Binary vs. Continuous Outcomes
Cameron highlights the crucial distinction between binary and continuous outcomes in A/B testing, emphasizing that many misleading headlines arise from misinterpreting binary results as continuous improvements. He stresses that while binary outcomes simply indicate whether an effect exists, understanding the magnitude of that effect requires a much larger dataset and a different analytical approach. Additionally, he discusses the ethical considerations and strategies for determining appropriate sample sizes in testing, particularly when revenue is at stake.In this clip
From this podcast
Data Skeptic
Bayesian A/B Testing
Related Questions