Jamie pointed me to this graph in the NYT:
Nice! Especially the margin of error, the subtle colorings and the use of the gray background. The y-axis labels are a little weird (why not simply 60, 80, 100, and 120), and I’m not sure how to think about the x-axis. (Given the scale on the y-axis, should we really care about a change of 1/2 mile per hour in wind speed.) Also, I don’t really understand what it means to measure changes in wind speed, if the storms are themselves categorized by wind speed! But, as a graph, it has many excellent features.
This actually is likely a vacuous observation (the graph). Your comment about the two axes is exactly on the mark.
If the distribution of storm strengths is relatively flat at lower speeds, but significantly dropping off at higher speeds (almost certainly true), then an overall increase in storm intensity would not increase the average speed in the lower buckets, but could increase the average speed in the upper buckets because they have a flatter distribution by virtue of the shoulder of the distribution passing them by.
It would have been much better to just see the empirical CDF (but you couldn't publish such a thing) or, if there is enough data, a histogram.
I hope that somebody didn't get a scientific publication out of this observation.
What is the URL of the original article?
Answering my own question, the reference is likely to be http://www.nature.com/nature/journal/v455/n7209/a…
This abstract makes it clear that the real paper didn't do anything as clueless as it appears. They measured the shift in wind-speed quantiles which is a very sensible thing to do.
The white horizontal lines are the cut-off points for hurricane categories. Not sure why they'd pick the numbers on the y-axis though – it doesn't include category 4 or 5 hurricanes that way.
should we really care about a change of 1/2 mile per hour in wind speed
I presume that's half a mile per hour per year, over twenty five years, making a diffrence between 1981 and 2006 that is one we should care about. Perhaps the scale should have been written to reflect that.
Yes, it is 0.5 mph/yr, i.e. the trend. Presumably that's what they were getting at when they labeled the axis "annual change".
NU,
I appreciate sarcasm as much as the next guy. But I think Derek has a point that this is not a natural measure to use. I think the change over the past 25 years (which would be 12.5 mph) is more relevant. Even if the original graph is correct, a rescaling can make it easier to read.
Andrew,
I wasn't actually intending to be sarcastic. I was agreeing that the way they labeled the axis is somewhat confusing in its intent. The change over a longer period of time is the more relevant and interpretable figure (hard to intuit what 0.5 mph/yr may add up to over a longer period).
Intonation is so hard to read without the sound of the voice…