NGAT common mistakes
NGAT · Understand the traps and how to avoid them.
Spending the same amount of time on every verbal item, so a single dense passage swallows the minutes that several short synonym or odd-one-out items would have taken.
Candidates treat the verbal section as a reading test and try to fully comprehend each passage before answering, rather than recognising that verbal-reasoning items are scored per item and that some items are answerable from a single sentence or from word-level knowledge.
What to do instead
Before reading, scan the question stem to decide whether it needs whole-passage comprehension or local word/sentence work. Set a hard per-item ceiling (for example 45-60 seconds on short items) and if the answer is not clear, mark a provisional choice and move on. Practise with a timer visible so the ceiling becomes automatic.
Answering vocabulary and odd-one-out items from the first association that comes to mind, without checking whether the chosen word's meaning actually fits the sentence's grammar and register.
Recognition memory fires fast on familiar words, and candidates confuse familiarity with correctness. Distractor options are often near-synonyms that differ in connotation, transitivity, or collocation, so the first plausible word is frequently wrong.
What to do instead
After selecting a word, silently insert it into the sentence and read the whole sentence for sense and grammar. For odd-one-out, write a one-line definition of the majority category and reject any option that does not fit that definition. Where two options seem close, look for the discriminating feature (register, countability, direction of meaning).
Setting up arithmetic step by step in written working and losing track of the question, so the final calculation answers an intermediate quantity rather than what was asked.
Under time pressure candidates default to school-style computation, which consumes time and creates many chances to drift. Multi-step percentage, ratio, and rate problems are designed so that intermediate values appear among the answer options.
What to do instead
Before computing, underline what is being asked and write its units. Estimate the answer to one significant figure first, then compute only as precisely as needed. If your computed value matches an option that is an intermediate quantity (for example a discount amount rather than a final price), re-read the stem before selecting.
Guessing on data-interpretation and table items without extracting the relevant row, column, or unit, often misreading a percentage as a percentage-point change or a thousands column as units.
Tables and charts in quantitative sections deliberately mix units (per cent, percentage points, thousands, millions) and include totals rows that candidates accidentally double-count or omit. Scanning quickly under time pressure makes unit errors invisible.
What to do instead
Adopt a fixed reading routine: read the title, axis labels and units first; circle the specific cell or series the question names; note whether the question asks for a value or a change. Sanity-check magnitude against the axis or total before choosing an option.
Treating each abstract figure as a whole picture and hunting for a 'looks similar' match instead of decomposing it into elements (number of sides, rotation, shading, line intersections, position of a dot).
Human vision matches overall gestalt quickly, and candidates trust that impression. Abstract-reasoning series and matrices are constructed by applying one or two rules to individual elements, so gestalt matching picks up surface similarity and misses the operative rule.
What to do instead
For every figure, list the attributes in a fixed order: element count, shape type, size, orientation, shading/fill, position, and line relationships. Compare the list between consecutive figures and state the rule in words ('the arrow rotates 90 degrees clockwise each step') before looking at the options. Eliminate options that violate the stated rule.
Assuming a single rule governs a matrix or series when the item actually applies two independent rules in parallel (for example shape changing down the columns while shading changes across the rows).
Once a candidate finds one working rule, they stop searching and force-fit an option that satisfies it but breaks the second pattern. Matrices with two simultaneous rules are a standard difficulty step, and the correct option is the only one satisfying both.
What to do instead
After finding a rule, test it against every row and column, not just the one that revealed it. If more than one option survives, deliberately look for a second attribute that varies systematically and check the surviving options against it. Practise items where exactly two attributes change independently.
Rotating or reflecting a shape mentally in one step and choosing the option that 'looks right', without tracking a distinctive feature (a notch, a dot, an asymmetric arm) to confirm the transformation.
Mental rotation is error-prone for complex or asymmetric figures, and mirror images are especially confusable with rotations. Candidates also forget that a paper fold or a net produces a specific chirality that a reflection cannot reproduce.
What to do instead
Pick one asymmetric landmark on the original and track only that landmark through the transformation, then verify a second landmark. For nets and folded paper, mark the orientation of a printed symbol and check which face it neighbours and which way it points. Where allowed, sketch the intermediate state rather than holding it in memory.
Losing track of the fold sequence or the direction of a rotation in multi-step spatial items, so an early error propagates and every later step is wrong.
Working memory is limited, and multi-step items require holding the state after each operation. Candidates compress steps to save time, skip writing the intermediate state, and then cannot tell where the error entered.
What to do instead
Write or sketch each intermediate state explicitly, one line per operation ('fold left half over right', 'rotate 90 clockwise'). Number the steps and re-check each against the prompt before proceeding. Slower, auditable steps are faster overall than restarting a corrupted mental model.
Accepting a conclusion because it is plausible or matches personal belief, rather than checking whether it follows necessarily from the stated premises, including premises that are counter-intuitive or hypothetical.
Everyday reasoning rewards plausibility and background knowledge, but logical-reasoning items often use artificial or false premises and require the candidate to treat them as given. Familiarity with the topic overrides formal validity.
What to do instead
Restate the argument in symbols or in stripped-down form ('All A are B; some B are C; therefore...'). Ask only whether the conclusion must be true if the premises are true, and explicitly bracket real-world knowledge. For assumption and strengthen/weaken items, identify the gap between evidence and conclusion before reading the options.
Rushing the final few items in each section and abandoning the elimination work that earlier items received, so accuracy collapses exactly when easy marks are still available.
Fatigue and clock pressure cause a shift from deliberate reasoning to fast pattern-matching. Candidates also fail to leave time to return to flagged items, so they never revisit provisional answers with fresh eyes.
What to do instead
Plan the section in blocks with a checkpoint (for example, at the two-thirds mark you should be through roughly two-thirds of the items). Reserve the last few minutes for flagged items only, applying the same elimination routine. In practice, run full-length timed sections so pacing under fatigue is trained, not improvised.