Coding of qualitative data?

 

What is coding of qualitative data?

Well, coding is a method used to analyze data by identifying themes or codes that appear in our qualitative data and then assigning intersections of data to those codes. 

Miles and Huberman state that codes are tags or labels for assigning units of meaning; codes are usually attached to chunks of varying size. Those might be words, phrases, sentences, or even whole paragraphs. 

The process is to break up very long and detailed qualitative data, like interviews, focus groups, and documents, into common themes. We can then read across different sources to compare what people are saying on one particular issue or topic. 

For example, we might have a code called “healthy eating,” and we can compare what different groups of people have said about healthy eating. So, for example, we might see whether older people were more into healthy eating than younger people or what different ideas people had about what healthy eating is. Coding is like putting things into categories. 

Once we've created our categories and sorted things into them, it's easier to find things later and see how items in one category vary. It's partly a method for analysis, but also a way of managing your data. Once you've been through and coded your data, it can be a lot easier to do the writing up later because it makes it easy to quickly find a quote to support something that you want to say.

Let’s do a little bit of coding here and see how that works. We’ve got an example data source of someone being interviewed about what they eat for breakfast:

“I’m a single mom with an eight-year-old toddler and breakfast is mayhem. Baby has porridge; I microwave up some Ready Brek with whole milk for him.”

So, I’ve created a couple of codes here of things I think might come up. You see we’ve already got one here for “porridge.” So, I’m going to select this text about porridge and drag and drop it onto the porridge theme. So, we’ve coded that section of text now to that theme. We could have chosen the whole paragraph or a larger piece of text or just this section here about porridge, but we chose the whole sentence. And we can also say that this whole sentence is also about milk, so we can drag that onto the milk code here as well. Now we’ve coded this text to be about milk and about porridge. That’s the basics of coding: going through, creating codes, and coding sections of text to those codes.

Now, the kind of coding we’re doing here is very basic and descriptive. For example, here: “I’m a single mom with an eight-year-old toddler; breakfast is mayhem.”

 Let’s put this onto the children’s code here because actually, I think that’s about children. But you’ve seen we’ve got some other themes here, so we can say that this is also about family structure, about wider issues, about social breakdown, and also about time. These are things that might come up, and we might want to do some kind of in vivo coding—questionably about “mayhem”—and we may want to even create a theme here called “mayhem.” 

So, various different ways that we can code up the data. You see, even this very simple sentence, these couple of sentences here, we’ve managed to code in a whole bunch of different ways. So, it can be a very long process of going through and doing coding, but this is the basics of how it looks.

Braun and Clarke describe codes as being a pithy label identifying what’s of interest in the data, while themes are an idea or concept making a common recurring pattern across a dataset, clustered around a central organizing concept. So, in our example here, “low-fat yogurt” might be a code, and “healthy eating” would be something more like a theme.

There are many different types of coding. 

The two big kinds are grounded theory or emergent coding, where you start with a blank sheet and have no preconceptions of what you’re going to find in the data. You’re completely open to having the data speak to you, and as you go through, you define the codes and themes that the data suggests to you.

 The other method is called framework analysis or structured coding, where you actually have a framework—your list of codes and themes beforehand—and you try to match the data to the codes you’ve already identified. In practice, most people do a flexible combination between both: they have some idea of the things they want to find in the data, but they’re also open to new things, so if something surprising and unexpected comes out, they can create new codes and themes to capture that.

There are also lots of different types of coding:

• Descriptive coding: literally coding what’s being discussed in a very literal basis.

• Thematic coding: creating themes and codes.

• Line-by-line coding: assign each line its own code, describing what’s going on in that line or sentence.

• IPA (Interpretative Phenomenological Analysis): looking to see how participants experience and make meaning of things in their world, often combined with line-by-line coding.

• In vivo coding: using the actual words and terms that participants use to develop codes.

• Discourse analysis: looking at how people express themselves, what types of words they use, and how different people express themselves.

There’s a whole textbook by Johnny Saldaña which is excellent to read on coding qualitative data, and it describes 34 different ways you can code data. If you’re stuck, you can try a completely different approach.

It’s important to remember that coding is nearly always an iterative, cyclical process. You go back and do it again and again, over and over, reading it through multiple times, doing different levels and types of coding, and also going through and coding your coding—putting your codes and themes into different groups and structures to pull together the different themes coming out from the data.

Coding is an important part of analysis, but it also requires further interpretation when it’s complete. You still need to say, “What does my coded data say?” You need to go through and read it, and all these different methods are really just different ways of getting to know your data better. They’re not going to do the analysis and interpretation for you, but they’re an important step in the process.

It’s also important to bear in mind that you don’t have to do any coding. Some people don’t like doing coding; some practitioners say that it abstracts people from the data and is a very reductive process. You’re taking very long, detailed, complex, nuanced data and extracting it to a very low level of analysis. Some people just advocate reading the data over and over until it makes sense in your head.

There are lots of software tools like Quirkos that will help you code qualitative data, but you don’t have to use any special software. Many people use spreadsheets like Excel to do line-by-line coding and put different themes for those, or in word processors like Word, just use the comments feature. Many people also just use pen and paper: print out the transcripts, use different colored highlighters to identify different themes, or even cut out sentences or paragraphs and put them into different envelopes or folders and group them together by theme. These are all ways to get to know your data better and start to sort and organize it. But dedicated software does make it a lot easier when you come to write up, to find those quotes, especially compared to doing it on paper and not knowing where they came from. If you want to see everything coded on “healthy eating” in all the qualitative software packages, that’s just a click away. Quirkos is just one of the ways that you can use to analyze your data—it’s very visual, intuitive, and easy to use

.

---

This transcript covers the main instructional content and examples from the video, summarizing the step-by-step process and key concepts in coding qualitative data. For a full verbatim transcript, you can use YouTube’s “Show transcript” feature as described in the search results

.