Hypomorphs (now commonly termed 'dominant negative') have a reduced (hypo) phenotype as compared the wildtype. Null (knockout) mutations, however, have the phenotype removed, so to say.
The reason that you would use null rather than hypomorph mutations to do double mutant analysis is because with the hypomorph, you would still see some effect of the phenotype you are trying to have eliminated.
Why do you want to eliminate a phenotype? Let's say you want to try to find the pathway for the genetic interactions between A and B. You have 2 animals you want to cross (I work with C. elegans worms, and use double mutants often). If we look at the wildtype homozygous of both animals, they will be the same. Animal 1: +;+/+;+. Animal 2: +;+/+;+. (where + denotes wildtype, and ';' means they are not linked, so the genes are independent of each other).
We will use 'a' to denote null mutation A, and 'b' to denote null mutation B.
Animal 1 will have mutation A. Animal 2 will have mutation B. So now Animal 1 will look like a;+/a;+, and Animal 2 will look like +;b/+;b. Notice that each locus is corresponding with itself - ie: in the region where Animal 1 is still +, Animal 2 is b, and vice versa (for null mutation 'a').
So then we cross Animal 1 with Animal 2 and get a;+/a;+ x +;b/+;b which will segregate in a 9:3:3:1 ratio (according to independent assortment of 2 independent genes - just draw a 4x4 punnet square to convince yourself).
We are interested in achieving the 1/16 a;b/a;b animal (also can be noted aabb, btw). And THAT is our double mutant.
Because we use a null allele, there is no chance of the wildtype phenotype "leaking" within our cross, giving us a false ratio or false assessment, as would be the case with a hypomorph (reduced, but not gone).
Now, when talking about genes in a pathway: I think this is what your professor was referring to. When we are mapping how genes interact, and in what order, we can either have the genes interacting in redundant or sequential pathways.
Redundant pathways: A -> C and B -> C. So either A or B can have an effect on C, meaning that having both is redundant, because one can do the job.
Sequential pathway: A -> B -> C OR B -> A -> C. So both are existing in the pathway in sequence, and both are needed.
Just as a side note, whenever you are using double mutants to find out if your pathway is redundant or sequential, you MUST compare it to a single mutant!!! Otherwise you WILL NOT be able to tell the difference between your results.
In a single mutant (A, for example), when using in a redundant pathway, you will still get phenotype C, since B is available. Just note that your C phenotype will be weaker.
When you then use a double mutant, knocking out both A and B, you will see NO expression of phenotype C. The same will apply if you knock out B instead of A.
Now, for a sequential pathway. If you use a single mutant (again, let's use A as the example) you will have no phenotype C, since A is needed to interact with B (whether upstream of downstream) in order to have an effect on C. And, if you use a double mutant (so knock out both A and B), then you will also have no expression of C, since both A and B are gone. (again, same applies if you use a B single mutant first)
So, for a redundant pathway: single mutant will have C phenotype, double mutant will not have C phenotype. There IS a change!
For a sequential pathway: single mutant will have no C phenotype, and double mutant will have no C phenotype. There is NO change!
If you would have used a hypomorph to do these experiments, then you would probably see some percentage of C expression in your double mutant experiments, making your results inconclusive.
I hope this helps!
[source: you can find this information in any genetics textbook (in my case, since i study worms, WormBook is a great source); I have given this information from my own experience. I have an honours degree in Genetics (Biochemistry and Medical Genetics) and am currently continuing my education in grad school (biochemistry and molecular biology).