Dean Vikram Amar (Illinois) has an excellent post on Above the Law about exam writing.  He offers four thoughts based on his experience as a professor, associate dean, and dean.  First, Dean Amar talks about the benefits of interim assessments:

Regardless of how much weight I attach to midterm performance in the final course grade, and even if I use question types that are faster to grade than traditional issue spotting/analyzing questions — e.g., short answer, multiple-choice questions, modified true/false questions in which I tell students that particular statements are false but ask them to explain in a few sentences precisely how so — the feedback I get, and the feedback the students get, is invaluable.

Second, Dean Amar articulates an argument in favor of closed book exams:

There are many possible explanations for this decline [in bar passage] (and no one knows for sure all that is going on), but surely having students experience some closed-book testing on various legal subject matters prior to the high-stakes bar exam is an idea worth pondering.  And if, as I suggest above, professors move beyond the single, final exam tradition, then there is more opportunity to employ a variety of test types.  For example, a course might feature a closed-book midterm and then an open-book final.

On this point, I would probably flip it around: make the interim assessments open book or even no-credit, and the final exam closed book.

Third, Dean Amar suggests that using “ripped from the headlines” fact patterns are generally beneficial, if done right.  He writes:

I think this is generally a great idea — it simulates what law students will be doing when they graduate, and also helps engage students during the exam itself by making them feel that they are weighing in on important issues of the day.

I agree with his words of caution, however.  “Real world” fact patterns can be messy and address issues the professor did not cover in class.

A somewhat related word of advice from me.  I recommend avoiding character names rooted in fiction, such as Game of Thrones, Sopranos, or Law and Order.  While constructing such a story may be fun for you, it can be distracting to students, who may feel compelled to be equally cute and witty in writing back.  In addition, avoid telegraphing equitable positions through party names, such as “Big Bad Corporation” in a Business Organizations problem.  And hopefully this goes without saying, be careful in criminal problems that you don’t give names to suspects or defendants that convey racial or ethnic bias.  For all of these reasons, I prefer the National Conference of Bar Examiners’ approach of not using names at all.  Instead, I use common nouns, like “painter” or “teacher” or “banker” to refer to characters.  When all else fails, use letters (Person A, B, etc.).

Dean Amar’s final piece of advice is to put students on proper notice.  I echo entirely his points here.

And finally, when it comes to putting in sufficient time and care in drafting to be fair to students, attention need be paid to the clarity of the instructions the students are given.  If answers to some questions will be weighted differently than answers to others, instructions should make that fact, and the relative weights, explicit.  If word limits for answers are prescribed, call them limits (not guidelines), and don’t offer a word-number range but rather a specific maximum.  And also specify the sanctions triggered when a student goes over (e.g., “I will stop reading when the answer hits the word limit,” or “I will downgrade an answer by a full letter grade for every X words over the word limit,” etc.)  Finally, instructions should guide students as to what they should do if they feel they don’t have sufficient information to properly answer a question.  Should they assume information and make their assumptions explicit?  Should they describe the additional information they would need or want, and explain its relevance?  I am not suggesting that there is one right way to address this and other foreseeable reactions students might have to particular questions; I am saying only that significant time that drafters spend on anticipating potential wrinkles students may encounter when they take the exam often saves more time (and hassle) during the grading and explanation-to-students process afterward.

In sum, giving careful attention to final exam writing is critical to ensure that the results do what we want them to do: to sort students based on ability and to give students and us feedback on how they performed in the class.  Indeed, on this point it is important to keep in mind the learning outcomes a professor established at the beginning of the semester.  The final exam should measure the extent to which those outcomes were achieved and by how many students.  Then, this can create a feedback loop so that the professor can use the results to improve the course for the next group of students.  For example, I saw on my Evidence exams this semester some difficulty with non-character impeachment, so I plan to cover this in more depth with my Fall class.

A final point.  Students take final exams seriously, and we should too.  This means creating them with care, administering them with proper protocols to ensure fairness and integrity, and grading them in a consistent manner using tools like rubrics.