I'm starting a new thread in this, because there seems to be a lot of confusion on what we are all talking about. I found several forum topics that have discussed this, but with no concensus on the problem or the answer.
Negative quiz grading is something that is used quite often in multiple choice, single-answer tests. Perhaps we need a new type of quiz that provides for this.
In the SAT, GRE and CLEP exams lets say that the student gets 1 point for each correct answer. If there are 5 possible answers, then a wrong answer will return a result of -.25 points. This is done to try to eliminate "guessing" or random checking of answers. Thus, in a quiz with 100 multiple choice questions, if the student correctly answers 80 questions they earn 80 points. By answering 20 questions wrong they lose 5 points ( 20 * .25). Thus, their final score is a 75. There is no "penalty" for not answering a question, the student just doesn't get any points for that question. That way, a student could choose to only answer 80 of the 100 questions. If they got 60 correct and 20 wrong, then they earned a 55 (+60 +(20*-.25))
In the existing quiz module, I can assign a negative percentage point value to an individual question, but it has no effect on the outcome. As an example, I create 10 questions, 5 answers each, with each question having a value of 1 point. Correct answer is assigned a +100% value, incorrect is assigned a -25% value. If I take the quiz and answer 6 correct and 4 incorrect, my reported score is 6, where it should be 5.
Thus, anyone who is developing study materials for standardized tests such as SAT or CLEP can not provide an accurate representation of the actual test, simply because they cannot assign a negative value. I have to also believe that there are many users who would want to integrate this type of testing in a normal environment.
Granted, using the adaptive mode I can do penalty points, but this has no relationship to the workings of a true test. They don't actually get the penalty until they retake the question. The student should be able to see a true representative score on a single attempt of a test.
Since the system already displays the negative percentage value in the individual answer areas, I would assume this was considered, but I have found nobody that has been able to make it work.
While I can't see a negative in providing this type of grading in the standard quiz module, perhaps I'm missing something (not unusual for me, ha ha). So, perhaps we can make a secondary module for "standardized testing".
Oh, and I'm using 1.6, but it seems the problem has always existed since there are comments going way back on this in other topic areas.
Thanks all, Dave