Empirical Paper Analysis Skill

Skill Description

Target User

Input Requirements

• PDF file of an empirical research paper
• Publication information (Authors, Journal, Date, etc)

Analysis Framework

1. 问题的提出 (Problem Statement)

Analysis Points:

• What is the primary research question? / What problem or phenomenon is being studied?
• Why is this question important (policy relevance, theoretical gap, methodological innovation, practical value)?
• What is the economic/legal intuition behind the research design?

2. 实证研究的核心难题 (Core Empirical Challenges)

Common Challenges to Look For:

• Selection bias: Observed vs unobserved outcomes (e.g., selective labels problem)
• Omitted variable bias: Unobserved confounders (e.g., judges' private information)
• Endogeneity: Reverse causality or simultaneity
• Measurement error: How to quantify abstract concepts (e.g., legal ideas, judicial attitudes)
• External validity: Generalizability concerns
• Data limitations: Missing counterfactuals, truncated samples, etc.

• Clearly state the problem
• Explain why it matters for causal inference
• Use examples/tables to illustrate if helpful

3. 识别策略与方法设计 (Identification Strategy & Research Design)

Key Elements:

• Identification strategy: Natural experiment, IV, RD, DID, matching, ML+causal inference hybrid
• Data source: Dataset description, sample selection, time period
• Empirical specification: Main regression model, key variables
• Robustness checks: Alternative specifications, placebo tests, sensitivity analysis
• Novel methodological contributions: Any innovative techniques?

• Are the identification assumptions plausible?
• Are there remaining threats to validity?
• How convincing is the causal interpretation?

4. 重要发现与结论 (Key Findings & Conclusions)

Structure:

• Main findings (with magnitude/significance)
• Robustness of results
• Heterogeneous effects (if any)
• Economic/legal interpretation
• Policy implications

• Use bullet points for clarity
• Include key numbers (effect sizes, significance levels)
• Reference important tables/figures

5. 学术价值 (Academic Contribution)

Dimensions:

• Methodological innovation: New identification strategies, measurement techniques
• Theoretical contribution: New insights about legal/judicial behavior, institutional design
• Policy relevance: Implications for legal reform, judicial training, algorithm adoption
• Interdisciplinary impact: Bridges law, economics, computer science
• Future research: Opens new questions or directions

Output Format

Generate a structured markdown document following this template:

# [Paper Title]

**Authors:** [List]
**Journal:** [Name, Year]
**DOI/Link:** [If available]

## 问题的提出

[Analysis following framework above]

## 实证研究的核心难题

### 难题一：[Name]
[Explanation]

### 难题二：[Name]
[Explanation]

## 识别策略与方法设计

### 数据来源
[Description]

### 识别策略
[Core identification approach]

### 方法设计
[Technical details]

## 重要发现与结论

- **发现一：** [Finding with magnitude]
- **发现二：** [Finding with magnitude]
- **政策含义：** [Implications]

## 学术价值

- **方法论贡献：** [Innovation]
- **理论贡献：** [Insights]
- **政策相关性：** [Relevance]

Special Instructions

• Academic Tone: Use precise academic language appropriate for PhD-level analysis. Assume familiarity with econometric concepts (DID, IV, RDD, etc.) and ML methods (GBDT, NLP, embeddings).
• Bilingual Output: Primary language is Chinese (as shown in the examples), but technical terms can be included in parentheses with English abbreviation when first introduced.
• Mathematical Rigor: Don't shy away from mathematical notation when describing models or identification strategies. For example:
• Regression specifications: $Y_i = \beta_0 + \beta_1 Treatment_i + X_i'\gamma + \epsilon_i$
• DID: $Y_{ijt} = \alpha + \beta(Post_t \times Treat_j) + \delta_j + \lambda_t + \varepsilon_{ijt}$
• Critical Thinking: Don't just summarize—analyze. Question assumptions, evaluate identification strength, consider alternative explanations.
• Tables/Figures: When referencing tables or figures from the paper:
• Describe what they show conceptually
• Highlight the most important results
• Don't try to reproduce full tables in text
• Scope: Focus on the five core sections. Don't add unnecessary sections.

Example Workflow

• Read the entire paper to understand the research question and context
• Extract the empirical strategy - pay special attention to identification sections
• Identify the key challenges the authors face
• Trace how they solve each challenge methodologically
• Synthesize the findings with appropriate interpretation
• Evaluate the contribution in context of the literature