Guide software can calculate the centroid of the guide star with sub pixel accuracy so you don't need really small image scale. I assume you're talking about a guide scope and not OAG? You'll probably get a variety of opinions but, assuming an imaging scope that has a FL that's not long enough to justify an OAG, I reckon you are better off with a small, light-weight, short FL guide scope that's less likely to cause flexure - even if the image scale is large compared to the imaging scope.
Here are some thoughts on the topic:
http://www.wilmslowastro.com/tips/autoguiding.htm
Cheers,
Rick.