CogAT common mistakes
CogAT · Understand the traps and how to avoid them.
Treating a verbal analogy as a loose topic association - choosing the pair that merely 'goes together in the same story' (e.g., 'dog is to bone' because dogs chew bones) rather than one that matches the exact relationship in the stem (e.g., 'dog is to puppy' as a parent–offspring pattern).
Analogy items reward the relation, not the theme. Under time pressure the mind latches onto the first plausible link it finds, and a thematically related distractor feels satisfying before the candidate has actually named the relation in words.
What to do instead
Before looking at the options, state the stem relation as a full sentence ('A ___ is a kind of ___', 'A ___ is the young of a ___', 'A ___ is used to ___'). Then test every option against that sentence and eliminate any that fails it. If two options survive, the relation is too broad - refine it (kind-of vs. part-of vs. function) rather than guessing.
Answering sentence-completion and verbal-classification items with the first word that 'sounds smart' instead of the word that fits the sentence's logic and tone, and skipping the check of whether the chosen word makes the whole sentence coherent.
Vocabulary items tempt candidates to pick the most impressive-sounding option. Distractors are often near-synonyms with the wrong shade of meaning or wrong grammatical role, and a strong-sounding word masks the mismatch.
What to do instead
Cover the options, predict your own word from context, then match. Check part of speech and tone (positive/negative/neutral) against the sentence. For classification, find the single shared category that excludes every odd word - name the category aloud before selecting.
Reading a short passage once and answering inference questions from memory, choosing an option that is true in real life but not stated or implied in the passage.
Passages are short, so candidates assume they can hold them in mind. Inference questions are designed so that plausible world knowledge competes with text-based reasoning, and memory of the passage drifts within seconds.
What to do instead
For each question, return to the specific sentence(s) that support the answer and reject any option that requires outside knowledge. Distinguish 'stated' from 'implied' from 'not supported' explicitly. If you cannot point to the supporting line, the option is wrong.
Rushing number-series and quantitative-comparison items by checking only one operation (e.g., only adding a constant) and then guessing when the pattern does not fit.
Simple series train the habit of 'find the difference', so candidates stop after one failed attempt instead of testing alternating operations, two interleaved sequences, or multiplicative/accelerating rules. Anxiety about time reinforces the early guess.
What to do instead
Use a fixed checklist: (1) differences, (2) second differences, (3) ratios, (4) alternating or two interleaved sequences, (5) squares/cubes or near-multiples. Write the differences under the terms. Only after the checklist fails should you estimate by elimination.
Solving word problems by grabbing the numbers and computing immediately, without identifying what is asked or what units the answer needs.
The numbers are salient and computation feels productive. Distractors are built from the most common wrong operation (adding instead of multiplying, forgetting a rate, mixing units), so fast arithmetic lands on a trap answer.
What to do instead
Restate the question in one sentence ('find total cost', 'find time', 'find difference') and note the required unit before computing. Estimate the answer's rough size first, then compute, then compare to the estimate. If they disagree, re-read the problem.
On quantitative comparison, computing both quantities fully when a relationship can be established by reasoning, and forgetting to test zero, one, fractions, and negatives before choosing 'cannot be determined'.
Candidates default to arithmetic because it feels safe, which burns time. Conversely, when they do reason, they test only positive integers and miss the cases that make the answer indeterminate.
What to do instead
Simplify both quantities symbolically first; only compute if simplification stalls. Before selecting 'cannot be determined', deliberately plug in a negative, a fraction between 0 and 1, zero, and one. If any legal value flips the comparison, the answer is indeterminate.
Treating figure analogies and classification items as 'find the picture that looks similar' rather than identifying the transformation (rotation, reflection, size change, shading, added/removed element) applied to the stem pair.
Visual similarity is fast and automatic, while transformation analysis is effortful. Distractors are deliberately similar in appearance but wrong in transformation, so appearance-based matching is systematically punished.
What to do instead
Name the change in words before scanning options: 'rotated 90° clockwise and shading inverted'. Check each option against every named attribute (shape, orientation, count, shading, position). Eliminate on the first attribute that fails.
Assuming a rotated figure is a different figure, or that a mirrored figure is the same as a rotated one, when judging paper-folding, figure-rotation, and fold-and-punch items.
Mental rotation is error-prone, and mirror images look deceptively like rotations. Candidates also forget that folding creates multiple symmetric holes, so they count one punch instead of its reflections.
What to do instead
Track orientation with a landmark (a dot, a corner, an arrow) and rotate it mentally step by step. For fold-and-punch, unfold one fold at a time, reflecting the hole across each crease in reverse order. Distinguish mirror (handedness flips) from rotation (handedness preserved).
Spending too long on one hard item in a timed section, then rushing the remaining items and making avoidable errors on easy ones.
Candidates interpret 'hard' as 'I must solve this to score well', but the exam rewards correct answers across the whole section. A single item can consume several minutes that would have secured multiple easier points.
What to do instead
Set a per-item time budget (roughly one minute for most items). If you are not converging, mark a best-guess answer, move on, and return only if time remains. Never leave a blank - an educated elimination beats no answer.
Changing a first answer on review without a concrete reason, or second-guessing a reasoned answer because a later option 'looks better'.
Anxiety near the end of a section produces doubt, and the second look happens without the original reasoning being re-derived. Research on answer-changing consistently shows that deliberate changes help but anxious changes hurt.
What to do instead
Only change an answer if you can state a specific error in the original reasoning (misread the relation, missed a rotation, arithmetic slip). Otherwise keep the first reasoned answer. Review flagged items, not every item.