Philosophy of Science

1. ObservationLockedLooking at something carefully and on purpose, and writing down what you see at the time - kept apart from your ideas about why.Opens after: Perception
2. HypothesisLockedA possible answer to a question about the world, which the world could show to be right or wrong. A starting point, not an answer.Opens after: Observation
3. PredictionLockedWhat you should see if a hypothesis is true, said before you look. To choose between two hypotheses, find a situation where they predict different things.Opens after: Hypothesis, Conditional statements
4. Scientific explanationLockedAn account of how or why something happens. A good one explains many different things at once - puddles, washing, sea salt - and makes predictions you can test.Opens after: Hypothesis
5. FalsifiabilityLockedA claim is falsifiable if some result could show it to be wrong. A claim that fits whatever happens tells you nothing; a claim that takes a risk and passes counts.Opens after: Prediction
6. ModelLockedA simpler version of something real, which keeps what matters for a purpose and leaves the rest out on purpose. Never exact, but a good one is useful for what it was made for.Opens after: Scientific explanation
7. Scientific theoryLockedIn science, a theory is a well-tested explanation of many facts, supported by many different kinds of evidence - not a hunch, and it doesn't "become a fact".Opens after: Scientific explanation, Strong and weak evidence
8. Testing against the worldLockedA claim about the world is settled by the world: an experiment sets up a situation on purpose to see whether a prediction comes true, and the result decides, not who made the claim.Opens after: Prediction, Evidence
9. The simplest explanationLockedWhen two explanations fit the facts, start with the one that needs fewer things to be true (Occam's razor). It is the best place to start, not a proof: test it.Opens after: Scientific explanation
10. Blind testLockedA test where the people involved don't know which is which, so what they expect can't change the result - as Clever Hans showed, knowing can change things even when nobody cheats.Opens after: Testing against the world, Confirmation bias
11. Fair testLockedA test where only one thing is changed and everything else is kept the same, so any difference in the result comes from that one thing.Opens after: Testing against the world
12. Predicting something newLockedThe strongest test of a theory is a prediction of something nobody has seen yet, which could easily have failed - as when Newton's theory told Le Verrier where to find Neptune in 1846.Opens after: Scientific theory, Falsifiability
13. Rescuing an idea with excusesLockedSaving an idea from every failed test with a new excuse that can't itself be checked. The idea ends up saying nothing; a fair change makes a new prediction you can test.Opens after: Falsifiability
14. What science can't answerLockedScience answers questions about what is and what would happen. Questions about what we should do also need values - what we care about. The facts make the choice better, but don't make it.Opens after: Testing against the world
15. Control groupLockedThe group in a test that doesn't get the thing being tested, to show what would have happened anyway. "It got better" isn't enough: compared with what?Opens after: Fair test
16. Sample sizeLockedHow many things or people a test is done on. One try can be luck; with many, lucky and unlucky results even out and the real difference, if any, shows through.Opens after: Fair test, Hasty generalisation
17. Random assignmentLockedUsing chance, like a coin toss, to decide who goes in which group. It makes the groups alike in every way - even ways nobody thought of - except the thing being tested.Opens after: Control group, Cause or coincidence
18. ReplicationLockedOther people doing the same test again, in the same way, to see if they get the same result. A finding earns trust when it is replicated, and loses it when it isn't, like cold fusion in 1989.Opens after: Sample size
19. Peer reviewLockedOther experts checking a piece of science before it is published. It catches many mistakes, but not all - so replication is still needed.Opens after: Replication, Expertise
20. Science corrects itselfLockedScientists make mistakes like everyone; science catches them - by checking, peer review, replication and asking big evidence for big claims - and puts them right openly.Opens after: Replication, Big claims need big evidence
21. Changing a big ideaLockedScience can change even its biggest ideas, as when the Sun replaced the Earth at the centre - slowly, over more than a century, as the evidence became too strong to ignore.Opens after: Science corrects itself, Updating beliefs
22. Scientific consensusLockedNearly all the experts in a field agreeing, because many independent kinds of evidence point the same way. It isn't a vote and it can change, but it is the best bet for someone who can't check.Opens after: Peer review, Independent sources
23. Refining a theoryLockedA new theory usually keeps what the old one got right and fixes where it went wrong: Newton's gravity still sends rockets to the Moon, and Einstein's explains what Newton's couldn't.Opens after: Changing a big idea, Predicting something new
24. Thinking like a scientistLockedThe habits of finding out: be curious and look carefully, guess in ways you can test, test fairly, share your work, change your mind - and don't fool yourself, the easiest person to fool.Opens after: Refining a theory, What science can't answer, Intellectual humility

Connections to other domains

• Scientific consensus needs Independent sourcesEpistemology
• Scientific theory needs Strong and weak evidenceEpistemology
• Observation needs PerceptionEpistemology
• Blind test needs Confirmation biasEpistemology
• Changing a big idea needs Updating beliefsEpistemology
• Testing against the world needs EvidenceEpistemology
• Science corrects itself needs Big claims need big evidenceEpistemology
• Peer review needs ExpertiseEpistemology
• Prediction needs Conditional statementsLogic
• Random assignment needs Cause or coincidenceLogic
• Sample size needs Hasty generalisationLogic
• Thinking like a scientist needs Intellectual humilityEpistemology