It's usually this time of year that makes me think the most about cumulative exams. In some ways, I really like the idea of a cumulative final exam, because it is an amazing opportunity to have the student to see the entire scope of course material, and understand how it all fits together. Granted, we have to write prompts and exams which ask this of the students, and we have to make this clear in class and in review sessions: "The reason we're doing a cumulative exam is for you to take a step back and be able to connect all of the course material, and see this last third of the class as it pertains to the first two thirds, instead of as an entirely separate segment."
Imagine if we had three exams then - would the second exam also be cumulative? I'd have think really hard about the ordering of the information in the course. I'd also have to think hard about how to proportion the information in the 2nd and 3rd exams, so as to require an understanding of the information in the previous exam(s) but to rigorously test their knowledge of the more-recent information. In short - all of this underscores my beliefs that tests should be written very, very carefully, and each question should map directly onto a learning objective for the unit, for the class, and for the field of study.
I've also heard some pretty convincing arguments against having a cumulative exam. It's much more stressful for the students, who are simultaneously being slammed with final projects and projects in all their other classes. Second, it means that you can't test the back-half of the class information as thoroughly as you tested the first-half of class information. This is because there is an entire test dedicated to the first-half of the course, but only a percentage of the second test dedicated to the second-half of the course.
In the end, even though I've opposed cumulative exams in the past, I actually think that I favor cumulative exams. Ultimately, the catalyst for this switch in perspective was the realization that if I wrote the tests carefully, a student could reflect on their past tests to guide their studying for the next test. For example, a student might receive feedback on the first test that she needs a more thorough understanding of a theory. Even though this is true, the student will be incentivized to use this feedback more thoughtfully if she knows that she may be tested on this theory in the second exam. This is a better exemplar of how feedback works in the "real world". This also means that the teacher's feedback will be used, instead of merely noted.
In other news, I went to the Celebration of Teaching Excellence, where I was one of four graduate students recognized for their work. I was truly awed at the stories I heard there about other graduate students, professors, and advisors. Hearing about what other instructors are doing inspired me to think even more critically about what I do in the classroom, and to continue to challenge myself to grow, both as a student and as a teacher. Congratulations to all of the other winners - keep up the great work!