Studying for a math test like it is a vocabulary list is why it goes badly, so here is what to do instead.
You read through the worked examples in your notes and every line makes sense. Then the test is in front of you. On question three you can’t work out where to start.
That’s not a memory problem, and it doesn’t mean you’re bad at math. You studied math the way you’d study a vocabulary list.
Flashcards are great for a fact with one right answer and no steps in between. An element symbol, a date, a definition. Very little of a math test looks like that. A math question asks for a sequence of moves under time pressure. The only practice that builds that is producing the moves.
When you read a solved problem, every step arrives already justified. The hard bit of a math problem is choosing what to do next. A worked example made every one of those choices before you got there. You’re watching somebody else drive.
There’s a two-minute test for this. Pick a problem you’re sure you understand. Cover the solution and do it on blank paper. If you can’t get the first line down, what you had was recognition.
This isn’t a math-only trap. In Roediger and Karpicke’s experiment, students who re-read a passage beat students who tested themselves when the test came five minutes later. A week later, the re-readers lost badly. Your notes feel smooth because you’ve already seen them.
The best practice problems you’ll find are on paper you already own. Homework, the quiz you got back, the review sheet. The test is written by the person who assigned all of that, and the structures repeat even when the numbers don’t.
Take the last three homework sets and the last quiz. Copy the problems onto one page, without their solutions. Homework in its original form is unusable, because the answers are sitting right there in your handwriting. A bare page of problems is a practice test you can sit cold.
Weight it toward what you got wrong or slow. One two-step equation is enough if you got them all first time. If all four systems-of-equations problems went wrong, put six on the page.
Then shuffle the order. Textbooks give you twenty factoring problems in a row. By the third one you’ve stopped reading the problem, because you already know the method. A real test doesn’t tell you the method. Rohrer and Taylor found that students who practiced mixed problem types beat students who practiced them in blocks, on a test a week later. The blocked group looked better at the time.
Most points lost on a math test don’t go to “I had no idea how to do this.” They go on one step inside a problem you otherwise handled. You set up the integral correctly and dropped a minus sign. You found the right common denominator and added the numerators wrong. Doing more whole problems to fix one broken step spends nine-tenths of your time on what already works.
Say you keep botching distribution over a subtraction. You turn −3(x − 4) into −3x − 12 instead of −3x + 12. Don’t do twenty full equations. Write out twelve distributions and nothing else, and give yourself two minutes. You hit the broken step twelve times. Full problems would have given you two.
The same works higher up. Say your chain rule keeps coming out as 5(3x + 1)4 with the inner derivative missing. Don’t do ten related-rates problems. Write ten expressions of the form (something)n. Differentiate only those, and check that the factor of 3 is on every line. Ten minutes. That’s the difference between studying three hours and missing the same thing again, and fixing it before dinner.
One part of math does behave like vocabulary. It’s the list of things you’re expected to have memorized. The quadratic formula, the derivative of tan x, sin2 θ + cos2 θ = 1, the value of sin(π/6). Those belong on cards. Word each one so the card asks you for the formula itself.
The first card memorizes a pair of numbers you’ll be shown once. The second pair memorizes a tool and when it fires, and that’s what the test checks. Writing flashcards that work has the general rules, and they apply here unchanged.
If a formula is still escaping you the night before, the real problem is older than tonight. It wasn’t on a card three weeks earlier. Ten minutes a day beats a long evening by a distance that’s hard to believe until you’ve done it. There’s a week-by-week version of that in how to study for a test.
A skill you’ve only done slowly, with your notes open and no clock running, isn’t the skill the test asks for. Three or four days out, sit a full set in one go. Timer running, notes closed, and no checking the answer after each question.
Work out your pacing from the real numbers first. Twenty questions in a forty-minute period is two minutes each. A problem you’re five minutes into is one to leave and come back to. Mark it in pen afterwards, and write the time beside anything slow. An answer that was right and took six minutes isn’t finished.
When you mark your practice, the score is close to useless. What you want is the first line where your work stops matching a correct solution. Circle it. That line is the bug, and everything after it is consequences. Nearly every circled line is one of four things, and they need four different fixes.
Labelling each wrong answer takes two minutes and decides your week. Three arithmetic slips and one method error is a completely different week from four method errors.
If you run out of homework problems to drill on, Machi.study has a Problems mode. It writes a fresh problem on a topic each time, and shows the worked steps once you have answered.
Make the formula cards over the weekend, then study them ten minutes a day. Early in the week, build the problem page and sit it cold. Mark it, and sort every wrong answer into the four categories. Spend the middle of the week on two-minute drills of the broken steps. Three days out, do a timed set. The night before, one pass through the cards, then sleep.
None of that takes longer than what you were going to do anyway. It spends the hours on what’s broken instead of on what already feels good.
Next: why cramming feels like it works, or how Machi.study decides when to show you a card.