Friday, August 21, 2009

How to Manipulate Infant Death Data

Yuri Maltsev explains:

... the United States has one of the highest [inant death] rates of the industrialized world only because it counts all dead infants, including premature babies, which is where most of the fatalities occur.

Most countries do not count premature-infant deaths. Some don't count any deaths that occur in the first 72 hours. Some countries don't even count any deaths from the first two weeks of life. In Cuba, which boasts a very low infant-mortality rate, infants are only registered when they are several months old, thereby leaving out of the official statistics all infant deaths that take place within the first several months of life.

3 comments:

  1. Remind me of my first job after leaving university many decades ago.

    I was a junior statistical researcher for a government construction agency. My bosses were off in another state at some pow wow, and one of the secretaries to the head of the department rushed in. A question had been raised in the legislature about how our agency compared to 'private industry' in terms of it's youth employment.

    They needed the answer within a few hours and there was no time to discuss the job with the bosses, who were uncontactable anyhow.

    So I wrote three nearly identical reports. One showed we matched the standard for private industry for youth employment. The next showed we had a lower youth employment rate and the third a higher rate. In each case I used a different definition of private industry for comparison. One was based on the home construction and repair industry statistics (better than us), another based on the general construction industy ('worse' than us) and another the housing, construction and building materials supply industry (the 'same' as us). I submitted all three reports and let our glorious leader say whatever he wanted. Apparently he was most pleased with the result.

    My guess is that a similar story holds for most government statistics.

    ReplyDelete
  2. Another example of statistical manipulation in the health care debate
    ... See here.

    (EtW)

    ReplyDelete