Communicating survey results fast and slow

The way market researchers have traditionally presented survey data in slide after slide of complex tables and charts can best be described as mind-numbing cruelty. The usual charts we see around are nowhere near as interesting as this one.  It is cruel to clients to impose dull complex charts on them. It is also a pretty ineffective way for our industry to communicate.  In this post, I offer some suggestions based on Fast and Slow Thinking.


Recently, we have been moving towards a more qualitative approach to presenting survey data, using diagrams rather than charts and tables where we can. However, there is always more to learn and experiment with, so I was delighted that one the papers presented at the Esomar Congress in Dublin a few weeks ago addressed just this issue. It was called Exploring the use of visuals in the delivery of research dataand it was by Adam Frost, Tobias Stuart and Jon Puleston. They conducted ‘around 30 different exploratory experiments into the visual representation and communication of data’, using a whole host of different charts and infographics. Their paper described the kinds of charts and infographics which lead to the best recall or encourage curiosity to read more.

How ‘thinking fast and slow’ helps

In his book Thinking Fast and Slow, Daniel Kahneman used the terms System 1 and System 2 to describe how our brains automatically make decisions quickly and emotionally, although we all have the ability to think slowly and in a more controlled way when the circumstances are right. System 1 is instinctive and emotional and prone to bias, while System 2 is more cautious and analytical.  System I is similar to the recall measures used in the paper mentioned above, while ‘curiosity to read more' is similar to System 2.

Five ‘ideal outcomes’ for readers (or audience members) when they see survey data

Building on this thinking, we can use Kahneman's ideas to work out how best to communicate survey results from a reader-centric perspective. We have identified 5 ‘ideal outcomes’ for readers (or audience members) when they see survey data.  

Get the gist. When the message is simple or familiar, readers will understand the gist of the message in an instant and move on.

Think when they need to. When the findings are new, complex or particularly important to them, readers will spend time looking and reading and thinking about the implications. 

No skipping. Readers won’t skip over important pieces of information.

No oversimplification. Readers won’t become victims to their cognitive biases and consequently mis-interpret complex and important data.

No over-thinking.Readers won’t spend time poring over a simple chart and over-thinking about what it means.

The reader-centric approach assumes that what may be simple and familiar to one person could be complex and new to someone else. 

Two ways Kahneman can help

System 2 has to be triggered. Kahnemen’s model is known as a ‘sequential dual processing model’ *, in that what he calls “lazy” System 2 (the slow one) doesn’t become activated, if it gets activated at all, until System 1 is underway.  Sometimes we need System 2 because System 1 can be “gullible" and prone to biases such as confirmation bias, where people see in the data what they want to see. The implication from Kahnemen’s work is that something needs to trigger System 2 to wake up, though there has been very little work done on what those triggers might be. (And remember, if people only see what they want to see in our data, that is our fault!)

People don’t start what they can’t finish. Kahneman says that "we are biased against actions that lead to regret" and will actively avoid losses. The important point for us is that loss includes lost or wasted time. In other words, to avoid the regret of wasted time our readers will avoid charts that look dull or difficult. We don't start what we can't finish.

The message is: choose your charts with System 1 and 2 in mind

For System 1, we should use bar charts in rank order; or columns not ranked-ordered; and if we use illustrations we should use clichéd ones that are easy to grasp in an instant (as suggested by Frost, Stuart and Puleston)

For System 2, novelty charts used sparingly attract more thoughtful attention, while visualisations on infographics also made people more curious to read more. We need to entertain, not just dump data.

So…if we want people to think "this will be easy; I will have a quick look at it" we need to present our data in a way that looks familiar. To some extent this is an argument for chart templates despite how mind-numbing endless charts all looking the same can be. However, ‘looking easy’ only gets System 1 activated. If you want to wake up System 2 – that is get people to think – then we also need to add something unfamiliar. 

A technical note: Kahneman's model is a type of dual processing model of which there are several. Kahneman's is different a) because he popularised it and b) because his is a sequential model in which System 2 can never start until System 1 has already kicked in. Other eminent psychologists disagree with this and say that System 1 and System are intertwined - but this is a topic for a later post.


We are good at this kind of thing, because we think.  Contact Sue.

Tags: Behavioural Economics, Communicating results