January 08, 2018
The Easy Money; Problems with Convenience Samples
Timothy Birdnow
Think planetary temperature data are accurate measurements based on physical data? Think again.
According to the article by David Wojick:
"Contrary to popular belief these are not measurements. They are the output of complex statistical models. These statistical models are every bit as questionable as the climate models they feed into, actually more so."
[,,,] "But the satellites show no such warming in the atmosphere over this period, where it should be if it were caused by greenhouse gases. The satellites show no warming at all over this crucial time. This zero warming strongly suggests that the surface statistical models are wrong.
Keep in mind that these global temperature statistics are no different than a voter poll prior to an election and we know how wrong they can be. An incredibly tiny subset of the overall population is being sampled. In this case the overall population is the temperature every place on earth at every moment over an entire year.
The pollsters know that a lot can go wrong. Apparently the alarmists that cite these crude temperature estimates as precise facts do not, or they choose to ignore the problems, in which case they are faking it.
There are at least ten things wrong with these statistical models. These flaws support the view that these crude temperature estimates are simply wrong. Some flaws are well known, like arbitrary adjustments an"
d the urban heat island effect. Other weaknesses are less well known, like local heat contamination, the use of area averaging and interpolation, or the use of sea water proxies, as well as taking the mean value to be true when we know it is not. These will be topics of later analyses.
But here I present the deepest flaw, which is not widely discussed. This is that the surface statistical models are operating on what is called in statistics an "availability†or "convenience†sample."
End excerpt.
It should be pointed out that the number of actual surface stations providing the raw data have declined considerably during the last decade, at a time when logic would dictate we would increase them to study the "problem". This decrease is made up by averaging two samples from two other stations, as if an average is all one needs to obtain precision. There isn't much point in taking actual temperature data if we are simply going to assume that a point equidistant between two others - perhaps fifty miles apart each way - is going to simply be the average of the two.
The article continues:
"Here is an example from NOAA’s recent "Global Climate Report†for 2016:
"The average global temperature across land and ocean surface areas for 2016 was 0.94°C (1.69°F) above the 20th century average of 13.9°C (57.0°F), surpassing the previous record warmth of 2015 by 0.04°C (0.07°F).â€
A hundredth of a degree is incredible accuracy given that temperatures around the globe on many days can differ by a hundred degrees or more F. In fact it is not credible. The truth is that these surface statistical models are not merely inaccurate, they are worthless. Here is why.
The math of statistics is based on probability theory. Thus one of the absolute requirements is that the sample be random. If the sample is not random then the math is not applicable.
In fact the samples used in the surface statistical models are nothing like a random sample of the Earth’s surface. They are heavily clustered near urban areas and airports in developed countries. The locations were not chosen to be a global temperature sampling system and they certainly are not. The oceans are even worse because there are no fixed stations. Most of the Earth had no fixed temperature recording stations during the period in question, and still have none. There is no random sample of the Earth’s surface temperature.
In short the surface statistical models use the data that is available, not a random sample of the population."
End excerpt.
Bear in mind that climatology was a sleepy backwater, a branch of meteorology, until the global warming scare brought big bucks into it. Now huge amounts of government money is doled out for the express purpose of "proving' the theory to justify draconian government regulations, and in fact the fundamental reordering of human society. Fudging data, or at least providing bad data, is inevitable.
It's like those media polls that showed Hillary Clinton winning the election by ten points or more; they weren't intended to provide accuracy but rather to create a bandwagon effect, to influence the electorate.
I simply cannot believe that we are still talking about this idiocy after all these decades. This is a WAr of the worlds scam that has lasted a generation now. And the Millenials buy into it hook, line, and sinker, despite zero evidence to prove any of this is true.
Think planetary temperature data are accurate measurements based on physical data? Think again.
According to the article by David Wojick:
"Contrary to popular belief these are not measurements. They are the output of complex statistical models. These statistical models are every bit as questionable as the climate models they feed into, actually more so."
[,,,] "But the satellites show no such warming in the atmosphere over this period, where it should be if it were caused by greenhouse gases. The satellites show no warming at all over this crucial time. This zero warming strongly suggests that the surface statistical models are wrong.
Keep in mind that these global temperature statistics are no different than a voter poll prior to an election and we know how wrong they can be. An incredibly tiny subset of the overall population is being sampled. In this case the overall population is the temperature every place on earth at every moment over an entire year.
The pollsters know that a lot can go wrong. Apparently the alarmists that cite these crude temperature estimates as precise facts do not, or they choose to ignore the problems, in which case they are faking it.
There are at least ten things wrong with these statistical models. These flaws support the view that these crude temperature estimates are simply wrong. Some flaws are well known, like arbitrary adjustments an"
d the urban heat island effect. Other weaknesses are less well known, like local heat contamination, the use of area averaging and interpolation, or the use of sea water proxies, as well as taking the mean value to be true when we know it is not. These will be topics of later analyses.
But here I present the deepest flaw, which is not widely discussed. This is that the surface statistical models are operating on what is called in statistics an "availability†or "convenience†sample."
End excerpt.
It should be pointed out that the number of actual surface stations providing the raw data have declined considerably during the last decade, at a time when logic would dictate we would increase them to study the "problem". This decrease is made up by averaging two samples from two other stations, as if an average is all one needs to obtain precision. There isn't much point in taking actual temperature data if we are simply going to assume that a point equidistant between two others - perhaps fifty miles apart each way - is going to simply be the average of the two.
The article continues:
"Here is an example from NOAA’s recent "Global Climate Report†for 2016:
"The average global temperature across land and ocean surface areas for 2016 was 0.94°C (1.69°F) above the 20th century average of 13.9°C (57.0°F), surpassing the previous record warmth of 2015 by 0.04°C (0.07°F).â€
A hundredth of a degree is incredible accuracy given that temperatures around the globe on many days can differ by a hundred degrees or more F. In fact it is not credible. The truth is that these surface statistical models are not merely inaccurate, they are worthless. Here is why.
The math of statistics is based on probability theory. Thus one of the absolute requirements is that the sample be random. If the sample is not random then the math is not applicable.
In fact the samples used in the surface statistical models are nothing like a random sample of the Earth’s surface. They are heavily clustered near urban areas and airports in developed countries. The locations were not chosen to be a global temperature sampling system and they certainly are not. The oceans are even worse because there are no fixed stations. Most of the Earth had no fixed temperature recording stations during the period in question, and still have none. There is no random sample of the Earth’s surface temperature.
In short the surface statistical models use the data that is available, not a random sample of the population."
End excerpt.
Bear in mind that climatology was a sleepy backwater, a branch of meteorology, until the global warming scare brought big bucks into it. Now huge amounts of government money is doled out for the express purpose of "proving' the theory to justify draconian government regulations, and in fact the fundamental reordering of human society. Fudging data, or at least providing bad data, is inevitable.
It's like those media polls that showed Hillary Clinton winning the election by ten points or more; they weren't intended to provide accuracy but rather to create a bandwagon effect, to influence the electorate.
I simply cannot believe that we are still talking about this idiocy after all these decades. This is a WAr of the worlds scam that has lasted a generation now. And the Millenials buy into it hook, line, and sinker, despite zero evidence to prove any of this is true.
Posted by: Timothy Birdnow at
11:09 AM
| No Comments
| Add Comment
Post contains 817 words, total size 5 kb.
23kb generated in CPU 0.0073, elapsed 0.2401 seconds.
35 queries taking 0.2352 seconds, 157 records returned.
Powered by Minx 1.1.6c-pink.
35 queries taking 0.2352 seconds, 157 records returned.
Powered by Minx 1.1.6c-pink.
