A few days back we quoted a piece on cousin marriage (my ellipsis):
Steve Jones, professor of genetics at University College London, agreed that there was a higher risk of defects (in cousin marriage) but drinking or smoking in pregnancy was “as bad if not worse”.
But now it turns out the statistics on smoking in pregnancy are junk:
Smoking in pregnancy is far less damaging to the unborn baby than commonly supposed, detailed analysis suggests…
The study by Emma Tominey, a research assistant at the Centre for Economic Performance at the London School of Economics, throws new light on government efforts to stop women smoking when they become pregnant.
(The study) shows that the worst effects are suffered by women from the poorest backgrounds, because in their case smoking is often combined with other unhealthy activities, such as poor diet and consumption of alcohol.
Middle-class women suffer almost no damaging effects, the analysis suggests, even if they continue to smoke throughout pregnancy.
That would explain why Mrs G’s smoking in pregnancy failed to stunt the growths of our altitudinous offspring.
It’s extraordinarily depressing to find that, until Ms. Tominey, none of the statistical surveys have controlled for the many other variables that might influence a baby’s birth weight.
Controlling is essential in all attempts to establish causality from statistical surveys. That’s because variables are interdependent – a poor mother might, as well as smoking, eat badly, drink to drown her sorrows, suffer physical abuse, and so on.
These factors are called dependent variables, and their effects must be removed before it’s possible to pontificate on the effects of any single behavior, such as smoking.
Ms. Tominey’s study tells us that hasn’t been done until now, and that people relying on those junk results, including Mr. Jones, are fools.