Use when defending the research design of a Comparative Political Studies (CPS) manuscript — cross-national/panel identification, case-based comparison and process tracing, experiments, or multi-method designs. CPS prizes comparative leverage. Strengthens the design; it does not write code.
SkillScore70Maintained
Automatic
◆Research pack
Last verified 9/1/2026
What it changes
Passed 16 of 23 test cases with the skill, and 12 of 23 without it.
Without skill52%
With skill70%
Checks it passed that the model alone missed
classifies the design as comparative-causal
specifies cps-data-analysis as the next step
specifies cps-data-analysis as the next stage
and 3 more in the full report
5 fixed11 passed either way6 still failing1 broke
Pass-rate lift
+17 pp
Tokens
+18%
Agent turns
0%
Response time
-13%
Model gemini-3.6-flashMeasured 9/1/2026v116 of 23 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 5 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 5
Prompt
I am developing a qualitative comparative study examining labor market institution divergence across two post-communist states that share baseline structural conditions. I need to justify my case selection using process tracing and rule out alternative economic explanations. Please write up the research design using the standard layout with section headings for Mode, Comparative leverage, Estimand or claim, Key assumption(s), Rival ruled out, Robustness/sensitivity, and Next.
What was checked, without and with the skill
Checked
Without skill
With skill
specifies cps-data-analysis as the next stage
Missed
Passed
specifies explicit process tracing tests such as hoop or smoking gun tests
Passed
Passed
includes the adjudication sentence formatted as "If the rival were true rather than my argument, the cross-case/over-time pattern would look like"
Missed
Passed
case-02 · gemini-3.6-flash · 9/1/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md94 linessingle file
Research Design (cps-research-design)
CPS is methodologically pluralist but demanding about each tradition. The design must credibly connect
the comparative argument (cps-theory-building) to evidence and rule out the leading rival
(cps-literature-positioning). This skill is mode-aware: pick the section that matches your work and
defend the comparative leverage — the variation across cases or time that identifies the claim.
When to trigger
Specifying the identification strategy, case selection, or experimental design
A reviewer questioned causal claims, case choice, external validity, comparability, or a confound
Designing a cross-national panel, a subnational comparison, or a natural experiment across borders
Justifying why the design adjudicates the rival account, not just shows an association
Identification first. State the estimand and the assumptions that license a causal reading
(parallel trends, exclusion, continuity, ignorability). Defend them; don't assert them.
Designs: cross-national panels with unit and period fixed effects; DiD/event study around reforms
(use modern staggered-adoption estimators, not naive TWFE); RD around institutional thresholds; IV
(first-stage strength, exclusion, weak-IV-robust inference); survey experiments fielded comparatively.
Comparability. Defend that the units are measured the same way across countries (V-Dem vs. Polity
coding, harmonized surveys); address country-level confounding and cross-national measurement error.
Inference: cluster at the level of treatment assignment (often country or country-year); few-cluster
corrections (wild bootstrap) when the number of countries is small; multiple-comparison adjustment.
Sensitivity: how strong must an unobserved country-level confounder be to overturn the result?
Case-based / qualitative comparison
Comparison logic: most-similar (control on shared traits, vary the cause) or most-different
(shared outcome despite different contexts) — justified by design, not convenience.
Case selection: typical, deviant, most/least-likely, paired comparison. Say what each case is a
case of and avoid selecting on the outcome.
Process tracing with explicit tests (hoop, smoking-gun, straw-in-the-wind); state what evidence
would have disconfirmed the mechanism.
Source transparency: archives, interviews, fieldnotes — plan documentation (see cps-transparency-and-data).
Use when defending the research design of an American Political Science Review (APSR) manuscript — causal identification for quantitative work, case selection and process tracing for qualitative work, experimental and survey-experimental design, or formal-empirical linkage. APSR judges each tradition on its own terms. Strengthens the design; it does not write code.