I tried to construct a formula, that would return some kind of weighted random number, so that you can use just one formula to get the results. The process went something like this:

If you pick just one random number (RAND()), there is a 50% change that the number is greater (or equal) than 0.5.
If you pick two random numbers, there is a 50% change that the maximum of those numbers >= 0.66666.
If you pick three random numbers, there is a 50% change that the maximum of those numbers >= 0.75.

Some kind of pattern is forming…

If you pick n random numbers, there is a 50% change that the maximum of those numbers is >= 1-1/(n+1)

Then I changed the perspective. Let’s say that a student has 8 completed assignments. I pick a random number, RAND(). If RAND() = 0.5, student should get a weighted value of 1-1/(8+1)=0.88888, but if the RAND() = 0, the weighted value should be 0 and if RAND()=1 => weighted value = 1.

So I should scale the result to return a kind of logarithmic value. More there are completed assignments, greater the change to get a better weighted value. So the final function to use is this:

=RAND()^LOG(1-1/(n+1),0.5)

- Asser

=MAX((B5>0)*RAND(),(B5>1)*RAND(),(B5>2)*RAND(),
(B5>3)*RAND(),(B5>4)*RAND(),(B5>5)*RAND())