Say me and my buddy throw 3 darts each. He only hits the target once and I hit it twice. So, in this sample, I hit the target 33% more than him.
How confident are you that I'm better at darts than him? Probably not very confident.
Now, what if we throw 30 more darts each? This time, I hit the target 20 times and he only hits it 10 times. In this scenario, I still hit the target with 33% more darts than he did.
But I bet you're a lot more confident in saying I'm better that I am better at darts than he is now. That's the power of power, my dudes.
Moral of the story: The more dart throws we add to our sample, the more confident you can be at saying the difference in skill between me and my amigo is "real" (aka. not due to chance; aka. statistically significant).
Now to bring it back to the question. They had a sample size of 40 here and, while there was a difference of 25% between the two groups, the result was not statistically significant. If we were to increase the sample size enough (just like we did in the darts example), a difference of 25% might actually be statistically significant.
https://step-prep.org/tutoring/
submitted by โlm4(23)
okay soooo 40 participants is too low of a sample size- but 80 back in the new B-adrenergic drug vs. albuterol question was fine? is there like a magic sample number that is considered large enough?