I think the first important question to ask is: is the research I am looking at directly applicable and relevant to other people? If stuff is highly applicable

I think the first important question to ask is: is the research I am looking at directly applicable and relevant to other people? If stuff is highly applicable and relevant, then the chance to get away with made up rubbish starts to drop quite a bit, because people will want to try for themselves - and fail. It gets more problematic when the application takes a long time to show effects, i.e. long observation periods - there more can be faked. Most published research is largely inconsequential and not interesting to but a tiny few, so "fraud" can easily hide in there. Yes, it increases the sum total of knowledge, but that might be about it. I am not saying most research isn't done well, just that there isn't a way to assess correctness at scale and only high-profile results might get a fast check (and even there...)