Optimizing qualitative research: Coding and tagging strategy

Image

The following is a guest post by Cherryl Pereira, Head of Content at Chisel Labs

Qualitative research is a method of gaining insights into how people think and act by studying what they say, do, use, or build. 

The data collected typically consists of videos, audio recordings, notes from observations, and conversations. 

Qualitative research can be used to identify emerging trends before they become mainstream or to understand the reasons behind why people behave in certain ways.

The qualitative researcher must code their findings so that it becomes useful for other researchers looking at similar topics in the future. 

This post will discuss coding strategies for qualitative analysis including highlighting common errors that are made when coding and tagging data as well as tips on how to ensure that your work is reproducible.

Quick Introduction of Qualitative Research
What Is Coding?
What Is a Tagging Method?
Methodological Steps to Implement the Coding and Tagging Process
Our Top Tips on Using Coding and Tagging Strategy
Final Thoughts

Quick introduction of qualitative research

Qualitative research is used to capture the meaning behind words, expressions, and actions. It is not focused on statistics or numerical data (quantitative research).

The purpose of qualitative research can be varied: it could be for evaluative purposes like focus groups; exploratory work, interviews; development projects, usability studies; marketing feedbacks, and more.

It’s important to understand that in these types of methodologies you are looking at behaviors rather than what people say they do. 

You may ask participants about their preferences but this does not mean they will act upon them. 

Hence, the researcher has to interpret accordingly after having observed behavior patterns within a group setting. 

The main difference between quantitative and qualitative research is that with quantitative research there is an independent variable. This variable is manipulated by the researcher to affect a dependent variable.

Qualitative research aims at understanding how individuals think, feel, and act in different situations through exploration. Instead of testing specific hypotheses as with quantitative research. 

Another key distinction between quantitative and qualitative research is that with qualitative research there are no control groups or randomization of conditions. Therefore, it’s harder for researchers to draw cause-and-effect conclusions from their findings.

Firstly, we will look into what exactly is coding and tagging in qualitative research.

What is coding?

Coding is the process of assigning qualitative data with a word or phrase that best describes it. 

For example, when you are listening to an interview about why consumers shop at Whole Foods instead of Trader Joe’s. 

Coding would involve labeling specific quotes as “shopping” and then analyzing each quote for common themes to find what motivates people to choose one store over another.

The simplest way to approach this task is by using Excel spreadsheets – however, not everyone has access to these tools. 

There are also free online software programs that researchers can use for their analysis such as NVivo, Atlas Ti, and Ethnography. 

These computer-assisted programs help researchers organize large amounts of qualitative research quickly into various codes (i.e. “shopping,” “personal preference,” etc.), which can then be used to do further coding (i.e., sub-categories of shopping) and searching within the data for specific quotes that fall into a certain code/category.

Why coding?

Coding is optimal to use when you want to look at more than one variable. 

For example, if you’re writing about two different types of medication and how each affects a user differently. By using coding, the respondent can tell you without hesitation which drug had what effect on their symptoms and quality of life.

Coding is also great for understanding multiple levels or stages of change that an individual has undergone in any given topic area (e.g., smoking cessation). 

It enables researchers to identify specific events as well as certain aspects of those experiences. Such as thoughts, feelings, and behaviors associated with them.

What is a tagging method?

Tagging aims at increasing the information density of a text by increasing its semantic specificity.

It is different from coding because it does not need pre-selected categories or concepts before data collection. 

Rather it relies on the researcher’s judgment to determine how detailed and specific each category needs to be.

Why tagging?

Tagging helps you organize your qualitative research quickly and efficiently without needing too many resources for that purpose. 

You can do this even if you are working alone as there is no limit in time, money or personnel required. 

It also allows researchers to easily identify relationships between objects/concepts across multiple texts simultaneously at low effort levels throughout the entire process (e..g., during transcription). 

This feature makes tagging suitable for large quantities of data where manual coding is not an option.

We learned about coding and tagging. Now, let us look into the steps of implementing this technique.

Methodological steps to implement the coding and tagging process

Step one: 

Get organized – create folders for each document type (e.g., interview transcripts, photos, etc.). It will make things easier to find later; Then you will have to decide if you have to go for the deductive or inductive approach.

Deductive: you might start by creating a codebook in which to define each category and its properties (e.g., it’s a label, where the data comes from)

Inductive: you will create categories as they emerge during your research process.

Step two:

Coding – You will have to do a preliminary coding of the data to get an overview and check that everything is consistent with your research questions.

Step three:

Tag the data – Here you will have to assign a code or tag for each document, line of an interview transcript, or paragraph of a memo. For example, if you are coding text documents, assigning numerical codes might be helpful;

Step four:

Cleaning up the raw material – depending on how much time you can spend, it might be wise to start from scratch or just use “copy & paste”. In any case, if there are documents mixed (within transcripts), you should separate them before starting anything else. The data which is not important has to be removed.

Step five:

Mark out where you think your codes should be placed. This helps in avoidance of codes being overlapped with other codes or changing line spacing too much (e.g., indenting, using different fonts, etc.). This is not as easy as it sounds! Make sure that each word/phrase can only have one code attached to it at any point during analysis.

Step six:

Use the appropriate tagger software to mark up all parts of that particular document that relate to those specific themes/categories/codes you’ve decided upon.

Just learning about steps wouldn’t suffice. You will also have to understand how to use the procedure efficiently.

Our top tips on using coding and tagging strategy

The steps in the coding and tagging strategy are to prepare, code, tag, analyze the research data (or documents), and finally write up your material. 

To do this you need access to a computer with an internet connection for each step of the process. You will also be using qualitative analytical tools such as Nvivo or Atlas-ti. 

It allows you to conduct manual coding alongside automated text analysis, without switching between different software programs. This makes it easy to integrate into any existing workflow.

It’s important not just to look at the words used but their context too. 

So consider how many times they appear throughout your texts/submissions, and more, whether there were specific patterns across the text or whether each time they were used it was in a new context. 

Your goal is to get into the mind of your participants.  Think about what their motivations are for using particular words, why certain words appear more than others etc.

Also, consider how language changes over time and if there’s anything you can do with any specific expressions that have evolved since data collection took place. 

For instance, did people use ‘hashtagging’ on social media when you collected the material?

You may also want to consider the frequency of terms across different data sources. For example, you might find that certain words are used more in one text than another. 

But, there is no reason why this would be unless it had an impact on your participants’ use of language or what they believed was important at any given time.

If you have access to transcripts for interviews and focus groups then look out for commonalities between these two sets of information. This approach can help inform how representative your coding system is.

Last thing on frequencies: when thinking about which concepts/themes etc. to code make sure each word represents only one concept. Otherwise, it can be difficult to identify when the same words are used for different concepts.

After you have coded your data, this is where things get interesting! It’s time to start thinking about how well each category represents the underlying phenomena that you want to understand. 

You may find out some themes or categories represent certain ideas in a very similar way (resulting in redundancy). Whereas, others capture distinctly different meanings (resulting in fragmentation).

If there are lots of overlaps between codes then consider merging them into broader categories. 

If they seem distinct and important enough then maybe create sub-codes instead. There isn’t one ‘right answer’ here but what an analytic induction approach does offer is flexibility. 

You can try it out and see what works best for your data. 

Much of the work in analytic induction is about trial and error. Hence, this approach allows you to play around with different ideas, concepts, or categories until they resonate most strongly with your particular dataset.

Final thoughts

It can be an overwhelming task to code and tag all of your qualitative data. 

However, with the help of some good tools, it becomes manageable. 

Hopefully, this post has shed light on what coding and tagging strategy is as well as how you can successfully implement it in your research process.