I'm in a similar situation: 3.39 GPA, testing around 176 average. I've looked extensively at the data on LSN, and to me it seems to indicate that, if you are going to split, high LSAT is much better.
My reasoning goes like this: If you look at the data points for top 14 schools on LSN, you'll notice a relatively huge cluster with high gpa, but mediocre (for top 14) LSAT scores. Then you'll see that there are often few or even no data points with low (for top 14) GPA and high LSATs. Now obviously, to protect means, schools would love to get the best of both worlds: GPA and LSAT at or above their current means. However, schools can't guarantee that admission equals attendance, and many of those 'ideal' candidates will be admitted to higher ranked schools.
So an inferior, but I would guess fairly common, way of protecting means would be to balance splitters of different kinds against one another. If the mean GPA is 3.5 and the mean LSAT is 172, then admitting one 3.7/168 and one 3.4/175 would leave the mean unchanged. Also, these are the type of students less likely to be admitted to higher ranked schools, so I would imagine this type of approach might also help protect yields.
So if this makes sense, and if this actually occurs (I'm no statistics major, so I wouldn't know), I would guess you are a breed of splitter that is more sought after. Still, one is probably going to be waitlisted quite often in this situation.