ICAC Investigator Written Test common mistakes
ICAC Investigator Written Test · Understand the traps and how to avoid them.
In the Chinese Language Proficiency section, candidates read a passage once and answer comprehension items from memory, then pick the option that 'sounds right' rather than locating the sentence in the text that supports it. They also skip over 轉折 and 因果 connectives (but, therefore, however, in fact) that flip the meaning of the sentence.
Under time pressure, candidates treat Chinese comprehension like casual reading. Because they are native or near-native speakers, they trust overall impression and never verify against the passage, so they fall for options that are true in general but not stated in the text, or that reverse the author's stance.
What to do instead
For every comprehension item, underline the exact clause in the passage that answers it and label the question type (main idea, detail, inference, tone). Circle all contrast and cause-effect connectives before answering. If no clause supports an option, eliminate it even if it feels plausible. Practise 20 passages with this annotation method and time yourself at roughly one minute per item.
In the English Language Proficiency section, candidates answer grammar and usage items by ear, choosing the option that 'sounds natural' in casual speech, and they misread double negatives, misplaced modifiers, or subject-verb agreement when a phrase separates the subject from the verb.
Everyday spoken English tolerates constructions that written Standard English marks wrong. Candidates who rely on fluency never consciously parse the sentence structure, so a long interrupting phrase between subject and verb defeats them.
What to do instead
Before choosing, strip the sentence to its core: find the finite verb, then find its true subject, crossing out prepositional and relative clauses. For modifier questions, ask what the modifier is meant to describe and check it sits next to that word. Do 15 items a day where you rewrite the sentence in your own words first, then match to the closest option.
In Numerical and Analytical Reasoning, candidates do all arithmetic in their head or on scrap paper without setting out units, and they answer the number the question seems to ask rather than the number actually requested (for example, giving a percentage when the question asks for the absolute amount, or forgetting to convert months to years).
Exam pressure pushes candidates to compute fast and grab the first number that appears. Without written units and a final check against the question stem, unit mismatches and 'answer the wrong quantity' errors go unnoticed.
What to do instead
Adopt a fixed written routine: write the formula, substitute with units attached, compute, then re-read the last line of the question and confirm your answer matches the requested quantity and unit. For ratio, rate and percentage problems, always write the base explicitly (percentage of what). Do timed sets of 10 and log every error by type.
In Logical Reasoning and Problem Solving, candidates read the stimulus once, jump to an answer choice that matches a keyword, and never diagram conditional statements ('if A then B') or check whether the argument confuses a necessary condition with a sufficient one.
Keyword matching feels efficient and works on easy items, so candidates never build the habit of formalising. Conditional logic reverses are the most common trap, and they are invisible unless the statement is written out as an arrow.
What to do instead
For any stimulus containing if, only, unless, all, or must, write the conditional in arrow form (A → B) and its contrapositive (not B → not A) before looking at options. For assumption questions, negate each candidate assumption and see whether the argument collapses. Practise 15 conditional-logic items daily until diagramming is automatic.
In Hong Kong Basic Law and National Security Law, candidates memorise article numbers and slogans but cannot apply a provision to a short scenario, and they mix up which body has which power (for example, who interprets the Basic Law versus who enforces the National Security Law, or the relationship between the Basic Law and ordinary legislation).
Study guides reward rote recall, so candidates learn labels without building a mental map of institutions and their functions. Scenario questions then expose the gap because recall alone cannot resolve 'which provision applies here'.
What to do instead
Build a one-page institutional map: for each key body, note its constitutional basis, its powers, and its limits. For each major provision, write a two-line scenario of your own and the correct outcome. Then practise applying provisions to short fact patterns rather than reciting article numbers. Cross-check every claim against the official text of the Basic Law and the National Security Law.
On integrity and ethics items, candidates answer as the 'ideal' officer would in a textbook, choosing the most severe or most virtuous-sounding option, instead of the option that follows proper procedure: report through the chain, declare a conflict of interest, decline the gift, or refer the matter to the appropriate authority.
Candidates assume ethics questions reward moral maximalism. In fact they test procedural judgment, and the most dramatic option (confront the person, investigate personally) is usually wrong because it bypasses the reporting and referral framework.
What to do instead
For each scenario, identify three things before answering: the potential conflict of interest, the correct reporting channel, and whether the candidate has authority to act. Prefer options that disclose, decline, record, and refer over options that confront or conceal. Study the ordinance and departmental conduct rules on gifts, hospitality, conflicts of interest and reporting, and map each to a scenario type.
On anti-corruption awareness items, candidates confuse the elements of the relevant offences, treating any dishonest act as corruption or any gift as a bribe, and they fail to notice when a scenario lacks a required element such as the corrupt agreement or the official capacity of the person involved.
Media coverage uses 'corruption' loosely, so candidates absorb an imprecise definition. Exam items are precise: they test whether each element is present, and an option that describes misconduct but not the specific offence is a deliberate distractor.
What to do instead
Write out the elements of each offence you are examined on as a checklist, then run every scenario through the checklist element by element. Note explicitly which element is missing in the wrong options. Distinguish corruption offences from related misconduct such as fraud, theft or breach of internal policy, and revise the definitions until you can state them without notes.
In Current Affairs and General Knowledge, candidates rely on vague impressions of recent news and answer questions about Hong Kong institutions, recent policy developments, or international organisations from memory that is months out of date, without checking dates or official names.
Candidates skim headlines rather than reading full reports, and they do not update their knowledge close to the exam date. Names of bodies, policy titles and effective dates are exactly what these items test, and approximate recall produces near-miss wrong answers.
What to do instead
Keep a dated log of major Hong Kong and national developments over the past twelve months, with the official name of each policy or body and its key date. Review reputable official sources weekly in the final two months. For each entry, write one line on what changed and who is responsible, and self-test the official names and dates.
Candidates allocate time evenly across sections or spend too long on one hard item, leaving later sections rushed, and they leave answers blank on multiple-choice items when there is no penalty for guessing.
Without a rehearsal under full timing, candidates do not know their natural pace per section. Pride keeps them stuck on a single item, and fear of being wrong stops them from making an educated guess at the end.
What to do instead
Sit at least two full-length timed mock exams and record time per section. Set a hard per-item cap (for example 60 to 75 seconds) and a rule to flag and move on, returning only after the section is complete. In the final minutes, fill every blank with your best elimination-based choice. Adjust section order only if the exam permits it.
Candidates revise by re-reading notes and highlighting, so they recognise material but cannot retrieve it under exam conditions, and they never review their wrong answers to find the underlying pattern of error.
Recognition feels like mastery. Re-reading raises familiarity without building recall, and candidates discard practice papers after scoring them instead of diagnosing why each wrong option attracted them.
What to do instead
Switch to active recall: close the notes and write out definitions, provisions and offence elements from memory, then check. Keep an error log with three columns (question type, why the wrong option tempted you, the rule that fixes it) and review it weekly. Re-attempt previously wrong items after a gap of several days to confirm the fix has stuck.