On this page
Most study plans move on when the calendar says so. Mastery learning moves on when you can actually do the thing. The idea is decades old, and it sits underneath many of today's adaptive practice tools, so it is worth understanding properly.
What is mastery learning?
Benjamin Bloom described it clearly in his 1984 paper "The 2 Sigma Problem". In a mastery learning class of about 30 students, the teaching is the same as in a conventional class. The difference is what happens after each unit:
A formative test, used for feedback rather than marks.
Corrective procedures for whatever each student got wrong.
A parallel formative test to check whether the student has now mastered the material.
Only then does the class move on. In the studies Bloom summarised, "the average mastery learning student was above 84% of the students in the control class", about one standard deviation better. Tutoring with the same feedback-corrective steps did even better, around two standard deviations. Those were short studies, over three-week blocks, so treat the sizes as encouraging rather than typical. Our profile of Bloom's 2 sigma problem covers the debate around them.
Does mastery learning work?
A larger test came from a 1990 meta-analysis by Kulik, Kulik and Bangert-Drowns of 108 controlled evaluations. Mastery programmes had positive effects on exam performance in colleges, high schools and the upper grades of elementary school. The effects appeared "stronger on the weaker students in a class". There were costs too: mastery programmes "may increase student time on instructional tasks", and self-paced versions "often reduce the completion rates in college classes."
That is the honest trade-off. You learn each unit more thoroughly, but it takes more time, and if nobody sets deadlines, some students fall behind.
How do adaptive tools apply it?
Adaptive tools try to automate the loop Bloom described: test, correct, retest, move on. The questions to ask are how finely they check your work and what they do next.
Kurt VanLehn's 2011 review sorted computer tutors by how closely they interact with a student's work. Systems that only check final answers are "answer-based". Systems that respond to each step of a solution are "step-based" or "substep-based". Intelligent tutoring systems, which mostly work at the step or substep level, reached an effect size of 0.76, close to human tutoring at 0.79.
So a good adaptive tool does more than count right answers. Ideally it notices which step went wrong, chooses your next problem to target that skill, and brings back older skills later so they do not fade.
A do-it-yourself mastery loop for one chapter
You can run Bloom's loop with a notebook. Here is an example for Current Electricity. The 4-out-of-5 bar is our suggestion, not a research standard.
List the skills (10 min). For example: Ohm's law and resistivity; series and parallel combinations; Kirchhoff's laws; meter bridge and potentiometer; power and heating.
Formative quiz (20 min). Five short problems per skill, from a source you have not practised from.
Mark by skill. 4 or 5 correct: mastered for now. 3 or fewer: needs correction.
Correct (30 to 60 min per weak skill). Re-read only that section, study one or two worked examples, and write down what you got wrong and why.
Parallel quiz. Five new problems on the same skill. Do not move on until you reach 4 out of 5.
Space it out. Revisit every mastered skill with two problems a week later, and again after a month.
What to watch for
Time. Corrective rounds take time. Leave buffer days in your plan for them.
Drill versus transfer. Getting 5 out of 5 on near-identical problems is not mastery. Make the parallel quiz genuinely different.
Deadlines. Self-pacing needs a calendar too, or a chapter can stretch for weeks.
If you are preparing for JEE Main or Advanced and want a structured version of this, Fermi Testprep offers adaptive mocks and a study plan that updates based on mastery. To get more from the mocks you already take, see our guide to using the data from your mock tests.
