Pie Chart Slices
How much time should a data analyst spend analysing data?
I was at a Creative Edinburgh Talking Heads event last night. Pixels and Bricks, it was called. It was basically a Pecha Kucha Night for Creatives, and there were ten speakers, who each spoke for six minutes and 40 seconds (that's 20 slides with 20 seconds per slide, for those of you unfamiliar with the Pecha Kucha format).
Some of the talks were a blur of frenetic madcap zaniness; others were altogether more measured. But the highlight of my evening was when freelance photographer Ellie Morag displayed this pie chart on the screen:
Photograph lifted from @CreativeEdin's Twitter stream without permission
Look at the tiny size of the yellow segment. That's the amount of time a photographer spends actually taking photographs. Crazy.
Anyway—naturally enough—I wondered if NHS information analysts would end up with similar pie charts if they drew them for their own work. Would the amount of time they spend doing actual data analysis be as small as Ellie Morag's yellow segment? Or would it be more like the size of the red segment?
Well, without knowing the answer to that question (but nevertheless speaking as a person who spends more time in the company of more NHS information analyst than most people do), here's my hypothesis. I think most NHS data analysts spend most of their time at work analysing data. Their number-crunching time is probably more like the size of the photographer's red 75% segment. That would certainly be true of the way I spent my own time back in the dim and distant days when I was an NHS-employed data analyst.
Now, as a freelancer, of course my pie chart segments have changed in size substantially, Now that I'm freelance, there's a slice that needs to be labelled "Trying to find people who will pay me" that needs to be on there and sometimes it feels as if that slice is about half the size of the circle and sometimes I'm so busy that I don't even have time to think about where the next cheque is coming from, and at times like that, that slice is therefore non-existent.
But there is a deeper point I want to make, and it is this. If we say that actual data analysis is number-crunching time, the time we're hunched over our PC or laptop keyboards, querying databases, doing all sorts of stuff in Excel or Tableau or whatever, should that time comprise 75% or more of our time? Or should we actually be spending more of our time "out there" "in the field", talking to the managers and clinicians in their natural habitat, as it were, finding out how they see their world, what their mental models of their world look like, asking them how they prefer to get their data, asking them if the report we sent them last week hit the spot or whether it just went in one ear and out the other?
I'm currently working on a theory that says data analysts might be better off spending more time than we currently do on a category of work that I'm loosely and provisionally calling "fieldwork". It's a category of work that fits in with what Schaun Wheeler talks about in his piece about Data Science and Anthropology:
I think anthropology's traditional tools—ethnographic field methods—offer the best opportunity to design products that facilitate decisions rather than just informing then.
In other words, if we want to make our data meaningful to the people who make decisions based on that data, then we need to engage in ethnography—fieldwork—if we are to find out how best to frame and present data in ways that help managers and clinicians the most.
We're moving towards a discipline that I find myself calling "Data Anthropology". Here is Schaun Wheeler again (from the same article):
Anthropology (if done right) produces stories clearly connected to on-the-ground concerns and perspectives but fails to demonstrate that those stories more generally hold true over time or across other populations. Data science produces stories that often have the time- and population-level generalizations down, but frequently fails to connect that output to the practical context and constraints of individual decisions.
All of this is a long way from a pie chart that raised a laugh at a Pecha Kucha Night. But this is the stuff that's keeping me awake at the moment. And if any NHS information analysts want to email me with an estimate of the size of the pie chart slice that's labelled "actual data analysis", then please do. You can get me at email@example.com.