You simply are releasing less information with “continue” than “continue to N=n”

Here are pictures of an old Twitter thread on group sequentials leaking information. You find out you don’t have really big or really small effects, but a lot of uncertainty remains

One of my colleagues talks about this one a lot. Largely because they both are often wrong. Most experimental drugs don’t work. Placebo groups usually get great care with less risk of side effects.

There are so many times in life we can’t help others in need. I got this one.