Comments on: Crisis chart of the day: The correlation between severity and probability http://blogs.reuters.com/felix-salmon/2010/01/14/crisis-chart-of-the-day-the-correlation-between-severity-and-probability/ A slice of lime in the soda Sun, 26 Oct 2014 19:05:02 +0000 hourly 1 http://wordpress.org/?v=4.2.5 By: DavidMerkel http://blogs.reuters.com/felix-salmon/2010/01/14/crisis-chart-of-the-day-the-correlation-between-severity-and-probability/comment-page-1/#comment-11235 Mon, 18 Jan 2010 16:01:13 +0000 http://blogs.reuters.com/felix-salmon/2010/01/14/crisis-chart-of-the-day-the-correlation-between-severity-and-probability/#comment-11235 Cute chart, Felix. On page 38 of the report it starts to describe how they generated it:

The Global Risks Landscape

The visualisation of risk on the landscape places risks
by severity of impact (measured in US$) on the vertical
axis and the likelihood of occurrence on the horizontal
axis over a 10-year time horizon. The numerical
assessment of these categories of risks is created
through qualitative assessment by the partners of the
report. The risks which appear in the upper right-hand
corner are those with the highest impact and highest
likelihood and are the focus of the narrative of this
report.

I’m a statistician by trade. Whenever I see a nifty graph, I think: “Wow, how would you generate a graph like that?” This was one of those graphs.

Surveying expert opinion is sometimes the best method for generating data because there is no other reasonable option. But expert opinion is prone to biases — man is a social creature; even scientists are prone to the biases of their peers, much more the rest of us.

As a result, I wouldn’t give this graph too much weight. If the experts are generally right, it will indicate some weak tendency toward the result, but as for numbers, I would be reluctant to put one significant figure, much less two sig figs.

One other note: I think the timespan of the figures in the report are meant to span the next 10 years. That is an aid toward understanding what their hypothetical probabilities mean.

]]>