Bocconi Test common mistakes
Bocconi Test · Understand the traps and how to avoid them.
Choosing an option because it is true in the real world or consistent with the candidate's own views, without locating the sentence in the passage that licenses it.
Bocconi passages often cover economics, science or social issues the candidate already has opinions about, so the brain auto-completes the passage with outside knowledge. The option 'feels' supported even though the text never says it.
What to do instead
For every option the candidate is tempted by, force a line reference: write the line number or quote the phrase that proves it. If no line can be produced within about 10 seconds, mark the option wrong and move on. Treat 'true but not stated' as strictly wrong.
Accepting a 'mostly right' option whose scope word contradicts the passage - e.g. picking an answer containing all, only, never, proves, or the main cause when the passage said some, often, suggests, or one factor.
Under time pressure candidates match on overall topic and gist rather than on strength. A 90%-matching option reads as correct and the single overstated word is skipped.
What to do instead
Before comparing options, underline the hedge/strength words in the passage (can, may, some, likely, one reason). Then circle the quantifier and modal in each option. Reject any option whose strength is stronger than the passage's.
Reading the stimulus first, forming an opinion about it, and only then reading the question stem - so the candidate strengthens when asked to weaken, or evaluates the conclusion when asked for an assumption.
The stimulus is longer and feels like the 'real' content, so it gets read first. Once a reaction has formed, the stem is skimmed and the task verb is not registered.
What to do instead
Read the stem first, every time, and copy the exact task onto scratch paper in three words ('WEAKEN', 'FIND ASSUMPTION', 'EXPLAIN PARADOX'). Re-read that note immediately before selecting the answer.
Selecting an option that contradicts or attacks a stated premise, instead of the link between premises and conclusion.
Premises are factual-sounding, and a plausible-sounding challenge to a fact feels like a strong objection. Candidates do not separate 'what the author asserts as given' from 'what the author concludes'.
What to do instead
Bracket and label the conclusion (C) and the premises (P1, P2) before looking at options. For each option ask: does it attack a P (automatically wrong in weaken/assume questions) or does it attack the P-to-C move (eligible)?
Mixing units and bases inside a calculation - treating a figure in thousands as if it were in units, dividing a percentage by a total expressed in millions, or taking a percentage of the wrong base (e.g. new value instead of original).
Tables and charts deliberately carry mixed scales (%, thousands, per 100,000, index = 100). The arithmetic is easy, so the candidate computes fast and the error is invisible in the final number.
What to do instead
Before computing, rewrite every extracted figure with its unit and convert to one common base. Write the base of every percentage explicitly as a fraction ('x / ORIGINAL'). Finish by checking the order of magnitude against the raw numbers in the table.
Confusing percentage change, percentage points and share of total, or averaging percentages that have different denominators.
All three are expressed with the % sign and the question wording is short. The familiar formula (new − old)/old is applied reflexively even when the question asks 'what proportion of total' or 'by how many points'.
What to do instead
Classify the question into one of three boxes before touching the calculator: CHANGE ((new−old)/old), POINTS (new − old, no division), SHARE (part/total). Also refuse to average percentages across groups of different sizes - go back to raw counts and re-divide.
Solving correctly for the intermediate variable and then selecting that value, when the question asks for a derived quantity (2x + 1, the sum, the perimeter, the value after a further change).
The equation naturally ends at x, so x feels like the answer. The last line of the question is read first and least carefully.
What to do instead
Before solving, write down what the question literally asks for and circle it. After obtaining x, physically re-read the last line of the question and perform the final substitution; check that the chosen answer is not simply x.
Applying a memorised formula or geometric property without verifying its preconditions - averaging rates instead of weighting, treating a triangle as right-angled because it looks so, assuming a sequence is arithmetic, distributing over a non-linear operation.
Formula recall is fast and rewarding; condition-checking is slow and feels unnecessary, especially with a drawn figure that 'obviously' looks right.
What to do instead
State the preconditions in one line before substituting (equal weights? right angle marked? constant difference verified for at least two gaps?). If a condition is unmarked in the figure, it is not given - do not use it.
Finding one rule that explains most of the series or matrix and answering immediately, without checking a second or third attribute (rotation plus shading, element count plus position, both rows and columns).
The first visible regularity produces an 'aha' that satisfies the pattern-recognition urge. Figures in these items almost always change on two or more dimensions simultaneously, so one-rule answers are usually among the distractors.
What to do instead
List the attributes (number, shape, orientation, fill, position) and test each one separately down the series or across rows, then across columns. Only answer when the candidate has two independent rules that are consistent everywhere and eliminate all but one option.
Reading a 3×3 matrix only row-by-row (or only as a left-to-right sequence) and therefore missing a column-wise or whole-figure rule.
Reading order is culturally automatic, and the first row usually contains enough variation to suggest a rule. Column rules are equally common and are simply never inspected.
What to do instead
Use a fixed scan order and say it out loud: rows first, then columns, then whole-figure symmetry/count. If a row-based rule leaves two surviving options, abandon it and re-scan by column before guessing.