CAT Data Interpretation Caselets: Build the Table the Setter Withheld
A caselet is a data interpretation set with the table removed. The numbers arrive as prose — embedded in sentences, spread across paragraphs, some stated and some only implied — and your first job is to rebuild the table the setter deliberately did not give you.
Candidates who try to answer questions directly from the paragraph lose to candidates who spend two minutes constructing a grid first. That is the entire difference in this question type, and it is almost never taught.
Extract before you calculate
Two minutes spent turning prose into a table is not overhead. It converts every subsequent question from a rereading exercise into a lookup — and a caselet typically carries four or five questions, so the investment repays several times over.
Quick Answer (30-Second Read)
First move
Tabulate
Before any question
Calculation
Approximate
Options are far apart
- Draw the grid first, with rows and columns named, then fill what the prose states.
- Mark derived cells differently from stated ones — you will need to know which is which.
- Check the totals row. A caselet almost always hides one value that a total reveals.
- Approximate aggressively. Exact arithmetic is rarely required and always slower.
- Read the units and the base of every percentage. This is where most caselets are lost.
Section structure per the official CAT bulletin at iimcat.ac.in.
CAT Mock Test — DI under the section that closes
Tabulating costs two minutes you only have if the rest of the section is under control. Practise it where that is true.
Why a caselet is harder than a chart
| Chart or table DI | Caselet | |
|---|---|---|
| Where the data sits | Already organised for you | Scattered through prose, some of it implied |
| First task | Read the axes and units | Build the structure, then populate it |
| Where candidates lose | Calculation speed | Extraction — a missed value invalidates everything after |
| Setup cost | Near zero | About two minutes, and worth every second |
| Reward for setup | Small | Large — four or five questions become lookups |
The final row is why caselets are worth attempting rather than avoiding. A chart set gives everyone the same head start; a caselet rewards the candidate who does two minutes of unglamorous work that most people skip because it does not feel like solving.
The extraction discipline
| Step | What you do | Why it matters |
|---|---|---|
| 1 | Read the whole caselet once without writing anything | Tells you what the rows and columns should be |
| 2 | Draw the empty grid, naming every row, column and unit | An unnamed column is a wrong answer waiting to happen |
| 3 | Fill only the values the text states outright | Separates given from derived, which you will need later |
| 4 | Add a totals row and a totals column | The hidden value is almost always recoverable from a total |
| 5 | Derive the remaining cells, marking them differently | If a derivation was wrong, you can find it without rebuilding |
The totals row is where caselets give themselves away
Setters withhold a value and supply a total that makes it recoverable. A candidate working question by question never notices; a candidate who writes the totals row as a matter of routine finds the missing number before reading a single question. Add the row even when nothing in the text mentions totals.
Mark derived cells with a circle, a different pen, or brackets — any consistent convention. When an answer comes out impossible, the fault is nearly always in a derived cell, and being able to see at a glance which cells were inferred turns a rebuild into a ten-second check.
A worked caselet
The caselet. A store sells three products — A, B and C — across two quarters. Total units sold across both quarters was 1,180. In Q1, A sold 120 units and B sold twice as many as A. C's Q1 sales were 60 units fewer than B's. Q2 sales exceeded Q1 sales by 100 units. In Q2, B sold 200, and A and C sold equal numbers.
Step 2 — draw the grid. Rows A, B, C. Columns Q1, Q2, Total. Everything is in units and no percentages appear, which is worth confirming before starting rather than discovering later.
Step 3 — fill only what is stated. Q1: A = 120. B = 2 × 120 = 240. C = 240 − 60 = 180. Nothing in Q2 is directly stated except B = 200.
Step 4 — the totals row does the work. Q1 column sums to 540. Q2 exceeds it by 100, so Q2 = 640 — and the overall total confirms it, since 1,180 − 540 = 640. Two independent routes to the same figure, which means the reading is right.
Step 5 — derive the rest. In Q2, B = 200, so A + C = 440, and since they are equal, A = C = 220. Row totals: A = 340, B = 440, C = 400, summing to 1,180. Grid complete in about two minutes, and every question is now a lookup.
Notice that step four earned its place twice. It produced the Q2 total, which nothing in the prose states directly, and it then verified that figure by a second route. That cross-check is the real reason to write a totals row: if the two routes disagree, you have misread a condition, and finding that at extraction costs seconds where finding it three questions later costs the set.
Are DI sets or reasoning sets costing you?
They fail differently — extraction versus representation — and the fix is different. The set-level breakdown separates them.
- Time inside each set
- Sets completed vs started
- DILR percentile
Approximating safely
Exact arithmetic is rarely required. The options in a DI question are usually far enough apart that a two-significant-figure estimate identifies the answer, and the candidate doing long division is spending ninety seconds to gain precision nobody asked for.
| Situation | Approximate | Do not approximate |
|---|---|---|
| Options differ by more than 10% | Freely — round to two significant figures | — |
| Options differ by 2–5% | Carefully, and check the two nearest | — |
| Options differ by under 1% | — | Compute properly; the setter chose those options deliberately |
| The question asks "approximately" | Always — it is an explicit invitation | — |
| A TITA question with a typed answer | — | No options to lean on; compute exactly |
Look at the options before calculating. Two seconds spent checking how far apart they are determines whether the next thirty seconds should be an estimate or a full computation, and that decision is worth more than any calculation shortcut.
Useful approximations worth having automatic: a percentage change is roughly the difference divided by the original, so a move from 180 to 200 is about 11%; dividing by 7 is close to multiplying by 0.143; and fractions like one-eighth, one-twelfth and one-sixteenth as percentages are worth knowing on sight rather than deriving.
Four traps in the data itself
| Trap | How it appears | Defence |
|---|---|---|
| Shifting percentage base | "30% of the total" then "40% of the remainder" | Write the base beside every percentage as you extract it |
| Mixed units | Some figures in thousands, some absolute | Convert everything to one unit at extraction, not later |
| Growth stated on different periods | A yearly rate compared against a quarterly one | Normalise to a common period before comparing |
| Ratios given without a total | "A to B is 3:2" with no absolute anywhere | Work in parts until an absolute appears; do not invent one |
The first trap is the commonest and the most costly. "Forty per cent of the remainder" means forty per cent of what was left after the previous deduction, not of the original total, and a candidate who takes both percentages against the same base produces a table that is internally consistent and entirely wrong. Writing the base next to each percentage as you extract it costs nothing and prevents the error completely.
Key Takeaways
- A caselet is DI with the table removed. Rebuild the table before answering anything.
- Two minutes of extraction turns five questions into lookups. It is not overhead.
- Always write a totals row, even when the text never mentions totals. That is where the hidden value sits.
- Mark derived cells differently from stated ones. It turns a rebuild into a ten-second check.
- Check option spacing before calculating. It decides estimate versus exact.
- Write the base beside every percentage. A shifting base produces a consistent, wrong table.
People also search for
What is a caselet in CAT DI?
A data interpretation set presented as prose rather than as a chart or table. The numbers are embedded in sentences, spread across paragraphs, with some values stated and others recoverable only by inference. The defining feature is that the organising structure has been deliberately withheld, which makes rebuilding it the first task — and candidates who instead answer questions directly from the paragraph lose to those who spend two minutes constructing a grid.
How do I solve caselet DI questions?
In five steps. Read the whole caselet once without writing anything, so you know what the rows and columns should be. Draw the empty grid with every row, column and unit named. Fill in only the values stated outright. Add a totals row and column. Then derive the remaining cells, marking them differently from the stated ones. Only after the grid is complete should you look at the questions, which by then are mostly lookups.
Why should I make a table for a caselet?
Because a caselet typically carries four or five questions, and a table converts each of them from a rereading exercise into a lookup. Two minutes of extraction repays itself several times over. It also protects against the failure mode specific to this question type: a value misread in paragraph three invalidates every answer that depends on it, and without a table that error surfaces only after several wrong answers rather than immediately.
What is the most common mistake in caselet questions?
Taking two percentages against the same base when the second one shifted. "Thirty per cent of the total" followed by "forty per cent of the remainder" means forty per cent of what was left after the first deduction, not of the original figure. A candidate who misses this builds a table that is internally consistent and entirely wrong, which is worse than an obvious error. Writing the base beside every percentage as you extract it prevents it completely.
Should I approximate in CAT data interpretation?
Usually yes, and the decision should be made by looking at the options first. When they differ by more than ten per cent, round to two significant figures and move on. When they differ by two to five per cent, estimate carefully and check the two nearest. When they differ by under one per cent, the setter chose those options deliberately and you should compute properly. On typed-answer questions there are no options to lean on, so exact calculation is required.
How long should a caselet take?
Around two minutes to build the table and roughly a minute per question after that, putting a five-question caselet at about seven minutes in total. If the extraction alone is running past three minutes, either the caselet is unusually dense or your grid structure is wrong — and in a forty-minute section that is the moment to consider whether another set is a better use of the remaining time.
Why should I add a totals row even when the caselet does not mention totals?
Because setters routinely withhold one value and supply a total from which it is recoverable. A candidate working question by question never notices the connection; one who writes the totals row as routine finds the missing number before reading a single question. The totals row also verifies your reading of the prose — if the numbers do not reconcile, you have misread a condition, and finding that at extraction is far cheaper than finding it three questions later.
Are caselets harder than chart-based DI?
They have a higher setup cost and a lower ceiling on difficulty once the table exists. A chart gives everyone the same head start, so the competition is on calculation speed. A caselet rewards two minutes of unglamorous extraction that most candidates skip because it does not feel like solving — which is precisely why caselets are often worth attempting rather than avoiding, particularly if your arithmetic speed is unremarkable.
What if a caselet gives ratios but no absolute numbers?
Work in parts rather than inventing a total. If A to B is three to two, carry those as three parts and two parts through the whole table, and the questions will either ask for ratios — answerable directly — or supply an absolute somewhere that lets you convert all parts at once. Assuming a convenient total like one hundred is a common shortcut and it fails whenever a later question requires the real figure.
How do I practise caselets?
By separating the two halves. Take a caselet and time only the extraction, stopping when the grid is complete — that number should come down toward two minutes with practice. Then answer the questions untimed. Most candidates discover their problem is entirely in extraction rather than in calculation, which points at reading the prose more carefully rather than at arithmetic drills, and those are very different fortnights of work.
Like what you read? Share with a friend
Senior Content Writer
Co-Founder at PrepGrind, working on education content, product development, and student-focused learning resources. I also write articles and guides for MBA/MBS entrance exams, covering preparation strategies, exam updates, and useful resources for students.
