Data Fit For the Purpose

By Bob Shapiro

When my daughter, Maria T, was in 5th grade, she took a gifted/talented class at her public grammar school. On Back-to-School night, I talked with her teacher for the course. Ken described the challenges he faced, including one student who found out that the value of PI couldn’t be calculated exactly.

She was terrified,” he said. “The girl was worried about the astronauts we were sending to the moon, whose flight plans depended on the

value of PI. ‘How could we put them in such danger?’ she asked.” Ken said he didn’t know what to tell the little girl.

I said he should have asked her, “How close is close enough?” PI has been calculated to thousands of digits, and all things being equal, the possible error based on PI would be far less than a millionth of the width of a strand of hair.

Of course, all else isn’t equal. The course calculation depends on other numbers with their own margins for error. Which is why the idea of a Mid-Course Correction was invented. This leads me to a subject we all learned (and forgot) in Middle School – Significant Digits.

Many government agencies take various measurements, and then combine them in ways that violate basic mathematical rules. For example, if a thermometer measures to the nearest degree, it is mathematical malpractice to come up with an average temperature to the nearest hundredth (or thousandth!) of a degree.

This would be bad enough, but it turns out that the overwhelming majority of the official weather stations being used for climate data cannot measure anywhere near to the nearest one degree!

Several years ago, a band of volunteers set out to audit the 1221 official weather stations, looking at their siting characteristics, to determine what the expected margin of error would be for each. The results are free for all to see on their web site, www.surfacestations.org. With over 80% of stations audited, only about 8% of the stations had an expected error under 1 degree, while 70% had expected errors over 2 degrees.

As bad as this looks, the US still is state of the art compared to temperature gathering around the world. And that is just for the 25% of the earth’s surface covered by land. For the earth’s oceans, there are 3600+ Argo Buoys, collecting temperature readings over a 10 day cycle. There are roughly 150 Million square miles of ocean, so each buoy takes 1 reading at each depth every 10 days, each covering a little over 40,000 square miles of ocean.

Most of the 3600 buoys have been collecting data for 10 years or less, so it will be a while before any trends can be determined from these data.

My point is not to pick on the climate change industry (well maybe a little). It is to emphasize that much of the data that our government uses to make laws generating monumental effects – often is lousy at best.

Action Item: We need to enact a law empowering an independent statistics watchdog with authority to invalidate any law which depends on faulty data for its justification. Each agency must be required to make unadjusted raw data available for all to see. Taxpayers pay for it, it belongs to taxpayers, let them see it.