alternative-hypothesis-check

スター11

フォーク1

更新日2026年2月22日 19:50

Use when analyzing results in Type D or Type H projects before making mechanistic or causal claims — requires systematic exclusion of confounders, batch effects, technical noise, and other alternative explanations

インストール

Codex または Claude でインストールこの Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。

ソース

EvoClaw

EvoClaw/amplify

GitHub リポジトリを開く Creator のリポジトリを見る

ダウンロード

Alternative Hypothesis Check

Discipline layer skill. Active during Phase 4–5 for Type D and Type H projects.

Iron Law

NO MECHANISM CLAIMS WITHOUT EXCLUDING ALTERNATIVE EXPLANATIONS

Systematic Check

For each major finding, systematically check:

□ Confounders — variables that could explain the association independently of your proposed mechanism
□ Batch effects [bioinformatics] — technical variation masquerading as biology (plate, lane, date, operator)
□ Technical noise — measurement artifacts, instrument drift, reagent lot differences
□ Sample selection bias — non-random inclusion criteria that correlate with the outcome
□ Multiple testing inflation — testing many hypotheses means some will be significant by chance; confirm FDR/Bonferroni was applied
□ Correlation ≠ causation — temporal precedence, dose-response, and intervention evidence must be present before causal language is used

When an Alternative Explanation Cannot Be Excluded

If an alternative explanation cannot be excluded:

→ It must be stated as a limitation in the paper → Language must be weakened ("associated with" not "causes") → It cannot be presented as a mechanistic finding

Checklist Format

For each finding, produce a table:

Finding	Alternative checked	Excluded?	Evidence
...	Confounder X	Yes / No	...

Rationalization Firewall

You will hear yourself say...	Reality check
"The effect is too strong to be confounding"	Strong effects can still be confounded. Check anyway.
"We controlled for the obvious variables"	Obvious to you. What about reviewer #2's favorite confounder?
"Correlation is enough for this paper"	Correlation is enough only if you DON'T claim mechanism.
"Batch correction was applied"	Did you verify it worked? Show before/after.

このリポジトリの他の Skills

同じリポジトリ

evaluation-protocol-design

EvoClaw/amplify

Use when designing the evaluation protocol for a method development project — locks primary metrics, datasets, seeds, statistical tests, and baseline list into an immutable contract

2026-02-2411

experiment-execution

EvoClaw/amplify

Use when G2 (plan freeze) and G3 (execution readiness) gates have passed — handles all implementation, baseline reproduction, experimentation, and iteration for both Type M and Type D projects; this is Phase 4

2026-02-2411

method-framework-design

EvoClaw/amplify

Use when research direction is validated (G1 passed) and the project needs method design, analysis framework, evaluation protocol, or story line — this is Phase 3 and must complete before any code or experiments

2026-02-2411

multi-round-deliberation

EvoClaw/amplify

Reusable protocol for multi-agent discussions that iterate until convergence. Used by results-integration (Phase 5) and paper-writing (Phase 6) to ensure discussions actually resolve issues rather than just noting them.

2026-02-2411

paper-writing

EvoClaw/amplify

Use when and ONLY when the user explicitly requests paper writing and G4 (write-ready) gate has passed — handles LaTeX structure, senior-level writing, reference verification, and iterative refinement

2026-02-2411

problem-validation

EvoClaw/amplify

Use when Phase 1 (research-direction-exploration) is complete and G1 gate has passed — subjects the research problem to adversarial defense-style questioning, classifies project intent, re-evaluates venue target, and produces a final feasibility ruling before any method design begins

2026-02-2411

Alternative Hypothesis Check

Discipline layer skill. Active during Phase 4–5 for Type D and Type H projects.

Iron Law

NO MECHANISM CLAIMS WITHOUT EXCLUDING ALTERNATIVE EXPLANATIONS

Systematic Check

For each major finding, systematically check:

□ Confounders — variables that could explain the association independently of your proposed mechanism

□ Batch effects [bioinformatics] — technical variation masquerading as biology (plate, lane, date, operator)

□ Technical noise — measurement artifacts, instrument drift, reagent lot differences

□ Sample selection bias — non-random inclusion criteria that correlate with the outcome

□ Multiple testing inflation — testing many hypotheses means some will be significant by chance; confirm FDR/Bonferroni was applied

□ Correlation ≠ causation — temporal precedence, dose-response, and intervention evidence must be present before causal language is used

When an Alternative Explanation Cannot Be Excluded

If an alternative explanation cannot be excluded:

→ It must be stated as a limitation in the paper → Language must be weakened ("associated with" not "causes") → It cannot be presented as a mechanistic finding

Checklist Format

For each finding, produce a table:

Finding

Alternative checked

Excluded?

Evidence

...

Confounder X

Yes / No

...

Rationalization Firewall

You will hear yourself say...

Reality check

"The effect is too strong to be confounding"

Strong effects can still be confounded. Check anyway.

"We controlled for the obvious variables"

Obvious to you. What about reviewer #2's favorite confounder?

"Correlation is enough for this paper"

Correlation is enough only if you DON'T claim mechanism.

"Batch correction was applied"

Did you verify it worked? Show before/after.

name	alternative-hypothesis-check
description	Use when analyzing results in Type D or Type H projects before making mechanistic or causal claims — requires systematic exclusion of confounders, batch effects, technical noise, and other alternative explanations