joap-data-analysis — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited joap-data-analysis (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
JAP analyses must be model-appropriate, fully reported, and reproducible. The house toolkit is SEM/CFA, multilevel (HLM) models, mediation and moderation with proper inference, and meta-analysis. The journal expects effect sizes with confidence intervals, fit indices, bootstrap CIs for indirect effects, full disclosure of how data were handled, and a clean confirmatory vs. exploratory separation, with data and code shareable under TOP.
SRMR) and reliability/AVE before interpreting the structural model; report measurement invariance when comparing groups or waves.
intervals** for key paths — not just p-values and stars.
only the Baron-Kenny steps or a Sobel z); for multilevel mediation use the appropriate (e.g., 1-1-1, 2-1-1, 2-2-1) decomposition and within/between separation.
grand-mean), model random effects, and justify the estimator; do not run OLS on nested data.
attrition) and reasons, all conditions, and all measures. Label confirmatory vs. exploratory.
artifact corrections, publication-bias checks, and a transparent coding protocol.
fresh session (see joap-open-science-and-transparency).
The servant-leadership package: a multilevel mediation in the field and a causal test in the lab.
Measurement model (field, N = 612 in 74 teams)
CFA: χ²/df = 2.1, CFI = .96, TLI = .95, RMSEA = .045, SRMR = .04
Reliabilities ω: leadership .91, safety .88, performance .87
Aggregation: ICC(1) = .16, ICC(2) = .77, r_wg(j) = .85 → team-level OK
Confirmatory (preregistered) — multilevel mediation (2-2-2)
a (leadership→safety) = .42 [.27, .57]; b (safety→performance) = .31 [.14, .48]
Indirect = .13, 95% Monte Carlo CI [.05, .23] → mediation supported
Direct (leadership→performance | safety) = .09 [-.06, .24], ns
Confirmatory (preregistered) — lab experiment (causal leg)
Servant vs control on safety: d = 0.46, 95% CI [0.21, 0.71]
H3 boundary: interaction with interdependence, ΔR² = .03, CI excludes 0
Exploratory (labeled): voice as a serial L1 mediator surfaced post hoc;
reported as exploratory, flagged for confirmation in future work.Why this passes JAP scrutiny: the measurement model is reported before structure; the indirect effect carries a bootstrap/Monte Carlo CI; nesting is modeled and aggregation justified; the experimental leg supplies causal warrant; and the post hoc serial path is honestly demoted to exploratory.
| Reviewer pushback | What it signals | JAP fix |
|---|---|---|
| "Mediation by Sobel/steps only" | outdated inference | report indirect effect + bootstrap/Monte Carlo CI |
| "OLS on nested data" | dependence ignored | multilevel model; report ICC, centering, random effects |
| "No fit indices / measurement model" | construct validity unchecked | report CFA fit + reliability before structure |
| "Which exclusions were preregistered?" | forking-paths concern | disclosure table: rule, count, preregistered vs post hoc, result with/without |
| "Is this confirmatory?" | HARKing concern | point to the preregistration; relabel post hoc analyses exploratory |
| "Reviewer 2 couldn't rerun your code" | reproducibility gate | ship a fresh-session run log (see open-science skill) |
convinces no JAP reviewer.
pre-reform.
and report so that interval is defensible.
Run the battery, don't just enumerate it. Full map: execution-with-mcp. JAP is organizational psychology — multilevel survey/field data and experiments; cluster at the right level and apply mediation/moderation discipline.
romano_wolf (step-down FWER) orbenjamini_hochberg — report the adjusted threshold.
oster_delta / sensemakr.wild_cluster_bootstrap (few clusters), twoway_cluster / conley;multilevel data → cluster at the right level.
audit_result(result_id) lists the missing checks and theexact suggest_function for each.
etable / did_summary_to_latex from 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.
【Measurement model】CFA fit + reliability + invariance reported? [Y/N]
【Main result】effect size(s) + CI(s) + meaning
【Mediation/multilevel】indirect-effect bootstrap/Monte Carlo CI; nesting modeled? [Y/N/NA]
【Disclosure】N-determination + all exclusions + all conditions + all measures? [Y/N]
【Confirmatory vs exploratory】clearly separated? [Y/N]
【Reproducible】scripts + codebook + fresh-session check? [Y/N]
【Next】joap-tables-figures../../resources/external_tools.md — Mplus, lavaan, lme4/nlme, psych, metafor/metaSEM, bootstrap/Monte Carlo tools../../resources/official-source-map.md — statistical and disclosure requirements~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.