Interview questions for critical thinking: 14 questions, probes and a 1–4 scale

Good interview questions for critical thinking ask the candidate to show you a moment when they did not take something at face value: a number that looked wrong, a claim from a senior colleague, a conclusion they had already reached themselves. Below are fourteen questions, nine about past behavior and five situational, each with follow-up probes, plus sample strong and weak answers, a 1–4 anchored scale and the mistakes interviewers make most often with this competency.

This page is deliberately narrower than how to interview for problem solving, which uses a live case to watch someone work toward an answer. Critical thinking is the part of that process where the candidate checks whether the inputs deserve to be trusted. The decision itself, committing to an option once the analysis is done, is covered separately in interview questions for decision making.

Critical thinking, defined as something you can observe

Definition: critical thinking is evaluating information, claims and your own reasoning before acting on them, and changing your conclusion when the evidence calls for it. In an interview you cannot see anyone think. You can hear three behaviors in how they describe past work and how they respond to a claim you put in front of them:

• Questioning the source. Asking where a number or claim came from, how it was measured, and what it leaves out.
• Separating fact from inference. Saying which parts of a story were observed and which were assumed, without being prompted.
• Updating on evidence. Changing a view, including one they argued for, when new information contradicts it, and being able to say exactly what changed their mind.

Notice what is missing from that list: sounding skeptical, using analytical vocabulary, or having a degree in a quantitative field. Those correlate loosely with the behavior at best.

Nine behavioral questions, with follow-up probes

Ask for one specific situation, then use the probes. The probes are where the score comes from; the first answer is usually the polished version.

1. Tell me about a time you were handed data or a report and something in it did not look right.
   Probes: What exactly made you suspicious? What did you check first, and what did you find?
2. Describe a time you changed your mind about something important at work.
   Probes: What were you sure of beforehand? What specifically changed it? How long did it take you to accept it?
3. Tell me about a time a senior person made a claim you thought was wrong.
   Probes: How did you test your view before raising it? What did you say, and what happened?
4. Walk me through a recommendation you made where the evidence was mixed.
   Probes: What evidence pointed the other way? How did you present that uncertainty to the people deciding?
5. Tell me about an assumption your team was working on that turned out to be false.
   Probes: Who noticed, and how? What would have tipped you off earlier, looking back?
6. Describe a time you were asked to explain why something happened, such as a drop in results or a spike in complaints.
   Probes: Which explanations did you rule out, and how? What was the first explanation people offered?
7. Tell me about a vendor, tool or expert claim you had to evaluate before your team relied on it.
   Probes: What did you ask for as proof? What would have made you say no?
8. Describe a piece of your own work where you later found a flaw in your reasoning.
   Probes: How did you find it? What did you do once you had found it?
9. Tell me about a time you were under pressure to reach a conclusion quickly.
   Probes: Which checks did you skip, and which did you insist on? Why those?

Five situational questions and a live claim to test

Situational questions are useful for candidates with little work history, and as a cross-check on a story that sounded rehearsed. Keep them grounded in the role.

10. "Sign-ups rose 30% the week we launched a new homepage. The team wants to credit the homepage. What would you want to know first?"
    Probes: What else could explain it? What single piece of data would change your view most?
11. "A customer survey says 80% of users love a feature. Would you use that to decide whether to invest more in it?"
    Probes: Who answered the survey? Who didn't? What would you compare it against?
12. "Your manager asks you to present a conclusion you think the data does not support. What do you do?"
    Probes: What would you say in the meeting itself? What if they insist?
13. "Two reliable colleagues give you opposite accounts of why a project slipped. How do you work out what happened?"
    Probes: What would you look at that neither of them told you? How would you avoid favoring the one you trust more?
14. "Here is a one-paragraph summary from our last quarterly review. What questions would you ask before accepting it?"
    Probes: Which sentence is doing the most work? What is stated as fact that is really an inference?

Question 14 is the most revealing. Write a short, realiztic paragraph for your own team with one planted flaw, such as a before-and-after comparison with no control, a percentage without the base number, or a cause inferred from timing alone. Give the candidate a minute to read it. You are not scoring whether they find your specific flaw; you are scoring whether they separate what the paragraph shows from what it merely asserts.

What strong and weak answers sound like

These are illustrative answers to question 1, written to show the difference in shape rather than quoted from real candidates.

Weak: "I'm very detail-oriented, so I always double-check reports. There was one where the numbers seemed off, so I flagged it to my manager and they sorted it out."

Nothing here is wrong, but nothing is evidence either. There is no description of what looked off, no check the candidate ran, and the resolution happened to someone else. Under probing, answers like this tend to either produce a real story or dissolve.

Strong: "Our weekly pipeline report showed conversion from interview to offer jumping from about 20% to 35% in one month. Nothing in the process had changed, so I pulled the underlying rows. Two recruiters had started logging phone screens as interviews, which shrank the denominator. I recalculated with the old definition, it came back to 22%, and I wrote a one-line definition into the report so the stage could not drift again."

The strong answer names what triggered the doubt (a big change with no cause), the check they ran, the alternative explanation they found, and what they did to prevent a repeat. You could verify every step of it with a follow-up question.

A second contrast worth listening for, on question 2: a weak answer describes changing their mind because someone senior insisted. A strong answer describes changing it because of a specific piece of evidence, and can say what they had believed before and why they were wrong.

A 1–4 anchored rating scale for critical thinking

Score the evidence you heard against these anchors, not your overall impression of how sharp the candidate seemed. If you already use a scale, the anchors below slot into it; the general principles are in our guide to interview rating scales.

Keep a "not assessed" option. If two questions produce no usable example, record that rather than defaulting to a 2. For wording to paste into other competencies on the same scorecard, see the behavioral anchors library, or build a full card with the free scorecard builder.

Common interviewer mistakes with critical thinking

• Scoring disagreeableness as rigor. A candidate who pushes back on everything, including your questions, can sound like a critical thinker. Ask what evidence each objection rests on. Reflexive skepticism with no check behind it is a 1 or 2, not a 4.
• Rewarding vocabulary. "Correlation versus causation" and "confirmation bias" are easy to say. Score what the candidate actually did in the example, not the concepts they named.
• Accepting a story where someone else did the checking. "We found out the data was wrong" is a team outcome. Ask "what did you specifically check?" every time.
• Using a puzzle instead of a job-relevant claim. Brainteasers mostly measure whether someone has practiced brainteasers. A flawed paragraph from your own domain tests the real behavior and is fairer to people from different backgrounds.
• Marking down "I don't know." A candidate who says what they don't know and how they would find out is showing the competency. Penalizing it trains candidates to bluff.
• Letting your own view of the right answer drive the score. On situational questions, a candidate can reach a different conclusion from yours through sound reasoning. Score the reasoning. The same principle runs through reducing bias in interviews generally.

Fitting these questions into an interview plan

Critical thinking rarely needs its own interview. Pick two behavioral questions and the live claim, assign them to one interviewer, and make sure no one else on the panel is asking the same questions under a different label. If you keep a shared bank, file these under a single competency code so scores stay comparable across interviewers; how to build an interview question bank covers the structure. When you write up the score, quote the specific check the candidate described. A note that says "strong analytical skills" cannot be calibrated; a note that says "recalculated the conversion rate after finding a stage-definition change" can.