In this talk, we will discuss the pros and cons of the weighted Z and the unweighted Z tests proposed for data analysis after sample size re-estimation in clinical trials. The weighted Z approach has been studied extensively by statistical researchers in the last 10 years, but many clinicians are hesitant to employ this test for comparisons of treatments. The sequential computed weighted Z tests can control the alpha level easily, but it violates the one-patient-one-vote principle. The weighted Z test, on the other hand, usually requires extensive simulations to justify alpha control. In this presentation, I will give in detail the motivation of the unweighted Z approach, and the penalty involved in practice.