are real consequences for not learning the material almost certainly is changing the behavior of teachers, parents and kids. Thus raising the scores of kids that were not held back.
IE even if the retention policy never held a single kid back, I would argue that it would still be boosting test scores. It's important to have standards and have consequences for failing to meet them.
Some NAEP data is public but the more detailed data generally requires institutional access. It would be amazing if an institution with access could get the data and run some of these comparisons. (And registering my guess: I predict Mississippi nine year olds have seen significant gains over the last two decades.)
"They could be explained if the low-performing students were excluded from the test."
This isn't true - the improvements at most percentiles are much larger than what we'd see if they were excluding the lowest scoring 10%. The by-percentile data is stronger evidence than you claim against the Wainer et al. argument.
The 50th percentile among the remaining students after removing the bottom 10% is the 55th percentile among all students (it is better than all of the excluded 10% and half of the remaining 90%). So, if the only change was that they started excluding the bottom 10% at some point, then we should expect the "50th percentile" now to match the 55th percentile from before the exclusion. In 2005, the 50th percentile score was 206 and 75th percentile was 230, so the 55th percentile was close to 210 or 211 (4 or 5 points above the 50th percentile, since there's almost a 1 point improvement per percentile in the 50th-75th percentile range, and it should be a bit smaller close to the middle of the distribution than it is farther out). "50th" percentile in 2024 was 222, which is +16 better than 2005 50th percentile and +11 or +12 better than 2005 55th percentile. So about 70% of the improvement since 2005 would be unaccounted for even if they had excluded the bottom 10%.
Excluding 10% could more than explain what the graph shows at the "10th percentile" (which would be 19th percentile considering the exclusion), but it could not explain the 2005 to 2024 improvement at the higher percentiles (25->32.5, 50->55, 75->77.5, 90->91). I suppose it could also explain the smaller improvements over shorter time periods at other percentiles, e.g. the 2015-2024 improvement at the 50th percentile (when scores improved from 217 to 222).
What allowed Mississippi (and other states) to -do- these reforms in the first place? Weak teacher's unions? Lack of activist parents? And, relatedly, is "rigorous standards for teachers and students" a kind of "secret sauce" that could extend these learning gains to other states and grade levels without a massive increase in public investment?
After replying to Sam Tobin-Hochstadt below, I wanted to post a mainline comment requesting clarification about Mississippi's retention policy.
According to the data Mississippi released in May of 2025 (link below), 22.7% of 3rd grade students failed to meet LBPA promotion requirements on this year's MAAP ELA test. Does this mean that more than 1 in 5 of Mississippi's 3rd graders were retained in 3rd grade for the 2025-2026 school year?
> To their credit, when I reached out to the authors about this mistake, they acknowledged it immediately and reached out to get it fixed before the article’s publication.
Before that article's publication or this article's publication? It looks like the mistake is still there - is the link to some kind of preprint? (I did notice the "lorum ipsum" text floating around.)
I’m pretty confident the knee jerk skepticism about the “Mississippi Miracle” is the result of the immense psychic injury a lot of blue staters suffered from having to entertain the idea that a bunch of dumb hicks are doing something better than they are.
I frequently get this vibe, yes. There's this overt disgust at having to take Mississippi seriously and people are insulted at even comparing their school outcomes to Mississippi's, let alone to that comparison being favorable to Mississippi.
I think you might have this one backward. People rooting for this Mississippi Miracle to be valid are going to find themselves embarrassed. Have you considered that you are not looking at this from a neutral position?
I have looked at this exhaustively from tons of different angles, reading everything I can find arguing both for and against there being a real effect here, and looking myself at tons of NAEP data for evidence of selection of the population taking the test. If there's a rebuttal that explains the data, I will share it immediately, because I care about getting the right answer here. This rebuttal is really, really bad, and the people who published it should be embarrassed.
If the Mississippi Miracle is real, that’s good. Hoping it’s true is objectively the morally correct position to have. You understand that right? It means we have successfully identified a replicable educational intervention. No one who is excited about that prospect will ever have cause to be embarrassed.
On the other hand, anyone who’s rooting against the success of the state with the blackest population in the country should pause for a moment and rethink their priorities.
Do you have any actual evidence underpinning your confidence that they'll be embarrassed?
Yeah, you might be too worried about getting schooled by a redneck if…
I had a college professor I know retort: "yeah, but they still have to live in Mississippi", which I think is exactly the kind of smug cope that's letting red states eat our lunch.
Viewing where to live through a blue state/red state lens is silly. You’re telling me you’d rather live in Spokane than Austin because of how everyone but your immediate neighbors votes?
I don't think that much thought went into it. I think it's just a way of avoiding having to think about how maybe maybe your brand of performative progressivism isn't the End of History, which would be uncomfortable.
That’s what I like about living in Madison. It’s a nice blue city in a state purple enough that there’s no danger of state policy going off the deep end to the left.
I also think that at least some of the psychic injury is the result of phonics being an integral part of the reforms. Lots of progressives really hate phonics.
At this point like 10% of my media consumption is Kelsey Piper pointing out that Mississippi did actually meaningfully improve education outcomes and Andy Masley pointing out that data centers don't actually use very much water. I'm awed at their persistence in the face of such poor epistemics on the part of their opponents and annoyed that this much effort is required to even put up a decent flight against pervasive motivated reasoning.
I genuinely dont understand the idea that retention amounts to selection bias in any way that a normal person would understand that concept.
Also it bears noting that if Mississippi’s policy is anything like Florida the retention level is basically they need to only be one year behind. My threadbare level 2s last year were basically proficient second grade students.
The few 1s I’ve had who can’t pass in summer school are literally in 3rd grade with early first grade or late k proficiency. They don’t know all the letters and sounds for them or have single digit percentile working memory or something else. What do standards even mean if we say they should pass like that.
> Jerusalem made me split this article in two because it was so long.
Boo!!! We want a 60,000 word Kelsey Piper essay that would make Scott Alexander weep!
This argument had me at “And what we see is that Mississippi has seen gains in every decile.”
I have a strong prior for claimed gains in education being artifactual but in this case, it does seem like Mississippi genuinely improved.
It can't be selection bias because the retention policy would merely shift the lower-performing students over by a few years max and the gains have been occurring for many years, and before the retention policy ever went into effect. It can't simply be that you're giving the poor students extra time or that you've optimized retention for the test because the gains have been occurring even in the students who don't get retained. You're not seeing any noticable jumps in the data, as would be the case if there were some huge change in methodology one year.
I don't know what explains all of the stats except that Mississippi made their students better readers. Maybe there's some weird combination of statistical artifacts going on making it seem this way when it isn't so but the simplest explanation at present is that the gains are real.
The pang I had when I saw a Kevin Drum chart. He would be blogging furiously about this. I miss him.
Hi Kelsey, I really enjoy your articles on child education. Please keep writing them. I'm not sure why people are trying to debunk the Mississippi Miracle theory. It seems pretty obvious that teaching kids the mechanics of phonically reading would be the best method to teach them. Further holding children back, especially in early grades, isn't a punishment; it's an opportunity for them to catch up.
I might have missed this in your reporting, but is there data on how much better children do on reading tests after being held back?
Do you or any other readers have suggestions on how parents can start teaching a 3 almost 4-year-old to start using phonics to read? Are there iPad apps I could use with him, or physical books/games?
Teach Your Child To Read in 100 Lessons is pretty good! We've tried a bunch of software programs (Starfall, Reading.com, Mentava) and some of them are fine but none of them are vastly better than any other. The core thing you want to do is just teach them the most common sounds made by each letter (watch a couple videos on how to do this, it's more helpful to kids to say 'b-' 'p-' than 'buh' and 'puh'. Once they've mastered that you want to model the skill of blending sounds together, and every app has a different approach to teaching this so you can just bounce around and see which one resonates with your kid.
I taught both my kids using the "teach your child to read in 100 easy lessons".
I agree it's great, and think every school should use it. Maybe 15 or 20 mins each day was all it took.
My daughter was reading Harry Potter by 3rd grade and was a high school level by 5th grade frequently reading the same books I read (I filter out anything with any adult content).
My son struggled a bit more (from stubbornness and not wanting to do the work). But we were persistent and now he's turning into a super reader as well.
Can you suggest a video on what you're talking about? My kid is starting to sound out short words, and I want to grease their path as best I can.
https://www.youtube.com/watch?v=vIN0HzzPWJY - this guy is great in general.
Thanks Kelsey!
I mostly agree that you have the stronger argument here but I think there are still real concerns.
1. I think the strongest argument against the ideas in your article is the strong pretrend. If Mississippi scores were going up before the reforms you describe, and then the reforms didn't change the trajectory, then it's hard (not impossible but hard) to attribute the results to the reforms.
2. Today, retention can absolutely lead to moving out of the data set in many contexts, given the significant expansion of non-public schools.
3. You are skeptical that kids passing the test might be retained for other reasons under the previous policy, but there are lots of possibilities -- retaining students for behavior issues, or for struggles with other aspects of the curriculum. Focusing retention on kids that fail the reading test could absolutely cause selection effects.
1). MS got significant mileage out of their NCLB reforms. 2013 was not the first step they took towards improving instruction, and I think was substantially an effort to standardize on stuff that was going well in 180 trial schools and that they were now confident enough in to take statewide. One reformer said in 2024 that 2013 "“t[ook] the Barksdale model that we had been vetting in those 180 schools, [and] took it to scale.”
2) Of course you should check private enrollment. Private school enrollment in MS fell over the time period in question.
3) For this to have the observed effect on the upper deciles they'd have had to be retaining students basically uniformly across reading ability levels. I absolutely don't buy this.
1. I think "Mississippi has a 20+ year history of successful incremental education reform" is a plausible read of the data but that's somewhat different from your original thesis in your article.
2/3. My point is that you made some weak claims in this specific post on these two points about retention -- it doesn't always turn kids into subsequent year test takers, and it could have changed composition even if the numbers stayed similar.
More generally, the other decile results suggest that Mississippi is doing something right, but I think it's more puzzling than you are giving it credit for. Both the general trends in education reform, and the specific reforms you discussed, should be primarily impactful for below-median students. After all, one of the stylized facts about the phonics debate is that the strongest kids learn reading easily under most methods, but better instruction is really needed for kids who struggle. Similarly retention should obviously not impact 90th percentile students. So I think it's clear there is something interesting going on here but I think there's less clarity yet on what.
One factor to note is that we're discussing three different tests.
- There's the federally administered NAEP that tries to construct a nationally representative sample. That's the one journalists and policymakers look at when they call miracles. (Apparently miracles only occur in 4th grade ELA NAEPs and never at any other age or in math?) Some subset of Mississippi students takes this test every couple of years.
- There's the Mississippi Academic Assessment Program (MAAP) for English Language Arts (ELA) test for grade 3. This is the standardized test Mississippi developed in-house and is the summative exam in April of the 3rd grade year that determines whether a kid is retained or promoted. Every single public school 3rd grader takes this test or the alternate (see below). Mississippi's initial report indicates that 22.7% of 3rd graders did not meet the requirement for promotion in 2025. [1] This is much bigger than figures reported in media. Kids who are retained receive a bunch of additional literacy interventions during the year they repeat. However, *retained students do not retake the MAAP ELA test for grade 3.*
- Which brings me to the third test that needs to be mentioned here, the MAAP-A. That extra A stands for "Alternate". Students who are retained take the MAAP-A, a different test aligned to standards originally designed for students with special needs. You can review the difference between the typical standards and the alternate standards to which the MAAP-A is aligned to get a sense of whether the alternate test is easier or not. [2] Simply put, retained kids are required to meet a lower standard for promotion the second or third time around.
I feel like this weakens the power of the retention itself but gives explanatory power to the treatments given to retained kids. It does seem like spending the time and effort to provide additional targeted services is worth it. The retention serves the purpose of identifying and making space for those services to occur (instead of remediation in 4th or whatever).
What I don't understand is how 1 in 5 third graders apparently fail to meet the legally required standard for promotion, but Kelsey and others report somewhere between 7 and 10% of kids are being retained in a given year. We seem to be missing ~10% of the kids who failed to meet the standard. Now, you'll notice in my link that this is the "Initial Summative Assessment Results" from May 2025 so maybe there is some work done over the summer to give those kids another chance? Waivers? I don't know. Looking at the detailed spreadsheet MSDOE provides doesn't make it any clearer because it says 75.6% earned a level 3 or higher, the threshold required for promotion. This means 24.4% of students scored below the threshold to be promoted. [3]
The more I look, the more confused I am. Does it mean there's fraud or that the miracle isn't real? No, it doesn't. I don't want to come off like a skeptic because the NAEP scores really are improving and I've adjusted my perspective in part because of Kelsey's great reporting. I just don't get how Mississippi is actually doing retention or where that missing portion of students is going.
[1] https://mdek12.org/wp-content/uploads/sites/33/2025/05/3RD_GRADE_MAAP_ELA_RESULTS_2025.pdf
[2] https://mdek12.org/sites/default/files/Offices/MDE/OAE/OSE/Info-and-Publications/2019_ms_aaas_ela_7_18_19_final.pdf
[3] https://view.officeapps.live.com/op/view.aspx?src=https%3A%2F%2Fmdek12.org%2Fwp-content%2Fuploads%2Fsites%2F33%2F2025%2F08%2F2025_maap_mediafile_ELA_MATH.xlsx&wdOrigin=BROWSELINK
This is actually something that I just read about (for my state, not MS). Indiana just passed a similar retention law to MS, and the first class was just tested. You can read about it here https://www.idsnews.com/article/2025/12/mccsc-third-grade-reading-rates-retention-bloomington-indiana-education-news
The upshot is that 104 students failed, but only 19 are being retained for next year. Of the remaining, 80 got an exemption and 5 enrolled somewhere else (the latter is why I think Kelsey should be more concerned about enrollment changes). Exemptions are mostly for being an ELL or special education student, or for having been retained already.
Makes me wonder if we should even call these retention policies. Maybe another name is better. It's really just about identifying students for special services. I kind of hate how much attention is given to retention what that seems like the least important component.
You would still have to contend with why everyone else went in the other direction. Such that, sure, you could make an argument that MS isn't really doing anything amazingly well. But it is not that uncommon for someone that is simply showing up and doing valid work will excel when everyone else is making mistakes.
I think the evidence that Mississippi is doing _something_ well is very strong. I think the evidence that we know what in particular that is is quite a bit weaker.
Right, but isn't the main argument here that it is all the things? I'd even support adding that it is specifically not doing the things that were done in other places that is most important.
Notably, this argument doesn't say that retention can't be a factor. It probably is. The argument is that it isn't the only factor, though. Right?
The general claim is that Mississippi shows that we can do better teaching kids to read, and that other places should therefore do better by adopting those practices. But if we don't know why Mississippi is doing better it's hard to adopt the relevant practices. Similarly we have always knows that some teachers are more effective, but just knowing that doesn't help anything because "get good" isn't a reform.
Right, so this is rebutted by finding places that tried to replicate and failed. It is not advanced one way or the other by just insisting it isn't real.
Put a different way, I get the skepticism. But specific criticisms have to also be treated as a hypothesis to test. And if you can't back it up with a successful prediction, then what do you have?
If all you have is "I have my doubts." I mean, fine? But where do you take that?
Was wondering could schools have improved the retention policy annually (better identification?) but I don't know how many years you can do that effectively.
Great work you are doing here Kelsey, easily worth the cost of my subscription.
I do want to emphasize one other point. Having a clear retention policy almost certainly is effecting other students as well not just the ones held back. Knowing there are real consequences for not learning the material almost certainly is changing the behavior of teachers, parents and kids. Thus raising the scores of kids that were not held back.
IE even if the retention policy never held a single kid back, I would argue that it would still be boosting test scores. It's important to have standards and have consequences for failing to meet them.
It would be nice if we could look at performance numbers by age instead of by grade, to bypass the retention question entirely.
Some NAEP data is public but the more detailed data generally requires institutional access. It would be amazing if an institution with access could get the data and run some of these comparisons. (And registering my guess: I predict Mississippi nine year olds have seen significant gains over the last two decades.)
"They could be explained if the low-performing students were excluded from the test."
This isn't true - the improvements at most percentiles are much larger than what we'd see if they were excluding the lowest scoring 10%. The by-percentile data is stronger evidence than you claim against the Wainer et al. argument.
The 50th percentile among the remaining students after removing the bottom 10% is the 55th percentile among all students (it is better than all of the excluded 10% and half of the remaining 90%). So, if the only change was that they started excluding the bottom 10% at some point, then we should expect the "50th percentile" now to match the 55th percentile from before the exclusion. In 2005, the 50th percentile score was 206 and 75th percentile was 230, so the 55th percentile was close to 210 or 211 (4 or 5 points above the 50th percentile, since there's almost a 1 point improvement per percentile in the 50th-75th percentile range, and it should be a bit smaller close to the middle of the distribution than it is farther out). "50th" percentile in 2024 was 222, which is +16 better than 2005 50th percentile and +11 or +12 better than 2005 55th percentile. So about 70% of the improvement since 2005 would be unaccounted for even if they had excluded the bottom 10%.
Excluding 10% could more than explain what the graph shows at the "10th percentile" (which would be 19th percentile considering the exclusion), but it could not explain the 2005 to 2024 improvement at the higher percentiles (25->32.5, 50->55, 75->77.5, 90->91). I suppose it could also explain the smaller improvements over shorter time periods at other percentiles, e.g. the 2015-2024 improvement at the 50th percentile (when scores improved from 217 to 222).
What allowed Mississippi (and other states) to -do- these reforms in the first place? Weak teacher's unions? Lack of activist parents? And, relatedly, is "rigorous standards for teachers and students" a kind of "secret sauce" that could extend these learning gains to other states and grade levels without a massive increase in public investment?
After replying to Sam Tobin-Hochstadt below, I wanted to post a mainline comment requesting clarification about Mississippi's retention policy.
According to the data Mississippi released in May of 2025 (link below), 22.7% of 3rd grade students failed to meet LBPA promotion requirements on this year's MAAP ELA test. Does this mean that more than 1 in 5 of Mississippi's 3rd graders were retained in 3rd grade for the 2025-2026 school year?
https://mdek12.org/wp-content/uploads/sites/33/2025/05/3RD_GRADE_MAAP_ELA_RESULTS_2025.pdf
> To their credit, when I reached out to the authors about this mistake, they acknowledged it immediately and reached out to get it fixed before the article’s publication.
Before that article's publication or this article's publication? It looks like the mistake is still there - is the link to some kind of preprint? (I did notice the "lorum ipsum" text floating around.)
The link is to a preprint, yes. The magazine article I believe will come out in January.
I was truncated in preschool