If we are testing drug effects (A placebo vs. B), and here are the two options:

1. randomly select n users for A and n for B;
2. randomly select 2n users take A, then take B.

What are the pros and cons for each and which one should we choose?

I’m thinking 1 probably makes more sense because the results will be totally independent. Meanwhile, 2 gives us more samples. However, I don’t know how to think about in a more scientific way. And what if A is not placebo?

Would appreciate your insights!