Arguing from qualitative data

arguing from qualitative data

One of the main queries I get from research students is about how to develop an argument using qualitative data. When you’re sitting with a stack of narratives, how do you shape them into something interesting and important? How do you construct a clear story without losing complexity, and while letting people speak for themselves as much as you can? This is difficult, painstaking work.

This post doesn’t contain advice about data analysis but about what happens after: how you create interpretations from data once you’ve synthesised them into categories or themes, once you’ve understood key trends and identified any particularly interesting or significant cases. You’ll probably have done some form of coding to get here, whether software-based or by hand. Of course, the distinction between analysis and interpretation is permeable and sometimes even arbitrary: interpretation often starts at the data collection stage (or in bad research, before it), when arguments begin to form in your mind. But in many projects there will be a point where it’s necessary to shift up a gear. What do you really want to say about these data, and crucially, why? The infographic above contains the headings or key principles below: click it to download full-size (you could use it as your screensaver or wallpaper if you like).

Examine your motivations

Are you mining your data to find evidence for what you already think? If so, it might be time to stop and reflect on your own positionality and any views you hold particularly dear. There is no such thing as unbiased, objective research, but it’s down to all of us to be as honest and thoughtful as we can. Also, are you preoccupied with being clever and making your mark, or are you committed to saying something useful which is grounded in your data? Academia tends to showcase the former at the expense of the latter – research has shown that the pressure to innovate in natural sciences often leads to ‘bad science’ which prioritises surprising findings that are often wrong. I think social sciences and humanities can fall into similar traps, if we’re not careful. Decide now to approach your interpretations with integrity and an open mind.

Go back to your rationale and research questions

This sounds obvious, but many students don’t do it! Why did you want to do this study, and what did you originally want to know? Of course, you’re not bound by your original aims: often the process of research shifts them because our data tell us unexpected things. We should be alert to this – and if we are being open-minded and honest as we interpret our data, we may well derive unforeseen conclusions. If you find your data answer different questions than the ones you started off with, you can change your questions. This is often best done through an honest narrative, in which you present your original questions at the outset of your dissertation and then explain why they shifted (note that if they have shifted substantially you might need to add more material to earlier parts of your dissertation such as your introduction and literature review). In any case, revisiting your original questions will help you focus on what your data say, whether you set out to say that or not.

Go back to the literature

Whether this is your theoretical framework (if you have one) or the empirical literature or both, you need to think about how your data speak to it. Do they confirm what has already been written or are there new stories, unanswered questions or anomalies to be explored? If you’re using a particular theory, are your data consistent with it or do they expose any gaps in how it can be applied? If you analyse data in enough depth you’ll usually find weaknesses in existing theoretical frameworks: the social world is complex, after all. Build on others’ work – this is how understanding becomes full and deep. If you need different theories or literatures to make sense of your data, find them.

Having said all this, your intuition is also an important tool: are there things in your data that strike you, that make you feel happy, sad, or uncomfortable? Why do you think that is? Follow the trail of breadcrumbs – perhaps your first instinct about the data will lead you back to a particular piece of literature, or you might want to do some additional targeted analysis. But be aware: using your intuition as a starting point is different from plucking a narrative out of the air because you think it sounds good. Interpretation needs to be worked through: resist the temptation to name, to speak, or to conclude before you’re ready.

(Re)examine your concepts

Interpretation is often a process of shuttling between theory and data. As you make these journeys, check you’re clear about the concepts you’re carrying, and how you’re using them. For example, do you know what ‘power’ might look like when you see it in your data? Don’t carry ‘black boxes’ – empty versions of concepts that can be inserted into an argument as if they tell us something (when they don’t). Agency is a good example of a concept that can easily become a ‘black box’. If you think you can identify agency in your data, shuttle back to the theoretical definition. Then shuttle forward into your data to consider if you can really see it in practice. What does agency look like? How do you know if an action is agentic, or if it is not? The theory should be able to tell you, if you have engaged with it properly.

Be honest about what your data actually support

In the context of marking criteria (and scholarly norms) which prioritise ‘originality’, students often create arguments which sound lovely but bear little relation to their dataset. Beware ‘armchair theorising’ which is not grounded in your research: this might be your pet idea, but are you sure you can evidence it? Beware buzzwords which explain nothing, merely describe the familiar in different terms, and/or are just thrown in when we don’t know what else to say. Steer clear of inventing your own terms or concepts unless you have the data to back them up – and this often takes years.

Know the difference between novelty and significance

Novelty: something we don’t already know, but we don’t necessarily need to. An example of a novel research project might be to interview women called Paula about how often they eat tomato ketchup. This would be brand new information, but not necessarily that useful. Significance: something that challenges received wisdom in a substantive way (which does not have to be ambitious – knowledge tends to advance in increments). To develop significant findings usually requires quite a bit of shuttling, as the most obvious story about your data isn’t usually the most significant.

For instance, you might interview 40 women architects. The majority might highlight pay inequity and persistent everyday sexism, but think that initiatives to encourage women to apply for promotion are helpful. This is important, although nothing we don’t already know. What might be more significant is that the two Black women in your sample had experienced specific forms of gendered racism (or misogynoir) which, amongst other things, meant that they had not been put forward for support with their promotion applications. These cases, when interpreted alongside other trends in the literature, might enable you to argue that equality initiatives tend to target white women, and that when these initiatives are deemed ‘successful’ this treats white women’s successes as a proxy for women as a whole, creating the illusion of collective progress and masking the specific difficulties Black women face. When arguing from your data, you might prioritise this story over the more pedestrian narrative we have heard many times before. This choice is a political one, and this is the value of qualitative research: it allows us to dig deeper than the majority story and explore the nuances of social issues.

Ask the ‘why’ questions

Exploring the nuances means engaging with the ‘why’ questions about the trends, anomalies and interesting cases in your data. This also means you need to understand the position of ‘voice’ in qualitative research, and its positives and negatives. The common practice of using social research to give people a ‘voice’ is a laudable (if perhaps doomed) attempt to elevate marginalised voices and avoid imposing ‘false consciousness’ on research participants. But there’s a difference between honouring people’s views and experiences and taking them at face value – and we have a political and moral obligation to examine truth claims rather than reproducing them in completely unadulterated ways. Consider the use of white people’s ‘concerns’ about immigration to justify Brexit and other right-wing social and economic projects. Consider the use of cis women’s rape trauma in advocating for trans women’s exclusion from women-only space. Engaging rigorously with qualitative data requires us to treat our participants with empathy and respect: but we should also set their views and experiences in context and explore how they are produced and framed.

Ask yourself: ‘so what?’

Once your argument starts coming together, ask yourself: ‘so what?’ How does it shed light on broader economic, social and/or political issues? This isn’t about micro- versus macro design: often in-depth research with very small samples can illuminate wider debates with more insight than much larger studies. The ‘so what?’ test refers to your mindset when you argue from your data. Are you looking for novelty or significance? Are you content to tell a nice story, or do you want to try to influence something to change? Again, your ambitions can be quite small, and it is often more practical to set your sights on something specific or local than to make claims which are too grandiose (which takes us back to the question of what your data actually support).

Write an abstract

When you have a decent emerging argument, try writing an abstract of your dissertation – this will help you to develop a narrative which is focused and makes logical sense. You can also outline chapters and sub-headings using Pat Thomson’s technique for avoiding ‘blocky’ writing: this is a really useful way to tie all the threads together. Then keep your abstract and outline handy as you write up your dissertation, so you can amend them and stay focused as your argument develops. You might even enjoy it – watching a research narrative emerge is exciting! What if you were able to construct a catalogue of police brutality against sex workers in your local area? Or show how a school has negotiated the hostile environment and protected refugee children in its midst? Or expose how media debates around ‘equal pay’ persistently erase the experiences of women of colour? Although your dissertation might not change the world, it might make it just a little bit better, and that’s a fantastic thing to achieve.