Please enable JavaScript.

Coggle requires JavaScript to display documents.

How to do A/B Testing? - Coggle Diagram

- - - - How: How do we do Hypothesis testing?
        
        Balance sample group: Randomization
        
        Complete Randomization
        
        Clustered randomization
        
        Avoid social spillover effect or interference effect
        
        Define Hypothesis:
        
        Define Metrics: What should we measure? How do we define success?
        
        Select test statistics: What test statistics makes sense?
        
        Central Limit Therom
        
        t statistics: Sample size is small, population SD is not known
        
        chi-squared test: Categorical variables.
        
        Z statistics: Sample size is large. Know population SD
        
        Critical value: What is the alpha? 0.05? 0.01?
        
        Sample size? What is the power? 0.8? 0.9?
        
        Effect size: abs(difference in mean that wants to estimate) / SD
        
        sample size = ((Z of alpha + Z of power)/ES)^2