Hi Matt,
Open this thread for overview of quiz-nodetypes which need supporting to cover spectrum of quiz question types (see: https://mw.brookes.ac.uk/download/attachments/7899751/Respondus+Question... for a great overview of question types).

Nodetypes:
- Long answer/ essay questions (in concept: http://drupal.org/node/165459 people to enter complete sentences or paragraphs for their answer).
- Short answer (require people to enter a word or short phrase)
- Multiple choice questions with multiple correct answers,
- Multiple choice questions with a single correct answer,
- True or false (are self-explanatory - enter the statement in the question wording field and then select either True or False as the correct answer),
- Fill in the blank (in development http://drupal.org/node/329931 presents people with questions that contain multiple blank spaces that must be filled in),
- Matching questions (in development http://drupal.org/node/329931 questions require people to match items from two columns)

Greetings,
Martijn

Comments

mbutcher’s picture

mbutcher commented

This is a great start for a discussion. Here are some pre-coding questions we should try to answer:

(1)
On short answer, let's list out the features that short answer questions do NOT have in common with either "long answer" or "fill in the blank".

Is short answer different enough that it should get its own question type?

(2)
This is a valid question for true/false, too: Is it different enough from multichoice that it should get its own question type?

(3)
With fill-in-the-blank, should we allow manual scoring?

We might do this to allow "fuzzy" answers. For example, a quiz question might be:

"What 17th century philosopher wrote 'Meditations on First Philosophy'?"

We would want to be able to support "Descartes" as an answer, and perhaps also "Renee Descartes", but not "Bob Descartes". Since test writers probably don't want to write regular expressions, manual scoring would be a good option.

summit’s picture

summit commented

Hi Matt,

My opinion about your questions:
(1)
On short answer, let's list out the features that short answer questions do NOT have in common with either "long answer" or "fill in the blank".
==> If the answerfield length and attachment yes/no are settings per question, then there only needs to be one quiz-nodetype "open-questions" which can be transformed to a long question or short question quiz-question through settings per question.

(2)
This is a valid question for true/false, too: Is it different enough from multichoice that it should get its own question type?
==> If again there is a setting which can make a multiple choice in a simple true/false question, then there only needs to be a multiple choice question type. Hopefully with the possibility to score different answers differently.

Thinking it more through a quiz should then be a set of quiz-nodes with there own settings, likewise cck-nodetypes have fields.

(3)
With fill-in-the-blank, should we allow manual scoring?
I think that there should be the possibility to have the different scores blank, and when the scores are blank manual scoring is defaiut. Would that work?
Again scoring per different fill-in-the-blank field appreciated, because then simple and difficult fill in the blank answers can be scored as wanted.

Greetings,
Martijn

karthikpnair’s picture

karthikpnair commented

Wat about questions for personality tests where we may be having 5 answer choices and each answer will be having different points say 10 points for choice a , 7 points for b and so on , then summing up all the points and then displaying the result or the persons's personality

mbutcher’s picture

mbutcher commented

karthikpnair: I am working on scoring options right now. In quiz 2, all multichoice questions were worth 1 point. With quiz 3, we need to be able to score things liberally. Long answer questions already support this. When you define the question, you can set the maximum score there.

Yesterday I finished rewriting the scoring system to support this.

It sounds like what you are asking for is variable scores depending on which answer is chosen. The *scoring* of this can be accommodated, but it will take some work implement a user interface.

Would the best way to do this be to add a field for each answer allowing a score instead of a checkbox?

I'm starting to think that maybe multichoice is getting to be too broad of a question type. As Martijn suggests, perhaps it would be better to break multichoice out into true/false, multichoice/multianswer, multichoice/single answer, and maybe something like personality/multichoice.

Any thoughts on that?

iLLin’s picture

iLLin commented

Wow if you go that far, I would suggest a way to add questions when you create a quiz. Instead of having to create questions, create the quiz, then link them... etc. If your gonna start restructuring things, I suggest something like that.

Then when you add questions and can specify options for the question. Type, Score, Action, Feedback... Dunno just thinking out loud as my first impression on the handling of this was sorta unfriendly. I understand why it does it that way, but a go between on the UI level should be handled diff.

Scoring types would include right answer score. That way you can weight different questions heavier than others.
Personality answer score, so you can weight different answers with a different score... and so on.

My two cents anyway.

raycascella’s picture

raycascella
he/him
commented
Issue summary: View changes
Status: Active » Closed (fixed)

Pruning the issue queue, the requests in the comments are fixed in the 7.x-5.x branch.