My suspicion on this is this - and I'd be curious to see if it looks anything like this when you look at the plays in your gameplan (on the run/pass distribution screen).
Let's say you have a run/pass ratio of 60/40. That means that run plays are weighted at 60 and pass plays are weighed at 40. So if you have one run play and one pass play in that situation, the run play would be weighed at 60 and the pass play would be weighed at 40, meaning there is a 60% chance that the run play would get called and a 40% chance that the pass play would get called. (This is of course ignoring the personnel distribution, or assuming that it is even across all personnel, or that both plays have the same personnel)
Now, let's say that you add another pass play to the mix - so you have one run play but two pass plays. The one run play has a weight of 60, but each pass play will have a weight of 40 - which means that the total weight is 140 points. So of those 140 points, the one run play actually only gets about 43% of the weight, and each pass play gets about 29% of the weight - but both pass plays together get almost 60% of the weight, so in effect even though you have a 60/40 distribution of run to pass the calculation gives you a 40/60 because of the play weight balance.
This actually is something I have given a lot of thought to, because I agree with you that (if this is indeed what is happening) it is counter-intuitive - the run/pass ratio should trump everything. I initially intended to put a governor on the play selection to force the play types to stay within their range, i.e. what a quality control coach would do in a game, but kind of got stalled because I was having a difficult time getting the math to work. This is probably something I need to revisit.