top of page

Calling for a revolution – we have to get rid of codes

Codes are the example par excellence for our constant banging on about strategies and tactics. Reminder: strategies are what you plan to do, and tactics are how you plan to do it.

When using a CAQDAS program, the tactics are very different in nature from the analytic strategies. A strategy might be to compare the men and women respondents in a study by separately conceptualizing the male responses from the female responses. The tactic to fulfil it will depend on the CAQDAS package you use, but it will involve software tools that allow you to collect together selected items of data that have been tagged or grouped by you in the software so that you can compare them on screen or in printed form.

Whatever the tools in your chosen software, they will have nothing to do with the subtleties of male and female characteristics or gender issues. Instead they will involve processing data in the software to accomplish your purpose. These are two extremely different ways of thinking, but because both are called ‘coding’ you unconsciously and unhelpfully think about them in the same way. For this reason we should stop using the term ‘code’ for the strategies level of our conceptualization work.


We have accidentally got into the situation of describing an analytic task – ‘code this piece of text’ – using the exact same word as a tactic in the software – “create and then attach codes to text”. But these two processes are entirely different. The strategy – to ‘code’ – covers a wildly varying multitude of possible mental activities. At one extreme is the matching of clear-cut pre-determined labels to items of data that qualify as having a strictly defined meaning. At the other is the tentative generation of an intentionally ambiguous short phrase that captures certain aspects of what was seen that have relevance to a particular dimension of the research question.  And this is not to mention how detailed and specific or how general and abstract a good code should be to serve a particular purpose.

This brings to mind the far Northern Sami people who have 180 words for different kinds of ice and up to 1,000 words for reindeer that have different characteristics (yes they really do – see Note 1). The English language only has the words “ice” and “reindeer”. Using the single verb “to code” to describe dozens of different mental processes in different styles of qualitative analysis is as meaningless in the field of qualitative analysis as it would be for the Northern Sami people to only have one word for all the different types of ice and reindeer.  

Which brings to mind Johnny Saldaña’s excellent scheme of 32 completely different kinds of codes for different purposes (Saldaña, 2016).


Most CAQDAS programs use the word “code” as a tactic in their program (see Note 2 for exceptions), and this  does have a single well-defined meaning, and it has nothing whatsoever to do with any of the meanings of the word used for analytic strategies. “Code” used as a tactic is a neutral thing or object we create in the software. It has no meaning of any kind, it is just a cog in a machine. Learning to use the software means learning what all the cogs are (of which codes are just one) and how they work and mesh with one another – i.e., how the software works. Learning to harness the software means representing specific analytic tasks by one ore more of these cogs. (This is what Five-Level QDA is all about – keeping the strategies separate from the tactics because they are qualitatively different, and translating back and forth between them).


Does it matter that the same word “code” is used for both of these extremely different processes? YES! Even a long time expert user of a CAQDAS program is working with a human mind, and human minds are dominated by a pull towards ease and convenience. The easiest way to think is that if two things have the same name, they are probably the same thing. This habit of thinking is very hard to break (Duhigg, 2014), and particularly in the midst of a qualitative research project which is already mentally challenging and is taking MUCH longer to finish than originally planned. And if it is hard for an expert imagine how hard it is for a newcomer to qualitative research.

Unless we consciously fight this habit of thinking we find ourselves conflating the single word “code” as a strategy with “code” as a tactic. This is what we call One-Level QDA – what-you-plan-to-do-and-how-you-plan-to-do-it as a single process. When “coding”, strategies then become the same as tactics, and tactics become the same as strategies. The problem with this is that strategies are harder to think about – emergent, ambiguous, iterative, etc. – and tactics are simple to think about – clear-cut, step-step-by-step operations in the software. So guess what, our analytic plan becomes gradually more and more cut-and-dried, more and more step-by-step, than our methodology calls for.


All I am asking is that we stop using the word “code” for analytic strategies, to help us break this habit. The word is perfect for software tactics, it has the right clear-cut connotations, like the rigid instructions written in computer code, or the four digits you use as a code to open your phone. For strategies let’s start using a different word. I like the word concept. The most fundamental idea of a concept is as a name for a group of similar or related things, whether concrete – like dog – or abstract – like truth. A concept can be thought of as a generic word for any collection of things, which covers all the myriad ways we currently use “codes” in our analytic strategies. Then there is no mental drive to conflate strategies and tactics, and the quality of our analyses and our skill in harnessing the software can only improve (Note 3).

Note 1: For the many different Sami words for ice and reindeer, see here.

Note 2: Transana uses the term ‘keywords’ and NVivo uses the term ‘node’.

Note 3: For more on this see topic the section Codes and Concepts in Chapter 5 of Woolf & Silver, Qualitative Analysis using ATLAS.ti/MAXQDA/NVivo: The Five-Level QDA Method.

Duhigg, 2014. The Power of Habit. New York: Random House.

Saldaña, J. (2016). The Coding Manual for Qualitative Researchers (3rd ed.). London. Sage Publications


bottom of page