GMAT Quantitative Reasoning Practice Test

The format, precisely

Twenty-one questions, 45 minutes, one question type. Every item is a Problem Solving question: a stem, five answer choices, one correct. The choices are values, expressions, fractions or ratios rather than sentences, which is a distinctive property of this section and one you can exploit. In the corpus we measured, the median Problem Solving answer choice contains zero English words. If you find yourself reading choices as prose in this section, you are in the wrong mode.

Two absences define the section as much as its contents. Geometry has been removed entirely, so triangles, circles, coordinate geometry and solids are not tested and revising them is pure waste. Data Sufficiency has moved to Data Insights, so the whole apparatus of statement evaluation, which used to occupy roughly a third of the old quantitative section, belongs to a different score now. If your study materials predate the Focus Edition, a meaningful share of them is aimed at content you will never see.

What is actually tested

The official score report classifies quantitative questions along three axes: content domain, question type and fundamental skill. Content domain has exactly two values, Arithmetic and Algebra. Question type has two values, Pure Contexts and Real Contexts, meaning a bare mathematical statement versus the same mathematics dressed in a scenario, and the split is roughly half each. Fundamental skills give the finer grain: Value/Order/Factors, Equal/Unequal/ALG, Rates/Ratios/Percent, and Counting/Sets/Series/Prob/Stats.

Underneath that taxonomy the concept distribution is strikingly uneven, and knowing the shape of it is worth more than any single technique. These are counts from the 797 question Official Guide corpus we measured, taken as they are tagged in the source.

Applied Problems is the single largest bucket by a wide margin. That is the honest headline of GMAT quant: most of it is real-world scenarios with numbers attached, not abstract manipulation. Statistics and Percents follow, and together those three account for 81 of the 271 Problem Solving items — about thirty per cent of the section, more than any other three named concepts combined. Meanwhile combinatorics and probability, which absorb a disproportionate share of most candidates' anxiety and revision time, sit at seven and six items respectively across the whole corpus.

The practical implication is uncomfortable but clear. If your weakest area is Applied Problems, that is where roughly one mark in seven lives — more than any other single concept offers — and no amount of combinatorics drilling will substitute. If your weakest area genuinely is probability, six items in the whole corpus is the ceiling on what fixing it can earn you. Our GMAT quantitative practice test reproduces this distribution rather than sampling evenly, so a section test tells you where your marks are actually going.

The shape of a Problem Solving question

Measured across 271 Problem Solving items in the official corpus, the median stem runs 36 words, with half of all items between 20 and 52 words and a full range from 1 to 101. The extremes are real: a one word stem is a bare expression to evaluate, and a hundred word stem is a multi-part scenario.

Look at the last three rows, because they carry the most useful finding on this page. Stem length is flat across difficulty. Easy items have a median of 36 words, medium 34, hard 36. A long question is not a hard question; it is a question with a scenario attached. Difficulty in Problem Solving comes from the number of reasoning steps between the stem and the answer, and nothing about the length of the text signals that in advance.

This has an immediate consequence for how you triage. Many candidates decide within two seconds, on the basis of how much text is on screen, whether a question is worth attempting. That heuristic is measurably wrong. A better one is to read the question fully and then ask whether you can name the first step. If you can, commit. If you cannot within about twenty seconds of finishing the read, the question is likely to cost you three minutes and it is a candidate for a guess and a bookmark.

The answer choices carry information too. Because they are numeric or symbolic rather than prose, they can often be used as part of the solution rather than only as a destination. Working backwards from the choices, testing a convenient value, or eliminating on the basis of magnitude, sign or units are all legitimate and frequently faster than the algebraic route. Nothing in the exam rewards you for solving a question the way a textbook would.

Pacing 21 questions in 45 minutes

Just under 2 minutes 9 seconds per question on average, which nobody can meter in real time. Use checkpoints instead. Question 11 should arrive with roughly 23 minutes remaining and question 16 with roughly 12. Check twice, not constantly; watching the clock is itself a tax on concentration.

1. Read the whole stem before touching the choices. Half-read stems produce the most expensive category of error, which is solving the wrong question correctly.
2. Identify what is being asked, specifically. If the question asks for 2x, write that down before solving for x.
3. Name the first step within twenty seconds of finishing the read, or plan to abandon.
4. Solve, using the choices as a tool where that is faster than the algebra.
5. Check that your answer is the quantity actually requested, then move on without re-reading.

Step two and step five look redundant and are not. The most common distractor mechanism in the entire quantitative corpus is answering the wrong question: solving for x when the item asks for 2x, or producing the part when the item asks for the whole. Those distractors exist because they catch a very large number of otherwise correct solutions. Writing the requested quantity down before you start, and checking it against your answer before you advance, removes a category of loss that no amount of additional content revision will touch.

The other recurring mechanisms are worth recognising by name, because seeing a familiar trap is faster than deriving it. One step short offers the value at the penultimate stage of the correct chain. Sign, direction or complement flips offer an increase where the question asked for a decrease, or 40 per cent where the remaining 60 was wanted. Right relation, wrong pair applies the correct operation to the wrong two quantities. Each of these is a specific candidate error, and each is deliberately represented in the choices.

Guessing, properly

There is no penalty for a wrong answer and every question is worth the same, so an unanswered question is strictly worse than a guessed one. The skill is not in guessing well; it is in guessing early enough that the time saved buys you a question you can actually do.

A guess made at twenty seconds costs you almost nothing and returns a one in five chance. The same guess made at three minutes and forty seconds costs you the better part of two other questions and returns the same one in five. Candidates rarely lose points on the question they abandoned; they lose them on the three questions at the end of the section they never reached because they refused to abandon anything.

• Eliminate on magnitude first. If a rough estimate rules out two choices, your guess is now one in three.
• Check units and sign. A surprising number of choices are wrong in a way that does not require solving.
• Do not look for patterns in the answer key. Our correct answer positions are randomised flat at 20 per cent each, so there is nothing to find.
• Bookmark every guess. If the section ends with time on the clock, the review screen and your three edits should go to bookmarked items first.

Using the review screen

Finish the 21 questions with time remaining and you reach a review screen listing every question and whether you bookmarked it. Three answer changes are available, each confirmed through a dialogue, and the counter does not reset. End the review and the section closes for good. If the clock expires before question 21, the review screen never appears at all.

Spend the edits on questions where you know what was unresolved: an arithmetic slip you suspected, a quantity you were not sure you had read correctly, a guess where elimination had got you to two choices. Do not spend them re-reading questions that simply feel uneasy in retrospect. Reworking a question you already answered under time pressure usually generates doubt rather than a correction, and the score report will show you how often your changes went the wrong way.

Reading a quantitative score report

Your section score sits on the 60 to 90 scale with a percentile beside it, and feeds into the 205 to 805 total alongside Verbal Reasoning and Data Insights. Below the score, four panels do the real work.

• Performance by content domain, splitting Arithmetic from Algebra.
• Performance by question type, splitting Pure Contexts from Real Contexts.
• Performance by fundamental skill, across the four skill labels.
• Question performance and time management: every question plotted by response time and outcome against a two-minute line, with a sortable table beneath it.

The Pure versus Real Contexts split is the most diagnostic and the most ignored. A candidate who scores well on Pure Contexts and badly on Real Contexts does not have a mathematics problem; they have a translation problem, and the fix is practice at converting scenarios into equations rather than more algebra. The reverse pattern, strong on Real and weak on Pure, usually means shaky fundamentals that scenarios have been hiding.

In the time scatter, the cluster to hunt for is slow and wrong. Those are questions where you paid full price and received nothing. Two or three are normal. Six or more means the abandonment decision is where your section is being lost, and that is a pacing problem to be trained directly rather than a content gap.

What to revise, and in what order

Because the content domain is only arithmetic and algebra, the revision list is short enough to write down in full. Work it in order of measured frequency rather than in the order a textbook presents it, and stop when the section test stops flagging a bucket.

1. Translating scenarios into equations. This underwrites Applied Problems, the largest bucket in the section, and it is a language skill more than a mathematical one.
2. Percentages, ratios and rates, including percentage change, reversing a percentage change, and combined work or travel rates.
3. Averages, weighted averages, median, range and standard deviation as a concept rather than a formula.
4. Number properties: factors, multiples, primes, divisibility, odd and even behaviour, and remainders.
5. Linear equations and systems, then inequalities, then quadratics and factorising.
6. Counting and probability last, because together they account for around a dozen items in the whole corpus.

Standard deviation deserves a specific note. It appears often enough to matter and almost never requires computation. What is tested is whether you understand that it measures spread, so questions ask which of two sets has the larger deviation, or what happens to it when every value is shifted or scaled. Candidates who memorise the formula and never build the intuition lose these questions while doing more work than the people who get them right.

Estimation is the other underrated entry. Six items in the corpus are tagged as estimation outright, but the skill pays across the whole section, because without a calculator an approximate answer is often enough to eliminate three choices in ten seconds. Practising deliberate rounding, and knowing when rounding is safe, converts a two minute calculation into a twenty second decision more often than any other single technique.

How this section fits the rest of your preparation

A quantitative section test takes 45 minutes and produces the same breakdown a full length test does for this section, without the fatigue cost. Sitting two or three a week while you work through the weak buckets is a far more efficient use of time than repeated full length sittings. Reserve the full length practice test for measurement every two to three weeks, and use it to check that section improvements are surviving contact with a 135 minute sitting.

Do not let quantitative crowd out the rest. Data Insights carries its own mathematics, including Data Sufficiency, simultaneous equations, statistics and sets, and it is scored separately on the same 60 to 90 scale. A candidate who is strong in arithmetic and algebra but has never practised Data Sufficiency is leaving points on a section they could win. The data insights practice test and the verbal practice test deserve the same weekly attention this one does.