VERY VERY VERY long story made short:

if the analysis is something like computing a function f on some subset of whole data (e.g. each peer has a portion of spectrum to analyze, like it happened with old SETI@home), I suggest you to check out ZK proofs, especially SNARKs, especially (again) function commitments.

(1) result is implicitly certified cause function commitments would permit to a peer to calculate the function and then share the result, together with a proof their computation is correct

(2) if your original problem is divided into functions of surely known computational difficulty, once calculated (see (1)) you are actually sure the peer producing the output has used an amount of computational power of the order of the function

This is a very good course about this stuff ongoing now, but it's quite demanding: https://zk-learning.org/