BigJim,
There are several ways to investigate such an issue.
First of all you must be in position to compare the "high percentage misses" between solitaire games (vs AI) and Pbem games (vs Human).
The above is somehow difficult (because of many reasons) and the most crucial factor is the experience of the player in both type of games (vs Pbem, vs AI).
What I propose you to do, is to experiment with test scenarios, in which you will create corridors (small maps) and you will choose and deploy the exact units (nation, type etc.)
Then you will start the game with player1 as human (player2 AI) and after several replays you will start the game with player2 as AI (player2 human).
The difference between these two modes is that the player who takes the first shot will also be the player who will make the "high percentage misses".
If you carefully examine the results of a series of these tests you will understand that there is no modification in the code (between the AI and Human).
In order to help you I did one series of tests with 6 Greek tank against the same Greek tanks (hit probability 62%-67%). My conclusion is that the game engine treats equally the Human and the AI.
cheers,
Pyros
here are the settings of the test: