Welcome new readers!

The "New to the blog? Start here" page will give you an overview of the blog and point you to some posts you might be interested in. You can also subscribe to receive future posts via RSS, Facebook or Twitter using the links on the right-hand side of the page, or via email by entering your address in the box. Thanks for reading!

Friday, April 16, 2010

When is an exam "too hard"?

By now, you may have heard about the biology professor at Louisiana State (Baton Rouge) who was removed from teaching an intro course where "more than 90 percent of the students... were failing or had dropped the class." The majority of the comments on the Inside Higher Ed story about it are supportive of the professor, particularly given that it seems like the administration did not even talk to her about the situation before acting. I tend to fall in the "there's got to be more to the story so I'll reserve judgment" camp but the story definitely struck a nerve with me, partly because I recently spent 30 minutes "debating" with a student about whether the last midterm was "too hard" and the whole conversation was super-frustrating.

To give some background: I give three midterms and a cumulative final, plus have clicker points and Aplia assignments that make up about 20% of the final grade. I do not curve individual exams but will curve the final semester scores if the class average is below 78%; I make whatever the class average is a C+ since my department has a general policy that we strive for class averages around 2.3 or 2.4 in the principles courses. Because 20% of the semester grade comes from things that are basically participation-based (i.e., most students get over 90% of those points if they just show up), I tell students that they will almost always do better overall than their midterm scores, even without a curve (and this is why I don't apply curves to individual exams).

Given all this, I tend to write exams simply based on what I think students should reasonably know and be able to do in 50 minutes, given what we have discussed in class; in the back of my head, I usually think that "good" students who have been consistently coming to class and general understanding stuff should be able to get around 85-90% and A's are for the students truly "get it". But sometimes I mis-gauge what my students can do and that is why I have a curve. The average on this last exam was admittedly quite low (63%), much lower than the typical average on my exams (the average on the first midterm was 75%), and that tells me that I over-estimated what my students could do. Many students felt that the main problem was that the exam was too long to finish in 50 minutes and that's probably a fair criticism - the exam had the same number of questions as all the others but there were a lot of multiple-choice questions about elasticity that I myself had told them were the sorts of questions that would be easier if they drew graphs to see the answer, and I didn't think about how much time that would take them. But since everyone had the same amount of time, that is the sort of thing that washes out in the curve.

However, this particular student kept wanting to argue that I should make the next exam 'shorter'. I told her that the problem was not just the number of questions but how difficult the questions were, and I under-estimated how difficult the questions would be/how long it would take people to answer them; she said she thought the questions themselves were fine but there should be fewer of them. She didn't seem to get that if the questions had been different/easier, the number of questions might not have been a problem. When I pointed out that the low average on this exam would be adjusted in the curve, she said that she didn't think it was "right" for so many students to do so badly and then have to hope the curve makes up for it. I don't necessarily disagree, which is why I write my exams based on what I believe students should actually know, given what we've done in class, but I also pointed out that I don't always know with certainty what my students do and don't understand beforehand - that's the whole point of giving an exam! We went around in circles for a while - she seemed to want me to say I'd make the next exam shorter/easier, which I wasn't going to say, and I think we both walked away frustrated.

I know there are some disciplines (particularly sciences) where it is routine for the average to be quite low, on an absolute scale, and god knows, grad school was like that. But when I have an exam where the average is lower than normal, I start wondering all sorts of things - were the exam questions too hard? too confusing? have I not taught the material well? are students not working as hard as they should be? But after a certain point (and I'd say that for me, teaching the same class more than three times is about that point), I don't think a low average is a reason to second-guess myself, it's just one of those things that sometimes happens...


  1. As usual, Jennifer, you raise lots of important issues in your post. Let me respond to one about tests being too long, because it is something I have chastised myself about nearly semester for twenty years!

    Although it clear that students who prepare well for tests usually are able to complete what I assign, there are often several students who don't finish. I don't feel good about giving them lower grades because they work more slowly, whereas other students earn similar low grades but obviously have learned much less. I feel even more guilty when my children complain about similar situations in their college courses.

    All this is remind myself to try not to make my tests a "speed drill" and more focused on assessing learning.

  2. I've had that conversation, as well. And I also have had to try to figure out what to do.

    For me, there is (at least) not the constraint of a departmental target or expectation for the overall GPA in a particular level of a course. I tell my students that I will be happy if they all do well enough to het high grades, but I'm prepared to live with a situation in which they don't...after 30 years, I'm pretty well able to estimate how hard the exams are.

    But sometimes things go wrong. I can occasionally write tests that turn out to be harder than I expected. My solution (and I do not recommend it for everyone, but it works for me) is to offer students the opportunity to take a second test over the same material. If they do better on the retest, they get the retest score. if they don't, they keep their initial score. And I do this within a week of the initial test. (I only do two in-semester tests and a final, by the way.)

    What amazes me is how often students who have not done all that well (defined as low Cs and lower) do not take advantage of the opportunity. And there's nothing I can do about that.

  3. @Mark: You raise a really good point. I think a big part of why my tests are hard for students is that I try to write questions where they really have to understand deeply in order to do well. And that tends to mean that some students, no matter how much time they have, are just not going to be able to figure out the answers. So it shouldn't hurt to make the tests a bit shorter, giving them more time but hopefully still assessing learning...
    @Doc: I'd be curious to know more about how you do the re-test. I've thought about doing something like this but am a bit baffled by the logistics - do you write a new test? do you take class time to do the re-test? I'd have to think hard about how to do this with my class of 500...


Comments that contribute to the discussion are always welcome! Please note that spammy comments whose only purpose seems to be to direct traffic to a commercial site will be deleted.