Surround speaker placement height

Hi all, This is the first time I've owned surround speakers, so I'm treading in new waters here. I've read several different things on the proper height of on wall surround speakers. First is, 2 ft above seated position, second is tweeter at ear level while standing, and third is 6-7 ft from the floor. All three positions would put the speaker at different heights.(The WAF would be in the negatives if I put three different holes in the wall.) The 6-7 ft comes from the speaker web site(Polk), the ear level while standing comes from someone who owns the speaker (FXi5), and the 2 ft above is as I understand it, a general consensus when using box speakers on stands as rears. Any more input would be greatly appreciated.
I've had them in all three position and all work fine. Install however makes her happy and enjoy.
I prefer surrounds higher (6 feet up) but it is largely preference - it makes concerts sound very real as that is usually where rear sounds come from (above you, from the side and behind in the stadium or hall).

Ideally they should all be at equal height with the mains and center - giving the optimum perfect sound while seated. But I think this ideal configuration requires an enormous room - or else only one listener gets the right sound field. I suspect that dipoles mounted on side walls are the best option in a small HT (I don't have dipoles).

So I have also placed my L and R surround well back (almost in the proper place for rears) and consequently I have them directed aggresively outwards to reflect sound off the side walls to the seated listeners - this is far from ideal (in correct) but I really don't like speakers in the middle of the room pointed right at your ears at the listening position (it really limits the sweetspot to ONE person in the middle of the couch and anyone on the side just gets too much of earful on one side - overdone surround and it becomes distracting, IMHO!!).

Take a look at my setup on the virtual system - far far from ideal but I find it acceptable and most WAF friendly.
Scan Dr. Toole's trio of papers on the Harman site. He addresses this issue in Part One, I think.