Hacker News Reader

Comment by pc86 | Hacker News Reader

It's not impressive if you don't understand statistics I guess.

You want to measure the impact of educational policies on students. So you control for other factors that influence educational outcomes, and you try to eliminate those effects.

Girls do better than boys on most testing. Asians do better than other races. Free & reduced lunch kids do worse. Special education and ESL students obviously do worse. So of course you're going to adjust for these populations.

The only reason you look at the raw scores of a rural district outside Biloxi that is 70% black and compare that to the raw scores of a suburban San Francisco district that is 50% asian and 35% white is that you either don't understand math or you're more interested in making a political argument than an educational policy one.

View on HN

Statistics can also be gamed, and this article reeks of it.

As other commenters are saying, Mississippi children's actual achievement rates are still bottom of the barrel. So while they might be statistically doing better, they're not actually doing better at the moment.

That's not to say that things can't turn around, or that this newfound investment into education might not bear fruit very soon.

But the OP made the choice not to address any of these issues; therefore, the article doesn't seem to have been made in good faith relative to the HN community's expectations.

As far as I can tell this article uses the exact same statistical analysis that every other educational achievement article in the world does. What are they doing that is statistically unsound? How does this article "reek" of it other than the fact that you don't seem to like the implication that a deeply conservative state is making positive strides in educational policy?

Mississippi's kids are doing 1.5 grade levels better in reading than they were 20 years ago. That's an objectively good thing. Maine's kids are doing 1.5 grade levels worse than they were 20 years ago. That's an objectively bad thing.

It makes sense to look at why.

But unless you control for it very, very carefully, you end up with results like "free and reduced lunch programs cause kids to do worse" - which isn't supported by the evidence, because there may be other factors that coincide with those who are eligible for them.

You'd have to do something like take the high-performing SF district and give half the students, randomly selected, free lunches.

Reduced-price lunches are a proxy for socioeconomic status. That it's a proxy means it isn't perfect but we're talking about the entire country you can absolutely draw conclusions from it based on the mountains of data we have available.

> unless you control for it very, very carefully, you end up with results like "free and reduced lunch programs cause kids to do worse"

Nonsense, you end up with results like "free and reduced lunch programs are correlated with kids who do worse", which they absolutely are!

Reply on news.ycombinator.com