Learn area
How Science Works
How claims become testable, how evidence is measured and compared, and how uncertainty, bias and synthesis constrain scientific conclusions.
Explore connections →How Science Works is the part of Learn that explains how empirical claims are tested and how evidence is interpreted: from defining a measurable proposition to causal inference, replication, risk assessment and evidence synthesis.
It is distinct from Reasoning Errors, which describe recurring failures in inference, and from Misinformation Mechanics, which explain how information is manufactured, stripped of context, amplified or made resistant to correction.
Each item below is a reusable methodological concept with its own scope, research trail and links to related Learn Items and Reasoning Errors. The groups are for navigation only; they are not separate Learn categories or additional hierarchy.
Method library
14 How Science Works concepts
Five navigation groups, one Learn area. Each card below is a standalone reusable Learn Item.
How Science Works
Claims & measurement
Turning ideas into testable propositions and making sure measurements capture the thing actually being claimed.
Testability, Operationalization & Falsifiability
How a vague idea becomes an empirical claim by defining what is measured, under which conditions and what observations would count against it.
Learn itemBurden of Proof & Evidential Thresholds
Why the strength and specificity of a claim determine how much evidence is needed, and why lack of disproof is not the same as support.
Learn itemMeasurement, Validity & Reliability
How evidence depends on whether measurements consistently capture the construct, exposure or outcome the claim is actually about.
How Science Works
Study design & causal inference
Using controls, randomization and explicit causal assumptions to distinguish effects from alternative explanations.
Controls, Randomization & Blinding
How comparison groups, random allocation and blinding help separate an intervention effect from expectation, selection, time trends and other alternative explanations.
Learn itemCorrelation, Confounding & Causal Inference
How scientists distinguish association from causation by using design, timing, comparison and explicit assumptions about alternative explanations.
How Science Works
Magnitude, uncertainty & applicability
Interpreting effect size, statistical uncertainty, dose, directness and generalization to the real-world claim.
Effect Size, Statistical Uncertainty & Significance
Why a p-value is not an effect size, why confidence intervals matter and why statistical detectability is different from practical importance.
Learn itemMechanism, Directness, Dose & Generalization
How mechanistic plausibility, relevant dose, direct outcomes and population similarity determine whether evidence transfers to the real-world claim.
How Science Works
Replication, synthesis & certainty
Testing whether findings survive new data and combining evidence without hiding bias, inconsistency or imprecision.
Replication, Reproducibility & Robustness
How repeated studies, re-analysis and alternative reasonable specifications reveal whether a result survives beyond one dataset or analysis.
Learn itemSystematic Reviews, Meta-analysis & Evidence Synthesis
How systematic methods combine multiple studies, why meta-analysis is not just citation counting and how synthesis inherits the strengths and weaknesses of included evidence.
Learn itemEvidence Quality, Risk of Bias & Certainty
How study limitations, inconsistency, indirectness, imprecision and missing evidence determine how confident we should be in a body of research.
How Science Works
Risk, models & scientific judgment
Interpreting hazard and risk, model assumptions, consensus and what different source types can actually establish.
Hazard, Risk, Baseline Risk & Absolute Risk
How the ability to cause harm differs from the probability of harm at a real exposure, and why relative changes need baseline risk to be interpretable.
Learn itemModels, Assumptions & Sensitivity
How scientific models simplify reality, why their outputs depend on assumptions and how sensitivity analysis tests which inputs drive conclusions.
Learn itemScientific Consensus, Disagreement & Updating
How expert convergence can summarize a mature evidence base without becoming an oracle, and why disagreement, uncertainty and revision remain part of science.
Learn itemWhat Different Source Types Can Establish
How study design and source type constrain what a document can support, from primary data and randomized trials to reviews, guidelines, patents and news reports.