Thematic analysis is a way of identifying themes or topics within a body of written data, for example from transcripts of interviews and focus groups, online survey responses or social media comments. The researcher combs through the data to identify recurring themes and patterns of meaning.
Traditionally, thematic analysis is a time-consuming, resource-draining process for researchers and analysts. Manual analysis is subject to researcher bias and human error. As a result, qualitative data is often distrusted or perceived as less valuable than quantitative data which is easier to analyse objectively.
Thematic analysis software has been around for years to assist in automating the process of generating themes. However, researchers often still need to put in substantial time to manually tag the data and train the software to recognise patterns, and the AI is often not advanced enough to accurately recognise human speech patterns beyond simple synonyms.
Recently, however, advancements in natural language processing (NLP) technology have enabled automated coding and theme generation with more accuracy than ever before.
The 6 Steps of Manual Thematic Analysis
There are various approaches to conducting thematic analysis, but one of the most popular forms, and the one we’ll describe here, is referred to as “Reflexing TA”. The reflexive TA method was developed by psychology researchers Virginia Braun and Victoria Clarke in 2006 and follows a six-phase process.
The first phase involves immersing yourself in the data by reading it over repeatedly. The purpose of this step is to familiarise yourself with the content and make initial notes of your observations on each data item and the entirety of the dataset.
The next step is to label your data with codes that define important and relevant features within the data. Coding must be done across the entire dataset, and this step should be repeated at least once more to prevent you from missing anything.
- Theme generation
This next phase involves reviewing all the codes and organising them into broader themes and patterns. Braun and Clarke define themes as a “pattern of shared meaning underpinned by a central concept or idea.”
- Theme review and development
At this stage, you should examine all the themes from the previous stage and compare them against the dataset as a whole. Identify which themes are relevant to the research question. Develop themes further or discard irrelevant themes.
- Refining, defining and naming themes
Develop a detailed analysis of each theme to truly understand their meaning. At this stage, you should come up with a succinct and informative name for each theme so that it’s easy to understand.
- The write-up
Finally, it’s time to develop an analytic narrative that combines all of the relevant themes. Reflexive TA was developed for use in an academic context, but this step also applies to writing informative reports for use in a business setting.
The Challenge of Manual TA
Now that you’re familiar with the six-step method, it’s easy to picture just how much of a time drain thematic analysis can be. It’s clear why researchers and analysts favour quantitative data. When qualitative data is used, research is often conducted on a smaller sample, providing less breadth and scope than would be possible if time and resources weren’t a factor.
However, qualitative data is extremely rich in information. Not only can it better describe people’s feelings, attitudes and experiences, but it can also bring to light new information that may otherwise be overlooked.
On top of that, there is also a vast amount of readily available untapped qualitative data online in the form of social media comments, posts and reviews. The greatest challenge researchers and analysts are facing today is battling with an insurmountable quantity of data streaming in by the minute. Manual TA is too time-consuming for practical use in many situations.
That’s where thematic analysis software comes in.
Thematic Analysis Software
There are many different types of thematic analysis software out there. Some of them simply make it easier to manually highlight and organise data. Quirkos is an example of TA software that helps you visually organise your data, but the heavy lifting (coding and theme generation) is still up to you, making it unsuitable for large data sets.
Others require a lengthy amount of time to train the software in an initial setup phase. This is also time-consuming and can be difficult for people who are unfamiliar with how the process works.
Natural language processing
NLP is the bridge connecting human speech to computer processing. Human language is extremely complex. There are many ways we can express an idea, experience or emotion. We can be subtle, we can be descriptive, and we can be obtuse. To decode meaning, we sometimes need to read between the lines.
Humans process this information without too much effort, but this task is extremely difficult for computers. That’s where NLP comes in. NLP programs enable computers to make sense of large amounts of natural language in the form of written text.
Symanto is at the cutting edge of natural language processing (NLP). Our technologies enable automatic topic detection without needing to spend hours pre-training software, making it the best option for large data sets.
While some TA software uses basic synonym detection, the NLP technologies we develop at Symanto are far more advanced and able to capture nuanced communication in unstructured, qualitative data without the need for excessive pre-training. This makes thematic analysis with Symanto faster and more accurate.