We joined Professor Ganna Pogrebna at the Sydney Behavioural Economics Network for her presentation titled Behavioral Data Science of Diversity and Inclusion in the Workplace.
Focussing on gender and race, the discussion explored how Behavioural Data Science, with the use of natural language processing, was deployed to better understand how we can improve diversity and inclusion both at the organisation and the society level.
Behavioural Data Science is a cross-disciplinary field which helps design modelling approaches to better understand three behavioural strands: (i) human behaviour e.g. consumer decision-making; (ii) algorithmic behaviour e.g. detecting algorithmic bias; and (iii) systems behaviour e.g. designing complex systems to maximize people’s wellbeing.
"It is NOT behavioural analytics or fancy marketing"
The field of Behavioural Science is largely conceptual, a theory is developed and then this theory is evaluated in a pilot study to measure impact. On the other hand Data Science is a methodological field, which revolves around the tools used to measure impact. Behavioural Data Science blends the two approaches, enriching the conceptual frameworks from Behavioural Science, with methods from Data Science, to test hypotheses at scale.
Project 1: Does Race ‘Exist’ in 200 years of Female Leadership
New publication (Not online yet) - Request Prof. Ganna Pogrebna for access
Analysing female leadership speech, this research project evaluated the data of 2,514 public speeches by 757 female leaders across 60 countries. The aim of this research was to understand whether misconceptions of racial minorities are proved at a social level. The methodology involved text analytics from the speeches to evaluate the key issues female leaders are interested in and identify clusters (e.g. Are Black or African American women talking about different issues compared to White women?; Are Asian women talking about different issues than Black or African American women).
Project 2: Leadership Role in Making Diversity and Inclusion a Part of Corporate Social Identity
Read the abstract - Request for preprint
In this project, the research team investigated how a leader can impact organisational culture so that they become diverse and inclusive. Using the Refinitiv Diversity and Inclusion Index ranking the top 100 companies across the world, the paper evaluated what gets a company on the list by analysing the mission statements of the top companies in their annual reports.
#1. Behavioural Data Science is not social marketing - Behavioural Data Science uses frameworks from Behavioural Science and methods from Data Science for large scale testing. Beyond this, you can use these tools to detect deficiencies in Data Science methodologies and modify algorithms by integrating Behavioural Science models with Artificial Intelligence.
#2. Race is an ill defined concept - Across the 17 selected topics of speech, there was no difference in the topics that female leaders were discussing. The research demonstrates that women, irrespective of race, were interested in a wide range of global issues. If there was racial specialisation, we would observe different clustering for the races across these topics. As there was no clustering, this demonstrates that race is an extremely poor predictor of behaviour.
#3. Stop ‘talking’ inclusivity & diversity - Simply talking about diversity and inclusivity on the company mission statements does nothing to predict whether a company remains in the top Diversity & Inclusivity list. Instead, concentrating on egalitarian values, like stating measures to improve the wellbeing for all citizens, customers and the wider community is a better predictor for positive inclusivity & diversity outcomes.
The presentation concluded with Prof Ganna Pogrebna reminding us - “Data does not always have a better idea. It’s how you analyse it and what intelligence you derive from it that is important.”
