GSET common mistakes
GSET · Understand the traps and how to avoid them.
Treating the Verbal Reasoning section as a vocabulary quiz and guessing on words in isolation rather than reading how each word functions in the sentence's logic.
GSET verbal items are sentence-completion and analogy-style tasks where the decisive clue is the contrast, cause, or restatement built into the sentence; candidates who fixate on a single hard word miss the structural signal and pick a word that is plausible but wrong in direction.
What to do instead
Before looking at the options, rewrite the sentence with a blank marker and label the relationship (contrast, support, definition). Predict a word of your own, then eliminate options that fail the predicted meaning or the grammatical form (noun vs. adjective, positive vs. negative charge).
Answering quantitative questions with mental arithmetic and no scratch work, then failing to re-check units, percentages, and whether the question asks for a rate or a total.
Under time pressure, candidates shortcut the setup and carry an early arithmetic slip through to the answer; the exam deliberately includes distractors matching common unit and percent errors, so the wrong answer feels confirmed.
What to do instead
Write the setup as an equation with labeled units before computing. Do a final sanity check: is the answer's magnitude plausible, and does it answer the exact quantity requested? Estimate first so a calculator or arithmetic error is visible.
Writing the Analytical Writing essay as a personal opinion piece or a list of examples with no explicit evaluation of the argument's reasoning.
Candidates transfer habits from school essays where stance and volume earn credit; GSET analytical writing rewards identifying assumptions, evidence quality, and logical gaps, so a fluent but unevaluative essay scores low.
What to do instead
Use a fixed structure: state the argument's conclusion, identify two or three unstated assumptions, test each against alternative explanations or missing evidence, then conclude on strength. Budget time for a short planning outline and a proofreading pass.
Reading the passage once for general gist and then answering comprehension questions from memory without returning to the text to verify the specific line.
Passages are dense and questions hinge on qualifiers such as 'primarily', 'except', and 'the author would most likely agree'; memory compresses these, so candidates pick an answer that is true in the world but not supported by the passage.
What to do instead
For each question, locate and underline the sentence or paragraph that decides it, and reject any option requiring information not in the text. Treat extreme wording ('always', 'never', 'proves') as a warning sign unless the passage itself is that absolute.
Reading a data table or chart by comparing raw counts across groups of different sizes instead of converting to rates or percentages.
Charts display absolute values prominently and base sizes are easy to overlook; the exam includes items where the larger count comes from the larger group, so raw-count comparisons invert the correct conclusion.
What to do instead
Before comparing, identify each group's base and compute the relevant proportion. Check whether the question asks for absolute change or percentage change, and re-read axis labels, scales, and units before computing.
Jumping to a conclusion in logical reasoning because it matches a real-world belief, without checking whether it follows from the premises as stated.
Everyday plausibility feels like validity, and candidates substitute their own knowledge for the argument's premises; the exam plants conclusions that are true in general but not entailed by the given evidence.
What to do instead
Restate the premises in symbols or short phrases and ask what must be true versus what could be true. For assumption questions, negate the candidate assumption: if the argument collapses, it is required; if it survives, the option is wrong.
Spending disproportionate time on one hard question early in a section, leaving too little time for later questions of equal value.
Sunk-cost thinking and the desire to finish every item push candidates to stay on a single problem; because all questions carry similar weight, one stubborn item can cost several easy marks.
What to do instead
Set a per-question time cap and move on when it is exceeded, marking the item for return. Practise full-length timed sections so the pacing habit is automatic, and answer every item since there is no penalty for a reasoned guess.
Skimming the question stem for keywords and answering the remembered version of the question rather than the one actually printed.
Stems differ by one word ('except', 'least', 'not', 'best supports'), and pattern-matching to a familiar question type produces answers to a question that was not asked.
What to do instead
Read the full stem and circle the operative word before touching the options. For 'except' and 'least' items, restate the task in your own words ('find the one that does not fit') and verify each option against that restatement.
Leaving the Analytical Writing response unproofread, with sentence fragments, agreement errors, and misspellings that obscure otherwise sound reasoning.
Time runs out during drafting and candidates submit the first version; scorers weigh control of standard written English, so mechanical errors lower the score even when the analysis is strong.
What to do instead
Reserve the final minutes for a targeted pass: check subject-verb agreement, complete sentences, and the spellings of repeated technical terms. Write shorter, controlled sentences when time is short rather than long ones you cannot revise.
Interpreting graphs with dual axes or indexed values as if both series shared one scale and unit.
Dual-axis charts and index numbers (base year = 100) are visually persuasive but numerically incomparable at face value; candidates read a crossing point as a real crossover or compare an index to a raw value.
What to do instead
Check the axis title, unit, and scale on each side before reading any trend, and convert index values back to levels if the question asks about magnitude. Note the base year explicitly and recompute any percentage change from the original figures.