Monday, April 4, 2011


It is always useful to remind ourselves of the correct use of denominators, like in today's offering from Statistical Modeling, Causal Inference, and Social Science:

The article also explicitly discusses the fact, previously discussed on this blog, that it's misleading, to the point of being wrong in most contexts, to compare the safety of walking vs cycling vs driving by looking at the casualty or fatality rate per kilometer. Often, as in this article, the question of interest is something like, if more people switched from driving to cycling, how many more or fewer people would die? Obviously, if people give up their cars, they will travel a lot fewer kilometers! According to the article, in Denmark in 1992 (!), cycling was about 3x as dangerous per kilometer as driving, but was essentially equally safe per hour and somewhat safer per trip.

Transportation safety is a tricky thing and it only gets trickier as you try to define the best possible measure of risk. However, it is worth noting that when results vary as widely as cycling does based on the selection of the denominator then it is worth reporting all of the possible metrics. Otherwise, the person presenting the data is makign a decision as to what is the most relevant comparison.

For example, opponents of urban density may see commuting distances as inflexible and be interested in the risk per mile. On the other hand, advocates for cycling may well point out that the decision to cycle may feed into the decision of where to live.

But it is a good point to remember just how easily convincing measures of association can be misleading.

No comments:

Post a Comment