Monday, January 31, 2011

Hard decisions call for ignoring predictions?

The biggest issue my children worry about - daily - is whether or not school will be canceled due to inclement weather. Of course, you can guess which decision they are rooting for ... My son is not a fan of the Superintendent of schools because, in his estimation, the superintendent does not call enough snow days. Still, I was surprised when my son said that the superintendent ignores weather forecasts! Amazingly enough:

However, as superintendent of our school district, I will not call a snow day based on a weather forecast. I will call a snow day based on existing weather conditions such as significant snowfall or dangerous wind chills. I will call a snow day based on the city’s ability to make streets passable and our maintenance staff's ability to make our schools accessible and our parking lots clear.
[emphasis added]. So - the National Weather Service digital forecast for the occurrence of precipitation was 82% accurate for Lincoln in the last month. (Aside - I'm not sure how Forecastwatch.com is calculating that number, but it seems to be a good number to me.) Sure, it shouldn't be the only factor involved in making a decision to close schools, but surely it is a useful source of information to make the decision farther ahead. As a parent I appreciate knowing as early as possible that school will be canceled so that I can make alternate arrangements. I can't see how a decision can be made the night before (as it was earlier this month) based on existing weather conditions. Has to be existing weather conditions PLUS A PREDICTION, and if the Superintendent isn't looking at the forecast for the next day, then I guess he's doing it in his head. Maybe he's in the wrong job if he can do a better prediction than the National Weather Service.
I think this is just more evidence that society at large a) doesn't understand the variability of nature, and b) devalues science that only makes probabilistic predictions. Conservation biology is stuffed.

UPDATE: They just called a snow day at 9:30 pm. Someone is using some kind of prediction.

Wednesday, January 19, 2011

Education is bad for you!

John Quinn pointed me to a blog post by Jason Collins about the effect of mathematical education on risk tolerance. Collins was musing about the consequences of a psychology paper from 2008 that demonstrated how one's innate concept of the number line shifts from a logarithmic scale to a linear scale as one is educated in mathematics. They went further, and conducted the same tests with people from Amazonia who had little contact with the outside world - sure enough, adults there also used a logarithmic scale for their concept of number.
Collin's contribution was to connect this to the use of logarithmic utility as a mechanism to model risk aversion in economics - the tendency to avoid a gamble even if the expected outcome is the same. If learning math makes you think linearly, maybe it also reduces risk avoidance! At least if you regularly make decisions by mapping out risk curves ...

Thursday, January 6, 2011

from 10% to certainty in 2 breathes or less

Having just spent a bit of time thinking about how risk and uncertainty are thought about in different disciplines, the sound bite at 9:50 of this video caught my attention! Dr. Larry Brilliant of the Skoll Global Threats Fund describes an expert estimate that there is a 10% chance of a flu pandemic that kills 100 million or more people in the next 10 years - the interviewer responds by saying "So its certain there will be a pandemic, it is just a question of the time frame". Those seem like two radically different statements to me.

Actually it reminds me of an interview with a Nebraska State Legislator on NPR this morning - paraphrased, he said that the BP spill in the Gulf of Mexico made legislators realize that pipeline technology could fail ... and hence they started paying more attention to the Keystone XL pipeline issue. Yes! Of course it can fail! If you drive enough miles, the cumulative probability of having an accident approaches 1! The idea that people need to be certain that an event will occur in order to start thinking about doing something about it is amazing.

Hookahs and Anecdotes

From Andrew Gelman's Blog:

The evidence is certainly all around you pointing in the wrong direction - if you're willing to accept anecdotal evidence - there's always going to be an unlimited amount of evidence which won't tell you anything.
This is in the context of a panel of experts wondering if Hookahs cause lung cancer - one of the esteemed panelists used the fact that an uncle lived to 90 while smoking a hookah every day. I think there is an additional psychological mechanism involved in accepting this kind of anecdotal evidence - it is the direct experience of the person making the claim. Unfiltered by statistics, other people's attention to detail, and possibly dodgy methodology. It is particularly easy to accept anecdotal evidence when the process in question is impossible to experience directly - like the population level risk of cancer - or in my case, density dependent reductions in population vital rates. Even when faced with their own data, plotted in a different way to demonstrate the population consequences, people cling to their own experience. And unfortunately, density dependence isn't something you can experience directly.

Tuesday, January 4, 2011

It's all Dragons in the mind

I just listened to a great podcast on the Psychology of Climate Change. Although Robert Gifford's "Dragons of Inaction" were cast in the framework of climate change, they are all relevant to environmental decision making generally.

Thanks to Kate Buenau for bringing this to my attention.