Population protocols are a model of distributed computation with very strict limitations on the memory of the single agents. A population protocol is called robust if it is resilient against interference to a certain degree. We expand on the term of robustness, adapting it to protocols with non-binary outputs that compute functions instead of predicates, tweak the existing protocols from [9] to fit our new definition of robustness, and arrive at a general construction for robust protocols for the parallel composition of functions. By that, we partly address whether there exist robust population protocols for Boolean combinations of Presburger predicates since a population protocol computing such predicates is a direct consequence of this. A simulator for population protocols with snipers was implemented, and its results were analyzed to empirically verify the theoretical claims.