I think BCD are all correct.
If we change the 99% var to 95% var, the sample will increase, so that the test will be more reliable (answer B), which mean both the probabilities of type I and type II error will decrease. And both B and D are talking about Type I error. So I think BCD are all correct.
