That means, when families decide where to live, one big factor is:

How good is the school in the neighborhood?

So how do you figure that out?

Well, you could visit the school...

See how the playground looks....

Or ask friends about it.

But all of these seem so imprecise.

Plus, it's kind of weird to visit dozens of playgrounds.

So, these days, you would probably Google the school.

And you'll almost certainly end up on this website:

GreatSchools.org. GreatSchools is a nonprofit school ratings website.

They provide a number rating for almost every public school in America.

The worst schools are a 1, and the best schools are a 10.

There are other websites and organizations that rate schools, but GreatSchools is the big one.

Well, we dug into the data to find out.

And what we found, is that this score often ends up showing something other than how good a school is.

GreatSchools started in the late 1990s.

They visited schools around the San Francisco Bay Area, and talked to each school's principal.

But that method of profiling schools wasn't exactly scalable.

They couldn't go visit every single school in the country.

That would be prohibitively expensive; it was hard enough as it was.

Matt Barnum is a reporter at Chalkbeat, a site that covers education.

He helped us report this story.

Going to every single school was just not a sustainable model.

Then, in 2001, something huge happened for GreatSchools.

Congress passed a national education law called "No Child Left Behind."

No Child Left Behind required schools to give state standardized tests,

each year, in grades three through eight, to figure out how many of them

met a standard called "proficiency."

"Proficiency" is basically each US state's standard for what kids should know at a certain grade level.

So that gave them a lot of data to rate schools all across the country.

After No Child Left Behind, GreatSchools had the numbers they could base their scores on.

Today, they say this proficiency score is the biggest factor

in the ratings it gives each elementary and middle school; accounting for about half the score.

But when we start looking at GreatSchools' data, we notice something.

Here are basically all the public elementary and middle schools in Denver and a few of its suburbs.

With the help of journalists at Chalkbeat, we sorted them by the percentage of

low-income students in each school.

These schools have more low-income students,

and these schools have fewer.

Now, let's highlight the schools that got a GreatSchools rating of at least a 7, which is what they consider above average.

You can see that almost all of these better-rated schools are more affluent.

And this is true in basically every city from San Francisco to Detroit.

And this correlation also appears when we sort these schools by racial demographics.

Here's Denver again. The schools on the left have more black and hispanic students,

and the schools on the right have fewer.

So which schools got a rating of 7 or above?

It's schools with more white and Asian kids, who tend to come from more affluent neighborhoods.

GreatSchools' ratings seem to confirm something that many people already assume:

Here's the problem: That's not necessarily true.

And to understand why, we need to look at one of the big issues with measuring proficiency.

America's neighborhoods are highly unequal.

And when children show up to school in affluent neighborhoods,

that are mostly white and Asian, they're better-prepared for school

than children in poor, mostly black and Hispanic areas.

So when we look at proficiency, we're actually often measuring how prepared these kids were coming in.

Not what happens inside the school.

So, if proficiency is more about measuring the student than the school,

how could we actually measure the school's performance?

Well, another common way of measuring schools is to see how much a school helps a student improve.

This is called a "growth score."

Now, Growth is still just based on test scores, which not everyone is a fan of.

But it's how many schools prefer we measure them.

The principals I talked to felt that it was more fair to judge them on growth.

If you're a school that is serving a lot of students who are coming in at a low level, you could

be doing a great job with those students, but it might not be showing up in proficiency.

Not all states report growth scores, which is a problem in itself. But most do.

And growth scores show that there are many low-income schools that are pretty good.

Here's every Denver school again.

And again, sorted by the percentage of low-income students.

And now, if we highlight the schools that have an above-average growth score,

which is data we got from GreatSchools' own website,

we can see that there are lots of schools in poor neighborhoods that are very good at educating their students.

And this is true in other cities, too: Like Indianapolis...

and Detroit.

But in the GreatSchools rating system, growth only matters about half as much as proficiency does.

Which is why these schools are rarely rated highly,

even if they do a great job teaching students.

We talked to GreatSchools. They said the important thing

is that they give parents a "broader picture of school quality,"

and that their site helps "underserved families" make good choices.

They also stressed that they've actually changed the rating system to

start considering growth, alongside proficiency.

And the data reflects this. But:

There's still this really stark correlation with race and class of

students, but it is lower than what it was before.

That's not entirely their fault. Not every state reports growth scores.

But even where growth scores are available, GreatSchools still weights proficiency a lot more.

And that muddles the difference between two definitions of "good" schools:

The ones that do the best job teaching students,

and the ones that get kids who are already high-achieving,

who tend to be from white and affluent families.

So maybe, the real question is:

Which one do parents actually want?