Back to Details

research-ideation

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Research Ideation

研究构思

Purpose

目的

This skill helps economists generate and refine research questions by applying economic thinking frameworks. It guides the process from observing phenomena to formulating testable hypotheses.
本Skill帮助经济学家运用经济学思维框架生成并完善研究问题，指导其从观察现象到提出可检验假设的全过程。

When to Use

适用场景

  • Starting a new research project
  • Brainstorming dissertation topics
  • Finding angles for policy analysis
  • Developing extensions of existing work
  • 启动新的研究项目
  • 构思论文选题
  • 寻找政策分析的切入点
  • 拓展现有研究成果

Instructions

操作步骤

Step 1: Understand the Starting Point

步骤1：明确起点

Ask the user:
  1. Do you have a phenomenon, puzzle, or data source in mind?
  2. What field/subfield of economics?
  3. What's your methodological preference? (Theory, empirical, structural)
  4. Any constraints? (Data access, timeline, etc.)
向用户询问：
  1. 你是否有特定的现象、谜题或数据源？
  2. 研究属于经济学的哪个领域/子领域？
  3. 你偏好哪种研究方法？（理论研究、实证研究、结构分析）
  4. 存在哪些约束条件？（数据获取、时间周期等）

Step 2: Apply Ideation Frameworks

步骤2：运用构思框架

Use these approaches:
1. The Puzzle Approach
  • What's surprising about current patterns?
  • What contradicts conventional economic wisdom?
  • Where do standard models fail?
2. The Policy Approach
  • What policies lack rigorous evaluation?
  • What natural experiments remain unexploited?
  • What interventions might solve important problems?
3. The Data Approach
  • What new data sources have become available?
  • What can existing data tell us that hasn't been explored?
  • What linkages between datasets are possible?
4. The Extension Approach
  • How can seminal papers be extended?
  • What mechanisms remain unexplored?
  • Can methods from one field apply to another?
可采用以下方法：
1. 谜题导向法
  • 当前的经济模式有哪些令人意外之处？
  • 哪些现象与传统经济学理论相悖？
  • 标准模型在哪些场景下失效？
2. 政策导向法
  • 哪些政策缺乏严谨的评估？
  • 哪些自然实验尚未被充分利用？
  • 哪些干预措施可能解决重要问题？
3. 数据导向法
  • 有哪些新的数据源可用？
  • 现有数据中还有哪些未被挖掘的信息？
  • 不同数据集之间可以建立哪些关联？
4. 拓展延伸法
  • 如何拓展经典论文的研究内容？
  • 哪些作用机制尚未被探索？
  • 能否将某一领域的方法应用到另一领域？

Step 3: Evaluate and Refine

步骤3：评估与完善

For each idea, assess:
  • Feasibility: Can this be done with available data/methods?
  • Contribution: What's new here?
  • Interest: Who cares about this question?
  • Identification: Can effects be credibly estimated?
针对每个构思，从以下维度评估：
  • 可行性：是否可利用现有数据/方法完成？
  • 创新性：研究的新颖之处是什么？
  • 相关性：哪些群体关注这个问题？
  • 识别性：能否可靠地估算效应？

Example Interaction

交互示例

User Query: "I'm interested in labor economics and have access to LinkedIn data"
Generated Response:
用户提问：“我关注劳动经济学领域，且可获取LinkedIn数据”
生成回复：

Research Ideas: LinkedIn Data × Labor Economics

研究构思：LinkedIn数据 × 劳动经济学

1. Network Effects in Job Search

1. 求职中的网络效应

Question: How do professional networks affect job transitions and wage changes?
Puzzle: Standard search models assume random matching, but most jobs come through networks. How much do networks matter for outcomes?
Approach:
  • Track profile updates signaling job changes
  • Measure network overlap between current/future employers
  • Instrument with exogenous network shocks (school alumni connections)
Contribution: Quantify the causal role of networks vs. unobserved ability
**问题：**职业网络如何影响职业转换与薪资变化？
**谜题：**标准搜索模型假设匹配是随机的，但大多数工作机会来自人脉网络。人脉对求职结果的影响程度如何？
研究方法：
  • 追踪显示职业变动的个人资料更新
  • 衡量当前雇主与未来雇主之间的网络重叠度
  • 以外生网络冲击（如校友关系）作为工具变量（IV）
**贡献：**量化人脉与未观测能力的因果作用

2. Skill Signaling and Returns

2. 技能信号与回报

Question: Do workers who list specific skills on profiles earn wage premiums?
Puzzle: Are listed skills signals of ability, or just cheap talk? What's the return to skill acquisition vs. skill signaling?
Approach:
  • Compare wages before/after workers add skills
  • Use skill endorsements as validation
  • Exploit skill assessment features as quasi-experiments
**问题：**在个人资料中列出特定技能的工人是否能获得薪资溢价？
**谜题：**列出的技能是能力的信号，还是仅仅是廉价的自我宣传？技能获取与技能信号的回报有何差异？
研究方法：
  • 对比工人添加技能前后的薪资
  • 将技能背书作为验证依据
  • 利用技能评估功能作为准实验

3. Gender Gaps in Self-Promotion

3. 自我宣传中的性别差异

Question: Do men and women describe equivalent achievements differently?
Puzzle: Lab evidence shows women understate accomplishments. Does this appear in real profiles and affect outcomes?
Approach:
  • NLP analysis of profile language (adjectives, quantification)
  • Compare language for workers in same role/company
  • Link to recruiter contact rates if available
**问题：**男性和女性对同等成就的描述是否存在差异？
**谜题：**实验室证据显示女性会低估自身成就，这种现象是否在真实的个人资料中存在，且会影响求职结果？
研究方法：
  • 对个人资料语言进行NLP分析（形容词、量化表述）
  • 对比同一岗位/公司中员工的语言差异
  • 若数据可用，关联招聘人员的联系率

4. Remote Work and Labor Market Geography

4. 远程办公与劳动力市场地理范围

Question: How has remote work changed the geographic reach of job matching?
Approach:
  • Compare pre/post-2020 job applications by distance
  • Measure changes in location signals on profiles
  • Study wage convergence across cities
**问题：**远程办公如何改变求职匹配的地理范围？
研究方法：
  • 对比2020年前后求职申请的距离差异
  • 衡量个人资料中地理位置信号的变化
  • 研究不同城市间的薪资趋同情况

Evaluation Matrix

评估矩阵

IdeaData FeasibilityIdentificationPolicy Relevance
Network effectsHighMedium (need IV)High
Skill signalingHighMediumMedium
Gender languageHighHigh (descriptive)High
Remote geographyHighHigh (COVID natural experiment)High
构思数据可行性识别性政策相关性
网络效应高中（需使用IV）高
技能信号高中中
性别语言差异高高（描述性研究）高
远程办公地理范围高高（COVID自然实验）高

Recommended Next Steps

后续建议步骤

  1. For empirical work: Start with Idea #4 (clearest identification)
  2. For policy impact: Idea #1 has most implications for job matching platforms
  3. For methodological novelty: Idea #3 combines NLP with quasi-experimental methods
  1. **实证研究方向：**从构思#4开始（识别性最清晰）
  2. **政策影响方向：**构思#1对求职匹配平台的启示最大
  3. **方法创新方向：**构思#3结合了NLP与准实验方法

Frameworks for Generating Questions

研究问题生成框架

The "5 Whys" for Economics

经济学中的“5个为什么”法

Start with an observation and drill down:
  1. GDP growth is slowing → Why?
  2. Productivity is stagnant → Why?
  3. Investment is low → Why?
  4. Uncertainty is high → Why?
  5. Policy is unpredictable → Testable: Does policy uncertainty cause low investment?
从观察现象入手，逐步深入：
  1. GDP增速放缓 → 原因是什么？
  2. 生产率停滞 → 原因是什么？
  3. 投资低迷 → 原因是什么？
  4. 不确定性高企 → 原因是什么？
  5. 政策不可预测 → 可检验假设：政策不确定性是否导致投资低迷？

The "What If" Generator

“假设情境”生成法

  • What if [policy X] were implemented?
  • What if [technology Y] became widespread?
  • What if [assumption Z] were relaxed?
  • 若实施[政策X]会怎样？
  • 若[技术Y]普及会怎样？
  • 若放宽[假设Z]会怎样？

The Cross-Field Pollinator

跨领域迁移法

Take a method from one field and apply to another:
  • IO techniques → Labor markets (how do firms choose wages like prices?)
  • Finance models → Education (returns to schooling as asset pricing)
  • Macro shocks → Micro outcomes (firm-level effects of exchange rate changes)
将某一领域的方法应用到另一领域：
  • 产业组织（IO）技术 → 劳动力市场（企业如何像定价一样设定薪资？）
  • 金融模型 → 教育领域（教育回报作为资产定价问题）
  • 宏观冲击 → 微观结果（汇率变动对企业层面的影响）

Common Pitfalls

常见误区

  • ❌ Questions that are too broad ("What causes inequality?")
  • ❌ Questions without clean identification ("Does education cause income?")
  • ❌ Questions without data ("Were medieval peasants happy?")
  • ❌ Questions already well-answered
  • ❌ 问题过于宽泛（如“什么导致了不平等？”）
  • ❌ 问题缺乏清晰的识别性（如“教育是否能提高收入？”）
  • ❌ 问题无数据支撑（如“中世纪农民是否幸福？”）
  • ❌ 问题已被充分研究

References

参考文献

Changelog

更新日志

v1.0.0

v1.0.0

  • Initial release with ideation frameworks
  • 初始版本，包含研究构思框架