スキル一覧に戻る

review-research

pjt222
更新日 2 days ago
5 閲覧
17
2
17
GitHubで表示
デザインaidesign

について

`review-research`スキルは、Claudeが研究資料に対して構造化された査読を実施し、方法論、統計手法の適切性、再現性を評価することを可能にします。このスキルは、論文原稿、プレプリント、研究プロトコルの査読、あるいは主張の根拠となる証拠の評価を目的として設計されています。主な機能には、バイアスの特定、実験デザインや論文の品質に関する建設的なフィードバックの提供が含まれます。

クイックインストール

Claude Code

推奨
メイン
npx skills add pjt222/agent-almanac -a claude-code
プラグインコマンド代替
/plugin add https://github.com/pjt222/agent-almanac
Git クローン代替
git clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/review-research

このコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします

ドキュメント

Review Research

Perform structured peer review of research work. Evaluate methodology, statistical choices, reproducibility, overall scientific rigour.

When Use

  • Reviewing manuscript, preprint, or internal research report
  • Evaluating research proposal or study protocol
  • Assessing quality of evidence behind a claim or recommendation
  • Providing feedback on colleague research design before data collection
  • Reviewing thesis chapter or dissertation section

Inputs

  • Required: Research document (manuscript, report, proposal, or protocol)
  • Required: Field/discipline context (affects methodology standards)
  • Optional: Journal or venue guidelines (if reviewing for publication)
  • Optional: Supplementary materials (data, code, appendices)
  • Optional: Prior reviewer comments (if reviewing a revision)

Steps

Step 1: First Pass — Scope and Structure

Read entire document once to understand:

  1. Research question: Clearly stated and specific?
  2. Contribution claim: What is novel or new?
  3. Overall structure: Does it follow expected format (IMRaD, or venue-specific)?
  4. Scope match: Work appropriate for target audience/venue?
## First Pass Assessment
- **Research question**: [Clear / Vague / Missing]
- **Novelty claim**: [Stated and supported / Overstated / Unclear]
- **Structure**: [Complete / Missing sections: ___]
- **Scope fit**: [Appropriate / Marginal / Not appropriate]
- **Recommendation after first pass**: [Continue review / Major concerns to flag early]

Got: Clear understanding of paper claims and contribution. If fail: Research question unclear after full read? Note this as major concern and proceed.

Step 2: Evaluate Methodology

Assess research design against standards for the field:

Quantitative Research

  • Study design appropriate for research question (experimental, quasi-experimental, observational, survey)
  • Sample size justified (power analysis or practical rationale)
  • Sampling method described and appropriate (random, stratified, convenience)
  • Variables clearly defined (independent, dependent, control, confounding)
  • Measurement instruments validated and reliability reported
  • Data collection procedure reproducible from description
  • Ethical considerations addressed (IRB/ethics approval, consent)

Qualitative Research

  • Methodology explicit (grounded theory, phenomenology, case study, ethnography)
  • Participant selection criteria and saturation discussed
  • Data collection methods described (interviews, observations, documents)
  • Researcher positionality acknowledged
  • Trustworthiness strategies reported (triangulation, member checking, audit trail)
  • Ethical considerations addressed

Mixed Methods

  • Rationale for mixed design explained
  • Integration strategy described (convergent, explanatory sequential, exploratory sequential)
  • Both quantitative and qualitative components meet their respective standards

Got: Methodology checklist completed with specific observations for each item. If fail: Critical methodology information missing? Flag as major concern rather than assume.

Step 3: Assess Statistical and Analytical Choices

  • Statistical methods appropriate for data type and research question
  • Assumptions of statistical tests checked and reported (normality, homoscedasticity, independence)
  • Effect sizes reported alongside p-values
  • Confidence intervals provided where appropriate
  • Multiple comparison corrections applied when needed (Bonferroni, FDR, etc.)
  • Missing data handling described and appropriate
  • Sensitivity analyses conducted for key assumptions
  • Results interpretation consistent with analysis (not overstating findings)

Common statistical red flags:

  • p-hacking indicators (many comparisons, selective reporting, "marginally significant")
  • Inappropriate tests (t-test on non-normal data without justification, parametric tests on ordinal data)
  • Confusing statistical significance with practical significance
  • No effect size reporting
  • Post-hoc hypotheses presented as a priori

Got: Statistical choices evaluated with specific concerns documented. If fail: Reviewer lacks expertise in specific method? Acknowledge this and recommend specialist reviewer.

Step 4: Evaluate Reproducibility

  • Data availability stated (open data, repository link, available on request)
  • Analysis code availability stated
  • Software versions and environments documented
  • Random seeds or reproducibility mechanisms described
  • Key parameters and hyperparameters reported
  • Computational environment described (hardware, OS, dependencies)

Reproducibility tiers:

TierDescriptionEvidence
GoldFully reproducibleOpen data + open code + containerized environment
SilverSubstantially reproducibleData available, analysis described in detail
BronzePotentially reproducibleMethods described but no data/code sharing
OpaqueNot reproducibleInsufficient method detail or proprietary data

Got: Reproducibility tier assigned with justification. If fail: Data cannot be shared (privacy, proprietary)? Synthetic data or detailed pseudocode acceptable alternative — note whether this provided.

Step 5: Identify Potential Biases

  • Selection bias: Were participants representative of target population?
  • Measurement bias: Could measurement process have systematically distorted results?
  • Reporting bias: Are all outcomes reported, including non-significant ones?
  • Confirmation bias: Did authors only look for evidence supporting their hypothesis?
  • Survivorship bias: Were dropouts, excluded data, or failed experiments accounted for?
  • Funding bias: Is funding source disclosed and could it influence findings?
  • Publication bias: Is this complete picture or might negative results be missing?

Got: Potential biases identified with specific examples from manuscript. If fail: Biases cannot be assessed from available information? Recommend authors address this explicit.

Step 6: Write the Review

Structure review constructively:

## Summary
[2-3 sentences summarizing the paper's contribution and your overall assessment]

## Major Concerns
[Issues that must be addressed before the work can be considered sound]

1. **[Concern title]**: [Specific description with reference to section/page/figure]
   - *Suggestion*: [How the authors might address this]

2. ...

## Minor Concerns
[Issues that improve quality but are not fundamental]

1. **[Concern title]**: [Specific description]
   - *Suggestion*: [Recommended change]

## Questions for the Authors
[Clarifications needed to complete the evaluation]

1. ...

## Positive Observations
[Specific strengths worth acknowledging]

1. ...

## Recommendation
[Accept / Minor revision / Major revision / Reject]
[Brief rationale for the recommendation]

Got: Review specific, constructive, references exact locations in manuscript. If fail: Review running long? Prioritize major concerns and note minor issues in summary list.

Checks

  • Every major concern references specific section, figure, or claim
  • Feedback constructive — problems paired with suggestions
  • Positive aspects acknowledged alongside concerns
  • Statistical assessment matches analysis methods used
  • Reproducibility explicit evaluated
  • Recommendation consistent with severity of concerns raised
  • Tone professional, respectful, collegial

Pitfalls

  • Vague criticism: "The methodology is weak" is unhelpful. Specify what is weak and why.
  • Demand a different study: Review research that was done, not research you would have done.
  • Ignore scope: Conference paper has different expectations than journal article.
  • Ad hominem: Review the work, not the authors. Never reference author identity.
  • Perfectionism: No study is perfect. Focus on concerns that would change conclusions.

See Also

  • review-data-analysis — deeper focus on data quality and model validation
  • format-apa-report — APA formatting standards for research reports
  • generate-statistical-tables — publication-ready statistical tables
  • validate-statistical-output — statistical output verification

GitHub リポジトリ

pjt222/agent-almanac
パス: i18n/caveman/skills/review-research
0
agentsagentskillsai-assisted-developmentclaude-codeskillsteams

関連スキル

executing-plans

デザイン

executing-plansスキルは、完全な実装計画があり、それを管理されたバッチでレビューチェックポイントを設けながら実行する場合に使用します。このスキルは計画を読み込んで批判的にレビューした後、小さなバッチ(デフォルトは3タスク)でタスクを実行し、各バッチの間に進捗状況を報告してアーキテクトのレビューを受けます。これにより、品質管理チェックポイントが組み込まれた体系的な実装が保証されます。

スキルを見る

requesting-code-review

デザイン

このスキルは、コードレビュアーサブエージェントを起動し、処理を進める前に要件に対してコード変更を分析します。タスク完了後、主要な機能の実装後、またはmainブランチへのマージ前などに使用すべきです。このレビューは、現在の実装と元の計画を比較することで、問題を早期に発見するのに役立ちます。

スキルを見る

connect-mcp-server

デザイン

このスキルは、開発者がHTTP、stdio、またはSSEトランスポートを使用してMCPサーバーをClaude Codeに接続するための包括的なガイドを提供します。GitHub、Notion、カスタムAPIなどの外部サービスを統合するためのインストール、設定、認証、セキュリティについて解説しています。MCP統合のセットアップ、外部ツールの設定、またはClaudeのModel Context Protocolを扱う際にご利用ください。

スキルを見る

web-cli-teleport

デザイン

このスキルは、タスク分析に基づいて開発者がClaude Code WebとCLIインターフェースの選択を支援し、これらの環境間でのシームレスなセッションテレポーテーションを可能にします。Web、CLI、モバイル環境を切り替える際のセッション状態とコンテキストを管理することで、ワークフローを最適化します。様々な段階で異なるツールを必要とする複雑なプロジェクトにご活用ください。

スキルを見る