Is It the Smoke or the Family?

My latest paper asks whether a mother’s smoking during pregnancy is related to how her children see the federal government as adults, and if so, why.

The Data

The National Longitudinal Survey of Youth 1979 has followed a national sample of Americans since 1979. The women in it reported whether they smoked during each of their pregnancies, and in 2008 their adult children were asked: “How often does the federal government do what most Americans want it to do?”

I wrote down the main test and how I would run it before looking at the result.

What I Found

Among 3,701 adults born between 1979 and 1990, those whose mothers smoked while pregnant with them rated the government as less responsive. On a five-point scale, the gap was about a fifth of a step after accounting for family background. That is modest, but it is larger than the difference tied to a large change (two standard deviations) in any of the measures of the mother’s background, and those measures explained only about a fifth of the gap.

Is It the Smoke?

There are three possible explanations, and each predicts something different.

Exposure before birth. If this were the cause, a sibling born after a pregnancy with smoking should trust the government less than a sibling born after a pregnancy without it. They did not. The comparison is small, fewer than 200 families, but it did pick up the known effect of smoking on birth weight (about 137 grams lighter), so it can detect a prenatal effect when there is one.

Growing up around a smoker. If this were the cause, children of mothers who smoked, but not during their pregnancies, should show the same gap. They did not. They answered like children of mothers who never smoked.

A marker of the family. If smoking in pregnancy identifies a certain kind of family, the mothers should show the same pattern. They did.

What It Marks

The gap is in trust, not interest. Adults whose mothers smoked while pregnant trusted government and other people less, but followed politics just as closely.

Smokers in general show the same pattern. In the parents’ generation, smokers, men included, rated the government lower than people who never smoked, and lower than their own never-smoking brothers and sisters. I also tested this in the General Social Survey from 1977 to 1994, with the prediction written down before I had the data. Adults who smoked had less confidence in the federal executive branch and less trust in other people.

The most likely explanation is that smoking through pregnancy marks families that share a distrustful outlook, and that children get it from the family rather than from the pregnancy. The data show the pattern more clearly than they show how it is passed on.

Open Questions and Limits

Three things the paper can’t answer:

• Whether distrust leads to smoking, smoking leads to distrust, or something else drives both.
• How the outlook passes from parent to child. The mother’s own answer explains only a small part of it.
• Why mothers who smoked, but not during any of their pregnancies, don’t pass on the same pattern.

The main measure is also a single question asked in 2008, a year when regard for the federal government was unusually low. Smoking was self-reported, and the results describe one generation of American families.

The full paper, with the methods and every result, is on the paper’s page.