Alas, the perfect midterm is something will never be found, but might, at best, be approximated at a moment in time. I haven't gotten there yet, but every time I write a midterm, I try.
I wrote one this morning. But not a perfect one.
A perfect midterm is one that let's students who've worked hard and learned lots really shine. Their work stands out, and the midterm gives you a way to reward it. And the students who haven't bothered to come to class or do any reading, they fail miserably, and you can see it as they stare blankly at the page, unable to make a meaningful beginning.
Most students do a fair bit of work, and in my classes, usually come to class and even participate pretty regularly. They're usually reasonably engaged. The midterm should help them show what they know, and ideally, should help them put things together a little better.
Because the most important aspect of a perfect midterm would be that students learned something from writing it. Sometimes, you can help students put things together better if you work on reviewing for the midterm well. But there are so many variables that I couldn't tell anyone how to do that except for my own classes.
Most folks have a fairly set format for midterms. Mine--which I adopted from working with my mentor in grad school, and which he adopted from his mentor, and which she, well, I have no idea where she got it--gives students two separate sorts of tasks. The first task is a short identification and explain what this has to do with what we've been working on in class; the second asks students to do a sort of mini-explication with passages.
There's one more thing about the perfect midterm: it's relatively painless to grade. Mine isn't, but it's not too horrible, either.
If ONLY I could think of a reasonably valid way to give a scantron midterm in an English class! But I never learned anything useful from taking scantron exams (and trust me, I took my share), while I did, in fact, learn from exams in many classes.
My favorite exam question of all time (from my mammology class):
at the bottom of one page of a short answer exam: What's the most important question in mammology today?
And on the top of the next page, otherwise blank: Answer it.
Of course, I wasn't one of those smart students who previews the test before I started writing, and I wrote in pen, and didn't go back to erase. I remember that I answered the first with a question about the development of body temperature regulation (endothermy) in mammals. And then I answered the second with a review about what we knew about the development of body temperature regulation (drawing also on a couple paleo classes), and then talked about why that development was important to the evolution of mammals.
I think it says something that I remember the question and more or less what I wrote after almost 25 years. I know I'd been thinking about such things in different classes, but the question helped me put together what I knew and made me realize that I had actually learned a lot.
I wonder now, thinking back, whether the answers tended to be interesting or irritating? Did most students write about a real problem and give thoughtful answers? Or did most write vapid questions with really uninteresting (though perhaps correct) answers? And how would you grade those? (And heck, maybe my own question was pretty vapid, but it seemed important to me at the time.) It's the potential pain of grading vapid questions that keeps me from using that sort of question in my own classes. But maybe I should try it some time? Just trust that at least some of my students will come up with really good, interesting questions?