Cramming: what works the night before, and what is already lost

Cramming works, briefly: one evening of testing yourself on a short list of ideas can raise tomorrow's mark, and most of that gain is gone within a week. Rereading the whole chapter tonight buys almost nothing. So triage: list the ideas, test yourself once on each, and spend the night attempting problems only on the ones you got wrong. Then sleep. Here is the evening, hour by hour. If the test is next week rather than tomorrow, start with the full week's plan instead; you have time this page assumes you do not.

Does cramming work? The honest answer

Both halves are true, and you need both.

Cramming is what researchers call massed practice: all the study in one block, close to the test. Across 254 studies of massed against spaced practice (Cepeda et al. 2006), massed study did produce learning. What it did not produce was learning that lasted. The same hours spread over separate days gave better recall at every delay the studies measured, and the gap grew as the delay grew. So an evening of cramming raises tomorrow's performance, and by next week most of the difference it made has gone.

The second half is the one students do not expect. Kornell (2009) had learners study flashcards either spaced out or massed, then tested them. Spacing beat massing for about nine in ten of them. Yet when asked which had worked better, most learners said massing. Cramming does not only fade; it feels more effective than it was while you are doing it, because the material is fluent in the moment. That feeling is not a measurement.

Tonight, cramming is not useless. As a way to learn the course, it is a poor one. The rest of this page is the first half; the last section is the second.

What is already lost tonight, and what is not

Being honest about what one evening cannot do is what makes the evening work, because it tells you where the hours go.

Lost, or nearly so:

• Deep understanding of an idea you have never met. You can learn a procedure's steps tonight; you cannot build the understanding that lets you handle the question in a shape you did not practise.
• Long-term memory of tonight's work. It will serve tomorrow and fade after.
• The sleep, if you take it. The last hours of an evening of study are worth less than the sleep they replace, which is the subject of the all-nighter guide below.

Not lost:

• Recall of the ideas you half know. A rule you learned three weeks ago and have not used is a retrieval away, and one honest attempt tonight brings it back for tomorrow.
• Procedures you have done before. A titration calculation you did in a lab is a worked example and a cold redo from being reliable.
• The easy marks. Definitions, set-ups, the first line of every calculation. These are cheap to secure tonight and are usually a third of the paper.

So the evening goes to the ideas you half know and the procedures you have met, not to the chapter you skipped. A new topic at midnight is a bad bet; a half-known topic at 7 pm is a good one.

The evening, hour by hour

A four-hour evening starting at 6 pm, for a test tomorrow morning. Clock times are there so you can see where the stop is.

1. 6:00 pm — The list (15 min). Write every idea the test covers as a question, one line each, from the course outline or the section headings. Not from the notes; from the headings. Ten to fifteen questions is usual for a unit test.
2. 6:15 pm — The diagnostic (45 min). One attempt at each question with the notes closed: a problem, or a blank page with the idea's name at the top. Mark honestly into three piles: can do, shaky, cannot. How to test yourself honestly has the marking rules; the short version is that "I would have got it" is the shaky pile.
3. 7:00 pm — Break (15 min). Eat something. Away from the desk.
4. 7:15 pm — The "cannot" pile (75 min). For each idea: read one worked example slowly, close it, redo the same problem from the question alone, then do a sibling problem with different numbers. Write every wrong line in an error log with the right line beside it. Three ideas in 75 minutes is a fair pace; if the pile is longer, take the ideas worth the most marks first.
5. 8:30 pm — Break (15 min).
6. 8:45 pm — The "shaky" pile (45 min). Recall first: the idea's name at the top of a blank page, write what you know, then one quick problem each. Add to the error log.
7. 9:30 pm — The error log, once (10 min). Read each wrong line, say the right line before you uncover it. Then stop.
8. 9:40 pm — Stop. Pack the bag, screens off, and be asleep by about 10:30. The stop time is not a suggestion; what one night without sleep does is the reason it is in the plan.

That is a little over three hours of focused work. It covers more than six hours of rereading would, because every minute is an attempt that gets marked, which is the thing the research says produces recall (Roediger & Karpicke 2006), and the sleep after it stores the work.

Worked example: a chemistry test on acids and bases tomorrow

A Grade 12 chemistry (SCH4U) student has a test tomorrow on two ideas from the acids-and-bases unit: acid–base theories with pH and strong versus weak acids, and buffers with titration. The 6:00 list has twelve questions. Here is how two of them sort, and what the 7:15 block does with the one that fails.

The diagnostic question for pH: what is the pH of a 0.010 mol/L solution of HCl? HCl is a strong acid, so it dissociates completely and [H⁺] = 0.010 mol/L. Then pH = −log(0.010). Since 0.010 = 10⁻², log(0.010) = −2, so pH = 2.00. The student writes exactly that in under a minute. Can do. One check later in the evening and nothing more.

Two common wrong lines a marker sees on this question, for the error log if either is yours: treating HCl as a weak acid and setting up an equilibrium table (the strong acid needs no table; the concentration of H⁺ is the concentration of the acid), and writing pH = 1.00 by miscounting the decimal place (0.010 is 10⁻², not 10⁻¹).

The diagnostic question for buffers: a buffer is 0.10 mol/L in acetic acid and 0.10 mol/L in sodium acetate, and the question gives the acid's pKa as 4.74. What is its pH? The student does not know where to start. Cannot. This idea goes first in the 7:15 block.

The 7:15 block on buffers:

1. One worked example, read slowly. The buffer equation: pH = pKa + log([base]/[acid]). Here the base is acetate and the acid is acetic acid, in equal amounts, so the ratio is 0.10 / 0.10 = 1, log(1) = 0, and pH = 4.74 + 0 = 4.74. A buffer with equal acid and base sits at its pKa.
2. Redo it cold. Cover the example and write it from the question alone. The student forgets which concentration goes on top. That is the wrong line for the log: "buffer equation: base over acid".
3. A sibling with new numbers. The same acid, pKa 4.74, but 0.20 mol/L acetate and 0.10 mol/L acetic acid. Ratio 0.20 / 0.10 = 2, log(2) = 0.30, so pH = 4.74 + 0.30 = 5.04. More base than acid, so the pH is above the pKa, which is the check: a buffer with more base is less acidic.

Three attempts, one wrong line logged, and the idea has moved from cannot to shaky in about twenty-five minutes. The 9:30 error log reads "base over acid" once more, and that is what the student carries into the room.

What not to do tonight

• Reread every page. It is the weakest method in the research on study techniques and it eats the evening.
• Rewrite the notes neatly. Copying is not recall. It feels like work because it is work; it is not learning.
• Watch a two-hour video. Watching is reading at someone else's pace.
• Start a new topic at midnight. It will not be understood by morning and it will cost the sleep that stores the rest.
• Caffeine past about 8 pm. It is still in you at 11.
• The all-nighter. The hours after 11 pm are low-quality hours, and they cost the recall and attention you need at 9 am; the all-nighter guide has the measured costs.
• Open an answer app and copy solutions. In our own reading of the review banks, the answers on answer apps are frequently wrong, and tonight you have no way to check which. A copied solution is also not an attempt, so it teaches nothing you can retrieve tomorrow.

One health line: if you cannot sleep most nights, not only the night before a test, that is worth telling a doctor or the school's counsellor. This page is about one evening.

Next time: the calendar that makes cramming unnecessary

The spacing research turns into dates. The rule from Cepeda et al. (2006) is that an idea comes back on separate days. A follow-up study of over a thousand people (Cepeda et al. 2008) asked how far apart the days should be, and found that the best gap grows with the time until the test: for a test about a week away the best gap between two study sessions was about one day; for a test about five weeks away it was about eleven days; and for tests two or more months away it was about three weeks. A gap that is too long costs a little; a gap that is far too short, which is what cramming is, costs a lot.

Days to the test	Times each idea comes back	Gap between visits
7	At least 2	1–2 days
30	3	About 10 days
70	3–4	About 3 weeks

For a test in a week, that is the week plan linked at the top. For a term, it means a session every few days in which old ideas come back, which is a calendar, not a feeling. The spacing schedule shows how to keep that calendar by hand or with cards, and a study plan you will keep fits it into a week that also has homework in it. Chapter 2 of the free playbook turns the same table into a calendar you can fill in.

Tonight, on one screen

1. The list as questions, from the headings, 15 minutes.
2. The diagnostic, notes closed, three piles.
3. The "cannot" pile: worked example, redo cold, sibling, error log.
4. The "shaky" pile: recall, one quick problem each.
5. The error log once, then stop by about 9:40 pm.
6. Asleep by 10:30. The sleep is part of the plan.