Wednesday, May 3, 2006

Infant Morality

The high rate of infant mortality among black women in New Haven would be a less vexing question if the there were as many black women as whites, according to a concerned caller from Westville.

The rate of infant mortality among black women is about 2.5 times higher than the rate in white women. This is troubling because the figure seems to hold across socioeconomic lines, does not depend on the relative rates of cigarette smoking, or the amount and timing of prenatal care.

In other words, white mothers who smoked and neglected their health -- yes, stupid things to do -- still did not have an infant mortality rate as high a blacks.

But, the thoughtful man from Westville argued, a valid comparison between the races could only be made if equal groups of black and white women were studied. Black women have twice as many babies as whites, he alleged, which throws off the statistics.

The infant mortality rates are expressed as the number of deaths per 1,000 births.

"But," the caller said, "If you had as many black women they have twice as many babies so the statistics wouldn't be so bad."

The logic is subtle, in a dunderheaded kind of way. Twice as many babies, so twice as many infant deaths.

What we have here is an issue of what statisticians call "sampling." Sampling error would be including an abnormally high or low number of black women with dead babies. However, increasing the sample size only lowers sampling error a little. But infant mortality is a description of existing data, not an exercise in probability, as the Westville man believes.

Moreover, the mortality rate is stated as a common ratio, per 1,000. That scales up or down the results accordingly. So the unfortunate truth is that infant mortality is higher in the black than the white community.

Do all of these numbers seem impossibly dry and boring?

If so, get used to it and learn about it or stop offering opinions on matters of statistics.