What I think I do when I research and write might not be what I really do. Here's what I think I do:
I think that I first start with a topic, a particular thing I want to study, and then generate ideas about it. I probably need at least 10 or 15 ideas that are specific to this topic. I have to have a critical mass.
The second stage is to know that I have enough ideas to write an article or chapter with. Why so many? Well, some of these ideas are going to be significant and others will be sub-categories of other, more significant ideas. Some will be discarded or put into footnotes, or used for other projects.
A third stage is writing it all out, a process of sorting through and seeing what ideas are going to be the predominant ones. Imagine you have 10 ideas and you could write them all out, in a number list. That wouldn't be an essay, just a list of ideas. But you need to know what order to treat those ideas, and what their interrelations are.
The final stage is to make sure the bibliography and references are complete enough so that the essay will be accorded academic legitimacy.
I find I don't have to worry at the stage of idea-generation which of the ideas are important, or what their relationships are going to be. That's what writing is for. I realize too that I use scholarship to legitimate my intuitions, after the fact, rather than using it as the basis for my work.
For example, something I'm doing now on Pound and Lorca. I want to show what elements of their early work they share by virtue of being imagistic modernists of a certain sort. So I can make a list of elements of Pound's poetics and develop ideas of how these relate to Lorca's work. How did Pound and Lorca understand melopoeia, for example? Simply by looking at the material I generate ideas, and if I use enough precision in the analysis the conclusions will flow out of that process. There's always the stage of doubt, if the ideas seem weak or insufficient in quantity to support a whole article or chapter.
There was a rather well-known Beckett scholar at my undergraduate institution. I took a course on modern drama and she had us write papers by taking a theme, and then studying it in three separate plays, in a kind of mechanical way. I later realized that she used this uncreative method in her own work. I was kind of shocked, because this did not allow for the organic development of ideas. She was erudite and well-respected in her field, but I could already tell she had a second-rate mind. Now I reflect back and think that I was being arrogant in thinking this at the time, but it was a sincere arrogance.