You had it. You marked the right answer, felt a flicker of doubt on review, switched — and missed a question you had already solved. If that sequence sounds familiar, you're not "bad at tests": you have a specific, measurable habit called the first-instinct override, and a ten-minute self-test tells you whether it's live.
What the first-instinct override is
The first-instinct override isn't a question type or a trap category. It's a property of your process: your first read picks the correct answer, and your review talks you out of it. It's one of the three behavioral leaks — habits that cost points independent of what you know — alongside anchoring and the time-sink.
Two details make it uniquely nasty. First, it strikes hardest on questions where two answers feel possible — exactly where doubt feels most justified. Second, it's invisible. Your score report grades your final answer and shows a miss. It doesn't show that you held the point first and handed it back. You can review every practice test you've taken and never see it: the evidence — your first pick — was erased when you changed it.
The honest science on changing answers
There's a real research literature here, usually filed under the "first instinct fallacy," and its headline finding surprises most students: on average, answer changes help. Studies of answer-changing — including eraser-mark analyses going back decades — find that changes go wrong-to-right about twice as often as right-to-wrong. The classic advice — "never change your answer, trust your gut" — is a myth; the fallacy in "first instinct fallacy" is the folk belief itself.
But the average is not you. A minority of test-takers shows the reverse pattern — their changes hurt more than they help — and the research does not say who is in that minority. The only way to know your group is to measure your own pattern, which takes one practice section and ten minutes.
Both pieces of folk advice fail: "never change answers" is wrong on average, and "review freely" is wrong for the minority. Measure first, then decide.
Why wrong answers win the re-read
If your first pick was right, why would a second look make it worse? Because of how wrong answers are built.
A well-made distractor's entire job is to feel plausible, and the most durable kind is the true-but-irrelevant choice: a statement that is factually accurate, consistent with the passage, and not an answer to the question asked. On a first read it loses, because you're checking each choice against the text and it doesn't answer the question. On a doubting re-read, the contest changes: now you're comparing choices against each other, hunting for reasons to switch, and "this one is definitely true" feels like a reason. The trap was engineered to survive exactly that scrutiny. The correct answer, meanwhile, is right for one boring textual reason — and boring doesn't flatter a doubting mind.
That's the override in slow motion: your first read chose for text reasons, your re-read switched for plausibility reasons. The trap didn't beat your reading — it beat your reviewing.
The self-test: grade the section twice
Run this on your next timed practice section — or on past sections, if you kept your work.
- As you take the section, record your first pick on every question (a small mark on scratch paper works). Then review and change answers the way you normally would.
- Grade it twice: once as if your first picks were final, once with your actual final answers.
- Compare the two scores.
Reading the result:
- Finals win. Your reviewing works. Keep changing answers when you have a reason — the 2:1 average is describing you.
- Firsts win. The leak is live. Every point of that gap is a point you earned and gave back.
- Tie, or one question apart. Not enough evidence. One question is noise, not a verdict — run it again on another section.
Check two or three sections before concluding anything — you're looking for a pattern, not a bad day.
What the leak costs
Points lost to the override are the purest form of wasted points: no knowledge gap, nothing new to learn, because you already solved the question. Worse, because the miss looks ordinary, students respond by studying more content — which cannot fix a reviewing habit. That mismatch is one way a score plateau survives months of honest studying: the work is real, but it's aimed at the wrong kind of miss. And the scale isn't trivial — two or three overridden answers in a section is on the same order as the test's entire noise floor.
The fix, briefly: the Receipt Rule
The fix is not "stop reviewing." It's raising the price of a change. The Receipt Rule: you may change an answer only when you can point to the exact word or phrase in the passage that proves your first pick wrong. A feeling is not a receipt. "The other one sounds better" is not a receipt. A concrete misread — "it says some critics, and I read it as most" — is a receipt.
That one requirement filters changes almost perfectly: legitimate changes come from catching a specific misread; override changes come from unease, and unease can't produce a receipt. The full practice protocol — how to drill the rule, track your changed-answer accuracy, and know when the leak is closed — is in the Receipt Rule guide.