Five Design Principles for Smarter Data Systems

In the past decade, school districts and states have spent more than a billion dollars to build and implement data systems. Data about student learning—and the systems that collect, organize, and report on this data—are what U.S. Secretary of Education Arne Duncan calls "the driving force [behind education] reform."

In the next five years, our nation will likely spend a billion dollars more. Influential persons and institutions, from legislators to governors to the Gates Foundation, extol the virtues of better data. Data systems are a key criterion for Race to the Top fund applications. And the American Recovery and Reinvestment Act—the stimulus package—injects an additional $250 million federal investment into longitudinal databases and requires states to assure that they are building these data systems as a condition for receiving stimulus funds.

The rhetoric around educational data is compelling: With better data, policymakers can identify effective schools and educators, expose problems, make better decisions about the allocation of resources, and build political will for reform. At the classroom level, better data will inform instruction—enabling teachers to better understand what approaches work for specific students—and lead to better teaching and improved learning.

Our first billion-dollar investment has yielded important results—building the capacity and systems for better collection and management of data. According to the Data Quality Campaign, by 2011, all 50 states will have longitudinal data systems that track student performance from year to year. Many districts have followed by building their own extensive data warehouses.

But, despite states' and districts' tremendous progress in building data systems, policymakers are not yet routinely using these new data to improve accountability systems, support performance management processes, evaluate programs, or influence resource allocation decisions. More importantly, the data is not yet being used where it matters most—in the classroom.1 A 2009 U.S. Department of Education report found that "even in districts with a reputation for leadership in using data, electronic data systems are barely influencing classroom-level decision-making."2 Many systems aren't even designed to provide teachers—or students and their families—with access. And much of the data—in particular, scores on annual state tests—are not useful or appropriate for informing day-to-day classroom instruction. Data may be everywhere, but the systems to access and use the data are disconnected and many times impossible to use.

We're now entering the next phase in our nation's thrust to use data to improve educational outcomes. The challenge is no longer whether or how to build institutional data systems, but to use better information about teaching and learning to improve outcomes for every student. A focus on the actual use of data must drive our next billion-dollar investment. And this focus has clear implications for how we think about, design, and implement data initiatives. Going forward, five principles should inform these initiatives:

  1. From Institutional- to Learner-Centered: It sounds simple and obvious, but designing learner-centered systems would mean dramatic shifts in current practice—away from compliance-focused data needs, such as reporting for federal programs, and toward information that students, families, and teachers can use to support day-to-day decisions about learning and achievement. Like a GPS navigation system, learner-centered systems would track a student's progress and provide individualized guidance on the paths and actions to take toward his or her learning goals. They would combine both hard data and qualitative information to give a full picture of learning and performance throughout a student's entire educational experience. The information would be timely and accurate. And, it would be specific enough to inform the actions of teachers, students, their families, and the variety of persons who support learning, including instructional coaches, tutors, after-school and youth workers.
  2. Information Flows Across Institutions: Students are increasingly mobile and not just across schools, districts, and states, but across a number of different learning opportunities within both the traditional school day and out-of-school programs. Yet, a teacher cannot use information for a transfer student when it's trapped in another district's data system; if you can't mobilize information, you cannot use it. A learner-centered system would operate across institutions, integrating important information from a wide variety of schools, programs, and interventions into a more complete and accurate depiction of a student's progress. It would allow adults who work with students to better communicate and to understand more about those students. And, it would enable schools, districts, states, and the federal government to improve the accuracy, timeliness, and efficacy of data-gathering. Internet-based platforms that enable this seamless exchange of information, rather than costly projects to integrate systems on a one-by-one basis, are critical to learner-centered approaches.
  3. Usefulness and Usability Drive Adoption: Good teachers strive to know their students better. To be successful, data initiatives must inform the work flow, incentives, and actual day-to-day practices of educators, but avoid "one-size-fits-all" solutions. And, unless systems are designed to be valuable to these educators in their daily work with students and provide insights into students' goals, use of data will be limited. The best tools won't require extensive training or campaigns to convince educators of their value—they'll be so intuitive and easy to use that adoption comes naturally. Importantly, educator use will ensure that data is more accurate, allowing for better information and decision-making at all levels.
  4. Common, Yet Open, Systems: Our nation's massive and ongoing investments in similarly focused health care initiatives, such as electronic medical records, which strive to use information technology and better data to radically reform and improve health care, provide an important lesson: Open, yet standards-based systems are critical. Here, good governance is essential to ensure that common policies, technical standards, privacy protections, and usage protocols across institutions are enacted and attended to on an ongoing basis. This facilitates use and allows educators to easily exchange information. But, monolithic systems are neither effective nor easily adopted. Local districts, schools, and a wide variety of other programs, from after-school to internships, must be able to tailor and customize systems for their particular information needs. The iPhone (and other "smart" phones) provide good examples: They adhere to recognized (or what might be considered standards-based) protocols so that they can operate across common voice, data, and GPS networks. The software, however, also enables the easy development and integration of an unending variety of useful applications, or "apps," that users can install to customize and extend the use of their phones. State longitudinal data systems, which currently exist separately from district systems, could be designed in a similar way—so that they provide the core of data for every district, but enable districts to easily tailor and extend beyond the core for their unique needs.
  5. Get the Right Data: The right data allows us to not only assess performance—for students, educators, and administrative systems—but more importantly, to understand the processes and changes that could lead to improvement. While improvement is still not optional, the message is no longer just "You're failing." Instead, it's "Here's where you are. Here's where you need to go. And here's data, information, and suggested actions to help you get there." Just as using data to inform student learning can empower students, providing high-quality data to educators, schools, and districts about their practices will also be enabling. The most successful performance management initiatives—those that actually change practice—will engage educators and put student success and responsibility at the center.

The potential for better data to improve education is real. Data and data systems are used to help inform and improve decision-making in almost every field of human endeavor—from health to sports to crime prevention to finance to even the production of motion pictures. Schools and districts, in cities such as Charlotte, New York, Dallas, and Houston, are beginning to see results. But it is astoundingly difficult to impact day-to-day classroom practices. And unless we design data systems with a primary goal of improving classroom teaching and learning, our investments will show little return.

Endnotes

1. For example, see the Data Quality Campaign's January 2010 "States' Actions to Leverage Data to Improve Student Success" survey (http://www.dataqualitycampaign.org/survey/actions) and the U.S. Department of Education's "Use of Education Data at the Local Level: From Accountability to Instructional Improvement," (http://www2.ed.gov/news/pressreleases/2010/01/01272010.html)

2. U.S. Department of Education, "Implementing Data-Informed Decision Making in Schools-Teacher Access, Supports, and Use," 2009, (http://www.gesci.org/assets/files/Knowledge%20Centre/Implementing%20Data%20Informed%20Decision%20Making%20in%20Schools-Teacher%20Access,%20Supports%20and%20Use.doc)

More in Accountability and School Improvement