Skip to main content

experimental-design

Experimental design, A/B testing, statistical hypothesis testing, and research methodology

Aller à l'installation

Informations de source

Dépôt
NeuralBlitz/Mito
Dernière activité de la source
22 mars 2026 à 13:29
Langue détectée de SKILL.md
anglais
Étoiles
0
Forks
0

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
experimental-design
description
Experimental design, A/B testing, statistical hypothesis testing, and research methodology
license
MIT
compatibility
opencode
metadata
{"audience":"data-science","category":"data-science"}
## What I do - Design rigorous experiments - Create proper control and treatment groups - Analyze experimental results statistically - Ensure internal and external validity - Calculate sample sizes - Avoid common pitfalls ## When to use me When running A/B tests, experiments, or any controlled study to measure causal effects. ## Experimental Design Types ### Randomized Controlled Trial (RCT) - Random assignment to groups - Gold standard for causal inference - Controls for confounding ### A/B Testing - Two variants (A = control, B = treatment) - Random assignment - Measure conversion/r engagement ### Multivariate Testing - Multiple variables simultaneously - Factorial design - Interaction effects ### Sequential Testing - Pre-specified stopping rules - Always-valid inference - Reduce sample size ## Key Concepts ### Hypothesis - **Null (H0)**: No effect - **Alternative (H1)**: Effect exists - One-sided vs two-sided ### Statistical Power - Probability of detecting true effect - 80% power is standard - Affected by: effect size, sample size, alpha ### Sample Size Calculation ```python # Example: Two-sample t-test from scipy import stats import numpy as np effect_size = 0.5 # Cohen's d alpha = 0.05 power = 0.80 n = int(np.ceil(2 * (stats.norm.ppf(1-alpha/2) + stats.norm.ppf(power))**2 / effect_size**2)) # n per group ``` ### Variables - **Independent**: Treatment - **Dependent**: Outcome - **Control**: Variables held constant - **Confounding**: Variables that affect both ## Common Designs ### Between-Subjects - Different people in each group - Pros: No carryover effects - Cons: Individual differences ### Within-Subjects - Same people in all conditions - Pros: More power, less variance - Cons: Order effects, carryover ### Factorial Design - Multiple independent variables - Main effects and interactions - 2x2, 3x2, etc. ## Analysis Methods ### Frequentist - T-tests - ANOVA - Chi-square - Regression ### Bayesian - Posterior distributions - Bayes factors - Credible intervals ### Bootstrapping - Resampling - Distribution-free - Confidence intervals ## Threats to Validity ### Internal - Selection bias - History effects - Maturation - Attrition - Instrumentation ### External - Sample representativeness - Ecological validity - Treatment diffusion ## Best Practices - Pre-register hypotheses - Use adequate sample sizes - Randomize properly - Monitor for issues - Report all results - Consider effect size - Replicate findings
Voir sur GitHub