State where the claim applies.
Population, setting, time and measurement boundaries prevent accidental overgeneralization.
Side 123
Scientific Writing
Research communication treated as part of reasoning: making claims precise, exposing the evidence chain, preserving uncertainty and designing prose so another reader can reconstruct the argument.
question→claim→evidence→structure→revision
04lenses
16working concepts
V0content
SS-1.0standard
A scientific sentence should reveal what kind of claim it makes.
Description, comparison, association, causal explanation and generalization require different evidence and wording.
Match language to evidence.
Words such as causes, suggests, predicts and is associated with are not interchangeable.
Use magnitude, not significance alone.
Effect size and uncertainty often matter more than a binary threshold.
Preserve what remains unresolved.
Qualification is not weakness when the evidence genuinely leaves alternatives open.
Sections should follow the logic of the investigation.
A paper is easier to evaluate when each part answers a different question in the evidence chain.
What is not yet known or decided?
The introduction should establish the gap without inflating novelty.
What was done and why?
Methods should contain enough operational detail for evaluation and, where appropriate, reproduction.
What was observed?
Results separate measured outcomes from interpretive claims as much as the design permits.
What does the result warrant?
Interpretation reconnects findings to the question while testing limitations and alternatives.
Sources should be attached to the claims they actually support.
Citation count is weaker than citation fit.
Use original studies for specific empirical claims.
Reviews are useful maps but can hide design details that matter to interpretation.
Use reviews and meta-analyses for broader patterns.
Synthesis still depends on inclusion criteria, heterogeneity and publication processes.
Separate another author's claim from your inference.
Citation should not blur evidence, interpretation and speculation.
Let displays carry structured evidence.
Figures should make the relevant comparison legible without requiring decorative complexity.
Revision is an analytical pass, not just copyediting.
The strongest revisions test reasoning, compression and reproducibility before polishing style.
Trace every conclusion backward to evidence.
Unsupported bridges become visible when each inferential step is named.
Remove words that do not change meaning.
Shorter is useful only if scope and qualifications survive.
Assume an intelligent reader without your context.
Define terms where ambiguity would alter interpretation.
Check whether another person could reconstruct the work.
Missing methods, transformations or exclusions are writing failures with scientific consequences.
Clear writing is not decoration added after the science. Writing exposes missing logic, vague scope and unsupported transitions because every sentence must specify what is being claimed and why.