This post is a response to a question about how to begin data analysis.

When you were little, I bet you played sorting games. You might have organised pencils into colours, or blocks into various shapes. Later on, you may well have sorted those same blocks into sizes as well as shapes, perhaps even added their material into the mix. You probably also transferred pattern-making into paper and pencil activities.  

Maths teachers understand this kind of pattern-making to be early mathematical thinking.

However, sorting games do more than provide a foundation for number. Sorting and categorising are thought processes that we use to understand the world in general – and in particular, they are the very processes we use when we are analysing  data – sorting and categorising the ‘stuff’ we have generated. 

We look to make patterns and establish commonalities in data by finding connections and associations. And we develop categories for our chosen patterns and groupings   – the equivalent of children’s sorting of red and yellow pencils  – and categories for the overall exercise – the equivalent of the child’s “Look Mum, I’ve done a big pencil sort”. So by the time we are adults we have internalised ways we orient ourselves to pattern-finding/making. 

It’s not unhelpful for we researchers to make these tacit orientations to pattern-making more explicit, so we can see if they are up to the job of data analysis, and to see which strategies are good for what tasks. 

If you want to do just this , you might like to check your orienting pattern-making strategies against these sentence starters – just for starters. (Note – these sentence starters are geared to qual data, but many of them also apply to quant.)

When we are finding commonalities, we generally say things like…

• This is similar to… because
• This is different from… because
• This relates to… because

We often then find ourselves refining our ideas of what our data might allow us to see – and we begin to look for particular things…

• What if…
• This raises the question of….
• What is meant by…
• How is it possible that…
• I wonder why…

And we begin to make some initial inferences from what we see in the data…

• It seems as if…
• I’m guessing that…
• I’d expect to see more of …
• If this… then probably also…

In quants research, we statistically test out these initial ideas and inferences, while in qual data we look for additional confirming or disconfirming material. In both traditions, frequency and/or quantity of the ‘this’ is important for a pattern to hold fast.

And then we begin to generate overarching analytic categories

• This is primarily about…

And evaluative summaries

• The most important things here are…

We often look for associations and categories using our pre-existing scholarly knowledge – in research this is usually from the literatures.

• This is connected with the literature on a … because …
• This also appears in b and there it is about…
• This reminds me of c which…

As we get on with association-making, we generally begin to think about explanations for the patterns we are seeing.

• Why is it like this?

Of course, we may find ourselves going back over the data and our analysis as we think…

• This doesn’t seem to fit. Why is that?
• I am confused about…
• I couldn’t quite grasp …
• I need to check back about…

But. But there are also things we are less likely to do/ask ourselves when we start grappling with data.

Unless we are doing practitioner research, practice research or a professional doctorate, where the use of prior non-scholarly knowledge is important, we usually don’t allow ourselves to dwell on associations such as…

• This happened to me when…
• When I experienced this I…
• The usual response/reason is…
• Before I have…

And unless we are doing ethnography or some kind of research in which our own affective responses are important, then we also don’t often think…

• I can picture…
• This made me feel…
• I now imagine…

Some people, and I am one of them, would suggest that both our prior knowledge and experience and our emotional responses are always involved in pattern making. They/we would argue that it’s better to acknowledge these influences, and see what lurking ideas and feelings are producing our researcher-self pattern-making – recognition that you can ‘see’ your own taken-for-granted ideas.

And a benefit. You can also sometimes find surprising clues about new patterns and new ways to write about them when you look at your experiences and emotions. Ask yourself for instance – what data made me feel sad – and see if any new associations and inferences occur.

Image: Studio Alijn