Home   About Kurtosis   Courses   Course Calendar   Booking Information   Ideas   Technique   Clients   Contact
     

Tables, charts, paragraphs and narratives.

We need to be both Type-A analysts and Type-B analysts

(4 comments)

I read Martin Amis’s novel The Information when it came out in paperback in 1996. I was a fan of Amis’s style. I’d already read The Rachel Papers, Money and London Fields and I liked the way he wrote. Amis wrote—still writes—great sentences and great paragraphs. But if I was ever in need of great chapters and great plots, I knew I had to look elsewhere. And I wasn’t the only one who thought that. When Iain Sinclair reviewed it in the Independent, he said: “The Information reads like 500 pages of smart, highly finished extracts. It doesn't add up.”

I used to think that being an information analyst was about producing “smart, highly finished extracts”. (Well, that’s one way of describing well-designed tables and charts.) And I used to think that well-designed tables and charts were the end-product of what data analysts did. Number-crunchers crunch numbers, and numbers, once crunched, usually look like tables or charts. So yes, it made sense that they were our end-product.

Moreover, I used to think that well-designed tables and charts were also what managers and clinicians wanted whenever they said they wanted “data”. Therefore if your objective was to communicate numerical information effectively, the trick for analysts was to make sure your tables and charts were well-designed. Problem solved.

But I was wrong. I no longer think that tables and charts—however well-designed— are our end-product. Managers don’t want tables and charts; they want explanations. They don’t want data; they want narratives.

You can put the best-designed data in front of NHS managers but most of them still won’t get it. Sometimes that’s because the data’s still too complicated; sometimes it’s because the managers are just too busy or too distracted to pay proper attention to it. But most of the time it’s because it’s not data they want; it’s a story.

So what made me change my mind?

Here are three strands.

I was re-reading Edward Tufte on data/text integration the other day. (Yes, you don’t need to tell me, I know that’s a hideously pretentious sentence). In the last chapter of The Visual Display of Quantitative Information, he has a section called Making Complexity Accessible: Combining Words, Numbers, and Pictures. He says:

“Words and pictures belong together. Viewers need the help that words can provide.”

And he tells us that we should think about data graphics (“data graphics” is the phrase Tufte uses for what the rest of us call tables and charts) being like paragraphs that need to be positioned correctly when they are part of a bigger piece of work. The key thing is that the well-crafted paragraph is part of a bigger picture. The bigger picture is the narrative. The story.

Here's the second strand. Another essential text on the data analyst’s reading list is Andrew Ehrenberg’s 1981 article The Problem of Numeracy. On the face of it, The Problem of Numeracy is little more than a clarion call for better designed tables and charts. Ehrenberg provides us with rules—over-prescriptive rules, as it turns out—about the micro-detail of designing tables. We need to round to two significant figures. We should present data in hierarchical order. We should compare numbers across columns rather than down rows. And so on. But it’s worth reminding ourselves that the sixth of Ehrenberg’s rules is: “Provide a narrative”. Here is what he says towards the end of his article:

“Few educated people lack the skills to read well-presented tables of data, particularly when they are also given a brief verbal summary…” (my italics)

So even Ehrenberg, the man who you thought just had things to say about table design, had things to say about the need for a narrative.

Another reason why I’ve changed my mind about what constitutes a data analyst’s end-product is to do with the way I’ve been thinking lately about what we mean by the phrase “gut feel”. Analysts hate it when they hear managers talking about “gut feel”. One of the most common complaints I hear from data analysts in the NHS is that they believe they’re often supplying data to managers who’ve already made up their mind about what they want to see. And if our data doesn’t support their “gut feel” then it gets discarded. I never miss an opportunity to let people know that my favourite cartoon is the one with the caption: “That’s the gist of what I want to say. Now get me some statistics to base it on.”

What managers mean when they say they have a “gut feel” about an issue or a problem isn’t something that can be illustrated with just one isolated table or chart. No, what they mean by “gut feel” is a narrative. This thing happened. Then that thing happened. Then, because of those two things, a third thing happened. A story. Usually a story with some cause-and-effect mechanism running right down the middle of it.

As data analysts, we have to get in amongst these narratives. We shouldn’t just stand passively on the sidelines, listening to other people’s narratives. If we are to supply the data evidence that supports or refutes them, then we, too, need to be involved in generating narratives.

But we can’t do that for as long as we view tables and charts as our end-product.

It’s a bit like novelists viewing sentences or paragraphs as their end product, when actually their end product is a story. Well-crafted sentences and paragraphs are obviously essential elements of that story; but on their own they don’t amount to much.

Well, for analysts, numbers are like sentences. And tables and charts are like paragraphs. But we need to go beyond paragraphs into narrative, plot, explanation and story.

For Martin Amis, it’s all about the sentence. The story takes second place. Here he is in an interview published in 1998:

“Anthony Burgess said there are two kinds of writers, A-writers and B-writers. A-writers are storytellers, B-writers are users of language. And I tend to be grouped in the Bs.”

A writer of Amis’s calibre can afford to be categorised as a B-writer. But analysts don’t have the luxury of choosing one or the other. Analysts need to be both A-analysts (storytellers) and B-analysts (users of numbers). Tables and charts on their own don’t change the world. In fact they barely even influence it slightly. What changes the world is narratives. Cogent explanations of why things happen, and why they happen sometimes but not all of the time, and why they happen in all sorts of different sequences and why some things are necessary for certain processes to happen and why some things are not necessary.

As analysts, we need to start thinking beyond numbers, tables and charts. We are like writers for whom words, sentences and paragraphs are not enough. We need to turn numbers into chapters and stories.

[14 January 2014]

 

Make a comment on this blogpost:

Name:

Job Title:

Organisation:

Email address:

 

© Kurtosis 2014. All Rights Reserved.
     

Comments on this article

14 January 2014:

As usual very thought provoking. And the thought it provokes for me on this train to Dundee is how do we get clinicians, managers and social workers to turn the data into their own story? How can an analyst without the subject matter expertise lay out the data in such a way that the managers etc can interact with it, annotate it and own it? Sometimes the story has to come from the data surely rather than data be selected to support it?

Alastair Philp

Information Consultant, NHS National Services Scotland: Public Health & Intelligence

14 January 2014:

Might've guessed you were a freakin' Amis fan!

Ruth Yates

Head of Business Intelligence, Staffordshire CSU

14 January 2014:

I agree wholeheartedly, and in response to Alastair's comment would argue that by being integral to a team, as opposed to a group of geeks in the basement (or Portakabin), analysts can co-produce the narrative with their clinical and managerial colleagues.

Kate Cheema

Specialist Analyst, Quality Observatory

14 January 2014:

Selecting Data: Dirk Gently would say that everything is connected to everything else - which may be very interesting but may not be very useful. Similarly in A&E we can count a thousand things every day but only some would be useful. We may choose to count the number of patients walking through the door to show the effect on the 4 hour target, rather than % of patients with blue eyes.

Ian Binns

Head of Information, East Lancashire Hospitals NHS Trust