GMAT Data Insights Practice Test

The format, precisely

Twenty questions in 45 minutes, an average of 2 minutes 15 seconds each, which sounds generous and is not. Several formats put a substantial block of source material in front of you before you can attempt anything, and one format has a median prompt roughly four times the length of a quantitative stem. The average is real but the variance around it is larger here than anywhere else on the exam.

The section is scored on the same 60 to 90 scale as Quantitative Reasoning and Verbal Reasoning and contributes equally to the 205 to 805 total. That equality is the argument for taking it seriously. A candidate who is comfortable in arithmetic and argument analysis but has never met a sortable table under time pressure is giving away a third of the exam to unfamiliarity rather than to ability.

The five formats

Data Sufficiency

A question followed by two numbered statements, and five fixed answer choices that never change: statement one alone is sufficient but two is not, statement two alone is sufficient but one is not, both together are sufficient but neither alone, each alone is sufficient, or the two together are still not sufficient. You are not asked to compute the answer. You are asked whether the answer could be determined, which is a different mental operation and the source of most of the difficulty.

Measured across 135 official Data Sufficiency items, the median stem runs 34 words, with half of all items between 22 and 50. Each statement has a median of 13 words and the two are close to symmetric, both sitting around that figure. That symmetry is deliberate on the exam's part: a statement one of eight words paired with a statement two of forty would signal which one is doing the work.

Unlike Problem Solving, Data Sufficiency stems do grow with difficulty, from a median of 28 words at easy through 37 at medium to 48 at hard. The reason is structural: harder items carry more setup precisely because the difficulty lives in what the setup does not pin down. A long Data Sufficiency stem is therefore a genuine warning, which is the opposite of the rule that applies in Quantitative Reasoning.

The answer distribution is also worth knowing. In the official corpus, the both-statements-needed and still-insufficient cases dominate, at 40 and 34 items respectively out of 135, because those are the interesting cases to test. We keep our Data Sufficiency key distribution near that observed shape rather than forcing it flat, which is a deliberate exception to how we balance every other format.

• Evaluate each statement in complete isolation first. Contamination between statements is the defining Data Sufficiency error.
• Answer the question that was asked. Sufficiency for a value is not the same as sufficiency for whether a value is positive.
• Look for the case that breaks it. One counterexample proves insufficiency; no amount of confirming examples proves sufficiency.
• Stop as soon as sufficiency is established. Computing the actual value is wasted time you cannot get back.

Table Analysis

A split screen. On the left, a prompt, a sort-by dropdown and a bordered table you can reorder by any column. On the right, an instruction and three statements, each answered with one of two mutually exclusive options, typically yes or no. All three have to be right for the question to score.

The sort control is the whole point of the format and it is routinely ignored. Almost every Table Analysis statement becomes trivial once the table is sorted by the right column, and almost every one is laborious if you read the table in its default order. Before evaluating a statement, decide which column would make it obvious, sort by that column, and read off the answer. Sorting is free and re-sorting is free.

Because all three statements must be correct, this is a format where partial understanding scores nothing. It is also a format where careless reading is expensive: statements are often phrased with a threshold, a comparison or a negation that changes everything, and the difference between at least and more than has decided a great many of these.

Graphics Interpretation

A chart at the top of a single-pane screen, then a sentence or two beneath it containing inline dropdown menus. You complete the sentence by choosing from each dropdown. There are no lettered choices; the answer is the combination you select.

The charts vary: scatter plots with or without a trend line, bar charts, line graphs, bubble charts and occasionally something less common. The reading skill required is elementary. The traps are not in the mathematics but in the axes, and specifically in scales that do not start at zero, logarithmic axes, units given in thousands or millions, and legends that assign meaning to size or colour as well as position.

Read the axes, the units and the legend before you read the sentence. Thirty seconds spent establishing what the chart actually shows prevents the most common failure in this format, which is a perfectly competent reading of the wrong quantity. The dropdown options themselves are usually a short ordered list, so once you know the quantity the selection is quick.

Multi-Source Reasoning

The heaviest format on the exam by reading load. A split screen with two or three tabs on the left, each holding prose, a table or a chart, and a question on the right with either five choices or a three-row grid of two-option statements. Several questions draw on the same tab set, so the reading cost is amortised in the same way a Reading Comprehension passage is.

The critical behaviour is opening every tab before answering anything. Candidates who answer from the first tab alone are the intended victims of this format; the material is deliberately arranged so that the tab you did not open contains the qualification that changes the answer. Skim all tabs to establish what kind of information each holds, then answer, returning to the specific tab you need.

This is where verbal skill pays inside a section most people think of as quantitative. Much of Multi-Source Reasoning is inference work over a small dossier: what does this discussion thread commit its participants to, what does the schedule rule out, what follows from the two together. If you have worked through the verbal practice test seriously, that transfer is real and immediate.

Two-Part Analysis

A prompt followed by a table with two response columns and a shared list of five or six options down the right-hand side. You choose one option for each column, and both must be correct. There is no answer letter; the key is the pair.

Measured across 48 official items, Two-Part Analysis carries by far the longest prompts in the corpus: a median of 158 words, with half of all items between 119 and 223 and a full range from 50 to 322. That is close to four times the median Problem Solving stem of 40 words. A Two-Part Analysis item is a scenario plus a selection instruction, and it needs to be budgeted at well above the section average.

The columns are usually related, often as a pair that must satisfy a joint constraint, so treating them as two independent questions is the standard mistake. Work out the relationship the pair has to satisfy first, then find the pair that satisfies it. Where the options are numeric, testing candidate pairs against the constraint is frequently faster than solving directly.

What Data Insights actually tests

The score report splits this section into Math Related and Non-Math Related content, and reports question type as Data Sufficiency, Graphs and Tables, Two-part analysis and Multi-source reasoning. Fundamental skills use the same quantitative labels: Value/Order/Factors, Equal/Unequal/ALG, Rates/Ratios/Percent, and Counting/Sets/Series/Prob/Stats.

Underneath, the concept tagging of the official corpus shows a familiar shape. Applied Problems is the largest bucket at 39 items, exactly as it is in the quantitative section, which tells you that the mathematics here is scenario mathematics rather than abstract manipulation. Inference and application work follow closely, then statistics, simultaneous equations and sets.

Add the inference, apply and evaluate buckets together and they rival Applied Problems in size. That is the most useful thing this table says: Data Insights is roughly half quantitative reasoning and half interpretation, which is why treating it as a maths section produces a plateau. The mathematics involved rarely exceeds what the quantitative section already requires, and simultaneous equations, sets and basic statistics cover most of it.

Pacing a section with five different question costs

The average of 2 minutes 15 is a pool, not an allowance. A Data Sufficiency item with a 34 word stem and two 13 word statements can be settled in ninety seconds. A Two-Part Analysis prompt of 200 words takes three minutes or more before you have understood what is being asked. A Multi-Source Reasoning set costs two minutes of reading before its first question and then delivers two or three questions cheaply.

1. Check the clock at question 10 with roughly 23 minutes left, and at question 15 with roughly 11.
2. Treat the first question against a Multi-Source Reasoning tab set as expensive and the following ones as cheap. Do not panic at the first.
3. Sort the table before evaluating any Table Analysis statement. It is free and it removes most of the work.
4. Read axes, units and legend on every chart before reading the question.
5. Use the calculator only when the arithmetic is genuinely awkward. Opening, dragging and typing costs more time than most candidates account for.
6. Bookmark and guess on any item where you have not understood the setup after ninety seconds. Understanding rarely arrives late in this section.

The interface itself deserves rehearsal, which is an argument for sitting section tests rather than only reading about the formats. Dragging a calculator out of the way, sorting a table, opening tabs, choosing from inline dropdowns and filling a two-column grid are all small operations, and each one costs a few seconds the first time you meet it. Twenty questions is enough for those seconds to add up to a question you did not reach.

How our Data Insights questions are built

All five formats appear in proportion on every Data Insights section test, and every item is written against the measured shape of the official corpus. Data Sufficiency statements are held close to symmetric at around 13 words each, because an asymmetric pair gives the game away. Two-Part Analysis prompts are written into the 106 to 209 word band. Difficulty is drawn in near-equal thirds, matching the split the Official Guide uses for its own practice set.

• Every distractor is traceable to a named candidate error rather than being filler.
• Data Sufficiency keys keep their natural distribution, which genuinely favours both-needed and still-insufficient.
• Every other format has its correct answers randomised flat, so there is no exploitable lean.
• Charts state their units and label their axes, including where a scale does not start at zero, exactly as official items do.
• Every question is tagged with content domain, question type and fundamental skill using the official report labels.

Reading the Data Insights report

The format-level breakdown is the reason to sit a Data Insights section test rather than relying on the 20 items inside a full length practice test. Twenty questions spread across five formats gives you three or four of each, which is too few to conclude anything. Four section tests give you a real sample of each format, and the report separates them.

Look first at the Math Related against Non-Math Related split. A large gap in favour of Math Related usually means Multi-Source Reasoning and the interpretive parts of the section are the problem, and the answer is close reading practice rather than more arithmetic. The reverse gap usually means the quantitative fundamentals need work, in which case the quantitative practice test is the more efficient place to do it.

Then look at response times by format in the per-question table. It is common to find that a single format is consuming a disproportionate share of the section, most often Two-Part Analysis or the first question of a Multi-Source set. If that format is also scoring well, the cost may be acceptable. If it is scoring badly, you have found the clearest possible instruction for the next fortnight of preparation.

Once the format-level picture is stable, check it under fatigue with a full length practice test. Data Insights placed third in your section order is a different experience from Data Insights sat fresh, and the section order you choose should be informed by which of those you handle better.