An Introduction to Qualitative Data Analysis Software
I’ve been very busy in the last few weeks and unfortunately, I was unable to summarize the progress of our research. We’ve started the interviews, and the results are very interesting so far. I’ll try to tell you all about it as soon as I have some free time.
In the meantime, I’d like to present the content of an introductory session on qualitative data analysis (QDA) software that I hosted as part of a methodology course at Université de Montréal.
I hope that it’ll help those who are less familiar with this type of software.
Introduction to Qualitative Data Analysis Software
1- About Me
My name is Pierre Blais. I have a Bachelor’s degree and a Master’s degree in Anthropology from Université Laval. Last spring, I completed my Ph.D. at Université de Montréal. I first started studying immigration during my Master’s degree. More specifically, I focused on the evolution of power technologies in the immigration policies of Quebec and Canada. Then, as part of my doctoral research, I studied the reasons that motivate French immigrants in Quebec to return to live in France.
During my Ph.D. studies, I got a job at Logiciels Ex-l-tec. As part of my work, I submitted a project to the National Research Council of Canada (NRC) to develop a new qualitative data analysis (QDA) software. The project was accepted, and I became the project manager … and that’s what I’ve done for the last three years.
2- A Brief Overview of How Qda Works
2.1- What Are QDAs For?
LiGRE is like Nvivo, MaxQDA, Atlas.ti or any other qualitative data analysis (QDA) software in that it allows to associate excerpts of texts to one or more keywords.
As such, this type of software can be really useful since it allows you to classify and organize the significant elements of large sets of texts and to navigate and filter their content more easily.
In anthropology and sociology, QDAs are most often used to analyze the content of interviews recorded or filmed during field research.
2.2- How Do QDAs Work?
In principle, this type of software is relatively simple to operate and can accommodate both proponents of inductive and deductive methodologies.
Since we’re in an anthropology class, I’m going to focus here on the inductive use of QDAs, which mainly consists in letting the data show us what is relevant and significant … and not submit them to a predefined conceptual framework.
2.2.1- Steps to Produce an Inductive Research with a QDA
- The first step is quite simple. Import your text data into the application that you’ve decided to use.
- The analytical phase can then begin. This step is often referred to as “data coding.” Invariably, this step consists in reading the contents of one of your interviews, locating a segment of text relevant to your research problem, creating a keyword and associating the text extract with the newly created keyword.
- Once a text extract is linked to a keyword, its status changes in the text. It will either be highlighted, colored or underlined depending on the software you use.
- This chain of operations is repeated each time you encounter a new text extract related to your research topic.
- Several text extracts can be linked to a keyword, so if the keywords have already been created, simply link the text extracts directly to those previously-defined keywords.
- Once the coding of the first document has been completed, repeat these operations for each project document.
2.2.2- Coding Arborescence
Also, most of these applications allow you to link keywords (parent-child relationship) and to organize them in a tree structure.
2.2.3- Saturation Point
Once a certain number of documents have been coded, you will reach a saturation point, which means that you no longer need to add keywords or edit the structure of the coding tree structure since all concepts linked to your research topic have been “discovered”, so to speak, and properly ordered.
Once this point is reached, the coding process is much faster since it only needs to link the new extracts to the existing keywords.
2.3- Why Use a QDA?
These programs allow you to find the text excerpts that you indexed quite easily. Usually, all you need to do is select a keyword to see the entire family of text excerpts linked to it. So, for example, if I select the keyword “discrimination”, all text excerpts linked to this keyword will appear. It will also show me in which exact document these excerpts are. It is therefore very convenient to search and quote important passages.
2.4- When Use a QDA?
The coding process can be lengthy, especially until the saturation point is reached. Also, QDA can be very expensive. For this reason, it is recommended to use this type of software when handling large volumes of data.
For example, under five interviews or texts, these programs are often useless. You can do the same job on your own and without much difficulty; however, they will prove to be real time savers when dealing with a dozen interviews and only become a necessity when dealing with over twenty interviews.
3. Practical Demonstration of LiGRE
Before starting the presentation, I’ll introduce some of the most interesting features of LiGRE.
1. First, and unlike most software on the market, LiGRE is a Web application not software. As such, LiGRE…
- does not require any installation or update. Just log in to the app to access it;
- is compatible with all operating systems (Windows, Mac, Linux);
- allows you to access your data anywhere, regardless of the platform you use. No inter-device synchronization is needed.
2. LiGRE is a relatively new application. As such, it lacks some of the more advanced features found in older QDAs; however, LiGRE also features exclusive and innovative gadgets. The most interesting one is the automatic transcription module. This module allows LiGRE to automatically transcribe the content of an interview that you have recorded or filmed. It can also automatically transcribe video content from sharing sites like YouTube, Dailymotion, Vimeo, etc.
3.1- Welcome Screen
Once you log on, the Home screen appears. This screen lets you create a new project or select an existing one.
At this stage, I must specify that all QDAs work by project. In a QDA, projects work the same way as folders in Windows or Mac. That is, they allow you to group and organize files of a similar nature. In addition, it is important to note that all documents in the same project will be analyzed with the same keywords and the same coding tree structure. This division was made to allow users to conduct several researches and analyze documents of different types simultaneously.
I’m currently working on a big project. Given that we’re still at the beginning stages and that the project doesn’t contain any sensible data as of yet, I’ll use this project for the demonstration.
When you select (or create) a project, you first access its dashboard. The right portion of the dashboard presents the project’s information. I will not spend more time on this aspect here. The left portion lists the options and tools available in your project. This is the most important portion since it is through those options that you can advance your project.
3.3- Add and Transcribe Data
Since it allows you to add data to your project, this option is the most important when you first start your work.
Data can be imported into a project in several ways:
- Text documents can be added by copying/pasting the content of a text file.
- If you have conducted audio and video interviews, you can import them and have the application transcribe them. Once their transcription is completed, you will receive an email informing you that the process is over and that you can start coding the result.
- Videos can be embedded from sharing sites like YouTube, etc.
Some things should be considered if you plan to use LiGRE to transcribe your interviews. Most of all, you should know that the quality of the recording and of what has been recorded will greatly influence the application’s ability to transcribe what you’ve submitted to it.
Regarding the quality of the recording, you don’t have to invest a fortune in a super voice recorder. You only need to use relatively new equipment. For example, an iPhone 6 will yield excellent results. The same can’t be said for the mp3 voice recorders lent by the university. You should also avoid using a one-way microphone in a noisy place. Ultimately, the golden rule can be summed up as follows: “Try to maximize the clarity of your recordings.”
Regarding the quality of what is recorded, the most important thing to remember is that LiGRE can only transcribe words if words are submitted to it. In other words, the result will be poor if you record someone who grumbles, who responds by onomatopoeia, who chews his words or who stops in the middle of a word.
I’ll use a video from YouTube as a demonstration. It’s a blog video in which a Canadian woman explains why she left Canada. Even if the video is not very long, its transcription will still take a few minutes. I’ll continue with the presentation, and will come back to it when it’s done.
3.4- Data and Codification
The second option lists all project documents and their status. As you can see, the document I added is being transcribed.
This page is important because it allows you to access the codification page, where the interviews are coded. To do this, simply select a document and click on “Continue coding”.
I’ll introduce you to this page using another YouTube video.. This is a video blog where another Canadian woman explains why she left Canada.
3.5- Codification Page
As you can see, this page shows the interview transcript in the right portion of the screen and the coding tree structure on the left side. Also, you can find some step-by-step tutorials at the top of the page to help you understand how it works.
As you can see, I’ve already created my coding tree structure. It’s important to note that a coding tree structure can only have three levels in LiGRE. Why three levels? It was our choice; it seemed enough to us. Other software allows more levels, others less.
In the application, we call these three levels: root, branch and leaf. Although they are at the last level, the leaves are the most important since you can only link text excerpts to them. Roots and branches are only there to help you organize the leaves. They are not necessary but add clarity.
To add an element to the coding tree, just click on “Edit coding tree.” Choose the type of item you want to add and specify the information you want to save.
To edit an item, select it from the drop-down list. Its information will appear. You then need to edit and save.
For the demonstration, I’ll add three leaves. As I said earlier, the leaves don’t have to be linked to branches or roots. You can create them and then organize them afterwards. Those leaves are automatically placed in the “orphan” segment of the coding tree structure. Once created, they are ready to use. So, we’re all set to start coding.
I’d now like to draw your attention to the text of the transcript. What do you notice?
It’s full of mistakes. There’s no punctuation. The transcription system cannot make punctuation for the moment. We’re working on it.
Another important point is that the text of the interview has been segmented into paragraphs. The system does this automatically, as soon as it encounters a pause or hesitation. The paragraph division does not always make sense, but it’s easy to fix with the features of the page, as we will find out later.
We’ll start by correcting the text. To do this, simply place your mouse on the paragraph that you want to edit. A menu to the right of the paragraph number appears. Choose the first option, “Play and edit the paragraph.
As you can see, we can listen to the interview segment corresponding to the text segment that was transcribed. You can slow down the flow to make it easier to correct the text.
I’ll start by specifying the speaker and will then make a few minor corrections such as punctuation. Once I’m done, I save everything.
Repeat the operation for each paragraph.
When some relevant text has been split into two paragraphs, it can easily be reunited with the “Merge paragraph” option.
Now, to code an excerpt, just select it in the text. A menu will appear under the paragraph. You then need to choose the leaf to which you want to link the excerpt.
Once this is done, the snippet takes the leave’s color, and you can see which leave the snippet is linked to by sliding your cursor over it.
Click on a coded segment to modify it.
Notes can also be added to the coded excerpts.
We’ll return to the previous page, “Data and Codification”, to find that the automatic transcript of the YouTube video that I uploaded is completed.
We’ll go see the result and quickly code some excerpts. But first, we’ll control the quality of the transcription by listening to some of it. As you can see, it’s not too bad.
3.6- Recurrence Analysis
The next and last option I am going to present here is the “Recurrence Analysis.” This page has many advanced options. I will focus here on the most important thing: that it graphically represents your codification.
You only need to select a branch to see a circular diagram that details the number of text extracts linked to each of its leaves. The exact number of excerpts related to a leaf can be seen by sliding your cursor over it. You can then access all text excerpts linked to a leaf by clicking on it.
It is therefore very convenient to quickly locate significant extracts in the content of your interviews.
I’d like to end this lecture by insisting on the fact that qualitative data analysis (QDA) software can be used for many more other complex and useful operations. My goal here, however, was specifically to give you an overview of their basic features.
This type of software differs in its advanced features. For example, although Atlas.ti does not offer automatic transcription, it allows you to code directly on audio and video files. There are obviously several other nuances differentiating the software, but the common core of that family of software is the codification feature, which, as we’ve seen, is quite simple. With the exception of the automatic transcription, I could have done all the operations that I have shown you with any other QDA software. Once you understand their basic principles, you can use any or all of the different QDA software on the market, whether it’s MaxQDA, Nvivo or LiGRE.