KRA CBT Aptitude Test cheat sheet

Percentage change = (new − old) ÷ old × 100; a rise then a fall of the same percent does not return to the starting value (100 → +20% = 120 → −20% = 96).

Successive percentage changes multiply: net factor = (1 ± a)(1 ± b); for +a% then −b%, net change = a − b − ab/100.

Work rates add: if A finishes in a days and B in b days, together they take ab/(a + b) days; time is the reciprocal of the combined rate.

Speed–distance–time: distance = speed × time; average speed for equal distances = 2xy/(x + y), not the arithmetic mean of the speeds.

Simple interest = P × r × t ÷ 100; compound amount = P(1 + r/100)^t, so CI − SI = P[(1 + r/100)^t − 1 − rt/100].

Profit% and loss% are computed on cost price; selling price = CP × (1 ± p/100); discount% is computed on marked price.

Ratio and mixture: alligation rule gives the ratio of quantities as (higher − mean) : (mean − lower) when mixing two priced components.

Divisibility shortcuts: by 3 or 9 sum the digits; by 4 check the last two digits; by 8 the last three; by 11 alternate digit sums differ by a multiple of 11.

Estimate before computing: round operands to 1–2 significant figures to locate the answer range, then check the option whose magnitude matches.

Main-idea questions ask what the passage is chiefly about; a correct main idea must cover the whole passage, not one supporting detail.

Inference answers must be necessarily true from the stated text; an option that is merely plausible or likely is not an inference.

Scope words decide correctness: 'all', 'only', 'never', 'must' make a statement stronger and usually wrong unless the passage states it explicitly.

Tone and attitude are signalled by evaluative adjectives and adverbs; neutral reporting versus advocacy is distinguished by such words, not by topic.

Assumption questions: the correct answer is an unstated premise needed for the argument to hold; negating it should break the conclusion.

Strengthen/weaken questions target the link between evidence and conclusion; irrelevant background facts change nothing and are distractors.

Para-jumbles: find the opening sentence (often definitional, no pronoun referring back), then chain by pronouns, connectives, and repeated key nouns.

Word-substitution and analogy items test the relation, not the meaning alone: identify the relation type (cause–effect, part–whole, degree) before choosing.

Number and letter series: check first differences, then second differences, then ratios; alternate-position series often hide two interleaved patterns.

Alphabet positions: A=1 … Z=26; reverse position = 27 − position, so the opposite letter of a letter at position n is 27 − n.

Syllogism rules: two 'all' premises give an 'all' conclusion; a negative premise forces a negative conclusion; two negatives give no valid conclusion.

Contrapositive is valid: 'if P then Q' entails 'if not Q then not P'; the converse 'if Q then P' is not valid.

Matrix and figure-series tasks: track one attribute at a time - number of elements, rotation (clockwise/counter-clockwise), shading, and position.

Seating and ordering puzzles: draw a line or circle, place the most constrained item first, and use 'immediately left/right' to fix adjacency.

Blood relations and direction sense: convert 'left/right' relative to facing direction; facing north, left = west, facing south, left = east.

Odd-one-out items: the shared property must hold for exactly the majority set; verify a candidate rule against every member before selecting.

String comparison: scan character by character in fixed groups (e.g., chunks of three) rather than reading whole tokens as words.

Common confusables: 0/O, 1/l/I, 5/S, 2/Z, 8/B, rn/m, vv/w, cl/d - check these pairs first in any mismatch task.

Transposition errors (swapped adjacent digits or letters) preserve all characters, so compare position by position, not just the set of characters.

In verification tasks, compare one attribute per pass: order of items, then spelling, then punctuation, then capitalisation.

Dates and IDs: validate format and check digit positions independently; a single wrong digit changes the whole entry's validity.

Counting symbols or occurrences: mark each counted instance as you go and re-count only the marked set to avoid double counting.

Speed–accuracy trade-off: an unanswered item scores zero, so flag uncertain items and return after completing all certain ones.

Read the axes and units first: values may be in thousands, lakhs, or percentages of an unstated base, which changes every derived figure.

Percentage of total = part ÷ total × 100; percentage change = (later − earlier) ÷ earlier × 100; keep the base year consistent.

Ratio comparisons across tables: convert to a common denominator or compute decimals to two places before ranking.

Averages in tables: weighted mean = Σ(value × frequency) ÷ Σ frequency; an unweighted mean of group averages is wrong unless group sizes are equal.

Pie charts show shares of a whole, so a slice's absolute value requires the total; two pies of different totals are not comparable by angle alone.

Growth rate over multiple years: CAGR = (end ÷ start)^(1/n) − 1, where n is the number of intervals, not the number of data points.

Approximate aggressively in multi-step questions: round to one significant figure, compute, and select the closest option.

Correlation is not causation: a chart showing two series moving together supports association only, not a causal claim.

Rank responses by priority: safety and legal compliance first, then accuracy and client impact, then team relations, then personal convenience.

Escalation rule: report immediately when there is risk to safety, law, or data integrity; handle routine matters at the lowest competent level.

Own errors promptly and correct them; concealing or self-correcting silently ranks below disclosure plus remediation.

When a colleague errs, address it privately and factually first; public blame and going over their head without cause rank low.

Conflicting deadlines: communicate early, quantify impact, and agree priorities with the decision-maker rather than silently dropping work.

Confidentiality: share information only with those who need it for the task; seeking consent or following policy outranks informal disclosure.

Constructive disagreement: raise concerns with evidence through proper channels, then support the final decision once made.

Best-answer questions favour professional, proportionate, documented action over extremes of doing nothing or overreacting.