GLAT common mistakes
GLAT · Understand the traps and how to avoid them.
Spending a fixed amount of time per question regardless of difficulty, so two hard items early in a section consume the minutes needed for five easy items later.
Candidates treat the test as a school assignment where every question deserves equal effort, and they fear leaving a blank more than losing easy marks downstream.
What to do instead
Pre-set a per-question ceiling (for example 75 seconds on a 45-question, 60-minute section) and practise a physical 'park and move' rule: mark the item, log a one-word reason, move on, return only in the final pass.
Answering the stem that was expected rather than the one printed, e.g. choosing the value of x when the question asked for 3x + 2.
Under time pressure the brain completes familiar patterns from practice sets, and the final line of the question is often skimmed because it looks identical to earlier items.
What to do instead
Underline or mouth the last five words of every stem before reading the options, and write the requested quantity on scratch paper as a symbol (e.g. 'find: 3x+2') so the final check compares the answer to that symbol, not to the first value solved.
Ignoring qualifying words such as EXCEPT, NOT, LEAST, and ONLY, then selecting the first true-looking option.
These words are short and unstressed, and recognition of a correct-sounding statement triggers the response before the negation is processed.
What to do instead
Adopt a mechanical scan: circle every negation or superlative in the stem, restate the stem in the candidate's own words with the negation in front ('which is NOT true'), and reject any option chosen before that restatement happens.
Answering passage or data-set questions from memory of the text instead of returning to the specific lines or table cells cited.
Candidates believe re-reading wastes time, and a general impression of the passage feels sufficient, especially when an option matches the overall tone.
What to do instead
Force a line-reference habit: for each option, name the paragraph, sentence, or table row that supports or contradicts it, and mark the option unsupported if no reference can be produced within a few seconds.
Doing multi-step arithmetic mentally and writing down only the final number, so a single slip is invisible and unrecoverable.
Mental arithmetic feels faster and candidates equate visible scratch work with weakness or wasted paper.
What to do instead
Mandate one line of written setup per step, including units and the operation performed, and reserve mental math for single-digit operations only; the written chain makes errors locatable in seconds during review.
Treating all answer options as equally likely and reading A through E in order, stopping at the first plausible one.
Sequential reading plus confirmation bias makes the first comfortable option feel correct, and later options are then read as variations of it.
What to do instead
Predict the answer before reading options, then classify each option as supported, contradicted, or irrelevant; eliminate on evidence rather than ranking plausibility, and only stop when four options have an explicit rejection reason.
Guessing immediately on unfamiliar item types, leaving no trace, and never revisiting them.
Anxiety about an unrecognised format is read as 'I cannot do this', and an unmarked guess is indistinguishable from a considered answer later.
What to do instead
Apply a three-pass structure: complete known items, mark and attempt unfamiliar ones with a best-guess letter, then use remaining time to re-attempt marked items with fresh eyes; every blank gets a recorded guess rather than nothing.
Failing to check the answer against the question's units, scale, or order of magnitude before selecting it.
Candidates verify that a computation 'worked' but not that the result is physical, so a rate expressed per month is submitted where the stem asked per year.
What to do instead
Run a ten-second sanity gate on every numeric answer: units match, sign is plausible, magnitude is within the range implied by the given values, and the answer is expressed in the requested format.
Changing correct answers during a final review sweep on the basis of unease rather than new evidence.
Fatigue and anxiety peak at the end of a section, and a vague sense of discomfort is misread as a signal that the original answer was wrong.
What to do instead
Permit a change only when a specific documented reason exists (misread stem, arithmetic recheck, missed constraint); otherwise record the original answer and leave it, tracking changed answers in practice to show whether revisions gain or lose marks.
Mismanaging the answer sheet or on-screen interface, such as off-by-one bubbling or losing time to navigation and flag tools.
Candidates rehearse content but not the delivery mechanism, so clerical actions consume attention that should go to reasoning.
What to do instead
Rehearse under identical conditions: same sheet layout, same on-screen tool bar, same flagging workflow, bubbling in blocks of five with a position check; run at least two full-length timed simulations before the test date.