[Aimmath-developers] Multipart questions ... a little trap

[Greg G. <gr...@ma...>]

Dear all,

I was just formulating a multipart question not too dissimilar to Neil's
primes example in doc/format.html and it occurred to me: `What happens
if, for a question, the answers are all examples of the same type, in such
a way that they can be entered in the answer boxes in any order, and we
are using the av> tag to ensure the answers are distinct ... *but* the
student can only think up one example and so enters that example for the
last part, marks it, then enters the same example in the 2nd last answer
box, marks it, etc. until all parts are answered?'

Answer: the student gets full marks if one does not have an explicit test 
in each part to check that the previous parts have been answered, though
once the student has finished this procedure all but the first part has
red feedback saying the answer to that part is not distinct from those
of the previous parts. (This is because AIM does not remark a question
that has been previously marked correct and so the previous perfect
score stands.)

In fact, the marking algorithm of the 2nd part of Neil's example as it 
stood (I've since added some code to prevent it) runs into error if the
first part is not answered. However, the question that I had been working
would have simply allowed the student to get full marks with one example.

Thus one needs to have explicit checks in each part to ensure that 
previous parts have been answered (if this is the rationale for the 
question). In Neil's example, the av> tag sets p1 to be the answer of
the first part. I have added a check that p1 is of type posint (the
type declared for the first part's answer) in the marking algorithm
for the second part.

name> primes
local> lowerbound,ans1,p1,p2
shift> 7645848653
t> Recall that Euclid proved that their are infinitely many primes
h> lowerbound := rnd(20..40);
note> sprintf("p2>p1>%a",lowerbound)
sq>
t> Find a prime $p_1$ such that $p_1 > @lowerbound@$
c> posint
av> p1
ap> $p_1 =$ 
s> ...
esq>
sq>
t> Find another prime $p_2$ such that $p_2 > p_1$
c> posint
av> p2
ap> $p_2 =$
s> [proc(ans)
     if not type(p1, posint) then
      # Make sure the student did in fact answer the first part!!
      `aim/t`("You haven't answered the first part yet!");
      RETURN(0);
     elif isprime(ans) then
      ...

Initially, I thought that perhaps some extra code should be included to
prevent a student from answering later parts if the former parts were
not already answered, but I think there are plenty of cases where it 
makes sense for students to answer the parts in any order ... so the
quiz coder just needs to be aware of the potential problem and write
questions accordingly.

  Regards,
  Greg