Use when executing and reporting the analysis for an Administrative Science Quarterly (ASQ) manuscript — qualitative coding and data-to-theory construction, or quantitative estimation and robustness. Makes the evidence-to-theory link transparent; it does not design the study (see asq-methods).
SkillScore86Maintained
Automatic
◆Research pack
Last verified 8/15/2026
What it changes
Passed 19 of 22 test cases with the skill, and 14 of 22 without it.
Without skill64%
With skill86%
Checks it passed that the model alone missed
Recommends calling audit_result on the result handle to list missing checks.
Recommends building a visual temporal display to model process dynamics over time.
Recommends theoretical contribution framing as the next step after empirical analysis.
and 2 more in the full report
5 fixed14 passed either way3 still failing0 broke
Pass-rate lift
+23 pp
Tokens
+22%
Agent turns
0%
Response time
-27%
Model gemini-3.6-flashMeasured 8/15/2026v119 of 22 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
We generated an estimation handle during automated statistical execution for our panel regression model. We want to identify missing robustness checks without re-specifying all models manually. What execution command inspects the missing checks?
What was checked, without and with the skill
Checked
Without skill
With skill
Recommends calling audit_result on the result handle to list missing checks.
Missed
Passed
case-15 · 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.md79 linessingle file
Data Analysis & Evidence (asq-data-analysis)
When to trigger
You have data but the path from data to theory is opaque
Qualitative: your quotes are decorative, not evidentiary; coding is undocumented
Quantitative: main results exist but robustness/alternative explanations are thin
Reviewers ask "how did you get from your data to these constructs?"
Branch A — Qualitative analysis (the data-to-theory link)
ASQ expects readers to see how raw data became theory — its guidelines stress that helping readers understand how the research was performed and ensuring the trustworthiness of published work are explicit aims (verify at journals.sagepub.com/author-instructions/asq). Qualitative rigor is judged on its own terms here, not held to a quantitative yardstick. Make the analytic ladder visible.
Transparent coding. Describe first-order codes (informant terms), second-order themes (researcher constructs), and aggregate dimensions — the Gioia-style data structure — or an equivalent (Eisenhardt cross-case, Langley process bracketing). State who coded, how disagreements were resolved, and how iteration proceeded.
Data-to-theory table. Provide a table linking representative raw evidence → codes → constructs, so the inference is auditable (see asq-tables-figures).
Power quotes vs. proof quotes. Use a few vivid "power quotes" in the body; place corroborating "proof quotes" in tables/appendix. Quotes must carry the claim, not illustrate it after the fact.
Evidence for each construct. Every theoretical construct should be backed by patterned evidence across informants/cases, with counts or prevalence where appropriate.
Negative cases. Report disconfirming instances and how they refined the theory.
Process display. For process theory, show the temporal/event structure (timeline, phase model, visual mapping) — as Barley (1986, ASQ) did in tracing how CT scanners restructured radiology departments over time.
Branch B — Quantitative analysis
Main models match the design (FE/RE, event-history, multilevel, network models); report clearly with appropriate standard errors (clustering at the right level).
Robustness that targets the theory's threats: alternative measures, alternative samples, alternative specifications, endogeneity checks, and modern staggered-DiD diagnostics if relevant.
Mechanism evidence. Don't stop at the reduced-form relationship — provide mediation/moderation or supplementary tests that probe why.
Effect interpretation. Report and interpret magnitudes in organizational terms, not just significance stars.
Alternative explanations are tested, not waved away.
Either branch — the "so what" of the evidence
Tie every analytic result back to the mechanism and the surprise.
Distinguish what the data can and cannot establish — overclaiming is a fast path to rejection.
Prepare the exhibits jointly with asq-tables-figures.
Execution bridge (StatsPAI / Stata MCP)
Run the battery, don't just enumerate it. Full map:
execution-with-mcp. ASQ wants a clean causal or well-identified observational design behind an organizational-theory contribution; reduced-form estimation fits the chain below, interpretive work does not.
Many outcomes / specifications:romano_wolf (step-down FWER) or
Use when building and refining exhibits for an Administrative Science Quarterly (ASQ) manuscript — qualitative data structures and data-to-theory tables, process models, and quantitative tables. Designs exhibits; it does not run the analysis behind them.