The purpose of an analysis is to bring about the artful disappointment of the reader’s expectations of your object. Less dramatically, we can say that there should be an interesting tension between the theory and the analysis presented in a paper. (Some people would say there are no disappointments in life, “only challenges”. I’m happy to put it that way too.) This is why it is so important to gain the reader’s trust in your methods section. If your readers are more likely to reject your results than let them challenge their expectations then you will not be able to write a convincing analysis. You will only be able to confirm their preconceived notions about reality.

When you are writing your analysis you should take your data for granted, as “given”. (That’s what data means.) Of course, you came by them honestly, having wrung them from experience through a lot of hard work (described, like I say, in your methods section). But once you get to the analysis, you’re not going to be making any apologies for assuming that the transcripts of your interviews are correct or survey respondents answered as they did. You are simply going to confidently base your analysis on them.

In the talk, I made a great deal out of the difference between “observations” and “interpretations”. Both kinds of statement make “factual” claims, they are statements of fact, and they should be as true as you know how to make them. (Just because it’s an interpretation, doesn’t mean you can just say whatever you want.) But observations are statements of facts that are given in your data — presented, perhaps, as tables of numbers, or quotations and paraphrases from interview transcripts, or descriptions based on your field notes — while interpretations are statements of your take on the facts — facts that are taken, if your will, or facts that have been accomplished by your analysis. These are facts that must be inferred from arrangements of the data you adduce.

In the Q&A someone asked whether this required a “leap”. My answer was, yes, there’s always a little leap of logic between an observation and an interpretation, from a statement of what happened to a statement of what it means, from what people say and do to what they believe or desire, hope or fear. But this leap should be carefully prepared by the work you did in the theory section to establish concepts as “categories of observation”, rules for how to pass from what you see to what you think. Logically speaking, the reader will be taking a leap (with you) from observation to interpretation. But it should feel like a very reasonable one. It should feel more like a step, actually.

Paragraph for paragraph, try to make your key sentences interpretations and the supporting sentences observations. [You can go back to the “how to write a research project” lecture to hear more about paragraphing.]

I got a chance to mention Hemingway’s plotting advice: make your a story “a sequence of motion and fact”. Every half page or so, every minute of your reader’s attention (which is to say paragraph for paragraph) you should have clear picture in your mind of what you want the reader to imagine, and that picture is either still or moving — tell the reader what is the case, or what is going on, describe either a state of affairs or a change in circumstances. That way you have a good way to keep your analysis engaging. It doesn’t matter what kind of data you are working with or what sort of analysis you are doing; you are going to be establishing a series of facts. But some of the facts come before and some come after a change. That change itself can be stated like fact — in motion. Keep that in mind when writing about it.

Like I said at the beginning, I had forgotten my notes, but looking at them afterwards I was happy to see I covered everything I had planned to say. Except that I had left out my epigraph — a line from Lisa Robertson’s poem Cinema of the Present. “Thus your data shimmers,” she declares. I hope the spirit of that idea was present in my talk. Tony Tost used to say that when he’s writing a poem he’s “basically just trying to be brilliant”. In your analysis, then, try to make your data shine!

[You can read another take on writing the analysis here.]

Here is what I say in the “How to Structure” lecture:

Here are some further posts on the subject:

“Observation, Interpretation, and Analysis”

“Saying, Doing, Meaning”

“Observation”

Here is the video from 2022: