- Low CAT RC accuracy is usually a diagnosis problem. More passages repeat the same error unless you know which cause is behind it.
- Log every wrong RC question under one of four causes: misread, over-inference, option trap or time.
- Each cause has a different fix, so a weekly review should change one habit, not the whole routine.
- VARC is timed at exactly 40 minutes with +3 for a right answer and -1 for a wrong MCQ. The official pages do not give the RC passage or question count.
- Targets for attempts and accuracy are estimates. Find your own by seeing where marks stop rising in your mocks.
CAT RC accuracy rarely improves just because you read more. It improves when you know why each wrong answer happened and fix that cause. A reader who misreads the question stem needs a different habit from one who over-infers, and a daily editorial helps neither unless the error log says that is the problem.
This post builds a four-cause error log, explains the usual option traps, and sets out timing and target ranges, all labelled as estimates. VARC is timed at exactly 40 minutes in CAT 2026, with +3 for a right answer and -1 for a wrong MCQ. The official pages do not state how many RC passages or questions it has. For the full method, see the VARC playbook.
CAT RC accuracy is a diagnosis problem, not a volume problem
Reading volume builds stamina and familiarity, and you need both. What it does not do is tell you why you picked the wrong option. Two readers can both score 60% on RC for opposite reasons: one reads the stem too fast, the other reads the passage well and then picks the option that sounds wisest.
Advice on the web also disagrees on targets, and none of it comes from the exam body. That is another reason to work from your own mocks, where the data is yours.
Build an RC error log with four causes
For every wrong RC answer, in every mock, tick exactly one cause. Four columns are enough.
| Cause | What it looks like | Fix |
|---|---|---|
| Misread | You answered a different question or missed a qualifier such as not, except, only or most | Underline the qualifier and restate the question in your own words before you read the options |
| Over-inference | You picked an option that is true or sensible but not supported by the passage | Point to the line that proves it. If you cannot, the option is out |
| Option trap | Half-true, reversed, out of scope or too extreme | Name the trap for each option you cross out |
| Time | Right method, but rushed or too slow on an earlier passage | Set a ceiling per passage and skip by rule |
Add the question type and the line that gave the right answer. The log format is the same one described in the mock analysis guide, with a cause tick instead of a free-text label.
| Mock and question | Type | Your answer | Right answer | Cause | Proof line |
|---|---|---|---|---|---|
| M8 P3 Q2 (example row) | Inference | C | B | Over-inference | Para 2 says 'some', option says 'all' |
Over-inference and extreme options
Over-inference is the quiet one. The option reads like common sense, so it feels right. Test it by asking whether the passage alone forces it. Words such as always, never, only and all are warning signs, but not proof: if the author is themselves extreme, an extreme option can be correct.
- (a) generalises from one city, so it is too extreme.
- (c) states a cause the author deliberately leaves open.
- (d) assigns an opinion the passage never gives.
- (b) matches the author's hedge.
Answer: (b)
Passage-type patterns: abstract, social science, argumentative
Check these patterns against your own log. They are tendencies, not rules.
| Passage type | Common risk | What to do |
|---|---|---|
| Abstract or philosophical | Losing the thread in dense terms | Label each paragraph's job and restate it in plain words |
| Social science or history | Mixing up who said what | Keep a mini map of each position and who holds it |
| Argumentative or opinion | Taking a reported view as the author's | Mark where the author turns, at words like but, however and yet |
Timing: when slowing down raises the score
Under +3 / -1, accuracy is worth more than it looks. Illustrative numbers, since the RC question count is not official:
- A: 7 right, 7 wrong = 21 - 7 = 14 marks.
- B: 8 right, 2 wrong = 24 - 2 = 22 marks.
Answer: B, by 8 marks, with fewer attempts. Slowing down paid.
Slowing down helps only when your log shows misread or over-inference errors. If time is your main cause, slowing down makes it worse. Then you need a ceiling per passage, not a slower pace.
Which questions to skip inside a passage
- Answer first: main idea and detail questions where you can point to the line.
- Delay: inference questions with long options, until the easy ones are done.
- Skip: any question where two options survive and neither has support in the passage.
- Never leave a type-in verbal item blank. It has no negative marking.
Guess an MCQ only after you can honestly remove at least one option. With one option removed the expected gain is positive, since the break-even chance of being right is 25%. With no option removed, and four options, it is zero at best.
Weekly review: what to change
Once a week, tally causes across your last three mocks. Change strategy only if three mocks show the same pattern, then change one thing.
| Most common cause | Change for next week |
|---|---|
| Misread | Restate every stem before you look at the options |
| Over-inference | Write the supporting line number beside every answer |
| Option trap | Cross out each option by its named trap |
| Time | Set a passage ceiling and skip by rule |
Retest the logged questions after three days and again after ten. If you miss one twice, the fix was wrong, so rewrite it.
Targets for attempts and accuracy: treat them as estimates
Published targets differ widely, and the exam body publishes none. Find your own: the point where more attempts stop adding marks because accuracy falls. These VARC-wide illustrations show the trade-off.
| Attempted | Accuracy | Right / wrong | Marks |
|---|---|---|---|
| 16 | 75% | 12 / 4 | 32 |
| 20 | 60% | 12 / 8 | 28 |
| 20 | 80% | 16 / 4 | 44 |
| 24 | 75% | 18 / 6 | 48 |
Twenty attempts at 60% score less than sixteen at 75%. More attempts help only while accuracy holds. Take a free CAT mock to get a first accuracy figure, then log the next three full mocks.