samigo: feature request: IF-AT questions (new scoring method with immediate feedback for MC)

An instructor would like to create an IF-AT multiple choice assessment in Tests & Quizzes. A description of IF-AT assessments can be found here: http://www.epsteineducation.com/home/about/

In IF-AT quizzes, students get immediate feedback as to whether their answer is correct or incorrect, and then they can change their answer until they get the question right. If they answer incorrectly, the number of incorrect attempts decreases their score for the individual question, resulting in partial credit for questions they answered incorrectly at least once. The quiz would also have to prevent students from continuing to change their answer to a particular question upon entering a correct answer.

A potential workaround would be to develop SAK-34754, allowing students to only answer questions they got wrong during a second quiz attempt. The instructor could use Feedback on Submission to indicate which questions were wrong or provide hints for them to arrive at the correct answers the next time and the students could make a new attempt at the remaining questions. Taking the average score of all attempts would then approximate the desired partial credit scoring.

The instructor provided the following additional information regarding the IF-AT assessment format:

I would like to recreate the process of taking these tests by paper as
closely as possible. Students are given a handout with all the
questions on it along with a "scratchers" sheet. These sheets are
available from the website you reference in your email. They function
similarly to lottery tickets where you scratch off what you think is
the correct answer and there is a star randomly placed in the
rectangle if you guessed correctly, and it's blank if not.

If this were computerized it would be MUCH more straightforward than
having to purchase these specialized scratchers and then match up the
questions with the pre-defined order on the scratchers. That said, I
definitely would want students to be able to see the entire test and
not be forced to make an answer to move on to the next problem. I
would say the most natural way to do this is to have to manually
register an answer choice for each problem, receiving immediate
feedback as to whether it was correct, and keeping a tally of how many
attempts they made. You would want to make it so they couldn't
continue making guesses once they got the answer correct. It would
also be nice to keep a tally of how many questions they have yet to
complete (i.e., having found the correct answer). So, you would
decouple moving around from question-to-question from actually
registering a response. The other cool thing is that they leave
knowing the score they got on the exam (which could obviously change
if the teacher decides to drop a question or curve the final values).

The scoring would need to be customizable. For example, in my classes
I typically use questions with 4 answers. If they get the answer
correct on one try, they get full credit, if they get the answer
correct on two tries, they get half credit, if they get the answer
correct on three tries they get one quarter credit. Finally, if they
only revealed the answer on their 4th try they get zero points, but
the at least learn what the correct answer is. Thus, the scoring is
not a linear effect of number of tries.

As someone who studies memory, I know there is a significant
improvement in long-term retention if students learn during the test,
itself, so from an educational standpoint this is a major improvement
relative to almost all standard testing approaches.