Cognitive Ability Test Example Questions (With Answers)

What cognitive ability tests measure and why they matter

Cognitive ability tests are the single strongest predictor of job performance across roles and industries. Meta-analyses by Schmidt and Hunter put the validity correlation around 0.51 - higher than educational credentials, interviews, reference checks, or personality tests.

When you hire a software engineer, a product manager, a sales director, or an operations lead, cognitive ability matters because it predicts learning speed, complex problem-solving, and adaptability under uncertainty. Yet many hiring teams have never seen what these tests actually look like.

Here are ten real-world example questions across verbal, numerical, abstract, and logical reasoning, with scoring rationale. Each reveals not just whether a candidate gets it right, but how they think.

Verbal Reasoning Examples

Verbal reasoning tests measure comprehension, vocabulary precision, and logical inference from written text. These correlate strongly with leadership roles, sales, and client-facing positions.

Example 1: Reading comprehension and inference

Passage: "The adoption of remote work policies by large tech companies has created unexpected consequences in urban real estate. Downtown office buildings designed in the 1980s now sit half-empty, reducing property tax revenue for cities already stretched thin. However, residential neighborhoods near transit hubs have seen a surge in young professional renters, increasing local demand for coffee shops, restaurants, and gyms. The net effect on city budgets remains unclear."

Question: Which of the following can be reasonably inferred from the passage?

A. Remote work is harmful to cities overall. B. Young professionals moving to neighborhoods benefits some city services more than others. C. Office buildings from the 1980s were poorly designed. D. Cities should ban remote work to protect tax revenue.

Answer: B. The passage explicitly notes declining office tax revenue but increasing commercial demand in residential areas. Option B acknowledges this mixed impact without overstating it. Options A and D draw conclusions not supported by "the net effect remains unclear." Option C is not supported; the passage doesn't evaluate design quality.

This question measures whether a candidate can extract nuance from a passage and avoid false certainty. Hiring teams use this to assess leaders who need to synthesize ambiguous information and communicate to stakeholders. A candidate picking A or D is oversimplifying; a candidate picking B understands systems thinking.

Example 2: Logical argument evaluation

Statement: "Employee engagement scores at our company have dropped 12% this year. A competitor recently implemented a four-day work week and reported improved engagement. Therefore, implementing a four-day work week will solve our engagement problem."

Question: What logical error does this argument contain?

A. It assumes correlation implies causation. B. It uses a sample size too small to draw conclusions. C. It cherry-picks a single competitor's success without considering other variables. D. All of the above.

Answer: D. This argument commits multiple logical errors: the competitor's success may correlate with the four-day week, but engagement depends on pay, management quality, team cohesion, and role fit - not just work schedule (A). One competitor's experience is weak evidence for organizational change (B). And the argument ignores whether that competitor faced the same engagement drivers as your company does (C).

This measures ability to recognize when a business argument is unsound before acting on it. Candidates who pick only A are reasoning shallowly. Candidates who pick D have spotted multiple failure modes and are more likely to design experiments or gather more data before major decisions.

Example 3: Vocabulary in context

Sentence: "The auditor's report was so voluminous that the finance team struggled to extract the key findings from the morass of detail."

Question: As used in the sentence, "morass" most nearly means:

A. Bog or swamp B. Confusing entanglement of complexity C. Incomplete information D. Deliberate obfuscation

Answer: B. While "morass" literally refers to a bog (A), in this context it's used metaphorically to describe a dense, confusing mix of details - a tangled mass of information. Option C (incomplete) misses the point - there's plenty of information, just too much to parse. Option D implies intentional hiding; the sentence suggests overwhelming volume, not deliberate concealment.

This question surfaces precision in language. Candidates who pick A have stopped at the dictionary definition without reading context. Candidates who pick B understand professional communication and can extract precise meaning from business documents - crucial for managers, analysts, and customer-facing roles.

Numerical Reasoning Examples

Numerical reasoning tests measure mathematical problem-solving, data interpretation, and financial acumen. These correlate strongly with technical roles, finance, operations, and roles requiring analytical depth.

Example 4: Mathematical problem-solving

Problem: A customer buys a jacket on sale for 25% off the original price. The sale price is $90. What was the original price?

Answer: $120. If the sale price is 75% of the original (100% - 25% = 75%), then Original Price = $90 / 0.75 = $120.

A candidate who answers $67.50 (subtracting 25% of $90 instead of solving backwards) is making a common error: applying the discount rate to the wrong base. This reveals whether the candidate understands the structure of percentage problems or just recognizes keywords and applies a formula. In hiring for finance, pricing analysis, or budgeting roles, this distinction matters - the wrong approach scales poorly to compound discounts or multi-step calculations.

Example 5: Data interpretation and estimation

Scenario: A SaaS company has 500 active customers. 40% are on the $50/month plan, 35% are on the $150/month plan, and 25% are on the $500/month plan. What is the monthly recurring revenue (MRR)?

Answer:

• $50 plan: 500 * 0.40 * $50 = $10,000
• $150 plan: 500 * 0.35 * $150 = $26,250
• $500 plan: 500 * 0.25 * $500 = $62,500
• Total MRR: $98,750

A candidate who answers $200 (averaging the three plans and multiplying by 500) has skipped the weighting step - a critical failure for operations, finance, or product roles. A candidate who gets $98,750 understands multi-step calculation and can work with weighted data. This is foundational for roles involving P&L, pricing, or revenue modeling.

Example 6: Ratio and proportional reasoning

Problem: You are hiring for a customer support team. Your current ratio is 1 support agent per 30 customers. You are expecting to grow from 3,000 to 5,000 customers over the next year. How many additional support agents should you hire?

Answer:

• Current team size: 3,000 / 30 = 100 agents
• Team size needed at 5,000 customers: 5,000 / 30 = 166.67, round to 167
• Additional hires needed: 167 - 100 = 67 agents

A candidate who answers "33 agents" (10% of 3,000) is using a percentage heuristic instead of applying the ratio. A candidate who answers "67" understands how to scale proportionally and can project resource needs for growth. This matters for operations, project management, and leadership roles where headcount and budget planning are central.

Abstract and Pattern Reasoning Examples

Abstract reasoning tests measure pattern recognition, spatial logic, and inductive reasoning without relying on learned knowledge. These correlate strongly with technical roles, engineering, software development, and roles requiring novel problem-solving.

Example 7: Pattern completion in sequences

Sequence: 2, 5, 10, 17, 26, ?

Question: What is the next number?

Answer: 37. The pattern is differences between consecutive numbers: 5-2=3, 10-5=5, 17-10=7, 26-17=9. The differences increase by 2 each time (3, 5, 7, 9). The next difference should be 11, so 26 + 11 = 37.

A candidate who answers 36 (adding 10 to the last number) has spotted a pattern but not the correct one. A candidate who answers 35 (a different arithmetic progression) has recognized that a pattern exists but failed to identify it correctly. A candidate who answers 37 has engaged in inductive reasoning - spotting the meta-pattern within the pattern. This reveals analytical rigor and the ability to recognize nested structures, valuable in software engineering, data analysis, and scientific reasoning roles.

Example 8: Spatial logic and categorization

Scenario: Consider these four items: hammer, screwdriver, wrench, pliers. Which of the following is NOT a valid categorization?

A. All are tools. B. All are primarily used to apply rotational force. C. All are used in construction and repair. D. All are handheld without requiring a power source.

Answer: B. A hammer applies force in a striking direction, not rotational. A screwdriver, wrench, and pliers all apply rotational force or rotational resistance. Options A, C, and D are valid - all four items fit each category. A candidate who picks B has recognized the exception within a set of true statements.

This tests category boundaries and logical precision. Candidates who pick A, C, or D have missed the flawed reasoning in the question. Candidates who pick B understand set logic and can identify the outlier - useful for QA engineering, data validation, taxonomy building, and any role requiring exactness.

Example 9: Abstract relationship reasoning

Analogy: Painting is to Canvas as Sculpture is to:

A. Stone B. Marble C. Chisel D. Pedestal

Answer: A (Stone, most generally). The relationship is "discipline to primary medium." Painting uses canvas as its primary material base. Sculpture uses stone/marble/clay as its primary medium. Option B (marble) is a specific type of stone, but stone is the broader, more parallel answer. Option C (chisel) is a tool, not a medium. Option D (pedestal) is a support structure, not the medium of the work.

This reveals whether a candidate can identify abstract relationships and find the closest parallel. In hiring for roles requiring analogy-making (strategic planning, architecture design, narrative communication), this matters. A candidate who picks C has confused tools with materials; a candidate who picks D has confused support with substance. A candidate who picks A understands category hierarchy and parallel relationships.

Logical Reasoning Examples

Logical reasoning tests measure deductive and inductive inference, often under time pressure. These correlate with all roles but especially with management, strategy, and client advisory positions.

Example 10: Deductive logic puzzle

Premises:

1. All people in our engineering team are strong problem-solvers.
2. All strong problem-solvers communicate clearly.
3. Some people in our engineering team work on infrastructure.

Question: Which conclusion is valid?

A. All people in our engineering team work on infrastructure. B. All people who communicate clearly are in our engineering team. C. Some people in our engineering team communicate clearly. D. Some people in our engineering team do not communicate clearly.

Answer: C. From premises 1 and 2, we know: if you're in engineering, you're a problem-solver, and if you're a problem-solver, you communicate clearly. Therefore, all people in engineering communicate clearly. Premise 3 tells us at least some people are in engineering; therefore, some people in engineering communicate clearly (in fact, all of them). Option A is false (only "some" work on infrastructure, not "all"). Option B reverses the implication incorrectly. Option D is false (all engineers communicate clearly, so none fail to).

This requires holding multiple logical chains in working memory and tracking necessity versus possibility. Candidates who pick A or B are making common inference errors. Candidates who pick C can follow multi-step logic and avoid overgeneralizing - valuable for roles involving legal reasoning, technical design review, or risk assessment.

Adaptive Cognitive Testing

The term "cognitive ability adaptive test" refers to assessments that adjust difficulty based on candidate performance. If a candidate answers a medium-difficulty question correctly, the next question becomes harder. If they answer incorrectly, the next question becomes easier.

Advantages of adaptive testing:

• Reduces testing time (fewer questions needed to measure ability accurately)
• Reduces frustration (candidates aren't presented with questions wildly above or below their level)
• Increases measurement precision (difficulty stays calibrated to candidate ability)
• Reduces ceiling and floor effects (fewer candidates scoring at the extremes due to test design, not actual ability)

Disadvantages:

• Less transparent to candidates (they may not understand why questions are varying in difficulty)
• Harder to compare across candidates if different candidates take different questions
• Requires more sophisticated test design and validation

Scoring, Percentiles, and Fair Use

Cognitive ability tests are scored against role-specific norms. A score of 50th percentile means the candidate scores as well as the median candidate for that role. Scores above 75th percentile indicate above-average reasoning ability for the role; scores below 25th percentile indicate below-average.

However, cognitive ability tests show meaningful subgroup differences in average scores across demographic groups. This is well-established in the research and has been a persistent finding for decades. Organizations should pair cognitive testing with other assessment methods and ensure fair recruitment pipelines upstream to mitigate adverse impact.

Best practice: combine cognitive ability with work samples, situational judgment, and culture-fit assessments. A high cognitive score doesn't guarantee job performance - motivation, experience, and team dynamics matter equally.

Administering Cognitive Tests Remotely

Cognitive ability tests are harder to cheat on than knowledge-based MCQs because they measure reasoning ability, not lookup-able facts. However, remote administration carries risks: candidates may seek outside help, use AI tools, or have a test-taker proxy complete the assessment.

Time pressure makes lookup difficult (there's no time to search Google mid-problem), but AI assistance is a real threat. The best defensive measure is integrity verification: capture keystroke biometrics and face continuity during the test. Anomalies - like instant-correct answers on difficult abstract reasoning questions, or face presence dropping during the assessment - surface when outside help is likely.

ClarityHire's cheat detection runs by default on all cognitive assessments, flagging solution-time anomalies and behavioral inconsistencies. This keeps unproctored cognitive tests reliable and scalable.

Using cognitive ability tests in your hiring

Cognitive ability testing works best when:

• You use role-specific norms (comparing a software engineer candidate against engineering norms, not executive norms)
• You combine it with predictive validity research for your specific roles - validating that high scores in your candidate pool actually predict on-the-job performance
• You compare vendor assessments carefully - different providers (Criteria Corp, SHL, Hogan, Pearson) weight reasoning domains differently
• You administer consistently to all candidates for the same role (avoiding bias in who gets tested)
• You interpret scores in context with experience, demonstrated skills, and culture fit

Cognitive ability is measurable, predictive, and scalable. Understanding what these tests actually measure - and what they don't - helps you make faster, more confident hiring decisions.

Explore ClarityHire's cognitive ability assessments to build your first screening, or contact our team to validate predictive validity for your roles.