Monday, December 13, 2010

The height of intelligence

The positive correlation between height and intelligence has been documented many times. It is attributed generally to deficits in early life nutrition. According to some this explains much of the height/earnings premium. So is there any evidence on this correlation for Ireland? Using the Growing Up in Ireland data I graph the students mathematics score against their height (adjusting for sex, SES and birthweight) along with a 95% confidence interval.
So yes there is a slight upward gradient but it flattens out above about 1.25 metres, possibly even dipping at the very top of the range, the raw correlation is .114. But on the whole it looks fairly flat and shorter people who want to be mathematicians should not be discouraged.


Peter Carney said...


Did you check the raw estimate for height and intelligence?

Be interesting to see the coefficients on SES and bithweight here?

Kevin Denny said...

Ok..the height effect seems to be:
an extra cm is worth 0.375 of a point up to about 1.36m and nothing thereafter.
An extra kilogram in birth weight is worth about 2.748 points up to about 3.6kg and is slightly negative thereafter.
SES is a set of dummies, medical card status, sex, household income in there too.
Of course there are just rough estimates with limited controls, no school stuff for example.
But being small does hurt it seems.

Liam Delaney said...

These are interesting results. One interpretation is that there is not a sufficient variation in SES-related factors that would influence stature among this generation of Irish children.

Kevin Denny said...

Well Liam there is a bit of variation:using mother's education going from bottom (mother with lower secondary education or less) to mother being a graduate is "worth" about 1.5cm.

Jason Loughrey said...

I have come across some similar research relating to my own work.

Paxon and Case in 'Stature and Status: Height, Ability, and Labor
Market Outcomes' make the point that advantaged children are not only taller on average but also experience earlier adolescent growth spurts. They use the term 'peak height velocity' and this is found to peak at age 12 for girls and 14 for boys. It is probably something to look out for in the next few waves.

Behrman and Rosenweig 2004 points out that birthweight divided by gestation rather than birthweight is a better measure of the healthiness of a child which might be worth a try as an independent variable.

Jason Loughrey said...


Best to leave out the comment on gestation period!
I see that the variable in the questionnaaire is categorical in any case.

I think that the first part of the comment is ok.