In a recent question, I left a comment saying that the actual topic of the question probably doesn't have much direct research, but that answers which provide research in similar areas could be acceptable. Right after I wrote that I thought: is that actually true? Can we provide somewhat-related evidence, and infer that it supports an answer one way or another?
Just to use this recent question as an example: the core of the question is the claim that if you are able to fast for long periods of time, you will also be able to overcome addiction. I'd be surprised if there are any robust studies on this exact topic: getting test subjects willing to abstain from food and continue to take drugs seems impractical and pretty unethical (not to mention the need for an addiction-focused control group). However, it seems reasonable that there are studies covering related topics: for example, perhaps there is a study of whether or not self-control skills in one area of life can translate to self-control in other areas, or studies regarding whether addiction rates increase or decrease in populations facing famine. If such studies exist, we could infer some relationship between the ability to handle hunger and the ability to handle addiction, which would help answer the original claim.
I've seen a couple meta questions (here and here) suggesting that "we don't know" answers can be acceptable, however I'm more wondering if "we don't know exactly, but…" answers are also acceptable. I'm sure they would be on a pretty subjective scale and am worried that answers like this will just lead to long debates in the comments about whether or not the research is actually relevant, so I think it'd be good to know whether to encourage or discourage this sort of answer.