Filling in the Gaps with ‘Predictive Modeling’ Research at JSM 2018 in Vancouver


Another week, another opportunity for Lea(R)n to convene with brilliant minds and academic stakeholders from around the world to share some of the innovative ways we’re helping advance education and revolutionize the future of edtech and its impact on student achievement.

This week, we’re presenting at the Joint Statistical Meetings (JSM), the world’s largest international gathering of statisticians. Held this year in Vancouver, British Columbia, JSM 2018 is hosting more than 6,500 attendees from 52 countries.


Two years ago, Lea(R)n Vice President of Research & Analytics, Dr. Daniel Stanhope attended the JSM conference in Chicago, where he made a presentation outlining the use of analytics and data dashboards for evaluating the impacts of educational technologies. In other words, he introduced what is now our rapid-cycle evaluation system, IMPACT™ Analysis.  

Here’s what he said then:

“Educational technology (edtech) is increasingly ubiquitous in schools and classrooms. Products are released constantly and billions are spent annually on them. Despite the immense investments, there has been no system for evaluating their impact.

We present herein a platform that systematically integrates data from multiple sources (e.g., product, administrative, and publicly available data), automates sophisticated analytics, and generates real-time dashboards and data visualizations. Administrators and educators can use the dashboards to understand the various impacts of edtechs, including which edtechs are being used, how much they are used, and how effective they are…”

It’s an incredible feeling to reflect on what LearnPlatform’s edtech management and rapid-cycle evaluation abilities were then and how schools and districts across the country are using these tools, along with current usage and achievement data, to make more informed decisions today.

It’s one of the reasons we’re so excited to attend JSM once again — the collaboration, insights and thoughtful discussions we experience here will help shape our innovations and research in the years to come, allowing us to continue advancing education and the use of educational technology by creating powerful solutions that solve real world problems for teachers, administrators and students everywhere.

Looking Into the Future of EdTech Rankings

At JSM 2018, Dr. Stanhope is presenting research spearheaded by our very own statistician, Joyce Cahoon, on “Predictive Modeling for Incomplete Rankings of EdTech Tools.”

This concept came to us after we released our latest EdTech Top 40 research — a list of the 40 most-accessed digital tools by the 100,000+ educators on LearnPlatform, identified through our Google Chrome extensions. Numerous teachers and administrators were intrigued with the research and asked if we could provide a similar breakdown for their district, to help support their purchasing and implementation decisions.

While the concept sounded great, there were two challenges. First, research has shown that product usage alone isn’t a strong predictor of student learning, so we needed to leverage IMPACT data gathered student achievement instead. However, for many students, we often only observe their learning outcomes for a small suite of products and we wanted to provide teachers and administrators with an “all inclusive” set of product rankings.

This led to our second challenge — the need to aggregate student performance on a larger list of edtech products when we only had access to their achievement on a small subset of that list.

In our efforts to overcome these challenges and provide ad-hoc assessments to districts that ranked the “Top EdTech Products” for their students’ academic success, we explored and experimented with a variety of methods and processes for aggregating student performance across a full set of digital tools when given incomplete data.  

While the analyses and algorithms behind these methods are far more complicated than I can explain in a short blog like this, the short version is that through working with data sets that were complete — meaning they had a full set of student performance outcomes for a set of tools — our research team found a way to predict full rankings of edtech tools for districts that give them a clearer picture of which tools are the best for various segments of their students (including those of different ages, learning abilities and socio-economic levels).

With this research, the Lea(R)n research team is continuing to find innovative ways to help educators fill in the gaps on which digital learning tools are the most valuable for their students, so they can continue making the most informed edtech purchasing and implementation decisions possible.

Want to stay up-to-date on our progress with predictive modeling or any other advancements we’re making in edtech management? Make sure to subscribe to our monthly newsletter (it’s free) for current news, blogs, resources and more. Or sign up for a no-cost LearnPlatform account here to see thousands of reviews and insights on the edtech tools approved in your school or district.