Quantitative Workplace Experimentation and Power Analysis Script

Role: Principal Experimentation Statistician with extensive expertise in causal inference and organizational trial design.

Context

• Study initiative: {{experiment_name}}
• Cohort profile: {{sample_population}}
• Key performance indicator: {{primary_metric}}
• Historical reference baseline: {{baseline_rate}}
• Target minimum detectable effect: {{minimum_detectable_effect}}
• Significance threshold: {{alpha_level}}

Task

Author an end-to-end Python statistical execution script that computes sample size requirements, performs pre-trial balance checks, and implements regression-adjusted hypothesis testing for {{experiment_name}} evaluated across {{sample_population}}.

Method

1. Import production-grade scientific libraries (scipy.stats, statsmodels, numpy, pandas) with deterministic random seeding.
2. Ingest parameters for {{primary_metric}}, {{baseline_rate}}, and {{minimum_detectable_effect}} to calculate statistical power curves across sample sizes.
3. Implement two-sided prospective sample size calculations targeting 80% and 90% power given {{alpha_level}}.
4. Construct data hygiene and outlier detection functions tailored to {{primary_metric}} distributional quirks.
5. Write baseline balance assessment routines (standardized mean differences) comparing control and treatment cohorts.
6. Implement Ordinary Least Squares or Logistic Regression estimation with robust Huber-White standard errors.
7. Compute point estimates, confidence intervals, and p-values alongside Cohen's d effect magnitude benchmarks.
8. Generate formatted CLI diagnostic summaries and structured logging of statistical conclusions.

Constraints

• Code MUST be fully functional, modular Python 3.10+ without pseudocode or undefined variable stubs.
• Script MUST handle both continuous and binary distributions based on {{primary_metric}} characteristics.
• MUST NOT use deprecated functions from older scipy or statsmodels releases.
• Provide clear inline docstrings adhering to Google Python style guidelines.

Output format

Provide a single production-ready Python script organized into:

1. Configuration block defining hyper-parameters from {{baseline_rate}}, {{minimum_detectable_effect}}, and {{alpha_level}}.
2. Power & Sample Size Calculation Module (min 20 lines).
3. Statistical Testing & Estimation Class (min 40 lines).
4. Execution Block (if __name__ == '__main__':) demonstrating end-to-end analysis on mock synthetic data.

Self-review

• Confirm all formulaic power calculations explicitly match standard two-tailed formulas.
• Verify that {{sample_population}} and {{primary_metric}} variable names are consistently utilized throughout the code.
• Ensure error handling exists for degenerate variance or underpowered sample sizes.