[[ Check out my Wordpress blog Context/Earth for environmental and energy topics tied together in a semantic web framework ]]

Sunday, December 12, 2004

Global Warming: the Numbers...

...apparently mean something to the EconoPundit. What they mean to me, I haven't a clue.

Clearly triggered by uncontained enthusiasm for the recent Michael Crichton novel State of Fear, the EconoPundit tries his hand at analyzing something outside his area of expertise (shades of Ross McKitrick). Inspired by the novel's pulp fictionality and moved to action by a seemingly authentic supplementary material "appendix", the E.P. wades through the Crichton bulletin board and finds a link to a global warming "database" site. Very pleased with himself on his ability to crack the password-protection of the evil climatology cabal's firewall, he proceeds to engage in expert number crunching. He starts by firing up the professional's choice, Microsoft Excel. The E.P. then proceeds to analyze with the insight gained through years of econometrics practice along with green-eyeshade determination.

After a brief flirtation with a scatter plot, the E.P. gets down to business and performs a time-series analysis on an arbitrary set of metrics.

Hmmm. Interesting that he chose a plot of "number of record high-temperature days per year". Questionable that he considered only data from the USA (what is the story about the drunk, missing car keys, and a lamp post?). Damning that he decides to use the sparser outlier data, max temperatures, of which only one occurs per year at each recording station.

In conclusion, the E.P. sez:
"So you tell me. Is this evidence of global warming? Proof fossil fuels are (via the greenhouse effect) changing the weather? What does this show? Anything?

UPDATE: And by the way: going to a database from link on a strange bulletin board, being surprised the huge database has loaded because you've (inadvertently) nullified the password screen by pressing "CANCEL," idly processing some of the data and being surprised at the resulting graph (which shows little sign of global warming) -- this is all just like stuff that happens in Crichton's novel!!!!!"

I would say this has all the earmarks of fine detective fiction styling, very capably emulating the simplistic plotlines of Nancy Drew, Hardy Boys, and (my favorite) Encyclopedia Brown. Recommended for Ages 6 and Up.

I discovered EconoPundit's analysis via an InstaPundit link. My own conspiracy theory: Why else would this weak an analysis get linked other than due to nepotism amongst the Pundit clan?


Professor Anonymous Anonymous said...

That "scatter chart" isn't a scatterplot. It's visibly a time series plot - the clue is that the x axis is showing time. His "time series" plot below with its polynomial trend (Chart - Add Trendline - Polynomial) is just the same chart as the "scatterplot", but with lines joining the points.

I think that Econopundit might have become confused and thought it was a scatterplot because some enemy of the state had set "line style" to "None" in Excel. In general I have long campaigned for the removal of the polynomial trendline option because it is overused by people who like the spuriously smooth lines. In almost all contexts, the "linear" or "moving average" trendlines are more accurate summaries of underlying data.


12:24 AM  
Professor Blogger WHT said...

Good points. I typically glance at a particular chart and figure out what the underlying assumptions are before making a decision regarding it's value. In this case, whatever the trends show, the premise of choice of data is extremely questionable.

5:42 AM  
Professor Anonymous Anonymous said...

I suppose my professors may have not known anything, but when I was in college a "scatter diagram" was the dataset illustrated in "xy" space. If time was "x", so be it. A scatter diagram with time on the horizontal.

The polynomial trend was plotted using TSP (EViews 5.1, actually), not Excel.

But heck, you guys are all smarter than me. Tell me where the warming is?


7:39 AM  
Professor Blogger WHT said...

They call it a scatter plot to convey that the diagram consists of a bunch of data that may or may not be correlated.

7:03 PM  
Professor Blogger Tim Lambert said...

Dear Mr Econpundit, to see global warming you might perhaps plot something that is a) a temperature and b) global. Like this

3:29 AM  
Professor Blogger Max said...

Well, and everybody else can come back, if they can make a computer model that can predict the temperature of the next 5 years correctly with a very small difference perhaps. If they cannot, they are no use to demonstrate anything....

If there is global warming or not is irrelevant, it is climate change that is relevant. There is climate change and it will come, whether humans can or cannot accelerate it by few hundred years. So, I take it with Douglas Adams: Don't Panic!

9:10 AM  
Professor Blogger WHT said...

Strange what happens when you link to a GIF titled "UNTITLED.gif"

I think it has since been changed to

9:41 AM  
Professor Blogger Steve Austin said...

Informative blog. I have a xml form blog.

8:57 AM  

Post a Comment

<< Home

"Like strange bulldogs sniffing each other's butts, you could sense wariness from both sides"