If the answer to a question is: "collecting ducts" (ie plural) will the program recognize "collecting duct" (ie singular) as the correct answer, and visa versa? If not, can I use a wild card (*) to allow either answer to be correct? And if the answer to that is "yes you can" how do I set up the wild card in the question?
Using Moodle 2.2
What happened when you tried this? What question type were you using?
I'm assuming you're using short answer here?
For collecting ducts, I would probably set the correct answer as *collect*duct*. That would mark things like the following correct:
I was collecting ducts.
collected a duct
collect a duct
For the second one, I would not make that a short answer question unless you want to start grading conditional statements like, "yes you can except for when pigs are flying." Just make that a multiple choice question with yes and no as the answers. Or set it up as true/false. If you really want to use short answer then *yes* would mark your answer correct.
Hi Joshua,
1.- The problem with the use of the wild card asterisk is that the number of student answers which will be considered correct is potentially infinite. In the case at hand, "*collect*duct*" will accept as correct, for instance: I hate collectors of ductible metal.
2.- I'm afraid you misunderstood that sentence from the OP's message "And if the answer to that is "yes you can" how do I set up the wild card in the question?".
Anyway, it would be better know to wait for the OP's reply to our questions and suggestions.
Joseph
Good call on my misreading. Whoops.
Your first point is taken but I doubt that its an issue. I use a lot of short answer questions for pre-lab quizzes that I have. I don't ever have students trying to put in crazy answers to see what it will take. They're too busy trying to get the question correct. Collect and duct are pretty specific words and so I find it unlikely that would be a problem. Of course we can't know without seeing the questions...
As an example, I ask a question where the answer is *ball*trash*. Out of 285 attempts last semester, no one put in an answer that was incorrect that got counted correct because of the way I coded the answer.
Yes, this is an important point. When you are trying to match freely entered students responses like this, you only care about distinguishing the right and wrong responses from amoung the types of things that students acutally type. You don't care about marking all possible random strings of characters.
That is really why the OU's pmatch question type works at all. Sally Jordan has some interesting things to say about that on her blog: http://www.open.ac.uk/blogs/SallyJordan/.
Therefore, the only way to create questions like this is
Then, you may find you want to adjust your scoring rules, or you may find everything is OK.
Tim "Review how well the marking worked, looking out in particular for false posititives (wrong answers that were marked right) and false negatives (right answers that were marked wrong). The response analysis in the statistics report should help here."
A very sound piece of advice!
May I add that, besides the excellent OU's pmatch question type, there are 2 other similar short answer question types relying on so-called "regular expressions" to analyse the student's answers.
Joseph's REGEXP question type has been around for a number of years.
Oleg's PREG question type is more recent.
Each of those 3 question types has its own features, and for anyone needing a better analysis system than the plain SHORTANSWER question type, it is worth to invest some time to compare the three.
Joseph
Of the three the regular expression short answer seems the easiest to understand.
If you know regular expressions, then that is true.
My collegues Phil Butcher and Sally Jordan did a detailed comparison of a propriatory system, pmatch and regular expression matching, and found that while all three could give high quality matching of sentence-length answers, of the three, pmatch was the one where it was easiest to write the matching rules: http://oro.open.ac.uk/20868/1/Butcher_&_Jordan.pdf, or http://www.sciencedirect.com/science/article/pii/S0360131510000461 if you can get through the publisher's paywall.
So, I think it is worth having both pmatch and regex. Depending on the type of answer you want to match, either might be better.
