Why We've Learned So Little From Big Data

Big Data

Why We've Learned So Little From Big Data

By Leonard Medlock     Mar 4, 2014

Why We've Learned So Little From Big Data

How do we make Pre-K teachers stakeholders in their students' college and career readiness 12 years into the future? The question, posed by Howard County School Supt. Renee Foose at this SxSWedu panel, perfectly describes the gap between practice and theory when it comes to Big Data in K-12 education.

Moderated by Nick Morgan, Executive Director of the Strategic Data Project at Harvard's Center for Education Policy Research, the mouthful of a panel -- Opening Doors to Advance Analytics for Your Agency -- provided equal parts perspective from research, industry, and practice on transitioning to a data-driven education. Here are a few takeaways from the panel, which also included Atnreakn Alleyne, Data Fellow at the Delaware Dept of Education and Jeff Watson of Versifit Technologies:

Data, Insights, and Action

Finding utility in data is just as much about asking the right questions as having good data. Alleyne explained that in Delaware, they share monthly data briefs with the public to ensure they're tackling the questions that matter. Of course it can be an unpleasant experience at times for administrators when the public sees empirical evidence that the least academically prepared students in Delaware are more likely to be placed with inexperienced teachers, or that low-SES students are less likely to enroll in college even when they are deemed college and career ready. But this level of sharing and community building can significantly influence policy and allocation of resources for traditionally low-performing school populations.

Technically Speaking, It's Complicated

District data wonks know all too well that a major issue with Big Data is getting to actionable data. If the expectation from Big Data is to have teachers attempt more targeted interventions on student learning outcomes, then this year and last year's test scores aren't enough. There also needs to be up-to-date demographic data, insights from other teachers serving the same student(s), and a full academic profile of that student's learning to-date. Assuming that this information is even captured, it's no trivial task massaging the data into a standard easy-to-read format that non-data scientists can understand. Accordingly, Jeff Watson and co. have been piloting technology tools with the Strategic Data Project that minimize the time needed to clean, manage, and merge data. These processes require a large amount of human capital and are limited by discontinuous testing cycles and regulatory requirements -- external factors that districts and schools often have little control over.

Building a Data Culture

Though Howard County Schools has only been experimenting with data-driven education since 2009, Superintendent Foose says she's lived the data for over 11 years -- including a stint as school bus driver -- and the reason they "haven't quite gotten it right is because the culture is wrong." In her time as Superintendent, she’s taken great steps to align curriculum, data, and technology teams which traditionally have existed in silos. The process is difficult, in part, because of incongruous data needs. Superintendents want an accurate picture of the school system. Principals need to manage school resources. And teachers need actionable real-time feedback for the classroom.

Howard County's approach to aligning these disparate stakeholders has been to look at the complete K-12 academic sequence for students that successfully matriculate from a postsecondary institution in 6 years or less. Taking these exemplars as a proof point, she says, is paramount in helping "1st grade teachers see that they have an impact on high school outcomes." It's that kind of end-to-end accountability and ownership in students' learning that helps the entire system buy-in to data-driven education.

So What's Next?

From a research and policy perspective, Alleyne cautioned that you have to create some kind of “demonstration project” that can serve as a proof point for building consensus and adapting policy. Likewise from a technology standpoint you first need to identify what are the different reporting needs to ensure you're measuring the right outcomes.

In terms of building a culture around data, Foose added that "too many data points are too many data points!" Instead she champions finding those 1-2 data points that can drive a district to take action. When people throughout the district begin to find their own data points, then "you know you have a culture in place."

Finally, Watson offered his case for more infrastructure investments. Perhaps alluding to how well technology companies are able to understand their users, he stressed that a more connected education agency has a better chance of producing data points that can be connected in a meaningful way.

What the presenters make clear is that edtech innovation, no matter how brilliant, cannot deliver on the promises of Big Data alone. To further student learning through technology, we first must ask how, when, where, and why we've failed or succeeded in the past. In other words, we've got a lot of learning to do about ourselves.

Learn more about EdSurge operations, ethics and policies here. Learn more about EdSurge supporters here.

More from EdSurge

Get our email newsletterSign me up
Keep up to date with our email newsletterSign me up