Deliberate Practice for Studying: The Honest Version

Deliberate practice is a narrow and demanding kind of training: individualised tasks selected by a supervising teacher, a clear performance goal, and immediate informative feedback. A 2014 meta-analysis found it explained 4 percent of the variance in educational performance. The same authors later published 5 percent. Either way, that is the number nobody quotes.

Which makes this one of the odder pieces of study advice in circulation. The phrase arrives attached to a promise about 10,000 hours and expert performance, and the promise is doing almost all of the persuading. The research underneath it is smaller, stricter, and openly disputed by serious people. It also leaves behind something genuinely useful, once you stop trying to use it as a theory of genius and start using it as a checklist.

What is deliberate practice, exactly?

The strict definition is the one worth learning, because it is the one the evidence was gathered under. Writing in Perspectives on Psychological Science in 2016, K. Anders Ericsson set out what the 1993 review he co-authored had concluded about the conditions of optimal learning: individualized practice with training tasks (selected by a supervising teacher) with a clear performance goal and immediate informative feedback was associated with marked improvement.

Read that slowly, because there are three separate requirements hiding in one sentence.

• Somebody else picks the task. A supervising teacher who can see your current performance chooses what you work on next. You do not choose it yourself.
• The task has a defined target. A clear performance goal, meaning a specific thing you are trying to do better, never a general intention to improve.
• The feedback arrives immediately. Not at the end of term, not when the paper comes back. Straight away, and informative enough to tell you what to change.

Now hold an ordinary revision session against that list. You picked the chapter. There was no target beyond covering it. The feedback, if any, comes weeks later in a grade. Ordinary studying fails all three conditions, which is not a small shortfall from the thing the research measured. It is a different activity.

Does deliberate practice work for studying?

Yes, and the effect in education is small enough that the honest answer has to lead with the number.

Brooke Macnamara, David Hambrick and Frederick Oswald gathered the evidence across five domains and published the result in Psychological Science in 2014. Their abstract reports that deliberate practice explained 26% of the variance in performance for games, 21% for music, 18% for sports, 4% for education, and less than 1% for professions. Education is the second-weakest domain on that list. The conclusion they draw from it is a single clause: deliberate practice is important, but not as important as has been argued.

There is a complication with those figures, and skipping it would be the easy thing to do. The 2014 article carries a corrigendum, published in Psychological Science in 2018. We could not read it, because the publisher serves that page behind a bot check. What we can read is the same three authors restating their own meta-analysis six years later in Frontiers in Psychology in 2020, pointing readers to that corrigendum as they go, and the numbers they give there are not the ones in the 2014 abstract. Identifying 88 studies, they write, they found that deliberate practice explained 14% of the variance in performance overall, and 24% for games, 23% for music, 20% for sports, 5% for education, and 1% for professions.

For a student the disagreement between 4 and 5 percent changes nothing. Both are small, and the ranking holds: education is near the floor while games and music are several times higher. The reason to show both is that this post is about a claim that got oversold, and quoting the more dramatic of two published figures without saying the other exists would be the same trick in the other direction.

What that percentage does not mean

This is the single most misread number in the whole debate, and getting it wrong would leave you with exactly the wrong conclusion.

The 4 or 5 percent is a between-person figure. It describes how much of the difference between one student and another is tracked by how much they practised. It is not a statement about how much you can improve. Those are separate questions, and the 2020 review is explicit about which one is contested. The authors set out two claims they call uncontroversial: that expertise is acquired gradually, and that training can lead to large, even massive, improvements. The contested question is a narrower one, about how far differences between people in accumulated training explain differences between people in performance.

The illustration they give is worth keeping. In a classic study from 1980, after more than 200 h of training, a college student improved his performance in a random digit memorization task from a typical 7 digits to 79 digits. Nobody in this argument disputes that training transforms what an individual can do. What they dispute is whether counting up practice hours explains why some people end up better than others.

So the small percentage is not permission to stop working. It is a warning that hours logged is a poor measure of anything, which is the same conclusion our guide to how many hours you should study a day reaches from the opposite direction.

Where does the 10,000 hour rule come from?

From a book, mostly, and the tidy version of that story is itself slightly wrong.

The 2020 review traces the line directly. Taking his inspiration from the 1993 findings, Gladwell wrote that 10,000 h is the magic number of true expertise, in the 2008 bestseller Outliers. That framing, a magic number that converts hours into expertise, is Gladwell.

The figure itself was not conjured out of nothing, though, and this is the part that usually gets dropped by people correcting the myth. The same review quotes Ericsson and co-authors writing in Harvard Business Review in 2007 that even the most gifted performers need a minimum of 10 years (or 10,000 h) of intense training before they win international competitions. A 10,000 hour figure does appear in the writing of the researchers themselves. The 2020 review also records where it came from: the 1993 study found that the most accomplished musicians reported having accumulated an average of around 10,000 h of practice alone by early adulthood.

Watch what happened to that number across those three steps. It starts as an average observed in a group of accomplished musicians, becomes a minimum for winning international competitions, and ends up a magic number for expertise itself. Meanwhile the actual definition of deliberate practice, the one quoted at the top of this post, contains no hour count at all. It specifies the conditions practice has to meet, and says nothing about how much of it you need. For a student the practical reading is blunt: hours are the input you can most easily fake to yourself, and they are the one thing the definition never asks about.

Why the framework itself is contested

Most articles about deliberate practice either sell the theory or debunk it. The accurate position is that this is a live, unresolved, and fairly sharp argument between two groups of researchers, and a student is better served by seeing the shape of it than by being handed a verdict.

The critics. After the 2014 meta-analysis, the Macnamara group published a second one on sports, and then, in a 2016 reply to Ericsson, stated that the view attributing individual differences in expertise largely to accumulated deliberate practice is not supported by the available empirical evidence. They found that deliberate practice accounted for a sizeable amount of variance in sports performance (18%), but it left a much larger amount unexplained. Their 2020 review goes further and attacks the theory as a theory: they document critical inconsistencies in the definition of deliberate practice, along with apparent shifts in the standard for evidence concerning deliberate practice, focusing on the empirical testability and falsifiability of the deliberate practice view. One illustration they give: after the 2014 meta-analysis appeared, Ericsson rejected 87 of the 88 studies that it included.

One detail from the 2020 review deserves singling out, because it lands directly on advice aimed at students. The critics note an inconsistency on whether a teacher is required to design deliberate practice, which is why they admitted both teacher-designed and performer-designed activities into their analysis. The very requirement that makes the definition demanding, the supervising teacher, is the one both camps have argued about.

Ericsson. His answer is that they measured something he never defined. His 2016 commentary replies to the sports meta-analysis, not the five-domain one that supplies the 4 percent, and he objects that it examines the use of the term deliberate practice to refer to a much broader and less defined concept including virtually any type of sport-specific activity, such as group activities, watching games on television, and even play and competitions. Summing up every hour of any type of practice, he writes, treats the impact of all types of practice activity on performance as equal, an assumption he says he has shown to be inconsistent with the evidence. On this reading a meta-analysis of loosely defined practice is no test of tightly defined practice, and since he rejected 87 of the 88 studies in the five-domain analysis as well, he reads its low numbers the same way: an artefact of the coding.

We are not going to adjudicate that, and you should be suspicious of any study blog that does. What matters for revision is that both readings survive the exchange. If Ericsson is right, the framework is strong and almost nothing students do qualifies for it. If the critics are right, the framework is weaker than advertised and the practice-hours model of achievement is wrong. Either way, logging hours of comfortable revision is not the activity that pays.

What survives, and what to do on Tuesday night

Strip out the contested claims and a small residue is left standing, agreed by both sides. Even while arguing that the theory is not scientifically defensible, the 2020 review grants that the deliberate practice view has important value in society, serving as a useful reminder to the layperson that training of some form is necessary to achieve a high level of performance in a domain. And nobody disputes the two conditions that make practice productive, because those come from the definition itself.

Two of the three original conditions are available to a student working alone.

1. Set a specific performance goal for the session, before it starts. Cover chapter four is not a goal. Write out the full glycolysis pathway with no notes, and get all ten steps in order, is one. You can tell the difference by whether you could be wrong about having achieved it. Our guide to setting study goals covers how to pick targets at this grain.
2. Make the feedback immediate and specific. Immediate in the definition means within seconds, which rules out most of what revision consists of. A quiz that marks each answer as you give it meets that literally. Marking a past paper against its scheme the same evening does not, though it is the realistic compromise for work that cannot be auto-marked, and it beats a grade three weeks later by a distance. Reading a chapter and feeling that it went well supplies no feedback whatsoever, which is the whole subject of the illusion of competence.
3. Substitute for the teacher with your own error log. This is the condition you cannot properly meet, so approximate it. The tasks a teacher would choose are the ones you keep getting wrong, and a written list of your own mistakes is the nearest available source of them. Reviewing a paper properly is the standard way to build that list, and how to review a practice test walks through it.

None of this is exotic. It overlaps heavily with the desirable difficulties, which arrive with a broader and better-behaved evidence base, and if you only adopt one framework this year, adopt that one. Deliberate practice earns its place here as a sharp question to ask of any study session: what is the target, and how fast will I know?

Where GeniusPal fits, and where it does not

The honest scope is narrow, so here it is with the limits first. The strict definition needs a supervising teacher selecting your tasks, and software is not a supervising teacher. GeniusPal cannot watch you work, diagnose why you went wrong, or decide what you should do next week. Any tool claiming to deliver deliberate practice as Ericsson defined it is overselling, and that includes this one.

What it can supply is the other two conditions, which happen to be the two ordinary revision drops. You upload your own notes and get questions back from them, so a session has a defined target instead of a chapter to get through, and every answer is marked the moment you give it. That is the immediate informative feedback half of the definition, applied to material you chose.

The error-log substitute is the daily review, which is assembled from the questions you have missed across every set you own. That is as close as a piece of software gets to the teacher-selected task: the list is built from evidence of what you actually got wrong, instead of from what you felt like revising. It costs nothing on any plan, which matters, because a feature that only paying accounts can reach is no use as general advice.

The plan detail, so you can judge the fit. Free gives two study-set generations for the account lifetime, a free set runs to ten questions and paid sets go up to thirty, and Free covers two full sittings of each set it makes. Quiz mode and the daily review cost nothing on any plan. Flashcards and written active recall are Student and up. On Student and Genius a quiz can also be set to a written mix, with settings called A few written, Half and half, and All written, which is the setting worth knowing about here: typing an answer from nothing is a harder performance goal than picking one from four. Student is 14.99 dollars a month and Genius is 59.99 dollars a year, shown as 5.00 dollars a month.

What it will not do is read an image-only file, so a photographed page is refused instead of turned into questions. And it has no idea what will be on your exam.

What this post does not show

Four limits, stated plainly, because this topic attracts very confident writing.

Three of the four sources here were read as abstracts only: the 2014 meta-analysis, the 2016 commentary, and the 2016 reply. Nothing above describes a method, a sample or a moderator that those abstracts do not state. The 2020 review was read in full, which is why the material drawn from it is more detailed.

The corrigendum to the 2014 article was not read at all. We report the 2014 abstract figures and the 2020 restatement side by side and leave the discrepancy open, because guessing which set supersedes the other would be inventing a fact.

Where this post quotes what the 1993 founding paper reported, it is relying on the later papers that describe it. That paper was not fetched, so treat those sentences as reported speech.

And the education figure is one number from one meta-analysis, in a literature whose central definition is actively disputed. It is a good reason to stop treating deliberate practice as the master key to academic performance. It is a poor reason to stop practising, for the between-person reason set out above: a statistic about the spread of outcomes across students has remarkably little to say about what tonight will do for you.