The evidence establishes
Use when the named source and calculation directly support the descriptive claim for the defined population and period.
Research protocol · Version 1.1
Research Method
How the Research Desk moves from a public question to a claim that readers can inspect, reproduce, challenge, and correct.
Direct answer
What makes a Research Desk claim reliable?
A claim is reliable only to the extent that readers can identify the population, measure, period, source, calculation, uncertainty, limitations, authorship, and correction history behind it.
The method does not treat a citation as proof by itself. It asks whether the cited evidence actually measures the claim being made.
The governing rule
Strength of language must never outrun strength of evidence.
A useful brief does more than cite a study. It identifies the exact population, measure, period, and uncertainty behind every central claim.
When evidence is mixed, incompatible, or incomplete, the conclusion must say so plainly. “We do not know” is a valid result.
Seven-step protocol
From question to public record.
- 01
Define the question
Name the population, geography, period, and outcome before searching for evidence.
- 02
Build the source set
Prefer primary government data and original peer-reviewed research. Record contrary findings, not only supportive ones.
- 03
Audit compatibility
Check whether race categories, age universes, survey designs, and measures can be compared across sources and years.
- 04
Show the calculation
Publish numerator, denominator, formula, rounding, uncertainty, and a downloadable table whenever feasible.
- 05
Separate result from interpretation
Describe the pattern first. Do not turn correlation into causation or a group average into a judgment about individuals.
- 06
Name the limits
State missing variables, measurement error, small samples, definition changes, and alternative explanations.
- 07
Version the record
Date every release and document material corrections without silently rewriting the historical record.
Claim language
Three conclusions, each with a different burden.
The evidence suggests
Use when the pattern is credible but the period, sample, compatibility, or causal explanation remains incomplete.
The evidence cannot determine
Use when the available source does not measure the question or cannot distinguish competing explanations.
Source hierarchy
Not every citation carries equal weight.
Primary statistical source
Government datasets, survey microdata, administrative records, or original data collected under a documented method.
Original scholarly research
Peer-reviewed articles, working papers, and technical reports with visible methods and data.
Authoritative synthesis
Systematic reviews, government research summaries, and scholarly books that accurately synthesize primary work.
Context only
News reports, organizational explainers, and commentary. These may frame a question but should not anchor a statistical claim.
Review status
Published for inspection. Not disguised as peer review.
Research Desk briefs are authored and editorially reviewed by Tyler Burns unless another contributor is named. They are not independently peer reviewed unless a brief later names the reviewers, scope, date, and outcome of that review.
The site’s .org address, academic citations, and links to government agencies do not confer institutional endorsement.
Submit a documented correctionCorrection architecture
What happens when a brief is wrong?
- Minor correction: typographical or link repair; logged with the modification date.
- Material correction: changed number, method, or interpretation; version increment and visible note.
- Retraction: the original conclusion cannot be supported; the page remains with a retraction notice and explanation.