Skip to content
Check My Logic
Check My Logic

Check My Logic

If someone sent you this link, they have concerns.

A plain-language reference to the ways reasoning goes wrong: formal and informal fallacies, cognitive biases and the biases that creep into research. Each entry gives you enough to understand it at a glance, plus the details, the evidence and the cases where it isn’t an error at all.

122 entries and 17 concepts so far · Show me a random one

Categories

Formal fallacies

An argument whose shape can’t guarantee its conclusion, whatever the argument is about.

Informal fallacies

An argument that fails because of its content or context, not its logical shape.

Cognitive biases

A predictable way human judgment drifts from what the evidence or logic supports.

Research biases

A systematic error in how a study is designed, run, analyzed or reported that pushes its results away from the truth.

Underlying ideas

Not errors themselves.

Valid forms

Correct argument forms. The formal fallacies are look-alikes of these.

Disjunctive syllogism
“P or Q. Not Q. Therefore P.” The valid way to reason by elimination from an either/or statement.
Modus ponens
“If P, then Q. P. Therefore Q.” The basic valid way to reason from an if-then statement.
Modus tollens
“If P, then Q. Not Q. Therefore not P.” The valid way to reason backward from an if-then statement whose result didn’t happen.

Principles

Ideas that explain why errors happen, or how to tell good reasoning from bad.

Burden of proof
The obligation to support a claim, to a standard that fits the situation, before others can be expected to accept it.
Chesterton's fence
Before removing a rule, custom or safeguard that seems pointless, find out why it was put there.
Confirmation bias
Favoring evidence for what you already believe, in how you look for it, read it and remember it.
Correlation vs. causation
Two things can rise and fall together without one causing the other, so a correlation is evidence about causes, not proof of one.
Goodhart's law
Once a number is used as a target, people change what they do to move the number, and it stops tracking what it was meant to show.
Motivated reasoning
Wanting a conclusion shapes how you search for, scrutinize and remember evidence about it, while the reasoning still feels objective.
Multiple comparisons problem
The more statistical tests you run, the more likely at least one comes out “significant” by chance, so a lone hit among many tests is weak evidence.
Necessary and sufficient conditions
A sufficient condition guarantees something; a necessary condition is something it can’t happen without.
Occam's razor
Among explanations that fit the evidence equally well, prefer the one that assumes the least.
Principle of charity
When what someone said can reasonably be read more than one way, answer the strongest plausible reading, not the weakest.
Regression to the mean
A case picked for being extreme on one measurement will usually be less extreme on the next, because the luck that made it extreme doesn’t carry over.
Unfalsifiability
A claim is unfalsifiable when no possible observation could count against it, so it rules nothing out and tells you nothing about what to expect.
Valid vs. true
Truth belongs to statements; validity belongs to arguments. A flawed argument can reach a true conclusion, and a valid one can reach a false one.
What you see is all there is
Judging from the information in front of you as if it were all the information there is.

Every entry

  • The claim that people explain their own behavior by the situation but other people’s by their character, which a large meta-analysis did not support.

  • Concluding that one option is false because the other is true, when the “or” allowed both to be true.

  • Concluding that the “if” part is true because the “then” part turned out to be true.

  • Drawing a conclusion from the wrong reading of a sentence whose grammar allows more than one meaning.

  • Letting a number you’ve just considered pull your estimate toward it, even when that number tells you little.

  • Accepting a claim because someone presented as an authority said so, when their say-so isn’t good evidence on that question.

  • Deciding whether a claim is true by how good or bad it would be if it were true.

  • Trying to win agreement by stirring up a feeling, such as pity, fear or outrage, in place of reasons that bear on the claim.

  • Using a threat to get someone to accept a claim or stop disputing it, as if the threat were a reason to think the claim is true.

  • Treating “natural” as if it meant good, safe or right, and “unnatural” as if it meant bad.

  • Treating something as better or truer simply because it’s new.

  • Treating the fact that many people believe or do something as proof that it’s true or right.

  • Treating someone’s poverty or humble way of life as a reason to think they’re honest or right.

  • Treating a belief or practice as right because it’s old, or because it has always been done that way.

  • Treating someone’s wealth or financial success as a reason to think they’re right.

  • Concluding that a claim is false because an argument for it was fallacious or badly made.

  • Treating a lack of proof as proof of the opposite, as in “it hasn’t been disproven, so it’s true” or “it hasn’t been proven, so it’s false”.

  • Concluding that something is false, or impossible, because you can’t imagine or understand how it could be true.

  • Being drawn toward an option because a clearly worse version of it is also on offer, although that worse option says nothing new about your real choice.

  • Judging how common or likely something is by how easily examples of it come to mind.

  • The claim that correcting a false belief makes people believe it more strongly, which large replications have rarely found.

  • Judging how likely something is from the details of one case while giving too little weight to how common it is in the first place.

  • Supporting a claim with premises that already assume it, so the argument gives a doubter no independent reason to accept it.

  • Judging whether an argument is logically valid by whether you already believe its conclusion.

  • Keeping a belief, at least in part, after the evidence it was based on has been shown to be false.

  • Seeing biases at work in other people’s judgments more readily than in your own.

  • Dismissing an idea because it’s old or from an earlier era, instead of asking whether it was ever shown to be wrong.

  • A spurious link between two things that appears when data are restricted to, or adjusted for, something both of them affect.

  • A third factor that affects both an exposure and an outcome makes the association between them a misleading guide to the exposure’s effect.

  • Judging that two things happening together is more likely than one of them happening on its own.

  • Concluding that there’s no real difference between two things because they shade into each other with no sharp dividing line.

  • Concluding that one thing causes another only because the two occur together.

  • Finding it hard to imagine not knowing something you know, so you overestimate how much others know or how easily they’ll understand you.

  • Concluding that one of two things is true because they can’t both be true and the other one isn’t.

  • Concluding that the “then” part is false because the “if” part turned out to be false.

  • Checking arguments against your view much harder than arguments for it, so the unwelcome ones reliably come out looking weaker.

  • The claim that people who are bad at something overestimate how good they are, because judging the skill takes the skill.

  • Assuming that a relationship seen across groups, such as states or schools, also holds for the individuals within those groups.

  • Valuing something more just because it’s yours, so you’d demand more to give it up than you’d pay to get it.

  • Using one word or phrase in two different senses within an argument, so the conclusion seems to follow when it doesn’t.

  • Claiming that a word’s “real” meaning is its original or older meaning, and that the way people use it now is therefore wrong.

  • Concluding that something of a kind exists from “all” or “no” statements, which can be true even when nothing of that kind exists.

  • Concluding that a whole or group has a property because its parts or members each have it.

  • Concluding that each part or member of something has a property because the whole or the group has it.

  • A syllogism that seems to link two groups through a shared term, when the “shared” term is really two different things.

  • Overestimating how many other people share your opinions, choices and habits.

  • Presenting a few options, usually two, as the only possibilities when there are others, or when more than one could be true.

    • Rejecting a workable option because it isn’t perfect, by comparing it with an ideal instead of with the alternatives actually available.

  • Making a different choice when the same facts are described differently, such as “90% succeed” versus “10% fail”.

  • Explaining what someone did by the kind of person they are, while underweighting the situation that produced it.

  • Research paid for by an organization with a stake in the answer tends to reach answers that favor it, through choices in design, analysis and reporting.

  • Expecting a run of one outcome in a random process to make a different outcome more likely next, as if chance kept score.

  • Judging a claim true or false by where it came from instead of by the evidence for it.

    • Rejecting an argument by attacking the person who made it instead of their reasons.

      • Rejecting a person or an idea because of a disliked person or group it’s connected to, instead of on its own merits.

      • Discrediting someone in advance, so that anything they say afterward is dismissed before it’s heard.

      • Dismissing a claim or criticism because the person making it doesn’t act consistently with it.

  • Overwhelming an opponent with more weak or false points than they can answer in the time available, so that whatever goes unanswered looks conceded.

  • Letting one overall impression of someone color your judgment of their unrelated traits.

  • Concluding something about a whole group from cases that are too few, or too unlike the group, to support a claim that broad.

    • Treating a story or a few personal experiences as if they settled how often something happens or what causes it, often against systematic evidence.

  • Believing, once you know how something turned out, that the outcome was more predictable than it really was.

  • Expecting a run of successes to continue because the streak seems “hot”, more than the process behind it justifies.

  • A syllogism whose conclusion covers all of its predicate group, when the premise naming that group covered only part of it.

  • A syllogism whose conclusion covers all of its subject group, when the premise naming that group covered only part of it.

  • Believing your actions influence an outcome that is really down to chance, or that you control events more than you do.

  • Believing you understand how something works, such as a zipper or a toilet, in far more detail than you could actually explain.

  • Seeing a link between two things that isn’t in the data, or seeing a weak link as a strong one.

  • Rating a statement as more likely true because you’ve encountered it before, even though hearing it again adds no evidence.

  • Favoring members of your own group over outsiders, in how you judge them and what you give them, even when the group is arbitrary.

  • Suggesting a claim by asking whether it’s true, so the suggestion spreads without the speaker having to support it.

  • Treating someone’s denial of an accusation as evidence that the accusation is true, so that nothing they say could clear them.

  • Asking a question that takes for granted something not yet established, so that any direct answer seems to admit it.

  • Weighing a loss more heavily than a gain of the same size, so that a fair or even favorable bet feels like a bad deal.

  • Concluding that two things are different because you know or believe something about one that you don’t about the other.

  • Concluding that the truth must lie somewhere between two opposing positions simply because they are opposed.

  • The claim that doing something good makes people feel entitled to something bad afterward, which large replications have mostly not found.

  • Concluding that something is true because it ought to be, or false because it would be wrong or unfair if it were.

  • Advancing a bold claim, retreating to a modest one when challenged, then going back to the bold claim once the challenge has passed.

  • Demanding more or different evidence once the evidence you asked for has been supplied, so the question never counts as settled.

  • Concluding how things ought to be from facts about how they are, as if the facts alone settled the question.

  • Giving bad events, feedback or information more weight than good ones of the same size.

  • Saving a generalization about a group from a counterexample by redefining the group afterward, so the counterexample no longer counts.

  • A researcher’s expectations shape what they record or how subjects behave, so the study partly produces the result it was meant to test.

  • Judging whether a decision was good by how it happened to turn out, rather than by what was known and reasonable when it was made.

  • Being more sure your judgments are right than your track record justifies, such as giving “90% sure” ranges that catch the truth far less often.

  • Treating one contributing factor as the whole explanation for an outcome that took several factors acting together.

  • Trying different ways of collecting or analyzing data until a result comes out statistically significant, then reporting that version as the finding.

  • Redefining a word with strong positive or negative force so that it covers what you want praised or condemned, while keeping the force.

  • Forecasting how long a task will take, or what it will cost, from the plan for it rather than from how similar tasks actually went.

  • Concluding that one thing caused another only because it happened first.

  • Giving extra weight to whatever is immediate, so that plans made for later are reversed when later becomes now.

  • Repeating a claim, often many times, in place of giving reasons or evidence for it.

  • Treating the chance that an innocent person would match the evidence as if it were the chance that a person who matches is innocent.

  • Studies with positive or statistically significant results are more likely to be published, so the published record makes effects look stronger than they are.

  • Presenting someone’s words without the surrounding context, so they appear to say something they didn’t mean.

  • When groups in a study remember their past with different accuracy, so a difference in what they report reflects memory, not what actually happened.

  • Answering a claim by raising a different issue, as if discussing that settled the original question.

  • Giving the credit or blame for a change to something that happened in between, when an unusual result was simply returning toward average.

  • Treating an abstraction, like “nature” or “the market”, as a real thing that wants, decides or acts, and reasoning from that.

  • Dismissing a problem or complaint because something else is worse, as if a bigger problem made the smaller one unreal or not worth fixing.

  • Valuing a good outcome about the same whether it is small or many times larger, as if the number involved barely mattered.

  • A distortion that arises when the cases studied differ systematically from the ones a conclusion is about, because of how they got into the data.

    • A distortion that arises when the people who drop out of a study differ from those who stay, in ways related to the result.

    • Drawing conclusions from the cases that made it through some filter, while the ones that didn’t are missing from view.

  • Crediting your successes to yourself and blaming your failures on circumstances, luck or other people.

  • Reading a word in an old text with the meaning it has today, when it meant something else at the time the text was written.

  • Answering a request to support your claim by demanding that the other person prove it false.

  • A trend that holds in every subgroup reverses or vanishes when the groups are combined, because the groups differ in size and makeup.

  • Rejecting a first step because it will supposedly lead, one step after another, to disaster, without good reason to think each step will follow.

  • Holding others to a rule or standard while exempting yourself or a favored case, without a relevant difference that justifies the exception.

  • Sticking with the current state of affairs, or a past choice, because it’s current, when you’d pick something else starting fresh.

  • Refuting a distorted or weaker version of someone’s position instead of the position they actually hold.

  • Sticking with something because of what you’ve already spent on it, when only what happens next can make it worth continuing.

  • Presenting the evidence that supports a conclusion while leaving out known evidence that points the other way.

  • Explaining a natural feature or process by the purpose it seems to serve, as if the benefit itself had brought it about.

  • Treating a pattern in data as meaningful when the boundaries that make it stand out were drawn after seeing where the data fell.

  • Ending a question or a discussion with a stock phrase that sounds like an answer but gives no reason, such as “it is what it is”.

  • Defending a wrong action by pointing to someone else’s wrongdoing, as if the other wrong made this one acceptable.

  • Concluding two groups are linked because both belong to a third group, when neither premise says anything about all of that third group.

  • Expecting your own future to turn out better than the evidence supports, or better than other people’s.

  • Concluding that two things alike in some ways must be alike in another, when the similarities don’t bear on that way or a relevant difference undoes them.