Thursday, August 11, 2005

Why the Web is Good for Information

I hope this isn't too political, but it's a response in a way to a post fellow Lutheran blogger Bunnie Diehl had on her web-log here.

I recently received a sharp lesson in why the web makes us better informed. In this article, which appeared in the New York Times, Nicholas Kristof argues that the United States has a lower infant mortality rate than Cuba, largely because of our "regressive" social policy. Here, however, Brian Carnell gives a thoroughly convincing argument, backed up by a deeper look at Cuban infant mortality statistics, that this so-called fact is actually a statistical artefact created by the fact that in the US highly premature babies (under 1,500 grams) are treated as savable, even though 30-40% don’t last out the day. Other countries, with fewer resources, write them off as still-born and don’t enter them into the "infant mortality statistics." So the Kristof column is based simply on an error in statistics. (Nota bene: I found the Carnell essay from NRO’s Corner.) Today, I read the NYTimes letter writers responding to the Kristof column. Some are "pro-Cuba," some are "anti-Cuba," but every single one responds as if Kristof had actually observed a real phenomenon. Not a single one realized that what we are seeing is simply a quirk of misunderstood statistics. Chalk one up for the web.

Originally posted at Here We Stand

UPDATE: The Brian Carnell link seems to be down. The Wall Street Journal here has a number of other links on Kristof's statistical sloppiness, including this, quite different, explanation of the 2002 rise.

I linked to this post later here.