LR consistently 7 wrong, not restricted to any question type in particular (My Method is complete recognition of question type and go from there with my set processes). RC is mixed but anywhere from 6 to 13 wrong and LG is also mixed with best 3 wrong but ...