Galton trivia


Jason Collins


June 24, 2011

Every time a new Francis Galton piece is published, I look forward to the Galton trivia. This time it is in an article by Steve Jones (HT: John Hawks):

He made statistical inquiries into the efficacy of prayer - he got into trouble for that for he found that those people frequently prayed for, like monarchs, lived no longer than anyone else.

He even made a beauty map of Britain, based on a secret grading of the local women on a scale from attractive to repulsive (the low point was in Aberdeen).

Jones uses Galton’s findings on height to lead into his argument that genetics has run into a problem. Geneticists are not finding specific genes for traits that we know are highly heritable.

Take adult diabetes, now a major health problem, and one that certainly runs strongly in families. Genome scans reveal scores of different bits of chromosome as possible culprits but together they explain just one part in 20 of the overall inherited liability to the disease.

The chance of being born with a predisposition to a common illness such as diabetes or depression is a gamble with huge numbers of cards.

So many small cards can be shuffled that everyone who falls ill fails in their own fashion and no gene says very much about whether or not you will get the illness (although the number of cheeseburgers you eat certainly does).

While no gene by itself is likely to say much about your chance of getting adult diabetes, Jones’s comparison between a single gene and cheeseburgers is not meaningful. We are not composed of a single gene - we have a genome. Despite genetic studies not finding the “missing heritability”, that heritability still exists. A more useful comparison might be whether family history is a better predictor than the number of cheeseburgers eaten. Or whether your twin has adult onset diabetes.

The other interesting thing about examples like this is that the environmental factor would have a genetic component. A host of heritable traits are likely to affect whether one eats a lot of cheeseburgers. This is one of the reasons why it is not easy to link gene X to a specific outcome.