(Looks around for potential collateral damage. Steps back. This can of worms is about to erupt into a hazardous mess…)
There are a lot of possible answers to this question. But first, it behoves us to understand some of the guidance we are dealing with in working our way through all of this.
- What kind of research are we talking about? Historical events? Technical questions? Psychological issues? All of the above? Something else?
- What is meant by "write she you know"?
- How much accuracy is required for fiction to be conveyed as realistic, to be accepted and believed?
I will, of course, start somewhere other than the beginning of the list. Just because. What does "write what you know" mean? Is this a suggestion that we should write about our own lives? That we should set stories within the confines of our professional or personal experiences? Or does it mean something else?
Any of those definitions would rule ouf the writing of anything fantastical or futuristic, yet those are some of the largest markets. As such, it must mean something else.
I posit, based on similar suggestions I have heard elsewhere, that it is about writing about feelings and issues we have experience of. It is about conveying echoes and shadows of our lives within a realm that may have nothing to do with the "real" one.
The obvious question about whether this should also apply to our technical experience - whether an activity, an area of study and proficiency, or otherwise - I will leave for the next point I tackle.
In my own writing, I deal with a lot of characters who do not relate to the world around them. They have messed-up relationships (if any at all), they feel a lack of agency within their environment, they fail to understand other's perspectives. Not all at the same time, but spread across a cast. Why? Because this is what I know about relationships. I could not possibly write a realistic loving and supportive relationship; I have no concept of what that looks - feels - like.
We come next to the question of accuracy. Again, I am going to cheat and paraphrase what I have come across elsewhere. The trick here is to go deep and wide. You do not need to get everything right, but you do need to convince your reader that the reality you are painting for them is accurate - trick them into believing this, in a sense. To do this, pick out one detail, one aspect of the subject that is important and relevant. It's probably a good idea to make this the firing mechanism of Chekhov's gun - something where the attention to detail ties into the final resolution. And be forthcoming about this detail. Don't skimp. Show your knowledge, show how much you have researched it. Or experienced it, if this is knowledge acquired prior to tackling your manuscript. Explain why things fit together the way they do.
And then… skirt over much of the other technical detail on other semi-related matters. Not by brushing them off, but carrying the authority established in that deep dive to all the things you will populate the wider scape with.
Of course, you can't afford to get anything wrong in the areas where you go wide - readers will pick out such glaring errors and then you will lose the established authority - so you need to do as much research for the subjects you only brush over as you did for the one you exposed in all its glory. You need, in effect, to become something of an expert in the subjects you investigate.
(For my own work, I spent a couple of days digging into formulations for ink, and for glass, into temperatures involved in glassmaking and the fuels that would allow this given technological constraints. None of which will make it into my manuscript, but peripheral details relating to it will, which allows it to be realistic. it tells me what areas I need to create fantastical solutions for, because the real-world options aren't up to scratch.)