This investigation of course suffers some limitations. Although broadly representative of the population of Minnesota when our participants were born, the sample is ethnically relatively homogeneous; results may not readily generalize to other populations. Our assessment of drinking during pregnancy consisted of retrospective self-report, which is obviously vulnerable to recall and social desirability effects. To minimize the latter, our interview introduces these questions by acknowledging that drinking during pregnancy was once relatively common. Reports of the number of drinks consumed on one occasion ranged as high as 15 drinks, fully 40% reported any drinking, and 10% reported weekly use, suggesting that participants were not overly minimizing their use. In addition, our other work with this measure supports its construct validity (Disney, Iacono, McGue, Tully, & Legrand, 2008). Moreover, maximum consumption during pregnancy was associated with the majority of mental health outcomes we investigated in the younger cohort. It seems unlikely therefore that the failure of these measures to account for our findings is due to insensitivity of the measures.
