🛝 Nobody's Checking the Homework

Welcome to Playground Post, a bi-weekly newsletter that keeps education innovators ahead of what's next.

This week's reality check: A five-university analysis of nearly 50,000 correlations just redrew the map of what actually predicts reading. The lesson marketplace used by 85% of American teachers is filling with AI-generated material nobody checked. And new Indiana data shows students lose ground when the person teaching math is anxious about math.

Data Gem

40.4% of 2024 research-doctorate recipients with employment plans headed to business or industry, up from 32.2% in 2014. The National Science Foundation is now spending $47 million over five years to train more than 250 doctoral students inside companies.

Not Every Reading Problem Is Dyslexia

A multi-university team led by Texas A&M analyzed nearly 50,000 correlations across 137 cognitive and achievement tests to answer one question: what actually predicts whether a child learns to read?

The answers break some assumptions. Visual processing, long a suspect in reading difficulty, played a minimal role. 

General intelligence mattered less than previous research suggested. 

The heavy lifters were auditory processing, vocabulary, and general knowledge, with reasoning, memory, and processing speed in supporting roles.

"Reading develops step by step, with each skill building on the one before it," said Daniel Hajovsky, the Texas A&M educational psychologist who led the study alongside researchers from four other universities. "Children make the most progress when adults understand and support the whole chain."

The chain is where current practice gets uncomfortable. 

Schools have spent years building screening and instruction around one famous break point. Hajovsky's data says the chain snaps in more places than that.

"Not all reading issues are dyslexia," he said. "If they have strong phonological processing and weak comprehension skills, phonics-based instruction is not necessarily going to help. It might require developing vocabulary and background knowledge."

He also flagged a working-memory connection: many late readers who eventually catch up have ADHD.

His suggested fixes - monitor comprehension closely, use graphic and semantic organizers, and have students generate and answer questions about what they read.

Strong phonics plus weak comprehension is not a phonics problem, and more phonics is not the fix.

For education innovators, this is a diagnostics market hiding inside a curriculum debate. Knowledge-building and vocabulary curricula get a fresh evidence base here, and screeners that incorporate auditory processing and memory can catch the struggling readers phonics-only tools miss.

85% of Teachers Shop Here. AI Slop Moved In

Teachers Pay Teachers, rebranded TPT in 2022, is education's answer to Etsy: an online marketplace where teachers sell lesson plans, activities, and classroom decor to other teachers. 

Its parent company says 85% of pre-K-12 educators in the US use it. 

The premise baked into the name is peer trust, materials made by teachers, for teachers.

Chalkbeat just reviewed what's on the shelves. 

A $4 poster represents George Crum, widely credited with inventing the potato chip, with a firefighter in a gas mask, above a caption explaining he "created the popular crispy snack when a customer is polated thinner fried potatees."

There was more. 

A $1.99 alphabet coloring worksheet missing the letter F. 

A $1.50 presidents poster illustrating Martin Van Buren's resilience with an asteroid hurtling toward anti-slavery demonstrators.

These are the telltale errors of AI-generated content that nobody vetted - and it breaks the marketplace's premise. 

"They don't have the time and energy to assess the quality," said Elon University professor Jeff Carpenter, describing teachers who see a highly rated resource and print it without really looking.

The sellers face the opposite incentive. 

Northern Arizona University's Catharyn Shelton notes the marketplace model rewards shops with thousands of pages of material, and AI makes that volume free. 

It's easy to slide into slop, she said, "because AI is so good at making things sound really good, even if the actual content or ideas aren't really there." TPT's parent company says algorithmic tools demote low-quality AI stores and that such products are a minuscule fraction of sales.

For education innovators, the trust layer is now the product. Automated curriculum QA that catches factual errors, and district-curated libraries that match TPT's convenience with actual vetting all attack the same gap.

Math Anxiety Isn't Just a Student Problem

Indiana asked its teachers how much they agreed with statements like "Of all the subjects, mathematics is the one I worry about teaching." 

Then researchers built a math-teaching anxiety score for over 3,000 educators teaching grades 3-8 and linked it to student test scores.

The distribution looks manageable at first: just over half of teachers were highly confident, nearly 40% middling, and 11% reported significant apprehension about teaching math. 

New teachers and early-grade educators were the least assured.

Anxious math teachers were not spread evenly. 

Students with disabilities, low-achieving students, and students from lower-income families were all more likely to be assigned one. 

And the anxiety showed up in results: an average student's scores dipped about one percentile with a less confident teacher, with girls, Asian American students, and English learners hit a bit harder. Across multiple anxious teachers in a row, the study suggests, the damage could compound.

The students who need the strongest math teaching are the most likely to sit in front of the least confident teachers.

Lead author NaYoung Hwang of the University of New Hampshire points to plausible mechanisms: less time on the subject, murkier explanations, teachers' own errors.

"When teachers are genuinely excited about the content, students tend to engage much more," she said. 

A National Council on Teacher Quality report backs the upstream diagnosis, finding most elementary prep programs don't spend enough time on core math content.

For education innovators, the study hands you the targeting mechanism for free: anxious teachers correctly self-diagnosed. Math-specific coaching and PD triggered by teacher self-report, staffing analytics that flag when struggling students are being concentrated with low-confidence teachers, and content-heavy early-career supports all follow directly.

⚡️More Quick Hits

This week in education:

• 22 states have overhauled math education since 2022 — Education Week's new tracker shows the reading-law playbook arriving in math: screening, intervention, teacher prep, and approved-materials mandates.

• West Virginia will retain 2,000 to 3,000 third graders this school year — a 2023 reading and math retention law takes full effect in 2026-27.

• Colorado's first Workforce Pell outcomes data is a "mixed bag" — wage returns vary widely across credentials as short as 150 clock hours, and the state had to name eligible programs by mid-August.

• 75% of educators expect student behavior to be a significant challenge this year — an EdWeek Research Center survey of 1,000+ educators also flags student engagement (70%) and budget uncertainty (42%).

To stay up-to-date on all things education innovation, visit us at playgroundpost.com.