Blog

HRW in 2022 – Uniting Data Science and Behavioural Science

26.04.2022

Our Quant philosophy at HRW is grounded in two areas – data science and behavioural science. We believe that combining both generates the most powerful and actionable insights, and this year we’ll be spotlighting three ways we’re taking key methodologies to the next level, bringing this philosophy to life.

Behavioural science can elevate analytics at many stages – but perhaps most importantly when it comes to how we capture our data. Poor sampling, badly written questions, and uncleaned data leads to the ‘garbage in, garbage out’ phenomenon, where any real trends are hidden by the noise of the poor inputs, and can even generate false findings.

For instance, you may remember in 2016 that Microsoft released an AI chatbot (called ‘Tay’) onto Twitter with the aim of learning and improving from the discussions it had with users. However, within 24 hours, Microsoft had to intervene due to internet denizens ‘teaching’ the chatbot to tweet offensive messages including using racist language and supporting genocide. Ultimately, (and unfortunately) this probably says more about certain groups of Twitter users than is does about the algorithm but demonstrates that getting the correct input is crucial!

When it comes to quantitative research, as well as talking to the right people, we  need to be asking the right questions as a foundation for our analytics. This year, we’re looking into how we can create better quant inputs by including questions that also tap into non-rational decision making.

Segmentation, as the founding methodology of HRW, is one area that’s particularly close to our heart, and a segmentation is most powerful if we include questions that truly tease apart the underlying drivers and motivations of physicians. In the past, underlying non-rational drivers were mostly left untouched as researchers focused on more ‘standard’ perceptions. However, by including bias-identifying questions in our surveys, we have developed an approach to segment on the basis of which behavioural biases are having most influence on a physician’s goals and prescribing. Needless to say, this approach (developed by our HRW Shift behavioural science team) can generate extremely actionable segments.

INTER:COM, our communications testing approach, is another area that we’re continuously developing, allowing us to better measure which ideas truly resonate with physicians (rather than simply being ‘liked’). Our new approach, currently in self-funded testing, will provide us with key metrics that have behavioural science at their heart – ensuring we can better determine which areas need improvement and how to do so.

And later this year we’ll be highlighting how a successful demand research approach relies on the acknowledgement of key behavioural biases which, if not mitigated, result in significant overclaim in use of new products.

We look forward to sharing more details on how uniting data science and behavioural science can enhance our research over the coming year!

 

By Richard Hutchings

< BACK TO BLOG