Korva is a scroll of the internet’s best ideas.
Five of six blue eyes isn't odd
The Blue-Eyed Crew
A scaffolding crew in the north of England noticed that five of its six bricklayers had blue eyes, and one of them claimed the pattern had held across a 15-year career on sites in the region. Blue eyes are rare worldwide, where 70 to 80% of people have brown eyes and only about 10% have blue. In the UK the figure is roughly 48%, rising to about 50% further north.
Running the Test
A formal test starts by assuming nothing suspicious links blue eyes and bricklaying, then asks how likely this result or a more extreme one would be if the background rate is 50%. Exactly five blue-eyed bricklayers out of six has a chance of about 9.4%, and all six has a chance of about 1.5%. Together that is about 10.9%.
Why It Proves Little
That is not low enough to count as statistically significant. Visiting 10 different construction sites would be expected to produce a result like this at least once. Confirmation bias may also be at work: someone who notices the pattern once will notice it again only when it appears strongly, and will remember the blue-eyed workers selectively.
Settling it for real would take a sample of at least 400 northern bricklayers.
Why is five percent the usual significance cutoff?
Statistician Ronald Fisher suggested the 5% level in his 1925 book Statistical Methods for Research Workers as a convenient rule of thumb. The number stuck because it was easy to apply with the tables available at the time. Many fields still use it. Particle physics, by contrast, requires results at roughly five standard deviations, about one chance in 3.5 million, before announcing a discovery.
Why do many comparisons create false patterns?
Each separate check carries its own chance of a fluke, and those chances add up. If someone runs 20 independent tests at the 5% level, the probability that at least one shows a fluke result is about 64%. This is called the multiple comparisons problem. Researchers who test many hypotheses often tighten the threshold, for example with the Bonferroni correction, which divides the cutoff by the number of tests.
Why do blue eyes look blue without blue pigment?
Eyes contain no blue pigment at all. Iris color comes from melanin, and blue eyes have very little of it in the front layer of the iris. Light entering that layer scatters, and short blue wavelengths scatter back toward the viewer, much as they do in the sky. A 2008 study by Hans Eiberg and colleagues at the University of Copenhagen traced most blue eyes to a single mutation near the OCA2 gene that arose roughly 6,000 to 10,000 years ago.
More from The Rest Is Science
