From jpsp-skills
Guides analysis and internal meta-analysis for JPSP multi-study manuscripts to JARS standard, covering effect sizes, robustness, and pooled estimates.
How this skill is triggered — by the user, by Claude, or both
Slash command
/jpsp-skills:jpsp-data-analysisThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
JPSP reviewers are methodologically sophisticated and the journal requires **JARS** reporting and
JPSP reviewers are methodologically sophisticated and the journal requires JARS reporting and
TOP Level 2 transparency. Two things set JPSP analysis apart from short-report work: (1) you are
analyzing a set of studies, and (2) the section expects you to integrate across them — an
internal meta-analysis is a core move, explicitly prioritized in IRGP guidance. This skill covers
execution and reporting; design lives in jpsp-study-design.
renv.lock, requirements.txt).These are the recurring post-credibility-revolution objections a JPSP section reviewer raises about results, and the move that defuses each. The fix is an analysis-and-reporting fix — design lives in jpsp-study-design.
| Reviewer says | What they distrust | The JPSP-fit fix |
|---|---|---|
| "Stars only — where are the effect sizes?" | You hid magnitude behind significance | Report d/r/β with 95% CIs and a sentence on substantive meaning, per JARS |
| "Each study reads in isolation" | No cumulative claim | Add the internal meta-analysis: pooled estimate + forest plot + heterogeneity (I²) |
| "This could be a confound" | Alternative causal account | A focused alternative-explanation analysis or a study that manipulates the confound |
| "Mediation is over-claimed" | Causal language on cross-sectional data | Bootstrapped indirect effect with CI, hedged: "consistent with, not proof of, the path" |
| "Was this predicted?" | Suspected HARKing | A confirmatory/exploratory table mapping each test to the preregistration; flag deviations |
Illustrative numbers — invented to show the reporting logic, not real results.
Three ASC studies estimate the same gratitude→construal effect. Per-study d (95% CI): S1 = 0.34 [0.10, 0.58], S2 = 0.21 [−0.02, 0.44], S3 = 0.40 [0.16, 0.64]. S2 alone is "non-significant" — the trap that invites a reviewer to call the package inconsistent. The internal meta-analysis instead reports a random-effects pooled d ≈ 0.31, 95% CI [0.18, 0.45], low heterogeneity (I² ≈ 12%). That one sentence — "across three preregistered studies, pooled d = 0.31 [0.18, 0.45]" — is the strongest summary in the paper and exactly what an ASC reviewer means by integrative analysis. Report S2 inside the pool, not as a footnote.
【Per-study effects】effect size + CI + substantive meaning
【Internal meta-analysis】pooled random-effects estimate + heterogeneity + forest plot
【Mechanism】indirect effect + bootstrapped CI (design caveats)
【Robustness】specs that could break it → what held
【Registered vs exploratory】clearly separated?
【JARS + repository】reported to standard + data/code/materials posted? [Y/N]
【Next】jpsp-tables-figures
../../resources/external_tools.md — metafor (internal meta-analysis), lavaan/PROCESS, effect-size + reliability tools../../resources/official-source-map.md — JARS, TOP Level 2, and IRGP integrative-analysis expectationnpx claudepluginhub brycewang-stanford/awesome-journal-skills --plugin jpsp-skillsGuides analysis and reporting for Journal of Applied Psychology manuscripts, covering SEM, multilevel models, mediation/moderation, and meta-analysis with proper effect sizes and confidence intervals.
Guides data analysis and reporting for JEP manuscripts: multilevel models, effect sizes with CIs, mediation/moderation, and full JARS disclosure.
Designs a multi-study package for a JPSP manuscript: sequences studies, powers each one, selects experimental/longitudinal/dyadic designs, and plans preregistration.