Empirical data is great, but proving your point from first principles that both parties already agree on can be extremely powerful, as well…
 
…And deriving things is just so much fun! :3
 
I’ve learned to derive special relativistic time dilation from the constancy of the speed of light with just one illustration and a few lines of math, then from there similarly prove length contraction as a special relativistic effect from time dilation.
 
Then you can go on to prove general relativity with only the equivalence principle, then use the Ehrenfest Paradox to prove the non-euclidean nature of spacetime, and an accelerating room in vacuum to prove gravitational time dilation, which two things make up everything you need to know for a Schwarzchild solution…
 
I could go on for hours and other things that are fun to derive, but I think I’ll stop here, since those are mostly mathematical derivations and this isn’t a good format to do those in.
 

Because I read in a book.

@when she felt her wings unfold  
Twilight definitely is statistically sexy.

a followup question to  

We have the data. We eliminated the alternatives, and controlled for all the confounding factors.  
You are best pony (p=0.05).

I don’t know! Stop asking these difficult questions!