I've noticed the data, and yet, one expects N+N and S+S to be more common even if socionics has the right idea. For one thing,
'coincidence of meanings' between MBTI perceiving functions and socionic irrational functions may be sort of low; for another, educational background skews results a bit, and would certainly have some bearing on who marries whom; lastly and most importantly, more similar minds would have readier semantic exchanges, reducing chaos early on, without necessarily implying superior compatibility. To wit, if everyone dates N and S types in equal proportions, then perhaps N types like other N types better; but if N types typically date other N types, then it would follow that N types are predisposed to marry N types, and the numbers would look better even if the divorce rates were average. I just find it sort of intriguing that, before creating socionics, Augustinaviciute examined marriages which had grown inexplicably antagonistic... if all unhappy marriages ended in divorce, the stats would tell us more. As things stand, it's hard to extrapolate the data to a statement of what sort of match would be
desirable. We can only say what is expected, and that doesn't disagree with a priori assumptions either
Also, forgive me, isn't your wife SEI? When you had mentioned her, you didn't sound discontented... would you be willing to comment further? Do you suppose you might have been happier with another intuitive?