Don't Blame the Subgroups

The NAEP achievement declines are widespread

As I’ve documented in separate pieces for The 74, the bottom has fallen out on American achievement scores in both math and reading. In 8th grade math, for example, from 2022-2024 the scores for the top 10% of students rose 3 points while the bottom 10% fell 5 points. The patterns are even uglier if you look out over the last decade, especially in certain states.

One pushback I’ve been hearing is that maybe these trends are being driven by changing demographics. For example, here’s Linda Darling-Hammond suggesting that the rise in English Learners and students with disabilities may be contributing to the sharp declines in the performance of low-performing students on NAEP:

Additionally, there has been an increase in the proportion of NAEP test-takers who are identified for special education and the share who are English learners. (The majority of English learners become proficient in English, at which point they are no longer counted in the EL group; so the EL category is, by definition, lower scoring.) From 2013 to 2024, the share of test-takers who are identified for special education has increased by 25% at the fourth-grade level and by 27% at the eighth-grade level. The share of test-takers who are categorized as English learners (i.e., not yet proficient in English) has doubled at the eighth-grade level (from 5% to 10%) and has increased by 30% at the fourth-grade level (from 10% to 13%). Most of that latter increase has occurred since 2019.

I find the percentage change numbers here distracting, but Darling-Hammond is correct that American schools are serving more students identified with special needs. It’s also true that English Learners and students with disabilities tend to score lower than non-English Learners and students without disabilities. So it’s fair to note that demographic changes are depressing the national averages.

And yet, demographic changes cannot fully explain our NAEP declines. Let’s start with just the simple averages. In 8th grade math, the scores for all EL students fell 8 points from 2013 to 2024. The scores for non-English Learners fell 9 points.

That’s interesting on its own. But it doesn’t show the full extent of what’s going on. The chart below breaks out the 8th grade math results for EL and non-EL students at different performance levels. As you can see on the left, the 90th percentile of both groups declined by a little bit, and higher-performing English Learners suffered a slightly bigger decline.

Now look at the right side of the graph. For both groups, the bottom 10% of students suffered particularly steep declines. And the bottom 10% of non-English Learners fell the farthest, a 16 point decline.

I’m too lazy to make this type of graph for every subgroup, but the same pattern holds across different student groups. Again, these are changes from 2013 to 2024 in 8th grade math, broken out by performance level:

SWDs

90th percentile: -5.8 pts

10th: -10.8 pts

Non-SWDs

90th percentile: -1.5 pts

10th: -17.4 pts

Low-Income Students

90th percentile: -5.6 pts

10th: -16.9 pts

Non-Low-Income Students

90th percentile: 0.0 pts

10th: -19.2 pts

Over at his Substack,

Tim Daly

documented the same type of pattern across parental education levels.

You can also see similar trends across racial categories:

Black Students

90th percentile: -6.5 pts

10th: -14.6 pts

Hispanic Students

90th percentile: -5.6 pts

10th: -18.4 pts

White Students

90th percentile: -1.8 pts

10th: -15.5 pts

What’s behind these trends? There are no simple explanations, and it’s probably due to several factors. I’ll have more to say on this later, but for now check out my article in The 74 unpacking a few leading theories, or check out Nat Malkus’ brief for AEI.