Links and Commentary 10/17/25

Gaps in math test scores based on race and ethnicity as well as household income remain large but appear to be narrowing. For example, the gap between students at high-poverty schools and the national average in third-grade math shrank by about a third from spring 2021 to spring 2025. The gap for Hispanic students in third-grade math shrunk by about 41%, while the gap for Black students was about 27% smaller.

The data released Tuesday comes from MAP Growth Assessments given three times a year to more than 7 million students in grades three through eight in 20,000 schools.

These students attend schools that choose to give the MAP test, while students who take the National Assessment of Educational Progress, or NAEP, reflect a statistically representative sample. The MAP data comes from the 2024-25 school year, while the most recent NAEP results come from spring 2024. NWEA weighted the results for students from certain subgroups to ensure the results were more representative.

Education giant Houghton Mifflin Harcourt will acquire well-known K-12 assessment provider NWEA, in a pairing of two of the biggest and most influential companies in the nation’s schools.

NWEA will operate as a division of HMH, and its assessment solutions will be integrated with HMH curriculum

“With accelerating billings growth, strong free cash flow and a transformed cost structure, we are at an important inflection point, and the time is right to move into the next phase of our long-term growth strategy alongside a partner that brings significant industry expertise,” [CEO] Lynch said in a statement.

This analysis found that both low and high increases in social media use throughout early adolescence were significantly associated with lower performance in specific aspects of cognitive function, supporting a prior finding that greater screen time was negatively but weakly associated with adolescent cognitive performance.⁵ Overall, the differences in mean scores were small relative to the NIH Toolbox standard deviation of 15. Previous literature has hypothesized that social media use replacing more educational activities or schoolwork may explain the association between social media use and lower cognitive performance.⁶ The specific associations between increasing social media use and poorer performance on the ORRT and PVT, which tests stored language knowledge, support this hypothesis. The finding that even low levels of early adolescent social media exposure were linked to poorer cognitive performance may suggest support for stricter age restrictions. Limitations of this study include self-reported social media data, potential residual confounding, and an observational study design that precludes inferring that social media trajectories cause worse cognitive performance. Future studies should examine how specific social media platforms and content relate to cognitive outcomes.

When I first started teaching, I had the ability to take whichever curriculum was presented to me, and make it my own. I could choose how I wanted students to engage with the required content standards. Instead of reading an excerpt from a basal reader, looking for the main idea, I could have students read novels and discuss themes, character development, and other literary elements. In the past, I had more professional agency to determine how I wanted to engage my students in the content. I could design transdisciplinary units, addressing multiple standards simultaneously. Unfortunately, this is no longer an option. Our district is doubling-down on using standardized curricular resources and standardized pedagogy to create standardized learning cycles. This means teaching a single content standard to all students using identical Tier 1 instruction, using standardized, formative assessments to determine which students are meeting (or not) that standard. Then, regrouping students for standardized Tier 2 interventions and extension support. All teachers are now required to teach the same thing on the same day in the same way repeatedly for 15-day learning cycles until everyone has mastered that discrete standard.

I am not engaging my students’ creativity or critical thinking skills. I am not tapping into their schema and connecting the content with their personal identities. I am not encouraging them to think for themselves or wrestle with complex ideas. I am merely managing academic tasks. Students are not reading well-written stories or having meaningful, student-centered discussions. They are not building on each other’s ideas, supporting their views with evidence, or engaging in dialogue with their peers. Students are reading AI-generated texts, locating details, selecting main ideas, and writing CERs. What’s worse is that these 15-Day Challenge cycles are set to repeat continuously throughout the year. One 15-Day Challenge after another until teachers have “taught” every academic standard required by the state, all to increase test scores.

…when it comes to “reading” or literacy, a fundamental problem is that the predominant approach treats reading comprehension—and writing—as sets of skills that can be taught in the abstract. Rather than teaching students to mine texts for whatever meaning they might have, most schools try to use them as vehicles for teaching skills like making inferences, finding the main idea of a text, or thinking critically. These are the skills that standardized reading tests purport to measure.

Surely, you may be thinking, those skills are worth teaching. It’s impossible to provide kids with all the information they’ll need to know in the future. Anyway, they can always use Google or AI to get answers to factual questions. Why not teach them skills they’ll be able to apply for the rest of their lives?

The problem is that skills like making inferences aren’t transferable. Readers can apply them only if they have enough relevant knowledge. It’s easy to make an inference about a text on a topic you know a lot about—so easy you may not realize you’re doing it. And it might be impossible to make an inference if the text is on a topic you’ve never heard of. So the key to developing those comprehension abilities is developing knowledge, which—especially at the elementary level—most schools have failed to do.

Some nights, I lie awake wondering what my students will be like after high school or college, having spent their entire academic careers filling in prescribed boxes. How will they succeed in the world with no practice in asking questions or thinking for themselves? Will these high reading scores gain them easy entry into college? Once admitted, how can they be expected to evaluate information, question assumptions, and form independent conclusions rather than simply memorize facts?

Parents do not equate standardization and testing with rigor and quality. When public schools become focused on standardization and achieving high test scores, parents see this as bad and become more receptive to alternatives. Anti-school conservatives are there, ready to take your kid to Disney to learn fractions from Snow White on the taxpayer’s dime. And, it appears that parents are willing to take on the burden of being the executor and guarantor of their children’s education if it means they escape systems of testing and standards.

I have been making the case against algorithmic grading long before ChatGPT showed up. This piece dates from 2013 and was predicated by the enthusiasm of the developers of MOOCs who saw grading automation as a way to deliver these courses “at scale.”

Those attempts failed, as did MOOCs, when it comes to taking over the lion’s share of our post-secondary educational system, not only because of technical limitations, but because people recognized there is value in the human exchange of reading and writing. We should not give this thing of value over to a simulation, no matter how convincing that simulation may now have become.

It is an abandonment of our humanity for no discernible gain beyond efficiency. As I’ve written in this newsletter previously, efficiency is not a value we should associate with learning. Learning is rooted in connection, engagement, and productive friction.

The expansive voucher program also has strained the state budget, with costs projected to top $1 billion this year, far more than originally projected. Because many voucher students were already attending private schools, including many from wealthy families, funding them is a new expense for the state. That’s drained dollars, critics say, that could and should go to public education in a state that ranks 47th in spending on K-12 per-pupil spending.

emphasis added

John Ward, the head of Arizona’s ESA program, told me that the state hasn’t authorized any mandatory assessments for its ESA students; he doesn’t even necessarily know where they go to school. Horne, his boss, put the matter to me squarely: “The philosophy of the legislature when they passed universal ESAs was that the accountability will come from parents.” There are no enforceable curriculum standards—and this, Ward told me, is part of the appeal. Parents of children in microschools tend to see standardized tests as reflecting “a curriculum that has been chosen and essentially developed by public school systems,” he explained. “Because [these schools] are doing something different than what the public schools are doing, those tests wouldn’t measure what they’re doing in their curriculum. It would be a poor metric applied to what they actually do.”