Plus the difference between data and stories

Good day, my good friend.

Short and sweet today. So less time reading, and the weekend is that bit closer as well. Here is today’s story lovingly selected for you.

James

The statistical significance of uncertainty

This paper by Guido Imbens in the Journal of Economic Perspectives is one of the most thought-provoking I have read in a long time. It concerns how we use statistical measures such as p-values when considering uncertainty, a particular interest to us transport planners. This is a searing reflection of the practice of using such points of certainty, and the context in which it is delivered. The author’s first observation is particularly reasonant:

I argue that in many cases, p-values and indicators of statistical significance do not answer the questions of primary interest. Such questions typically involve making (recommendations on) decisions under uncertainty.

But his final point is the most worrying:

…the overemphasis in empirical work on statistical significance has led to abuse of p-values in the form of p-hacking and publication bias.

I have not seen any evidence of either of these in the work I have been involved in. Nor do I ever hope to see it. But we must be ever-vigilant.

Something interesting for today

I have seen this everywhere, and so cannot attribute the original source. But it says it all, really.

If you do nothing else today, do this

Read the IPPR report on how to decarbonise fairly.