Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› cps-data-analysis
cps-data-analysis
GitHub用于 Comparative Political Studies 稿件的数据分析技能，涵盖估计、不确定性评估、稳健性及多方法三角验证。确保在跨国数据中透明、可复现地处理聚类、测量误差等挑战，规范结果报告标准。
Trigger Scenarios
执行主要结果、稳健性和异质性分析
回应审稿人对标准误、规范或测量准确性的质疑
决定主文与附录的内容分配
将定量估计与案例证据进行三角验证
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill cps-data-analysis -g -y
SKILL.md
Frontmatter
{
"name": "cps-data-analysis",
"description": "Use when running and reporting the analyses for a Comparative Political Studies (CPS) manuscript — estimation, uncertainty, robustness, and multi-method triangulation on comparative data. Sets analysis norms; it does not choose the identification strategy (see cps-research-design)."
}
Data Analysis (cps-data-analysis)
Once the design is fixed (cps-research-design), this skill governs how the analyses are run and
reported so a CPS reviewer trusts them. Comparative data bring distinctive hazards: few clusters
(countries), cross-national measurement error, missing data that differ by regime, and the temptation to
over-read a panel correlation as causal. The standard is modern, transparent, and replication-ready.
When to trigger
- Estimating the main results, robustness, and heterogeneity
- A reviewer questioned standard errors, specification, measurement, or fragility of the result
- Deciding what goes in the main text vs. the supplementary/online appendix
- Triangulating quantitative estimates with case evidence
Analysis priorities (in order)
- Main estimate that matches the design. The headline specification should be the one the identification argument justifies — not the one with the biggest coefficient or most stars.
- Honest uncertainty. Cluster at the assignment level (usually country / country-year); with few countries use wild-cluster bootstrap or randomization inference. Report CIs, not just stars.
- Measurement transparency. Name the source and coding of each comparative variable (e.g., V-Dem, Polity, CSES, Manifesto Project); show robustness to alternative codings of the key construct.
- Robustness as a coherent story. Alternative specifications, samples, codings, and estimators that probe the threats named in the design — not a scattershot table of every variant.
- Heterogeneity by theory. Subgroups/scope conditions pre-specified by the mechanism
(
cps-theory-building), not data-mined; adjust for multiple comparisons. - Mechanism evidence. Tie the quantitative result to the mechanism — mediation cautiously, or case evidence in a multi-method design.
Comparative-data hazards to address explicitly
| Hazard | Symptom | Fix |
|---|---|---|
| Few clusters (countries) | over-rejection, tiny SEs | wild-cluster bootstrap / randomization inference |
| Cross-national measurement error | results flip across codings | show robustness to V-Dem/Polity/alt scales |
| Differential missingness | sample changes by regime type | report attrition; multiple imputation with caution |
| Time-series confounding | spurious trend correlations | unit + period FE; over-time placebo |
Failure-mode audit
Run this audit before interpreting the main coefficient:
- Concept equivalence: Does the key variable mean the same thing across regimes, languages, regions, or institutions? If not, report measurement-invariance checks, alternative codings, or scope limits.
- Selection into observation: Are only more democratic, richer, more peaceful, or better-measured cases observed? Report the observation process and show how estimates change under credible sample restrictions.
- Temporal dependence: Are observations mechanically persistent across years? Use lag structure, unit trends, event-time plots, or placebo leads to avoid re-labeling persistence as effect.
- Cluster leverage: Does one country, region, election, conflict, or reform episode drive the result? Show leave-one-cluster-out or influence diagnostics for claims that hinge on few cases.
- Subgroup multiplicity: If theory predicts heterogeneity, pre-specify the dimensions and report how many comparisons were examined.
The output should connect each failure mode to a design threat. Do not add a robustness table unless it
answers a named threat in cps-research-design.
Execution bridge (StatsPAI / Stata MCP)
Run the battery, don't just enumerate it. Full map:
execution-with-mcp. CPS is comparative politics — cross-national and sub-national designs; emphasize identification and clustered / multiway inference.
- Many outcomes / specifications:
romano_wolf(step-down FWER) orbenjamini_hochberg— report the adjusted threshold. - OVB sensitivity:
oster_delta/sensemakr. - Inference:
wild_cluster_bootstrap(few clusters),twoway_cluster/conley; multilevel data → cluster at the right level. - Re-fit off one handle:
audit_result(result_id)lists the missing checks and the exactsuggest_functionfor each. - Exhibits:
etable/did_summary_to_latexfrom the handle — no retyped numbers.
Keep the decisive checks in the body and the exhaustive battery in the supplement. See the executed chain in the JF execution walkthrough.
Checklist
- Headline specification = the one the design justifies
- SEs clustered at the assignment level; few-cluster correction where needed
- Every comparative variable's source and coding named; key construct robust to alt codings
- Robustness probes the design's named threats; not a kitchen sink
- Heterogeneity pre-specified by theory; multiple testing addressed
- Main text vs. appendix split is deliberate; every appendix result is referenced
- All results reproduce from the script destined for the CPS Dataverse
Anti-patterns
- Treating a cross-national panel correlation as causal without the design to back it
- Default OLS SEs with 20 countries (massively over-rejects)
- Cherry-picking the coding of the key variable that gives significance
- Robustness theater — many variants that never test the actual threat
- Data-mined subgroups reported as confirmed heterogeneity
- Results in the paper that the deposited code does not reproduce
Output format
【Headline result】estimate + CI, with the design it rests on
【Inference】clustering level + few-cluster correction if any
【Measurement】sources/codings + alt-coding robustness
【Failure-mode audit】concept equivalence / observation selection / temporal dependence / cluster leverage / multiplicity
【Robustness】the design-threats probed
【Heterogeneity】theory-driven subgroups + multiple-testing fix
【Reproducible?】script regenerates every exhibit [Y/N]
【Next】cps-tables-figures
Supplementary resources
../../resources/code/— clean → estimate → robustness → tables skeleton (Stata + Python)../../resources/external_tools.md— estimation and inference packages (R / Stata / Python)
Version History
- 1839142 Current 2026-07-05 12:38
