So how have you controlled for income?

An interesting result from an academic paper:

Researchers found that high levels of exposure to a common type of pollution in pregnancy was \”significantly associated\” with anxiety, depression and attention problems when the children were six or seven.

Academics at Columbia University in New York measured maternal levels of a particular type of pollution caused when petrol or diesel is not burnt properly.

All 253 women they studied, none of whom smoked, were exposed when pregnant to the chemicals, called polycyclic aromatic hydrocarbons (PAHs). Levels varied widely and they found that women exposed to more traffic pollution during pregnancy tended to have children with more behaviour problems.

Sadly I cannot find the paper as the first thing to check is \”how have they controlled for income?\”

Unsuprisingly, low income people tend to live where traffic is heaviest: house prices do change dependent upon whether you\’ve the South Circular or a cul de sac outside your front door.

Low income peeps tend to live where there are worse schools etc as well. There are many things that could link behavioural problems at 6 or 7 years of age with poverty. Things that would be associated with that higher exposure to traffic fumes. But all meaning that unless we\’ve dealt with that income issue, we don\’t actually know whether it is the traffic fumes or not causing the problems.

So, if anyone sees the paper can you let us know how they dealt with it?

11 thoughts on “So how have you controlled for income?”

  1. The paper is here:

    Covariates were retained in the models as potential confounders if they exhibited a relationship (p ?0.1) with motor or mental development, regardless of their association with PAH exposure. The final models included an indicator for PAH exposure, the child’s exact age at test administration, child’s sex, ethnicity, gestational age at birth, quality of the home (caretaking) environment, and prenatal exposure to ETS and CPF measured as described above.

    Tim adds: So they’ve not controlled for income at all then?


  2. The data collected included “socioeconomic information related to income and education”. They say their analysis included all available confounders if at all significant, so the lack of a control for income should be because they tried it and it didn’t matter. The cohort looks fairly homogeneous, so it’s not astonishing that income isn’t significant once “quality of home” has been included (its measurement is discussed here – )

    The p-level for the PAH result is <0.01 . There's no obvious reason not to take this study seriously.

  3. This is a perfect example of what Dierdre McCloskey calls the “cult of statistical significance.” They got a statistically significant result: being in the bottom 25% of air quality reduces the quotient by about 2%, and the quotient itself is relatively strongly related to ability. But to say that result is “important” would be stretching it too far, even before you get into issues with model misspecification.

  4. The Pedant-General

    I’ll be there’s a lot riding on that “quality of the home (caretaking) environment”.

  5. Ceri,

    > P>=0.1??

    I doubt there is a single person in the House of Commons who would have clue as to what this meant. As far as they are concerned it is more policy based evidence making they can use to create more bureaucracy and tax motorists more.

    I remember when a similar campaign ran to get rid of the lead in petrol. If you believed half the claims made for that our inner cities should be full of child geniuses. Hasn’t exactly happened has it.

    This is all part of the death spiral of our civilisation, where science is ignored.

  6. Ian Reid:

    Of course it worked–see how smart you’ve become now that we’ve got rid of the nasties?

  7. Conducting experiments on humans using a control sample to ensure scientific accuracy is fraught with problems.

    Just ask any of the bastards tried for war crimes in 1946.

  8. I would want to see that they have controlled for caffeine intake, some parents give their kids vast quantities of Coca-Cola. Furthermore there is a pretty strong causal link between caffeine, anxiety and hyperactive behaviour. Guess you can’t blame that on car drivers though.

  9. Kind of odd really, but I’ve noticed over the years that women tend to like the smell of gasoline better than men.

    Most of that odor is PAH’s.

    Perhaps women choose to be near this sort of pollution?

    Odd, I know. But worth considering.

Leave a Reply

Your email address will not be published. Required fields are marked *