Scientific method
scientific reasoning · philosophy of science
The set of procedures by which science produces and tests claims — and a subject of dispute, since no single account describes what scientists across fields actually do.
In practice
A team runs a test and gets a null result. Whether that refutes the idea or the instrumentation is exactly the problem no procedure settles, and it is why replication matters more than any single experiment.
The common mistake
Believing the textbook sequence describes practice. Working scientists rarely follow it, and philosophers have spent a century on why it does not hold together.
The scientific method is usually taught as a fixed sequence: observe, form a hypothesis, predict, test, revise. That description is useful for beginners and does not survive contact with how science has actually developed.
The standard account and its problems
Observation is not neutral. What you record depends on what you take to be relevant, which depends on the theory you hold. There is no stage of pure data collection before theorizing begins.
Confirmation is weak. Instances agreeing with a hypothesis do not establish it — Hume's problem of induction. Any finite body of evidence is compatible with more than one general claim.
Tests are never isolated. A prediction follows from the hypothesis plus assumptions about equipment, background conditions and auxiliary theories. A failure tells you something in that bundle is wrong, not which part.
The main proposals
Popper replaced confirmation with falsifiability: make bold conjectures, try hard to refute them, keep what survives, and never claim proof.
Kuhn argued that the description should follow the history, which shows long stable periods of work inside a framework, interrupted by revolutions — see paradigm shift.
Imre Lakatos proposed research programs with a protected core and a belt of adjustable assumptions, judged over time by whether they predict new phenomena or merely absorb failures.
Paul Feyerabend argued that no methodological rule has gone unbroken by a successful scientist, and that insisting on one would have blocked several major advances.
What is left
No single procedure. What the accounts share is a set of commitments that hold across fields: claims answer to evidence rather than to authority, methods are stated so others can check them, results have to survive people trying to break them, and any claim can in principle be revised.
That is weaker than a method and is what does the work. See first principles and second-order effects.
Concept web
Open the full webQuestions
What is the scientific method?
The procedures by which science produces and tests claims. The textbook sequence of observe, hypothesize, predict and test is a simplification that philosophers of science have substantially revised.
Why is there no single scientific method?
Because observation is theory-laden, confirmation cannot establish universal claims, and tests never isolate a single assumption. Different fields also proceed in genuinely different ways.
What do the accounts agree on?
That claims answer to evidence rather than authority, methods are stated so others can check them, results must survive attempts to break them, and any claim can in principle be revised.