How to Solve an Analogy Reasoning Question

How do you solve a four-term verbal analogy?

For a four-term verbal analogy, read the first pair in order, state its relation clearly, carry the same roles and direction to the third term, then test every option in that sentence. In thermometer : temperature :: scale : weight, a thermometer measures temperature; a scale measures weight. “Both relate to measurement” is too broad to check an answer. Name what each word does and how the words connect.

Study the first pair before looking at choices; describe its relation; apply that sentence to the third term to predict the missing word; then test each candidate and reject those that change the relation. “How to Answer Verbal Analogy Questions: 10 Worked Examples” recommends this sentence-first method and refining the sentence if multiple choices survive. This is guidance, not proof one script is best for every item. Check for the same roles and direction: association with the third term is not enough. The sentence cannot guarantee a unique answer. An unfamiliar word may block inference, or two choices may fit the wording. Acknowledge that limit instead of inventing a distinction after choosing. One answer describes performance on one item under its conditions; it does not establish general reasoning ability.

Sources: How to Answer Verbal Analogy Questions: 10 Worked Examples

What is the relationship between understanding, inferring, and mapping?

A four-term verbal analogy asks you to preserve a relation across two pairs. First understand the terms, infer how the first two connect, map their roles onto the second pair, and apply that structure to complete the analogy. These are useful names for separable operations, not four guaranteed mental stages everyone consciously experiences. The article “Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?” presents encoding, inferring, mapping, and applying as component processes for analyzing analogy problems. Encoding means examining the terms and information given so you understand what each means in context. Before comparing words, ask whether you know them well enough to use their relevant meanings. That uncertainty matters because an inference built on the wrong sense can sound tidy and still miss the relation. Identify the sense that makes the pair coherent, using only information and instructions available. Inferring means identifying a meaningful connection between the first two terms, based on what you understood. A broad association is not yet enough. In thermometer:temperature, both words relate to science, but that does not explain their connection. A more useful inference is: “A thermometer measures temperature.” This sentence gives each term a role and a direction. It also creates a test: when you substitute the second pair, should the same statement hold? Mapping links that inferred relation to the second pair. In thermometer:temperature::scale:weight, the matching structure is measuring instrument followed by measured quantity: a thermometer measures temperature; a scale measures weight. What carries across is the relation, with its roles in the same order. This helps rule out a candidate chosen because its words share a topic, setting, or word class with the prompt. Applying means using the mapped relation to complete the problem and checking whether the candidate preserves its important features. A proposed pair that reverses which term measures the other has changed the roles. Another pair may name an object used in a laboratory, preserving the topic but not the relation. The full sentence exposes these differences. If the sentence only works after you quietly change its meaning, the candidate is a weak fit. Keep the original terms in view while checking so that a familiar association does not replace the relation you first identified. This approach treats analogy as relational reasoning: finding a meaningful structure in one set of elements and looking for a corresponding structure in another. “Analogy, Higher Order Thinking, and Education,” a review in WIREs Cognitive Science, discusses analogy through relations across material rather than surface resemblance alone. A precise paraphrase is a practical check, not proof that an item has only one defensible interpretation; wording and choices still matter. Did you misunderstand a term, infer only a loose association, map a sound relation onto the wrong roles, or fail to check the completion? The Frontiers article presents the operations in an instructional framework for tertiary education; it is not a controlled comparison showing this self-check is best for every item. Use the framework to inspect your work, not to diagnose general ability. A miss may reflect an unfamiliar word, a role reversal, or wording that leaves the relation unclear. One response alone cannot distinguish these causes or describe your general reasoning ability. Ask instead: what did this item require, what did I understand, and where did my proposed relation stop fitting?

Sources: Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?; Analogy, Higher Order Thinking, and Education

Why is a category label weaker than a complete relation?

A category label is a useful starting guess, but it leaves too much unsaid to test an answer. “Tool,” “opposites,” or “part and whole” names a broad family of possible links. A complete relation sentence specifies what the first word does in relation to the second, including which role comes first. That precision lets you check whether a candidate preserves the structure instead of merely belonging to the same topic. Consider this original teaching illustration, not a test item or research finding: scissors:cut::broom:sweep. “Both are tools” describes a category shared by scissors and broom, but it does not explain why cut pairs with sweep. A more useful sentence is, “Scissors are used to cut; a broom is used to sweep.” The relation is an implement paired with its characteristic action. If “dust” appears as a tempting candidate for broom, it is associated with sweeping, but it does not fill the action role in the sentence. The difference is between a connection that can be made in the world and the particular connection expressed by the first pair. The sentence also preserves direction. Read the first pair as “scissors are used to cut,” with the implement before the action. Reversing the terms gives “cut is done with scissors.” An analogy may allow a relation to be paraphrased, yet the mapping still has to be consistent across both pairs. A broad label such as “function” can hide whether a term is the agent, instrument, action, result, or object affected. Stating the relation makes the roles visible and helps catch a candidate that fits only after quietly reversing them. A second invented contrast shows why “part and whole” is not enough on its own. In “page:book,” a page is part of a book. In “book:page,” the order is whole then part. If the starting pair is ordered page:book, a candidate completion should preserve part-to-whole order. A reader who records only the label “part and whole” may accept either orientation; a sentence such as “A page is one part of a book” rules out the reversal. The aim is not to force every relation into a rigid formula. It is to say enough to make the proposed mapping checkable. This focus on structure matches “Analogy, higher order thinking, and education,” a review describing analogical reasoning as aligning, comparing, and mapping structured relationships. The review concerns reasoning across educational contexts, not a direct trial of one verbal-test strategy. Still, it supports a useful distinction for solving: shared subject matter or word type alone is weaker evidence than a relation whose roles correspond. “How to Answer Verbal Analogy Questions: 10 Worked Examples” likewise recommends expressing the first pair’s link as a sentence and using it to test choices; that is practical advice, not proof that one routine works best for every item. Use a category as a hypothesis, then sharpen it into a proposition: who or what does what to, with, or as what? Carry those roles and their order to the second pair. A candidate is stronger when the same sentence fits without changing its meaning. If a tempting option shares the topic but cannot occupy the required role, name that mismatch. The category gets you started; the complete relation gives you a falsifiable test.

Sources: Analogy, Higher Order Thinking, and Education; How to Answer Verbal Analogy Questions: 10 Worked Examples

How should you test a candidate without guessing by association?

Insert each candidate into the same complete relation sentence and check whether it preserves the first pair’s roles, order, and specificity. Reject an answer that is merely connected to the third term or requires changing the relation midway. This makes the comparison testable: each option faces the same interpretation instead of receiving a custom explanation after you have chosen it.

Consider an original practice example: compass : direction :: thermometer : ___. Before looking at choices, say, “A compass indicates a direction; a thermometer indicates a temperature.” The first word in each pair is an instrument; the second is what it indicates. Predict “temperature,” then test illustrative options: heat, temperature, weather, and glass. Heat is related to a thermometer, but it shifts from the quantity indicated to a possible condition. Weather is a broader context for using the instrument. Glass may be a material in an instrument, not what it indicates. Temperature preserves the same instrument-to-quantity relation. These options are invented to explain the method, not taken from a test or study.

A candidate can share a setting or common use with the third term and still fail. Ask not merely “Could these words go together?” but “Can the same sentence describe both pairs?” The Practice Aptitude Tests guide recommends stating the relation, predicting a completion, and checking choices. This is practical advice, not a comparative trial showing the routine is best.

Direction matters. “A key opens a lock” is not the same ordered relation as “a lock is opened by a key.” Reversing both pairs can preserve the relation if roles reverse consistently: key : lock :: password : account can fit “a key provides access to a lock; a password provides access to an account.” But account : password does not fit that ordering. Plug terms into their places instead of relying on a loose category such as “access.”

Keep the sentence fixed during elimination. For every option, ask whether it fills the same role, follows the same direction, and matches the same level of specificity. If one survives only after changing “indicates” to “contains” or “is used near,” it has not passed the original test. If no option fits, revisit the wording and your interpretation. Choices can reveal that the first sentence was too broad; revise only when the pair supports the change, then test all options again. Do not give a favorite candidate a special explanation you would reject for another.

In “GRE Verbal Analogy Items: Examinee Reasoning on Items,” recently tested GRE examinees thought aloud on eight items varied by vocabulary difficulty, relation difficulty, and stem-answer correspondence. The report’s abstract says vocabulary and correspondence affected examinees’ processes and correctness, while relation difficulty affected correctness. This small study does not compare the sentence routine with other methods or prove it universally optimal. It supports a narrower point: understanding the words and matching the relation both matter, so association alone is insufficient.

If two options still fit, refine the relation only with evidence from the pair. “Indicates” may be too broad if it clearly means “measures.” Do not invent an unstated condition to force a unique answer. If two choices remain equally consistent, the practice item may be under-specified. When required, select the strongest fit, then note the competing reading.

The framework in “Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?” separates understanding terms, inferring a relation, mapping it, and applying it. Sentence testing makes these operations inspectable: name the relation, carry its roles across, and check completion. The framework is not a direct trial of this method; it helps explain why association is weaker than matching structure. A useful answer states the relation, shows each term’s role, and identifies what the nearest alternative changes.

Sources: GRE Verbal Analogy Items: Examinee Reasoning on Items; How to Answer Verbal Analogy Questions: 10 Worked Examples; Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?

Where can the solving process break down?

When an analogy answer goes wrong, separate four possible problems before deciding what to practice: a term was not understood; the relation was inferred too broadly; its roles or direction were mapped incorrectly; or the wording supports more than one candidate. These checkpoints help review an answer; they are not a diagnosis or a validated checklist. The ETS report *GRE Verbal Analogy Items: Examinee Reasoning on Items* offers a bounded reason to distinguish them. Recently tested GRE examinees thought aloud while solving eight items, varied by vocabulary difficulty, relation difficulty, and the correspondence between the first pair and the answer pair. The report says vocabulary and correspondence affected both the process and correctness, while relation difficulty tended to affect correctness but not how examinees reached a solution. This small, specific study cannot establish which obstacle is most common in other tests.

First ask whether the words were understood. A solver may know a word but miss its relevant sense; an unfamiliar term can block comparison before relation inference. During timed testing, follow the stated rules and use only clues available in the prompt. Do not look up a word if outside resources are prohibited. If meaning remains uncertain, record that instead of calling it a reasoning failure. In untimed practice, a dictionary can clarify vocabulary after the attempt, but it cannot show that the original answer was reached from the information allowed under test conditions. Separate review can then target vocabulary or relational reasoning as appropriate.

Next inspect the first-pair relation. A label such as “connected” or “used with” may be too broad to constrain the answer. Ask what each term does in the relation and state that link in a sentence. If the sentence could admit many unrelated completions, the inference needs greater precision. Not every difficult item has one obvious relation. The ETS study varied relation difficulty and found it tended to affect whether participants answered correctly. It did not test these checkpoints or show that a sentence prevents every error. The sentence records your inference for review before you see the key.

Then check mapping: did the candidate take the same role, in the same direction, as the corresponding term in the first pair? A reversal can retain the same concepts and still be wrong. Grammar can help expose such a mismatch. For example, compare the subject and object of a sentence, or notice whether a verb describes an action or its result. But grammatical shape alone is not enough: a candidate can fit the same word pattern while expressing a different relation. Test whether the candidate preserves the stated relation. ETS reported an effect of stem-key correspondence, but did not establish one cause for mapping errors or validate grammar as a remedy.

Finally, consider whether the item itself distinguishes the remaining choices. If two candidates still fit, revisit the wording and test both under one interpretation. Do not silently change the sentence for one choice, or add an unstated condition to rescue a preferred answer. The research framework in *Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?* describes encoding, inferring, mapping, and applying as useful components of relational reasoning; it is an instructional account, not a trial of this review routine. These checkpoints turn an incorrect answer into a more specific observation: uncertain vocabulary, a loose relation, reversed roles, or competing readings. Write down which one best explains the attempt, then choose practice that addresses that step. One item cannot determine general language ability or reasoning capacity.

Sources: GRE Verbal Analogy Items: Examinee Reasoning on Items; Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?

What if two answers fit the relation sentence?

Make the relation sentence more precise using only the terms and wording in the item, then test every remaining candidate against that same sentence. If two still satisfy one well-supported relation, the item may be under-specified. Do not invent a distinction after choosing a favorite. On practice, this is feedback; on a formal test, follow its directions and choose the best-supported option if one is required. A choice can help refine an uncertain relation, but it should not write the relation for you. Begin with a provisional sentence from the first pair. Revise it only if that pair supports a missed feature, then retest every option under the same wording and roles. This prevents a preferred answer from receiving a more generous explanation than its alternatives. Imagine the first pair is “thermometer : temperature” and the incomplete pair is “scale : ___.” The sentence “This instrument measures this quantity” supports an answer naming weight. An option naming mass might also fit, depending on the item’s definitions. Do not narrow the rule to “measures body weight” unless the stem supplies that context; a distinction should come from the item, not an after-the-fact explanation. Check the stem for wording that narrows the relation, such as a specified purpose, direction, or degree. Apply that wording consistently to each surviving answer. Then ask whether rejecting one requires information absent from the question. A candidate may be associated with the third term without matching the first pair’s relation. If two candidates preserve the same roles and specificity without extra assumptions, the sentence has not separated them. In a practice set, note the uncertainty instead of treating an answer key as proof the wording was unique. If instructions require an answer, choose the best-supported option and move on; an examinee may need to decide even when an explanation is imperfect. The distinction between generating a completion and identifying an intended relation appears in research. “Response Generation Norms for Verbal Analogies,” a 1977 report, describes 71 university undergraduates asked to produce two completions for each of 150 incomplete verbal analogies. Researchers recorded response probabilities and separately classified whether initial responses represented the intended semantic relation. Responses varied across analogy bases and relation classes. This shows the outcomes are separable and responses differed by item. It does not show that a particular current multiple-choice item has two correct answers: participants generated two reasonable completions in an untimed task using test-derived bases. These response frequencies are not current norms. “GRE Verbal Analogy Items: Examinee Reasoning on Items,” an ETS report, used think-aloud protocols with recently tested GRE examinees working through eight items. Its publisher abstract says vocabulary and correspondence between stem and key affected processes and correctness, while relation difficulty tended to affect correctness. It cannot decide whether a disputed item is ambiguous, but shows difficulty can involve multiple features. The test-preparation guide “How to Answer Verbal Analogy Questions: 10 Worked Examples” also advises making the relation sentence more specific when multiple options survive. This is advice, not proof every question has one indisputable answer. Keep the conclusion proportional: multiple completions in an open-response study show variation in that task, not ambiguity in a particular multiple-choice item. Yet an answer key does not make every rationale sound. Record whether uncertainty came from a word, the relation, or the wording separating candidates. This directs the next attempt toward the difficulty instead of defending the first choice.

Sources: Response Generation Norms for Verbal Analogies; GRE Verbal Analogy Items: Examinee Reasoning on Items; How to Answer Verbal Analogy Questions: 10 Worked Examples

What does practice improve—and what remains unproven?

Practice can make the format more familiar and help you see which step in your solution needs attention. The evidence supports a narrower claim than “practice raises intelligence”: in two experiments with school-age students, analogy training improved performance on verbal analogy tasks, and one experiment found transfer to a related analogy format. That does not establish that this particular sentence routine raises general intelligence or that the same results apply to adults. After practice, ask not “Am I smarter now?” but “Can I use the relation on a new, related item, and where does my method still fail?” A gain on the task you practiced is a direct effect. Near transfer means improvement on a different but related task; farther transfer asks whether learning carries into a more distant outcome. “Training in Analogical Reasoning” reports two experiments using componential training. The first involved fourth graders described as having average reading ability. Training had a significant positive effect on A:B::C:? verbal analogies, but no significant effect on inferential comprehension, the farther outcome measured. The second involved gifted eighth graders and average-ability tenth graders; it reported direct effects on A:B::C:? items and near transfer to A:B::?:? verbal analogies. These results show that analogy performance can respond to instruction in the studied groups. They do not establish lasting gains, adult effects, or change in a general intelligence measure; the accessible abstract gives no effect sizes. A related format is a closer test of transfer than a claim about reasoning across everyday life. “Doing better” can name several outcomes. After repeated exposure, a person may understand directions sooner, recognize familiar relation forms, know more relevant words, or apply a strategy more consistently. Each can help on analogy questions. A higher result on practiced items alone cannot tell which change occurred, whether it carries to unfamiliar content, or whether another ability changed. The first experiment’s contrast between analogy performance and inferential comprehension is a reason not to treat one outcome as a stand-in for all the others: training helped on the former, while the latter did not show a significant effect. For practice, keep a record that separates these possibilities. After each fresh item, write the relation sentence you used, the nearest candidate you rejected and why, and the step that caused difficulty: understanding a term, inferring a relation, mapping its roles, or checking the choices. On a later, new example, ask whether the same reasoning operation works when the words and surface topic change. If it does, that is success on that new item, not by itself evidence of broad transfer. This log is a practical way to inspect your work, not a validated training program or measure of intelligence. Use varied, original items rather than memorizing an answer key. Explain why the relation fits and why a tempting alternative does not. This can reveal whether you are relying on a shared topic, a remembered pairing, or a relation with the right roles and direction. These are cautious practice recommendations derived from the task distinctions above, not claims that varied practice or self-explanation has been shown here to produce a particular score gain. Test IQ Free offers a separate low-stakes activity: its adaptive test has 50 original questions across pattern reasoning, number patterns, and spatial reasoning. It does not report verbal analogy as a domain, so success or difficulty there is not an analogy score. Its displayed estimated IQ range is provisional and comes from a conversion with no validated adult reference population; it is not normed, a confidence interval, or an established comparison with the population. The optional paid report uses that same unvalidated conversion, so payment adds explanation rather than validity. Use the test for reasoning practice and domain feedback, not to decide whether analogy practice changed general intelligence. The evidence supports a modest verdict: practice can improve analogy-task performance for the student groups and formats studied, with transfer to a related analogy format in one experiment. It does not support a promise that this routine raises general intelligence or that analogy gains spread to distant abilities. In your next session, use a fresh item and ask: can I state the relation, preserve its roles, and explain why the closest alternative fails?

Sources: Training in Analogical Reasoning; Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?

What can one answer tell you, and what should you do next?

One analogy answer tells you how you handled that item’s wording and conditions. It cannot establish an IQ level, diagnosis, worth, or fixed potential. A correct response may reflect sound mapping; an incorrect one may reflect vocabulary, relation, role, or wording problems. “Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?” describes relational reasoning assessment through multiple items and reported scores. That is a different basis from interpreting one answer, and it does not turn a single item into a measure of general ability.

For a useful next attempt, return to the worked explanation and identify the step that caused trouble: understanding the terms, stating the first pair’s relation, carrying its roles across, or comparing choices. Then try a fresh, original analogy and write the relation as a sentence before looking for a match. Keep the sentence and the words that support it. If two candidates still fit, record that uncertainty instead of inventing a distinction after selecting one. This makes practice feedback specific without promising that practice changes general intelligence.

If you want low-stakes reasoning practice, Test IQ Free offers a 50-question test across pattern, number, and spatial reasoning. Its displayed estimated IQ range is provisional and comes from an unvalidated conversion with no validated adult reference population. It is not normed, a confidence interval, or an established population comparison. The test does not assess verbal analogy performance, and its result is not equivalent to a professional instrument. The optional paid report adds explanation, not validity. After any item, ask: which exact relation sentence did I test, and which words support it?

Sources: Relational Reasoning in Tertiary Education: What Is Its Value and How Can It Be Assessed and Trained?

Questions readers ask