Wednesday, 9 March 2011

How to cook a data set



Take a look at this chart, which has been doing the rounds of Climate Change denier blogs and other discussion forums under the title of 'NASA satellites reveal no Global Warming for 30 years'. The original source appears to be here, a blog called C3 Headlines.

Pretty conclusive heh? Four out of six data points showing cooling easily balancing out the two that show warming. The source of the data, the very respectable National Space Science and Technology Centre, adds a touch of respectability too.

So what's the trick?

Well, the data isn't made up. The original numbers are all here and, although it takes a bit of time to wade through them, the figures used on the chart are correct.

What's been done is a bit of good old fashioned cherry picking. That's easy enough to do, but what some eagle eyed denier has spotted is that the cherries that are worth picking lie at 5 year intervals. Neat.

The result is a bit like the Bible Code. If you have a big enough set of numbers then, rather like the million monkies with their typewriters coming up with Hamlet, there's always going to be a pattern that shows what you want to find. If you searched hard enough you'd probably find the results of the Six Nations rugby tournament hidden somewhere.

This trick wouldn't have worked last month, where the figures would have been 0.19, -0.04, -0.18, 0.03, -0.13 and 0.02 - a clear warming trend, even though this has been a cold winter.

The trick certainly didn't work last February either, when the data set would have shown an even stronger warming trend: 0.18, -0.09, -0.01, -0.24, -0.32, -0.07.

Games like this can be played all day, but they don't really prove anything. The only way to tell if we're warming or cooling is to do a proper statistical analysis, and that's shown on the bottom line of the data - a warming trend of 0.14 degrees per decade. At 15 years this trend becomes statistically significant with a confidence level of 95%. So, we are warming.

But I'm not going to leave it there.

If any climate scientist had tried to use a trick liked this to 'prove' Climate Change, then it would have been all over for the discipline. C3 Headlines can get away with this stuff because it serves a particularly gullible and self interested client group.

Scientists by contrast have to be open, honest and upfront. When Professor Phil Jones and the UAE were accused of fraud after Climategate their entire body of work was on trial.

The likes of C3 Headlines though are expendable. They put out their bogus data, which can ultimately be denied itself by those who fund it and quote from it.

As Jay Inslee, a Washington Democrat, lamented on Tuesday, if only the media would stop reporting the debate about Climate Change as if it was a divorce trial and start showing how barren the cupboard was on the denier side.

26/03/2012

Note: C3 Headlines appears to have responded to this post. I'd like to reply to their reply - but as they haven't actually answered any of my points I can't. Sorry guys, you got caught. Live with it.

2 comments:

Terry K said...

Hi Martin,

this is a very nice blog. I'm going to come back and have a better read when I have time.

By way of exchange, this one is mine.

http://terry-kidd.blogspot.de/

greetings from JetexJim

Martin Porter said...

Thanks.

Maybe one day this blog will work out what it is actually about!