Written with Jonathan Adamson @ordinaryjon who introduced me to the concept of linking transactional analysis and data viz.
So you know data viz and what charts work, and what charts don’t. And you definitely know Tableau. But this may not be enough to produce a successful viz. Why? Because to get there you have to deal with the sometimes tricky relationship between you and the potential user of your resulting viz?
To produce a successful visualisation, there is a general agreement in the literature that you need a successful collaboration between the “viz experts” and the “domain experts”.
In essence, you need to start a conversation. That can be difficult when strangers from opposing areas of the business, or from a completely different business, try and work together for the first time. Dealing with this relationship seems to be somewhat neglected in the viz literature. Reduced to abstracting the issue down to a solvable problem or a linear series of transactions.
Yet it can take a bit of time and skill to break down the barriers. For the “viz expert” to show they have something to offer, that they are not just showing off with Tableau and are starting to understand the specific domain issues.
A useful way of thinking this through is to use Transactional Analysis, developed by Eric Berne in the 50 & the 60s. I talked about this approach in my recent presentation at the London Tableau Conference, “More than Pixels on a Screen” and it was the part of the presentation that I got the most questions about afterwards.
There is an apocryphal tale of Berne talking to a fellow passenger on a plane ride in the 60s who, when Berne said he practised in Transactional Analysis, responded,
“oh yes, I know all about that; parent-adult-child, that’s it isn’t it?”
When Berne asked what his fellow passenger did for a living he responded that he was an astronomer, to which Berne said,
“oh, I know all about that too: twinkle, twinkle little star.”
So, acknowledging the over-simplification here, a very, very basic overview of Transactional Analysis (TA) analyses human behaviour and communication around the transactions between people, usually an ‘agent’ (in this case the “data viz” expert), and a ‘respondent’ (the client).
It uses three states (or ‘ego’s) which are parent, adult and child. In terms of data viz we could think of this in the following ways:
Data viz expert (parent ego)
I’m the analyst, I’m the expert with the data, I know the best way to present it.
Client (child ego)
I don’t understand maths, I’m not an analyst, I don’t know what to do, you have to tell me
Alternatively, it could also go the other way:
Data viz expert (child ego)
I don’t know anything about…(social care, highways etc), I can’t do anything unless you know what you want.
Client (parent ego)
What do you know about social care/highways etc, you can’t tell me anything I don’t already know.
Neither of these scenarios has a very high likelihood of producing successful data viz.
A Transactional Analysis approach to data viz might suggest that the scenario most likely to result in successful data viz is in the adult-to-adult states, and that’s what you should aim for.
Data viz expert (adult)
I have skills around analysis and data viz and I want to work with you to understand you/your business better.
Client (adult ego)
I understand my business really well, especially some of the problems and I want to draw on your expertise as an analyst/data viz expert to help improve my business.
I’ve tried to find a way of understanding why some of my data viz is more successful than others. And one of the common factors in successful data viz work is that an adult-to-adult relationship exists between analyst and client. So how does that feel different?
We analysts start to learn about the business, and users start understanding and suggesting viz. We respect each other’s roles. We start overlapping as the roles become less defined and the knowledge transaction is much greater. And that’s when we have the biggest chance of success.
If an adult-to-adult relationship leads to better data viz, then how do we create the environment which makes it more likely to develop? What skills do you need to develop an adult-to-adult relationship?
In the scenarios we looked at earlier, the agent is the data viz expert and the client is the respondent. Let’s not worry too much about what the respondent (client) does as we can’t control that. That’s something for another article.
So how does the data viz expert avoid adopting a parent role?
Important characteristics for any data viz expert are humility and self-awareness. If you know what you are good at, then it can be easy to say what you aren’t good at and what you also know nothing about. People who find that difficult either lack ability or confidence, or both.
Filmmaker and criminologist Roger Graef was asked what he learnt from studying at Harvard,
“What Harvard has given me…is the confidence to say, ‘I don’t know, explain it to me’. I know how to ask the question without feeling small.”
Of course, we don’t all enjoy a Harvard education. But we can all allow ourselves to show a little vulnerability in our knowledge. As well as cede some ground to our respondent (client).
Just because you know a bit about data viz, it does not mean you know the best way to present someone else’s data.
How to avoid taking on a child–like role?
It is obvious you can’t be an ‘expert of everything’. You can’t be an expert on social care and climate change and sports analytics. Or the overabundance of other data sets which might enjoy better visualisation. Yet, you can help shed new light on those areas, by showing an interest to learn more. In fact to make real progress, you have to understand the business almost as well as the “domain experts” do.
Think of the adult-to-adult relationship like a good conversation.
It works best when you are self-aware and not self-absorbed. You have something interesting to say. You also think other people have something interesting to say too. You both benefit from an adult-to-adult conversation, leaving your egos at the door, and producing great data viz.