Sometimes, correlation does imply causation

Project specs

Format

2D

Client

COSTNET

Contact

Ernst Witt

Country

Netherlands

Length

04:30

Summary

You often hear “correlation does not imply causation” in order to warn the listener about spurious relationships that may be observed in everyday life. The number of homicides seems to correlate with ice cream sales, but clearly this is not a causal relationship. Given that all we can see in real life are mere correlations, can we ever be sure of causality? This video will describe a network inspired definition of causality in order to derive a method, known as the PC algorithm, to detect causal interactions.

Researcher Profile

COSTNET is a European COST collaboration of more than 500 statisticians and other quantitative scientists from 34 countries on the development of novel inference methods for network data science. It is lead by Prof. Ernst Wit (USI/RuG) and the core management team involving Gesine Reinert (Oxford), Goeran Kauermann (LMU, Munich), Veronica Vinciotti (Trento), Claire Gormley (UCD, Dublin), Clelia di Serio (UniSR, Milano), Steffen Lauritzen (Copenhagen), Anuska Ferligoj and Vladimir Batagelj (Ljubljana), Arnoldo Frigessi (Oslo).