Intuitive measurement
Information processing is at the heart of much behavioural science yet, until now, measurement tools have not kept up
At the heart of much of the discussion about behavioural science is the way in which we ‘process’ information. There is much focus on the way we may do this more automatically or more deliberatively, famously captured in Daniel Kahneman’s ‘System 1’ for ‘thinking fast’ and ‘System 2’ for ‘thinking slow’.
In many ways this is uncontroversial: if we were not able to operate in a more automatic fashion then we would find getting through the day hard work as we constantly deliberate over minor details. It makes much more sense to operate in a more routinized way once we have established ‘what works’.
Of course, this simple insight gets a little reified, or over-stated, as people then consider ‘automatic’ means ‘not conscious’ rather than operating more intuitively: For example, while I may be able to do a weekly shop quickly without too much thought, I have yet to come back from the shops to be surprised to see what I have purchased ‘non-consciously’! Further, this is not a binary distinction, but instead of spectrum with the terms intuitive and deliberate reflecting the different ends of this spectrum. For further discussion of the nature of these processes, see the Ipsos Dynamic Decision Making Model.
Measurement of processing
It is well known that we use a range of heuristics and biases to facilitate more intuitive processing: more on this in a later post. But for now, we want to examine how we can measure the degree to which we operate more intuitively or more deliberatively.
Why might we be interested in this? Well, if someone is operating in a more intuitive fashion as regards their current choices then they are likely to be less alert to other possibilities. If we have developed a certain choice for a specific type of chocolate cake that we like to eat, which happens to be a full-fat, unhealthy option – why would we deliberate on competitor options? But if we are operating a little less intuitively, let’s say that we have tried a specific brand of cake but we didn’t find it fully satisfying, we may want to scan the shelves, therefore, being a little more deliberative about our choice and as such open to other possibilities.
The same may apply to other possible options. There may be a cake which is low fat healthy option which is on the shelves. As we intuitively associate chocolate cake with sugar, cream, and not with healthy options, then we may intuitively dismiss it as an choice.
With this in mind, we can see the way in which ‘processing’ can help us to understand our readiness to change from our current behaviours and choices and which alternatives we may be willing to consider.
Tools we use
How do we measure where we sit on the intuitive – deliberative spectrum? One way we can capture this is through response time measurement of course – the rationale being the faster our response, the more intuitive it is. While useful, this does not tell us very much about the nature of our intuitive response or, importantly, how to disrupt it. To this end, we have a scale that we use to measure this consisting of four items:
· Confidence in the choice
· Ease of recalling reasons for the choice
· Feeling of ‘rightness’ in the choice
· Curiosity about alternative choices
These simple questions, applied to current choices and (with slightly different wording) alternative choices, allow us to understand the readiness to change but also guidance on how to ‘disrupt’ more intuitive processing. So, for example, if people are not curious about alternative choices then the guidance is to pique their curiosity with samples or highlight interesting features such as new flavours. We calculate a change veracity score which references the difference between the degree of intuitive-deliberative processing on their current choice compared with alternatives.
What we do with it
Having this measurement allows us to understand what type of action we can take:
If we want to reinforce the more automatic nature of people’s current choices then we can consider ways to use heuristics and biases to facilitate this through the design of comms, for example
They can also be used to facilitate incremental shifts (e.g. a chocolate ginger cake option)
But to enact a more substantial behaviour change (e.g. buying the healthy chocolate cake option) then we need to identify ways to disrupt processing slowing down and engaging more deliberative thinking. Of course, the question is how to do this.
Understanding the dimensions that are important in the decision making of chocolate cake buying is important. Maybe Emotion is important (feeling happy) or Cultural Values (treating myself is legitimate reward), or Outcome Expectations (I know it will taste good). The point is here that while Processing is often presented as a separate, compartmentalised function, the reality is that it is deeply integrated with all the other mechanisms that underpin behaviour.
Beyond nudges
Identifying these dimensions and then using them to create interventions is needed for these more significant shifts which will not be gained by simple ‘nudges’ which work through more intuitive processing. A nudge may encourage someone from a chocolate option to a chocolate-ginger flavour by tapping into intuitive influences, but these are unlikely to work for the healthy option, where a fundamental disruption is likely to be needed.
Measuring the impact of an intervention on our processing
Finally, this measure is useful when seeking to measure the impact of an intervention on our processing: when we design an RCT to measure the effect of an intervention on behaviours, we may look at the behaviour itself (ideally) but we are often reliant on measure of intention (which we covered here). Having this as an additional measure, (the degree to which the intervention disrupted processing), offers an additional insight into the degree to support the evaluation of the intervention’s effectiveness.
Conclusion
Processing is often presented as the mechanism at the heart of behavioural science yet there is very little that is used for effective measurement outside of response times. A survey based approach means that we can measure it across a wide range of contexts. This offers multifaceted measurement of the processing that unfolds, allowing us to explore it in a more nuanced way that moves beyond the typical, simple binary distinction.