Caselet DI
A caselet is a data interpretation set in which the data is given as a short paragraph of prose rather than as a table or a chart, so you have to extract the numbers and their labels yourself before you can answer anything. Solving one is therefore two jobs: build the table the paragraph withheld, then read every answer off it.
Caselet DI is the hardest form of data interpretation in a bank paper, and the difficulty is entirely in the extraction. Candidates who answer question by question re-read the paragraph five times; candidates who spend a minute drawing the table first answer all five questions from it in under two.
Try it against the clock. Data Interpretation (Caselet) runs at 40s easy, 50s medium and 65s hard. The clock is stamped and judged on our server, so the limit you see is the deadline that is actually enforced — and it starts when you tap Start, not while the question is loading.
The word is the diminutive of case — a small case study. On an exam paper it is the name of a question type: a scenario two to five sentences long that states, in words, the figures a table would otherwise have laid out for you. “Caselet”, “caselets” and “caselet DI” all mean the same thing, and it is always part of the data interpretation section rather than a topic of its own.
Pronounced KAYSS-lit (/ˈkeɪslɪt/) — two syllables, stress on the first, rhyming with bracelet. It is often typed as two words, case let, and searched that way; the two-word form is the same question type, not a different one. There is no standard Hindi or Marathi translation — question papers and coaching notes use the English word, transliterated केसलेट, and gloss it descriptively: गद्य में दिए गए आँकड़ों का सेट in Hindi, गद्य उताछ्यात दिलेली आकडेवारी in Marathi — both meaning “data given in a prose passage”.
So a caselet asks nothing a table DI set does not ask. Every question on one is still a total, a difference, a ratio, an average or a percentage change. What it adds is a comprehension step: the figures arrive as sentences, and often as relationships rather than values — twice as many, the remainder, in the ratio 3 : 2 — so the set is unanswerable until you have turned them back into rows and columns.
Caselets appear in IBPS PO and Clerk quant papers and in SBI at both stages, and they dominate SBI mains Data Analysis & Interpretation, where a single caselet can carry a whole five-question set. They also turn up in SSC CGL Tier 1, though less often and with lighter arithmetic.
Resist starting on question one. A caselet usually defines its categories in the first two sentences and only then gives the numbers, so a value read early often means something different from what it appeared to mean.
The row labels are the groups the paragraph talks about and the column labels are what it measures about them. Write that grid down and fill every number the paragraph states. This one minute is what the entire topic rewards, and the table serves all five questions rather than one.
Caselets deliberately withhold values that are recoverable — a row total minus its known entries, or a stated ratio applied to a known figure. Fill these in before starting, because a question will need one and finding it mid-question costs more.
Once tabulated, the questions are the usual totals, differences, ratios and percentage changes. Approximate when the options are far apart, and convert percentage questions into fractions.
| The paragraph says | Write |
|---|---|
| Twice as many X as Y | X = 2Y — a relation, not yet a number |
| 30% of the total are X | X = 0.3 × total, once the total is known |
| The remainder are Z | Z = total − everything named so far |
| The ratio of X to Y is 3:2 | Split the known total in 3:2 |
| X exceeds Y by 40 | X = Y + 40 |
| 40% more than Y | X = 1.4Y — 'more than' is on Y, not on the total |
| A total is stated | Anchor it first — most other cells derive from it |
Three shapes account for nearly every caselet in a banking or SSC paper. They differ only in how much of the table the paragraph withholds — which is another way of saying they differ in how long the extraction takes.
One scenario, one group of figures, usually a total plus a couple of percentages or a ratio. This is the commonest prelims form. The table is two or three rows and you can often hold it in your head — write it anyway, because the fourth question is invariably the one that needs the cell you did not bother to derive.
Two or three scenarios sharing a structure — the same categories across different years, cities or branches — each with its own paragraph. Here the table is genuinely two-dimensional, and comparison questions across the paragraphs are the entire point of the set. Never solve these paragraph by paragraph: build one grid with a column per paragraph.
The paragraph omits values that are recoverable from what it does state — a total minus its named parts, or a ratio applied to a known figure. The set is built so that filling those gaps first makes every question quick, and skipping the step makes each one slow. Treat a blank cell as a task, not as information you were not given.
Generated by the same engine that mints the ComputePrep daily. Each answer is computed from the numbers printed in the question, and the walkthrough below each one is the engine's own working — not a solution written afterwards. Reload this page's live drill and you get different numbers.
A company employs 2400 employees, of whom 20% are men.
Of the women, 25% work part-time.
How many women work part-time?
A train carried 1600 passengers, of whom 60% are adults.
Of the children, 60% travelled without luggage.
How many children travelled without luggage?
A bank branch serves 3000 account holders, of whom 50% are salaried customers.
Of the self-employed customers, 40% hold a fixed deposit.
How many self-employed customers hold a fixed deposit?
The wrong options are not random numbers. Each one is the result of a specific careless error for this topic, so picking one tells you which habit is costing you marks.
It means re-reading the paragraph once per question and re-deriving the same values each time. The table is built once and used five times.
The single commonest caselet trap. 20% of the remaining passengers is 20% of what is left after the first group is removed, not 20% of the headcount — and the number you get by taking it on the total will be sitting there as an option.
A 3:2 ratio is not three and two. It needs a total before it becomes numbers, and the total is usually stated elsewhere in the paragraph.
Fill everything recoverable while tabulating. Discovering mid-question that a value was available all along is the commonest reason a caselet set overruns.
A caselet will state one figure in lakhs and the next in thousands, or a rate per month against a total for the year. Normalise every number to one unit as it goes into the table, not when a question finally forces the issue.
The specific wrong numbers this topic produces, and what each one tells you about the step you took. If you have just got a question wrong and want to know which habit did it, start here.
Because you are reading the passage once per question. Build the table first — every row, column and total the prose names, including the ones no question has asked for yet — and then answer all five from the table. Re-reading the paragraph five times is what costs you the four minutes.
No, and this is the trap the whole format is built around. If 40% of 500 went to A, the remainder is 300, so 25% of the remainder is 75, not 125. Write the remainder down as its own number the moment the sentence introduces it, before you read on.
Not until you have found the multiplier. A 3 : 2 split of 200 is 120 and 80, and the parts 3 and 2 are only shares. Treating the ratio numbers as counts gives an answer right in shape and wrong in size, and the bare ratio parts are usually printed as options.
The drill above is the live, server-timed version — a fresh caselet every time, judged on our clock. Two things carry it off the screen:
A data interpretation set where the data is given as a paragraph of prose rather than a table or chart. The questions are the same as any DI set; the extra work is extracting the numbers and their labels into a table yourself before answering.
Caselet means a small case study. In an exam it is the name for a short written scenario that states figures in sentences instead of laying them out in a table, and then asks data interpretation questions about them. Caselet, caselets and caselet DI all refer to the same question type.
Read the whole paragraph once without answering anything, name the entities it talks about, and draw a table with those as rows and what it measures as columns. Fill in every stated number, derive the recoverable cells from totals and ratios, and only then answer the questions — reading each one off the table.
Yes. Caselets are a standard part of the data interpretation section in IBPS PO and Clerk and in SBI PO and Clerk, and they are especially heavy in SBI mains Data Analysis and Interpretation, where one caselet can carry a five-question set. SSC CGL asks them too, less often and with lighter arithmetic.
Yes, but only at the extraction step. Once tabulated, a caselet is an ordinary DI set and the same approximation and fraction habits apply. If you find them hard, the thing to practise is the tabulation, not the arithmetic.
ComputePrep is free, needs no account, and is not affiliated with IBPS, SBI, SSC, RRB or any exam authority.
