I'm a professor at U Michigan and lead a course on climate change problem solving. These articles often come from and contribute to the course.
By: Dr. Ricky Rood , 7:07 AM GMT on February 17, 2015
Nell, Dudley and Snidely: Uncertainty
In last week’s article I wrote:
Probability and likelihood are notoriously difficult ways to communicate in quiet consultation, and even more difficult in newspapers, on the radio, television and online. Probability and risk are just made for conflicting headlines. The conclusions are, therefore, by definition, uncertain, and uncertainty can always fuel both sides of a rhetorical or a political argument.
I got a very nice comment from Scott Sabol of WJW FOX 8 about the “uphill battle attempting to communicate uncertainty both in day-to-day weather forecasts and the describing of the components of extreme weather events/climate change influence without alienating the audience.” I have seen several blogs since the blizzard forecasts of January 23 – 26 that focus on the need to better quantify and describe the uncertainty associated with winter storms. Uncertainty is subject of this article.
Here’s a still growing record of the Northeast blizzard news cycle on my Tumblr site. This record includes some of the blogs referenced in the previous paragraph that discuss the need for better communication of uncertainty.
In the fall of 2014, I taught a small course on uncertainty, and specifically, on placing uncertainty of climate change in context with other sources of uncertainty in applying climate knowledge to planning and policy. My starting point in many uncertainty discussions is from the uncertainty fallacy; namely, that the quantification and reduction of uncertainty is the primary barrier that hinders the use of scientific knowledge in decision making. During the 1990s, many proposals and measurement missions were sold on the promise of “reducing uncertainty.” If you consider all of the complex processes that make up the climate system and their simplified representation in models, then casual statements that uncertainty will be reduced by any one investigation are not likely to hold up. Uncertainty might be better understood and be better described, but reduction is unlikely. Further, reduction does not assure better usability of knowledge, and in most cases it is not required.
One of my favorite classroom experiences is when the business students in class describe to the scientists and engineers that they are always making decisions in the face of great uncertainty. They want to know how climate uncertainty stacks up against other sources of uncertainty. They also what evidence that changes in the uncertainty descriptions will be incremental; that is, for example, from one assessment to the next, the description is largely the same.
If you listen to the NPR series on Risk and Reason, you will get a feeling of the difficulty of communicating uncertainty and the difficulty that people have in using information about uncertainty. In that series, there are those who advocate never using numbers describing uncertainty in policy contexts, and then there are those thinking of clever and effective ways to communicate numbers to individuals making important decisions. One take away is that how people use information about uncertainty is highly personal. There are often strong elements of fear and want.
Also, in many cases people have an agenda of how they want to use uncertainty – to make something happen or to keep something from happening ( a Rood blog, an ancient Rood blog, and yet another Rood blog, enough).
The quest for uncertainty quantification and highly quantified descriptions of uncertainty to assist in decision making is a mistake often made by scientists. In the cohort of clients I work with, the vast majority is simply not prepared to work with highly quantitative descriptions of uncertainty. Even more to the point, when climate uncertainty is placed into context with other sources of uncertainty, the quantification is overkill. There are studies that suggest, for instance Tang and Dessai (2012), that highly quantified descriptions of uncertainty can, on average, reduce the usability of climate information.
All of these factors together lead to at least one robust conclusion, there is no way to communicate uncertainty in a usable way to everyone. Therefore, you need have several strategies for communicating uncertainty, and you need to frame those strategies for different audiences. In the work that I have done with experts in public health, there is always the discussion about how to communicate a risk, for example, heat waves to the public. There is also the discussion of how to communicate information to first responders and to emergency health providers so that they will be on the lookout for heat-related afflictions. I am not aware that there is any discussion to communicate to anyone the numbers from epidemiological statisticians that one type of heat index has some fractional advantage in predicting heat-related afflictions.
An important point is the need to make a special effort to communicate to those who are trained professionals and have a framework in which to interpret and use uncertainty information. In the case of a weather emergency, one imagines that large cities might have such professionals. One of the most interesting responses that I saw in the Northeast blizzard news cycle was one where funding for experts, interpreters, in providing guidance on the use of forecasts had been eliminated. I don’t know the complete knowledge chain from weather forecast to shutting down a city, but this type of expertise is critical at some place in that knowledge chain.
My whole raison d’être these days is training interpreters on how to use climate knowledge in problem solving. Many of the same principles apply in how to use weather forecasts and how to use science-based knowledge in general. The Northeast blizzard news cycle has been and continues to be a real-world example for both climate and weather. The continued snow storms in Boston, for example, are a wonderful example of relentless patterns of weather that demonstrate that weather is not “random.” However, the biggest lessons are on uncertainty, communication and exaggeration for the benefit of telling a story.
I stated, above, that how we use uncertainty is highly personal. I have used climate knowledge and weather uncertainty to choose the location of a house on the Chesapeake Bay as well as to decide whether of not to take a kayak out into a hurricane. In neither case did I feel I was taking on large risk. This weekend, I (over)heard what seemed to be a discussion of two people deciding not to vaccinate their son because they had determined that their son had exceptional natural immunity. A relevant weather-related example of personal choice and, perhaps, the subconscious is the evidence that people take hurricanes named after women less seriously than hurricanes named after men. It made me think of naming winter storms and what Venus or Vesta might suggest compared to Jupiter or Mars. That, of course, led to Nell, Dudley and Snidely.
One thing that I count on from scientific organizations is a dispassionate description of events and uncertainty. Winter storms, especially if we are going to personify them, need a dispassionate, standard scale to describe them. The weather service has several scales that are effective for hurricanes, tornadoes and storms at sea. Winter storms offer a difficult detail, namely, the rain-ice-snow line, whose boundaries are tricky and important. Climate change offers the additional difficulty that characteristics of storms are changing and expected to change more. Therefore, placing storms within recent and historic context seems like a potentially usable piece of information. We need qualifiers, not number-heavy quantifiers. We don’t need to explain numerical dispersion errors in models to the masses. We don’t need to break down all of the pieces – to speak loudly and more slowly.
From the point of view of the climate scientist and the roles that climate change plays in a particular storm – it is always true that public communication is walking into a maelstrom where people have many agendas of how they want to use uncertainty – to make something happen or to keep something from happening. I have had colleagues tell me that there is an imperative to participate in ever loudening ways to convey the knowledge of climate change. This does not appeal or seem effective to me. Those conversations of deliberate disruption and doubt need to be identified for what they are and left in their stewing pool. We need to persistently differentiate the important aspects of climate change, isolate the deliberate disruption, and more effectively expose that which is important about climate change in the many conversations that are emerging.
The views of the author are his/her own and do not necessarily represent the position of The Weather Company or its parent, IBM.