I am a little bit confused. I always model square holes like this: 
But is it correct to model those holes in that way?:
I would be really happy to find out which method is the best. Because method like this:
Is really attractive, and easy to bevel, meanwhile modeling like this:
requires less poligons. So what to choose?
Both are possible but in my opinion it's better to choose the first solution (on the left) for this kind of object:
For example if you want to cut off a second window on the left of the existing one it will be much easier with this kind of topology. Same thing if you want to add some shapes like an edge, etc:
Also you should not worry too much about the amount of polygons in that case, it won't make a big difference.
But for example for the end of a pan handle it would make no sense to use this kind of topology because it is inadequate, it would prevent to easily add edge loops:
In that case a more correct topology would be the one you show, this kind of topology will allow to easily add an edge loop with a simple CtrlR :
