Method
The SAT mistake log, done properly: why yours stopped working
· 9 min read
If you have studied for the SAT for more than a month, someone has told you to keep an error log. If you are like almost everyone, you started one, kept it faithfully for eight or nine questions, and have not opened it since. That is not a character flaw. The standard advice describes a habit that is structurally hard to sustain and, even when sustained, does less than it promises.
It is worth separating the two failures, because they have different fixes.
Failure one: the log asks for effort at the worst possible moment
Writing up a mistake takes five minutes of careful thought. You are asked to spend those five minutes immediately after getting something wrong, which is precisely when you least want to look at it again. The natural move — and every student makes it — is to note it quickly and move on to a question you can get right, because that feels like progress.
So the log fills up with entries like "careless error" and "forgot the formula", which are not diagnoses. They are ways of closing the tab.
Failure two: it logs the question, not the misconception
Most logs are organized by question: here is the problem, here is what I did, here is the right method. The trouble is that you will never see that question again. What you will see is another question built on the same misunderstanding, wearing completely different clothes.
A log organized by question gives you twenty entries that each look unique. The same twenty entries organized by misconception usually collapse into four or five recurring errors, which is a far more useful thing to know — and a far shorter list to fix.
| Logged as a question | Logged as a misconception |
|---|---|
| Q42: got the price after two discounts wrong | I add successive percent changes instead of multiplying the multipliers |
| Q17: wrong sign when solving for x | I move a term across the equals sign without flipping its sign |
| Q8: picked the wrong answer on a word problem | I solve for the variable and stop, without checking what was asked for |
The right-hand column is short, and each line is actionable. The left-hand column grows forever and tells you nothing you can act on before the next test.
Failure three: re-reading is not practice
Even a well-kept log is usually reviewed by reading it. You look at the entry, the correction feels obvious, and you conclude that you have fixed it. This feeling is unreliable. Recognizing a correct method when it is in front of you is a much easier task than producing it under time pressure on unfamiliar material, and the confidence you get from the first tells you almost nothing about the second.
There is a large body of work on this in cognitive psychology, and the practical conclusion is consistent: retrieval beats review. You have to make yourself produce the correction from scratch, on something you have not seen, for the practice to count.
What a working log looks like
- One entry per misconception, not per question. Merge duplicates ruthlessly — the merged count is the useful signal.
- Write the entry as a rule you broke, in the first person and in the present tense: "I do X when I should do Y."
- Note which wrong option you actually picked. On a good question each distractor corresponds to one specific error, so your choice names the misconception for you.
- Test each entry on a new question, not by re-reading the old one.
- Test it again after a few days, once you have started to forget. A correction you can produce on day three is a different thing from one you could produce ten minutes after reading the explanation.
Why we built this into the product instead of writing a guide about it
Every item on that list is achievable by hand, and approximately nobody achieves it, for the reason at the top of this piece: it asks for the most work at the moment of least motivation. So we are building it so that none of it is your job. Miss a question and the option you took is recorded for you, and you are told why that specific route fails. The rematch engine then generates a fresh question carrying the same trap, so you have to produce the correction rather than recognize it, and reschedules it — three days later, then a fortnight after that, and tomorrow instead if you miss. Connecting that engine to the question bank is what we are working on right now; the practice, the explanations, and the record of what you took are live today.
The method is not ours and is not secret — it is retrieval practice plus spaced repetition, aimed at your own errors instead of at a deck of facts. What is ours is that it runs whether or not you feel like maintaining it.
How many entries should an SAT mistake log have?
Fewer than you would expect. Logged by misconception rather than by question, most students converge on four to eight recurring errors per section. If your list is thirty entries long, it is almost certainly logging questions.
How often should I review my SAT mistakes?
Reviewing is the wrong verb — you want to be retested, not to re-read. Roughly three days after the correction, then again a fortnight after that retest, is a reasonable schedule: each interval is measured from the last successful recall, not from the original mistake, so the gap grows as the memory gets stronger.
Should I log questions I got right but was unsure about?
Yes, and this is under-used. A guess that lands is the same gap as a guess that misses; it just has not been punished yet. Log the reason you were unsure.
Is a paper notebook or a digital log better?
Neither format fixes the two real problems — that entries get written as "careless error", and that reviewing them is re-reading rather than retrieval. Pick whichever you will actually open, then make sure each entry names a rule you broke and gets tested on new material.
James Zhang
Founder, kaoshen.co