Let's talk about (academic) tests

In My Humble Opinion
1

I wasn’t sure whether to put this in GD or IMHO. Maybe we should have a forum placement exam? :stuck_out_tongue:

I don’t get tests – I’ve never really gotten tests. I know the first response is probably “well, you’re probably just not a good test taker.” And you’re right, I’m not. I’ve gotten plenty of A’s on tests, it hasn’t hurt my GPA or anything to any noticeable degree, and it’s usually only an issue if the class itself is bad anyway – but if I had to name my worst academic aspect, it’s tests and test-taking.

I know school itself isn’t representative of the real world, and isn’t expected to be. But I feel that tests are somehow even more artificial and much less representative of anything than the rest of school.

It’s not that all tests are bad. They have their place, they’re just used far too often and where they shouldn’t be. I can understand that some thing are so foundational, so time/pressure sensitive, or so sensitive to screw-ups that you need to make sure the material is known and under quick recall. Testing basic arithmetic? Fine, I can even understand minor quizzing on the absolute most fundamental calculus (derivative of x^2 and such). I support testing doctors and nurses because screwing up under pressure could mean someone’s health or life. I can see Astronauts needing to really know their stuff in case something goes wrong while in space. Foreign language naturally requires a lot of rote, so testing there makes sense too.

In our system we just seem to have this test mentality. A short reading quiz to make sure you actually read the material? Okay, whatever. But I don’t get English classes that have you analyze Shakespeare under a time constraint. I’m failing to see how this is a useful metric of anything. Even if it was open book, it’s mostly testing how fast you can make something up (unless you’re lucky enough to have a really profound and deep insight within 5 minutes or reading the question) and how well you can remember random quotes and what Scene/Act they’re from.

I’ve known plenty of STEM professors. The only ones who know trig identities off the tops of their heads are ones who use trig all the time, or who are strange alien creatures who find trig fun. Many of them say they look them up when they need them. So why are high schools giving kids 10 minutes to solve 50 trig identity problems?

I just don’t get why, for K-12 and lower-division college courses at least, we have this environment completely dominated by tests. Like every course NEEDS to make sure you can answer questions on a time constraint with minimal information at hand, like I’ll have to somehow get my way off of a deserted island by using a fourier series some day. Even a lot of jobs have tests nowadays (at least in the programming realm as I’ve been told). I just don’t get it.

So, argue with me or explain to me: what’s so great about tests? I’m not even talking about standardized tests (which I dislike even more), I just mean plain ol’ in-class tests.

2

Because they don’t know how to teach math.

Ideally, a test will require you to clarify and organize your thinking, so you can solve problems under pressure. It should lead to a deeper understanding by forcing you to carefully consider how you approach a problem and what the essential steps are to consistently solve it. Otherwise you can often just muddle through. Memorizing identities though is a shallow level of understanding.

3

I often test students about two levels below where I think they should be. My quizzes, for example, are to see if they’ve read the stories or poems we’re discussing each day, not what the stories mean or which fine and obscure techniques the poets used, though they really be able to handle stuff like that. When a student does badly on my quizzes, the reason is usually they haven’t read the literature at all.

Step ONE–read–the-literature.

4

Is your issue evaluating students–which has its own problems–or the specific constraints of timed, in class tests? I am going to assume the latter.

I see two functions of tests.

One, and this is not a little concern, is that the only way I can be sure a given student actually did the work is if I sit there and watch them do it. Even then they may be cheating, but it helps. If a credit for a course is to certify that Suzie can do X, then the only way for me to see if Suzie can do X is to watch her.

Two, tests create a certain type of teachable moment that is really valuable for some kids. They learn while they test because they suddenly have skin in the game: instead of passively receiving information, they are having to interact with it, make connections, figure it out. It’s easy to say “they could do that at home, on homework”, but many don’t. The isolation of a test, the focused questions, the necessity of taking your best shot instead of giving up and letting someone else make the call are all valuable things . There’s many kids who walk out of a test smarter than they walked into it.

I do agree that time limits are not particularly useful. I will always let kids come back to finish a test. I do recognize that this introduced the potential for cheating, but I think most kids use the time sincerely, and I think it’s worth it to give them that chance for deeper learning than to stop the cheaters.

5

Another relevant reason for testing is that it’s an assessment on how well the course outcomes have been ingrained into the student’s thinking. At my college and other colleges, instructors and professors have to state how they know the students ‘got’ the material, and testing is one way to do it.

The test scores usually correlate very well to how well the students did in other portions of the class - discussions, quizzes, group and individual projects, etc. If an ‘A’ student has a disastrous test, it usually means the test questions are written badly, and they get changed for the next semester (and sometimes the question is thrown taken out of the grading).

6

Tests are a feedback loop for both the students and teacher. If the test is appropriate for the material taught, it gives an idea of how much a student knows. I’d also track test questions closely. If a lot of students missed the same question, then either the question was bad or I wasn’t teaching that part of the course properly. Also there were times when the best students were missing a particular question but the not-so-good students were getting it right. That can cause some head scratching!

Also I can tell you from experience, if it weren’t for tests, many students would not study the material at all.

7

Jragon, what would you recommend instead of tests? How would you assess how well students know and understand what they’re supposed to know and understand, or how well they can do what they’re supposed to be able to do?

The answer to this question, and the extent to which tests should be a big deal, may well vary from one subject or class to another. In my own area (mathematics), well-designed tests are simply the best way I know to tell how well students have actually mastered the material.

If your beef is specifically with the in-class, timed nature of tests, I can say two things in defense:

#1: Very often the difference between someone who kind of knows something and who really understands something, or between someone who’s doing something for the first or second time and someone who’s done it before enough times that he’s really gotten the hang of it, is how quickly he can do it—whether he can do it without taking a lot of extra time to think it over, and in a situation where he may be under a certain amount of pressure.

#2: Having students take tests during class time is usually the only way I know for sure that what I’m getting is the work of the students themselves.

8

I think my problem is more what is tested than how it’s tested. For a math class, I don’t think making people do math problems is the best way to test something.

I think I like the way my upper division classes in college did it. I took a graphics course, for instance. While it was called “graphics” it was mostly a computational geometry and advanced linear algebra course. And yet, during the final or midterm I never once had to so much as add two numbers. No matrix multiplication, never once was I told to “calculate the radiosity matrix…” or anything else we covered. Instead, the instructor assumed that if we needed to know the specifics of how something is calculated we can look it up. Instead, he tested our broad, high-level knowledge and if we know how to apply it. Things like “using a Z-buffer, how would you…” or “how could you simulate a mirror without actually simulating reflection?” Or on the more mathematical slide, “assuming two things cannot be coplanar, prove…”

IMO, these questions are simply better, because they boil down the problem into parts that test whether you learned the course material.

This gets around a problem a class I was in had with a physics test. Over 85% of the class forgot the quotient rule and got a problem wrong on a test. However, basically everybody involved knew how to solve the problem (take a derivative, find the maxima), they just hadn’t used the quotient rule in a while. I’m in a math heavy field and I hadn’t needed the quotient rule in 3.5+ years, and was completely lucky that I remembered it. That’s why a lot of problems that use words like “calculate…” can go so wrong. A student may absentmindedly mis-add 3 and 5, an obscure property the didn’t know they forgot because it hadn’t come up in 5 years may suddenly be needed. It seems to me “Why does multiplying this term of a parabola do what it does (use mathematical notation if required)?” is a much better problem that “calculate the roots of this equation.” It boils the question down to a conceptual, rather than an executional level.

Like I said, I know plenty of math professors, and most of them say that they look up trig identities and tricky derivatives, so I fail to see why it’s important to be able to perform these on command – even if the test doesn’t revolve around them, why we’re asking questions that involve using them without a reference sheet. I know when I’m programming I always have a terminal open to use the man pages, or the language’s package documentation open. I just don’t see the utility in testing things at anything below a conceptual level.