Using tech and data science to make sense of election results

Maryam Ahmed

March 20, 2019, 9:30 a.m.
Roger Ockrent Room (Chateau)

Maryam, a Data Scientist at BBC News, gives a hands-on demonstration of the use of data science to add context to election results. Share her experiences of the excitement of election night analysis in the newsroom, and learn of the pitfalls when producing analyses for a mass audience.

Public scrutiny and understanding of election results are both crucial in a transparent democracy. Metrics such as vote share, swing and turnout are reliable indicators of a party’s success at the ballot box, but do not explain which demographic groups may be driving large swings or unexpected results.

Traditional methods for adding context to election results have proved inaccurate in recent electoral cycles. Analyses provided by pundits are not necessarily grounded in evidence and may be based solely on intuition or experience.

Similarly, exit polls and surveys conducted after the 2016 US Presidential Election produced varying results due to methodological flaws. Data science and machine learning techniques have been used by outlets including BBC News, the New York Times and the Economist to analyse the demographic factors underlying election results.

A recent analysis commissioned by BBC News disproved the popular notion of a surge in young voter turnout for Labour party leader Jeremy Corbyn during the UK 2017 General Election, demonstrating the potential for statistical techniques to cut through noise and hype surrounding election campaigns and results.

Within the Civic Technology space these methods have great potential to help researchers, community organisers and the public to understand the forces behind confusing or surprising election results.

Learn how to use and interpret methods including regression analysis and random forests to make sense of election results. No coding knowledge is needed, but participants will benefit from a good grasp of high school mathematics and bringing their own laptops.