How a statistical paradox helps to get to the root of bias in college admissions

Simpson’s paradox is a statistical phenomenon in which a trend appears in small data sets, but differs or reverses when those sets are combined into a larger group. One of the most fascinating examples of the paradox comes from a study about gender bias in graduate admissions at the University of California, Berkeley in 1973, when roughly 44 per cent of male applicants were accepted, compared with only 35 per cent of female applicants. These figures appeared to show an obvious bias against women, but when the data were broken down by department, they actually showed a slight bias in favour of women. This animation from MinutePhysics explains just how Simpson’s paradox occurs and, in the case of Berkeley, how the paradox highlighted a deeper societal bias that pushes women towards departments that are more crowded, have less funding, and offer poorer employment opportunities. 

Video by MinutePhysics

Support Aeon

‘I am a friend of Aeon because I value freedom ... freely provided, intelligently presented information liberates us all.’

Roland M, USA, Friend of Aeon

Aeon is a registered charity committed to the spread of knowledge and a cosmopolitan worldview.

But we can’t do it without you.

Support Aeon

Essay/Physics

Going nowhere fast

After the success of the Standard Model, experiments have stopped answering to grand theories. Is particle physics in crisis?

Ben Allanach

Idea/History of Science

Network visualisations show what we can and what we may know

Christopher Warren

Video/Evolution

Take a shimmering, surreal swim with what might be the Earth’s oldest animals

7 minutes

Idea/Biology

Animal pain is about communication, not just feeling

Mirjam Guesgen

Essay/Space Exploration

What if ET is an AI?

After centuries searching for extraterrestrial life, we might find that first contact is not with organic creatures at all

Caleb Scharf

Video/Ecology & Environmental Sciences

How to maintain infrastructure – the stunning collective intelligence of ant engineers

1 minutes