Comments on: Experiments in teaching: am-I-ready-for-this? quiz outcome

By: Adam Smith

Adam Smith — Thu, 14 Oct 2010 15:12:25 +0000

Disclaimer: I am not actually one of your students.

In my experience teaching a similar (not required, but “core”) class at Penn State, I’ve found that lots of students who would have done fine on a test like that were still unprepared for the class. [There are also "false negatives", although I think those are less problematic since if a student really feels she understands the material, she can always talk to you personally or take the class next year.]

Here are the “undergraduate” skills that I’ve found most consistently problematic:

- turning a vague description of an algorithm into a clear, unambiguous one (e.g. pseudocode)
- writing proofs
- distinguishing correct proofs from incorrect ones
- analyzing simple algorithms using ideas like loop invariants or induction
- writing/understanding recursive algorithms

That is, almost all of my incoming students can do the mechanical steps (apply the master theorem, sorting functions, etc) pretty well. It is the slightly less structured tasks at which they perform very poorly unless they have taken the “right” kind of undergraduate algorithms class.

I haven’t tried in-class tests, but I find that the score on the first few homework problems where students have to write algorithms and proofs is a decent predictor for overall performance in the class. Short tests like the one you gave are also skewed against incoming students with poor English skills, who can’t read and process the questions fast enough.

Perhaps, since you will have the data at the end of the semester, you could look at what best predicts the difficulty a student will have in your course: the first homework or the in-class test? I realize that homework has its own problems as a means of evaluation: plagiarism, etc.

By: Glencora

Glencora — Thu, 14 Oct 2010 14:57:29 +0000

For additional context:

In my experience, if something does not affect a students grade, then they (on average) will not put in enough, or any, effort. Even a small effect on the grade (5%) appears to work.

Last year, I had an ‘Assignment 0′. Like this quiz, it covered background material with harder questions afforded by the extra time the students had. Having this assignment seemed to have no effect. Students who had (significant) difficulty with the quiz and perhaps should have waited another year to take the class did not.

By: anonymous

anonymous — Wed, 13 Oct 2010 22:54:33 +0000

It is really quite absurd to give an exam on a semester’s worth of work when you have not taught the material to the students nor explained the level of detailed knowledge expected yet plan to include it in an evaluation of their performance in your class!

By: honez

honez — Tue, 12 Oct 2010 22:47:37 +0000

This quiz should not count toward our grade. It could still be a guage to use to decide whether the class is for you or not, but if had not been worth any points, it would have taken the pressure off.

By: victim of master theorem

victim of master theorem — Tue, 12 Oct 2010 02:33:09 +0000

I agree with Soroush. For many students who are new to the graduate program at a state university, the pressure of suddenly changing lifestyles combined with, what appears to be, a very important deciding factor in our success for this class can lead to false negatives. I was one of the middle-of-the-road students and I think that about half of my wrong answers were simply stupid mistakes. However, I’m not sure a take-home test would be the best solution either. It would be too easy for students who value grades over understanding to fudge their way through without double checking what they do and do not actually know.

Perhaps the best solution would be somewhere in between? Find a way to do the quiz in class but take the pressure off?

By: Shahed

Shahed — Tue, 12 Oct 2010 01:24:44 +0000

well, it was useful.
I would not argue on if it were “easy” or “hard”, but it atleast ensured reading (revisions of the previously learnt materials) that might be necessary for this course onwards. But I agree, the results might not a good representation of the class performance.

By: by-partite

by-partite — Tue, 12 Oct 2010 01:15:12 +0000

The questions were quite easy. Except the wording of “a matching”!?1? But the idea was good to bring students up to speed with the foundations.

PS: I wanted to write here earlier, but didn’t know if you’d mind

By: Soroush

Soroush — Tue, 12 Oct 2010 00:57:31 +0000

It wasn’t useful I think. It was pretty easy. Instead you can prepare an assignment with harder question, and ask students to work on them individually at home. This kind of “quiz” has lots of stress with it, and the result might be fake.

p.s:will you post a blog regards “solving problem session” at the end of quarter?

btw, I still have problem with q.3.e