** Using WeBWorK **

In almost all cases, I like to give my students unlimited attempts to get each problem correct. The only exception is when I am doing true/false or multiple-choice questions. For example, I gave them a homework problem last week that was basically this:

Here are ten series, tell me if each one converges or diverges.

Obviously, I can't give them unlimited attempts on this. They could all get 100% in about 60 seconds without giving the problems any thought at all. So what I did was to set $showpartialcorrectanswers = 0 so that they could not see which were right and which were wrong. Then I give them something like 5 tries to get the whole batch of 10 problems. That gives them plenty of chances to correct mistakes, but prevents brute-force methods. I've given this problem in past semesters and thought it was great. When told that they did not yet have full credit, students were forced to re-evaluate *all* their answers, which not only helped them find mistakes, but also helped reinforce the ones they already had.

But this time...

Some students noticed, and I had told them before on previous assignments, that once you had gotten an answer correct, you were given credit for the answer even if you later submitted over it with a wrong answer. So some clever students discovered --- and were nice enough to tell me about --- the loophole: submit all "converges", then change them all to "diverges" and re-submit. Then you'll be guaranteed to have answered each question correctly at some point and thus have 100%.

Is there any way to turn off the above behavior, and have WeBWorK only count the most recent attempt? It appears that the answer has something to do with $problem->status in the file lib/WeBWorK/ContentGenerator/Problem.pm, but I'm not quite confident enough to go in there and start changing things.

Better yet, is there any other way to give multiple attempts on true/false questions? I know that sounds like a funny question, but I've given this question to previous (less clever/devious) classes and it's been a very successful one, in the sense that it's gotten them to work hard.

Thanks very much,

Doug

<| Post or View Comments |>

WeBWorK does not work this way, i.e. "Some students noticed, and I had told them before on previous assignments, that once you had gotten an answer correct, you were given credit for the answer even if you later submitted over it with a wrong answer."

What the standard problem grader does is give the student the maximum of (1) the problem score on the current attempt and (2) the recorded score from previous attempts. Thus to get full credit, a student must do the problem completely correct at least once. In fact for scoring purposed WeBWorK does not record results from indiviual parts of problems. Also, if you want, you can write custom graders. Quite a few series problems use a custom grader that, as the grader reports, works as follows.

To get full credit, all answers must be correct. Having all but one correct is worth 50%. Two or more incorrect answers gives a score of 0%.

Arnie

<| Post or View Comments |>

Students were showing 90% credit, then they'd change one and still have 90%. I took that to mean that they had changed a right one to wrong, but were still getting credit for the right one. But that's not necessarily the case. I guess I had incorrectly interpreted that kind of behavior on previous problems too.

I feel much better now, as I can continue to use these problems, which I really like. And thanks also for the tip on custom graders. I'll look into that.

Doug

<| Post or View Comments |>

<| Post or View Comments |>

    install_problem_grader(~~&std_problem_grader);

take care of this problem? The student must get all the parts right to get credit for the problem. I always include this in T/F and multiple-choice questions. I just tested entering all true and then all false, and got 0% score both times.

<| Post or View Comments |>