Oh (North) Carolina!

[jimmy olsen]

California on the other has given up trying to resist the sea.

www.msnbc.msn.com/id/47660947/ns/technology_and_science-science/#.T8tcp8V5fhw

[Razgovory]

I note that this graph stops 20 years ago, I wonder why. 

Just more fake science.  If you only look at a very narrow time in history you can see a correlation, expand the horizon and it disappears.  A snapshot in time ain't shit.

So what data would convince you of the validity of the theory.

[Hansmeister]

I note that this graph stops 20 years ago, I wonder why. 

Just more fake science.  If you only look at a very narrow time in history you can see a correlation, expand the horizon and it disappears.  A snapshot in time ain't shit.

So what data would convince you of the validity of the theory.

For starters, a multiple regression model with a 95% confidence factor, you know, which is usually considered the minimum for proof of a scientific theory, except of course in global warming theory, where we are supposed to accept it on the environuts sayso, based on computer models that don't actually work.

[Razgovory]

If such a thing exist will you change your mind?

[grumbler]

For starters, a multiple regression model with a 95% confidence factor, you know, which is usually considered the minimum for proof of a scientific theory,

Scientific theories are never proven, only disproven.

Please, make up more shit to justify your pre-selected position.  This is funny, funny shit.

[Razgovory]

I'm going to admit, I don't know what a "multiple regression model", is.  Or if can have a "95% confidence".  This is a statistics thing, so someone else will have to pick up the slack here.  I honestly don't know if such a thing exists or if Han's requirements are reasonable.

[CountDeMoney]

If such a thing exist will you change your mind?

Only if Moses brings the tablets down from on high and changes the party platform in Tampa this summer.

[Barrister]

I note that this graph stops 20 years ago, I wonder why. 

Just more fake science.  If you only look at a very narrow time in history you can see a correlation, expand the horizon and it disappears.  A snapshot in time ain't shit.

So what data would convince you of the validity of the theory.

For starters, a multiple regression model with a 95% confidence factor, you know, which is usually considered the minimum for proof of a scientific theory, except of course in global warming theory, where we are supposed to accept it on the environuts sayso, based on computer models that don't actually work.

It's been a long time since I got my BSc, or took any statistics courses, but getting those kinds of results really depends on the type of science you are using.  There is of course only one earth - we can not run hundreds of tests to determine the precise results.  My own field of study was geology, where you have a simialr problem - you can not run experiments on the millions of years time frame that would be required.  So instead you extrapolate from the available evidence.

[Admiral Yi]

What kind of confidence level are the tests currently producing Hans?

[PDH]

Actually, since the 1990s there have been multiple regression analyses of flora and fauna reacting to temperature changes.  A quick look at Nature shows this.  However, Hans will always be able to hide in the fact that the data is only 100+ years old, and therefore not reliable.

Oh, Hans.  The reason that graph "only" goes to 1990?  It was plotted in the early 2000s, if you would have looked at the data.  There are more recent graphs as well, including the ones called for recently that incorporate data from far more collection points.  (I am sure, as a climate scientist, you recall the problem some statisticians raised about needing more)

[DGuller]

http://en.wikipedia.org/wiki/File:Climate_Change_Attribution.png

This figure, based on Meehl et al. (2004), shows the ability with which a global climate model (the DOE PCM [1]) is able to reconstruct the historical temperature record and the degree to which the associated temperature changes can be decomposed into various forcing factors. The top part of the figure compares a five year average of global temperature measurements (Jones and Moberg 2001) to the Meehl et al. results incorporating the effects of five predetermined forcing factors: greenhouse gases, man-made sulfate emissions, solar variability, ozone changes (both stratospheric and tropospheric), and volcanic emissions (including natural sulfates). The time history and radiative forcing qualities for each of these factors was specified in advance and was not adjusted to specifically match the temperature record.

Also shown are grey bands indicating the 68% and 95% range for natural variability in temperature relative to the climatic expectation as determined from multiple simulations with different initial conditions. In other words, they indicate the estimated size of variations that are expected to occur due to fluctuation in weather rather than changes in climate. Ideally the model should be able to reconstruct temperature variations to within about the tolerance specified by these bands. Some of the remaining misfit may be accounted for by the ~0.05 °C uncertainty in the temperature reconstruction. However, though the model captures the gross features of twentieth century climate change, it remains likely that some of the differences between model and observation still reflect the limitations of the model and/or our understanding of the histories of the observed forcing factors.

In the lower portion of the figure are the results of additional simulations in which the model was operated with only one forcing factor used at a time. A key conclusion of the Meehl et al. (2004) work is that the model response to all factors combined is to a good approximation equal to the sum of the responses to each factor taken individually. This means it is reasonable to talk about the temperature change due to individual aspects of the evolving man-made and natural influences on climate. The zeros on both plots are set equal to 1900 temperatures, and it is apparent that most of the 0.52 °C global warming between 1900 and 1994 should be attributed to a 0.69 °C temperature forcing from greenhouse gases partially offset by a 0.27 °C cooling due to man-made sulfate emissions and with other factors contributing the balance. This contrasts with the warming from 1900 to 1940 for which the model only attributes a net increases of 0.06 °C to the combined effects of greenhouse gases and sulfate emissions.

To be devil's advocate, this graph doesn't prove much.  Finding a model that fits observed history is the easy part.  Finding a model that predicts the future as well as the past is the hard part.  Overfitting a model is a very real danger in such exercises.

[CountDeMoney]

Please, make up more shit to justify your pre-selected position.  This is funny, funny shit.

[The Brain]

http://en.wikipedia.org/wiki/File:Climate_Change_Attribution.png

This figure, based on Meehl et al. (2004), shows the ability with which a global climate model (the DOE PCM [1]) is able to reconstruct the historical temperature record and the degree to which the associated temperature changes can be decomposed into various forcing factors. The top part of the figure compares a five year average of global temperature measurements (Jones and Moberg 2001) to the Meehl et al. results incorporating the effects of five predetermined forcing factors: greenhouse gases, man-made sulfate emissions, solar variability, ozone changes (both stratospheric and tropospheric), and volcanic emissions (including natural sulfates). The time history and radiative forcing qualities for each of these factors was specified in advance and was not adjusted to specifically match the temperature record.

Also shown are grey bands indicating the 68% and 95% range for natural variability in temperature relative to the climatic expectation as determined from multiple simulations with different initial conditions. In other words, they indicate the estimated size of variations that are expected to occur due to fluctuation in weather rather than changes in climate. Ideally the model should be able to reconstruct temperature variations to within about the tolerance specified by these bands. Some of the remaining misfit may be accounted for by the ~0.05 °C uncertainty in the temperature reconstruction. However, though the model captures the gross features of twentieth century climate change, it remains likely that some of the differences between model and observation still reflect the limitations of the model and/or our understanding of the histories of the observed forcing factors.

In the lower portion of the figure are the results of additional simulations in which the model was operated with only one forcing factor used at a time. A key conclusion of the Meehl et al. (2004) work is that the model response to all factors combined is to a good approximation equal to the sum of the responses to each factor taken individually. This means it is reasonable to talk about the temperature change due to individual aspects of the evolving man-made and natural influences on climate. The zeros on both plots are set equal to 1900 temperatures, and it is apparent that most of the 0.52 °C global warming between 1900 and 1994 should be attributed to a 0.69 °C temperature forcing from greenhouse gases partially offset by a 0.27 °C cooling due to man-made sulfate emissions and with other factors contributing the balance. This contrasts with the warming from 1900 to 1940 for which the model only attributes a net increases of 0.06 °C to the combined effects of greenhouse gases and sulfate emissions.

To be devil's advocate, this graph doesn't prove much.  Finding a model that fits observed history is the easy part.  Finding a model that predicts the future as well as the past is the hard part.  Overfitting a model is a very real danger in such exercises.

Pray tell how we are supposed to know the future, Nostragullerus.

[Viking]

For starters, a multiple regression model with a 95% confidence factor, you know, which is usually considered the minimum for proof of a scientific theory, except of course in global warming theory, where we are supposed to accept it on the environuts sayso, based on computer models that don't actually work.

Sometimes I shout at the TV tearing my hair out. This is one of these cases.

Why is it that you require 95% confidence for the case for Global Climate Change but seem to accept 0% for your politically preferred explanation?

If these are your terms then no explanation other than the Climate Change Hypothesis comes anywhere close to explaining the data.

[Admiral Yi]

That's not an unreasonable position from Hans.  A significance test is only testing the proposition that the covariance of the two variables is statistically different than zero.

Does anyone know what significance the t tests are generating?