Research skills
Move from a published claim to a tractable project
Research starts before the novel calculation. Learn to identify exactly what a source claims, expose its assumptions, reproduce a decisive result, compare disagreement fairly, and define a next question small enough to finish.
Five cumulative outputs
Choose the skill you need next
Each step produces something the next step can use. Familiarity with the topic is not a substitute for the actual source map, calculation, comparison, or brief.
- 01Trace
Find the claim behind the citation
Use it whenA sentence is repeated widely, but you do not yet know which source establishes which part.
You produceA precise claim, supporting passages, assumptions, and limits on what follows.
Open step - 02Classify
Separate scope from confidence
Use it whenWords such as established, excluded, promising, or open are doing too many jobs.
You produceA short explanation of what is established, under which assumptions, and what remains unresolved.
Open step - 03Reproduce
Test a result, not just a code path
Use it whenYou need to reconstruct a derivation or run and add a check with a different failure mode.
You produceA reproducible protocol, expected outputs, tolerances, and a validation check with a different failure mode.
Open step - 04Compare
Locate the real disagreement
Use it whenTwo results seem incompatible, a bound is null, or uncertainty language hides different objects.
You produceA matched comparison, uncertainty decomposition, and a test that could distinguish explanations.
Open step - 05Bound
Turn the remaining gap into a project
Use it whenA broad interest needs one answerable question, a deliverable, and a stopping rule.
You produceA one-page research proposal with the necessary background, method, possible difficulties, and next steps.
Open step
Four habits
Keep the conclusion no stronger than the evidence
- Match objects before results.Align observables, states, conventions, orders, data, and domains before calling two numbers inconsistent.
- Separate running from validation.A successful execution checks the environment; convergence, benchmarks, and independent methods test the scientific claim.
- State the limits of your conclusion.Identify a tempting stronger conclusion that the available results do not support.
- Give every project a stopping rule.Decide when to continue, revise the method, seek help, or stop before adding more background or computation.
Where these pages stop
Connect research skills to a scientific question
The examples in this sequence teach durable research methods. For dated literature assessments and open questions, use Research. A specialist pathway supplies the scientific background for the calculation itself.
A good first move
Take one sentence you rely on and trace it to its source
The result should identify what is claimed, where it is established, which assumptions it uses, and what the source does not establish.