What study methods actually work, according to Make It Stick?

An AP economics class went from a 20% to a 60% pass rate. The teacher switched to heavy retrieval practice and let students generate wrong answers in class. McDaniel says the turning point was the day students took a practice exam and saw their own results.

A Florida law school went from the state's lowest bar-pass rate to its highest in three years. It rebuilt courses around quizzing and self-testing instead of one exam at the end, and the authors stress the school did not change its admission criteria.

Answering wrong still strengthens later learning, provided the answer is corrected. Roediger calls it test-potentiated learning. The older assumption in his field ran the other way: produce a wrong answer and that is the thing you will remember.

Learning styles have no good supporting experiment, and their own study found mismatching can help. McDaniel says no good experiment he knows of shows that matching instruction to a style improves learning; a study the authors published the year before found mismatching sometimes produced better learning, possibly as a desirable difficulty.

A blocked problem set practices one of the three steps a test requires. With one solid at a time and the formula supplied, students never practice recognizing the problem type or retrieving the formula — so the homework looks perfect and the exam does not.

Why does rereading feel like learning when it is not?

McDaniel's answer is about fluency, and it is the most useful idea in the episode. When you reread, the passage gets easier to process at every level — the words, the sentence structure, the extraction of ideas. Your brain reads that ease as evidence of knowledge. In his words, familiarity is a very misleading cue, and it gives you the illusion of knowing. The trap is that the fluency is real; it is just fluency at levels that do not reach deep understanding.

So the student sits with the chapter thinking I know this, I know this, I know this. Nothing in the process ever contradicts that feeling, because nothing in the process ever asks for the material back. McDaniel's verdict is the line worth taping to a desk: with nothing pulled back out of memory, a student has no information at all about what they actually know.

Nick brought the extreme version of it to the interview. A college student came to his office struggling in microbiology, getting C's, D's and F's while feeling like she knew the material. She had been rereading for all her free hours. When she hit a question on the exam, what came back to her was the color of the highlighter she had used on that passage — pink — and not the content. McDaniel's reading of it is exact: if she can retrieve the color but not the content, that tells her the content is not in memory well enough to retrieve.

Neither author tells students to stop reading. Roediger's framing is that there is nothing wrong with reading, highlighting and underlining as a start, and that making things stick takes other techniques on top. McDaniel adds the part students miss: after retrieval practice, your rereading changes. You spend your time on the sections you could not recall instead of spreading it evenly, or worse, over the parts you already knew.

What is retrieval practice, in practice?

Roediger defines it as practice at getting information out of memory rather than only getting it in. Either the teacher tests you or you test yourself. He notes that teachers talk about building a storehouse of knowledge as though depositing it were the job, when the thing you will be asked to do later is withdraw it.

The low end of the technique is the multiplication flash card: six times nine on one side, you try to produce the answer, you check. His point is that the same move scales up. Read five pages, close the book, and say what you can remember. Use the questions at the back of the chapter. McDaniel adds one more, from the learning-strategies course he teaches: a free recall of the material before you reread it. A student of his reported back that it made her far more active in the material when she did go back and read.

Roediger's sports version lands with students who think memory is a different kind of thing from skill. No one expects to learn to skate by watching skaters or reading about skating; you have to get on the ice. Memory is the same — you practice retrieving, because retrieving is the task.

McDaniel lists the payoffs in a bundle, running straight on out of that same mechanism. Retrieval itself produces better long-term memory. It increases your accuracy about what you know, and that makes your study time better aimed. And having to try to retrieve also makes the later learning of the material you could not retrieve more efficient. Three separate benefits, one habit.

Why does getting it wrong still help, and what is the condition?

McDaniel hands this one to Roediger as his area, and Roediger calls it test-potentiated learning. The surprise, he says, is that you can give an answer, get it wrong, and still come out ahead. The older view in his field was the reverse: a wrong answer was the one thing you must not produce, because it would be the thing that stuck and the thing you later remembered.

The condition is the whole ballgame, and he states it plainly: correct the answer afterwards and the potentiation still happens. Guess, miss, then check. A wrong answer left uncorrected is not the experiment. A wrong answer corrected beats never having been asked.

McDaniel explains the mechanism in terms of what failure makes you do. When you cannot get something out of memory and then look at it again, you start working out where the obstacle was and why you could not remember it, and you connect the material to things you already know. That is deeper processing than being handed the same page to read again.

Roediger is honest that the finding sits badly with intuition, and says he has been told as much — that testing beating restudying is deeply counterintuitive, because school is built on the assumption that studying is how you learn and a test only shows what you already know. His correction is one sentence: testing is a means of learning better.

What are desirable difficulties, and why do the forgetting curves run in opposite directions?

Roediger credits the term desirable difficulties to a psychologist at UCLA, and then gives the result it describes. Take people who have studied something once or twice. Have one group restudy repeatedly and the other test themselves repeatedly. Test everyone immediately and the repeat-studiers win.

Wait a couple of days or a week and it reverses. The group that did retrieval practice — and in the version he describes, without feedback — started lower and barely lost anything; their forgetting curve is very slight. The massed-study group started high and dropped steeply. Two curves, opposite directions, same material.

That shape is the definition of a desirable difficulty: something that hurts you on an immediate test and helps you on a later one. Roediger says interleaving does the same thing. Hurts initially, helps in the long run. And he is clear about who the advice is for: students who want to know the material for the long run, not just get through the test coming up in an hour.

He also explains why cramming survives. Block the practice, cram the night before, and you can get through an immediate test — especially a multiple-choice or true-false test, where you only have to recognize something as familiar. Once a test asks you to produce the answer with no recognition cues to lean on, the cramming stops working.

What does blocked practice look like, and why do coaches and teachers like it?

Roediger's first example is hockey. Ask a coach how players practice and you hear 10 slap shots from the right, 10 from the left, 10 from the middle. Then ask whether that ever happens in a game. It does not. Batting practice is the same shape: 10 fastballs, 10 curveballs, 10 sliders, when the whole skill in a game is reading what the pitcher is doing and anticipating what is coming.

McDaniel extends it from single skills to whole phases of play. A team with a three-hour practice gives an hour straight to breakouts out of the defensive zone, an hour to offensive sets, an hour to moving the puck through the neutral zone. The players do get more fluid as the hour goes on. Players and coaches read that as the practice working.

McDaniel's correction is the one he makes about rereading as well: that fluency is not necessarily indicative of robust, long-term learning. Smooth today, gone next week. And blocked practice removes the hardest part of the real task — you already know which phase of the game you are in, so you never practice recognizing the cues that tell you.

The interleaved version cuts the three hours into segments of 10 or 15 minutes, switches between breakouts, offensive sets and the neutral zone, and circles back. Now the practice contains the thing the game is made of: transitions arriving in seconds, with the right response having to come back from memory fast.

What did the authors change for the Navy SEALs?

Roediger says one surprise of the book was the mail. Sports coaches wrote. Musicians wrote. And the head trainer of the Navy SEALs wrote, saying they had read the book and thought they needed help with how they train. The authors spent two days with them, going through what the SEALs were doing and how they might modify it according to the principles in the book.

The main change was the move off blocked practice toward what they call transfer appropriate practice: think about what the criterion task is, what the ultimate goal is, and make the practice correspond to it as closely as you can. The same logic that says a batter should face mixed pitches says a team should train the way it will operate.

Roediger is also candid that some of it could not be fixed, for an ordinary institutional reason. SEALs have to practice jumping from a helicopter well offshore and getting to the beach, and they can only get helicopters from the Air Force on certain days. So the practice gets massed onto those days, because that is when the helicopters are there.

That constraint is worth holding onto, because every school has its own version of it. The question is not whether you can interleave everything. It is whether the parts you control are arranged to resemble the task you are preparing for.

What does a blocked problem set hide, and does mixing it up fix that in a real course?

This is the sharpest classroom point in the episode, and it is Roediger's. Take computing the volume of a solid. In a typical problem set you work one type of solid at a time, and the formula is supplied. So you practice one step: putting numbers into a formula you have been handed.

The test does something else. It shows you a solid and asks for the volume. Now there are three steps. Recognize what kind of solid this is. Retrieve the right formula from memory. Then compute. Two of the three were never practiced, because the problem set did them for you.

The consequence runs in both directions. Students think they really know the material, because they did the homework perfectly. The teacher thinks she did a great job teaching it, for the same reason. Then everyone is disappointed by the exam results — and Roediger's view is that the outcome is perfectly predictable if you use blocked practice. His prescription is not to throw blocking out: you probably need some at the start to get the hang of a procedure, and after that you mix it up so the two hidden steps get practiced too. Nick found the same thing by trial and error in SAT work. When a student has three problem types giving them trouble, the fix is not a page of those three types, it is whole mixed sections where the student has to notice which problem they are looking at.

McDaniel says the volume-of-solids work was laboratory work, and that the interesting studies now are running inside physics courses. Physics is notorious for students who know the formulas and cannot tell which kind of problem is in front of them — a work problem, an energy problem, whatever it is. The design he describes crossed over mid-semester. One group got homework with problems mixed across units. The other got the standard version: this week's problems from this week's unit. Halfway through, the groups swapped.

On the first exam, the students with mixed homework scored higher. On the second exam the result turned over, and the group newly moved onto mixed homework was the one ahead. Same students in both halves. The only variable was how the practice had been scheduled, not how smart anyone was.

McDaniel also names the cost to the teacher, which is why this is hard to adopt. Teachers watching interleaved practice see students struggle more and look less fluid, and conclude they are teaching badly. What the students are actually getting is the part the exam will test: how to figure out what kind of problem this is.

Why do students judge mixed practice backwards?

McDaniel brings up a study by one of their colleagues on learning to identify painters' styles. In one condition you study paintings by different artists interleaved. In the other you study all the paintings by one artist, then the next artist, then the next.

Study after study, experiment after experiment, the interleaved condition produced better recognition of who painted what. And the students uniformly rated the blocked condition as the better teacher, because it was easier and felt more comfortable. Comfort converted straight into a judgment about learning.

That is the tension students have to get over, and McDaniel is clear that there is only one thing he has seen do it. They have to take a test and see that they perform better after interleaving. Objective feedback is the key to convincing a student that the method is working, and he says many instructors who want students to adopt better strategies skip that step entirely.

It applies to the student in Nick's office too. McDaniel's note on her was that even if the app had mixed her practice properly, she still could have felt it was going badly — until an exam told her otherwise. Feelings are not the instrument. The score is.

How do you mix up practice when nobody will do it for you?

The authors' answer is a partner — a confederate, in the word used on the tape. You need someone who knows the problems you have to solve over the next five, six, seven weeks, who will mix them up and present them to you. Second best: pull problems out of the assignment at random so you do not know what is coming before you look.

The household version is McDaniel's favorite story, and it is a real one. His department chair had daughters whose math homework came home as sheets of all one problem type. She refused to let them do it that way, cut the sheets up, shuffled the problems, and handed them back. Their math improved after that.

There is a punchline that tells you how the method feels from inside. Roediger and McDaniel were invited to the family's Christmas dinner, and the daughters would not come say hello, because their mother had told them who suggested the scissors. McDaniel's summary is that the parents got the best of both worlds. The short-term pain was entirely the daughters'.

And the details matter more than people expect. Nick's microbiology student went looking for a tool that would do this and found Quizlet, which randomizes the questions and brings back the ones you miss. What it does not randomize is the order of the answer choices, so she was learning that a given question's answer was B rather than learning the content. The devil, as Nick says, is in the details of how any of this gets put into practice.

Do learning styles hold up to the evidence?

McDaniel separates two claims that usually travel together. Students will happily take the survey and describe themselves as auditory or visual learners, and it may well be true that a person has a preferred way of taking in information. The second claim is the one the whole industry rests on: that matching instruction to that preference improves learning. On that one he is flat: he knows of no good experiment showing that the match improves anything, and says there is no evidence at all for it.

Their own work goes one step further. In a study the authors published the year before this conversation, teaching that ran against a student's stated style sometimes produced better learning than teaching that fitted it. McDaniel's guess is that the mismatch adds a desirable difficulty — the student has to think a little harder and so learns more. He keeps the hedge on it himself: that was one study.

He also points to a study at Indiana University that queried students on their learning styles and then looked at how they actually studied. The techniques students chose did not even match the styles they had claimed — while students may still demand that teaching match those styles.

McDaniel's diagnosis of why the belief survives is that good instruction gets credited to the wrong cause. Strong teaching tends to be multimedia: diagrams, text, listening, hands-on experiments, even virtual reality. All of that improves learning, period, and none of it works because it matched someone's style. Meanwhile teachers tell him that 20 years in the classroom has taught them matching works, with no objective evidence behind the impression. McDaniel adds the structural costs: schools send teachers to summer institutes to get certified in identifying styles, and he and Roediger have sat in a high-school Q&A listening to teachers describe parents who call in to say their child is underperforming because the teaching does not meet the child's learning style. He calls it a useful scapegoat, and notes the obvious contradiction nobody resolves: a teacher facing students with seven different learning styles cannot teach to all of them anyway.

What happened at Harvard Medical School?

Roediger heard that Harvard Medical School was revamping its curriculum and that Make It Stick was playing a part in it, so the authors asked to interview the people involved. They talked with the medical school's dean of education.

The dean opened with the thing Roediger keeps repeating: it is remarkable how many people arrive at a medical school that selective with no idea how to study. The first two years bury a student in information that has to be known cold, and the habits that got them admitted will not do that job.

So the school rebuilt the curriculum around the book. The detail that makes it concrete is what happens at the meeting of admitted students before the fall. One student asked what they should read over the summer. The dean was expecting to be asked about a physiology text. The answer was to read Make It Stick instead, because the curriculum was built around it and the students needed to know how to study.

Roediger has two smaller versions of the same thing. His own daughter went to medical school and said the same about her classmates: all they did was reread, and then the material was not there to retrieve on the test. And at a physical, a resident shadowing his doctor mentioned the book, which he had read before starting medical school. Roediger told him that was his book.

How did a Florida law school go from last to first on the bar exam?

This is the case with the most institutional weight on it, and it begins with the co-author who is not on this episode. Peter C. Brown, the third author of Make It Stick, was the one invited down to speak at the law school, which the authors name on the recording as Florida International University, because the school was already trying things along these lines.

The starting position was the bottom. A new dean looked at a law school ranked near the bottom on the things that matter, including students passing the bar. The curriculum changed toward more quizzing, which is a bigger shift in law than it sounds: in many law schools every course is a single test at the very end. Grades went up every year.

Then they applied it to the bar exam itself. The school had the lowest pass rate on the Florida bar. After three years of the new approach — and, the authors stress, with no change to its admission criteria — its students were passing the bar at the highest rate in the state. The telling keeps its own hedge: the author recounting it says he forgets how many law schools Florida has.

The explanation is almost tautological once you hear it. The bar exam is a test. If you want people to pass a test, have them test themselves and test them. And the point runs past the exam: every time a lawyer works on a case, that is a test, the same way every patient a doctor sees is a test — mixed symptoms, several candidate explanations, and a decision that has to be retrieved from memory.

How did an AP economics class go from a 20% to a 60% pass rate?

In this teacher's AP economics class, 20% was the usual share of students who passed the exam. He leaned hard on retrieval practice and ran the room so that students produced wrong answers out loud, and the share passing reached 60%. McDaniel tells it on the tape as one of the cases the authors are collecting for a follow-up book.

The part the teacher emphasized was not a technique, it was a culture. Students have to understand that wrong answers are fine, that you are allowed to struggle a little bit to get the concept, and that struggling is where the learning starts. Without that, nobody volunteers an answer they are not sure of, and the whole method stalls.

And the turning point was objective feedback again. The best day he had ever had in school, the teacher said, was the day his students took a practice exam and saw for themselves how well they were doing. They bought in. After that, by his account, everybody could not wait to do retrieval practice, answer questions and write on the whiteboard.

McDaniel pairs it with a high-school algebra teacher who mixed his topics up instead of teaching one at a time. The test that mattered was the one he gave when students came back in the fall. Parallel courses taught the usual way showed the normal summer drop; his course showed almost none. What excited the teacher most was something else he reported: his students became adept at making connections across different aspects of math, he had never seen students make these connections, and they were not only making them but pushing to make new ones. He said they became like active mathematicians.

Does all this testing make test anxiety worse?

Roediger takes the objection seriously and calls it a reasonable point. If your advice is more testing, the obvious worry is more anxiety. The answer in their own data turns on one design choice: in the middle-school work, the quizzes were zero stakes and the students knew it. They were there to learn from, not to be graded by.

The surveys came at the end, from the study their colleague Pooja Agarwal led. Students were asked to compare this class with the rest of their schedule: more anxiety about tests, the same, or less. 80% landed in the bottom two boxes — the same as their other classes, or less.

Roediger's contrast is a system organized the other way. He describes a single two-hour exam that decides whether a student goes to college and which one, and a Japanese psychologist telling him students are paralyzed by it. His read is that one way to reduce the fear is to make tests routine enough that no single one carries a life in it.

Nick's version from SAT work is the reframe: a test is a snapshot, not a verdict. A starting score in the middle of the pack is a piece of feedback and a starting point. He says students who carry that frame back into school see anxiety fall and performance rise, because the stakes are still visible but the test has stopped feeling like a personal judgment.

What did the authors find in their own university's professional programs?

McDaniel offers this as an extension, and it is the most uncomfortable finding in the conversation, because it is about students who are doing exactly what school trained them to do. In the professional programs at their own university — physical therapy among them — the authors found a negative relation: the heavier a student's reliance on rereading and memorizing, the worse that student did on the deep questions, the ones asking for application, synthesis, integration and understanding.

Those students were fine on the factual questions. They fell down on the application questions. And their retention fell too. The programs run learning checks that put the same questions in front of students four months later, told openly as no-stakes, and the memorizers came back worse on those as well.

The reason it matters now is that the tests are moving. McDaniel says schools are shifting toward exams that require understanding and synthesis rather than recall, so the strategies that used to produce success are becoming less effective. His point is not only that rereading fails retention. It is that rereading does not produce understanding in the first place.

Two structural fixes come up. McDaniel wants retention itself measured — he says he cares less about what a student knows right away than about what they know and can use four months later, and that if we measured that, teaching and studying would both reshape themselves around it. He illustrates the gap with a business school dean who asked for help because students pass the statistics course and then cannot apply it when a later course needs it, so the faculty teach it again. The second fix is cumulative exams. McDaniel relays Roediger's position here: drop the cumulative exam and you have undercut learning before the course starts, because his studies point to students holding material better when they simply expect one to be coming — and McDaniel adds, carefully, that he may not have that detail right.

If it works this well, why has not every school adopted it?

Because it did not even spread inside the school where it worked. The authors spent about five years with a middle school in Columbia, Illinois, across the Mississippi from them, where three or four enthusiastic teachers ran classroom experiments. The experiments worked. Roediger expected the results to persuade the other teachers, and they did not. Colleagues kept teaching as before, maybe adding a little testing around the edges.

Part of it is cost to the teacher: quizzes have to be written and graded, and a full redesign is too big a change to make at once. Part of it is upstream. The authors note that this material is not taught in education schools — there may be a course in educational psychology that barely represents the work — even though they and many others have been at it for 20, 25 years.

Both of them think the change comes from the bottom, not the top. An administrator ordering every teacher to mix everything up would produce pushback; a single instructor who tries it and can show what happened becomes a mentor to the others. The algebra teacher's school went that way — once colleagues saw how well his students were doing, they came and asked how to do it, and the school's culture shifted from one classroom.

It also takes nerve. McDaniel is frank about imagining a teacher deciding to interleave algebra topics instead of doing a day of fractions and moving on, with students pushing back, parents asking whether it will work, and no proof until the semester is over. Nick's read is that the demand is building from underneath anyway. Andrew Huberman put out a solo episode on effective learning that passed a million views within a couple of days, and Nick's estimate is that 80 to 90% of it covers ground Make It Stick already covers. If you want the teacher-facing version, the authors point to Powerful Teaching by their colleague Pooja Agarwal and the teacher Patrice Bain, and to Daniel Willingham's books.

Method and sources: This guide is built from one episode of The Nick Standlea Show, Make It Stick: The Science of Effective Learning, and from its transcript. The spine is the episode's 22 key takeaways, each of which was checked against the recording; the detail and the examples come from the transcript itself, and timestamps are given where the recording supplies them. Every figure on this page is a figure said on the tape. Where a speaker hedged — one study, I may not have that detail right, I know of no strong evidence — the hedge is kept. Claims are attributed to the person who made them, and where the recording carries no speaker cue the claim is attributed to the authors jointly rather than to a guess: Henry Roediger and Mark McDaniel are different researchers with different specialisms, and several findings here belong to a named colleague or to another lab rather than to either of them, including the painters-style experiments by a colleague, the middle-school anxiety surveys led by Pooja Agarwal, and the term desirable difficulties, which Roediger credits to a psychologist at UCLA. One note on authorship: Make It Stick was written by three people — Peter C. Brown, Henry L. Roediger III and Mark A. McDaniel. Only Roediger and McDaniel took part in this episode. Brown was not a guest, and where he appears on this page it is because the authors mention him in the conversation.