The imbalanced power calculator, developed with Streamlit, Python, and R, was designed to address the need for a quick power analysis in the context of unequal group sizes in online experiments (aka A/B tests). While most online calculators focus on balanced experiments (where treatment and control sizes are the same), this tool allows users to assess the statistical power implications of intentional imbalances. The tool utilizes the R pwr package. It is accessible through a user-friendly interface and publicly deployed on Streamlit Community Cloud.

Link to the app

Github

Here is a screenshot of the application:

Input parameters such as control group size, imbalance factor, conversion rates, and significance level enable quick insights on the balance / power trade-off.

Updated: