GRE General Test · GRE & GMAT Analytical Reasoning
GMAT Data Insights
Data literacy through multi-source reasoning, tables, graphics and two-part analysis.
Data Insights is the GMAT Focus section that rewards reading before computing: five question formats, one on-screen calculator, and most of the points sitting in whether you sorted the right column, named the right base, or filled both halves of a two-part answer.
- GRE General Test
- Hard level
- 5 concepts
- 5 practice questions
1Five formats in one section
Data Insights combines five question formats into one equally weighted Focus section: Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, and Two-Part Analysis. The calculator is available here; knowing which format you are in decides whether you sort, synthesise, read a scale, or fill two linked answers.
Data Sufficiency shares the section but has its own depth of method — sufficiency, not the value — so this topic treats it as one of the five labels and leaves the answer-choice logic for that topic.
Figure. Five formats share one Data Insights score. Equal bar lengths mark equal weight; the opening move still depends on which frame you see.
Recognising the format
- Scan the stem frameTwo statements labelled (1) and (2) with the fixed five choices is Data Sufficiency; two response columns is Two-Part; tabs of sources is Multi-Source.
- Scan the stimulusA sortable spreadsheet is Table Analysis; a chart or graph with dropdowns is Graphics Interpretation.
- Pick the first moveSort, jump to a tab, read axis units, or evaluate sufficiency alone — the format names the opening move before any arithmetic.
| Format | Stimulus | First move |
|---|---|---|
| Data Sufficiency | Question + statements (1) and (2) | Test each statement alone before combining |
| Multi-Source Reasoning | Two or three tabs of text, tables, or both | Index which tab holds each fact |
| Table Analysis | Sortable multi-column table | Sort by the column the statement cares about |
| Graphics Interpretation | Chart, graph, or diagram with fill-ins | Read axis labels, units, and legend first |
| Two-Part Analysis | Two linked response columns | Satisfy both columns; one correct column is not enough |
A question shows a bar chart of quarterly sales and asks you to complete two dropdown sentences about the chart. The format is
- Table Analysis
- Graphics Interpretation
- Two-Part Analysis
A chart with fill-in or dropdown sentences is Graphics Interpretation. Table Analysis needs a sortable spreadsheet, and Two-Part Analysis needs two linked answer columns rather than chart dropdowns.
2Multi-source: index the tabs before answering
Multi-Source Reasoning puts facts across two or three tabs — email, memo, table, or short passage — and asks questions that each need a different subset. The trap is rereading every tab for every question; the method is to note, once, where each fact lives so you can jump straight to it.
Synthesis means combining a number from one tab with a constraint from another. The on-screen calculator is provided in Data Insights, but it cannot tell you which tab held the base of a percentage.
Figure. Label each tab once (who, numbers, dates). Later stems jump to the needed tab — they do not force a full reread.
How to work a multi-source set
- Skim and labelOpen each tab once and write a three-word label: who, what numbers, what dates — so later questions do not force a full reread.
- Park contradictionsIf two tabs seem to disagree, note both locations; the question may be testing which source applies, not which arithmetic is faster.
- Pull only what the stem needsFor each question, list the facts required, jump to those tabs, then compute — the calculator comes last.
In Multi-Source Reasoning, the most reliable first pass across the tabs is to
- Compute every total that appears before reading any question
- Note which tab holds each fact so later questions can jump straight to it
- Ignore the text tabs and work only from any table that is present
The durable work is indexing where facts live. Pre-computing every total wastes time on quantities the set never asks, and skipping text tabs drops constraints that often decide the answer.
3Table Analysis: sort, then derive the metric
Table Analysis gives a sortable multi-column table and true/false-style statements about it. Sort by the column the statement cares about before evaluating — maxima, minima, and threshold counts fall out of the order instead of a slow scan.
When the metric is not a column — revenue from price and units, density from mass and volume — derive it for every row first, then sort or compare. The product that looks largest on one factor is often not the largest on the product.
Figure. Revenue = price × units for each product. B is tallest; C (highest price) is shortest.
How to clear a table statement
- Name the metricUnderline what the statement claims — a given column, or a derived quantity such as price × units.
- Derive if neededIf the metric is not already a column, compute it for each row before comparing.
- Sort, then judgeSort on that metric (or the relevant given column) and only then mark the statement true or false.
Top revenue is not top price
A table lists five products with unit price ($) and units sold: A 40 and 30; B 25 and 50; C 60 and 18; D 15 and 80; E 50 and 22. Which product generated the most revenue?
- Revenue = unit price × units sold for each rowfive products
- A: 40 × 30; B: 25 × 50; C: 60 × 181200; 1250; 1080
- D: 15 × 80; E: 50 × 221200; 1100
- Largest revenue among 1200, 1250, 1080, 1200, 1100B at 1250
Pro tip. C has the highest unit price and D the most units sold; neither wins on revenue. When the asked metric is a product, derive it for every row — sorting price or units alone answers a different question.
In the five-product table above, product C has the highest unit price. Its revenue rank among the five is
- First
- Last
- Neither first nor last
C's revenue is 60 × 18 = 1080, the smallest of the five (1200, 1250, 1080, 1200, 1100). Highest price and lowest revenue can sit on the same row — which is why the derived column has to be built before ranking.
4Graphics Interpretation: units first, then percent change
Graphics Interpretation attaches dropdowns or fill-ins to a chart. Misreading the scale is the most common error — check axis labels, units, and any legend before computing. A bar labelled 2 may be $2M, 2 thousand, or 2 percent; the arithmetic is identical and the answer choices are written for the wrong reading.
Percent change from a chart is still (new − old) / old × 100%. The endpoints must come off the graphic after the units are settled; the base is always the earlier value you are changing from.
Figure. Endpoints for the percent-change ledger: $2M to $5M is a 150% rise on the 2018 base.
How to read then compute
- Lock the unitsRead the axis title, tick unit, and legend. Write the unit next to each endpoint before any subtraction.
- Take the endpointsRead the exact start and end values the question names — not a neighbouring tick that is easier to see.
- Percent changeCompute (new − old) / old × 100%, with old as the base you are changing from.
Profit from $2M to $5M
A line graph shows a company's profit rising from $2M in 2018 to $5M in 2022. What was the approximate percent increase over that span?
- Change = 5M − 2M3M
- Divide by the original: 3 / 21.5
- 1.5 × 100150%
- Percent increase from 2018 to 2022150%
Pro tip. Read exact endpoint values off the axis before computing percent change. The same rise measured against $5M as base would be a 60% fall the other way — different question, different base.
Profit falls from $5M to $2M. The percent decrease is
- 150%
- 60%
- 40%
Change is 3M on a base of 5M: 3/5 × 100 = 60%. Answering 150% reuses the rise calculation with the base still on 2M after the question has moved it to 5M. 40% is the leftover share of the new value, not the percent decrease.
5Two-Part Analysis: both columns must be right
Two-Part Analysis presents two response columns — often related quantities, a pair of values, or a matched verbal pair — and you choose one entry in each column. Credit requires both halves; a perfect left column with a wrong right column scores nothing.
Treat the pair as one condition: solve for the relationship that ties the columns, pick the left entry, then pick the right entry that satisfies the same condition. Checking only the column that looks easier is how a nearly finished question still returns zero.
Figure. Both columns must satisfy the same joint condition. A perfect left half with a wrong right half scores nothing.
How to fill both columns
- Name the joint conditionWrite the equation or verbal rule that both columns must satisfy together.
- Fill one columnPick the entry that fits the condition; do not stop to celebrate a match on one side.
- Fill the other, then re-checkPick the paired entry and verify both together — one correct column is still a wrong answer.
On a Two-Part Analysis question you are sure of the left column and unsure of the right. The score if the right column is wrong is
- Full credit for a half-correct response
- No credit — both columns must be correct
- Half credit, proportional to one of two columns
Both entries must be correct. There is no partial credit for one column, so certainty on the left does not bank any points until the right column also satisfies the joint condition.
Notes
- Five Formats: Data Insights combines Data Sufficiency, Multi-Source Reasoning, Table Analysis, Graphics Interpretation, and Two-Part Analysis into one section.
- Integrated Reasoning: Questions demand synthesizing information from tables, graphs, and text, and an on-screen calculator is provided in this section.
- Sortable Tables: Table Analysis lets you sort columns; use sorting to quickly answer true/false-style statements about the data.
- Two-Part Analysis: You choose one option in each of two columns, often for related quantities, and both must be correct.
Formulas
- Formats: Data Sufficiency, Multi-Source, Table Analysis, Graphics Interpretation, Two-Part Analysis.
- Percent change: \dfrac{\text{new}-\text{old}}{\text{old}}\times100\%.
- Read graphics precisely: check axis labels, units, and scale before computing.
- Two-Part rule: pick one entry per column; both must satisfy the condition.
Exam traps & shortcuts
- Sort tables by the relevant column before evaluating each statement to spot maxima, minima, and thresholds fast.
- For Graphics Interpretation, read the axis units and any legend first—misreading the scale is the most common error.
- In Multi-Source Reasoning, note where each fact lives across the tabs so you can jump straight to it per question.
Reference tables
One first move per format. Reconstruct from the concepts above rather than memorising in isolation.
| Format | First move | Watch for |
|---|---|---|
| Data Sufficiency | Test (1) alone, then (2) alone | Sufficiency, not the numeric value |
| Multi-Source | Index which tab holds each fact | Synthesise across tabs; calculator last |
| Table Analysis | Sort on the relevant column or derived metric | Price × units before ranking revenue |
| Graphics Interpretation | Lock axis units and legend | Percent change divides by the old value |
| Two-Part Analysis | State the joint condition | Both columns must be correct |
Same formula as elsewhere in quant; the DI-specific failure is reading the wrong unit off the axis before the division.
| Quantity | Relation | Watch for |
|---|---|---|
| Percent change | (new - old)/old \times 100\% | Base is always the value you change from |
| Rise $2M → $5M | 3/2 \times 100\% = 150\% | Endpoints in the same unit |
| Fall $5M → $2M | 3/5 \times 100\% = 60\% | Not the 150% from the reverse direction |
Recap
Read only this the night before a Data Insights section.
- Five formats
- Data Sufficiency, Multi-Source, Table Analysis, Graphics Interpretation, Two-Part Analysis. The format names the first move.
- Multi-source
- Index which tab holds each fact once; jump per question. Calculator last.
- Tables
- Sort on the column that matters. If the metric is missing, derive it per row — highest price is not highest revenue.
- Graphics
- Units and legend before arithmetic. Percent change is (new − old)/old × 100% on the old base — $2M to $5M is 150%.
- Two-part
- One entry per column; both must satisfy the joint condition. One correct column scores nothing.
Practise GMAT Data Insights
Reading is free and needs no account. Practice, mocks and progress live in the app.
- 5 exam-style questions on this topic, with explanations
- A 5-question practice set that ends the chapter
- Timed mocks scored with the real marking scheme
- Readiness tracked per topic, kept on your device