Thought analysis: what your working reveals that your score doesn't
A score of 640 tells you forty questions went wrong. It does not tell you that eleven of them were questions you had right and talked yourself out of.
Thought analysis is the practice of studying how a question was answered rather than whether it was answered correctly. It reads the record of your working — the text you highlighted, the choices you struck out, the answer you changed, the seconds you spent — and identifies the habit that produced the result. A score report throws all of that away and keeps a single number.
Why the score is the least useful thing about an attempt
Two students score 600 on Reading and Writing. The first is steady: they get the easier material right and run out of knowledge somewhere around the harder Craft and Structure questions. The second is erratic: they answer the hardest inference questions correctly and drop four transitions questions they could do in their sleep, because they were rushing by question 22.
These are the same score and opposite problems. The first student needs to learn more. The second needs to slow down for ninety seconds. A score report cannot tell them apart. The record of how they worked can, immediately.
What gets recorded
Every action you take on a question is evidence about your reasoning:
| What is recorded | What it reveals |
|---|---|
| Text you highlighted | Whether you found the part of the passage that actually decides the answer, or marked the part that merely sounded important |
| Choices you struck out | Whether you eliminate on evidence or on feel — and whether you eliminated the correct answer early |
| Answer changes | Whether second thoughts help you or cost you. For most students it is one or the other, consistently |
| Time per question | Where the minutes go, and whether the questions you rush are the ones you get wrong |
| Order and revisits | Whether flagging and returning is working, or just fragmenting your attention |
The four patterns worth knowing about
Right and quick
Fluent. This is knowledge you own. It is also the evidence that you can afford to spend time elsewhere — a student who answers half the section in forty per cent of the time has twenty minutes they are not using.
Right but slow
The most under-diagnosed state on the test. You got it right, so the score says nothing is wrong. But a question that takes you two and a half minutes untimed is a question you will get wrong on test day, because on test day you will not have two and a half minutes. These are worth drilling for speed rather than for accuracy — which is a completely different kind of practice.
Wrong and instant
A snap answer is closer to a guess than to a considered miss. It usually means you recognised a surface feature — a word from the passage appearing in a choice, a familiar-looking equation — and answered the pattern rather than the question. The fix is not more content knowledge. It is refusing to answer until you have read all four choices.
Wrong after changing from the right answer
The most painful one, and the easiest to fix once you can see it. If your log shows you routinely change correct answers to incorrect ones, the instruction is simply: stop. Trust the first reading unless you find a concrete reason in the text.
What eliminations say that nothing else does
Striking out choices is the one part of test technique that is visible in the record and invisible in the score. A student who strikes two choices and picks correctly between the remaining two is reasoning. A student who strikes nothing and picks is guessing with extra steps. And a student who strikes the correct answer, then picks from what is left, has a specific misconception that can be named — you can look at what they eliminated and see what rule they think they are applying.
How to do this without software
You do not need a tool to start. Sit a paper, and for every question you get wrong, write one sentence in a notebook answering: what did I think the question was asking? Not what the right answer was — what you believed at the time. After twenty questions the repeats become obvious, and repeats are what you fix.
What software adds is the part you cannot reconstruct from memory: the timings, the order, and the fact that you changed your answer on question 14 nine seconds before the module ended.
Questions people ask
- Isn't this just reviewing your mistakes?
- Reviewing mistakes tells you the right answer. Thought analysis tells you the route you took to the wrong one. The second generalises to questions you have not seen; the first does not.
- Does it work for Math as well as Reading?
- Yes, though the evidence differs. On Math the useful signals are time, answer changes and whether you attempted the question at all. On Reading and Writing, highlighting and elimination carry most of the information.
- How many attempts before the patterns show?
- Two full papers, or roughly a hundred questions. Fewer than that and you are reading noise.
Colorbook records all of this while you practise, and the report after each paper reads it back to you. It is free, and it does not need an account.
2,079 questions, full-length adaptive mocks, and a report on how you worked.
Start practising — free, no account