Leveraging Dollars with Data

A data warehousing project coming online in late 2012 will help Virginia lawmakers forge education policy based upon hard data instead of anecdote and ideology.

By  James A. Bacon

Tod Massa is big man with big ideas. His goal is to build a “Longitudinal Data System” (LDS) that will track the progress of Virginians from pre- Kindergarten through high school, into college and then through their working lives. With a bear-like frame and a broad, bearded face, he speaks passionately about the potential of the $17.5 million project to “fundamentally transform” education and workforce-development policy in the Old Dominion.

When crafting policy, law makers rely upon anecdote and hunch, ideology and politics, and patchy, often-inconclusive research. The state has lots of data but it resides in different silos so it’s of limited use. The Department of Education (DOE) maintains data on K-12 students such as enrollment, graduation rates and test scores. The State Council for Higher Education in Virginia (SCHEV) keeps track of college enrollments, degrees earned and financial aid. The Virginia Employment Commission has access to employment and wage/salary data. But none of the silos connect.

The DOE can’t tell, for instance, how many graduates of a particular high school went on to earn a college degree, much less how much they earned when they entered the workforce. By December 2012, however, the LDS will link the separate databases, allowing analysts to get definitive answers to that type of question. If it can dispel the uncertainty that fogs so much debate over education and workforce development, a modest investment in data warehousing will pay for itself hundreds of times over.

“It’s an exciting time in higher education and state government,” says Massa. Virginia, like other states, has been engaging too long in “drive-by school improvements,” he says, spending vast sums of money to little effect. Soon it will be possible to answer many questions not with a survey, not with a sample, but with hard data on what is. Pet theories from all across the ideological spectrum will bite the dust and a more effective educational system will emerge.

As director of policy research and data warehousing for SCHEV, Massa represents one of three key agencies that are collaborating on the project. Virginia, he says, is one of 20 states participating in a federal initiative funded under the American Recovery and Reinvestment Act (widely known as the “stimulus” bill). What will be distinctive about Virginia’s linked databases is a mechanism for “de-identifying” individual identities in the database. Names, social security numbers and other tags will be stripped out and replaced with numerical codes. In the Old Dominion, he insists, it will be “close to impossible” for anyone’s privacy to be violated.

A top priority in educational policy right now is to measure how effectively teachers, principals and schools educate students. Virginia’s LDS will allow analysts to track, among many other things, how well a teacher’s students perform in standardized test scores. That kind of study has been accomplished on a small scale before, but Virginia will be able to roll it out statewide. Such data can be used, with appropriate adjustments for those who teach more challenging students, as the basis for rewarding good teachers and weeding out bad ones.

The list of questions is endless. What’s the payoff from investing in universal pre-K? Are college grads earning enough to pay back their student loans? Are certain majors valued more highly in the workplace?

Peter Blake, interim SCHEV director, gives another hypothetical example: Should the state channel its financial support for higher ed into undergraduate programs or should it support graduate schools, even though they cater extensively to out-of-state residents? One way to approach that question is to find out how many graduate students at Virginia colleges and universities end up getting employed in Virginia and how much money they earn. It might be easier to justify state support, he says, if it can be shown that grad schools contribute to the build-up of Virginia’s human capital.

Massa envisions researchers rummaging through the data and looking for patterns and outliers. For example, only 30% to 50% of the students attending Virginia community colleges complete a degree (with an in-state institution) within 10 years. But the figure for Richard Bland College, a two-year college in Petersburg, is 77%. Why is that? Is there something unique about the student population, or is the college doing something different? Perhaps other community colleges can learn from its example.

What really gets Massa stoked is the idea of letting scholars and members of the public use the data to generate their own findings. Qualified researchers will be given access to all the data under contract and, he hopes, a broad selection of the data will be made accessible to anyone with a Web browser.  He sees citizens being able to build their own data dashboards, create reports and publicize their findings. “I want the data shared,” he declares. Ninety percent of the citizen analysis may be worthless – but 10% might be spun into public-policy gold.