Editorial Policy #
BehavioralStrategy.com is the living canon for Behavioral Strategy. Jason Hreha is accountable for its editorial position. Each public claim must match its evidence, retain its limits, and remain open to correction.
Evidence ladder #
The site classifies evidence by what it can support. A higher category does not remove the need to report population, context, method, and limits.
| Class | What it means | What it can support | What it cannot support by itself |
|---|---|---|---|
| Illustration | A constructed example, scenario, or calculation used to explain a concept. | Comprehension of a method or formula. | A claim that an event occurred or that the method caused an outcome. |
| Usability evidence | A person or group could understand, complete, or reuse an artifact under stated conditions. | A bounded claim about comprehension or use. | Effectiveness, adoption at scale, or improved outcomes. |
| Observational case | A documented event, association, or before-and-after pattern without a credible assigned comparison. | A descriptive or hypothesis-generating claim. | Causal attribution or a general effect size. |
| Comparative test | A test with a defined comparison, denominator, window, and assignment or counterfactual. | A context-bound comparative result when the design and analysis support it. | Automatic generalization beyond the tested population and setting. |
| Independent replication or evaluation | An outside party controls the method, analysis, and final account and tests or evaluates the claim. | The strongest available support within the replicated scope. | Universal effectiveness or claims outside the evaluated design. |
Usability and adoption show that people can understand or use an artifact. They do not prove that the discipline improves outcomes. Broad effectiveness claims require suitable comparative evidence or independent replication or evaluation.
Null, negative, contradictory, and favorable findings follow the same publication rules. A result is not removed because it weakens a preferred claim.
Claim and source rules #
The public Evidence Ledger is rendered only from /_data/evidence_ledger.yml. Each evidence ID must be unique and resolve to one canonical record. Public pages must not maintain a second claim table.
Quantitative claims must include or link to:
- the population and target behavior;
- the denominator, observation window, and assignment or comparison method;
- the measure and calculation used;
- a named calculation with its data source or a source receipt;
- a canonical Evidence Ledger ID when the claim is used across pages; and
- uncertainty, confounders, and limits that affect interpretation.
A number without adequate support must be removed or labeled as an illustration. A qualitative claim must state whether it is an interpretation, observation, working hypothesis, or independent assessment. Citations support only the claims that their methods and results address.
Corrections #
Reports of possible errors can arrive through the contact page. The review then follows these steps:
- Preserve the affected text and source state for review.
- Check the claim, evidence ID, cited source, public page, and generated or machine-readable copies.
- Contain a material unsupported claim without waiting for a routine release cycle.
- Correct, narrow, relabel, or remove the claim according to the available evidence.
- Update every controlled copy and rebuild generated derivatives.
- Add a dated entry to the Corrections log when the change affects meaning, evidence status, attribution, or a material conclusion.
- Record the affected URL, old claim, correction, reason, status, and any open follow-up.
Routine spelling or formatting changes do not need a public log entry. Corrections do not erase the repository history. If a deposited artifact is wrong, a corrected version is published rather than silently replacing the old version.
Funding and conflicts #
Material funding, employment, client, coauthor, business, family, and close personal relationships must be disclosed when they could affect the creation, selection, interpretation, or review of evidence. A paid review can improve a draft, but it does not become independent authority. An undisclosed or unresolved material conflict keeps an independence claim unverified.
Sponsorship, access, an invitation, or a commercial relationship does not guarantee coverage, favorable treatment, a link, or a positive conclusion. Editorial decisions follow the evidence and the canonical role of the page.
Client and private cases #
A client or private case requires permission for publication and a clear statement of who supplied the data and who controlled the final wording. Confidential, personal, or commercially sensitive information must not enter the public repository. De-identification does not make weak evidence stronger or make a controlled account independent.
Client cases must use the same evidence ladder as other cases. A testimonial, private compliment, selected success story, or supplied result does not establish effectiveness. If permission, provenance, denominators, source receipts, or material limits cannot be published, the claim must be narrowed, kept private, or excluded.
AI use #
AI tools may help with search, transcription, organization, checking, or draft revision. A named human remains responsible for every published statement. The editor must verify factual claims against accessible sources, inspect citations, and review generated text for unsupported inference, false precision, missing limits, plagiarism, and disclosure risk.
Do not enter confidential client data, personal inquiry data, credentials, API keys, or restricted source material into an unapproved AI system. AI output is not a source receipt, an independent evaluation, or evidence of effectiveness. Disclose material AI involvement when it affects the provenance or reproducibility of a public artifact.
Editorial independence from invitations and commercial interests #
The site may offer a restrained path for interviews, talks, research, institutional work, or other approved invitations. Invitation fit, status, revenue, or relationship value does not control doctrine, evidence classification, corrections, or publication decisions.
The site does not trade favorable coverage, founder attribution, citations, evidence status, or editorial changes for access, payment, backlinks, speaking opportunities, or advisory work. When an invitation or relationship is relevant to a public claim, the relationship is disclosed and the claim remains subject to normal evidence review.
Review and maintenance #
Major doctrine, founder, evidence, application, and specification changes require their named owner and the applicable authority review. External results remain outside the site’s control. Missing evidence stays unknown rather than becoming a favorable assumption.
This policy was materially revised on 2026-08-09. Future material revisions update the date and preserve the reason in repository history or the public corrections log.