Does money buy test scores?

There’s a comfortable story where wealth and school achievement rise together: prosperous states fund better schools, families have more slack, and the scores follow. It’s a reasonable instinct — money really does buy a lot of things. I just wanted to see how much of it you can actually see in the data, so I lined up every state’s 2024 NAEP scores against its median household income, its GDP per capita, and its per-pupil school spending.

Short version: there is a relationship between how much a state’s households earn and how its kids score — but it’s a loose one, a correlation around 0.3 to 0.4, not the tight line the intuition promises. And the deeper you go, the more the interesting part turns out to be the states that break the pattern, not the ones that follow it.

01Yes, richer states do score higher — loosely

Start with the version of the claim that holds up. Plot each state’s median household income against its 2024 NAEP score and the cloud does tilt uphill: the wealthier end of the axis sits a little higher. For eighth-grade reading the correlation is +0.39; for fourth grade it’s a softer +0.28. Positive, real, and worth respecting — income is a genuine correlate of achievement.

But “loose” is the operative word. A correlation of 0.39 means income accounts for only about 15% of the differences between states; the other 85% is everything else. You can see it in the scatter below — the band is wide, and some of the most interesting states sit furthest from the trend line.

02It’s household income that matters — not “the economy,” and not spending

Here’s where the naive version starts to wobble. If wealth drove scores, you’d expect the broadest measure of a state’s economy — GDP per capita — to predict at least as well as household income. It doesn’t. GDP per capita is a weaker correlate (about +0.35 vs. +0.39 for reading), because energy and agriculture states like North Dakota, Nebraska, and Wyoming post big output per person that never shows up in the median family’s paycheck. What lands in households tracks scores better than what shows up in the state’s ledger.

And then the one everybody assumes is the lever — per-pupil spending — comes in essentially flat. Against fourth-grade math the correlation is actually slightly negative (−0.07); against reading it’s a weak +0.1 to +0.2. The lowest-spending states in the country, Utah and Idaho at roughly $10k a pupil, post some of the highest scores; the biggest spenders — New York, D.C., Vermont at two to three times that — sit mid-pack.

03The twist: adjust for cost of living and the income link gets stronger

A dollar isn’t a dollar. Ninety-nine thousand in Massachusetts and fifty-four thousand in Mississippi don’t buy the same life, so I divided every state’s income by its local price level (the BEA’s Regional Price Parities) to get real, purchasing-power income. I half-expected this to wash out the relationship — lift the low-cost South, drop the pricey coasts, and flatten everything.

The opposite happened. Adjusting for cost of living made the income–score correlation go up across the board — eighth-grade reading climbs from +0.39 to +0.46, and fourth-grade math from +0.17 to +0.26. Correcting for what money actually buys locally sharpens the signal rather than erasing it. Spending, adjusted the same way, stays stubbornly near zero. Here’s the full grid — every wealth measure against every test.

04So what actually moves the needle?

Not the checkbook — at least not at the coarse, state-to-state level this data can see. If income only explains a sixth of the gap and spending explains none of it, then five-sixths of what separates a top state from a bottom one lives in the residuals: curriculum, instruction, policy, demographics, how early kids learn to read. Mississippi is the poster child — near the bottom in income and spending, near the top in fourth-grade gains after a decade of hard phonics reform. The money didn’t change; what the state did with it did.

None of this means money doesn’t matter — real income clearly correlates, and no one runs a school on good intentions. It just means “richer states score better” is a weak rule with loud exceptions, and “spend more, score more” barely survives contact with fifty state data points. If you want to know why one state reads better than another, the budget line is where you stop looking, not where you start.

The tidy version — wealthy states read and count better — isn’t wrong so much as thin. Income nudges scores; local purchasing power nudges them a touch more; the school budget, the thing we most associate with “investing in education,” nudges them essentially not at all. Which is strangely hopeful: if dollars were destiny, a poor state would be stuck. Mississippi’s decade says otherwise.