Frame the question
A research unit begins by stating what is being estimated, observed or tested, what is outside scope, and which claims would be justified by the resulting evidence.
Methods
The method is part of the evidence.
IAI treats experimental design, provenance, validation and claim boundaries as part of the research itself. A result is more useful when the conditions under which it was produced remain visible.
Research sequence
From question to defensible claim
Not every study uses an identical protocol, but the underlying discipline is consistent: define the question, control what matters, preserve evidence and keep the claim proportional to what was actually observed.
-
01 -
02 Form the method prospectively
Where the problem permits it, experimental controls, comparison structure, sample size, stopping rules, exclusions and analysis plans are formed before execution rather than tuned around the result.
-
03 Make the boundary observable
Important experimental transitions should be measured rather than assumed. Logical reset, process termination or a clean-looking environment are not automatically treated as evidence of independence.
-
04 Preserve the evidence
Execution records, instrumentation, artefacts and relevant intermediate outputs are preserved with enough provenance to support later inspection and reconstruction.
-
05 Validate the result
Validation checks whether the claimed result is supported by the produced evidence and whether the artefacts, reports and validators themselves behave as intended.
-
06 Separate finding from decision
Evidence may support a finding without automatically authorising adoption, publication or institutional change. Scientific result and institutional decision remain distinct.
Epistemic separation
Not every statement has the same authority.
IAI separates what was observed from what was inferred, recommended or formally decided. This prevents a measurement from silently becoming a conclusion, or a conclusion from silently becoming institutional authority.
- 01 Observation
- What was directly measured, recorded or produced.
- 02 Finding
- What the evidence supports under the stated method and assumptions.
- 03 Interpretation
- How the finding is understood in the wider research context.
- 04 Recommendation
- What course of action the evidence suggests.
- 05 Decision
- What the authorised institutional process actually approves.
Reproducibility
The research record should survive the telling of the story.
- 01
Authoritative execution is separated from narrative explanation.
- 02
Machine-readable state is validated against explicit schemas or contracts where appropriate.
- 03
Rendered reports remain projections rather than new sources of truth.
- 04
Tool and runtime choices are recorded so that the method can be reconstructed.
- 05
Unexpected failures and failed instruments are retained when they are scientifically relevant.
Uncertainty
Uncertainty is recorded, not edited out.
Negative results, failed instruments, unresolved mechanisms, conditional conclusions and limits on generalisation can all be scientifically useful. Preserving them helps distinguish what the research established from what still requires investigation.
Governance boundary
Method supports authority; it does not replace it.
Validation, reproducibility and evidential strength can support an institutional decision, but they do not make that decision automatically. Where formal authority is required, the authorised governance process remains explicit.