analyzing-data_skill

Overview

This skill is a statistical analysis toolkit for researchers conducting hypothesis tests, regression, correlation, Bayesian analyses, power calculations, and APA-style reporting. It streamlines test selection, assumption checking, effect size computation, and generation of publication-ready statistical summaries. Use it to move from raw data to defensible statistical conclusions and clear reports.

How this skill works

The skill inspects your research questions and dataset to recommend appropriate tests and power calculations, runs hypothesis tests (t-tests, ANOVA, chi-square), regression and correlation models, and performs Bayesian analyses when requested. It automates assumption checks (normality, homoscedasticity, independence), computes effect sizes and confidence intervals, and formats results into APA-style output ready for manuscripts or reports.

When to use it

• When choosing and running hypothesis tests (t-tests, ANOVA, chi-square).
• When performing regression, correlation, or model diagnostics for empirical data.
• When conducting Bayesian analyses or reporting Bayes factors and posterior summaries.
• When calculating effect sizes, confidence intervals, and conducting power analyses.
• When you need APA-format tables and narrative summaries for publication or thesis reporting.

Best practices

• Pre-register hypotheses and analysis plans before running confirmatory tests.
• Inspect and clean data first; handle missing data and outliers before inferential tests.
• Always run assumption checks and report diagnostics alongside test results.
• Report effect sizes and confidence intervals, not just p-values; include power analysis where relevant.
• Use Bayesian methods for flexible inference or when prior information is available; document priors and sensitivity checks.

Example use cases

• Compare group means with independent or paired t-tests and report Cohen’s d and 95% CI in APA format.
• Run one-way or factorial ANOVA, check homogeneity and normality, then report follow-up contrasts and effect sizes.
• Fit linear or logistic regression models, check multicollinearity and residuals, and produce formatted model summaries for publication.
• Conduct a power analysis to determine sample size for a planned experiment based on expected effect size and desired power.
• Perform a Bayesian re-analysis of an existing result, report posterior distributions, credible intervals, and Bayes factors.