GSET cheat sheet

Analogy stem relationship types: part-to-whole, function, cause-effect, degree, category-member, antonym; state the relationship in a sentence before scanning options.

Bridge rule: a valid analogy bridge must fit both pairs in the same direction and the same grammatical role; a bridge that fits only the stem pair is invalid.

Antonym items: negate the stem word, then test each option for closest opposite meaning; eliminate options that are merely unrelated rather than opposite.

Sentence equivalence: both correct choices must produce sentences alike in meaning, so pick the pair of options that are near-synonyms and fit the blank's grammatical slot.

Two-blank and three-blank items are scored only when all blanks are correct, so verify each blank independently against its own clause's logic.

Transition signals: 'although, despite, whereas' invert meaning; 'thus, hence, consequently' preserve it; 'indeed, moreover' intensify or extend it.

Polarity check: mark each clause as positive or negative; blanks usually take the same polarity unless a contrast transition intervenes.

Word-in-context items: substitute each option into the sentence and choose the one matching the author's usage, not the most common dictionary sense.

Percent change = (new − old)/old × 100; successive changes of a% then b% give a net change of a + b + ab/100 percent.

Compound interest: A = P(1 + r/n)^(nt); simple interest: A = P(1 + rt). Doubling time is approximately 70/r years for small r in percent.

Arithmetic sequence: aₙ = a₁ + (n − 1)d; sum = n(a₁ + aₙ)/2. Geometric sequence: aₙ = a₁r^(n−1); sum = a₁(1 − rⁿ)/(1 − r) for r ≠ 1.

Ratios scale by a common multiplier k: if a:b = 3:5, write parts as 3k and 5k, then solve for k using the stated total or difference.

Work rate: combined rate equals the sum of individual rates; if A finishes in x hours and B in y hours, together they finish in xy/(x + y) hours.

Distance: d = rt. For average speed over equal distances, use the harmonic mean 2r₁r₂/(r₁ + r₂), not the arithmetic mean.

Geometry: circle area πr², circumference 2πr; triangle area ½bh; Pythagorean triples 3-4-5, 5-12-13, 8-15-17 recur frequently.

Overlapping sets: |A ∪ B| = |A| + |B| − |A ∩ B|; for three sets subtract pairwise intersections and add back the triple intersection.

Issue task: take a defensible position, concede one limitation of it, and rebut the strongest counterargument; a fully one-sided essay scores lower than a qualified argument.

Argument task: attack the reasoning, not the conclusion's truth; name the specific unstated assumption, then explain what evidence would test it.

Standard flaw patterns: hasty generalization from a small sample, post hoc ergo propter hoc, false dilemma, appeal to authority, survivorship bias, confusing correlation with causation.

Alternative-explanation move: show that a different cause could produce the same stated evidence, which defeats the argument's inference without new data.

Each body paragraph should carry one claim, one piece of support, and one sentence linking back to the thesis; paragraphs without a link read as a list.

Transitions that signal analysis rather than narration: 'because', 'therefore', 'however', 'even so', 'granting that', 'the flaw lies in'.

Length is not a scoring criterion; raters reward clear organization, varied sentence structure, precise vocabulary, and few mechanical errors.

Reserve time to proofread: sentence-level errors in grammar and spelling lower the score independently of the quality of reasoning.

Main-idea questions ask for the passage's central claim, not its topic; a correct answer covers the whole passage rather than one paragraph.

Primary purpose questions take an active verb plus object: 'to argue that', 'to reconcile two views', 'to question a hypothesis'.

Inference answers must be necessarily true given the text; an option that is merely plausible or likely is wrong.

Tone and attitude questions hinge on qualifiers: 'somewhat', 'arguably', 'surprisingly' signal hedged or critical stances rather than neutrality.

Structure questions map the passage: claim → counterclaim → evidence → concession → restatement; locate where the author shifts from reporting to evaluating.

Detail questions are answered by returning to the cited lines and matching the exact scope; watch for answer choices that broaden or narrow the claim.

EXCEPT and NOT questions: three options are supported by the text and one is not; verify each of the three against the passage.

Beware of answer choices that recycle passage vocabulary but reverse the logical relation, a common distractor pattern.

Read the axis labels, units, and legend before computing; a chart in thousands versus millions changes every derived answer.

Percent of total = part/total × 100; percentage-point change is the arithmetic difference of two percentages, not their ratio.

Weighted average = Σ(value × weight)/Σ(weight); use it whenever group sizes differ, since a simple mean of group averages is biased.

Index numbers: an index of 120 against a base of 100 means a 20 percent increase; never read the index value itself as a quantity.

Growth rate between periods = (later − earlier)/earlier × 100; for multi-year spans report the compound annual rate, not the total divided by years.

Tables with totals: check whether a row or column labeled 'total' is included in the data; double-counting it inflates every computed share.

Estimation strategy: round to two significant figures and compare the result to the answer choices; exact arithmetic is rarely needed.

Stacked bars and pie slices show composition, not absolute values; convert to raw numbers using the stated total before comparing categories.

Conditional logic: 'if P then Q' establishes only not-Q → not-P; affirming the consequent (Q therefore P) and denying the antecedent (not-P therefore not-Q) are invalid.

Necessary versus sufficient: P is sufficient for Q when P guarantees Q; P is necessary for Q when Q cannot occur without P; the two are not interchangeable.

Weaken questions: the correct answer makes the conclusion less likely given the premises, often by supplying an alternative cause or attacking a sampling assumption.

Strengthen questions: the answer closes a gap between evidence and conclusion; it need not prove the conclusion, only raise its probability.

Assumption questions: negate the answer choice; if the argument collapses under the negation, that choice is a required assumption.

Flaw questions name the error in abstract terms; match the description to the argument's move rather than to its subject matter.

Parallel reasoning: abstract the original to symbols (P → Q, Q, therefore P) and select the choice with the same structure and validity status.

Must-be-true answers follow from the premises with certainty; 'could be true' answers only need consistency with the premises.