I am interested in the effects of ever attending office hours on midterm grades.

(a) I randomly pick half of your names from a hat and then write your names on a piece of paper. I then immediately place this piece of paper in a locked cabinet and forget all about it. If I wanted to use this random assignment as an instrument to evaluate the effects of attending office hours on midterm grades, what do you think would be the bigger problem, relevance or validity?

(b) I monitor who has been coming to class and send special, individualized emails to students who come to class the most often, urging them to come to office hours. If I wanted to use the sending of an email as an instrument to evaluate the effects of attending office hours on midterm grades, what do you think would be the bigger problem, relevance or validity?