CAST Permanent common mistakes
CAST Permanent · Understand the traps and how to avoid them.
Spending the same time on every verbal item, reading the full passage before looking at the statements, so the last items in the section are left blank.
Candidates treat the passage as a comprehension text to be understood end-to-end, mirroring school reading habits, and underestimate how long four options take to check against a dense paragraph. The section is timed per-item in practice, so this pace guarantees unfinished items.
What to do instead
Read the question stem first, then scan the passage only for the sentence that decides the statement's truth. Practise with a hard cap of about 45–60 seconds per item; when the cap hits, mark a best guess and move on. Do two full timed sets and record how many items you left blank - the target is zero.
Answering verbal 'cannot say' items by importing real-world knowledge or personal opinion instead of judging only what the passage states.
The three-way true/false/cannot say format rewards strict textual literalism, but candidates instinctively fill gaps with plausible assumptions (e.g. assuming a named company is profitable because the passage says it grew). This is the single biggest source of lost marks on this item type.
What to do instead
For every option, point to the exact words in the passage that would prove or disprove it. If no sentence does either, the answer is 'cannot say' regardless of how likely it seems. Drill 20 'cannot say' items where the true answer is the least intuitive option and review why each distractor fails.
Treating 'true' and 'false' as 'supported' and 'contradicted' loosely, so statements that are merely consistent with the passage get marked true.
Candidates conflate 'the passage does not contradict this' with 'the passage states this'. The distinction matters because a statement can be compatible with the text yet not entailed by it - the correct answer is then 'cannot say', not 'true'.
What to do instead
Apply a strict entailment test: would a lawyer reading only this passage be forced to agree the statement is correct? If it is merely possible, choose 'cannot say'. Build a personal log of ten items you got wrong and label each as 'assumed', 'over-read' or 'misread the stem'.
Misreading quantifiers and comparatives in verbal stems - 'all', 'some', 'most', 'only', 'not' - and answering the statement as written rather than as qualified.
Under time pressure the eye skips small function words, and the brain reconstructs the sentence from the topic words. A statement about 'some managers' gets judged as if it said 'all managers', flipping the correct answer.
What to do instead
Underline or mentally box every quantifier and negation in the stem before scanning the passage. Make a one-page list of the quantifier pairs that flip answers (all/some, most/some, only/at least) and run a 15-item drill where the only difference between options is the quantifier.
Solving numerical items by full written calculation on paper, including long division and multi-step percentages, instead of estimating and eliminating options.
Candidates carry over school maths habits where exact working earns marks. Here only the final option matters, and written arithmetic consumes three to four times the time an estimate needs, leaving later items unreachable.
What to do instead
Before calculating, round the numbers to one or two significant figures and see whether one option is clearly closest. Only compute exactly when two options are within a few percent. Time yourself on 15 items with a 60-second limit and note how often estimation alone was sufficient.
Getting percentage change, percentage points and reverse percentages backwards - for example computing the change relative to the new value, or adding a mark-up and a discount as if they cancel.
These three operations look similar in wording and the base value is easy to misassign. A 20% rise followed by a 20% fall is not a return to the start, and candidates who do not fix the base explicitly get a plausible-looking wrong option.
What to do instead
Write the base value next to the calculation every time ('change ÷ original'). Memorise the multiplier method: a 20% rise is ×1.2, a 20% fall is ×0.8, and the combined effect is ×0.96. Drill ten chained-change items and ten 'percentage points versus percent' items.
Ignoring units, currency and scale in tables and charts - mixing thousands with millions, or reading a bar chart axis that starts above zero as if it started at zero.
Tables in this test often label columns in thousands or use different currencies per column, and the numbers themselves are unremarkable. Candidates grab the digits and skip the header, producing an answer off by a factor of a thousand or misjudging a visual difference.
What to do instead
Before touching any number, read the row and column headers, the axis labels and the units line aloud in your head. Convert everything to one unit at the start of the item. Practise five table items where the only trap is a units mismatch.
Trying to name or classify each abstract figure ('this is a rotation, this is a count of sides') instead of testing one rule against the whole sequence at a time.
Naming a rule feels productive but consumes time, and the first rule noticed is often a coincidence that holds for two figures and fails on the third. Candidates then force the wrong rule onto the answer options.
What to do instead
Work through a fixed checklist in order: number of elements, element type, position/movement, size, shading/fill, rotation, symmetry, intersections. Test each candidate rule against every figure before accepting it. Do ten sequences with the checklist written beside you until the order is automatic.
Choosing an answer option that fits the sequence but ignoring the direction of the pattern, so a mirrored or rotated-by-90-degrees option is picked instead of the correctly rotated one.
Once the rule is identified, candidates relax and pick the first option that 'looks right'. Distractors are deliberately built as the right shape with the wrong orientation or the right orientation with one element missing.
What to do instead
After identifying the rule, state the expected answer's properties explicitly (e.g. 'arrow points left, two dots shaded') before looking at the options, then eliminate any option that violates one property. Practise by covering the options, predicting the figure, then uncovering.
Answering situational judgement items with what an idealised manager would do in theory, or with the most decisive-sounding action, rather than with the behaviour the competency framework rewards in that specific scenario.
Candidates without public-sector experience default to generic leadership instincts - escalate immediately, or act unilaterally to show initiative - and both can score badly where the expected behaviour is to consult, gather information first or follow procedure. The 'most effective' and 'least effective' options are frequently both plausible.
What to do instead
Learn the handful of competencies the test targets (delivering at pace, making effective decisions, working together, managing a quality service, communicating and influencing) and map each option to one. Prefer options that gather relevant facts and involve the right people before acting; distrust options that are dramatic, blame others or bypass process. Read the scenario for the role you are given and answer as that role, not as a hypothetical ideal.