Research question, audience and evidence boundary
What makes public legal statistics comparable across offices and years? The central analytical task is definitions, reporting coverage, denominators and selection bias. The page is intended for researchers, inventors, analysts, university teams and decision-makers who need to examine underlying records rather than accept search-result summaries as conclusive findings. Its examples illustrate a possible workflow; they are not reports of a completed engagement, a claim of proprietary data access or proof of results for a specific client.
Research-specific verification matrix
| Verification phase | Evidence operation | Reviewer challenge |
|---|---|---|
| 1. Definitions | Name the measured population and numerator: filings, judgments, grants, cases disposed, rights in force or persons served. | Absence of a reported value need not mean zero filings or cases. |
| 2. Definitions | Capture the statistical agency’s glossary, revisions, data-source description and release frequency before creating graphics. | Changing court processes or classification definitions can produce artificial trends. |
| 3. Definitions | Distinguish statistics by office from those by applicant origin and identify regional-office equivalent counting. | Differences in office workloads do not alone establish system efficiency. |
The primary research object is not a number but a defined statistical series. An office-of-filing measure is different from an applicant-origin measure. Changes in reporting coverage can masquerade as increases or decreases in activity; readers need a disclosure of offices missing from each reporting year.
Worked example and the unit of analysis
Primary reference for checking this procedure: WIPO statistical data descriptions. This source supports background verification; exact proposition-to-passage review remains necessary.
A presentation treats a rise in patent filings as evidence that the quality of innovation improved. The evidence may only demonstrate changed filing activity, with no direct observation of innovation quality.
A comparison table should preserve the numerator, denominator, units, agency, reporting vintage and any revision history. Before drawing a trend line, test whether a classification rule changed within the period. If the aim is to discuss productivity, distinguish workloads, outcomes and duration rather than infer administrative efficiency from volume alone.
For the illustrative scenario above, the evidence table should have explicit columns for the source identifier, the particular proposition tested, observed record content, competing explanation and the decision that remains open. The proposed workflow is not a completed case study.
Five-stage evidence workflow
1. Define and delimit
Name the measured population and numerator: filings, judgments, grants, cases disposed, rights in force or persons served. Before searching, record the questions that are intentionally outside scope. An excluded jurisdiction or period might be perfectly reasonable for an exploratory study yet unacceptable for a legal clearance decision. The brief should make that difference legible to someone who has not attended the initial discussion.
2. Collect and preserve
Capture the statistical agency’s glossary, revisions, data-source description and release frequency before creating graphics. Retain raw record identifiers and original language next to normalized fields. Translations, vendor exports and interface-specific labels simplify review but can erase distinctions. Preserving the unmodified provenance means an analyst can return to the material even after an export format or search interface changes.
3. Reconcile and test
Distinguish statistics by office from those by applicant origin and identify regional-office equivalent counting. Test a small counterexample set: a record that should be included, one that should be excluded, and an ambiguous borderline record. This gives reviewers something concrete against which to challenge the grouping rule, rather than assuming the procedure worked because a chart appeared plausible.
4. Interpret with competing explanations
Test missing values, unreported offices, retroactive revisions, inflation and population denominators where meaningful. Write down at least two explanations for the important pattern and specify which additional observation would discriminate between them. Different publication practice, database refresh and real changes in technical activity can generate superficially similar curves; the report should not flatten them into one story.
5. Report and escalate
Label descriptive trends separately from causal explanations, forecasts and legal conclusions. Separate verified facts, analytical judgments and unresolved questions in the finished output. Include the conditions under which a result ought to be refreshed, and state whether the next step is deeper technical research, office-register validation or professional legal review.
Official sources and what they establish
WIPO statistical data descriptions. Consult this source for the relevant published method or official record and check the document version and date. The example here is illustrative, not a claim that this page reports a completed search.
WIPO global IP statistics overview. Consult this source for the relevant published method or official record and check the document version and date. The example here is illustrative, not a claim that this page reports a completed search.
For this question, preserve the distinction between documentation about a data system and the underlying record returned by that system. Include the exact version, issue date or identifier needed to validate consequential findings.
Limitations, contrary evidence and professional review
Absence of a reported value need not mean zero filings or cases. The finding should be qualified against the relevant source and decision rather than generalized across territories or dates.
Changing court processes or classification definitions can produce artificial trends. The finding should be qualified against the relevant source and decision rather than generalized across territories or dates.
Differences in office workloads do not alone establish system efficiency. The finding should be qualified against the relevant source and decision rather than generalized across territories or dates.
This public guide addresses the research process rather than a particular client matter. Refer any disputed statutory interpretation, enforceability or professional clearance issue to the responsible authority or qualified adviser.
Useful output and related research
An illustrative research output is a harmonized statistical comparison with disclosure of denominator choice, missingness and vintage. The file should define its scope, explain material exclusions and give readers identifiers sufficient to reproduce the analysis. It should not imply an engagement has already occurred.
Independent review should revisit the source material most likely to change the result. For this topic, challenge a positive example, a near-boundary example and an omitted record. Document which conclusions survive the challenge and which require correction.
For a scoped research inquiry, describe the technical question, source universe, applicable jurisdiction, deadline assumptions and expected analytical output. The discussion is informational and does not promise a regulated opinion, patent grant, legal clearance or market outcome.
For more context: Parent research collection · Methodology library · Source library · Research inquiry





