People Doing Statistics Badly #2

This is part two of an open-ended series. Part one is here.

Let’s talk about babies and the problem of incomplete context. There is an article over at the Atlantic that talks about the approximately 9000 trips to the ER associated with babies falling out of high-chairs and the need to abolish the high-chair from the American family. What the article doesn’t discuss is all the other causes of infant/toddler morbidity.

I’m not suggesting babies falling out of high-chairs isn’t something to which we can apply better engineering controls. Far from it. In fact, the author of the article in question makes a number of imminently practical suggestions tied to the observation that an infant toppling over on the floor is exposed to much less risk of injury than one who topples off a high-chair.

On the other hand, the article begs the question of the level of risk to children required before action should be taken (up to and including government intervention). So here are some questions that must be answered before I can take this seriously as a threat:

  1. How many babies are there in America in any given year?
  2. How many of them are in families which use high-chairs?
  3. What is the risk of serious injury in terms of babies injured per thousand babies per year for high-chair incidents?
  4. What are the other major sources of infant/toddler morbidity and mortaility and what is the risk expressed as in (3)?

There is a final question which must also be answered: “At what point have we made the world safe enough?” Many will argue this is only when everyone is immortal and can never sufer injury. That includes everyone who argues we should do “everything we can” because they ask for an open ended expenditure which can only be satiated in the noted state of the world.

Until we start answering these sorts of questions in a rigorous and meaningful way the statistics presented in the article will continue to be without the necessary context. Without context, the article has emotional appeal but lacks the sort of substance which may reasonably expected to sway the thoughtful reader to action.


About this entry