Relvora Prep

GRE Quant lesson · Data interpretation

GRE data interpretation: reading charts and tables, percent change and what the data can't tell you

How to answer GRE Data Interpretation questions: reading bar, line, stacked and scatter charts, ratios, percent change, combining a chart with a table, growth-rate charts, and deciding what must be true from a table.

The core idea

Read every value against the scale and its units before you compute. Percent change divides by the earlier value; percents compound; a share is an amount only once you know the total. Before you answer, ask whether the display really fixes the quantity.

Core ideas

Words used here

  • Scale: the numbered axis. Gridlines are its horizontal lines; values may also sit on the halfway mark between two gridlines.
  • Stacked bar: one bar split into segments that add up to the bar's total.
  • Scatterplot: one dot per item, placed by two measurements.
  • Percent change: (later value − earlier value) ÷ earlier value × 100.
  • Percent share: a part as a percent of a total.
  • Percentage points: the plain difference between two percents (30% to 35% is 5 points, a 16.7% increase).
  • Read before you compute. Note the title, the units ("$ thousand", "millions") and the scale. GRE charts are drawn to scale, and every value in these questions sits on a gridline or a halfway mark.
  • Percent change divides by the earlier value.
  • Shares need a total. 35% of a total of 400 is 140; 35% alone is not an amount.
  • Percents compound. Successive changes multiply: (1 + r₁)(1 + r₂).
  • Ask what the data fixes. Percents fix comparisons; only totals or levels fix amounts.

On the GRE.

  • Data Interpretation sets: several questions often share one display. Spend a few seconds on the display first; the questions then go quickly.
  • Calculator: the on-screen calculator does the division in a percent change. The traps are which numbers to divide, not the arithmetic.
  • Numeric Entry: follow the rounding instruction exactly ("to the nearest whole percent"), and round only at the end.
  • Quantitative Comparison and "must be true": the answer can be "cannot be determined" when the display gives percents but the question needs amounts.

Try it. A bar chart shows a company's sales of $80 thousand in 2021 and $116 thousand in 2024. By what percent did sales increase?

Show the answer

(116 − 80) ÷ 80 × 100 = 36 ÷ 80 × 100 = 45%.

Reading values from a chart

Read each value against the scale, then do exactly the arithmetic the question asks. In a stacked bar, a segment's value is its top edge minus its bottom edge. In a scatterplot, draw the gridlines for each condition and count the dots on the right side.

Why it works. A chart is a table drawn to scale. Once each value is read correctly, the question is ordinary arithmetic; nearly every wrong answer comes from reading the wrong value or the wrong scale.

How to spot it: "according to the graph", "by how much", "how many of the …".

Example. Sales of four stores, in $ thousand: A 140, B 90, C 170, D 110. By how much were C's sales greater than D's?

020406080100120140160180200Sales ($ thousand)ABCD14090170110170110
Read C at 170 and D at 110 on the scale: the difference is 60, or $60 thousand.
  1. C reads 170 and D reads 110.
  2. 170 − 110 = $60 thousand.

Stacked bars. If a bar's bottom segment runs from 0 to 40 and the next from 40 to 70, the second segment's value is 70 − 40 = 30, not 70.

Trap: reading a value one gridline off, or adding the two bars (280).

Edge cases.

  • Units: "$ thousand" means 170 is $170,000.
  • Halfway marks: with gridlines every 20, a bar halfway between 100 and 120 reads 110.
  • Scatterplots: decide strict inequalities first; a dot exactly on a cutoff doesn't count for "greater than".

GRE-level example. A stacked bar for 2023 has segments Hydro 0 to 30, Wind 30 to 80 and Solar 80 to 100 (GWh). In 2024 they are Hydro 0 to 40, Wind 40 to 85 and Solar 85 to 110. Did Wind generation rise or fall?

  1. 2023 Wind: 80 − 30 = 50. 2024 Wind: 85 − 40 = 45.
  2. Wind fell by 5 GWh, even though its top edge rose from 80 to 85.
2023Hydro 30Wind 50Solar 201002024Hydro 40Wind 45Solar 25110
Wind is the middle segment: 80 − 30 = 50 GWh in 2023 and 85 − 40 = 45 GWh in 2024. It fell, though its top edge rose.

Check yourself. In the store chart, what were the combined sales of A and B?

Check your answer

140 + 90 = $230 thousand.

Ratios between values

Read the two values, write them in the order the question names (first to second), and reduce.

Why it works. A ratio compares two quantities directly. Swapping the order gives the reciprocal, and comparing one part with the whole gives a different ratio altogether.

How to spot it: "the ratio of … to …", "how many times as large".

Example. In the store chart, what is the ratio of C's sales to D's sales?

  1. C : D = 170 : 110.
  2. Divide both by 10: 17 to 11.

Trap: 11 to 17 (the order reversed) or 17 to 28 (C compared with C and D together).

Edge cases.

  • "How many times as large" is the same ratio as a single number: 170 ÷ 110 ≈ 1.55.
  • Units cancel when both values use the same scale, so the ratio has no units.
  • Different series: read both values in the same year or category unless the question says otherwise.

GRE-level example. What is the ratio of B's sales to A's sales?

  1. B : A = 90 : 140.
  2. Divide by 10: 9 to 14.

Check yourself. What is the ratio of A's sales to the four stores' total?

Check your answer

Total: 140 + 90 + 170 + 110 = 510. A to total: 140 to 510 = 14 to 51.

Percent change from a chart

Percent change = (later − earlier) ÷ earlier × 100. To compare percent changes, divide each change by its own starting value. For a combined total, add first, then take one percent change.

Why it works. A percent change measures a change against where it started. A rise of 50 from 200 is a quarter of the start; a rise of 30 from 60 is half of it.

How to spot it: "by what percent", "percent increase", "increased by more than 20%".

Example. Series P rose from 200 to 250; series Q rose from 60 to 90. Which had the greater percent increase?

  1. P: 50 ÷ 200 = 25%.
  2. Q: 30 ÷ 60 = 50%.
  3. Q, although P rose by more units.
P200+50+25%Q60+30+50%
P rose 50 on a start of 200 (25%); Q rose only 30, but on a start of 60 (50%). Divide each rise by its own start.

Trap: comparing the rises in units (50 against 30).

Edge cases.

  • "More than 20%" means the later value is more than 1.2 times the earlier one; exactly 1.2 times is not more.
  • Combined totals: P and Q together went from 260 to 340, an increase of 80 ÷ 260 ≈ 30.8%, not the average (25% + 50%) ÷ 2 = 37.5%.
  • Decreases: a fall from 250 to 200 is 50 ÷ 250 = 20%, not 25%.

GRE-level example. Three hubs shipped 200, 160 and 90 thousand parcels in 2024 and 238, 196 and 105 thousand in 2025. Which grew by more than 20%?

  1. Cutoffs (1.2 times 2024): 240, 192 and 108.
  2. 238 < 240: no. 196 > 192: yes. 105 < 108: no.
  3. Only the second hub. The first hub had the biggest rise in units (38) but only a 19% rise.
050100150200250thousand parcelsHub 1Hub 2Hub 320251.2 × 2024238196105240192108above
Cutoffs are 1.2 × 2024: 240, 192 and 108. Only the second hub beat its cutoff (196 > 192); the first missed by 2.

Check yourself. A value rose from 80 to 92. By what percent did it increase?

Check your answer

12 ÷ 80 × 100 = 15%.

Combining two displays

When one display gives totals and another gives percent shares, each part's amount is its share times the total for the same year. Work out each year's amount before comparing years.

Why it works. A share has no size until it is applied to a total. A falling share of a growing total can still be a growing amount, and the reverse.

How to spot it: a chart of totals beside a table (or a second chart) of percents; "what was the amount for …", "by how much did … change".

Example. Total sales were $400 million in 2024 and $460 million in 2025. The North region's share was 35% in 2024 and 30% in 2025. How did North's sales change?

  1. 2024: 0.35 × 400 = $140 million.
  2. 2025: 0.30 × 460 = $138 million.
  3. North's sales fell by $2 million, although total sales rose.
2024North 140Others 26035% of 4002025North 138Others 32230% of 460
North's share fell from 35% to 30%, but of a bigger total: 0.35 × 400 = 140 and 0.30 × 460 = 138, $2 million less.

Trap: applying the change in share (−5 points) to one year's total, or using one year's total for both years.

Edge cases.

  • The rest of the total: 100% − 35% = 65% belongs to the other regions.
  • Matching years: take the share and the total from the same year.
  • Percentage points vs percent: a share falling from 35% to 30% is 5 points, a 14.3% drop in the share.

GRE-level example. With the totals above ($400 million, then $460 million), another region had 25% of sales in 2024. What is the smallest 2025 share that keeps its sales from falling?

  1. 2024 amount: 0.25 × 400 = $100 million.
  2. In 2025 it needs $100 million of $460 million: 100 ÷ 460 ≈ 21.7%.
  3. Any share of at least about 21.7% keeps its sales from falling, so a share that dropped from 25% to 22% still means higher sales.

Check yourself. In 2025, what were the sales of all regions other than North?

Check your answer

70% of $460 million = $322 million (or 460 − 138).

What must be true from a table

A statement must be true only if it holds for every situation the table allows. Percents of one total fix comparisons between rows; percent changes fix comparisons between years; no percent fixes an actual count.

Why it works. Imagine the unknown total as 100, then as 1,000,000. Anything that changes between the two is not determined; anything that holds for both, and for every total, must be true.

How to spot it: "which of the following must be true", a table of percents with no totals.

Example. A table shows, by region, each region's percent of a company's 2025 employees and its percent change from 2024: A, 40% and +25%; B, 30% and −10%.

  1. 2025: A has 40% of the total and B 30%, so A had more. Must be true.
  2. 2024: A had 40 ÷ 1.25 = 32 and B had 30 ÷ 0.9 ≈ 33.3, in units of 1% of the 2025 total. So B had more in 2024. Must be true (whatever the total).
  3. "A had more than 500 employees in 2025": the total isn't given. Not necessarily true.

Trap: treating the percents as counts, or assuming a statement about amounts follows from the shares.

Edge cases.

  • A region with a positive percent change had more employees in 2025 than in 2024: that must be true.
  • The company's total change can be found: add each region's 2024 amount in units of the 2025 total and compare with 100.
  • "Could be true" is a different question: then any statement the data doesn't rule out qualifies.

GRE-level example. In the table above, did A's employee count change by more than B's?

  1. In units of 1% of the 2025 total: A changed by 40 − 32 = 8; B changed by 30 − 33.3 ≈ −3.3, a change of about 3.3.
  2. A's change is greater, whatever the total. Must be true.

Check yourself. Must it be true that B had fewer employees in 2025 than in 2024?

Check your answer

Yes: a −10% change means B's count fell, whatever the actual numbers are.

Charts of percent change

A chart of yearly percent changes shows how fast something grew, not how big it was. Every positive bar is growth, however short. Successive percents multiply, and with no starting value no actual amount can be found.

Why it works. Each bar is a growth factor: 10% means × 1.1. Multiplying the factors gives the overall change, and an unknown starting value multiplies every later level, so it cancels out of comparisons between years but not out of actual amounts.

How to spot it: a y-axis labelled "change from the previous year (%)".

Example. Membership grew 10% in 2022, 5% in 2023 and 20% in 2024.

  1. Membership rose every year, including 2023: the short bar is slower growth, not a fall.
  2. From the end of 2021 to 2024: 1.1 × 1.05 × 1.2 = 1.386, a 38.6% rise, more than 10 + 5 + 20 = 35.
  3. The increase in 2023 was L × 1.1 × 0.05 = 0.055L and in 2022 it was 0.1L, where L is the 2021 level: the 2022 increase was bigger.

Trap: reading the 2023 bar as a decline, or adding the percents.

0510152025change from previous year (%)20222023202410520still growth
Every bar is growth, the short 2023 bar too. Together: 1.1 × 1.05 × 1.2 = 1.386, a 38.6% rise, more than 10 + 5 + 20 = 35.

Edge cases.

  • Comparing two years' levels: if every bar between them is positive, the later year is greater.
  • An actual amount ("more than 1,200 members") can't be found without a starting value: the answer is (D).
  • A negative bar is a real decline.

GRE-level example. Compare Quantity A, the percent increase from 2022 to 2024, with Quantity B, 25%.

  1. From 2022 to 2024 the factors are 1.05 (2023) and 1.2 (2024): 1.05 × 1.2 = 1.26.
  2. Quantity A is 26%, greater than 25%: (A).

Check yourself. Was membership higher at the end of 2023 or at the end of 2022?

Check your answer

2023: the 2023 bar is positive (5%), so membership grew.

How hard questions disguise it

For harder questions (L4–L5)

L4 questions ask what the data can't tell you: an amount from percents alone, an actual level from a growth chart, or "must be true" statements that mix determinable comparisons with undeterminable amounts. L5 questions combine series or displays: add the series first, then take one percent change, or turn shares and totals into amounts before comparing years. Write down which numbers each step uses, and check that they come from the same year and the same units.

Practice

Practice these in timed sets on the data interpretation practice page.

Common traps

  • Misreading the display

    Check the scale (gridlines and the halfway marks between them), the units in the axis label, and which bar, year or segment the question names. A stacked segment's value is its height, not the height of its top edge.

  • Doing the wrong arithmetic with the values

    'How much greater' is a difference, a ratio keeps the order the question names, and 'A to B' compares A with B, not with the total. A percent share becomes an amount only after you multiply it by the total.

  • Getting percent change wrong

    Percent change = (later − earlier) ÷ earlier. A bigger rise in units can be a smaller percent rise, and 'more than 20%' excludes exactly 20%.

  • Adding or averaging percents

    Two years of 10% and 20% growth is 1.1 × 1.2 = 1.32, a 32% rise, not 30%. The percent change of a combined total is not the average of the separate percent changes.

  • Reading a percent-change chart as levels

    On a chart of yearly percent changes, a shorter positive bar is slower growth, not a decline, and no bar tells you an actual amount. Without a starting value or a total, actual amounts can't be determined.