Statistics on the GMAT Focus sits in a strange position: every candidate meets it, almost nobody drills it, and the questions that appear are far more predictable than the average word problem suggests. The Quant section tests arithmetic, algebra, and word problems in roughly equal measure, but within those families a statistics cluster is statistically over-represented relative to the time most prep plans spend on it. A candidate who can recognise a statistics stem in the first ten seconds, choose the right working framework, and avoid three predictable traps can quietly pull a 5–7 point swing on a 60–90 score band without changing anything else in their preparation. This guide is written for that candidate: someone who has done the arithmetic work, has practised Problem Solving and Data Sufficiency, and now wants statistics handled the way a senior tutor would handle it at the whiteboard.
What follows is a working map of the statistics question types that the GMAT Focus actually delivers, the way each stem signals its family, the standard attack for each, and the silent failure modes that cost points. The goal is not to recite definitions. The goal is to make statistics the section of the test where you stop second-guessing and start banking points on autopilot.
Where statistics lives inside GMAT Focus Quant
The GMAT Focus is a single 45-question Quant section, adaptive across an easy and a hard module, scored on a 60–90 band. Each module pulls from a balanced question bank covering arithmetic, algebra, geometry, word problems, and statistics. Statistics in the strict textbook sense — mean, median, mode, range, standard deviation, variance, weighted averages, and probability — is a recurring sub-family rather than a labelled section. In a typical 45-question sitting a candidate will see somewhere between four and seven statistics-adjacent questions, depending on the module difficulty. That sounds small. In practice, statistics questions are the cluster most likely to carry a disproportionately large hit on confidence because the stem often looks like an arithmetic problem until the second read.
Three observations matter before any tactics. First, the GMAT Focus does not require a formal statistics course. Every concept tested sits inside what a strong secondary-school syllabus covers: arithmetic mean, weighted mean, median, mode, range, frequency interpretation, basic probability, and the intuition behind standard deviation without heavy computation. Second, Data Sufficiency questions on statistics test reasoning more than calculation, which is good news for candidates who freeze on long numbers. Third, the test rarely asks for a numerical standard deviation. It asks whether you can compare the spread of two sets, or decide if adding a value changes the mean in a predictable direction. Recognising that single fact changes how you read a stem.
A practical consequence: the most efficient statistics preparation is not a chapter review, it is a pattern-recognition drill. Candidates who memorise the stem signals and rehearse four or five canonical attacks routinely pick up 3–5 extra correct answers per sitting. A candidate reading this who has already done 200+ practice questions is closer to a statistics score ceiling than they realise.
Mean, median, mode: the three calculations and the one decision that matters
Arithmetic mean is the most-tested concept, but it is the mean questions framed around "the new mean after adding or removing a value" that quietly decide scores. The standard attack is mechanical. If a set of n numbers has mean M, the total sum is n × M. To find the new mean after adding a value x, compute (n × M + x) ÷ (n + 1). To find the new mean after removing a value x, compute (n × M − x) ÷ (n − 1). Every variation reduces to this identity. Candidates who write the sum equation first, then divide, almost never lose these questions. Candidates who try to short-cut by averaging averages lose them often.
Median and mode are usually tested together with mean in a comparison stem. The classic shape presents five numbers and asks which of mean, median, or mode is greatest, smallest, equal to a given value, or changed by an operation. The trap is sequencing: candidates who compute mean first, then median, then mode, then compare, burn 90 seconds. The faster approach is to look at the numbers and rank the three measures by inspection. For a roughly symmetric short set, mean and median are close. For a set with one large outlier — say 2, 3, 4, 5, 26 — the mean is pulled upward, the median stays in the middle, and the mode (if any) is independent. One read of the spread, and the answer is often visible without any computation at all.
Common pitfalls and how to avoid them
- Forgetting to recount the divisor. When a value is added, the count goes up by 1. When a value is removed, it goes down by 1. Off-by-one divisor errors are the single most common statistics mistake and the easiest to fix: write n explicitly on the page.
- Confusing the mode with the most frequent value versus a unique mode. A set like 1, 2, 2, 3, 4 has a clear mode (2). A set like 1, 1, 2, 2, 3 has two modes. The GMAT Focus stems always specify "unique mode" when needed; if the stem does not say unique, the question is testing something else.
- Mixing up median and mean under time pressure. A 30-second mental checkpoint: is the question asking for the middle value when ordered, or the arithmetic balance? The verb in the stem tells you. "Middle value" or "middlemost" is median. "Average" is mean.
Weighted averages: the lever-and-balance framework
Weighted average questions are the highest-yield statistics pattern on the GMAT Focus because they hide inside word problems that look like mixtures. The canonical stem describes two groups with different means and asks for the combined mean, or describes a combined mean and asks for the ratio of the group sizes. The lever rule is the cleanest attack. If group A has mean a with weight wA and group B has mean b with weight wB, the combined mean sits closer to the larger group. Equivalently, the distance from the combined mean to a is proportional to wB, and the distance to b is proportional to wA. This produces a single ratio without any algebra: wA / wB = (b − combined mean) / (combined mean − a).
Worked example: a class of 60 students has a mean score of 70 on a test. The boys' mean is 65 and the girls' mean is 75. How many boys are in the class? Apply the lever rule. Distance from combined mean to boys' mean is 5. Distance from girls' mean to combined mean is 5. So wBoys / wGirls = 5 / 5 = 1, meaning equal numbers. With 60 students total, that is 30 boys and 30 girls. The arithmetic is one line. The pattern recognition is everything.
Two practical notes. First, the GMAT Focus frequently disguises weighted averages inside "a shop sells two products" or "a team has two groups of workers" wording. Read the stem for two distinct means and a combined mean before reaching for any formula. Second, Data Sufficiency versions of the same pattern reduce to: can the candidate deduce the ratio of the two group sizes? Statement (1) alone is usually insufficient because it pins only one variable, and Statement (2) is usually sufficient because it gives the second ratio or the second mean. Recognising that pattern saves minutes in DS.
Range and standard deviation: the comparison questions
Range questions are quick: subtract the smallest value from the largest, then compare. The trap is not computation; it is misreading which set is being asked about. Candidates who read "Set A and Set B" carefully never lose range. The GMAT Focus occasionally tests range indirectly by asking which transformation — adding a constant, multiplying by a positive constant, removing an outlier — changes the range. Adding a constant to every value does not change the range. Multiplying every value by a positive constant scales the range by the same factor. Removing an outlier can shrink the range dramatically. Three facts. Memorise them once.
Standard deviation is where candidates panic unnecessarily. The GMAT Focus does not require a numerical standard deviation. It tests whether a candidate can compare the spread of two sets, or predict how a transformation changes the spread. The decision rules are short:
- Adding the same constant to every value in a set does not change the standard deviation.
- Multiplying every value by a positive constant k multiplies the standard deviation by k.
- Multiplying by a negative constant multiplies the standard deviation by the absolute value, since spread is always non-negative.
- Increasing the spread of a set (pushing values further from the mean) increases the standard deviation.
Worked comparison: Set X is {2, 4, 6, 8, 10}. Set Y is {1, 4, 7, 10, 13}. Both have a mean of 6 and a range of 8. But Set Y's values are more clustered around the mean relative to the extremes, so Set Y has a smaller standard deviation. A candidate who can see that without computing either standard deviation is operating at the level the test rewards. The same instinct solves Data Sufficiency stems that ask "is the standard deviation of Set A greater than that of Set B?" — Statement (1) is sufficient when it pins the exact composition of both sets; Statement (2) is insufficient if it gives only one set's shape.
Probability questions: the conditional and the independent
Probability on the GMAT Focus is always a small-integer calculation. The most common shapes are independent events, conditional probability expressed as "given that…", and probability phrased as "at least one" or "none". The cleanest attack for independent events is to compute the complement. If a question asks "what is the probability that at least one of A, B, C occurs?", compute the probability that none of them occurs, then subtract from 1. The complement is almost always faster than enumerating cases, and it generalises to any number of independent events.
Worked example: a bag contains 4 red and 6 blue marbles. Three marbles are drawn without replacement. What is the probability that at least one is red? Compute the complement: probability that all three are blue = (6/10) × (5/9) × (4/8) = 120 / 720 = 1/6. Therefore the probability of at least one red is 1 − 1/6 = 5/6. Two multiplications and a subtraction. Enumerating the cases would have taken five times as long.