Categories
Research Methods

Deductive-Inductive Combination in Thematic Analysis (Tutorial)

Do you want to apply a combination of deductive and inductive thematic analysis in your qualitative research?

Is that even possible, and what should you keep in mind?

In the next few minutes, I’ll show you how to combine deductive and inductive coding, what authors like Braun and Clark, who are the most cited authors on thematic analysis think about it, and how to use this knowledge to perfect your qualitative study.

Inductive and Deductive Category Formation for Qualitative Content Analysis

If you’re familiar with my videos on thematic analysis, you know that this method distinguishes between two types of code and theme formation for analysing qualitative data.

#1 Building Themes Inductively

Here, you derive abstract codes from your data, that is, for example your interview transcripts. As the saying goes, the themes “emerge from the material.” This approach is also often referred to as “bottom-up” coding.

#2 Applying Themes Deductively

In this type of thematic analysis, you create a list of pre-defined themes from a theory or other literature.

Then, you approach the data with these themes in mind and systematically allocate your data to each theme. You can also count how often each theme occurs in your data set.

But what about a deductive-inductive combination? Does this have any advantages.

The answer is: Yes!

Deductive-Inductive Combination for Thematic Analysis

Since thematic analysis was originally invented to be an inductive method, but it’s not always easy to form new codes and themes purely from the ground up, a deductive-inductive combination is often a good way to get the best of both worlds.

Background

Why is this approach useful?

First, pure induction is powerful but difficult. For most research questions, pure induction isn’t the ultimate solution. The problem of induction, which frustrated David Hume and Karl Popper, still persists. Inferring a general rule from a single case is highly problematic. But, even if you are OK with that, it is quite difficult for beginners to confidently develop themes “out of nothing.” At the same time, supervisors often encourage you to read about theories. Incorporating existing theory in inductive research is quite a challenge, and you need a lot of experience to do it correctly.

Second, doing only deductive coding also has significant disadvantages. The theoretical framework you choose to work with is practically predetermined. Deduction doesn’t allow for breaking out of this framework, which means surprising insights that might contradict prior knowledge are not considered. These surprising insights are what make many research projects interesting and can potentially provide greater value for existing theory and literature.

Thus, a combination of both logics can be considered for thematic analysis.

deductive inductive combination

A deductive-inductive thematic analysis

In deductive thematic analysis, you choose your themes based on prior theoretical knowledge to guide the research process. The data is then coded and allocated into this structure (e.g., the main themes). You start with the same thing in your deductive-inductive thematic analysis.

But then, you use surprising findings or data that do not fit these themes to form new, inductive subthemes. If you find a lot, you can even add a main theme to your list of themes.

This can create theoretical added value, contributing something new to the existing knowledge – all while you enjoy the comfort of the theoretical framework you used for the deductive part of the coding.

Example

To better understand how a deductive-inductive combination works in thematic analysis, let’s look at a brief example.

Theory

Assume your study involves Identity Theory. It broadly states that individuals adopt certain characteristics, values, and norms to view themselves as unique compared to others. According to Burke and Stets (1999), three theoretical dimensions are involved in the identity formation process: Investment, Self-esteem, and Rewards.

These three theoretical dimensions could serve as a structure for your thematic analysis. According to deductive logic, three main themes would emerge:

#1 Main Category: Investment in… X (X=your topic)

#2 Main Category: Self-esteem based on… X (X=your topic)

#3 Main Category: Rewards from… X (X=your topic)

You could now structure your material, such as interview passages, according to these three main themes using regular coding techniques.

Now, you could inductively add subthemes by grouping your codes.

Or, based on data that doesn’t fit any of the three main themes, add another main theme based on that “left-over” data.

Data

Imagine you have the following quote in your interview transcript:

“Earlier, I thought I would have to give up my job completely to be a mother. Now I’m more confident that I can do both if I want, or if I want to work part-time or flexibly and arrange childcare without it affecting my career.”

This quote initially falls under Self-esteem, as it deals with self-perception and confidence in overcoming obstacles.

For this quote, you could create the subtheme “self-evaluation of own competencies.” This new subtheme could be placed under the main theme “Self-esteem in … X (X=your topic).”

You might find more quotes that fit into this subtheme, or perhaps quotes that form a completely new main theme – though this is unlikely with a well-established and often-tested theory.

You work inductively through your data until your system of main and subthemes forms a complete picture.

What Braun and Clark Say About the Deductive-Inductive Combination?

The authors of the most cited paper on the method have made clear that the method was originally intended to follow an inductive logic. With their focus on being “reflexive” as a researcher underscores this.

However, it is exactly this what makes qualitative research so flexible. It is not about following a guide from start to finish and never deviate from it. It is about tailoring the analytical approach to your needs and what makes the most sense in your situation.

Therefore, Braun and Clark would agree that combining both logics can be of value if it makes sense in your context.

Outside the Braun and Clark bubble, many researchers see deductive-inductive code formation as a very legitimate way to conduct qualitative research, especially, if you are a beginner.

As I always emphasize, you have the freedom to tailor your approach to your needs, as long as you act systematically and can logically justify your decisions in your methods chapter.

Is Deductive-Inductive the Same as Abductive?

You might have heard of the third type of reasoning: Abduction. This comes into play when you encounter a particularly surprising result and try to explain it with what makes most sense based on the information you have at this point. Then you infer a rule or category that best explains this result. However, this is quite risky as you can’t verify if the rule is correct.

A typical example of abduction is the detective Sherlock Holmes. He looks for clues at the crime scene and abductively infers how the events might have unfolded. These inferences are often very bold but bring him closer to the truth.

However, abduction in qualitative research isn’t really something you can plan systematically. You can’t know if you’ll find surprising results where an abductive inference could help. Here, we can only adopt abduction as a general attitude towards surprising results.

The deductive-inductive method, on the other hand, can be systematically planned and implemented.

Leave a Reply

Your email address will not be published. Required fields are marked *