Sarah Abel
март 2017.

How useful are programmes like NVivo and Atlas.ti for coding qualitative data?

1 ответ

I have taken a couple of courses on Computer-aided qualitative data analysis software (CAQDAS), and have used it in my research. Once I used the program Dedoose for a paper with a co-author, and we started with it from the moment of organizing our data, and used it until the very end. Another time I started analyzing a data set of 36 interviews with Atlas.ti, but quit early on and went back to old school methods for the rest of my analysis.

After these experiences, I have concluded that  benefits of CAQDAS include:

  • Making it easier to work in groups on one set of data, 
  • Making it easier to organize your data on your computer,
  • Making it easier to retrace your steps when you're trying to do analysis. 
  • In addition, if you get stuck conceptually or have writer's block, I find it's easier to "get going" again if you go back to coding in a program than if you try to go back to a stack of papers with highlighted text and scribbles on the margins.

Atlas.ti is a popular program, yet it is very expensive. Many universities have subscriptions, and some allow you to access them from home for free or for a limited charge. There are also student discounts. The cheapest program out there from what I know is called Dedoose, it's about $10 a month for students. It's slower than Atlas, but has most of the same functions and perhaps even better visualization tools. Dedoose's conceptual innovation is something called a "descriptor," which you can read about on their website. Both Atlas and Dedoose have strong communities of users and developers who are quick to communicate about ways to improve the programs, so you can join those and chat away about CAQDAS with others, if you like.

One question to ask yourself is if you're prepared to invest some time into learning how to use one of the programs. The usual thing you hear at the outset of courses on qualitative data analysis software is that it makes qualitative work easier to a degree, but the learning curve is quite steep. As the programs evolve, the learning curve gets less steep, and once you get the hang of what coding actually is, the rest I find to be quite intuitive across different programs. If you want to see how difficult it is to master, I would sit down with someone who uses the software and would click around with them until you feel comfortable. The hardest thing is uploading files and converting various formats, but the coding itself is quite easy. 

Another question to ask yourself is whether you like "coding" as a qualitative technique, and if you use it outside the software as well. If you work in a grounded theory style and intuitively start coding text whenever you're facing a pile of papers, CAQDAS is likely to be a godsend. If you're unsure about what coding is and how you feel about it, you might find Kathy Charmaz's book "Constructing Grounded Theory" helpful, or Johnny Saldana's Coding Manual.

I have also found that it is useful to divide the functions and uses of CAQDAS into two categories: handling data and analyzing data. Data analysis can, in turn, be subdivided into two categories: mechanical analysis and conceptual analysis. Of course, the line between handling and analyzing data is blurred. When does the analysis actually begin? Perhaps with sorting data?


A prerequisite for quality qualitative analysis is good organization of what is often a large mass of data. CAQDAS helps precisely in this “data organization” dimension of qualitative work.

Effectiveness of data organization and code organization can be increased by the use of CAQDAS through the inherent neatness of computer coding (no need to cross out codes or make numerous paper copies of data), through the use of the query tool to find and link existing codes, and through faster visual categorization that automatic features of Atlas.ti software allows (“network view”). “Effectiveness” does not imply that working with data is faster in CAQDAS than it is by hand; CAQDAS is still a time-consuming mode of analysis and requires some steps in data preparation as well as a lot of time in mastering all the computer skills needed to navigate and use it comfortably. A period of learning, adjustment and trial and error is necessary before one is capable of using CAQDAS for comfortable and helpful data organization and interpretation.

A very important way in which CAQDAS can be helpful is in situations in which you work with a vast and complex variety of different empirical data sources, from videos to texts to audio files to images and so on. CAQDAS allows you to store all of this in one HU file, which can be helpful cognitively: it allows you to see and to treat your empirical base as one whole, which structures your thoughts during your analysis.


Mechanical aspects refer to the technical mark-up of texts with codes, generating lists of codes, searching for key terms, and so on.

Conceptual aspects of analysis include the process of thinking up codes, assigning codes, identifying key ideas of the text.

It follows from this that learning CAQDAS is not synonymous with learning how to do qualitative analysis. Mechanically generating codes and doing searches in and of itself does not produce analysis.

In this regard, it is important to take note that coding both in CAQDAS and without CAQDAS is done by the researcher; the researcher decides on the types of codes and on the length of quotations used. Moreover, CAQDAS was created by researchers for researchers based on the ways in which researchers work or prefer to work. In this way, human agency remains a key feature of coding in CAQDAS. CAQDAS has not become a methodology in and of itself, but remains a tool in the hands of social scientists.

The analysis that one conducts after organization and coding of data in CAQDAS can be easily linked to the initial empirical data because one can share and distribute hermeneutic units between computers and show other researchers or readers what their coding and data organization looks like. If used as evidence of one’s analytical claims in the papers produced based on the empirical data, the files from CAQDAS can be a way to make such claims more transparent. Sharing hand-made documents is not as easy or as straightforward because they are often messier, harder to distribute and contain less standardized modes of organizing data than HU files. This transparency can add a degree of legitimacy to the analysis produced and can also yield stimulating methodological reflections on the entire process of organization, coding, interpretation and analysis of a given researcher.

The presence of this in-between documentation between raw data and the academic context containing analysis can manifest itself in a near, open-for-viewing file which readers and reviewers could potentially access (and some journals even require this), and gain very important insight into how the researcher dealt with the data.

Thus, by increasing effectiveness and transparency of data organization and analysis, CAQDAS can sometimes be helpful for qualitative data analysis, but since the researcher retains ultimate agency in deciding how to use CAQDAS for their purposes, it does not yield qualitative revolutionary changes for how social science research is done at large.

So CAQDAS does not magically provide solutions to analytical problems social scientists face when working in qualitative data analysis or in a grounded theory approach. Nor does CAQDAS dehumanize qualitative data analysis through mechanization of interpretation. CAQDAS simply provides (in some cases) a more effective way to organize data and code, and allows for an increase in the transparency of analysis by making it easier to link analysis to empirical data. The researcher, however, always retains reflexive agency in relation to data and in the process of interpretation, whether using CAQDAS or not. 

Olga Zevelevaотвечает на ваши вопросы в своейПрямой линии