Sorry if this has already been asked, I did try a search.
Is it possible to set a question to have multiple true/false statements?
Hi
I am looking for a way of setting a question that has a stem followed by 5 true/false statements. I can format them as 5 seperate questions, but this is a bit ugly and takes up a lot of, screen space & time.
As this is a format that we commonly use and as a result have a very healthy bank of, we are keen to continue with if possible.
All advice greatly recieved.
Regards
Robert
I am looking for a way of setting a question that has a stem followed by 5 true/false statements. I can format them as 5 seperate questions, but this is a bit ugly and takes up a lot of, screen space & time.
As this is a format that we commonly use and as a result have a very healthy bank of, we are keen to continue with if possible.
All advice greatly recieved.
Regards
Robert
Robert,
In the standard quiz, choose "multiple choice question"
Fill in the stem, then add an image if required.
Next select "multiple answers allowed".
Done.
Only problems that I have found are:
1. that there must be at least one correct answer, and
2. you must have some form of penalty for wrong answers (otherwise the user can just select all answers as correct and guarantee to score full marks!)
For importing from another format I would recommend using GIFT format (search on the forums for GIFT).
Regards
Colin
In the standard quiz, choose "multiple choice question"
Fill in the stem, then add an image if required.
Next select "multiple answers allowed".
Done.
Only problems that I have found are:
1. that there must be at least one correct answer, and
2. you must have some form of penalty for wrong answers (otherwise the user can just select all answers as correct and guarantee to score full marks!)
For importing from another format I would recommend using GIFT format (search on the forums for GIFT).
Regards
Colin
Having thought about it, this would generalise out to a kind of question grid or table. This is fairly common in satisfaction type questionnaires where a set of questions all with the same (multiple choice) answers are down the left and the answers are along the top, the radio buttons forming the rest of the grid. True/False would just be a special case of this.
An interesting development project...
viz:
A1 A2 A3 A4
Q1 o o o o
Q2 o o o o
Q3 o o o o
Q4 o o o o
(the o's are meant to be the check boxes).
That would be ideal Howard as it would allow for a category 'don't know' along with 'true' and 'false'. For the future no doubt.
Hi, I also need such a generalized question.
The "True/False" should also be a "True/False/NoAnswer" as, if there is penalty for incorrect answer, the student has (I think) the right to leave it in blank when in doubt.
Such a question would have simultaneous multiple feedback. (build by n out of 3xn possibilities)
______________________________________
Colin,
I don't understand you:
What is true is that without penalization a big set of TF questions, answered at random, will yield statistically a 5 on a 10 basis. For a particular attempt (and mostly with few questions) the score ranges from -10 to 10.
The "True/False" should also be a "True/False/NoAnswer" as, if there is penalty for incorrect answer, the student has (I think) the right to leave it in blank when in doubt.
Such a question would have simultaneous multiple feedback. (build by n out of 3xn possibilities)
______________________________________
Colin,
I don't understand you:
..otherwise the user can just select all answers as correct and guarantee to score full marks!, ...
Correct and True are different concepts. The user can select True or can select false and in any case, it can be correct or incorrect depending on the question. What is true is that without penalization a big set of TF questions, answered at random, will yield statistically a 5 on a 10 basis. For a particular attempt (and mostly with few questions) the score ranges from -10 to 10.
Dear Jaime,
Sorry. I obviously haven't explained well enough. I'm referring to how the current multiple answer question works, not the proposed future question with separate true/false, which I would also welcome.
If you have a multiple answer question with, say, 3 correct out of 5 options, each correct option scores 33.33% of the total marks for the question. The ideal student will identify these 3 correct answers and score full marks. However, if the student ticks all 5 answers as correct and there is no negative marking for incorrect answers, the student will still score full marks for the question.
...unless I've misunderstood how the questions work.
Colin
Sorry. I obviously haven't explained well enough. I'm referring to how the current multiple answer question works, not the proposed future question with separate true/false, which I would also welcome.
If you have a multiple answer question with, say, 3 correct out of 5 options, each correct option scores 33.33% of the total marks for the question. The ideal student will identify these 3 correct answers and score full marks. However, if the student ticks all 5 answers as correct and there is no negative marking for incorrect answers, the student will still score full marks for the question.
...unless I've misunderstood how the questions work.
Colin
OK Colin,
In the current multiple choice question, it is obvious that using "multiple answers allowed" the creator must assign penalization in wrong answers and the student should be warned.
Other ways, the question may be cheated easily.
If the creator of the quiz doesn't want to apply penalization, at least he should select how many answers are right. In any case, it will be appreciated that the question One or multiple answers?:"multiple answers allowed"/"one answer only" be replaced by How many multiple answers?:"1"/"2"/"3"/ ... and be controlled at submission time. And the "You must fill out at least two choices" will be then "You must fill out at least n+1 choices"
I always apply penalization to avoid scoring if the student answers at random.
In the current multiple choice question, it is obvious that using "multiple answers allowed" the creator must assign penalization in wrong answers and the student should be warned.
Other ways, the question may be cheated easily.
If the creator of the quiz doesn't want to apply penalization, at least he should select how many answers are right. In any case, it will be appreciated that the question One or multiple answers?:"multiple answers allowed"/"one answer only" be replaced by How many multiple answers?:"1"/"2"/"3"/ ... and be controlled at submission time. And the "You must fill out at least two choices" will be then "You must fill out at least n+1 choices"
I always apply penalization to avoid scoring if the student answers at random.
If I have five options available and three need to be ticked (ie: they apply to the statement, or question), then each correct answer gets 33.3% and each incorrect answer gets -50%. this means that if all the options are ticked, the student gets 0. It can mean that the student ends up with -100%, but then they would have to completely misunderstand the question.
Philip
Has anyone figured this out yet? Like someone early in this thread, my paper quizzes so far have a lead-in prompt and five possible completions, all to be marked individually. In theory, there might be a prompt with NO correct completions listed, or a prompt where ALL FIVE are correct (so it's not the case that if n answers are correct we should have to supply n+1 possibilities). I don't allow an "I don't know". So, there are 32 possible "Gestalt" answers, only one of which gets 100%, one of which gets 0%, the rest getting increments of 20% according to their being on "the right side" for each completion option.
I struggled for a while trying to see how moodle's "allow multiple answers" could accomplish what I want, and by now I'm convinced it's not currently possible. I'm dissatisfied with the suggestion that for 5-prong questions with 3 right answers I should assign 33% credit for the correct ones and -50% for bad ones. Someone who fills in all five should get NOT zero, but 60%; after all, to three of the five binary choices the correct answer was marked.
If I assign 20% for the correct ones and 0% for the incorrect ones, the grades come out right for exactly those students who mark no "false positives". If I assign 33.3% to the correct options and -20% to the incorrect ones, then the scores are correct for all the students who "overaffirm" without missing any correct options, but otherwise the scores are incorrect.
If anyone has a pointer for how to get the straightforward behavior I'm hoping for, please do chime in!
I struggled for a while trying to see how moodle's "allow multiple answers" could accomplish what I want, and by now I'm convinced it's not currently possible. I'm dissatisfied with the suggestion that for 5-prong questions with 3 right answers I should assign 33% credit for the correct ones and -50% for bad ones. Someone who fills in all five should get NOT zero, but 60%; after all, to three of the five binary choices the correct answer was marked.
If I assign 20% for the correct ones and 0% for the incorrect ones, the grades come out right for exactly those students who mark no "false positives". If I assign 33.3% to the correct options and -20% to the incorrect ones, then the scores are correct for all the students who "overaffirm" without missing any correct options, but otherwise the scores are incorrect.
If anyone has a pointer for how to get the straightforward behavior I'm hoping for, please do chime in!
I have used multi multiple choice for about a quarter of my quiz questions over the last term, and have been rather frustrated with the current scoring method. For the student, each choice represents one decision that must be made. So it seems unfair that with four choices, sometimes selecting one wrong choice will result in a score of 0 (3 choices correct), and in other cases a score of 67% (1 choice correct).
The problem with current scoring in multi multiple-choice is that credit is given only for selecting the correct responses. It needs to be revised so that credit is also given for not selecting the incorrect responses. Rather than assigning e.g., 33.3% for each of 3 correct choices and -100% for one wrong choice, it would make much more sense to assign each 25% and specify whether each choice is correct (true) or incorrect (false). Then each response receives its correct weight.
True, with this scenairo you have to guess very badly to get 0, but you have to guess very well to get 100%. Leaving all blank will typically give some positive score as will marking all the answers, but in regular multiple choice always marking "a" (or whatever) will also normally yield a positive score. The best prevention for credit with no response is the T-F-Don't Know method (defaults to Don't Know), since that requires a resonse to earn points.
I would really like to see multi multiple-choice scoring improved. The old scoring method could be left as a default option, e.g., by setting a "score as TF choices" radio button above the list of answers. If selected, each answer is given equal weight automatically according to whether it is designated Correct or Incorrect (or T/F). Displaying a "Don't Know" could also be an option. Implementing a Multiple True/False would also work, since conceptually there isn't a great difference between multi multiple choice and a single stem followed by several T-F responses. So I would welcome a solution either way -- revised multi multiple-choice scoring or multiple T/F, whichever is easier to implement (and add to GIFT).
Jim.
I would really like to see multi multiple-choice scoring improved.
Have a glance at confidence based marking, over at http://www.ucl.ac.uk/lapt/index.htm
I attended a presentation about this, a really innovative way of getting over the guessing problem in m/c tests and bringing in a far more reflective approach to this type of question, by asking the learner to not only choose the correct answer but also indicate how confident they are that they're right.
I don't suppose anyone's figured out how to do this in Moodle?
Have a glance at confidence based marking, over at http://www.ucl.ac.uk/lapt/index.htm
I attended a presentation about this, a really innovative way of getting over the guessing problem in m/c tests and bringing in a far more reflective approach to this type of question, by asking the learner to not only choose the correct answer but also indicate how confident they are that they're right.
I don't suppose anyone's figured out how to do this in Moodle?
Very intriguing. I would love to try it, but I don't see any way to do it in Moodle at present. You might want to request it as a feature addition in the bug tracker (don't know what the guidelines are for that). Anyone else interested in this capability?
Aslo, is there any data on how this affects students with severe test anxiety? I would expect it might exacerbate the problem, but that's just an initial hunch.
Jim.
Also, is there any data on how this affects students with severe test anxiety
It's a while a go but I recall folks asking about gender differences in the meeting - the evidence I think indicated no significant difference - it provides a feedback mechanism so that those who are relatively hesitant see an incentive to becoming more confident about their opinions, whilst giving lower grades to those who had been too confident about what was, in fact, incorrect; so sterotypically, it encourages the girls to speak up and makes the boys think twice before they do. The version I heard about (and linked to above) was developed for medical education, where over confidence can be, erm, problematic.
It's a while a go but I recall folks asking about gender differences in the meeting - the evidence I think indicated no significant difference - it provides a feedback mechanism so that those who are relatively hesitant see an incentive to becoming more confident about their opinions, whilst giving lower grades to those who had been too confident about what was, in fact, incorrect; so sterotypically, it encourages the girls to speak up and makes the boys think twice before they do. The version I heard about (and linked to above) was developed for medical education, where over confidence can be, erm, problematic.
Regarding Miles suggestion regarding implementation of confidence marking in MCQ or Mtrue-FalseQ I was wondering if any progress has been made since august?
If not what would you expect if confidence marking is implemented in MCQ or MTFQ?
Jean-Loup Castaigne
If not what would you expect if confidence marking is implemented in MCQ or MTFQ?
Jean-Loup Castaigne
Hi James,
regarding test anxiety Bruce Chopin has demonstrated in his 4th model that, to make it short, student arrange their choice in degree of 'likelyness" or in percentage to be correct or false (100% sure that proposition is false, this one 50% being right and this one 80% being true so I choose that one but I'm not 100% sure). So one can assume that choosing an answer and giving his confidence in his choice is the same mental process.
But like always to reduce test anxiety and evaluation bias students should be trained trough the evaluation procees.
On the other hand teaching should be focused on learning and not on assessment as research should focus on assessment of learning instead of assessment for learning... (EC report) but that's another debate.
Jean-Loup Castaigne
regarding test anxiety Bruce Chopin has demonstrated in his 4th model that, to make it short, student arrange their choice in degree of 'likelyness" or in percentage to be correct or false (100% sure that proposition is false, this one 50% being right and this one 80% being true so I choose that one but I'm not 100% sure). So one can assume that choosing an answer and giving his confidence in his choice is the same mental process.
But like always to reduce test anxiety and evaluation bias students should be trained trough the evaluation procees.
On the other hand teaching should be focused on learning and not on assessment as research should focus on assessment of learning instead of assessment for learning... (EC report) but that's another debate.
Jean-Loup Castaigne