Trouble with a dichotomous decision study

Any help appreciated: Have a study with video clips and a Yes / No answer. There is a correct answer. There are ~20 (but this can be changed) clips evenly distributed between showing the finding ("yes") and the normal ("no"). I've been looking online for any examples of analysis of similar things and struck out. I expect close to 100% correct answers and would like to show that this is not due to chance. There are 4 groups taking the test, each with about 20 people. These are separated by a year of experience (total 4 years of training). This is a blinded study (individuals will be identified by year of training only).