Relvora Prep

GMAT Data Insights lesson · Multi-Source: combining sources

GMAT Multi-Source Reasoning: combining sources

How to combine a memo, a table and a chart in GMAT Multi-Source Reasoning: code keys, units, adjustments, must-be-true ranges and unsupported claims.

The core idea

Multi-Source Reasoning splits the facts across tabs, and every statement needs at least two of them. Read every tab first and note its keys, units and rules; then, for each statement, find the tabs it needs, translate and convert, and only then compute or compare.

Core ideas

Words used here

  • Tab: one source in the set: a text (a memo, notes, an email), a table or a chart.
  • Key: a list in one tab that says what the codes in another tab mean.
  • Measure: what a memo says to compare, for example tickets sold per member of staff.
  • Must be true: true for every value the tabs allow, not just for some of them.

A Multi-Source Reasoning set gives you two or three tabs about one scenario, and every statement needs facts from at least two of them. One tab alone can't answer a statement, so a statement that seems to need only the tab you are looking at is a warning sign: something in another tab (a key, a unit, an adjustment) changes the answer.

Read every tabNote keys, units, rulesFind each statement's tabsTranslate and convertCompute, then check
Read all the tabs once before the statements, then work each statement from the tabs it needs
  1. Read every tab once before the statements. In each one, look for the things that change other tabs: a code key, a unit ("in thousands"), a rule ("leave these out", "members visit free"), a target or a range.
  2. For each statement, list the tabs it needs. A rate times an amount needs the table and the chart; a ranking needs the memo's measure too.
  3. Translate and convert first. Turn codes into names, thousands into exact numbers and adjusted figures into the ones the memo wants.
  4. Then compute or compare, and check the answer against the exact words of the statement ("more than", "at least", "must").
Tab What to look for How it changes the other tabs
Memo or notes keys, rules, adjustments, targets, ranges says what a code means and which figures count
Table row names, column headers and their units gives the rates, counts and targets
Chart axis label and its unit, series names gives amounts, often in thousands

Looking up a fact across tabs

The table often shows a short code (a depot, a crop, a zone), and only the memo says what each code means. Translate the code with the key, then read the rows.

Why it works. Neither tab says "Field 15 is sown with wheat": the table says Field 15 has code W, and the memo says W means wheat. Put together, the two facts give a third that neither tab states.

How to spot it: a column of letters or short codes in the table, and a line in the memo such as "Crop codes: W = wheat; B = barley; R = rye".

Example. Memo: Crop codes: W = wheat; B = barley; R = rye. Table:

Field Crop code Hectares Tonnes harvested
Field 3 B 18 95
Field 8 W 25 210
Field 11 R 14 60
Field 15 W 31 240
Field 22 B 22 150
Field 26 W 12 105
  1. "The field with the most tonnes harvested is sown with wheat." The most is Field 15 (240). Its code is W, and W = wheat: Yes.
  2. "Every field that is sown with wheat had more than 15 hectares." The W rows are Field 8 (25), Field 15 (31) and Field 26 (12). Field 26 has only 12: No.
  3. "Exactly 2 of the 6 fields are sown with barley." The B rows are Field 3 and Field 22: Yes.

Trap: stopping after the first two W rows in statement 2. Fields 8 and 15 are both above 15 hectares, but the statement is about every wheat field, and Field 26 is one of them. Find every row with the code before you decide.

Try it. With the same tabs: (a) "Every field that is sown with barley harvested more than 90 tonnes." (b) "The field with the fewest hectares is sown with rye."

Show the answer
  1. (a) The B rows harvested 95 and 150 tonnes. Both are more than 90: Yes.
  2. (b) The fewest hectares is Field 26 (12), coded W, which is wheat: No. Field 11 (14, code R) is only second fewest; it is the tempting wrong row.

Check yourself. A memo says Z1 = the river zone. The only two rows coded Z1 show 680 and 600 pupils. Is "Every school that is in the river zone had more than 600 pupils" true?

Check your answer

No. 600 is not more than 600, so the school with exactly 600 fails the statement.

Computing across tabs

A total that no tab shows is often a rate from one tab times an amount from another: a price per visitor times the number of visitors, a yield per hectare times the hectares.

Why it works. The table gives "dollars per visitor" and the chart gives "visitors". Multiplying cancels the visitors and leaves dollars, but only if both numbers are about the same site, the same group and the same unit.

How to spot it: a column with "per" in its header (or a fare, a charge, a yield), and a chart of the amounts.

Example. Table: Average spending per visitor, May.

Site Shop spend per visitor ($) Café spend per visitor ($)
Abbey 9 6
Castle 7 11
Mill 4 8
Priory 10 5

Chart: Visitors in May (thousands).

02468101214161820Visitors (thousands)AbbeyCastleMillPriory181291512 thousand
Castle's bar is 12 on an axis in thousands: 12,000 visitors

What was the total café spend by visitors to the Castle site in May?

  1. Table, Castle's row, café column: $11 per visitor.
  2. Chart, Castle's bar: 12 thousand, that is 12,000 visitors.
  3. 11 × 12,000 = $132,000.

Trap: forgetting the thousands gives 11 × 12 = $132. The other wrong choices each take one wrong cell: the shop column (7 × 12,000 = $84,000), Abbey's bar (11 × 18,000 = $198,000), Mill's row (8 × 12,000 = $96,000), or both columns added ((7 + 11) × 12,000 = $216,000).

Try it. What was the total shop spend by visitors to the Priory site in May?

Show the answer

Shop spend per visitor at Priory: $10. Priory's bar: 15 thousand = 15,000 visitors. 10 × 15,000 = $150,000.

Check yourself. A chart in thousand cubic metres shows 20 for business customers, and the table charges $5 per cubic metre. What is the total charge?

Check your answer

20 thousand cubic metres = 20,000 cubic metres. 5 × 20,000 = $100,000.

Comparing across tabs

A memo often says how to compare ("ranked on tickets sold per member of staff"), sets a target and makes an adjustment ("leave out the festival shuttle tickets"). The chart and the table give the figures. Apply the memo to the figures before you compare.

Why it works. A ranking per member of staff divides each station's tickets by its staff, so a big station with many staff can rank below a small one. An adjustment changes the figure that is divided, so it can change the order too.

How to spot it: words like "ranked on", "per", "meets the target", "leave these out" in the memo.

Example. Memo: Stations are ranked on tickets sold per member of staff. A station meets the target if it sells at least 2,000 tickets per member of staff. Colne's figure in the chart includes 3,000 tickets sold for the festival shuttle; leave these out of every comparison, including the target. Chart (tickets sold, thousands): Ashby 18, Bramley 22, Colne 16, Denton 12. Table (staff): Ashby 9, Bramley 12, Colne 6, Denton 5.

Station Tickets counted Staff Per member of staff
Ashby 18,000 9 2,000
Bramley 22,000 12 about 1,833.33
Colne 16,000 − 3,000 = 13,000 6 about 2,166.67
Denton 12,000 5 2,400
Colne in the chart13,000 countedColne counted13,000 ÷ 6
Leave out the 3,000 shuttle tickets first, then divide by Colne's 6 staff: about 2,166.67
  1. "Colne ranks first of the 4 stations." Without the adjustment, Colne would be first (16,000 ÷ 6 ≈ 2,666.67). With it, Denton's 2,400 is the highest: No.
  2. "Bramley ranks above Denton." Bramley has the biggest total (22,000), but per member of staff it is the lowest (about 1,833.33 against 2,400): No.
  3. "The number of stations that meet the target is 3." At least 2,000: Denton, Colne and Ashby (exactly 2,000): Yes.

Trap: each statement has a shortcut that flips it. Skipping the memo's adjustment makes Colne first; comparing the chart's totals puts Bramley on top; and reading "at least 2,000" as "more than 2,000" leaves Ashby out, giving 2.

Try it. With the same tabs, does Ashby rank above Bramley?

Show the answer

Ashby: 18,000 ÷ 9 = 2,000. Bramley: 22,000 ÷ 12 ≈ 1,833.33. Yes, Ashby ranks above Bramley, even though Bramley sold more tickets.

Check yourself. If the memo's target were "more than 2,000 tickets per member of staff", how many of the four stations would meet it?

Check your answer

2: Denton (2,400) and Colne (about 2,166.67). Ashby is exactly 2,000, which is not more than 2,000.

Inferring what must be true

Some statements ask what must be true. The tabs fix some numbers exactly and leave others inside a range ("between 20% and 30% of paying visits included an audio guide"). Select Yes only if the statement holds for every value the tabs allow.

Why it works. A range gives a smallest and a largest possible value. "More than N" must be true when even the smallest value is above N. If N lies inside the range, the statement could be true, but it needn't be, so the answer is No.

How to spot it: "must be true", "can be determined", and a memo with "between … and …", "at least", "no more than".

Example. Memo: Members visit free. Every other visit is charged that month's adult ticket price. In each month, between 20% and 30% of the paying visits included an audio guide rental. Chart (visits, thousands): May 30, June 40. Table: member visits (% of all visits) May 20, June 35; adult ticket May $12, June $15.

  1. Paying visits: May 30,000 × 80% = 24,000; June 40,000 × 65% = 26,000.
  2. Audio guides: May 20% to 30% of 24,000 = 4,800 to 7,200; June 20% to 30% of 26,000 = 5,200 to 7,800.
May: 4,800 to 7,200June: 5,200 to 7,8004,0005,0006,0007,0008,0004,5006,000
4,500 lies below all of May's range (must be true); 6,000 lies inside it (could be true)
  • "More than 4,500 audio guides were rented in May." Even the smallest possible figure, 4,800, is more: Yes.
  • "More than 6,000 audio guides were rented in May." Anything from 4,800 to 7,200 fits, so it could be 5,000: No.
  • "More audio guides were rented in June than in May." June had more paying visits, but the ranges overlap: June at 20% (5,200) and May at 30% (7,200) fits the tabs, and then May has more: No.
  • "There were more visits by members in June than in May." June 40,000 × 35% = 14,000; May 30,000 × 20% = 6,000. These are exact: Yes.

Trap: answering Yes to "could be true". A statement that holds at one end of a range but not at the other is not a must. The reverse slip is answering No when the exact value is unknown but every possible value gives the same answer, as in the first statement.

Try it. With the same tabs: (a) "More than 5,000 audio guides were rented in June." (b) "Ticket revenue was higher in June than in May."

Show the answer
  1. (a) June's smallest possible figure is 5,200, more than 5,000: Yes.
  2. (b) May: 24,000 × $12 = $288,000. June: 26,000 × $15 = $390,000. Yes, and it is exact, not a range.

Check yourself. Suppose the memo's range were 25% to 30% instead. Must more audio guides have been rented in June than in May?

Check your answer

No. June's smallest is 25% of 26,000 = 6,500, and May's largest is 30% of 24,000 = 7,200. 6,500 is less than 7,200, so May could still have more.

Spotting a claim the sources don't support

The hardest sets ask which claim (in a draft report, a press release, a newsletter) the tabs do not support. Check each claim against the tabs it needs. Four are supported; the one that fails is the answer.

Why it works. Each claim is a small calculation: a rate, a rise, a count against targets, a total. Done carefully, each one comes out true or false, so exactly one claim fails.

How to spot it: "Which of the following is NOT supported", "inconsistent with", "contradicted by".

Example. Notes: The chart gives 2026 loans in thousands; the table gives exact figures. A branch reached its target if its 2026 loans were at least the target.

Branch Loans in 2025 Registered members Target for 2026
Abbey 30,000 2,500 32,000
Bridge 21,000 1,600 24,000
Castle 36,000 4,000 34,000
Dock 18,500 1,500 20,000

Chart (loans in 2026, thousands): Abbey 32, Bridge 22, Castle 38, Dock 20.

Which claim is NOT supported?

  1. "Bridge had the most loans per registered member in 2026." Per member: Abbey 32,000 ÷ 2,500 = 12.8; Bridge 22,000 ÷ 1,600 = 13.75; Castle 38,000 ÷ 4,000 = 9.5; Dock 20,000 ÷ 1,500 ≈ 13.33. Bridge is highest: supported, even though Castle made the most loans.
  2. "Dock's loans rose by more than 8% from 2025 to 2026." 20,000 − 18,500 = 1,500, and 1,500 ÷ 18,500 ≈ 8.1%: supported.
  3. "The number of branches that reached their 2026 targets was 3." Abbey (32,000, exactly on target), Castle and Dock (20,000, exactly on target): 3, supported.
  4. "Dock had more loans per registered member than Abbey in 2026." About 13.33 against 12.8: supported.
  5. "Total loans across the 4 branches fell from 2025 to 2026." 2026: 32,000 + 22,000 + 38,000 + 20,000 = 112,000; 2025: 30,000 + 21,000 + 36,000 + 18,500 = 105,500. The total rose: not supported.
202518,500the base202618,500
Dock's rise of 1,500 is a percent of 2025's 18,500 (about 8.1%), not of 2026's 20,000 (7.5%)

Trap: each supported claim has a slip that makes it look false. Dividing Dock's rise by the 2026 figure gives 1,500 ÷ 20,000 = 7.5%, under 8%. Counting only branches above their targets gives 1, not 3. Comparing loans instead of loans per member makes Castle the leader.

Try it. With the same tabs, is "Bridge's loans rose by more than 5% from 2025 to 2026" supported?

Show the answer

22,000 − 21,000 = 1,000, and 1,000 ÷ 21,000 ≈ 4.8%. That is not more than 5%: not supported.

Check yourself. Three bike hubs: one fell from 10,000 hires to 9,000, another from 12,000 to 11,000, and the third rose from 20,000 to 24,000. Two of the three fell. Did total hires rise or fall?

Check your answer

Total: 42,000 before, 44,000 after. Rose. Compare the totals, not how many hubs went up or down.

Practice

Practice combining sources on the practice page.

Common traps

  • Translating a code wrongly, or missing a row in the group

    When the table shows only codes, a statement about 'the Northgate depot' is about every row coded N, and no other row. Translate with the memo's key, then tick off every row with that code before you decide.

  • Taking a figure from the wrong row, column or series

    The table and the chart list the same names, so it is easy to take the neighbouring bar or row, the other rate column, or the other series. Name the row, the column and the series out loud before you read a number.

  • Mixing units between tabs

    A chart in thousands and a table of exact counts don't mix until you convert: a bar of 12 is 12,000. Read each axis label and column header for its unit before you combine them.

  • Skipping the memo's adjustment or condition

    If the memo says to leave something out (a one-off sale, the visits that are free), do it before you divide or compare. The unadjusted answer is almost always one of the tempting choices.

  • Comparing totals where a rate is asked, or answering from one tab

    'Per member of staff' means divide first. The biggest bar is often not the biggest rate, a bigger share of a smaller total can be fewer people, and a total can rise even when most of its parts fall.

  • Confusing 'could be true' with 'must be true'

    When a tab gives a range, test both ends. A statement must be true only if it holds at every value the sources allow; if it holds at both ends, it must be true even though the exact value is unknown.

  • Getting a threshold or a count wrong

    'At least the target' includes a value exactly on the target; 'more than' doesn't. Count rows that pass with ticks, not by eye.

  • Measuring a rise against the wrong year

    A rise from last year to this year is a percent of last year's figure. Dividing by this year's figure gives a smaller percent and can make a true claim look false.