Critical Thinking Test
Two lines and a conclusion. Say whether it follows, even when the conclusion sounds plainly wrong.
Settings
Changing one restarts the attempt, and scores set under different settings are not comparable.
Quick start
- 1Press Start. Two lines appear, then a conclusion under them.
- 2Treat the two lines as true, however odd they sound.
- 3Decide whether the conclusion has to be true if they are.
- 4Answer Follows or Does not follow. Press 1 or 2 to do it on the keyboard.
- 5Read the line saying what settles it, then press Next argument.
Does the conclusion follow?
Two lines, then a conclusion. Take the two lines as given, however odd they are, and say whether the conclusion has to be true.
Why there is no comparison
Pull of what sounds right (points). Lower is faster. No published distribution exists for this task.
Every other test here draws your score on a bar of published results. This one cannot. Nobody has measured enough people on this task to say what a typical result is, so a bar here would be a picture of a number we made up.
Numbers for this task do circulate. The ones we could find name no study and do not agree with each other. We would rather show you nothing than repeat one of those.
No published distribution, and not CognitiveDrill data either. Nobody has measured this task on a large enough sample to say what a typical result is, so this page shows your number and does not rank it. Our own norm is published once a cohort reaches n = 1,000.
Try next
About this test
Reasoning from a line you know is false
Half the arguments here start somewhere untrue. All birds can fly. All metals float. That is the task rather than a slip in the writing, and it is the one instruction worth reading twice.
The question is never whether the lines are true. It is whether the conclusion has to be true if they are. Those are different questions, and everyday reading runs them together because in ordinary life the premises usually are true.
So an argument can be perfectly sound and end somewhere absurd. If every metal floats and iron is a metal, then iron floats. The conclusion is false about the world and it still follows.
That is why the page never asks you to agree with anything. It asks whether one thing follows from another, which is the only part of this a test can mark.
Four kinds of argument, in equal numbers
A run draws the same number from each of four groups. Two of them are comfortable and two of them are not, and the difference between the pairs is the whole measurement.
- It follows and the conclusion sounds right. Both signals point the same way, so these are the easy ones.
- It follows and the conclusion sounds wrong. Logic says yes and your ear says no.
- It does not follow and the conclusion sounds right. Your ear says yes and the argument does not support it. This is the trap.
- It does not follow and the conclusion sounds wrong. Both signals point the same way again.
The number this page reports
Your headline is one accuracy minus another. How often you were right when belief and logic agreed, minus how often you were right when they pulled apart. Zero means the wording made no difference to you.
Both figures are on the card as well, because the difference alone hides which side did the work. Ninety against fifty and fifty against ten are the same gap and not the same run.
Evans, Barston and Pollard set this up in 1983 and found a large gap in the group they tested. Valid arguments with comfortable conclusions were accepted by roughly nine in ten. The same arguments with absurd conclusions were accepted by little more than half.
The broken arguments told the other half of the story. About seven in ten people accepted one when the conclusion sounded right, against roughly one in ten when it did not.
Why the order of the lines is a setting
By default the conclusion sits under the two lines, which is how an argument is usually laid out. The setting moves it above them.
Reading the conclusion first gives your ear a head start. You know where the argument is going before you know what it rests on. The search for support then starts from a position rather than from the lines.
Markovits and Nantel found the belief effect held up even when people were asked to produce a conclusion themselves rather than judge one. The pull is not a trick of the answer format.
The two orders are not comparable. Pick one and stay with it if you want to see whether anything changes with practice.
Why there is no percentile on this page
Nobody has published a distribution for this gap. The 1983 study reports acceptance rates for its own items in its own sample. The arithmetic behind your number is not the arithmetic behind theirs.
The items matter too. Swap these twenty four arguments for twenty four others and the gap moves. How comfortable a conclusion feels is a property of the sentence as much as of the reader.
There is a further problem with small numbers. A run of sixteen puts eight arguments on each side, so one lucky guess is worth twelve and a half points on your card.
For a number with a real distribution behind it, what counts as a good digit span has one. That is one task, measured the same way on thousands of people.
What a large gap does not mean
It is not a measure of intelligence, and nothing here has been checked against one. It is not a verdict on your judgement at work, and nobody has shown that this task predicts anything outside the page.
It is not a screening test either. Nothing here detects or rules out any condition, and no score is a diagnosis. If something has genuinely changed in how you think or decide, that belongs with a doctor.
A large gap does mean something narrower and more useful. On these sentences, today, the sound of a conclusion moved your answer. That is worth knowing before you sign off on a report.
Improving on this page is improving on this page. The same warning applies to every number here. The piece on what a good reaction time is makes that argument with an older one.
Questions
What is a good score on a critical thinking test?
Zero is the honest target here, because the headline is a gap rather than a total. It means believable wording made no difference to how often you were right.
Why are some of the premises obviously false?
Because the task is judging whether a conclusion follows, not whether it is true. A false premise is the cleanest way to separate those two questions.
Is this the Watson-Glaser test?
No. That is a published assessment with its own items and its own norms. This is a free page in the same family of tasks, and no score here transfers to it.
Can my score be negative?
Yes. It means you were more accurate when belief and logic disagreed, which usually comes from slowing down on the awkward ones.
Does the time per argument matter?
Nothing here is timed against you, and the median is reported because it is informative. Belief driven answers tend to be the quick ones.
Will I get better with practice?
Probably, on these arguments. There are twenty four in the pool, so a second run repeats most of them and part of the gain is memory.
Is this an IQ test?
No. Nothing on this page has been checked against any measure of intelligence, so any claim in that direction would be invented.
Does it work with a screen reader?
Yes. The lines and the conclusion are ordinary text, and the two answers take the 1 and 2 keys.
Sources
- Evans JStBT, Barston JL, Pollard P (1983). On the conflict between logic and belief in syllogistic reasoning. Memory & Cognition, 11(3), 295-306. Link
- Markovits H, Nantel G (1989). The belief-bias effect in the production and evaluation of logical conclusions. Memory & Cognition, 17(1), 11-17. Link
- Klauer KC, Musch J, Naumer B (2000). On belief bias in syllogistic reasoning. Psychological Review, 107(4), 852-884. Link