Sections

Commentary

Student data privacy: Moving from fear to responsible use

Data has always been an inherent part of the educational process – a child’s age, correlated with her grade level, tracked to specific reading or math skills that align with that grade, measured by grades and tests which rank her according to her peers. Today this data is ever more critical. Education professionals seek understanding from what the data reflect on the teacher’s role and influence, evaluating student outcomes across classrooms. Parents seek similar measures on individual K-12 schools and districts, and desperately seek insight into the value of education at individual colleges and universities to justify the cost or debt incurred.

In the last two years, there has been a perfect storm on the topic of student data privacy. The role of technology within schools expanded at an unprecedented rate, general awareness of consumer data security and breaches increased, and student databases at the state or national level were established or proposed, which drew great public scrutiny and fear. This maelstrom yielded a tremendous output of legislative activity targeted at education technology companies, that was overwhelmingly focused on protecting and limiting the sharing and use of student data—in rare instances, to the point of forbidding research uses almost completely. There are signs that this wave of fear-driven response has finally crested, and that more measured conversations are occurring; conversations that prioritize the fundamental requirement for appropriate privacy and security, but with a clear focus on the invaluable role of research and analysis and the need to enable it.  

In December 2015, the Senate unanimously passed the Strengthening Education Through Research Act (SETRA), which would reauthorize the structure for federal education research in the Institute of Education Sciences. This vote is one of the recent signs that Congress takes seriously the research value of student data. Another encouraging moment occurred in March, when the House Committee on Education and the Workforce held a hearing which addressed stakeholders’ concerns about student data. Witnesses included parents who are concerned about understanding what is collected and how that data is used; researchers advocating access to datasets for necessary analysis; and educators describing what has been happening at the state level over the last few years to build State Longitudinal Data Systems (SLDS). Unfortunately, as of now, SETRA is stalled (or “held hostage”), potentially because of continuing distrust about broader student privacy concerns.

There is significant risk in not leveraging this information in responsible ways to improve what is almost undeniably a seriously flawed, and certainly highly unequal, public education system. Certainly all student data collection must be designed and implemented to protect personal information. However, current technologies generate highly detailed data, about large groups of students, over time and across locations, and properly handled, this provides an unprecedented chance to identify patterns that lead to success on both the individual and micro level, as well as system-wide, with reference to a broad set of factors. For example, the U.S. ranks 21st worldwide in science despite our unparalleled levels of spending, but big data may hold the answers to achieving better results that would be impossible based on individual teachers making personal observations.

The question before us moving forward isn’t “whether” to collect data—but how to define what data is needed, what to do with the collections, and how to secure the benefits of research in ways that do not put student privacy at excess risk. The fundamental role of education and its intrinsic value is one of the few areas where there is bipartisan consensus including recognition of the invaluable role of data in both creating and evaluating the success of our educational programs.

What are the opportunities from greater repositories of student data?

Let me count the ways. Smart programs check to see if students are understanding the material, are they answering questions correctly? How quickly, or is it taking them a long time to answer questions? Do they keep making mistakes? Data-based sorting and tracking decisions can counter systemic bias and the effects of unconscious prejudice. It even takes data to show us that more technology isn’t always the answer in an educational setting.

In recent decades, more extensive data and research have led to societal improvements in the realms of health, finance, government, transportation, and more. As we at the Future of Privacy Forum summarized in a recent paper, research studies in recent years have been able to identify patterns and outcomes in many educational programs that are then helpful for policymaking. For example, studies have shown poor outcomes from once-popular “zero tolerance” discipline policies; that the average poverty level of the student body is a greater predictor of poorer performing schools than per-pupil expenditures; and how multi-factor algorithms are a fairer predictor of selecting students for advanced programs than individual teachers. Some studies help evaluate new educational practices, demonstrating that “deep learning” techniques improve both short- and long-term student success outcomes; and that the right intervention at the right time may make all the difference for students in both getting into, and finishing, college. The increased types and quantities of data being generated now are an immense resource to analyze the details of the educational system, whether it’s individual academic outcomes, district-wide policies leading to graduation rates, or statewide performance metrics on particular curricula.

If we properly use this data, then having the discussion about what educational challenges need to be tackled first, and how, becomes more manageable. Not every state, school, or district will face the same challenges, but the value of research informs the best strategy for useful change with the best hope of the desired outcomes.