How do we reach the conclusions we do?

Recently I was asked how do anthropologists actually reach the conclusions they do? In other words, when we are researching a topic, how do we decide what it all means? This seemed like a good topic to talk about here, especially since I had been meaning to include some more general anthropology stuff on the site. It also raises some more general issues that I’ll discuss at least briefly here.

In talking about research methods and data analysis, some anthropologists talk about things like dependent and independent variables. They talk about constructing hypotheses in much the same way a biologist or chemist might do. You develop an idea, you figure out what you need to find out to test your idea, and you go and do it. You work out what is dependent on what – or at least, the proposed relationship you want to test. This description, though, raises more questions than it answers. How does all this actually translate into real-world practice? (The short answer, of course, is that real life is much more messy.)

I think there are actually two issues here. The first is data analysis – how do you make sense of what you have learned? What does it all mean? The second is conveying your explanation. How do you tell your readers what you think, and why what you think is correct? I’ll only touch on the second one here, leaving it for another essay to discuss more fully. (The related question of how do you actually carry out anthropological research I’ll also put in another essay.)

To me, data analysis is much more of an art than a science, at least on the surface. I’ve never yet been able to teach (or explain) how you actually do it. It takes lots of work, time, and thinking, but beyond that, I'm not sure I can chart out how it is done. Perhaps because there are no real hard and fast rules for data analysis. This doesn’t mean that anything goes, and that some explanations aren’t better than others. (Let me make clear that I am talking about qualitative analysis – looking at data such as interviews, archival records, etc. I’m not going to go into quantitative analysis – number crunching – as it is of much less interest to me. There are certain rules, methods, tests than you can use for qualitative research, but even then, this won't necessarily tell you what it all means.)

I’m going to assume that you have developed what is sometimes called a "guiding question." That is, you have a research interest, and you know (more or less) what you want to investigate. This doesn’t always (usually? often? ever?) stay the same throughout a research project, but without a least an idea of what you are going to be doing, you won’t know what to focus on, what sort of data to look for, and so on. You need to start somewhere, and you'll never get grant money if you don't have at least a guiding question and some tentative thoughts. You need also to demonstrate that you know the relevant literature - who's done what. You should also be able to explain why your project is of interest and importance. Unfortunately, they won't give you money to do research if you just want to go hang out and see what's going on. (Trust me, I've tried...)

OK, with all that said: on to the analysis. First of all, reaching an understanding of an issue takes time. Time that can easily be measured in months or years. I'm not sure I fully understand even research I now largely regard as completed. I've simply had to move on, although I still keep an eye on stuff that interests me.

However, hopefully once you have some data, but while you are still in the field, you should be asking yourself "what does this all mean?" Or: "What is going on here?" It is best to at least start this in the field, so you can try to direct (or redirect) your research better. This is the start of data analysis, of trying to make sense of it all. As I've already said, there are no hard and fast rules. Rather, there's a lot of thinking, cogitating and contemplating. Pondering and puzzling help as well. (I often go for long walks when thinking about data.) Having said that, there are three general pieces of advice I keep in mind when thinking about data. The first I learned as an undergraduate, studying engineering: trust your gut reaction. The second is a question from graduate school: what is the plausible mechanism? Finally: does it pass the smell test?

To explain: As one of my engineering professors pointed out, what we think of as a gut reaction, or instinct, is often data analysis and evaluation going on at a non-explicit level. We are calling into play our previous experiences and knowledge to evaluate the question, the problem, whatever. A gut reaction won’t always be right, but it’s a good place to start. It's an indication that you've already been thinking about an idea, if not explicitly.

Plausible mechanisms. This was the favorite concept in a course I took in graduate school on "explanation in anthropology." The course was hell to go through, but it was a valuable experience. The professor’s main point could largely be boiled down to this: Can you provide a mechanism that explains what you are proposing and does so in a manner that makes sense? In other words, if you think people shape their identities based on their group’s history (for example), can you show how they do this? You may well be right, but can you walk us through how it happens? I don’t do this as often as I should, but even if you don’t, it’s a good approach to keep in mind.

And finally, what I think of as "the smell test." This links us back to the first two pieces of advice. When all is said and done, does the conclusion you reach make sense? Does it smell right, or is something rotten? Is the mechanism really plausible? Was the gut reaction right, or was it the chili speaking? It is important to realize this is NOT the same as saying all conclusions must be common-sensical or intuitive. Some of the best (or at least most interesting) explanations in anthropology are far from obvious. But when you read them, or hear about them, they seem right. The novelist and semiotician Umberto Eco argues that while you may not be able to prove a specific interpretation correct, you can rule some out as wrong. He was writing about texts, but I think this goes for explanations in the social sciences as well. (See his The limits of interpretation, published by Indiana University Press in 1990.)

This still probably hasn’t answered any questions. Perhaps because there are no answers. At least no simple ones. Reaching an understanding of a topic in anthropology is an iterative process. The more time you spend in the field, or reviewing data, the easier it is to come to an understanding of what you think is going on, and where you need to go next. To do so, just spend time thinking. Reread field notes. Take a break to clear your mind. Think about the data. Code the data and review it. (I use EndNote, a text-based database designed for reference bibliographies. I also use it to type up fieldnotes, and when I do so, I add keywords. I can then go back and read notes on any topic over the years.) This helps suggest patterns. Try looking at data from different angles, and data from different sources to see if they point you in the right direction. Read other stuff - both on the region you are working in and on the theoretical focus. Read other anthropology or history. Inspiration or the key that will help you unravel a point often comes from unexpected places and at unexpected times.

In the end, I find that the fine points of data analysis and conclusion-drawing are inextricably linked to the writing process. In many cases, to work out sticky or stubborn points, I have to write in order to think. So I'll leave further comments for the essay on writing and conveying your ideas.

Back to the top