Behavioural Science and Data in the Real World: A UCL BIS Seminar
Imagine yourself as a busy executive, working in a sector like finance. These days, behavioural science is a buzzword from which there is no escape, and it seems to be the solution to most problems faced by modern businesses. However, for it to work, you need reassurance that the approach is designed to meet your needs. For this you need data. Eoin Campbell and Azim Arsiwala work at Cowry Consulting, Azim specialising in data science-led behavioural interventions and Eoin in experimental design and business solutions. Cowry work with all sorts of large companies to promote the use of data-driven behavioural science, with the aim of helping companies make better and more ethical decisions, they are making academic research usable for humans and by extension businesses.
“Cowry work with all sorts of large companies to promote the use of data-driven behavioural science, with the aim of helping companies make better and more ethical decisions”
On Thursday 25th February, Azim and Eoin were kind enough to give us a seminar about how behavioural science is used in the private sector, how it interacts with data science, and how this is done in practice.
They began by elaborating on how Cowry use a lot of System 1 vs System 2 evidence (what they term the Homer and the Spock brain) when engaging with businesses. Building on this, Cowry have made behavioural science more accessible to businesses by summarising over 150 heuristics and biases research into their top 14“c-factors”. This is something they believe is more digestible for clients, therefore increasing the use of behavioural science in the private sector. Private sector workers are often busy, so having a simple referable toolkit is likely to achieve this. Eoin presented evidence that shows developing personalised pension videos can help people engage more with their savings. While this is impressive, introducing an algorithm is postulated to improve results even further, by adding and extra layer of personalisation.
An example of how Cowry have done this is through their ‘behavioural fingerprint’, which integrate human learning and machine learning. This helps show ‘what works in behavioural science and business’, in the real world from their previous case studies. The behavioural fingerprint analyses what biases and heuristics are most present in a sector, across different companies. Importantly, this data can be cluster-analysed to determine whether biases applied to create behavioural solutions are clustered together, i.e. are certain biases always used in tandem to create behavioural solutions? It also helps to determine what are the most common biases in certain sectors, which helps similar companies impactfully apply behavioural science. Evidently, data science can help to quantify behavioural science problems.
“Cowry have done this is through their ‘behavioural fingerprint’, which integrates human learning and machine learning”
In the future, Cowry are planning on adapting their algorithm to identify biases and heuristics in specific contexts, e.g., letter or emails, thereby making their algorithm more and more usable to businesses. Moreover, they want to build an algorithm that uses behavioural science to predict business success, e.g., what biases work where? Azim points out that ‘the potential for fingerprint right now is limited only by the data we can collect’, meaning that as data science improves, so can behavioural science.
Another area of interaction outlined was sentiment analysis. Sentiment analysis is commonly used but it’s rarely used in conjunction with behavioural science. Azim explained that employing behavioural science in sentiment analysis modelling would lead to better identification of words that predict sentiment, which can then be used to create better dictionaries. Finally, behavioural science can then be incorporated within the sentiment algorithm to better predict sentiment. For example, Integrating loss aversion within a sentiment model.
Cowry are in the process of building lots of data science tools driven by their expertise in behavioural science, to support specific business needs, something which has had positive feedback across industries. Cowry’s approach to bringing these two fields together is carefully calculated, ethical, and ambitious, and we are very excited to see the future progression of the interaction of these two fields. Thank you to Azim and Eoin for a riveting talk!
If you want to read more about Cowry’s work, check out their website here: https://www.cowryconsulting.com
Stay tuned for more exciting events from the UCL Behavioural Innovations Society!
Twitter: @UCLBIT
FB: UCL Behavioural Innovations Team
LinkedIn: UCL Behavioural Innovations Team
Instagram: @ucl.bit
