Dick Hansen nailed the problem. Here's what can happen...

As the servo rotates, the deflections will not be the same in the middle due to the rotation of the pushrod - even if the neutral point and full throws are the same for both halves. The longer the pushrod, the smaller the difference will be. personally, I gave up on using forked pushrods because I'm anal-retentive and even if I couldn't see a difference, it bothered me to know that there is one.

Now I go with two servos.

Another way, as Dick said, is to mount the servo sideways or have the servo go to a bellcrank mounted vertically. I've done the bellcrank thing, but it's more weight, more slop and more things that can go wrong. So the two servo method works for me.