Tuesday 28 July 2015

Micronutrients in the first trimester of pregnancy


Here is an interesting study, which finds what looks like a big effect. Ethiopian Jews moved to Israel, some by a massive airlift which brought in women in various stages of pregnancy. This natural experiment allowed the researchers to look at the possible effect of vitamin supplements given to pregnant mothers. Amusing title, as well.

Victor Lavy  Analia Schlosser, and Adi Shany. Out of Africa: Human Capital Consequences of In Utero Conditions. Draft July 16, 2015.


The authors say: Children who were in utero in Israel starting from the first trimester are about 12 percentage points more likely to obtain a Baccalaureate diploma than children who were in utero in Ethiopia during the first and second trimester but spent the rest of the pregnancy in Israel. This is a large effect since the average Baccalaureate rate of children who arrived at the second and third trimester is only 20 percent. Children who arrived to Israel during the first trimester also engage in more challenging study programs during high school relative to those who arrived at a later pregnancy stage. For example, they obtain 3.2 more credit units relative to those who arrived at the third trimester, an effect of about 33 percent. These individuals also attain 0.4 more credit units in Mathematics and 0.5 additional units in English, implying a gain of more than 50 percent. They are also 12 percentage points less likely to repeat a grade and 7 percentage points less likely to drop out of high school. 

We find that the effect of better environmental conditions in utero is larger and significant mostly among girls.

Try looking at Tables 3 onwards. The contrasting groups are in rows, not columns, which confused me when I looked at it as a welcome distraction after becoming punch drunk after posting on the “glass floor” LSE social exclusion report. For some reason, in the traditional brackets where researchers usually give the standard deviation these authors have put the standard error of the mean, which is much smaller. Surprisingly disorienting.

This study is of particular interest to me, because a PhD student of mine looked at whether the Iraqi invasion of Kuwait caused pregnant mothers to give birth to children with any behavioural disorders, and was unable to find any apparent disturbance. All the children were well nourished, and brought up in good material circumstances. However, health services were very poor at time of the invasion, and still no detectable effects (no long term look at adult outcomes).

It appears that vitamins might be doing something useful after all. Give it a close reading while I go off to bed.


  1. "the standard error of the mean": my freshman lecturer said that it was just like The Holy Roman Empire which was, you'll remember, not Holy, not Roman and not an Empire.

  2. Psychology researchers do not generally put the standard deviation in the brackets. They put the confidence interval there. Not the same.

    Other fields often use standard errors, which are closely related to confidence intervals (ratio about 2).

    1. Emil, things may have changed. In ancient times people gave means and standard deviations, the latter often in brackets. Confidence intervals came later. I don't mind conventions moving on, but the standard deviation helps me visualise the distribution.

    2. Yes, but it is not a robust measure so using it to visualize the distribution may be misleading (if it is non-normal).

      It does not tell you about the certainty of the estimated mean value, which is usually what people are after.