Two effects, both measured
Testing beats rereading
In 2006 Roediger and Karpicke had students study prose passages and then either take free-recall tests without feedback or restudy the material the same number of times. Tested five minutes later, the restudy group did better. Tested two days and a week later, the students who had been tested remembered substantially more than those who had restudied, even though restudying had left students more confident in their memory. The authors' conclusion is the reason flashcards exist: testing is a powerful means of improving learning, not just assessing it. Every card you answer is one of those tests.
Spacing beats cramming
The same year Cepeda, Pashler, Vul, Wixted and Rohrer published a meta-analysis of the distributed practice effect: 839 assessments in 317 experiments from 184 articles. Spaced learning episodes beat consecutive massed ones, and the review added a finding about timing: the interstudy interval that produced the best retention increased as the retention interval increased. In plain terms, the longer you need to remember something, the longer the gaps between reviews should be.
The ranking
Dunlosky and colleagues reviewed ten study techniques in 2013 for Psychological Science in the Public Interest. Two received a high utility rating because they help learners of different ages and abilities across many kinds of material: practice testing and distributed practice, which are exactly what a flashcard routine combines. Elaborative interrogation, self-explanation and interleaved practice were rated moderate. Five were rated low utility: summarization, highlighting, the keyword mnemonic, imagery for text learning and rereading, the last two being what most students report doing most. The authors' suggestion for rereading is practice testing in its place.
An expanding schedule
A spaced schedule reviews a card soon after learning it and then at growing intervals. The numbers below are a common convention, not a measured optimum, chosen so that each gap is roughly double the last; what matters is the shape. Starting on Nov 2:
| Review | Days since the last review | Day | Date |
|---|---|---|---|
| 1 | 1 | 1 | Nov 3 |
| 2 | 3 | 4 | Nov 6 |
| 3 | 7 | 11 | Nov 13 |
| 4 | 14 | 25 | Nov 27 |
| 5 | 30 | 55 | Dec 27 |
| 6 | 60 | 115 | Feb 25 |
Six reviews carry a card 115 days, almost four months, and the last three of them are weeks apart. A card you get wrong goes back to the start of the ladder; a card you get right climbs it. That rule is what makes the daily workload manageable: most of a mature deck is due rarely, and the cards you keep missing are the ones you see most.
The five-box method on paper
The same idea without an app. Five boxes, each reviewed on its own rhythm; every new card starts in box 1. Right answers move a card one box up, wrong answers send it back to box 1.
| Box | Reviewed | Reviews of a card that stays there for 90 days |
|---|---|---|
| 1 | every day | 90 |
| 2 | every 3 days | 30 |
| 3 | every week | 12 |
| 4 | every 2 weeks | 6 |
| 5 | every month | 3 |
A card that reaches box 5 is seen 3 times in three months instead of 90, which is where the time saving comes from. The flashcard maker runs the sorting for you: mark each card Know or Don't know, repeat only the missed ones until none are left, and print cut-out cards if you prefer the boxes.
Writing cards that teach
- One fact per card. A card with three answers is three cards; you will get one part right and mark the whole thing wrong, or the reverse.
- Ask a question, do not show a heading. The front has to force recall. "Capital of Australia?" works; "Australia" does not.
- Write the answer in your own words, short enough to check at a glance.
- Make both directions when both matter. Vocabulary needs foreign to native and native to foreign; the flashcard maker's Show first: Back option studies the reverse.
- Add context to lonely facts. A date with a one-line reason why it mattered is recalled more easily than the date alone.
- Delete cards you always know. A deck full of easy cards hides the hard ones.
A routine that survives the semester
- Review before you add. Due cards first, new cards only if there is time; new cards are tomorrow's reviews.
- Cap new cards per day and lower the cap when the daily session grows past what you will do every day.
- Answer out loud or on paper before flipping. Recognising the answer once it is shown is not recall.
- Keep sessions short and daily. The spacing effect rewards many short sessions over one long one; a focus timer set to twenty or twenty-five minutes is a natural unit, and two of those beat one hour-long sitting.
- Match the gaps to the deadline. For an exam months away, let the intervals grow; for a test this week, keep them short. Cepeda and colleagues' finding is the rule: the longer the retention interval, the longer the optimal spacing.
Frequently asked questions
What is spaced repetition?
Reviewing material at increasing intervals instead of all at once. The research name is the distributed practice effect: Cepeda and colleagues' 2006 meta-analysis of 839 assessments in 317 experiments found that spaced learning episodes beat consecutive massed ones, and that the interval between study sessions which produced the best retention grew as the time to the final test grew. Flashcards are the simplest way to apply it, because each card can carry its own schedule.
Why are flashcards better than rereading?
Because answering a card is a test, and testing improves memory more than restudying. Roediger and Karpicke showed in 2006 that students who took recall tests instead of rereading remembered more after two days and a week, even though rereading had looked better after five minutes and left students more confident. Dunlosky and colleagues' 2013 review rated practice testing and distributed practice as the two high-utility learning techniques and rereading and highlighting as low utility.
How often should I review flashcards?
Often at first, then less and less. An expanding schedule such as 1, 3, 7, 14, 30 and 60 days is a common pattern; the exact numbers are a convention, not a finding. Cepeda and colleagues' result points the other way too: the longer you need to remember something, the longer the gaps between reviews should be, so cramming the night before fits a test tomorrow and not an exam in June.
What is the Leitner system?
A way to run spaced repetition with paper cards and a few boxes. Every card starts in box 1, which you review daily. A card you get right moves up a box, and each higher box is reviewed less often; a card you get wrong goes back to box 1. The schedule emerges from the sorting: cards you know drift to the rarely reviewed boxes, cards you do not know keep coming back.
How many cards a day is realistic?
The daily load is the reviews due, not the deck size, and with an expanding schedule most old cards are due rarely. New cards are what set the pace: each one you add today comes back several times over the next weeks. Start with ten to twenty new cards a day, watch the review count, and lower the intake if the daily session passes what you will actually do every day. A missed week is worse than a slower start.
Sources
- Cepeda, Pashler, Vul, Wixted and Rohrer (2006), Distributed practice in verbal recall tasks: a review and quantitative synthesis, Psychological Bulletin (839 assessments in 317 experiments; spacing beats massing; the optimal interstudy interval increases with the retention interval)
- Roediger and Karpicke (2006), Test-enhanced learning: taking memory tests improves long-term retention, Psychological Science (repeated studying wins at 5 minutes; prior testing produces substantially greater retention at 2 days and 1 week, despite higher confidence after studying)
- Dunlosky, Rawson, Marsh, Nathan and Willingham (2013), Improving Students' Learning With Effective Learning Techniques, Psychological Science in the Public Interest (practice testing and distributed practice rated high utility; elaborative interrogation, self-explanation and interleaving moderate; summarization, highlighting, the keyword mnemonic, imagery for text and rereading low)