If we want 2-sigma learning, we have to break school
In 1984, educational psychologist Benjamin Bloom described a holy grail that schools have been chasing ever since.
According to Bloom, an educational model that uses mastery learning and one-to-one tutoring can substantially improve outcomes in education. Bloom’s paper is titled the “2-sigma problem” because it proposes an improvement in learning outcomes of up to two standard deviations. To be clear, this kind of shift would be outrageous. At scale, a 2-sigma improvement to learning would be the single biggest breakthrough in the history of modern education.
The 2-sigma promise sent a ripple through the education world. Some people find the results misleading. Others are actively trying for 2-sigma. But so far, no one has delivered results that prove the theory. Doing so would mean getting both mastery learning and one-on-one instruction right for a massive number of students. The challenge? It’s flat-out hard to scale the benefits of one-to-one tutoring.
At Alpha, we think 2-sigma results are possible if you’re willing to ditch the traditional school model and use adaptive learning software. And after reviewing the wave of education research that has come since Bloom’s ripple, we’re even more optimistic about tutoring, mastery, and the 2-sigma problem.
Let’s take a look.
One-on-one tutoring works because of fine-grained feedback.
Tutors are effective because they can intervene with feedback often and at just the right moment. But not every student in the country can have their own human tutor.
However, maybe the most important aspect of tutoring isn’t the instruction from a human. In his exhaustive review of the literature on the 2-sigma problem, Jose Rincon claims that there is “suggestive evidence” that the reason one-on-one tutoring works so well is because of the fine-grained feedback that tutors provide.
Alpha example: Adaptive apps continue to get better at giving students timely feedback. Apps like Albert and Khan Academy provide quick interventions and feedback that tell students exactly how they did on each problem.
Use apps to scale fine-grained feedback, potentially the most important aspect of tutoring.
Constant testing is the key to learning.
This idea is also known as the “testing effect.” Specifically, repeated retrieval (being asked to recall concepts over a long period) and spaced repetition (being asked to recall something over increasingly longer intervals of time) have strong scientific support.
- A 2008 paper mentioned a study in which “repeated retrieval increased final recall by 4 standard deviations.”
- Rincon suggests that most of the power we attribute to mastery learning is really just the power of the testing effect in disguise.
- This review of the literature mentions studies whose results favor spaced repetition over massed practice (cramming).
- Also, a 2016 study found that students who practiced spaced repetition increased their retention rate three-fold.
Being tested often is good for learning.
Alpha example: Apps are great at testing. And they test often. In an Algebra II course of Khan Academy, a student will see 79 quizzes and 13 unit tests. Compare this to an Algebra II class at a traditional school, where students will be tested half as much. Oh, and Khan is just one of three apps that an Alpha student will complete for Algebra II.
Adaptive learning demolishes static learning.
A 2020 study found that 75.5% of students who used adaptive learning “scored above the mean of a control group.”
Alpha example: Where possible, we aim to use the most adaptive app available. The MAP test that students take three times a year is also entirely adaptive.
The standard of mastery matters.
We’ve covered the idea of mastery learning before, the main takeaway being that, when done in the right environment, a mastery model can have an astonishing impact.
An important question within a mastery model is at what standard a student is considered to have mastered the content. This varies by institution. A study described in this 2011 paper shows evidence that raising the mastery standard from 80% to 90% can make a massive difference.
Alpha example: Alpha’s program is fueled by a bold belief in mastery learning. We utilize the mastery standard built into each app, therefore, Alpha mastery standards are extremely rigorous.
Adaptive learning apps work at scale because they test often, adapt to the student’s needs, and give feedback, all things a great tutor might do. Additionally, we apply a mastery standard to all of our app-based coursework.
Going a step further, we believe that if you want to do this well, you must think differently. Mastery learning and adaptive learning apps work wonders together if you’re willing to break the existing school model, which means shattering rigid schedules. Give learners the time and freedom to direct their own learning.