Selection and the effect of prenatal smoking

Abstract

There is a debate about the extent to which the effect of prenatal smoking on infant health outcomes is causal. Poor outcomes could be attributable to mother characteristics, which are correlated with smoking. I examine the importance of selection on the effect of prenatal smoking by using three British cohorts where the mothers' knowledge about the harms of prenatal smoking varied substantially. I find that the effect of smoking on the probability of a low birth weight birth conditional on gestation is slightly more than twice as large in 2000 compared with 1958, implying that selection could explain as much as 50% of the current association between smoking and birth outcomes.

Figure 1. Prenatal Smoking by Cohort and Mother’s Age

Figure 2. Prenatal Smoking by Cohort and Mother’s Education

Note: The bars display the distribution of mothers across the categories and correspond to the left axis. The lines show the percent who smoked by category and correspond to the right axis.

Figure 3. Prenatal Smoking by Cohort and Father’s Social Class

Figure 4. Prenatal Smoking by Cohort and Mother’s Social Class

Note: The bars display the distribution of mothers across the categories and correspond to the left axis. The lines show the percent who smoked by category and correspond to the right axis.