A Fair Comparison
How is ABC Elementary doing compared to XYZ Elementary across town, or JKL Elementary in that district across the river?
Schools, districts and even classrooms are stacked up against each other all the time. During meetings. In the media. At the supermarket. Yet, "very few comparisons are actually appropriate," says Richard J. Wenning, director of operations and accountability at the Colorado League of Charter Schools.
Apples-to-oranges analyses mean little to nothing. While they can be relatively harmless in conversation, decisions based on them spell educational disaster. But researchers now have access to a tool that makes apples-to-apples benchmarking more possible and extensive than ever before.
The Growth Research Database, a repository of longitudinal academic growth and demographic data collected from Northwest Evaluation Association member districts in most states, is the machine that can make these comparisons happen. With data from nearly 1,500 school districts and other educational agencies, the GRD can help in pinpointing educational trends and patterns at any level--the individual student, classroom, district, state, region or nation.
This is something to celebrate, researchers say. "A lot of studies out there are based on a small number of schools, a small number of teachers or principals. We need a higher degree of statistical reliability," says Kenneth Wong, director of the National Research and Development Center of School Choice, Competition, and Student Achievement, based at Vanderbilt University in Nashville. With student-level data, studies don't usually account for variability among peers, adds Wong, who is using GRD data for a current project.
Rather than a focus on students meeting No Child Left Behind or other standards, the GRD offers a look at "the growth kids make in various kinds of classrooms, schools, from one year to the next," explains Peter Hendrickson, assessment manager of Evergreen School District in Vancouver, Wash.
The quantity and long-term tracking of data are what Michael Flicek of NWEA member Natrona County School District in Casper, Wy., says makes the database different. It's got the "potential to answer questions you really couldn't begin to answer in the past," notes Flicek, the district's director of assessment and research.
He serves, along with Hendrickson, on an NWEA advisory committee formed to help the nonprofit develop its database. The committee is considering questions that might be answered with the GRD and helping to establish its research agenda. For example, a district might think it's serving free and reduced lunch students in Title I schools well, but how does that progress compare to other, similar schools?
"There are a lot of questions that are in common for all districts," Flicek says. "Those are the kinds of things that will be front and center."
While the database was first announced in 2003, it's just gaining momentum now. Brian Bontempo, the initial project manager and a current committee member, says he has seen interest grow from a few researchers showing up at a members seminar, to, two years later, "a packed house" of all kinds of administrators and researchers seeking information on how they could begin conducting program evaluations and research using the GRD.
A Packed Data House
The current lineup in the Growth Research Database reads like a shopping list of items that data-hungry administrators won't want to pass up:
Individual student results from NWEA Achievement Levels Tests and Measures of Academic Progress, two achievement scale assessments based on state standards and designed to be highly reliable measures
Student demographic information such as ethnicity, socioeconomic status and specific educational programs in which the student is involved
School data such as faculty size and composition, class size, grade level and school-wide educational programs
Census data such as state and local population and ethnic makeup.
Add data gathered by a researcher on a mission, and the tool's power is magnified. Districts could, for instance, survey students about how hard they tried on a test and teachers could share their specific instructional approaches, says Hendrickson. At Evergreen schools, he could see exploring the growth characteristics of English-language learners and special needs students--at-risk groups that together make up nearly one-quarter of the student population.
Once a district or other researcher has a project in mind, NWEA staff will find a virtual comparison group of students from the database who match key characteristics of students in the study group. It enables a look at how students are growing relative to a very similar group of students, explains Gage Kingsbury, director of research at NWEA. While comparing, say, a high-performing, affluent suburban Chicago district to other districts in Illinois may not make sense demographically, it would be meaningful to compare that district to other, similar districts in suburbs of major cities.
It's not just about taking one school and comparing it to another. With virtual comparison groups, each student is matched up randomly to a student with the same characteristics elsewhere, and together those other students make up a "school" with the same characteristics as the initial group. As Bontempo, now principal of Mountain Measurement, explains, finding a real student, such as eighth grader Johnny, elsewhere means locating a student with similar test scores, from a school and classroom of the same size. And since Johnny is one of the oldest students in his class, the comparison student should be, too.
These kinds of comparisons are easiest when the study group is made up of NWEA member district students, since the database holds their information. But the GRD is news for other districts, too. First off, comparison group studies, while more limited, are still possible. (Pricing varies by the project.) And then there's the universal interest in results of others' research on educational trends and programs.
The GRD is "pretty much open for business right now," Bontempo says. "If I were a district administrator interested in control-group comparisons, I'd be knocking on the door."
Help on the Research Front
Some researchers have already sat down to visit with the database.
There's Al Sanoff, for instance, who has been overseeing a Columbia Teachers College project on the relationship between teacher background and classroom outcomes. When someone told him about the GRD last year, he knew it would help to have data from a subset of representative school systems.
The overall project entails collecting information from teachers about the kinds of courses they've taken and how helpful they were, Sanoff explains. Of the NWEA research team providing student achievement data, he says, "We've told them what we need, and we're looking for their guidance."
The project's second report, which will incorporate GRD data, is expected late this year.
Also underway is Wong's analysis of school choice and its relation to student improvement. The study is tracking the progress of students in charter, "regular" public and independent schools. Right now, there are more than 150 charters in the database, with more to come--a good number to work with, he notes. (When making comparisons at the student-to-student level, Kingsbury says finding about 20 similar students is typically the aim.)
Some questions Wong is hoping to answer: Under what choice arrangements do we see student improvement? Are there systemic effects in the public school district? When there are charter school clusters in a district, is the system seeing better performance because schools are competing for students?
Wenning of the Colorado League of Charter Schools is taking a different approach to the same topic with a project currently in proposal stage. "We want to furnish every school that uses NWEA assessments with information that compares each student in their school as a whole to comparison groups nationally and in the state," he says.
Participating schools can anticipate a set of reports, along with the chance to attend a workshop on interpreting initial results, late this summer. It's all about sharing best practices of schools.
NWEA researchers are also busy with their own GRD projects. An April 2005 report on the impact of No Child Left Behind covers how well the law is beginning to meet its promise. In other words, to what extent has student achievement growth changed since it was signed in 2002?
The report (posted along with others at www.nwea.org/research/national.asp) reveals, for instance, that student growth in every ethnic group has decreased slightly since NCLB first took effect. The growth of Hispanics in every grade and subject area tends to be lower than the growth of European-American students with the same initial test scores.
Researchers will continue tracking NCLB trends and reporting on progress.
Testing Teaching Materials
Another type of research the database can help with can be a boon to instructional materials developers and districts alike. Besides allowing publishers to conduct for independent program evaluations, the tool allows districts to compare a program they're using to students in other districts who aren't using that program, Kingsbury says.
During Hendrickson's many years as a curriculum director, he mainly relied on publisher-supplied testimonials or anecdotal information from other districts. "It's rare to see achievement data connected to a program," he says. Moreover, there's the crucial matter of trusting that data. "Whenever you can have a measure that's independent of the person who's making the sale, generally there's more credibility attached to it."
The GRD could also examine what happens when several programs are used together in the education of students in a school or district. "If you think about a district as an eclectic mix of programs, one of the questions that's been difficult to answer is, How does that eclectic mix of programs affect students?," Kingsbury says. It could get administrators thinking along the lines of how effective one program could be for a particular set of students, in the context of what else is being done to ensure students learn that subject area.
What About Privacy?
When individual student data is accessible to outsiders, it's not surprising that the question of privacy comes up early on in conversations about the Growth Research Database.
That's why NWEA has constructed it with privacy in mind. For example, Kingsbury says, "Researchers don't have students' or parents' names." Wong adds that he sees numbers but not which schools they represent.
Maintaining privacy procedures will be crucial as the database expands. "We're starting to peel this onion with indicators we already have," Bontempo says. Another helpful layer, he adds, would contain data on the programs students are actually being put through--not just what's been adopted but what the quality of instruction and degree of alignment to the program show.
As for the entire onion, Kingsbury says, "We'd like to create a community of practice among researchers to share findings and share ideas." He envisions various groups of researchers, from K-12 and elsewhere, working together.
On the horizon is a research Web site to facilitate that discussion. That way, he notes, no one has to reinvent the wheel.
Melissa Ezarik is features editor.