Help:QDA best practices

This document outlines some suggested best practices for Qualitative Data Analysis.

Do a first reading

When working with a new dataset, read through some of the documents before starting to code. This will help to gain a sense of the main themes that are likely to emerge.

Use the search function

If you are uncertain about the importance of a theme, or discover that you overlooked an important theme earlier on in your coding, use the search function in your software to find and code passages on related topics. Atlas.ti web includes a built-in thesaurus that will allow you to locate passages containing similar search terms.

Use prefixes to group codes

Use prefixes to group codes in Atlas.ti. In contrast to NVivo, which uses a hierarchical coding structure (code tree), Atlas.ti uses a flat code list by design. This is because Atlas.ti aims to support an inductive approach to coding, whereby codes are generated from the data and later grouped, rather than being organized from the outset into conceptual categories. Since codes are listed alphabetically in Atlas.ti, an easy way to group them is by using prefixes. For example, we can have groups such as "WELL-BEING: satisfaction with work" and "WELL-BEING: poverty".

Always keep working definitions

Always keep working definitions of codes. As you work through a dataset, your understanding of the represented themes and concepts should evolve, and precisely stated definitions allow you to distinguish between similar themes or concepts -- allowing you to apply codes consistently, but also to combine and split codes as needed.

Working definitions will also help communicate the coding structure to assistants and team members who collaborate in the analysis, or to subsequent researchers who may review your coded dataset. If you are using an inductive (grounded) approach, codebook containing definitions can also be useful when building theory at a later stage in the project.

Start with specific codes

When working inductively, start with codes that are relatively specific -- but not so specific that they are unlikely to be applicable to other documents. Keeping as much relevant detail as possible at an early stage will ensure that important ideas and themes are not overlooked, while at the same time, it is essential to avoid what QDA researcher Susanne Friese calls "code swamp" -- hundreds of codes that are applied only once or twice across an entire dataset.

If in doubt about the specificity of a code, consider applying two more general codes rather than one specific one: for example, instead of coding "female migrant trader who enjoys her work", you could apply the separate codes "female trader", "migrant trader", and "enjoys work" to the same passage. Later on, you can use the code manager and report functions in the software to explore the combinations of these tags.

Use codes to organize workflow

Use codes to organize workflow in addition to content. Codes can be used to mark quotations or documents that you wish to review later, assign to someone else in your team, identify as "coded", exclude from the analysis, or label as particularly unique or important (e.g., a passage that you would consider quoting in your research).

Various symbols and prefixes can help organize these codes in a way that makes sense to you: for example, you might use a question mark prefix for codes such as "? check" or "? unclear", and the "@" symbol to label items for action by individuals in your team, such as "@Nikita review". The "comment" field for these codes can be used to provide further details.