Let me ask you something. Doesn’t it strike you as the least bit odd that global warming would stop in 1999? Just up and stop one year? Does that seem characteristic of natural phenomena? No waning, just up and stop one day.
Allow me to explain why any article that claims that global warming stopped X exact number of years ago should be immediately suspect. Since you’re an accountant (aren’t you?), I assume you know something about statistics, right? If I take a data set, say temperature records going back a hundred years, and I find an upward linear trend, do a statistical significance test, and find that I have a very much statistically significant trend. Next suppose I take a 50 (say, 1920-1970) year chuck out of that 100 year data set and run the numbers again. Curiously, I will find that the trend appears less statistically significant than from the 100 year trend; the p-value is greater. Then, if I take just 20 years (maybe 1960-1680) the p-value gets even bigger! Eventually, if I make the data set (timescale) small enough, the resultant trend appears to be statistically insignificant; maybe, that timescale happens to be 15 years; that is, only on a timescale of >15 years does the trend appear significant. Does this mean that the 100 year upward trend is nonsense, because it doesn’t show up in any 15 year selection of the data? Obviously not. For one, larger data sets are inherently more reliable than smaller ones. The signal to noise ratio increases as timescale decreases. Same as with stock market trends or yearly climate cycles. If you took temperature readings in the northern hemisphere for every 6 hour period between Sept 1 and Dec 30, you would find that almost none of those 6 hr periods show a significant downward trend in temperature. Would you consider this proof that Winter has ceased to happen? Just because the temperature change concomitant with it cannot be observed on a small 6 hr timescale?
Wanna know how they came up with that ‘15 years’ claim? The did everything backwards, basically. They set the p-value equal to 0.05 (assuming they were doing a t-test for the slope of the curve; maybe they used another kind of test), standard indicator of statistical significance, they then plugged the data from temperature readings into the significance test, and what they got was the number of years over which they could say that the trend was no longer ‘statistically significant.’ 15 years was the largest timescale they could do without the p-value going. I have a suspicion if you looked at the data for 16 years going back to '98, you’d find that the trend was in fact statistically significant again.
Of course, a picture is worth a thousand words:
Edit: graphic didn’t load for some reason. I’ll just post the link:
skepticalscience.com/graphics.php?g=47
It only takes a couple seconds to see the whole ‘animation.’ I forgot that it wasn’t just a static image.
Anyway, still think it’s reasonable to look at that last flat short-term ‘trendline’ and declare that it’s stopped?
You say ‘and counting.’ If I’m right (and I’m willing to bet you good money that I am) they’ll still be saying it stopped “15 years ago” or maybe 13 or 16 or whatever, in 2025. And if they’re still doubters, in 2050. Funny how that number never seems to get much bigger than that.
