I'll concentrate on this from a cooking/coffee-making point of view, but if experimenting on people you need to be careful and follow the appropriate local guidelines, even if you're only giving them substances you can buy in a supermarket. The design of the experiment must be done under supervision. That side of things would be off-topic here for good reasons, but we can consider preparing drinks.
You'll obviously need to follow up the references but Wikipedia says that decaf coffee only gets down to ~3 mg per cup. This means that if you want zero caffeine, you'll have to start with something other than coffee unless you can prove the claim on Wikipedia (which I've flagged as original research) that this was residual caffeine in the machine.
If we assume for the moment that you can get an acceptably low baseline, one approach I've heard of is to brew this near-zero coffee for all the experiements, then add caffeine to the others. This will also reduce the taste difference (all your coffee will taste rubbish, rather than people thinking "this is disgusting, it must be the decaf") which is important for a blind test. It won't however eliminate differences in taste as caffeine is btter. Adding pure caffeine is one approach if you have access to it, but dosing small quantities is difficult. Caffeine tablets are a better source as they typically contain 50 mg (in the common low strength versions) which means you can add 0.5 – 4 tablets per serving.