The Center for Data Innovation spoke with Courtney Monk, co-founder and CPO of Schoolytics, a Washington, D.C.-based startup that helps teachers and administrators use data to track student engagement and achievement. Monk discussed some of the challenges associated with data-driven education.
This interview has been edited.
Gillian Diebold: What value does data bring to education?
Courtney Monk: At the most basic level, data in education helps educators to know four things. The first is whether/what/how much students have learned. Building on that, data helps you measure what impacts your interventions and investments have had on that learning, whether they be positive, negative, or neutral. This then helps you discover why students have or have not learned what you expected. And that tells you what you can do in the future to improve learning.
These four logical steps apply whether you’re thinking about a single student, a classroom of students, or a whole school or district. Piecing together these answers is important because, of course, educating our children well helps them to become productive adults (which creates individual and collective value). But also, because we invest public resources into our education system, we should know if it’s producing the desired learning outcomes for students.
The first point—what kids have learned—is the most important step, and the one that school administrators get stuck on, for a couple of reasons. Educators’ and researchers’ ability to measure learning accurately is inherently imperfect. For example, end-of-year standardized test scores are infrequent and are not an exact reflection of achievement. And, it’s often easier to track and collect data on inputs, like the number of dollars spent, than outputs, such as a student’s ability to think critically, which can be harder to quantify or may be more complex.
That said, thanks in part to teachers and students adopting more educational technology, we’re beginning to gather richer data from different sources, and we are collectively at a turning point with respect to data. Teachers and school leaders can now move beyond infrequent standardized test scores and bring in more holistic information from the entire learning picture: attendance data, homework data (from the learning management system), formative assessment data, grades, behavioral data, and more. In addition, as we connect that data to what we call standards—a set of concepts that students need to master in a given subject and grade level—we can get a more precise understanding of students’ learning gaps, progress, growth, and mastery.
As we pull together this data and create a more complete historical view of students, then we move on to steps two through four: empowering educators to evaluate interventions to understand why students are learning or not, and then determine what to change to foster more learning.
Diebold: How does Schoolytics empower students and teachers?
Monk: Schoolytics is built to solve the interoperability challenges for educators. This means pulling together a bunch of data from different platforms that don’t automatically talk to one another. Interoperability challenges really drive people crazy; for teachers, they have to go into multiple places to find little bits of information that they have to piece together manually on their own.
Our mission is to make data effortless for educators. We create an invisible data layer that the end-users (administrators, teachers, parents, students) don’t have to manage but makes it possible for them to use and explore data without the usual headaches of manual work. Administrators interested in making smart policy decisions across grade levels and schools use Schoolytics to spot macro-level trends in the data and cut the data by different categories to answer new questions quickly and easily.
Our platform saves time for teachers by automating manual tasks (such as sending students a list of missing assignments). In fact, when we surveyed our users, they indicated that Schoolytics saves them 30 to 60 minutes per week! This kind of teacher support and advocacy is really important right now because educators feel overwhelmed and near burnout. Schoolytics also helps educators to track student engagement and get a view of every student’s progress in real-time. In many schools, data is often reserved for administrators, so we are especially excited to offer teachers the ability to look at their own real-time data and use it to improve their practice and support students. We have robust internal evidence that when teachers use Schoolytics in their classes, their students do complete more assignments—which is an important piece of the learning puzzle.
For students, our goal is to give them agency over their own learning. Students very often are hungry for knowledge of their progress, but, given the many competing demands on teachers’ time, may not be getting regular feedback or grades. Although the Schoolytics student experience is in its infancy, we are really excited about how students are motivated by seeing their own data, and how much they appreciate the clarity of their to-do list. It is funny how much a simple view across their classes of which assignments are incomplete can help them stay organized and excel in school.
Diebold: What technologies are behind the Schoolytics platform, and are there any emerging technologies you envision the platform eventually adopting?
Monk: We are big Google fans at Schoolytics! We use Google Cloud Platform to run our backend infrastructure and have a React frontend. One of our products for school districts is actually to build them an affordable, flexible data warehouse, which is a critical need in the education space. There are some really clever, innovative ways that BigQuery allows different users to access data that we’ve been able to take advantage of.
More importantly, though, is our orientation towards data and data engineering. Both my co-founder Aaron and I are data scientists by training, which means that we have a lot of experience making meaning and information out of raw data. This has set us up really well to build Schoolytics, where we transform operational data into analytical data. For example, logs of individual student assignment submissions, aggregated over time and across classes, become a metric of the percentage of assignments students are completing, which indicates to school leaders and teachers whether students are generally keeping up with their work and mastering content.
As far as future technologies, so many things nowadays are easy to plug in and use right away, as opposed to trying to custom engineer everything ourselves. For example, we will likely leverage third-party software for in-platform messaging and have started using a new tool for user onboarding called Stonly.
And of course, we are thinking about artificial intelligence and machine learning now that we have over 11,000 teachers using the platform, and lots and lots of data. (This is of course an area of interest for us as data nerds.) Our first undertaking in this space was engagement prediction, and in the near future, we will probably be standards tagging for content and personalization for students, i.e., recommendations for what students could do to close gaps in their learning, or what content teachers could use to support each student to excel. Although we could also do a sort of chatbot tutor for kids, informed by students’ assignment histories…we’ll see!
Diebold: Bringing technology into the classroom often comes with concerns about student privacy. How can education technology best address those concerns?
Monk: At Schoolytics, we focus a lot on privacy and security, and we take our duty to protect student data very seriously. We take great care to ensure that data is safe, and in fact do more encryption than a lot of much bigger companies. We employ encryption and anonymization processes to ensure all data in storage and in transit align with best practices for protecting confidentiality and data integrity, and we use encryption algorithms such as AEAD (Authenticated Encryption with Associated Data) to handle personally identifiable information.
The regulatory framework is still in development, but it’s good that states are starting to require privacy policies and data sharing agreements. It’s important for lawmakers to understand the basic tenets of data privacy and play a strong role to establish guardrails in this space.
Diebold: What are the biggest challenges with data-driven education?
Monk: I think there are a few big challenges. The first is interoperability. In both K–12 and higher education, there is a mix of old legacy systems and new educational technology tools. These systems usually don’t connect well or talk to each other easily, for a couple of reasons. First, the legacy systems are built on an antiquated tech stack, and many have not invested in building developer APIs, which means you can’t get data out of those systems easily. And second, some players, both new and legacy ones, choose to make it difficult to export student data outside of their platforms. This might be because they want to “lock-in” users, or it’s valuable for them to maintain control of the data. Either way, it makes it really hard for actual teachers and administrators to combine data from different systems.
The second challenge is data overload. Students are using more and more devices, such as Chromebooks, and other digital tools. The pandemic definitely accelerated this trend, but it was already happening before COVID started. This means there is actually a lot of data floating around, which I think is on the whole a good thing, but it can be overwhelming for educators who are trying to figure out what to make of it all. The lack of interoperability exacerbates the feeling of being inundated by data because it’s coming from so many systems, instead of just one unified platform.
Third, limited resources are a big challenge. With the exception of large school systems like the Los Angeles Unified School District (LAUSD), districts have not been able to invest much in data management or analytics. Last I checked, data scientists and engineers are expensive to hire! And district technology teams have big strains on their time, not least because they are typically also tasked with IT and hardware support for students and teachers. Schools are therefore generally ill-equipped to leverage data for even simple analytics, let alone take advantage of AI/ML models. They really need turnkey solutions.
Lastly, building a “data culture” continues to be a challenge for data-driven education. Many of the people who work in schools have had insufficient training on how to use data. And educators, especially teachers, have either had to make do without any good analytical tools at all or been stuck with tools that are too rigid and poorly designed. That means we still have a long way to go to create a “data culture” in education, in which data is used regularly and effectively with teachers and students, and not as a “gotcha” tool. Teachers are understandably cautious when it comes to using data, given legislative trends from the past two decades. My hope is that Schoolytics will contribute meaningfully to building a “data culture” within our education system, by making it easier for educators to access and use data to achieve their teaching and learning goals.