It seems to me that Roger Pielke Jr. is now persona non grata at Nate Silver’s 538 blog. Silver, whose statistics-based predictions of the last presidential election were stunningly accurate, hoped to capitalize on his celebrity by starting a blog which would present interesting and sometimes important posts which were based on competent analysis of actual data. What a concept! Unfortunately for Silver (in my opinion), he failed miserably when choosing Pielke as his point-man for the global warming/climate change issue. Many of us noted the kerfuffle which followed Pielke’s first foray, with Silver himself admitting publicly that it wasn’t up to the standard he hoped to set.
I wandered over the other day, checked out posts filed under “climate change,” and although I’m gratified to see that Pielke isn’t their point man any more, I wasn’t exactly impressed. The latest is about the use of the phrase “global warming” vs “climate change” by U.S. politicians. First of all, I think it kind of misses the point — which phrase is used is far less important, in my humble opinion, than how often either is uttered, by whom, and in what context.
Second, I didn’t really find much analysis in that post. Yes, there were numbers and yes, they were relevant to the subject — so I suppose I shouldn’t be too critical. Maybe I’m just too immersed in analysis-in-depth to know what the appropriate level is for a “popular” blog. Or, maybe Nate Silver should hire me.
Still, use of the phrases “global warming” and “climate change” is interesting, and happily, Google Trends provided me with data about their use as search terms so I didn’t have to do too much work (after all, Nate Silver isn’t paying me). Let’s look at the use of those phrases as Google search terms over the last 10 years or so, right here in the good old USA.
Here’s the data:
The red line shows the prevalence of searches for “global warming,” the blue line for “climate change,” and the black line is the sum of both. Note that Google Trends doesn’t give actual counts, they scale the data so that the highest value is equal to 100. The highest total is over 100 because I added the separate counts together.
Clearly “global warming” has been a much more prominent search term than “climate change.” But that may soon no longer be true. If we look at the fraction (in percent) of searches for “climate change” we note that it has risen steadily in popularity (compared to “global warming”) since about 2007:
If the present trend continues, “climate change” will soon surpass “global warming” as the search term of choice.
But I’m less interested in the choice of term, more interested in the total searches (the black line). There’s an overall rise up to about 2007, a decline since then, and to my eye it looks like there’s an annual, seasonal cycle present. It also looks like there’s more variation when the level is high than when it’s low, leading to the phenomenon of heteroscedasticity, which can often be removed by log-transforming the data. Here’s the logarithm of the totals (log-10 to be more intuitive rather than log-e which is my intuition):
This makes the presence of a seasonal cycle rather evident. But what is the seasonal pattern? I was expecting more interest during summer, when it’s hot, so I smoothed the data (the red line in the above graph) to remove the secular trend, then “folded” the data with a period of one year to show what that looks like (I’ve plotted two full cycles of the seasons, for clarity):
That’s a surprise. Summer is the season with the fewest searches, and there’s also quite a dip around Christmas, which is obvious if we average the many years’ values:
I have a strong suspicion that what we see here is the seasonal cycle of the school year. When school starts in September, searches go up — when it ends in May/June, they go down — and when students get vacation around the Christmas holiday season, there’s a brief span with very little activity.
Naturally, I subtracted the average seasonal cycle from the log-transformed data, then log-untransformed (i.e. exponentiated) back, to produce what I’ll call anomaly values, which give a much sharper image of the secular trend:
I’ve also plotted a red dot on those values which might suggest an “outburst” of activity.
Wherefore the changes? I can only speculate — but it’s my blog so I’ll speculate freely.
First, let’s consider the rise to 2007 and decline thereafter. I suspect that the rise is due to Al Gore’s film An Incovenient Truth and the publicity which preceded and followed it — including an academy award. This suggests that the person most responsible for making Americans aware of the issue is Al Gore.
As for the decline, I suggest that it does not reflect declining interest in the issue. Rather, I submit, it’s simply due to the fact that people are so much more aware of it today they don’t need to search for “global warming” — they already know what it is (at least roughly). Those who are seeking information these days through google search are probably being more specific in their search terms, looking for things like “sea level rise” and “heat waves” and “hurricane Sandy” rather than “global warming” or “climate change.”
As for the outbursts of search activity, I have hypotheses for at least two of them. The peak in early 2007 coincides with winning the Academy Award for Best Documentary Feature. The outburst in late 2009 coincides with the Copenhagen Summit on climate change, a news event which seems to have spurred public interest in the topic.
I suppose I could dig through news reports to identify what events may have triggered all the other outbursts of activity, and to define more precisely just how statistically significant those outbursts are. But I’ll leave that as an exercise to readers. After all, Nate Silver isn’t paying me for this.