Causal Inference

causal-inference

causal inference — rigorous statistical analysis methods and tools for academic and applied research.

Causal Inference

Apply statistical analysis techniques with statistical rigor and reproducible workflows.

Overview

This skill provides comprehensive guidance for causal inference in both academic and industry research contexts. It covers methodology selection, implementation, analysis, and interpretation following established scientific standards.

Research quality depends on methodological rigor. This skill ensures that analyses use appropriate statistical tests, handle assumptions correctly, report results accurately, and produce reproducible outputs that can be verified by peers.

The skill bridges the gap between theoretical knowledge and practical implementation — translating research questions into executable code, appropriate statistical models, and publication-ready visualizations.

When to Use

Activate this skill when the user needs to:

Design and execute statistical analysis analyses

Choose appropriate statistical methods for research questions

Implement computational workflows for data analysis

Generate publication-ready tables, figures, and reports

Validate research methodology and check assumptions

Reproduce or extend existing research findings

Core Capabilities

Method Selection

Recommends appropriate statistical analysis methods based on data characteristics, research questions, and assumptions. Explains trade-offs between different approaches and their suitability for the specific analysis.

Implementation

Produces correct, efficient code for statistical analysis analyses using standard libraries. Handles data preprocessing, model fitting, result extraction, and visualization in a single reproducible pipeline.

Validation & Diagnostics

Checks model assumptions, detects violations, and applies corrections. Runs diagnostic plots, residual analyses, and sensitivity checks to ensure result reliability.

Reporting

Generates publication-ready output following APA/journal style guidelines. Includes effect sizes, confidence intervals, and appropriate caveats about generalizability.

Example Prompts

Users might ask:

"Run a statistical analysis analysis on the survey data"

"Check if the assumptions for this statistical test are met"

"Create a publication-ready figure for the results"

"Design a power analysis for the proposed experiment"

"Review the methodology section of this paper for issues"

Best Practices

Pre-register analyses — Define analysis plans before collecting data to prevent p-hacking and HARKing

Check assumptions first — Every statistical test has assumptions — violating them produces unreliable results

Report effect sizes — P-values alone are insufficient — always report effect sizes and confidence intervals

Use reproducible workflows — Script every analysis step so results can be exactly replicated by others

Document decisions — Record every analytical decision and its justification in a research log

Common Pitfalls

⚠️ Using p < 0.05 as the sole criterion for significance — consider effect sizes, confidence intervals, and practical significance

⚠️ Not checking for multiple comparison problems — running many tests inflates false positive rates

⚠️ Confusing correlation with causation — observational data requires careful causal reasoning

⚠️ Cherry-picking results — report all analyses, not just the ones that support your hypothesis

Output Format

Generates analysis scripts, statistical output tables, diagnostic plots, and narrative interpretation. Follows reproducible research standards with version-controlled analysis pipelines.

Prerequisites

Research data in accessible format (CSV, database, API)

Clear research question or hypothesis

R/Python with relevant statistical libraries installed