Relvora Prep

GMAT Data Insights lesson · Filter & Sort

GMAT Data Insights: sorting and filtering a table

Table Analysis on the GMAT: sort by a column, find a row's position, apply a filter before comparing, and test two columns at once.

The core idea

Sorting answers 'what position', filtering answers 'among those that…'. Sort by clicking the column header, count positions from the end the statement names, and always strike out the rows a condition removes before you compare.

Core ideas

Sort by clicking a column header (click again to reverse). Positions are counted from the top of the sorted table. A filter keeps only rows meeting a condition; everything after it is about those rows only.

Table Analysis questions come in a few shapes:

The statement says… You do
"position 3 when sorted from greatest to least" sort, count from the top
"of the rows with X above 27, …" filter first, then compare
"in the top three in both columns" sort twice, keep rows on both lists
"the top two account for more than half" add the top rows, compare with half the total
Read the conditionfilterStrike out rowsclick headerSort what's leftanswerCount or add
Filter before you sort, and only then compare
  1. Read the whole statement and underline any condition ("of the warehouses with…").
  2. Strike out every row the condition removes.
  3. Sort the remaining rows by the column the statement names, in the direction it names.
  4. Count positions, compare, or add, using only the rows that are left.

A row's position from the top, sorted greatest to least, is 1 + the number of rows with a greater value. You can find it without sorting at all.

Ranking by a column

Why it works. Sorting lines the rows up by one column, so a position is just a count. Every row with a greater value sits above yours, so sorted greatest to least, your position is 1 + the number of rows with a greater value.

How to spot it: "When the table is sorted by Doctors from greatest to least, Bram is in position ___".

Example. Doctors at six clinics. Sorted from greatest to least, what position is Bram?

Clinic Doctors
Alton 16
Bram 18
Corby 14
Dene 21
Esher 6
Frith 22
0510152025DoctorsFrithDeneBramAltonCorbyEsher22211816146position 3
The same clinics sorted from greatest to least
  1. Sort greatest to least: 22, 21, 18, 16, 14, 6.
  2. Bram's 18 is third: position 3.
  3. Check: two clinics (Frith, Dene) have more doctors, and 1 + 2 = 3.

Trap: counting from the smallest end puts Bram 4th. "Greatest to least" means position 1 is the largest.

Try it. Five libraries lend 340, 415, 290, 388 and 402 books a week. Sorted from least to greatest, what position is the library that lends 402?

Show the answer
  1. Sort least to greatest: 290, 340, 388, 402, 415.
  2. 402 is fourth: position 4.
  3. Check: three libraries lend fewer books, and 1 + 3 = 4. ✓

Edge cases.

  • Flipping the direction. With n rows, position p from the top is position n + 1 − p from the bottom. In the six clinics, Bram is 3rd from the top and 6 + 1 − 3 = 4th from the bottom.
  • Ties. Two rows with the same value can sort either way. Check whether the answer depends on which one comes first.
  • A computed column. If the statement ranks by a rate ("doctors per 1,000 people"), work out that column first, then sort it.

GMAT-level example. Five towns. Sorted by clinics per 10,000 people from greatest to least, what position is Arlo? And sorted by number of clinics?

Town Population Clinics
Arlo 40,000 12
Bevan 25,000 9
Colm 60,000 15
Dorr 50,000 17
Esk 20,000 8
  1. Clinics per 10,000 people = clinics ÷ (population ÷ 10,000). Arlo: 12 ÷ 4 = 3.0. Bevan: 9 ÷ 2.5 = 3.6. Colm: 15 ÷ 6 = 2.5. Dorr: 17 ÷ 5 = 3.4. Esk: 8 ÷ 2 = 4.0.
  2. Sort the rates from greatest to least: Esk 4.0, Bevan 3.6, Dorr 3.4, Arlo 3.0, Colm 2.5. Arlo is in position 4.
  3. Sort the clinic counts: Dorr 17, Colm 15, Arlo 12, Bevan 9, Esk 8. Here Arlo is in position 3.
  4. Check: three towns (Esk, Bevan, Dorr) have a higher rate than Arlo's 3.0, and 1 + 3 = 4 ✓.
012345Clinics per 10,000 peopleEskBevanDorrArloColm43.63.432.5position 42nd by count
Sorted by the rate, Esk (fewest clinics) leads and Colm (second-most clinics) is last

The rate column and the count column give different orders: Esk has the fewest clinics but the highest rate. Sort the column the statement names.

Check yourself. A table has 7 rows. A row is in position 2 when sorted from greatest to least. What is its position when sorted from least to greatest?

Check your answer

7 + 1 − 2 = position 6.

Filter, then find the greatest

Why it works. A filter changes which rows the question is about. The greatest value among the kept rows can belong to any of them, so the overall leader matters only if it survives the filter.

How to spot it: "Of the warehouses with more than 27 thousand m² of floor area, Ironton shipped the most orders."

Example. Is the statement above true?

Warehouse Floor area (thousand m²) Orders (thousand) Area > 27?
Ashford 28 290 kept
Brookline 25 360 out
Carver 31 305 kept
Dunmore 22 210 out
Ironton 29 324 kept
Eastfield 27 340 out (equal)
  1. Filter: keep floor area above 27. That leaves Ashford (28), Carver (31) and Ironton (29). Eastfield, at exactly 27, is out.
  2. Compare orders among those three: 290, 305, 324.
  3. Ironton's 324 thousand is the greatest, so the statement is true.
050100150200250300350400Orders (thousand)AshfordBrooklineCarverDunmoreIrontonEastfield290360305210324340most of keptfiltered outfiltered out
The two tallest bars belong to warehouses the filter removes

Trap: Brookline ships the most overall (360), and Eastfield is second (340). Both are removed by the filter. Forgetting it, or keeping Eastfield's 27 as "more than 27", makes the statement look false.

Try it. Of the bus routes longer than 12 km, which has the fewest riders?

Route Length (km) Riders per day
R1 14 820
R2 9 610
R3 12 540
R4 17 905
R5 13 760
Show the answer
  1. Filter: keep lengths above 12. That leaves R1 (14), R4 (17) and R5 (13). R3, at exactly 12, is out, and so is R2.
  2. Compare riders among those three: 820, 905, 760.
  3. The fewest is R5, with 760.
  4. Check: the two routes with fewer riders, R3 (540) and R2 (610), are both removed by the filter. ✓

Edge cases.

  • "More than" leaves the boundary out; "at least" keeps it. Read the exact words before you strike a row that sits on the cut-off.
  • The filter column and the compare column differ. Filter on floor area, then sort by orders. Don't sort by the filter column.
  • Two conditions. "And" keeps rows that pass both; "or" keeps rows that pass either.

GMAT-level example. Of the hotels with at least 120 rooms and a guest rating above 4.0, which fills the most rooms on an average night?

Hotel Rooms Occupancy Rating
Alba 150 72% 4.3
Birch 110 90% 4.6
Cove 200 58% 4.1
Dune 120 85% 4.0
Elm 180 70% 3.8
Fern 140 80% 4.4
  1. First condition, at least 120 rooms: Birch (110) is out. Dune's 120 counts.
  2. Second condition, rating above 4.0: Dune (exactly 4.0) and Elm (3.8) are out. That leaves Alba, Cove and Fern.
  3. Rooms filled = rooms × occupancy. Alba: 150 × 0.72 = 108. Cove: 200 × 0.58 = 116. Fern: 140 × 0.80 = 112.
  4. The most is Cove, with 116 rooms filled.
  5. Check: 116 > 112 > 108 ✓. Cove has the lowest occupancy of the three, but the most rooms.
020406080100120140Rooms filled per nightAlbaBirchCoveDuneElmFern10899116102126112most of keptfiltered out
Elm fills the most rooms overall, but its 3.8 rating removes it; Cove leads the hotels that are left

The trap answers are Elm (most rooms filled, but the rating filter removes it) and Birch (highest occupancy, but too few rooms, and occupancy isn't what the question asks).

Check yourself. In the warehouse table above, which warehouse with at least 27 thousand m² of floor area shipped the most orders?

Check your answer

"At least 27" keeps Eastfield, so the kept orders are 290, 305, 324 and 340. Eastfield (340 thousand) ships the most.

Ranking by two columns

Why it works. Each sort is its own ranking, and two columns measure different things, so a row's place in one tells you nothing about the other. "In the top three in both" means the row is on both lists: you need both sorts.

How to spot it: "Bexley is among the top three stores in both revenue and profit."

Sort by the first column and write down the top rows, sort by the second and write them down again. Only rows on both lists are "in the top three in both".

Example. Six stores. Which stores are in the top three for both revenue and profit?

Store Revenue ($ million) Profit ($ million)
Avon 48 5.2
Bexley 52 6.1
Crane 45 7.0
Delph 39 6.5
Eston 50 4.1
Frome 41 3.8
  1. Sort by revenue: Bexley 52, Eston 50, Avon 48, Crane 45, Frome 41, Delph 39. Top three: Bexley, Eston, Avon.
  2. Sort by profit: Crane 7.0, Delph 6.5, Bexley 6.1, Avon 5.2, Eston 4.1, Frome 3.8. Top three: Crane, Delph, Bexley.
  3. On both lists: only Bexley.
Rank By revenue By profit
1 Bexley Crane
2 Eston Delph
3 Avon Bexley

Trap: Eston is 2nd in revenue but 5th in profit. Leading one column says nothing about the other, so "Eston is in the top three in both" is false. Bexley leads revenue but is only 3rd in profit, so "the revenue leader has the greatest profit" is false too.

Try it. In the same six stores, which stores are in the top four for both revenue and profit?

Show the answer
  1. Top four by revenue: Bexley, Eston, Avon, Crane.
  2. Top four by profit: Crane, Delph, Bexley, Avon.
  3. On both lists: Bexley, Avon and Crane. Eston is out on profit, and Delph is out on revenue.

Edge cases.

  • "In at least one" is the other way round. It keeps rows on either list. The stores' two top-three lists hold five different stores, since only Bexley is on both.
  • "Ranks higher by X than by Y" compares two positions of the same row: Crane is 4th by revenue but 1st by profit.
  • A tie at the cut-off can put four rows into a "top three". Look at the values just below the line.

GMAT-level example. Five cities. Decide whether each statement is true.

City Bike lanes (km) Riders (thousand per day)
Avila 120 18
Brest 95 22
Cadiz 140 15
Dover 80 9
Ennis 110 20

(a) The city with the most bike lanes is in the top two for riders. (b) Exactly one city is in the top three for both columns. (c) The city with the fewest bike lanes also has the fewest riders.

  1. Sort by lanes: Cadiz 140, Avila 120, Ennis 110, Brest 95, Dover 80.
  2. Sort by riders: Brest 22, Ennis 20, Avila 18, Cadiz 15, Dover 9.
  3. (a) Cadiz leads lanes but is 4th for riders: false.
  4. (b) Top three by lanes: Cadiz, Avila, Ennis. Top three by riders: Brest, Ennis, Avila. Two cities, Avila and Ennis, are on both: false.
  5. (c) Dover is last in both lists: true.
Rank By lanes By riders
1 Cadiz Brest
2 Avila Ennis
3 Ennis Avila
4 Brest Cadiz
5 Dover Dover

Check: Brest is 4th by lanes but 1st by riders, and Cadiz is the reverse ✓. Writing both lists side by side answers all three statements at once.

Check yourself. In the stores table, the store ranked 2nd by profit is Delph. Is it in the top three by revenue?

Check your answer

No. Delph's revenue of 39 is the lowest of the six, so it is 6th by revenue.

Top rows' share of a total

Why it works. "More than half" compares a sum with half the total. The total is the top rows plus the rest, so the top rows pass half exactly when they are bigger than the rest. That is why the shortcut below works.

How to spot it: "The three warehouses with the most workers employ more than half of all the workers."

Example. Workers at seven warehouses: 40, 39, 31, 26, 25, 19, 11. Do the top two employ more than half? The top three?

01020304050WorkersW1W2W3W4W5W6W740393126251911
Sorted from most to fewest workers; the top three are marked
  1. Total: 40 + 39 + 31 + 26 + 25 + 19 + 11 = 191. Half is 95.5.
  2. Top two: 40 + 39 = 79. Less than 95.5, so no.
  3. Top three: 79 + 31 = 110. More than 95.5, so yes (about 57.6%).
All 191top three 110other four 81Half95.5
The top three's segment passes the halfway mark

A shortcut: compare the top rows with the rest. Top three 110 against the other four 81: more than half, since 110 > 81.

Trap: judging by eye. Seven bars where the top two look big can still fall short of half. Add them.

Try it. Five products sold 34, 22, 18, 15 and 11 thousand units. Do the top two products account for more than 55% of the units sold?

Show the answer
  1. Total: 34 + 22 + 18 + 15 + 11 = 100 thousand.
  2. Top two: 34 + 22 = 56 thousand, which is 56% of 100.
  3. 56% > 55%, so yes, just.

Edge cases.

  • "At least half" includes exactly half. "More than half" doesn't.
  • "Top" means the largest values, not the first rows as listed. Sort first.
  • A share of a filtered group uses the filtered group's total, not the whole column's.

GMAT-level example. Six stores. Of the urban stores, do the two with the most sales account for more than 60% of urban sales?

Store Type Sales ($ thousand)
A Urban 420
B Rural 510
C Urban 380
D Urban 150
E Rural 290
F Urban 250
  1. Filter: keep the urban stores, A, C, D and F.
  2. Sort their sales: 420, 380, 250, 150. Urban total: 1,200.
  3. Top two urban: 420 + 380 = 800.
  4. 60% of 1,200 = 720, and 800 > 720, so yes (800 ÷ 1,200 ≈ 66.7%).
  5. Check with the shortcut: the other two urban stores sold 250 + 150 = 400, and 800 is twice that, so the top two hold 2/3 ✓.
Urban 1,200top two 800other two 40060% mark720
Within the urban stores, the top two's 800 passes the 60% mark of 720

The trap answers use the wrong total: 800 ÷ 2,000 (all six stores) = 40%, or the top two of all stores, B and A, with 930 ÷ 2,000 = 46.5%.

Check yourself. In the seven warehouses above (total 191 workers), do the top four employ more than 70% of the workers?

Check your answer

Top four: 40 + 39 + 31 + 26 = 136. 70% of 191 = 133.7, and 136 > 133.7, so yes (about 71.2%).

How hard questions disguise it

For harder questions (L4–L5)

At L4 and L5 the sort is rarely on a column you can click. The statement ranks by a rate you must compute first, such as a ratio of two columns (ratios) or a change between two years (percent change), and the order of that computed column is usually different from the order of either raw column. Filters stack: two or three conditions joined by "and", with one row sitting exactly on a boundary. Some questions ask how many rows pass the filter (counting rows that meet conditions), or for the median of the rows that are left (mean and median), and share questions use the filtered group's total (part and whole). Write the kept rows down once, then answer each statement from that short list.

Practice

Practice sorting and filtering questions in timed sets on the sorting and filtering practice page.

Common traps

  • Counting from the wrong end

    'Sorted from greatest to least, position 2' is the second largest. A row's position is one more than the number of rows with a greater value.

  • Forgetting the filter

    The overall leader in a column is often a row the condition removes. Filter first, then compare.

  • Checking only one column

    'Top three in both columns' needs two sorts and both lists. Leading one column says nothing about another.

  • Judging a share without adding

    Whether the top rows make more than half of a column is a sum: add them and compare with half the column total.