You are here


Show Me the Numbers

How data drives school and student improvement

One textbook was once good for all students. Many teachers, with no guidelines, just winged their lessons. And administrators knew some students struggled with reading, but couldn't pinpoint if the problem was comprehension or vocabulary.

It was the landscape of the good ole days of American education. And it left a plethora of students behind.

About 100 superintendents, chief information officers, curriculum directors, and other education experts gathered on the roof of the Hudson Hotel in New York City on a sweltering, soupy June day to discuss a hot and fairly new approach to education to ensure every student achieves.

The new strategy: collecting and analyzing data. Data gathering has been around for years, but greater technology and higher academic demands from the federal government are creating bigger demands for details and for personifying data, ridding systems of assumptions.

"The world is a more data-driven place," says Jonathan Harber, president and CEO of SchoolNet, provider of data-driven decision-making solutions for K-12 school districts. "With the new world there is a drive for increased academic achievement. And the more that schools can personalize at the student level, the better they will be at getting closer to the goal. ... Technology also helps in finding different learning styles. A personalized curriculum is better than having one textbook for all kids."

The rendezvous in New York was the EduStat Summit 2004, sponsored by SchoolNet, where education leaders debated and offered before-and-after pictures of programs and agreed that aligning curriculum to standards, sharing information across the Internet, and using data would most definitely bring wisdom. And of course, the federal No Child Left Behind act is driving more administrators to devise programs, particularly around data, that will ensure every child is kept in line.

How does data work for you?

"While schools know data, they don't know how to integrate it, personify it," says Joseph Jacovino, chief accountability officer for Philadelphia City School District. "Data without connection to individual children does not help create significant interventions on their behalf. Because schools have traditionally only looked at cumulative data from a variety of assessment and demographic points they, many times, fail to take the important step of creating individual student profiles."

Over the past two years, Philadelphia, which had 160 schools in need of academic improvement last year and 35 percent of its students mobile, has used various strategies to improve achievement. To ensure reforms are well-aligned and needs are met, administrators track data so that mid-course corrections can be made.

Benchmark testing in Philadelphia transformed how data is used throughout the year, Jacovino says. Curriculum is set up in six-week cycles. At the end of the fifth week, students have a benchmark test that helps determine how well students mastered concepts and standards in the five weeks prior. Teachers will use results to either remediate or enrich instruction in the sixth week based on what concepts and standards his/her class has or has not mastered, Jacovino says. Teachers can also deduce if students need after-school help.

About 250 miles west in Pittsburgh, a check and balance system to accurately assess data can change the outcome. "We need quality, clean data," says Superintendent John Thompson.

Two years ago, Pittsburgh schools created a program whereby every student's identity is "never lost," says Elbie Yaworsky, chief technology officer for the Pittsburgh Public School District. Using its own Real Time Information program and SchoolNet, each student's strengths and weaknesses are clear on tests. For example, a teacher can pinpoint every C student in math class. A teacher did find that one female student aced math tests when it dealt with pure math, but faltered in math class, Yaworsky says. It turns out the student struggled in reading and did not understand written math questions, Yaworsky says. "Now what is clear is that ... this child of brilliance was held back by reading limitations," he says. "And we set up a prescriptive plan for that student" including an after-school reading program, he says.

The data from RTI is used in Pittsburgh's Managed Learning System Architecture, which is a standards-based technology program that allows all elements of an educational IT system to work together. For example, in the seventh week in math, teachers can launch a test to see where students should be and measure their deficiencies on a particular day, Yaworsky says.

While Philadelphia has a new standardized reading and math curriculum in place, about 25 percent of the students were in catch-up, remedial classes over the summer, a jump from last year. Philadelphia Schools CEO Paul Vallas reportedly said it was due in part to tougher promotion and graduation policies and a new curriculum that raised the academic bar.

In Beaufort County School District in South Carolina, where half the students receive free and reduced-price lunch and 12 percent are English language learners, change started in the mid-1990s but only in the last two years have technology and data been used to modify instruction, according to Superintendent Herman Gaither.

The state itself was a very "poorly performing" education system before it started high-stakes testing in grades 3-8 and a 10th-grade exit exam in 2000, he says.

In the mid-1990s, Gaither wanted to "quadruple expectations" of students. The district developed its own local standards, which were later aligned to state standards. The district also uses a Training Wheels program that sends instructors into teachers' classrooms to show them, hands-on, how to incorporate technology into instruction, such as using PDAs in elementary school science and laptops for middle school.

SchoolNet created a data warehouse system for Beaufort County so all systems are linked, including a program that tracks money as well as a state program that tracks students. "It allowed us to put our finger on a problem in a more finite way," he says.

Last year, the district started using a testing software program based on state standards. It's an electronic test given to all students roughly every month, testing what concepts they know and how successful instruction has been, Gaither says. The results are disaggregated so that teachers can see if one student understands concept one and not concept two or if they mastered standard five. "Pockets of this kind of data are in use everywhere" across the district, Gaither says. "Principals... are looking constantly at what data shows them."

Is it a lack of comprehension or vocabulary downfalls or something else that hurts a student's reading ability? With the test, teachers will pinpoint the real problem, he says.

Making assumptions a thing of the past

In Chula Vista, Calif., there was an intense need to infuse data-driven decision making. "We created the notion of student-based decision making," says Libia Gill, former Chula Vista Elementary School District superintendent and now senior fellow at American Institutes for Research, a not-for-profit organization that performs research and analysis and provides technical support for education among others.

Learning First identified Chula Vista as one of five districts, in a study, Beyond Islands of Excellence, as exemplary in 2003. All five districts show continuous improvement in student achievement while using data to drive decision-making.

In Chula Vista, success is about asking questions and never assuming something before all angles are covered, Gill says.

For example, Gill wanted to learn the experience and skills new teachers have coming from one learning institution over another and then tailor the district's support for each teacher. She discovered how self-reflective new teachers were from one institution and compared them to another institution that put out teachers possibly without a clue as to how to use technology. The data is collected manually. The question becomes: Is there a pattern?

In another example, if reading scores falter one year, she encourages administrators to ask questions. "How do you know it's the program?" she asks. If teachers have poor instructional strategies, no matter what the content is, students "won't get it."

Or in math, if scores are lagging the questions to ask are: Is the teacher using appropriate materials? Does he or she need more training?

"To immediately jump to the conclusion that she's bad at math and the district should get rid of her ... is not wise," Gill says.

It's the same idea with student suspensions for tardiness. Administrators are encouraged to look deeper. If it's the same student then possibly the questions could be: Is suspension going to make this student more apt to come to school on time? What kind of behaviors do we correct with this?

"You need more than one source of data and it takes time to collect," Gill says. "The idea is not to be so quick to jump to conclusions without a thorough process."

After nine years in Chula Vista, Gill went to New American Schools, where she assisted states and districts in developing strategies for improving student achievement, which is what she did in Chula Vista. New American Schools is merging with AIR in January.

Is alignment the silver bullet?

"Absolutely," says Cathy Mincberg, former chief business officer at Houston Independent School District and now founder and CEO of Center for School District Effectiveness, a new nonprofit company designed to reform school systems. "If you can regurgitate every date around the Civil War ... but you can't describe the causes, what good [is that]? Alignment is very specific. A lot of materials are correlated. We cover fractions, but do we cover fractions the way the state wants us to know them and the way the state tests them? This is not a trick. It really is stretching to a higher order of thinking skills. It's about analyzing and thinking not just regurgitating the facts."

Years ago, when Mincberg was a biology teacher in HISD, she says teachers didn't have guidelines in what to teach. A third of the teachers followed textbooks, which had huge holes, a third followed some instructional materials the district distributed or a teacher or department person suggested, and another group "dreamed up what to teach every day," Mincberg says. "We expected kids to achieve without any kind of real map."

Texas is ahead of the game when it comes to No Child Left Behind. About 10 years ago, the state started testing and slicing up data among different groups of students to see weaknesses and strengths.

About eight years ago, then-Houston Superintendent Rod Paige wanted to create a brochure for parents, informing them what their children should learn in a particular grade, Mincberg says. The curriculum department, at the same time, was working on the new Texas Assessment of Knowledge and Skills, or TAKS, standards.

Administrators made data work for them. They use software that takes test data from the Stanford 9, a norm reference test, TAKS, and a benchmark test taken four times annually, which breaks down each student's results by each class and each school.

The benchmark test also gives teachers data during the year, not after the year when students are on the next grade level, Mincberg says. Reports are accessible through a Web portal.

From this, teachers can see a pattern or a slip-up uncharacteristic of a straight-A student. Maybe students are struggling with a family or personal problem and they need counseling or another class to jump-start their academics, she says.

Mass personalization

Although all the numbers are still not in, over the past year in Beaufort County School District fewer students scored below basic level on state standardized tests than in the past two years, Gaither says. And about 75 percent of all students scored at basic or above. About 10 to 15 percent of students are scoring at proficient level and about 2 or 3 percent are at advanced levels.

"I think we are capable of doing better," Superintendent Gaither says. "I think we will do better. A lot of things in our system support instruction and our kids. ... Our goal this year is to reduce the number of kids below basic and increase students in proficient and advanced levels."

Harber predicts that in the next five to 10 years, "mass personalization has to be the end goal." "From data come extraordinary results," he says. "No parent will stand to send their kids to school if they can't go online and see their child's work and be able to talk to teachers [via computer]."

Angela Pascopella is features editor.

Related Information