e1

E1-Quantitative Analysis Guide with Code Generation & Sensitivity Analysis VS-Enhanced with Full 5-Phase process: Avoids obvious analyses, explores innovative methodologies Expanded to include qualitative analysis (thematic, grounded theory, content, narrative) Absorbed E4 (Analysis Code Generator) and E5 (Sensitivity Analysis - Primary Study) capabilities Use when: selecting statistical/qualitative methods, interpreting results, checking assumptions, generating code, sensitivity analysis Triggers: statistical analysis, ANOVA, regression, t-test, power analysis, assumption checking, effect size, thematic analysis, grounded theory, content analysis, narrative analysis, NVivo, ATLAS.ti, coding, qualitative data, R code, Python code, SPSS syntax, sensitivity analysis, robustness check

インストールへ移動

リポジトリ: brycewang-stanford/Auto-Empirical-Research-Skills
ソースの最終更新活動: 2026年4月3日 02:07
検出された SKILL.md の言語: 英語
スター: 3,291
フォーク: 432

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

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。

直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。

npx skills add https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills --skill e1

コマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。

ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

SKILL.md

ソースの指示 · 読み取り専用プレビュー

検索に戻るインストールに戻る

auto-empirical-research-skills

Route empirical-research requests through the Auto-Empirical Research Skills catalog when this whole repository is installed as one skill in Codex, CodeBuddy, Claude Code, or another IDE. Use to choose and load the right vendored AERS skill for causal inference, econometrics, replication, data acquisition, manuscript writing, peer review and referee responses, citation checking, de-AIGC editing, or full empirical-paper workflows without reading the entire repository at once.

2026-08-06

de-aigc-skills

中英双语学术降 AIGC / bilingual academic de-AIGC skill. Removes AI-generated writing signatures from empirical papers in economics, management, and the social sciences — in both English and Chinese. Covers Turnitin AI, GPTZero, Originality.ai on the English side and 知网 AMLC, 万方, 维普 on the Chinese side. Uses a six-step loop (intake → audit → claim-evidence check → differentiated rewrite → five-dimension self-score → cold-reader recheck) with two pattern libraries (22 English + 17 Chinese patterns), section-by-section strategies for empirical papers, and hard protections that keep every number, coefficient, and citation intact.

2026-08-04

kaggle-research

Use when a research task needs reproducible Kaggle discovery, metadata inspection, bounded public-data downloads, competition or kernel discovery, model discovery, or an explicitly approved Kaggle write/delete operation through the official CLI.

2026-07-31

このリポジトリのすべてを見る

E1-Quantitative Analysis Guide

Agent ID: E1 (formerly 10) Category: E - Publication & Communication (Analysis Methods) VS Level: Full (5-Phase) Tier: Flagship Icon: 📈📊

Overview

Comprehensive guide for both and analysis methods appropriate for research design and data characteristics. Applies to avoid monotonous analyses like "recommend t-test" or "just do thematic analysis," presenting methodological diversity optimized for research questions across paradigms.

E1-Quantitative Analysis Guide

Agent ID: E1 (formerly 10) Category: E - Publication & Communication (Analysis Methods) VS Level: Full (5-Phase) Tier: Flagship Icon: 📈📊

Overview

--- ## Qualitative Analysis Methods (NEW in v5.0) ### Thematic Analysis **Approach**: Braun & Clarke 6-Phase Framework ```yaml thematic_analysis: phases: phase_1_familiarization: activities: - "Read and re-read data" - "Note initial ideas" - "Immerse in content" output: "Familiarization notes" phase_2_coding: activities: - "Generate initial codes systematically" - "Code interesting features" - "Collate data relevant to each code" output: "Coded data extracts" tools: ["NVivo", "ATLAS.ti", "MAXQDA", "Dedoose"] phase_3_searching_themes: activities: - "Collate codes into potential themes" - "Gather data relevant to each theme" output: "List of candidate themes" phase_4_reviewing_themes: activities: - "Check themes work with coded extracts" - "Generate thematic map" output: "Refined themes and thematic map" phase_5_defining_naming: activities: - "Define and refine each theme" - "Generate clear definitions" - "Name themes" output: "Theme definitions and names" phase_6_writing: activities: - "Final analysis" - "Select vivid extracts" - "Relate to research question and literature" output: "Scholarly report" quality_criteria: - "Theoretical coherence" - "Richness of interpretation" - "Member checking (optional)" - "Audit trail" software_comparison: nvivo: strengths: ["Rich visualization", "Matrix coding", "Framework matrices"] best_for: "Large qualitative datasets" atlas_ti: strengths: ["Hermeneutic unit", "Network views", "Query tools"] best_for: "Grounded theory and complex theory building" maxqda: strengths: ["Mixed methods", "Visual tools", "TeamCloud"] best_for: "Mixed methods research" dedoose: strengths: ["Web-based", "Collaboration", "Mixed methods"] best_for: "Team-based coding"

bayesian_analysis: core_concept: "Update beliefs with data using Bayes' theorem" packages: r_packages: brms: description: "Bayesian Regression Models using Stan" strengths: ["Flexible syntax", "Multilevel models", "Great documentation"] example: | library(brms) fit <- brm(y ~ x + (1|group), data = data, family = gaussian(), prior = c(prior(normal(0, 10), class = b))) rstanarm: description: "Applied Regression Modeling via Stan" strengths: ["Easy syntax", "Pre-compiled models", "Fast"] python_packages: pymc: description: "Probabilistic programming in Python" strengths: ["Flexible", "Large community", "Integration with ArviZ"] example: | import pymc as pm with pm.Model() as model: beta = pm.Normal('beta', mu=0, sigma=10) sigma = pm.HalfNormal('sigma', sigma=1) y_obs = pm.Normal('y_obs', mu=beta*x, sigma=sigma, observed=y) trace = pm.sample(2000) use_cases: prior_incorporation: description: "Incorporate existing knowledge as priors" example: "Meta-analysis results as priors for new study" small_samples: description: "Better uncertainty quantification with limited data" advantage: "Regularization prevents overfitting" complex_hierarchical: description: "Natural fit for multilevel/hierarchical models" advantage: "Partial pooling and shrinkage" advantages: - "Quantifies uncertainty via posterior distributions" - "Incorporates prior knowledge formally" - "No p-values or significance testing" - "Intuitive probability statements (e.g., '95% probability effect > 0')" reporting: elements: - "Prior specification and justification" - "Posterior distributions (median, 95% credible intervals)" - "Convergence diagnostics (Rhat, ESS)" - "Posterior predictive checks"

machine_learning: paradigm_shift: "Prediction-focused, but can support causal inference" techniques: random_forest: use_for: "Variable importance, non-linear relationships" interpretation: ["Feature importance via Gini/permutation", "Partial dependence plots"] packages: ["randomForest (R)", "scikit-learn (Python)"] support_vector_machines: use_for: "Classification with complex boundaries" kernels: ["Linear", "Polynomial", "RBF"] packages: ["e1071 (R)", "scikit-learn (Python)"] neural_networks: use_for: "Complex non-linear patterns, image/text data" architectures: ["Feedforward", "CNN", "RNN/LSTM"] packages: ["keras/tensorflow", "pytorch"] gradient_boosting: use_for: "High-performance prediction, structured data" implementations: ["XGBoost", "LightGBM", "CatBoost"] advantage: "State-of-the-art performance on tabular data" validation_strategies: cross_validation: k_fold: description: "Split data into k folds, rotate train/test" typical_k: "5 or 10" stratified: description: "Preserve class proportions in each fold" use_when: "Imbalanced outcome variable" leave_one_out: description: "Use n-1 observations to predict 1" use_when: "Very small sample sizes" holdout: description: "Single train/test split (e.g., 80/20)" use_when: "Large datasets" bootstrap: description: "Resample with replacement" use_for: "Uncertainty estimation, small samples" interpretation_tools: shap_values: description: "Shapley Additive Explanations" advantage: "Game-theoretic, consistent feature attribution" packages: ["shap (Python)", "fastshap (R)"] use: "Explain individual predictions and global patterns" feature_importance: methods: - "Permutation importance (model-agnostic)" - "Gini importance (tree-based)" - "Coefficient magnitude (linear models)" partial_dependence: description: "Marginal effect of feature on prediction" packages: ["pdp (R/Python)", "iml (R)"] lime: description: "Local Interpretable Model-agnostic Explanations" use: "Explain individual predictions via local linear approximation" causal_ml: double_machine_learning: description: "Use ML for nuisance parameters, preserve inference" packages: ["DoubleML (Python/R)"] causal_forests: description: "Estimate heterogeneous treatment effects" packages: ["grf (R)", "EconML (Python)"] targeted_learning: description: "Efficient estimation of causal parameters" packages: ["tmle (R)", "tmle3 (R)"]

Research Paradigm? │ ├── Quantitative │ │ │ └── Dependent Variable Type? │ │ │ ├── Continuous │ │ │ │ │ └── Independent Variable Type? │ │ │ │ │ ├── Categorical (2 levels) │ │ │ ├── T > 0.8: t-test (modal) │ │ │ ├── T ≈ 0.6: Welch's t-test / Bayesian t-test │ │ │ ├── T ≈ 0.4: Mixed-effects / Bootstrap │ │ │ └── T < 0.3: ML classification + SHAP │ │ │ │ │ ├── Categorical (3+ levels) │ │ │ ├── T > 0.8: ANOVA (modal) │ │ │ ├── T ≈ 0.6: Welch ANOVA / Bayesian ANOVA │ │ │ ├── T ≈ 0.4: Mixed-effects / HLM │ │ │ └── T < 0.3: Random forests + variable importance │ │ │ │ │ └── Continuous │ │ ├── T > 0.8: OLS Regression (modal) │ │ ├── T ≈ 0.6: Robust / Bayesian regression │ │ ├── T ≈ 0.4: SEM / Causal inference (PSM, IV) │ │ └── T < 0.3: Causal forests / Double ML │ │ │ └── Categorical │ │ │ └── T > 0.8: Chi-square/Logistic (modal) │ T ≈ 0.5: Multinomial/Ordinal logistic │ T < 0.3: Bayesian logistic / Neural networks │ └── Qualitative │ ├── Interpretive Goal? │ │ │ ├── Describe experiences/meanings │ │ ├── T > 0.8: Basic thematic analysis (modal) │ │ ├── T ≈ 0.5: Interpretative Phenomenological Analysis (IPA) │ │ └── T < 0.3: Hermeneutic phenomenology │ │ │ ├── Build theory │ │ ├── T > 0.8: Generic grounded theory (modal) │ │ ├── T ≈ 0.5: Charmaz constructivist GT │ │ └── T < 0.3: Situational analysis / Critical GT │ │ │ ├── Analyze narratives/stories │ │ ├── T > 0.8: Thematic narrative analysis (modal) │ │ ├── T ≈ 0.5: Structural narrative analysis │ │ └── T < 0.3: Dialogic/performance analysis │ │ │ └── Count/quantify content │ ├── T > 0.8: Descriptive content analysis (modal) │ ├── T ≈ 0.5: Directed content analysis │ └── T < 0.3: Computational text analysis + ML

## Qualitative Analysis Guide ### Research Context | Element | Details | |---------|---------| | Research Question | {Question} | | Data Type | {Interviews / Focus groups / Documents / Visual} | | Sample Size | {N participants / texts} | | Paradigm | {Interpretive / Critical / Constructivist} | --- ### Recommended Analysis Method **Primary Method**: {Thematic Analysis / Grounded Theory / Content Analysis / Narrative Analysis} **Selection Rationale**: - {Fit with research question} - {Paradigmatic alignment} - {Data characteristics} **Software Recommendation**: {NVivo / ATLAS.ti / MAXQDA / Dedoose / Manual} - **Rationale**: {Why this software} --- ### Analysis Process #### Phase 1: {Phase name} **Activities**: 1. {Activity 1} 2. {Activity 2} **Output**: {Expected output} **Quality Check**: - [ ] {Quality criterion 1} - [ ] {Quality criterion 2} #### Phase 2: {Phase name} [Repeat for all phases] --- ### Coding Framework #### Initial Coding Scheme (if deductive) | Code | Definition | Inclusion Criteria | Example | |------|------------|-------------------|---------| | {Code 1} | {Definition} | {When to apply} | {Quote example} | | {Code 2} | {Definition} | {When to apply} | {Quote example} | #### Coding Process **Approach**: {Inductive / Deductive / Abductive} **Coder Training** (if multiple coders): - Training materials: {Description} - Practice rounds: {N rounds} - Disagreement resolution: {Process} **Inter-coder Reliability Target**: - Measure: {Krippendorff's α / Cohen's κ / % agreement} - Target: {≥ 0.80 / ≥ 0.70} --- ### Trustworthiness Criteria | Criterion | Strategy | Implementation | |-----------|----------|----------------| | Credibility | {Member checking / Prolonged engagement} | {Specific plan} | | Transferability | {Thick description} | {Specific plan} | | Dependability | {Audit trail / Reflexive journal} | {Specific plan} | | Confirmability | {Reflexivity / External audit} | {Specific plan} | --- ### Results Reporting #### Theme Structure **Theme 1**: "{Theme name}" - **Definition**: {What this theme represents} - **Sub-themes**: {If applicable} - **Illustrative quotes**: - "{Quote 1}" (Participant X) - "{Quote 2}" (Participant Y) #### Thematic Map

Parameter	Value
Expected effect size	[d = / η² = / f² = ]
Significance level (α)	.05
Power (1-β)	.80
Required sample size	N = [calculated value]

Metric	Small	Medium	Large
d	0.2	0.5	0.8
η²	.01	.06	.14
r	.10	.30	.50
f²	.02	.15	.35

Mechanism	Application Timing	Usage Example
Forced Analogy	Phase 2	Apply analysis methodology patterns from other fields by analogy (e.g., Physics → Social Science)
Iterative Loop	Phase 2-3	4-round analysis method refinement cycle
Semantic Distance	Phase 2	Discover semantically distant analysis technique combinations
Temporal Reframing	Phase 1	Review methodology development from past/future perspectives
Community Simulation	Phase 4-5	Methodology feedback from 7 virtual statisticians

name	e1
description	E1-Quantitative Analysis Guide with Code Generation & Sensitivity Analysis VS-Enhanced with Full 5-Phase process: Avoids obvious analyses, explores innovative methodologies Expanded to include qualitative analysis (thematic, grounded theory, content, narrative) Absorbed E4 (Analysis Code Generator) and E5 (Sensitivity Analysis - Primary Study) capabilities Use when: selecting statistical/qualitative methods, interpreting results, checking assumptions, generating code, sensitivity analysis Triggers: statistical analysis, ANOVA, regression, t-test, power analysis, assumption checking, effect size, thematic analysis, grounded theory, content analysis, narrative analysis, NVivo, ATLAS.ti, coding, qualitative data, R code, Python code, SPSS syntax, sensitivity analysis, robustness check
version	12.0.1

e1

ソース情報

インストール方法

ソースファイルを確認

このリポジトリの他の Skills

このリポジトリの他の Skills

⛔ Prerequisites (v8.2 — MCP Enforcement)

Checkpoints During Execution

Fallback (MCP unavailable)

E1-Quantitative Analysis Guide

Overview

⛔ Prerequisites (v8.2 — MCP Enforcement)

Checkpoints During Execution

Fallback (MCP unavailable)

E1-Quantitative Analysis Guide

Overview

VS-Research 5-Phase Process

Phase 0: Context Collection (MANDATORY)

Phase 1: Modal Analysis Method Identification

Phase 2: Long-Tail Analysis Method Sampling

Phase 3: Low-Typicality Selection

Phase 4: Execution

Phase 5: Suitability Verification

Typicality Score Reference Table

Quantitative Analysis Method T-Score

Qualitative Analysis Method T-Score

Input Requirements

For Quantitative Analysis

For Qualitative Analysis

Output Format (VS-Enhanced)

Homogeneity of Variance

[Additional assumptions...]

4. Power Analysis

A Priori Analysis

Sensitivity Analysis

5. Analysis Code

6. Effect Size Interpretation

7. Multiple Comparisons (if applicable)

8. Results Reporting Format (APA 7th)

Phase 5: Suitability Verification

Grounded Theory Analysis

Content Analysis

Narrative Analysis

Advanced Quantitative Methods (NEW in v5.0)

Bayesian Analysis

Machine Learning for Inference

Analysis Method Selection Flowchart (VS Enhanced - Expanded)

Qualitative Analysis Output Template

Absorbed Capabilities (v11.0)

From E4 — Analysis Code Generator

From E5 — Sensitivity Analysis (Primary Study)

Related Agents

Self-Critique Requirements (Full VS Mandatory)

v3.0 Creativity Mechanism Integration

Available Creativity Mechanisms

Checkpoint Integration

References

System References

Quantitative Methods References

Qualitative Methods References

Software References