More often than not, the purpose of building an online community is to gather large amounts of qualitative data. This post will look quite broadly at how you analyse and code qualitative data.
Analysis of qualitative data involves looking for themes, patterns and relationships within data. Taking an organized and thorough approach to the dataset ensures that nothing is missed out or wasted. Whilst it is important to take a structured approach to analysis, it is equally important not to become too formulaic – flexibility and fluidity is crucial.
The goal of analysis is to ‘distil the essence’ of the data you have available, and this essence will vary based on the type of project you are working on. For example, an advertising or PR company would want you capture the creative or inspirational essence of research, whereas insight and reliable evidence might be the goal with public sector clients.
To get to the essence of the data it is necessary to dissect, scrutinise and label chunks of content and place them under relevant headings so that relationships can be discovered. This process is called coding and there is a lot of software that can help you do this, from free open source programs to expensive feature rich packages. Alternatively, a spreadsheet or word-processor can often be just as effective if you are working with a small dataset.
When coding, the first distinction to establish is whether you will take a deductive or an inductive approach to your project. Inductive means that your code scheme will come from the data itself whereas deductive refers to approaching your data with some theoretical ideas or concepts.
Most of the time you will be using both approaches. A deductive approach will allow the coding to be defined by a combination of a researchers own personal knowledge (acquired from their own reading and experience), the research brief and the contents of the discussion guide. This would be followed by an inductive approach where the contents of the dataset go on to determine part of the coding scheme.
The questions that you ask when coding qualitative data can be categorised into three approaches:
- A literal approach – what words, dialogues, actions etc are used?
- A interpretative approach – more analytical: what are the values, norms, rules, etc of what is being said?
- A reflexive approach – what role have you as a researcher had on the study e.g. does the way questions are worded have an effect on the response?
With this in mind you can begin to approach your data by asking these basic questions:
- What is going on?
- What are people doing?
- What are people saying?
Once these basic questions are addressed you can begin to probe a bit deeper:
- What assumptions are people making and what are they taking for granted?
- How does context affect what is said or done?
- What particular words are being used and how frequently?
- What concepts are people using to describe things?
- What words are used in which contexts?
- What are the repeated words and how often are they repeated?
- What metaphors and analogies are people using?
- Where are people attaching significance?
- Where are people placing emphasis?
- Are there instances of slang or colloquialisms?
- How are transitions used– such as pauses, changes of tone, new sections and topics?
- What pronouns do people use e.g. do they use I or us to describe an emotion?
- How does a term or expression compare to something similar or different in other instances in the text?
- What linguistic connectors are used to create relationships between words – e.g. because, before, after, next and ‘for instance’?
- Are there any noteworthy omissions - something you expected to see but is missing?
- Are there contradictions between different parts of the data?
- What interesting stories are in the data?
These questions will help you create a code-framework that should cover all the salient parts of the dataset. Remember to keep an open mind when coding and don’t think of your code-framework as fixed – codes can be expanded, combined, split into subcategories or simply thrown away. Bear in mind that the same chuck of text can be coded more than once and that there are no rules on how large or small a highlight should be.
During the your second or third round of coding, new relationships will emerge and your code scheme will become more refined. You will want to begin using codes to highlight good verbatim’s that you will want to highlight in your final report.
Once you have chopped up your data and organised it into a strong coding frame you are ready to reassemble into a report.