Premiergrad Start practising

StrategyQuantitative · Verbal · Data Insights

Why “careless errors” aren’t really careless

“Careless” describes how a mistake feels after you see it. It does not explain what produced it.

If the same kind of “silly mistake” happens repeatedly, it is not random noise. It is evidence of a process that can be named and changed.

Five mechanisms hiding under “careless”

1. Target loss

You found a useful value but not the requested value.

Example: selecting the $12 increase when the question asks for the percentage increase.

Correction: label the target and units before solving; state what the result represents before selecting.

2. Constraint loss

You solved the equation but ignored positive, integer, distinct, nonzero, or feasibility restrictions.

Correction: keep a visible constraint ledger and test candidates against it.

3. Representation error

You knew the concept but translated the situation incorrectly.

Example: using the final price as the denominator for percentage increase.

Correction: write the relationship in words before inserting values.

4. Case loss

You checked one scenario and treated it as proof.

Example: testing x = 2 for “Is x² > x?” without checking values between 0 and 1.

Correction: list structural cases—sign, parity, boundary, overlap—before testing.

5. Timing-driven shortcut

You skipped a step because the clock felt threatening, or spent so long on one item that later work became rushed.

Correction: use decision checkpoints and a bounded guess rule, not a vague promise to “go faster.”

Why “slow down” is incomplete advice

Slowing down may reduce some execution errors, but it can also create a section-level timing problem. The useful intervention is a small checkpoint placed before the specific failure.

The checkpoint should take seconds because it is matched to a known pattern.

Diagnose one mistake precisely

A price rises from $80 to $92.

The surface topic is identical. The remedies are not.

Problem Solving Very Hard · percent / fractions, decimals

A coin dealer purchased 200 identical silver coins at $11 7/8 per coin and sold them all at $27 per coin. The dealer paid a 5 percent brokerage fee on the total purchase price and an 8 percent sales commission on the total selling price. Which of the following is closest to the dealer's percent gain on this investment?

Answer: B

Total purchase price: 200 x $11.875 = $2,375. Purchase fee: 5% of $2,375 = $118.75, so total investment cost = $2,493.75. Total sale price: 200 x $27 = $5,400. Sale fee: 8% of $5,400 = $432, so net proceeds = $4,968. Gain = $4,968 - $2,493.75 = $2,474.25. Percent gain = $2,474.25 / $2,493.75 x 100 = about 99.2%, closest to 99%. The 109% option subtracts the sale fee but omits the purchase fee; 117% adds the purchase fee but omits the sale fee; 127% omits both fees; 50% divides the gain by net proceeds.

Original GMAT-style question written for Premiergrad and checked by an independent blind solve. Not an official GMAT question.

What Lumen looks for

Lumen reads the answer you chose, not just whether it was wrong. On a question like this it separates three things that all look identical in a score: a concept you have not met, a reasoning step that went sideways, and a correct method pointed at the wrong target. Those need three different next weeks, and only the third one is fixed by re-reading the topic.

A single answer is weak evidence and Lumen says so. It becomes a diagnosis when the same shape shows up across several questions — which is the part a page like this cannot do for you.

A one-week careless-error audit

For each wrong or lucky-correct answer, record:

At the end of the week, count mechanisms—not topics alone. If Requirement Misread appears in Quant, Verbal, and Data Insights, train the cross-section process.

Related resources: Requirement Misread, Hidden Constraint, Missing Case, Premature Calculation, and How to Build a GMAT Error Log.

Try the free diagnostic quiz and replace “careless” with something you can actually train.

Diagnose my recurring errors →

Related

You have the method — now get the evidence

A framework is worth more with your own attempts in front of it. A short diagnostic produces the evidence this page tells you how to read.

Run a short diagnostic Everything on this page stays free and open — no account needed to read it or to attempt the questions above.

Last reviewed 2026-09-17