What’s the right way to judge a school?
Across the country, states and school districts have devised their own systems of letter grades and color-coded dashboards based on test scores and graduation rates. But arguably the most visible and influential school rating system in America comes from the nonprofit GreatSchools, whose 1-10 ratings appear in home listings on national real estate websites Zillow, Realtor.com, and Redfin. Forty-three million people visited GreatSchools’ site in 2018, the organization says; Zillow and its affiliated sites count more than 150 million unique visitors per month.
GreatSchools’ stated mission is to help all parents, especially those who are low-income, make more informed decisions about where to send their children to school.
But GreatSchools ratings effectively penalize schools that serve largely low-income students and those serving largely black and Hispanic students, generally giving them significantly lower ratings than schools serving more affluent and more white and Asian students, a Chalkbeat analysis found.
And yet, according to GreatSchools’ own data, many schools serving low-income, black, and Hispanic populations are doing a good job helping students learn math and English. But those schools still face long odds of getting an above-average rating on GreatSchools — likely because their students are arriving far behind.
The result is a ubiquitous, privately run school ratings system that is steering people toward whiter, more affluent schools. A recent preliminary study found that as the site rolled out an earlier version of its ratings, areas with highly rated schools saw increases in home prices and rises in the number of white, Asian, and better-educated families. After three years, the study found, property values in those areas increased by nearly $7,000, making it more difficult for low-income families to buy into the areas.
As parents decide on where to live, “these scores really were on the top of the list of almost all of the parents that I talked to,” said Sharique Hasan, the Duke researcher who authored the study of GreatSchools. “They wanted to be in a school district with schools that were rated a 9 or a 10.”
GreatSchools revamped its ratings in 2017. Chalkbeat’s analysis shows that the correlation with demographics has declined somewhat as a result of the shift — but is still substantial.
GreatSchools rejects the notion that its ratings contribute to segregation. “On the contrary, we believe that information drives equity. Parents, especially low income parents, deserve to know how their schools are doing,” said its CEO Jon Deane. “There are real issues at play here, but keeping parents informed, so that they can act on behalf of their children, isn’t the issue.”
Here is the paradox at the heart of GreatSchools ratings: They are available to all, which means they can help low-income families choose a school or pressure officials to make improvements. But they can also help affluent families cement access to areas other families cannot afford, while bolstering stereotypes that schools in certain neighborhoods are uniformly of poor quality.
“I think GreatSchools has a lot of responsibility to think deeply about how their platform is being utilized or not in relation to patterns of segregation,” said Francis Pearman, a Stanford professor who has studied the relationship between school choice and housing. “The notion that a quality school is necessarily one that is white and affluent is a problematic stance.”
Watch a video produced in partnership with Vox based on the reporting in this story.
Two decades in, GreatSchools rates public schools in every state
When Bill Jackson started GreatSchools in 1996 as a project of a Silicon Valley business association, he personally visited schools in the Bay Area and talked to principals about how their schools worked.
“We wanted to build a public appetite for a view of school quality based on a school’s own conception of what’s important,” Jackson told Education Week in 1999, a year after the organization became an independent nonprofit.
As GreatSchools expanded, first across California and then nationwide, that approach proved unsustainable. Fortuitously, the organization gained a new source of information: test scores for nearly every public school in the country, thanks to new requirements in federal education law. “The passage of the No Child Left Behind Act put a lot of wind in the sails of GreatSchools,” Jackson, who served as CEO until 2016, says now.
Around the same time, charter schools and other school choice policies were gaining traction. More choices meant information about school quality was more valuable — and sparked concerns that such information might be most accessible to well-connected parents.
By democratizing that information, which otherwise might have lived only in unwieldy spreadsheets on government websites, “GreatSchools had intentions to provide some equity,” said Janelle Scott, an education professor at the University of California, Berkeley.
GreatSchools also partnered with real-estate websites, starting with Zillow in 2012, to embed its ratings, since schools are often crucial in homebuying decisions. GreatSchools, a nonprofit, draws a licensing fee from those partnerships, which accounts for less than 20% of its revenue, according to a spokesperson. Most of its funding comes from philanthropic dollars, including foundations supportive of school choice.
The Walton Family Foundation, for instance, has poured nearly $25 million into GreatSchools since 2004, citing the need to help low-income families choose among schools and advocate for better ones. (Walton is also a funder of Chalkbeat.)
At the same time as GreatSchools was growing, all the new data from state tests had kicked off debates across the country about how to use it. Proficiency rates don’t take into account where students started, or how much schools helped them learn. Researchers developed ways to isolate the “value added” by a school, but those metrics proved controversial too, because they also rely on test scores, can be difficult to understand, and may bounce around from year to year.
Jackson recalled experimenting with different approaches, including directly accounting for poverty. But this didn’t resonate with families: “I’ll never forget one parent in Redwood City when we did a focus group who said, ‘So you mean this is a good school for poor kids?’”
So proficiency rates would remain the primary ingredient in the GreatSchools formula — which meant most schools serving students from low-income families would continue to rate poorly.
Jackson, though, said the organization did try to ensure the ratings got into the hands of less-affluent families. In addition to real estate websites, GreatSchools ratings are embedded on GoSection8, a site for people using federal housing vouchers. GreatSchools says that its own website is fully accessible in Spanish, is written in an easy-to-understand manner, and allows for clear comparisons among nearby schools. Some of this came on the advice of Innovate Public Schools, an advocacy group that works with low-income families and has supported the growth of charter schools.
GreatSchools also sees itself as a tool for those parents to demand better from their schools and their school systems. Jemima Hernandez, a San Jose parent, uses GreatSchools ratings to recruit parents as a volunteer for Innovate.
When families see the often-low ratings their children’s schools get, “They can’t believe it,” she said. “Then they start to talk to you.”
But GreatSchools was dogged by criticism that its scores were crude measures of school quality that penalized schools for serving low-income students.
When the organization overhauled its ratings in 2017, it included a host of new metrics. A GreatSchools representative said at the time that the new ratings would “more accurately reflect what’s going on in a school besides just its demographics.”
It was a striking acknowledgement of the flaws in the prior system. The new ratings also marked the start of a new approach for GreatSchools.
Across several cities, GreatSchools ratings give schools with whiter, wealthier students a clear leg up
Two years into this new system, Chalkbeat took a closer look.
We examined the ratings of elementary and middle schools in Chicago, Denver, Detroit, Indianapolis, Nashville, New York City, Phoenix, and San Francisco, combined with several of each city’s suburbs.
The results are striking. On average, the more black and Hispanic students a school enrolled, and the more low-income students it served, the lower its rating.
The average 1-10 GreatSchools rating for schools with the most low-income and most black and Hispanic students is 4 to 6 points lower than the average score for schools with the fewest black and Hispanic students and fewest low-income students.
In most places, only a tiny fraction of schools with the most low-income and most black and Hispanic students score a 7 or better, the number that earns an “above average” label from GreatSchools. (New York City was a notable exception. There, a relatively large number of predominantly black and Hispanic schools — particularly charter schools and especially Success Academy schools — earned above-average ratings.)
Chalkbeat focused on metro areas because families tend to choose schools and homes locally. GreatSchools argues that only statewide comparisons are appropriate; in six states where GreatSchools provided data, a substantial connection between student poverty and school ratings remained — though in some cases it was smaller than the connection in the metro area Chalkbeat focused on in the same state.
Chalkbeat’s analysis also confirmed that the ratings’ connection with demographics is weaker than it was a few years ago, thanks to new factors included in the ratings. The most notable addition, for schools in states that calculate it, is test score “growth,” or how much students progress in a year compared to others who started at a similar level. Most researchers say growth scores are a more accurate way to measure a school’s performance, and accounting for growth is also where schools serving students who arrive behind can shine.
“The ability to look at growth, college readiness, equity — these subratings that we have — really helped elevate schools that have higher [low-income] populations,” said Deane, GreatSchools’ CEO.
Differences by race and income persist, though, and they are quite large. Probably in part because many schools serving low-income students of color are genuinely struggling. But it remains difficult even for seemingly successful schools serving many low-income black and Hispanic students to score well, too.
Take Denver’s Knapp Elementary School, where most students are Hispanic and come from low-income families. Educators there describe Knapp as a collaborative place where they are respected and asked to push themselves to help students, nearly two thirds of whom are learning English as a second language.
Knapp has won statewide accolades for its ability to help its students improve on state tests. By fifth grade, Knapp’s share of students reading on grade level is higher than the district’s.
Principal Shane Knight, who is in his seventh year at Knapp, is proud of the school’s environment and its academic success. “If you come into Knapp, you feel that level of genuine care for kids,” he said.
Under GreatSchools’ formula, Knapp is rewarded for its students’ growth, earning a 9 out of 10 in that category. But overall, Knapp still earns a 4 — and a “below average” label.
On real estate websites featuring GreatSchools, users encounter the overall score, not the details. And if you type Knapp’s ZIP code into GreatSchools’ search bar, the school won’t even appear initially, because the site prioritizes schools with higher overall ratings. The schools that do show up are in a different Denver ZIP code and in most cases have whiter, more affluent student bodies.
Knapp fares poorly on two other, less widely used third-party rating sites too. SchoolDigger gives the school 2 of 5 stars, while Niche awards it a C. Those sites don’t consider student growth at all.
The low ratings frustrate Knight, who worries families who might help integrate the school will overlook it.
“Are you going to choose a house in the neighborhood where Knapp is? Maybe if the rating is good, but if the rating’s poor, I’m going to look to a different neighborhood,” he said.
Why is Knapp’s rating mediocre? GreatSchools’ formula.
Growth, where Knapp scored a 9, counts for only about 25% of its overall rating. Proficiency rates remain the biggest factor: 45%. There, Knapp earns a 3, because most of its students don’t meet the proficiency bar on state tests.
The rest of the formula isn’t even based on Knapp’s students. That’s the “equity” score, which is supposed to gauge test score gaps within a school. But Knapp is too homogenous — racially and economically segregated — for an equity rating to be calculated, GreatSchools explains. So the site assigns it the average equity rating of other schools in the state with similar proficiency rates: a 2, which counts for nearly 30% of Knapp’s overall rating. (Carrie Goux, a spokesperson for GreatSchools, said about one in 10 schools nationally receive such adjusted equity ratings.)
In a sense, GreatSchools penalizes Knapp for being segregated, even as GreatSchools ratings could make it more difficult for the school to attract a diverse student body.
Knapp is not alone in the Denver area. Among 88 schools that serve predominantly low-income, black and Hispanic students, just one scored above average on GreatSchools’ overall rating; looking at growth alone, 25 of them scored above average. That likely reflects the fact that low-income students of color arrive to school with lower achievement levels due to numerous factors, including the effects of poverty and racism.
Knapp’s scores highlight the challenges that high-poverty schools face in attaining a high rating, but also GreatSchools’ difficult balancing act. Some argue that proficiency should be a major factor in school ratings, as it is now in the GreatSchools formula, to hold all students to high standards.
Goux, the GreatSchools spokesperson, said the organization works to ensure all facets of school performance are visible. “We have displayed the sub-ratings prominently so parents can see how schools are doing across several important measures,” she said, pointing to data showing that these sub-ratings are widely viewed on GreatSchools’ site. “We know parents engage with the sub-ratings and are looking past the Summary Rating to understand schools on multiple levels.”
Goux also said that school searches can be easily adjusted across a number of dimensions — including distance and academic growth — to match a user’s interests.
Finally, she pointed to six schools in other cities that have a high share of low-income students and also earn high ratings. Most of these examples, though, are atypical in that they require at least some students to perform at a certain academic level to enter or remain enrolled.
Tomas Monarrez, an education researcher at the Urban Institute, says GreatSchools’ approach doesn’t provide an accurate picture of how a school is performing. The enduring connection between poverty and GreatSchools ratings in Denver and beyond “is evidence that these measures of school quality are contaminated.”
“Are these schools good because they are actually providing value for these kids?” he said of highly rated schools. “Or are these schools just enrolling more privileged populations?”
GreatSchools now — and in the future
What GreatSchools shows users matters, because millions encounter the ratings each year in their search for homes and schools, both on GreatSchools’ site and on real-estate sites.
Alex Robinson is among them.
She used the website to pick a new school for her son, a second grader; it helped her realize that a charter school she hadn’t considered was a short drive from her home in South Los Angeles. The school was rated a 5, whereas many other schools in the area scored only a 1, she said, and her son is doing well after the switch.
“GreatSchools ended up being my best resource,” said Robinson.
Other research has shown that GreatSchools ratings can affect where families who receive housing vouchers choose to live — an indicator that low-income families are using the scores to pick schools, too. Families offered GreatSchools ratings ended up in areas where schools earned an average rating of 4, as opposed to 3.7, a modest but real improvement.
The study, and Robinson’s story, are victories for GreatSchools, which argues that it offers a nuanced overview of school quality that is accessible to all families.
Still, labeling schools in poor communities and communities of color as bad could make it even tougher to create integrated schools, which research has found benefit students. “Segregation operates through stigma,” said Genevieve Siegel-Hawley, a professor at Virginia Commonwealth University who studies race and education.
Plus, there’s the research showing that the expansion of GreatSchools across the country has made residential segregation somewhat worse. (Notably, though, that study was done prior to the ratings shift in 2017, and there does not appear to be any research that looks specifically at GreatSchools’ effect on school segregation.)
“We show that broader access to information increased segregation because high-income families could more readily leverage school ratings to move to neighborhoods with better schools,” wrote researchers Sharique Hasan and Anuj Kumar. “In this case, knowledge was indeed power, but only for the powerful.”
Of course, there are many other powerful, longstanding forces in the way of more integrated schools.
A primary one is entrenched residential segregation, the product of government policies that have restricted who can live where and who can accumulate wealth. Potential changes to enrollment rules and school zone lines that would integrate schools often prompt backlash from more advantaged parents. Racism and bias may stop white families from choosing schools with more students of color.
While the Fair Housing Act forbids real estate agents from steering potential homebuyers of different races to different neighborhoods, a recent Newsday investigation found that many agents on Long Island were doing just that, sometimes using schools as a vehicle to do so.
These issues long predate GreatSchools. Still, some lawyers consulted by Chalkbeat said the fact that the GreatSchools ratings are so correlated with race raises questions about whether their use by real estate companies could run afoul of the Fair Housing Act, because the ratings effectively nudge families away from communities with more black and Hispanic residents.
“It’s a legal gray area,” said Morgan Williams, the general counsel for the National Fair Housing Alliance. “I would discourage real estate service providers from providing information that may be used as a proxy for the racial demographics of a neighborhood.”
(Spokespeople for both Zillow and GreatSchools said their respective sites encourage families to look at multiple factors and visit schools in person; a spokesperson for Realtor.com said the ratings are among many data points it offers to homebuyers.)
Yet even some critical of GreatSchools say it would not serve parents to eliminate or obscure data about school performance.
“Making that information easier to attain enables [parents] to isolate themselves more if they wish to do so, which is worrisome,” Monarrez of the Urban Institute said. “But I would also be worried about taking that possibility and saying that’s a reason for information not be put out there at all.”
Could GreatSchools do things differently? It could put more weight on growth. There are some signs that families would choose more integrated schools and districts if offered that information.
Myron Long, who is starting a charter school in Washington, D.C., says emphasizing growth measures can shift widely held perceptions of schools serving students of color. “It changes the narrative behind schools and how we talk about them,” he said. “It has the opportunity for parents to see there are more options in the schools that they choose.”
But that approach has its own drawbacks. Several states, including California, don’t calculate growth scores, impeding efforts to make ratings depend on them more. And encouraging more affluent families to go to high-performing schools in low-income neighborhoods could have unintended consequences, such as displacing other students.
Meanwhile, any changes could anger affluent communities that have benefited through property values from the way the ratings are currently calculated. Changes could also encourage skepticism among those with deeply ingrained notions, perhaps biased ones, about what is and isn’t a quality school. That could mean GreatSchools loses credibility with a powerful constituency of families.
For its part, GreatSchools says it remains engaged with questions about how to measure school quality.
Last month, dozens of researchers and policymakers gathered at MIT to talk about school choice. Several speakers made the same point: Judging a school’s quality by looking at overall student performance is misleading.
Usually, school ratings say more about peers than schools, MIT Professor Parag Pathak told the audience. “That may actually end up further segregating schools,” he said.
“Peer ‘quality’ is not school quality,” Josh Angrist of MIT explained later.
Watching from the audience was Deane, GreatSchools’ CEO. His takeaway? “A strong reinforcement” of data that shows how much students learn over time, he told Chalkbeat.
“The question is, are we moving in the right direction towards providing the information that is going to give us a broader picture?” he said. “We feel an incredible responsibility to provide that information to the users as best we can. So of course we’re going to continue to evolve.”
Watch the video produced in partnership with Vox based on the reporting in this story.
Curious about how Chalkbeat did the data analysis for this story? We’ve got more details, plus answers to frequently asked questions about our findings and approach.