Let me jump in here and suggest this. First you need to establish the actual tank - you can do a fishless cycle while discussing this project further.
Second, allow me to be blunt and show you an image of how clownfish are stocked in rearing systems:
The reality is that regular partial water changes (i.e. weekly 25-50%) and appropriate filtration are all that is required....it is not uncommon to have 5-10 clownfish juveniles per GALLON of water in a growout system as a breeder. There is nothing unethical about it.
The next thing to consider is that Christine is right on track when she talks about raising the sample size. What you will find if you simply divide the fish into 2 sets of 2 is that you'll wind up with a single dominant and submissive fish in each group. This will cause one fish to grow much faster than another.
My advice is to seek out a person who is breeding clownfish in the area. Hopefully this person can make available to you a decent group (perhaps a dozen) 3-4 month old Ocellaris clownfish (they should be at the 1" size at that point). You'll be able to conduct a trial without any issues, and if you have 6 fish on each side of the divider, you'll have a much more meaningful result. I don't know how much time you have to work with personally, but I do know that such an experiment is going to show more noticable results the longer you're able to run it.
I look at this and don't see how 3 sets makes sense unless you were testing food. Thus, 2 sets (food a and food b) makes more sense. I do wonder about the actual "control", as this is just a case of which food performs better. Now, more tanks with more fish to increase sample sizes would increase the validity of the data you collect. All said, 6 fish on each side is better than 3, and 3 is better than 2, and 2 is better than one. You have to remember that genetically, there are fish that will simply grow faster...having nothing to do with the food. James Fatheree's book on Giant Clams shows a really good demonstration of this, where all clams, all siblings, reared together under the same conditions, exhibit a distribution of growth rates. Since your fish are not genetically identical, you could quite easily end up with a "super fast grower" on the brine shrimp side, and a runt on the "premium food" side.
Another concern - feeding to saturation throws in another variable - fish A may simply be able to consume more than fish B. Or Food A is more palatable than Food B. What are you going to do to counteract that?