Introduction
A research project conducted by Silberzahn &
Uhlmann in 2015 invited 29 teams of expert level data analysts to process the same dataset of soccer players and referees and answer
the same question using an odds ratio metric: are darker skin-toned players more likely to receive red cards from referees?
Through different distribution methods and selections of combination of covariates, the 29 teams came up with 20 different odds ratios ranging from 0.89 (more unlikely) to 1 (unlikely) and to 2.9
(more likely).
A flat Sankey diagram on the right illustrates uncertainty in analytical outcome arises from diverging analytical approaches.
The below experimental 3D simulation brings motion to the process and illustrates such uncertainty is to some extent deterministic.
