MSI Concours d'Admission common mistakes
MSI Concours d'Admission · Understand the traps and how to avoid them.
In mathematics, the candidate starts computing immediately on multi-step problems (systems, optimisation, sequences) without first writing down the unknowns, the constraints, and what quantity is actually being asked for, then answers an intermediate value or a related but different quantity.
Under time pressure the reflex is to 'get going' on the numbers; MSI-style problems often embed a translation step (word problem to equations) and the final question is stated last, so candidates latch onto the first quantity they can compute and treat it as the goal.
What to do instead
Impose a fixed 20-second planning pause: write 'Given / Find / Constraints' in three short lines before any calculation, and re-read the final sentence of the question and underline the requested quantity. At the end, check that the boxed answer matches the 'Find' line, including units or nature (a count, a rate, a probability).
In physics, the candidate substitutes numbers into a formula before converting all quantities to SI units and before checking dimensional consistency, producing answers off by factors of 10, 1000, or with mixed units (g with kg, cm with m, minutes with seconds).
Formulas are memorised as symbol strings rather than as relations between quantities with dimensions; school habits of working in 'convenient' units persist, and calculators hide the error because the number looks plausible.
What to do instead
Adopt a two-pass routine: first write the symbolic expression, verify dimensions by hand (e.g. energy must reduce to kg·m²/s²), then convert every given value to SI in a single conversion line before substituting. Sanity-check the order of magnitude against a known reference (a 1 kg object falling 1 m gains about 10 J).
In chemistry, the candidate balances equations by trial and error on the full equation instead of identifying the limiting reagent and working in moles, and confuses mass, amount of substance, and concentration when moving between them.
Stoichiometry requires a chain (mass → moles → mole ratio → moles → mass or concentration) and candidates skip links, especially dividing by molar mass and by stoichiometric coefficients; the limiting-reagent step is silently omitted.
What to do instead
Use an explicit four-column table (species, n_initial, n_change via coefficient ratio, n_final) and always compute moles for every reactant before applying the ratio. State the limiting reagent in writing, then convert the product moles to the requested unit. Practise the mass–mole–concentration triangle until it is automatic.
In biology, the candidate answers with everyday or teleological language ('the cell wants to', 'to survive') and gives a description of the phenomenon instead of the mechanism at the requested level (molecular, cellular, organismal), so the answer is vague and loses precision marks.
Biology is learned narratively; candidates memorise outcomes rather than causal chains, and the exam asks for the mechanism (e.g. how a gradient is maintained, how an enzyme is regulated) rather than the result.
What to do instead
For each major process, rehearse a causal chain in three to five links with named actors (membrane protein, ion, enzyme, gene product). Ban purposive verbs from answers and replace them with 'because X causes Y via Z'. When a question specifies a level of organisation, answer only at that level and name the structures involved.
In French language and literature, the candidate writes a dissertation or commentary without an explicit problem statement (problématique) and a stated plan, producing a catalogue of ideas or a paraphrase of the text rather than an argument answering a question.
Candidates treat the exercise as 'say everything I know about the topic' and are not trained to convert the subject into a tension or question; paraphrase feels safe because it is always true to the text.
What to do instead
Before writing, draft one sentence: 'The question is whether X, given that Y', then two or three thesis-style headings that each answer part of it. In commentary, every quoted feature must be followed by an interpretation ('this metaphor suggests...'), never left as description. Reserve the introduction's last two lines for the announced plan.
In English, the candidate translates idiomatic French structures word for word (false friends, 'actually' for 'currently', 'I am agree', missing auxiliaries, wrong tense after 'since') and produces grammatically French sentences with English vocabulary.
Interference from the first language is strongest exactly where the two languages look similar; candidates read for meaning and never audit their own output for structure, and false friends are invisible without explicit contrastive practice.
What to do instead
Keep a personal list of high-frequency false friends and contrastive structures (since/for, present perfect vs passé composé, do-support in questions and negatives) and drill them in sentence transformation. In written production, leave two minutes to proofread only for verb forms and agreement, reading each sentence's subject and verb aloud mentally.
In history and geography, the candidate writes a narrative of events with dates but no explanation of causes, scales, or actors, and in geography ignores the map or document provided, answering from general knowledge instead of from the evidence.
History is revised as a story and geography as a list of facts; document analysis is a distinct skill that requires reading the legend, scale, source, and date before interpreting, and candidates skip that step under time pressure.
What to do instead
Structure every answer around a driving question and a hierarchy of causes (long-term structures, conjunctural factors, triggering events) with named actors. For documents, write a short identification block first (nature, author, date, context, scale) and cite at least two specific elements from it in each paragraph.
In philosophy and general culture, the candidate produces a themed essay that juxtaposes authors' positions without defining the key notions, and cites works vaguely or invents titles and dates to fill space.
General culture is revised as a stock of references; candidates believe quantity of citations earns credit, whereas the exam rewards conceptual definition, argumentative rigour, and precise, verifiable references.
What to do instead
Begin by defining the central notion and distinguishing two of its senses, then build a dialectic (thesis, objection, resolution) in which each reference does argumentative work. Keep a reference sheet limited to works you can state accurately (author, title, one precise claim) and prefer fewer, well-explained references to many vague ones.
In logical reasoning, the candidate reads the puzzle once, starts drawing or enumerating before extracting the constraints, and then misreads quantifiers ('some', 'all', 'necessary', 'unless') or reverses conditional statements.
Speed pressure pushes candidates to solve while reading; natural language quantifiers are ambiguous in everyday use, and the converse of a conditional feels intuitively equivalent even though it is not.
What to do instead
Use a fixed protocol: list the constraints numbered, translate each into formal notation (A → B, not-B → not-A), and only then solve. Test conclusions by constructing a counterexample rather than by intuition. For syllogisms, draw minimal diagrams for 'all' and 'some' and check whether the conclusion holds in every compatible arrangement.
Across all domains, the candidate allocates time uniformly across the paper, runs out of time on the sections carrying the most marks, leaves multiple-choice items blank when there is no penalty, and does not return to flagged questions.
Candidates rehearse content but never rehearse the exam as a timed exercise, so they lack a strategy for triage; perfectionism on a hard early question consumes the budget, and fear of being wrong prevents guessing on items with no negative marking.
What to do instead
Before the exam, determine the mark value and time budget per section, and rehearse full papers under timed conditions at least twice. During the exam, apply a one-pass rule: answer everything you can within a per-question time cap, flag the rest, then return. Always fill every item when there is no penalty, and reserve the final five minutes for transferring answers and checking that no question is unanswered.