Personalized, Adaptive, and Mastery Learning

EdSurge has a good recent article by Sal Khan on the intersection of mastery learning and technology. As Kahn defines the term, "Mastery learning is the idea that students should adequately comprehend a given concept before being expected to understand a more advanced one."

One goal of personalized and adaptive learning technologies is to provide reactive instruction, to the degree possible giving students the specialized attention that might be otherwise afforded by a tutor or teacher.

Now with software, we can begin to deliver on Benjamin Bloom’s dream: help teachers identify and meet the needs of each student so that all students can master important concepts. Too often, students struggle in calculus due to gaps in their algebra skills, or in algebra because they never fully mastered fractions. A tool like Khan Academy can provide students with unlimited practice and instructional support for each skill so they can be sure they’ve truly mastered a concept before moving on. And we offer teachers data on how their students are doing so they can identify gaps and provide tailored instruction. That way they can spend less time making differentiated worksheets for each student and more time interacting with and inspiring their students. After all, nothing beats getting more quality time with an incredible teacher.

Audrey Watters gives us a nice overview of the main players in adaptive technology in "What Should School Leaders Know About Adaptive Learning?" Her post is a couple of years old, but still looks relevant, including this definition of adaptive learning:

The term “adaptive” actually covers a number of characteristics in software (indeed, the term is applied quite liberally in marketing new tools). Traditionally, adaptive learning tools focus on the following components (PDF): “monitoring the activities of its users; interpreting these on the basis of domain-specific models; inferring user requirements and preferences out of the interpreted activities, appropriately representing these in associated models; and, finally, acting upon the available knowledge on its users and the subject matter at hand, to dynamically facilitate the learning process.”
​
In other words, adaptive software focuses on the domain – the information that’s supposed to be taught; on the student’s responses – often using “bayesian algorithms” to determine the likelihood of a student getting the next question right or wrong; and on the instruction – what sorts of hints or feedback a student might need to move forward.

Critics are often heard to complain that all of this investment in technology has yet to yield positive quantifiable results. For example, a June 2016 article by Blake Montgomery looks at a Harvard study of Dreambox Learning, and the broad initial statement is similar: "There is little evidence yet that educational software is actually helping students progress more rapidly." But from there things start to get much more positive. In particular:

1. Students who spent more time on the DreamBox software saw larger gains in achievement, and those who followed the company's lesson recommendations saw faster gains.DreamBox operates by engaging students in math lessons, gauging their progress then recommending other DreamBox lessons in order to, theoretically, optimize mastery. Following these suggestions, Harvard found, helps students more so than repeating previous lessons.
​
2. Researchers wrote, "The DreamBox progress measure was positively associated with achievement gains on state tests and other interim assessments." Measuring students beginning the year at the 50th percentile in math scores, Harvard found that Rocketship students reached the 54th percentile by the end of the year and that HCPSS students advanced to the between the 54th and 55th percentile. The measured students were using DreamBox the average amount for their schools.

So perhaps we're starting to turn the corner with respect to achieving significant measurable results from EdTech.