The Iterative Qualitative Data Analysis Process with ATLAS.ti: A Dialogue Between Description, Reflection, Analysis, and Interpretation

October 29, 2020

In simplified terms, you can approach your work with ATLAS.ti either in a lineal way or in an iterative way. A lineal approach can be represented by segmenting and coding first, and postponing for later the process of reflecting, asking questions, and ultimately making sense of the data. I see this approach as a trap that prioritizes speed over in-depth understanding and can lead you to a potentially large number of codes but without really developing an understanding of the data. To avoid that trap, I suggest to approach your work with ATLAS.ti by establishing a dialogue between working at the data level, which entails paying good attention to the quotation as the minimal unit of analysis, and a higher level of abstraction represented by tasks related to exploration, organization, analysis, and interpretation.

This iterative and dialogical approach I am suggesting can be represented in the figure below:

Figure 1. The Iterative Process in ATLAS.ti

In this figure, the data analysis process is indicated by points in time. You could say that at time point 1, you work at the data level. After a while, you stop and at time point 2 you look at the work you have done so far, learn about it, and use that knowledge to inform your continued data level work at time point 3. And so on, so forth. According to this approach, the data level work, which focuses on the quotation, informs higher-level tasks related to data exploration, organization, analysis, and interpretation. And, at the same time, these higher-level tasks inform data level work. In practical terms, this translates into constantly reflecting upon the data, asking questions, and writing throughout the process.

The table below lists the ATLAS.ti tasks that that can be used as we approach our work with ATLAS.ti in this iterative or dialogical way. Although I list these items in this table, it is very important to remember that you, the researcher, will decide what ATLAS.ti tools to use and when. That decision is informed by your research design and the methodological framework within the qualitative that informs your data analysis work. As an illustration of this, if you are following an interpretative approach in data analysis, it could very well be that you will not use quantification tools such as word clouds, word lists, code frequency counts (Code-Document Table) or co-occurrence matrices. At the same time, you may decide to emphasize rich descriptions of your quotations over coding. Or, if you are following a grounded theory approach, the theory building allowed by code-to-code linkages may play a more relevant role than if working from other methodological traditions. Or, if working from a discourse analysis perspective, it could very well be that you will make good use of the quotation-to-quotation (hyperlinking) tool. Thus, do not take the content of this table as a prescription you have to follow; instead, think of it as possibilities to explore.

Tasks in the Iterative Approach to Data Analysis with ATLAS.ti

Figure 2. Tasks in the iterative process

In Figure 2, under the ‘Data Level Work’ column, I list all of the tasks I think enrich your work with quotations. Let me start by saying that I like to think of quotations as the voice of the participant and, as such, they play a key role in qualitative data analysis with ATLAS.ti. If we agree with Creswell (2013:47) that a universal of qualitative research is understanding participants’ meanings, then, clearly, we do need to pay attention to the quotation. And that involves a number of tasks: creating the quotation, reflecting upon its content, adjusting its size as necessary, coding, and defining codes well. I would say that by actively reflecting upon the quotation content in the form of quotation comments you are building a potentially deep understanding of the data. But, of course, you will decide whether or not you should dedicate time to such detailed work at the data level. In fact, I have had plenty of students, particularly mixed-methods research teams, that tell me that they do not really have time to spend at the quotation level. Coding for them takes priority.

As you stop after a while working at the data level, take a look at the work you have done. As shown in the Exploration, Analysis, and Interpretation columns in Figure 2, this may involve examining quotations in context, creating reports of quotations by code (including other complementary information as well, such as quotation comments, codes used, and the identity of the document from where the quotation comes). You may also take a look at word frequencies in the form of word clouds or word lists, and from there you may want to examine the context within which some of those words occur. You may also want to examine individual documents, quotations, or codes in a systemic way, by placing them in networks and looking at the relationships they have with other items of the analysis project. And as you do that, ask yourself how that systemic view illuminates your understanding of the whole. You may also revisit the structure of your codes, memos, and networks, and start asking questions to the data. And, fundamentally, you will write in memos, your understanding of the data so far. For this writing you will draw from the reports you create, the visualizations you examine, the word frequencies you explore, the context within which words occur, and the questions you ask the project through analysis tools.

Following, you go back to working with quotations, this time trying to make use of what you learned as you explored, organized, and interrogated the data. After a while you will stop again and once again explore the data, interrogate it, and keep on writing in your memos.

Keep repeating these cycles until you feel satisfied with your understanding of the data. And it will not be difficult to know when to stop because, after all, you are building your understanding throughout the process. And, importantly, that understanding is always well grounded on the voice of participants.

About the Author:

Ricardo B. Contreras is an applied anthropologist and director of ATLAS.ti Training & Partnership Development. In that position, he teaches ATLAS.ti and provides support to users internationally. Ricardo’s undergraduate degree is from the Universidad de Chile and his master’s and doctoral degrees are from the University of South Florida. Ricardo has been teaching ATLAS.ti for more than 10 years. You may reach Ricardo at [email protected].

Reference Cited

Creswell, J. W. 2013. Qualitative Inquiry & Research Design: Choosing Among Five Approaches. Los Angeles: SAGE.

 

 

 

Share this Article

The all-new ATLAS.ti 9 and ATLAS.ti Cloud have been launched!

From now on, all ATLAS.ti 9 licenses include full access to ATLAS.ti Cloud at no extra cost.

ATLAS.ti 9 represents a huge step forward: Implementation of new technologies (e.g., A.I./machine learning), a sleek and unified design between platforms, and laser-focused concentration on user experience and usability set new standards.

A flexible and easy-to-use license management system makes it a breeze to manage licenses seamlessly between computers, and to distribute individual seats under a multi-user license.