Use when executing and reporting the analysis for an American Sociological Review (ASR) manuscript so it survives expert masked review — honest uncertainty, robustness, and evidence handling appropriate to quantitative, demographic, comparative-historical, or computational sociology. Guides analysis norms; it does not fabricate results.
SkillScore85Maintained
Automatic
◆Research pack
Last verified 8/15/2026
What it changes
Passed 23 of 27 test cases with the skill, and 16 of 27 without it.
Without skill59%
With skill85%
Checks it passed that the model alone missed
Uses the section tag 【Next】.
Uses the section tag 【Main result】.
Uses the bracketed header 【Main result】.
and 7 more in the full report
7 fixed16 passed either way4 still failing0 broke
Pass-rate lift
+26 pp
Tokens
+45%
Agent turns
0%
Response time
-24%
Model gemini-3.6-flashMeasured 8/15/2026v123 of 27 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 7
Prompt
When inspecting regression result handles in statistical automation tools to catch omitted diagnostic tests, which function checks the handle for missing checks and returns specific execution recommendations?
What was checked, without and with the skill
Checked
Without skill
With skill
Identifies the audit_result function for identifying missing estimation checks.
Missed
Passed
case-22 · gemini-3.6-flash · 8/15/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md128 linessingle file
Data Analysis (asr-data-analysis)
ASR reviewers are methodologically demanding across very different traditions. Whether your evidence
is regression coefficients, life tables, archival sequences, or coded fieldnotes, the analysis must be
transparent, well-documented, and reproducible to the extent your data allow. Design decisions live in
asr-research-design.
When to trigger
Running main and supporting analyses; building the results/findings section
A reviewer asked for robustness, heterogeneity, alternative specifications, or more evidence
Documenting how qualitative claims are grounded in the data
Making the analysis reproducible before sharing materials
Analysis norms ASR expects
Quantitative / demographic
Report uncertainty and magnitude, not just significance — intervals and substantive effect
sizes; respect survey design (weights, clustering).
Robustness that probes, not decorates — alternative measures, samples, estimators, and
specifications that could break the result; say what you learn.
Heterogeneity with discipline — pre-specify or justify subgroups; adjust for multiple
comparisons; don't mine an interaction and theorize it post hoc.
Measurement — validate constructs, report reliability, show results aren't an artifact of a
coding/scaling choice (especially for inequality and well-being measures).
Comparative-historical / ethnographic
Make the chain of evidence explicit: link each claim to specific sources, observations, or
cases; present negative/disconfirming evidence.
Use evidence tables, timelines, or coded excerpts so reviewers can trace claims to data.
Computational / text-as-data
Document model/version, hyperparameters, seeds, preprocessing; validate against human-labeled
samples; report stability. Don't treat model outputs as ground truth.
Reproducibility while you work
One master script regenerates every table/figure from raw/constructed data (quantitative).
Set and report seeds for any stochastic step.
Pin software/package versions (renv.lock, requirements.txt, recorded ssc/net installs).
For qualitative work, keep a documented codebook and analytic memos.
Use when executing and reporting the analysis for an American Journal of Political Science (AJPS) manuscript. AJPS will have a third-party verifier re-run your exact code against the numerical results in the main text before publication, so analyze reproducibly from the first line. Guides analysis and reporting norms; it does not fabricate results.