## Spinning the climate

##### Jun 03, 2009

Mike L. pointed us to this pair of "climate change model pie charts", with the brief comment "Yuck".

What they are doing is to use the spinning wheel analogy to present probabilities (odds).  Not a good use of pies either.  Histograms do the job with minimal fuss:

I collapsed the 2-2.5 and 2.5-3 degrees sectors since every other one is a one-degree interval.  We see immediately that the effect of the policy is to shift the probability distribution to changes of fewer degrees.

Reference: "Climate change odds much worse than thought", Science Daily, May 20 2009.

You can follow this conversation by subscribing to the comment feed for this post.

Global warming is a tremendous hoax.

I've got to disagree with you on this one. This chart is aimed at the layperson, who probably (heh) doesn't have the intuitive grasp of probabilities that readers of this blog have. By making an analogy to a roulette wheel, a physical device that most people have at least a passing familiarity with, this chart conveys the fundamental message: "we don't know exactly what the temperature change will be, but which wheel would you rather spin?"

Furthermore, I believe this chart is often presented on posterboard with an actual, physical spinner on it, inviting the audience to actually interact with it to get a sense of the probabilities involved.

Pie charts are often misused, but this is one does the job well, in its intended context - educating the public.

Your histogram should go from small changes at the bottom to large changes at the top.

I blogged about this ugly visualization in Whatever, 25 May 2009. I didn't say much, just that they went way to far to try to support a weak metaphor.

i think the color works in the original and that makes it easier for me to understand than your histogram.

Orange and red = hotter with no policy

Blue and green = cooler with policy.

Tarek: I'd agree if this chart was interactive and allowed us to click and spin it. When it is frozen on the page, like a pie chart, the analogy to the roulette wheel is weakened.

Speaking of which, anyone knows of any experimental validation that such spinning wheels indeed generate the theoretical distributions? That, for example, the landing angle is independent of the starting angle?

I agree with Tarek on this one. There's a large segment of the population (*cough* lottery *cough) that have almost no comprehension of probability, and the wheel really helps. It would be nifty if it spun, but that's a much easier leap to make.

Regarding validation that wheels generate theoretical distributions, I suspect it's a lot like dice. In theory if you control absolutely every aspect of a dice throw it will be perfectly repeatable, but in practice minuscule changes are allowed to make huge differences in the outcome. I would hope some obsessive has compiled the histories for Wheel of Fortune over its decades long run.

One advantage of the pie charts you have lost is the ability to quickly read the median expected temperature change.

Another aspect worth noticing is the use of colour. Blue to red is an intuitive scale for warm/cold.

However should 1 or 2 degree warming be coloured a nice cool blue, which intuitively would represent cooling? Or a greeny yellow which would intuitively represent a mild and comfortable warming? Or a moderate orange representing a significant inconvenience?

And at the upper end various scientists have stated that a point ranging from 2 to 6 degrees is 'dangerous' which I think would best be represented by a strong red or browny burnt red.

It would all depend on what you believe about the impacts of climate change, and what message you were trying to convey.

And I'd also like to add that I'd seen this chart before and subconsciously interpreted the difference as being about double the warming for the no policy option. However your histogram shows that I have underestimated the difference and increased my perception of the difference.

The original chart and the histogram are both good-and both bad. As several people have pointed out, the spinner analogy is a wonderfully concrete way to convey probability distributions. But as others point out, the histogram is better for precise quantitative examination of the data.

To me, this is a good illustration of the fact that you can't judge a chart without knowing something about its intent and audience.

I'm not sure that showing WAGs with high levels of precision is at all helpful. We probably need to be careful with data visualizations that we don't overemphasize the degree of certainty behind the probabilities -- or that we allow data labels like "policy" and "no policy" to be the only explanations of the information presented.

The comments to this entry are closed.