As a first level test, if this is a pick 1 from 4 multiple choice or similar, completely random responses are probably as useful as anything else.
These should enable stress testing of correct answers, incorrect answers, and responses at inappropriate times when input is not expected which covers, quite a lot of the bases. It will also give you a good idea of the meantime between failures to determine how often you should restart the, system the show is running on.
A second QLab workspace can generally be used for this, as, it allows simulation of OSC and Midi interactions, and also screen interactions through scripting.
Mic