Critical Thinking Questions
For a teacher, a student, a manager or any reader who wants the questions that test a claim before believing it, whether they are put to a class, a colleague or yourself. The first three groups follow the order a check goes in: pin down the claim, look at the evidence and where it came from, then dig out the assumptions, the other explanations and who gains. The last three carry the same moves into a classroom, the news and decisions at work, and every question has a note on what a good or a worrying answer sounds like and what to do with it.
Want questions from the whole vault instead? Try the random question generator.
The questions
Each question, and why to ask it
The claim
What exactly is being claimed, in one plain sentence?
Why ask it
Until someone can write that sentence, there is nothing to test and the argument is about a mood. A good one says who or what does what, and how much. "Screens are bad for kids" is the worrying kind: which screens, how many hours, and bad for what?
Is this a fact, an opinion or a prediction?
Why ask it
Each gets checked a different way: a fact against records, a prediction against what has happened by its date, an opinion against its reasons. Label the sentences one by one, and watch for a forecast or a judgment written in the grammar of a fact.
What is the conclusion, and which sentences are the reasons given for it?
Why ask it
Underline the conclusion once and number the reasons. No reasons means you are reading an assertion. Reasons that never arrive at the conclusion mean the subject changed somewhere, so find the paragraph where it happened.
What do the key words mean here, and would everyone in the room define them the same way?
Why ask it
Safe, fair, significant and works are the kind of word two people can use for an hour while meaning different things. Have two people write their definitions separately. If the two differ, settle that first, because the rest of the argument sometimes disappears once the word is fixed.
Could this claim be shown to be wrong, and what would that take?
Why ask it
A claim that survives every possible result is not strong, it is empty. Ask the person making it to describe the finding that would embarrass them. If they can, you have a test to run; if every outcome confirms it, treat it as a belief and move on.
How far does the claim reach: always, usually, sometimes or once?
Why ask it
"Some" has a way of becoming "most" as a claim is repeated. A careful speaker picks one of the four words and stands by it. "You know what I mean" is a sign the claim is bigger than its support.
Compared with what?
Why ask it
Faster, cheaper, more and record high all lean on a baseline that is often left out: last year, the competitor, doing nothing. If the speaker has no comparison ready, treat the claim as unfinished until one turns up.
Is this one claim, or several bundled together?
Why ask it
"The program is cheap, popular and effective" is three claims that need three kinds of support. Split them and check each alone. A common pattern is solid evidence for the least important one, lent quietly to the other two.
Evidence
What is the evidence, and can I look at it myself?
Why ask it
A good answer points somewhere: a document, a dataset, a named witness, something you could go and count. "Everyone knows" and "it has been shown" point nowhere, and the right response is to ask again, politely, for the place.
Where did this information come from originally?
Why ask it
Trace it back until you reach someone who saw, measured or decided the thing. Each retelling tends to drop a qualifier, so the original is often narrower than the version that reached you. If the trail stops at "people are saying", you are relying on a stranger's judgment and should know it.
How would this person or organization know what they say they know?
Why ask it
Seeing it, holding the record, calculating it and hearing about it are four different grades of knowing. Ask which one applies. An expert speaking outside their own field is at the hearsay end, with a better vocabulary.
Is this a single story, or a pattern someone has counted?
Why ask it
One vivid case shows a thing can happen and says nothing about how often. Ask, out of how many? When stories are all that is on offer, the claim you can accept is "this has happened at least once", and no stronger.
Could the cause run the other way, or could a third thing be driving both?
Why ask it
Ice cream sales and sunburn rise together because hot weather drives both, and neither causes the other. Before accepting that A causes B, say "B causes A" and "C causes both" out loud. If either sounds as plausible, what you have been shown is a link.
How old is this information, and what has changed since?
Why ask it
Information can be right on the day it was gathered and wrong by the time it reaches you. Look for the date the data was collected, which may be years before the date on the article quoting it. Then ask what has moved in between: prices, rules, technology, the people in charge.
Who or what was counted, and who was left out?
Why ask it
A satisfaction survey reaches the customers who stayed, and a poll reaches the people who answer polls. The missing group is often the one that would change the result. Good sources say who was left out without being asked.
Is there evidence pointing the other way, and what does the argument do with it?
Why ask it
Look for the paragraph that takes on the awkward findings and says why they do not sink the case. If there is none, the author either did not look or hoped you would not, so spend five minutes searching for the opposite claim before you decide.
Could this source have said anything else?
Why ask it
A company's page about its own product was always going to praise it, and a campaign flyer was always going to back its candidate. That does not make them false. It means their saying so adds little, and you need one source that was free to disappoint you.
How was this number measured, and how exact can it really be?
Why ask it
Decimal points make a figure sound measured even when it is a guess multiplied by a guess. Ask what was counted, by whom, and how far off it could be. Be more comfortable with a source that gives a range than with one that offers precision it cannot have.
Do several independent sources agree, or are they all repeating one?
Why ask it
Ten articles that trace back to one press release are one source counted ten times. Follow two of them to where they got it. Agreement only counts when each had its own way of knowing.
Assumptions
What is being taken for granted that nobody has said out loud?
Why ask it
Look for the step between reason and conclusion that only works if something unsaid is true. "Sales fell after the redesign, so the redesign failed" assumes nothing else changed that month. Say the hidden step aloud and see whether its owner still wants it.
What else could explain this?
Why ask it
Come up with at least three explanations before choosing, and make one of them dull: chance, timing, or a change in how things were counted. Someone who cannot produce a second explanation has not tested the first one. They have only met it.
Who benefits if I believe this?
Why ask it
A motive does not make a claim false, and people with something to gain are right every day. What it changes is how much checking the claim deserves before you act. A stated interest is a reason to relax a little, and a hidden one is a reason to look harder.
If this is true, what else should be true, and is it?
Why ask it
True claims leave other traces. A store that is "busier than ever" should also have longer lines, more staff hours and emptier shelves by evening, and when those are missing the claim needs another look.
Does the conclusion follow from the reasons, or does it only sit near them?
Why ask it
Grant every reason for a moment, then ask whether the conclusion is now forced, likely or merely possible. Plenty of persuasive writing has accurate facts and a conclusion those facts would support equally well in the other direction.
Are these really the only two choices?
Why ask it
"Either we raise prices or we close" and "you are with us or against us" leave out the middle and everything off to the side. Ask for a third option, even a poor one. If one comes easily, the either-or was steering you toward an answer and was never a full list of the choices.
Am I judging the argument or the person making it?
Why ask it
Swap the speaker in your head. Would the same sentence persuade you coming from someone you distrust, or irritate you coming from someone you admire? If your verdict flips, you were weighing the messenger, so go back to the reasons.
Do I want this to be true?
Why ask it
Wanting is not evidence either way, but it predicts where you stopped checking. When a claim flatters your side, your purchase or a decision you already made, give it the scrutiny you would give the reverse claim from someone you dislike.
What is the strongest version of the view I disagree with?
Why ask it
Write it so that someone who holds the view would sign it. If your version sounds foolish, you have not found it yet. Read the most careful person on that side and answer them, since beating the weakest version settles nothing.
What would change my mind?
Why ask it
The useful answer is specific and given in advance: a particular result, record or event. If the honest answer is "nothing", you hold a commitment and not a conclusion, which is allowed, but stop presenting it as where the evidence led. Then put the same question to the other person and compare.
How sure should I be, from a guess to nearly certain?
Why ask it
A yes or no hides most of what you know. Put a rough number on your confidence, then say what the remaining doubt is about. Being able to name the doubt matters more than the number, because it tells you what to check next.
What does it cost if I act on this and it turns out to be wrong?
Why ask it
If being wrong would cost you an afternoon, thin evidence will do. If it would cost your savings or someone's safety, it will not. Compare the price of acting on a false claim with the price of ignoring a true one, since the two are rarely equal. The expensive side is the one to check first.
In class
How do you know?
Why ask it
Four words, and worth asking after right answers as often as wrong ones. "The book says so" is a start; a student who can say how the book's author found out has gone a step further. Keep the tone curious, because it is easy to make this sound like an accusation.
Can anyone think of a counterexample?
Why ask it
One clean counterexample defeats "always" and "never", so students should hunt for one before defending a rule. When they find it, have them rewrite the rule so that it survives. The rewriting is where most of the learning is.
How did you get that answer, step by step?
Why ask it
A right answer reached by a wrong route fails on the next problem, in math and science especially. Have the student talk it through while a partner listens for the step they cannot follow. Praise the catch, whoever makes it.
Who would see this differently, and what reason would they give?
Why ask it
Name a specific person: a farmer, a factory owner, a child living at the time, a scientist from a rival lab. Vague "other people" produce vague answers. A strong response gives that person a reason, and a weak one gives them only a feeling.
Which part of that did you observe, and which part did you conclude?
Why ask it
The plant by the window is taller: that is observed. Light makes plants grow: that is concluded. Have students draw a line down the page and sort their sentences into two columns, and expect first drafts to put conclusions on the observation side.
What would you need to find out before you could decide?
Why ask it
This rewards a student for not answering yet, which classrooms seldom do. Good replies are specific and checkable, such as how many people were asked or who paid for the survey. Collect them on the board and let the class rank which missing fact matters most.
If we changed one condition, would the conclusion still stand?
Why ask it
Change the century, the country, the sample size or the sign of a number and see what breaks. This is how students find the part of an idea that is holding the rest up. It also turns a memorized fact into something they can reason from.
Which of these sources would you trust most, and what are you basing that on?
Why ask it
Hand out two or three short accounts that disagree about the same event. Listen for criteria, such as who was there, when it was written and what the writer stood to gain, over impressions like "it sounds official". Keep the class's criteria on the wall and reuse them all year.
What did you think about this before the lesson, and what moved you?
Why ask it
A closing question for a lesson or a unit. "Nothing moved me" is a fair answer if the student can say what evidence they met and why it fell short. A changed view with a named cause is some of the clearest evidence of thinking a quiet student will give you.
In the news
Does the headline say the same thing as the article under it?
Why ask it
A headline may be written by someone other than the reporter, and it has less room. Read down to the paragraph where the claim is actually stated and compare the verbs. "Linked to" in the text can turn into "causes" by the time it reaches the top.
Is this reporting, analysis, opinion or advertising?
Why ask it
Look for a label near the byline, then check the outlet's own page on how it marks each kind, since practice varies. Paid pieces can be laid out like articles. Where there is no label, see whether the piece quotes anyone who disagrees with it.
Who is quoted, and who was not asked?
Why ask it
Count the voices. A story about a strike that quotes only management, or only the union, is half a story however accurate each quote is. Notice too whether the people most affected get to speak or are only spoken about.
Is the source named, and if not, does the story say why?
Why ask it
Withholding a name can be legitimate, and a careful outlet explains the reason and how the person is in a position to know. "Sources familiar with the matter" and nothing more is a weaker grade. Outlets set their own rules on this, so look up theirs.
Is this photo or video what the caption says it is, from the time and place it says?
Why ask it
Old pictures get attached to new events, and generated ones are harder to spot by eye than they were. A reverse image search takes a minute and can show where the picture appeared before. Until you have run one, do not pass the image along as proof of anything.
Is that number a count, a rate or a change, and which one did the story leave out?
Why ask it
"Cases up by 200" means one thing in a village and another in a country. A risk that doubles may have gone from one in a million to two. Ask for the figure the story did not choose and see whether the headline survives it.
Why am I seeing this now, and who put it in front of me?
Why ask it
A feed is built to hold your attention, a friend shares what moved them, an editor picks what is new. None of those is the same as important or typical. If a story reached you because it made people angry, raise your bar before believing it.
How does this story make me feel, and was it built to?
Why ask it
Strong anger, fear or delight is the moment to slow down, because those are the feelings that get a post shared before it is read. The feeling may be fully deserved. Check the facts first and you get to keep it.
What do other outlets, including ones I rarely read, say about the same event?
Why ask it
Read two more accounts before forming a view on anything that matters to you. What all three agree on is the firmest ground you have. Where they differ shows each outlet's angle, your usual one's included.
What is confirmed, what is claimed and what is a guess at this point?
Why ask it
Early reports of a fast-moving event are often revised. Sort the story's sentences into those three piles and hold the last two loosely. Better outlets mark the difference themselves and note their corrections in the open.
At work
What problem are we actually trying to solve?
Why ask it
Ask it before anyone opens a slide. If three people in the meeting write three different problem statements, the argument about solutions is really an argument about that. A good statement fits in one sentence and names who has the problem.
What are the options, including doing nothing?
Why ask it
A decision with one option on the table is a request for approval. Ask for two real alternatives, and cost the do-nothing case honestly, since it is the baseline every other option claims to beat.
What does this data not tell us?
Why ask it
Dashboards measure what was easy to measure. Ask what the customers who left would say, what happened before tracking began, and which number could look healthy while the thing itself gets worse.
What will we see in three months if this is working, and what would make us stop?
Why ask it
Set the success sign and the stop sign before launch, in writing, with a date. Chosen afterward, both drift toward whatever happened. A team that will not name a stop condition has told you how the review is going to go.
Who disagrees with this, and have we heard their reasons in their own words?
Why ask it
The objection as summarized by the plan's sponsor is seldom the real objection. Invite the skeptic in, or read what they wrote. If nobody disagrees at all, find out whether that is agreement or whether people have stopped saying so.
Would we start this today, knowing what we know now?
Why ask it
Money and months already spent do not come back whichever way you choose, yet they pull hard on the decision. A "no" followed by "but we have come so far" is the pattern to catch. Ask what the next dollar buys, and ignore the last one.
Imagine this failed a year from now. What was the most likely reason?
Why ask it
Treating the failure as already past lets people name risks they would not raise as objections. Have everyone write one reason before anyone speaks, then read them out with no rebuttals. A reason that turns up on three slips of paper is where the safeguard goes.
Can this decision be reversed, and what would reversing it cost?
Why ask it
A choice that is cheap to undo deserves a quick call and a short trial. One that is hard to undo, such as a hire, a contract or a public statement, deserves the slower questions on this list. Meetings often give the first kind the time the second one needed.
Who is affected by this and is not in the room?
Why ask it
Name them: the support team who will take the calls, the customer on the old plan, the new hire who inherits it. Then get one of them to react before the decision is final. Their answer is cheap to hear this week and expensive to hear next quarter.
What would have to be true for the other option to be the better one?
Why ask it
It turns a fight between two camps into a list of conditions that can be checked. Have each side write the conditions for the option it opposes, then go and find out which ones hold. People concede a point more easily to a fact than to a colleague.
Using critical thinking questions on a real claim
Practical guidance for the conversation itself
The order to ask in
Claim first, evidence second
Most arguments that go in circles skipped the first group. Two people trade evidence for an hour and then find they meant different things by one word. Spend the first two minutes on what exactly is being claimed and how far it reaches, and the evidence questions get much shorter.
Match the effort to the stakes
A friend's restaurant tip needs none of this. A claim you are about to repeat in public, spend money on or build a decision on deserves a full pass. For everything in between, three questions cover most of the ground: what exactly is being claimed, how would they know, and what else could explain it.
Start with your own side
The questions bite hardest on what you already believe, and that is where they are least likely to be asked. Before putting one to anybody else, run "Do I want this to be true?" and "What would change my mind?" on your own position. It also earns you the right to ask.
Write the answers down
An answer held in your head adjusts itself to whatever you find next. One line on paper, written before you look, does not. This matters most for the questions about what would count against a claim and what you expect to see if it is true.
Asking other people
In a classroom
Pick one question for a week or a unit, put it on the wall and ask it until students start asking it of each other. Use it on correct answers as well as wrong ones, or the class learns that the question means "you are mistaken". Wait longer than is comfortable after asking, and let students turn the same question on you and on the textbook.
With younger children
The wording gets shorter and the moves stay the same. "How do you know?", "What else could it be?" and "Who told you, and how did they find out?" are short enough for the early grades. Try them on an advertisement or a cereal box before trying them on anything the child cares about.
In a meeting
Aim the question at the plan and not at its author: "What would we expect to see if this is working?" lands better than "How do you know this will work?". Ask one or two, early, while the plan can still change. If you are the senior person present, ask them of your own proposal first so that others see it is safe to.
With family and friends
One question, asked because you want to know, does more than seven asked to win. "Where did you see that?" opens a conversation, and "That's not a real source" closes it. If the person does not want to check, leave it. You were never going to settle it at the dinner table.
Where it goes wrong
Doubting everything equally
Treating every source as equally suspect takes as little effort as trusting them all, and it leaves you just as easy to mislead. The purpose of the questions is to sort: some claims come through with their sources, methods and corrections intact, and those have earned more confidence than the ones that did not.
Questioning only the other side
It is easy to become very good at finding holes in views you dislike and to wave your own through. Keep a rough count for a week of which claims you checked. If they all came from one direction, the habit you have built is advocacy.
Mistaking a flaw for a refutation
An argument with a weak step can still end at a true conclusion, and naming a fallacy does not settle the fact. When you find a hole, say what it means: the claim is unsupported so far. Whether it is false is a separate thing to find out.
Waiting for certainty
The questions can always turn up one more thing to check, and at some point a decision has to be made on the evidence available. Say how sure you are, what the doubt is about and what would make you revisit it. Then act, and keep the note.
Making it a habit
One claim a day
Take a single claim from your day, such as a headline, a line in a report or something you said yourself, and spend five minutes on it with three of the questions. Small and regular beats a long session once a semester. After a few weeks the first questions start arriving unprompted.
Keep a record of being wrong
Write down predictions and firm opinions with a date and a rough confidence, and look back at them. People who never check tend to remember being right. A short list of misses shows where your judgment runs ahead of your evidence, which no reading about bias will tell you.
When you cannot check it yourself
Nobody can verify most of what they rely on, so much of critical thinking is choosing whom to trust. Look at a source's record, whether it corrects itself in public and whether it was free to tell you the opposite. For a question about your own health, money or legal position, take it to a qualified professional who can look at your situation, and bring these questions with you.