Preamble
To test a seasickness remedy you 'volunteer' a batch of students to embark in a small boat on a wild sea.

Controls
It is not good enough to give them the pill. You don't know how many would have been seasick, or not, without the pill. You gotta have controls, subjects who don't get the pill:

Randomize
Don't cheat by giving the pill to subjects who look like good seamen, and using as controls subjects who look dismayed at the mere idea of going out in a boat. Avoid this, and other biases, including ones that didn't occur to you, by randomly dividing subjects into pill and control.

Blind
Even controls don't suffice. Seasickness is partly psychological. If you really believe the seasickness pill works... It works. Blind the experiment. Make up a dummy (called placebo) which looks like the real thing. Randomly decide who gets which. And don't tell them. Afterward compare the treatment and control groups.

Double Blind
At what point do you judge a victim seasick? Look uncomfortable? Turn green? Go to the rail? Go to the rail and perform? Someone has to make the call, and it shouldn't be a person who expects to make a fortune marketing this pill. Not only shall the subject not know which pill (blind), whoever calls the shots shall not know either (double blind). Lock the document telling which subject got which pill in a safe till the experiment is over.

Probability, Etc.
A cynic about patent medicine doesn't believe the pill is any better than peppermint candy. Who got seasick and who didn't was predetermined when you randomly assigned pills to subjects. Suppose out of 8 subjects, 4 are prone to seasickness, 4 are not, and the four weaklings chanced to be the ones who got the placebo. One time in 70, the four weaklings get the placebo. One time in 70 you will be fooled.

Conclusion
If indeed all four controls get seasick but none of the four treated subjects get seasick, you are in this fork:

EITHER

The pill really helped

OR

Something unlikely to the extent of 1/70 happened.

That's the evidence: Take your pick.