Diagramming in ATLAS.ti: Moving Towards a Holistic Understanding of the Data
Author: Ricardo B. Contreras
In previous Newsletter and blog articles, I have discussed data-level work in ATLAS.ti. I have highlighted the benefit of commenting, renaming and hyperlinking quotations; and have discussed several of the day-to-day operations involved in coding. In this issue of the newsletter, I will discuss diagramming and in a future issue of Inside ATLAS.ti I will examine memo writing. All of these topics represent what I consider to synthesize the core of data-level work in ATLAS.ti: the integration of data segmentations, writing, coding and diagramming. In yet another article I will explain what I mean by this integration and the role it plays in an iterative process of data analysis with the software.
Linking one element to the other is an intrinsic aspect of qualitative data analysis, and this is what we do in ATLAS.ti. Friese (2014:221) states that a project with ATLAS.ti is a system of linkages. As we link one element to another, we are building our understanding of the data. Elements can be linked to each other through two types of links: an unnamed link (also known as a ‘second-class relation’) and a named link (also known as ‘first-class relation’). These linkages are displayed graphically in the form of network views.
Network Views of Unnamed Linkages
Unnamed links represent the connections between elements that do not reflect any meaning beyond the simple representation of a connection. The following are examples of these kinds of linkages:
Network views of document groups/families
The network views above show something very simple: that four documents are grouped together according to shared attributes (i.e., they are interviews). The lines that link the documents to the node that represents the group or family, do not have names. Therefore, it is not possible to derive any meanings to this association beyond the simple representation of the grouping.
Network views of quotations
Another example is given by the linkage between quotations, the codes connected to them, and the documents from which they come. See below:
As it is shown above, a quotation has been coded by three codes, and it comes from a document that is coded by nine codes. In other words, that interviewee is discussing nine topics, and in relation to three of them she said what is represented by that quotation. Once again, the lines that link one element to the other do not have names and they just represent the fact that there is a connection between them; however, nothing can be implied about the meanings behind those linkages.
Network views of memos
Another example of network views showing unlinked linkages is the one showing a memo linked to a set of quotations and codes. This network view can be interpreted as following: What I am writing in this memo is illustrated well by these three quotations and is related to the concepts represented by these three codes. See the figures below:
Network Views of Named Linkages
ATLAS.ti also offers the possibility of building network representations connecting codes to codes, and quotations to quotations, through meanings. In these network views, the connecting lines do have names, and those names represent a meaning that the researcher has defined through interpretation. Hence, the establishment of these linkages is an intrinsic component of the data analysis process whereby the researcher has to determine the way in which concepts, themes, and the words of the participants relate to each other. These linkages are semantic. In the case of the code-to-code linkages, they can represent concept maps, cognitive maps, or folk taxonomies. See below a concept map representing the researcher understanding of funding processes, derived from the analysis of the data.
In the other hand, the network view representation of quotation-to-quotation linkages (also known as ‘hyperlinks’) resemble argumentation maps. That is, these network views show how arguments relate to each other. The researcher establishes these linkages as she asks herself questions such as:
- How do study participants construct their arguments?
- How do arguments contradict other arguments (if they do at all)?
- How do they support each other?
- Is what one person is saying expanding upon what the other person is saying? If so, how?
- How do some arguments illustrate other arguments?
Like in the case of code-to-code semantic linkages, hyperlinks are the product of interpretation and they can only be established by the researcher, not automatically by the software. See below the following example.
In this example, a picture quotation illustrates what two interviewees said about their health centers providing pediatric care and family medicine services. At the same time, what one of these interviewees said directly supports what another study participant said in regards to the same subject.
Network Views Showing Both Named and Unnamed Linkages
It is also possible to incorporate into a network view both unnamed and named linkages between elements of the project. For example, you may want to see graphically the codes connected to a quotation (or, in other words, what the study participant is talking about in a particular segment of the interview). If those codes are linked to each other semantically, then the network view will show those linkages as well. See below:
In this example, in quotation 5:1 the interviewee is saying that the clinic in which she works receives foreign funding and that there is partnership in service provision. The network view also shows that, as part of interpretation, the researcher determined that ‘foreign funding’ is a source of funding. Hence, in a single network view, both unnamed and named linkages are displayed.
As you work on your analysis project with ATLAS.ti, it is always a good idea to examine in the form of network views the linkages that you have created in the regular process of working with your data. These network views allow you to look at the components of the analysis project not as individual pieces or fragments, but rather as a system of relationships. They facilitate the process of building a holistic representation of your findings.
At the same time, as you work on your project, stop once in a while and ask yourself if you can establish semantic relationships between the concepts that you are exploring in the analysis (or that emerge from the data), and as you do that, build maps that reflect your understanding of the point of view of study participants. You may also represent, through hyperlinking, how you understand the way in which study participants build their arguments.
It is important to visualize your work in the form of network views because through them you may have insights that you might not have been able to have by looking at fragments, at pieces disconnected from each other. Those insights constitute the core of an analysis process; through them you are building the holistic representation of your data that is essential in every qualitative data analysis process.
Friese, Susanne. 2014. Qualitative Data Analysis with ATLAS.ti. London: SAGE Publications Ltd.
About the Author
Ricardo B. Contreras is an applied anthropologist with research interests in migration, community health, and qualitative methodology. He is a consultant in ethnography and qualitative methodology and director of the training division of ATLAS.ti Scientific Software Development GmbH. He lives in the city of Corvallis, Oregon, in the United States. Ricardo can be reached by email by clicking here.