What is coding?
Codes are tags or containers for complex qualitative data; they are a way to index, sort, link and interrogate data.
The process of coding involves finding ways of summarizing key meanings within your data.
The same piece of text can have multiple meanings or codes.
Within this section we will cover the conceptual grounding for qualitative coding and set through the basic processes for creating codes and coding within NVivo.
What are codes?
- Codes are tags or containers for complex qualitative data; they are a way to index, sort, link and interrogate data.
- The process of coding involves finding ways of summarizing key meanings within your data.
- The same piece of text can have multiple meanings or codes.
Deductive vs inductive coding
In deductive coding, you approach your dataset with a pre-established list of codes, often based off existing theoretical frameworks or even the research tools themselves (e.g. the questions that were asked). This is a more top-down, etic approach that is most useful for confirming or testing existing theories, or imposing organizational and structure into your file (e.g. coding by interview guide or research questions).
Conversely, inductive coding is when you allow codes to emerge from your engagement with the dataset itself, with no preconceived categories. This is a bottom up, emic approach to coding.
Your own approach may be heavily influenced by your discipline, research aims, and research questions.
However, most work will be situated in a hybrid space between the two.
Coding as iterative
Sorry folks, qualitative coding is far from one-and-done. Coding is a cyclical, ongoing process. Interviews cannot be read and coded just once. As we re-read our datasets, we continue to develop new codes and subcodes, and refine our understanding of the data. Each time we read our material, we gain a deeper level of understanding each time we read our transcripts. With each pass, we will notice new things, which may prompt us to read new academic articles or ask different questions. This work will in turn shape what we see and understand in our next reading of the dataset. Nvivo can help us organise and interrogate codes but coding is led by the researcher.
The practice of coding
Remember, when coding in Nvivo:
- Codes can overlap
- Codes will sort themselves to be alphabetical list – so clear, direct, short codes are helpful.
- Consider your future self (and any collaborators), and try to annotate your coding by always including a code description: what you think this code means and what should be included here
- Codes should change as your project changes. They can be combined and rearranged.
- Don’t be afraid to make a code because you’re uncertain if it will be relevant or important enough. They’re a tool to think with and can always be adjusted later.
- Be careful: try not to code when tired, bored, or distracted – you must always be actively thinking (the ‘organic computer’ [your brain] is the primary tool of analysis).
Autocoding or using AI for coding?
You might have heard about autocoding and potential use of AI in Nvivo 15.
Autociding is currently a basic - but often useful- functionality in Nvivo. It can help you organise the text into cases or themes based on pre-defined criteria using a deductive apporach (we’ll show you these below!)
However it is the researcher’s job to interpret and make meaning from data.
In Nvivo version 15 there is an AI helper, which can detect common words or patterns of text in documents, and can be used as a sort of data-driven inductive coding assistant. Again, this is currently quite basic and shouldn’t replace the essential work of the researcher making sense of the data themselves.
But to find out more here at the AI features in Nvivo 15, see these videos from Lumivero:, and an online youtube video.
Different approaches to coding - learn more
See resources tab for more information about various approaches to coding.