How cetinss/baron-kenny-mediation-analysis works
This is a research-analysis package and collection of study scripts, not a deployed service. The principal workflow builds a cleaned analytic sample from the synthetic caffeine dataset, estimates Baron–Kenny paths with bootstrap inference, evaluates hypotheses and diagnostics, then writes reports. Separate branches compare specifications, probe sensitivity, and validate related analyses against LEMURS and NHANES data. The README documents these capabilities; sampled excerpts show only part of their implementation, so uncertain internal call links are omitted.
Main entry points: cli.py Analysis CLI · outputs Analysis Outputs · data.py Sample Builder · sensitivity_analysis.py Sensitivity Analysis
Components 20
Primary Analysis
- Analysis CLI
src/bkmediation/cli.py - Synthetic Dataset
data/synthetic_coffee_health_10000.csv - Sample Builder
src/bkmediation/data.py - Analysis Configuration
src/bkmediation/config.py
Estimation and Checks
- Descriptive Statistics
src/bkmediation/descriptives.py - Mediation Estimator
src/bkmediation/mediation.py - Hypothesis Rules
src/bkmediation/hypotheses.py - Model Diagnostics
src/bkmediation/diagnostics.py
Robustness Studies
- Specification Comparisons
src/bkmediation/comparisons.py - Sensitivity Analysis
src/sensitivity_analysis.py - Benchmarks
src/bkmediation/benchmark.py
External Validation
- LEMURS Dataset
data/lemurs/survey_syn_5.csv - LEMURS Validation
src/lemurs_validation.py - NHANES Data
data/nhanes - PHQ-9 Analysis
src/nhanes_mediation_analysis.py - Allostatic Load Analysis
src/nhanes_mediation_AL.py
Reports and Artifacts
- Run Provenance
src/bkmediation/provenance.py - Report Writers
src/bkmediation/reporting.py - Analysis Outputs
outputs
Other components
- Researcher
Connections 23
- Researcher → Analysis CLI: runs commands
- Analysis CLI → Sample Builder: builds sample
- Sample Builder → Synthetic Dataset: reads rows
- Analysis CLI → Mediation Estimator: estimates paths
- Analysis CLI → Model Diagnostics: checks assumptions
- Analysis CLI → Hypothesis Rules: evaluates rules
- Analysis CLI → Report Writers: writes reports
- Analysis CLI → Specification Comparisons: compares samples
- Analysis CLI → Benchmarks: runs benchmarks
- Analysis CLI → Run Provenance: records versions
- Report Writers → Analysis Outputs: writes tables
- Run Provenance → Analysis Outputs: writes metadata
- Sensitivity Analysis → Sample Builder: loads sample
- Sensitivity Analysis → Mediation Estimator: runs baseline
- Sensitivity Analysis → Analysis Outputs: writes checks
- LEMURS Validation → LEMURS Dataset: reads survey
- LEMURS Validation → Analysis Outputs: writes validation
- PHQ-9 Analysis → NHANES Data: loads tables
- PHQ-9 Analysis → Analysis Outputs: writes analysis
- Allostatic Load Analysis → NHANES Data: loads tables
- Allostatic Load Analysis → Analysis Outputs: writes analysis
- Analysis Configuration → Sample Builder: sets defaults
- Analysis Configuration → Mediation Estimator: sets inference