So is your question theoretical, where you want to know if it is actually more difficult from a statistical calculation standpoint or does it comes more as a gestalt question.
As for the latter, and per Miner's link - I would say it can be very context based. For example I have recently ran for the first time genetic analyses and if you have a sound hypothesis, interactions can be abundant. E.g.) for what I was running - if you had two particular genes you under performed on most all cognitive measures.