A Tirade Against Exams

I don't like exams

Summer break has just begun. I managed to get away without any exams this semester, for the first time. In the past weeks, like every end of semester, I find myself thinking what an awful, ridiculous system these exams really are, especially in university. I’d like to try and articulate why.

I can imagine a university where exams are hardly even relevant because people only study things they find interesting, and only so long as they are interested. Such places exist (take Tokyo Shure for example).

However, most officially-recognized undegrad programs are still based on instructors providing students with pre-packaged chunks of information, and then judging whether each student has properly digested the information. This post will be about exams in that context; my point of reference will the linguistics BA program at the University of Leipzig. As far as I know, it’s as good an example as any of a normal undergrad program in science.

Exams are bad experiments

So why are exams a bad idea when you want to check whether a bunch of science undergrads understood what you taught them? Well, one part of the problem should be obvious to anyone with even a rudimentary understanding of science: exams are not very good experiments. There is no way to control for interference of irrelevant, extraneous factors. When scientists conduct a study, in any field and with any methodology, they seek to control for irrelevant interferences. For example, when psychologists test hand-eye coordination, they’ll do something like only taking right-handed people with healthy hands and eyes, in order to make sure that the results aren’t skewed by irrelevant differences between individuals.

You can’t do anything like that in exams. For example, one of my exams once took place at a time when I was infatuated with someone. I spent about a quarter of the exam staring into blank space and thinking about things quite unrelated to linguistics. As you might expect, my grades for that semester were not spectacular. This was not a reflection of how well I understood the material in question, but rather a reflection of how capable I was of concentration at the time of the exam.

Exam stress: an antidote for learning

Not only can’t exams control for interference, they create a strongly interfering, totally irrelevant factor: stress.

Exams cause those who take them to get stressed out, usually for weeks in advance.

Google the words “stress” and “learning” together. The first result I got (of some 25 million) was this site, which says “Stress can disrupt learning and memory development”. Huh. That sounds like a great way to lower people’s performance on a test.

One obvious remedy is to train people so they’re used to taking tests and don’t get so stressed out. This is what traditional schools do, and perhaps why they do it.

For some reason, that really doesn’t work for most people. I’m guessing that the way schools make a big deal out of exams rather trains people to think exams are a big deal and worry about whether they’ll pass. What also doesn’t help much is that the resulting grades are relevant to one’s progress in a degree program as well as one’s chances of getting accepted for further studies or a job.

Exams are bad science

But even if we accept that it’s schools’ job to prepare students for the stress of university exams, what are those exams preparing them for? Surely, it can’t be their future work as scientists. Exams are good preparation for bad science.

A scientist’s job is essentially the opposite of exam-taking.

Exam-taking is swallowing a more experienced person’s presentation of information (course material), then regurgitating small bits of it as closely as possible to the original (“the right answers”). Science is carefully considering information (raw data) and other people’s presentations of information (prior work), carefully deciding whether or not to swallow it, then, optionally, producing a novel presentation of the information (research), which is considered useless if it’s in small bits that are exactly like they were when you got them.

The whole idea of one person telling the beginners how it is and expecting them to accept it is bad science. Obviously, my instructors are far more experienced and knowledgeable than me in their respective fields. Still, it would not be very good if I accepted everything they taught me unquestioningly.

If I take my role as a budding scientist seriously, I should critically examine everything I am taught and decide for myself whether I agree or disagree (and why). Exams tell me the opposite, and it takes real effort to continue thinking critically while I am expected to soon be able to reproduce the instructor’s view.

Worse still, in some introductory courses, the theory being taught is not perfect: instructors use simplified or “toy” versions of the theories being taught, or perhaps just a rather recent theory which is more a work in progress than the final word about anything. Either way, attentive students might notice inconsistencies or incoherences. This is good for undergrads; they can be inspired and take the theories further. That value is diminished by needing to swallow theories whole for a test.

Some suggestions

I could probably think of another point or two against exams, but instead I will dedicate the end of this post to pointing out a few things that might make the situation better:

Abolish the importance of exam grades.

This is the most important thing, but also likely the most difficult. Exam grades should not be available to anyone but the student and instructor. It might make sense to indicate on a degree whether the holder’s grades were consistently above average — this might potentially be an indication of extraordinary ability. But knowing that even the best exams are inaccurate and susceptible to extraneous variables, it does not make sense to prefer B students to C students.

Make feedback the goal of all exams.

Finding out I got a C on an exam doesn’t help me improve. Telling me what the weak and strong points of my exam were, could. I learned a lot on the few occasions where I’ve asked an instructor to go over the exam and tell me what my mistakes were. This value as a learning tool is wasted by not presenting all exam takers with feedback. (Some instructors do this, but all should.)

Make some or all exams optional.

If the goal of exams is to give feedback, then save it for those who want it. Mandatory exams create unnecessary stress. There are plenty of other ways to run a system like the modular European Credit Transfer and Accumulation System, in which it is essential to judge whether a student really took part in their courses.

Replace some exams with real work.

Writing a term paper takes more effort than writing an exam, but you learn new things from it and experience something akin to actual academic work. Some disciplines have other “simulations” of real work which could be graded as tests. Sure, this requires more effort per test from the staff, but grading something other than an exam may be a welcome change. And perhaps a system could be created where more advanced students grade beginners’ work and get graded for their grading work (being real academic work practice itself).

Filter students in conversation, not testing.

I get the impression that one of the main reasons I had to take so many exams in the first year of my studies was to filter out students who are not really interested in the program they chose. (I’ve mentioned before that some people choose their major at random here, and if that’s as common as I think, filtering is a good idea.)

I imagine a ten-minute conversation with each student after their first semester could replace some or all of that testing. If the courses didn’t do the trick, simply asking the students if they want to continue with this major, and if yes then why, will get them thinking about those questions themselves. With all of the second chances people are given after failing, it’s their choice anyway; a short conversation could save a lot of exam creation, administration and grading. And of course, this could be done by advanced students as well as by faculty.


Clearly, all change in the university system is slow. Certainly, there are many different changes that can be made. I hope I have provided a few good points of critique and a few good ideas on how to improve the system. Further ideas, comments, and criticism are most welcome in comments.


9 thoughts on “A Tirade Against Exams”

  1. I think that exams should not be the holy grail that they are now.
    MIT, for one, use a system of exams but in the end you recieve a diploma that says “MIT graduate” without grade or being recognizes in any way for your grades. You must , however, to produce letters of recommendation from professors in order to proceed to grad school.
    I think this system is better because while it does give you feedback and allows your professor to know what’s your status.

    This way it allows them to know you have atleast a passing level of knowledge (and thus not entitled for the big bad boot) but there is no chase of “if I don’t get this test high i’m not going to get a good enough diploma to get a job/valedictorian/magna cum lauda”.

    In a society run by statistics, there is no escape from tests and grades but we still need to shift the focus towards personal interaction between a student and a professor.

    1. Also, in ethnonolgy and many other social sciences that relate to human behaviour in any way, people have come to the opinion that trying to produce an artificial surrounding as in an experiment is ‘fighting against windmills’. It’s an ancient technique. From quantitive analysis to qualitative analyses. I guess we will just have to wait a few decades until the system has come to the point when it can use its achievements.

      Elaye, you are right when you state that, in modern society, people cannot escape from tests and grades and they do rely on the importance of numbers a lot, because it can be useful. But still, in a competition at university it definitely is preventing people who are really good in there subject but who shit their pants before exams from doing what they are good in. So why not establish a different and more valid system which measures the variables that should be measured?

      Sure it takes time, but you have to start somewhere.

  2. But
    a) the university must make sure that you really know something about the subject when the give you a degree (and you should not only know something but a kind of curriculum – at least for undergrads)
    and b) written exams seem to be much more objective than (random) conversations: exmas are documented and stored (i.e. the grading is replicable in principle) and some kind of personal bias is ruled out to some degree (some professors may not like piercings or some style of dressing – as a human you can’t get rid of your prejudices).

    1. Those are fair points. But to (a): term papers and other types of tests, not to mention degree theses, provide a better picture of one’s knowledge. And to (b): I suggested conversations for filtering out the students who chose the wrong program, not for grading. Even if exams are resistant to lecturer bias, they’re still very far from being good indications of ability. Test-taking is itself a skill and there’s no real good reason for it to be considered so important across all fields of study when it is utterly irrelevant to the professional life that follows.

  3. hi michael,

    your thoughts remind me a lot of a classic column called the professor fails the test by Roger Schank who went from AI to education. The final part below the questions is spot-on, I think, as well as hilarious – he is not able to correctly answer a textbook question about his own work ;-)

    While we never administered testing procedures in our seminars (opting for project work and reflections), after having spent some time “on the other side of the lectern” I am not sure we can get rid of testing anytime soon for a variety of reasons having to do with the very culture of academia:

    -> severe lack of resources.
    -> disciplining effect.
    -> make students learn some arcane stuff as a shared ritual of passage into the community. The more arcane the better.
    -> did I mention resources?

    btw the (“completely voluntary”) 10 minutes talks to beginners are overly politicized at the moment here in Austria as they seem to be the conservatives’ favored strategy for getting a foot into the door to get rid of the open admission policy altogether…

    Also there are structural problems even beyond those mentioned by Sebastian. It turns out that interviews are perhaps the most culture-laden ways of selecting for people in that they are most highly skewed in favor of people coming from well-of and academic families.

    While I like the idea of mentoring in principle, here it might just give the people with the (“wrong”) dialect and habitus the impression that they are not wanted…

    Your basic points about science resonate a lot with me… In curriculum development for our master program we seriously talked about how to acomplish “unlearning”/”deprogramming” of the kind of “skills” (e.g. regurgitation) that creep into the character of students during their bachelor… ;-)

    Karen Kastenhofer, whose original background is in biology and who is now doing social studies of biology, has done a study tracking how the first semester of biology in Vienna changes students’ expectations and values. In an interview with a professor he tells her that in a working group of four students for him the optimal number of students who think for themselves is 1 (at most) since this is quite enough… the others should simply do as they are told and there would be unproductive discussions as soon as there are two of them…

    not that I agree with this person at all, but we should recognize that a lot of science is about rote-work and that there may be more method to the testing madness than we are giving credit for…

    (a point made recently by Jeffrey Schmidt in his Disciplined Minds)

    on that happy note

  4. An excellent perspective. Exams are supposed to be a valid evaluation , and they are not. Perhaps the intention behind the evaluation is flawed in the first place. “We want to make sure you are learning”. Is not the student their own best judge? But this is only if the student wants to learn.

    Do I need to take an exam to learn linear algebra, if I am wanting to learn it because I actually want to use the techniques? How I wish that I could have taken a course on linear algebra when it came to be that I did want to use it. Instead, I had to pour through books and teach most of it to myself again.

    I think andreas is right.
    -> severe lack of resources.
    -> disciplining effect.
    -> make students learn some arcane stuff as a shared ritual of passage into the community. The more arcane the better.
    -> did I mention resources?

    But aren’t 2 and 3 connected in an unsettling way?

Comments are closed.