Uncertainty in Vienna... #EGU2014 by Murray Lark

In a previous post Murray Lark discussed some recent work by BGS and the Geological Survey of Ireland on how to communicate uncertain information derived from geochemical surveys. Murray's post got wide attention from colleagues and those in the community so we thought we'd catch up with Murray after his session at EGU2014 and see how it all went:

Good discussions around the poster displays
The EGU meeting was a collaborative enterprise with colleagues from BGS (Fiona McEvoy), Rothamsted Research (Dr Alice Milne) and Wageningen University (Dr Gerard Heuvelink). It brought together statisticians and mathematicians along with earth scientists from a range of disciplines, and a psychologist, Dr Adam Harris, from University College London, who is particularly interested in how uncertain information is perceived. 

To give you a flavour of proceedings here are some questions that came up. Whilst there are no easy answers I've picked out some points from our exciting and lively discussions (facilitated by a great mix of people and interests):

Q1. How do non-specialists interpret the fixed verbal expressions ("likely", "very unlikely","about as likely as not") that the Intergovernmental Panel on Climate Change (IPCC) uses to convey uncertain information to policy makers and the public?  Are the translated expressions interpreted in the same way in different countries?

A1. There is a tendency for non-specialists to interpret the verbal expressions "regressively", that is, large or small probabilities tend to be interpreted closer to "50:50".  That really matters.  For example, when the IPCC recently stated that "It is very likely that there is a substantial anthropogenic contribution to the global mean sea level rise since the 1970sthis means that the probability is over 90% (which is roughly the same as the probability of getting more than one "Head" in ten tosses of a fair coin).  Clearly it matters that voters and politicians clearly understand the strength of the evidence.

Q2. Geological mappers know that the lines they draw on maps are not completely certain, and they take that into account when they interpret maps.  Can we reproduce the expert mapper's understanding of uncertainty to explain the reliability of geological boundaries to data users?

A2. Yes, by using methods of "expert elicitation" groups of geological mappers can sit down and come up (usually) with a consensus  on how to represent their interpretation of uncertainty.  Some soil surveyors at the meeting were keen to join in.

Q3. What kind of statistical plots work best at explaining uncertain information about greenhouse gas emissions to policy makers and industry representatives — and does it depend on how good they were at maths?

A3. Not surprisingly the best way to present uncertain information depends on the kind of question you are asking, and which method you prefer does depend on how good you are at maths.  The take-home message is "know your target audience", but the research identified more specific guidelines too.

Q4. We can use powerful statistical methods to map properties of rocks or soils from data, and the more measurements we make the better the map will be.  Can we show the relationship between numbers of measurements and the uncertainty of the map in a way that will help a non-specialist decide how much money it is worth spending on data collection?

A4. A method to express the reliability of maps of properties on a simple intuitive scale from zero to one has been developed at BGS.  You can read the details here

Please get engaged and leave any questions or thoughts in the comments box below, if you were at the meeting in Vienna what did you think, can you add any more to the Q&A's?

by Murray Lark

For "I know" seems to describe a state of affairs which guarantees what is known, guarantees it as a fact. One always forgets the expression "I thought I knew". Ludwig Wittgenstein "On Certainty" translated by D. Paul and G.E.M. Anscombe.  One morning on the way to the congress I called in on the house that Wittgenstein built for his sister Margaret.