I gave you an example of a t-test and conclusion earlier in the thread, and even contacted you privately. You can even run a t-tests on the data you provided using the info I sent, and I did one for the first examples you provided. Just read up on t-test, they show exactly why percentage difference calculation for your type if data is a wrong approach. Alternatively you can Google it. Even MrSmith gave you a thread to build your work on.
No need to be rude 'mate if you disagree with people who want to help. We already know that you're biased, so there's no need to demonstrate more of that either