The meaning of R2 in pictures: Mann v. Steyn/Simberg/CEI/NRO
At the heart of the climate debate, is maths/statistics.
What the climate realists often point out is, that many/most/all ? the of the we are doomed statistical papers are either based on MODELS and not data, or the data papers have not been co-authored or even "peer reviewed" by statisticians.
Can you see a problem.
And if you want to know if a data set/sets is valid, then R2 must be used to see if the conclusions are robust.
I suggested this would best be done with visuals. Of course these could be combined with words. So, I’ve made some visuals. With these, I think working with a good lawyer, we could create a discussion that shows people (a) what we are talking about, (b) why it matters. To optimize the discussion we would need to details of what Mann told the NAS (which I will admit to not knowing.) But I think this could be done.
To start In words: R2 scores matter in Mann’s work because Mann uses things like “tree ring width” to attempt to tease out the earth’s temperature. The R2 score between “tree ring width” and the earths’ temperature “how much” of earths’ temperature might be detectable by looking at a tree relative to “how much” some other unknown factor (e.g. precipitation, sunlight, wind) might be affecting the width of tree rings. If R2 is low, then other factors have a large effect. If it is high, then temperature has a large effect.
Because the meaning of R2 is well known to scientists, engineers and statisticians, they often look at this value when evaluating data. Of course, they tend to resort to math, and start saying things like “R2 is the square of the correlation coefficient, R”. (Likely lay people who read articles with titles like “chocolate found to be good for you!”, will have heard the word “correlation coefficient”, though perhaps not.)
However, lay people don’t often use jargon terms like “R2″, and frequently, they are presented with algebra. But in fact, it’s entirely possible to do this visually. So, let’s do that.
Let’s start by showing a figure that illustrates a relation between “ring width” and “temperature” if it turned out that this ‘R2′ thing has R2=0. That is the “correlation” is exactly zero.
For this primer with big pictures then please read this page.source
Can you see a problem.
And if you want to know if a data set/sets is valid, then R2 must be used to see if the conclusions are robust.
I suggested this would best be done with visuals. Of course these could be combined with words. So, I’ve made some visuals. With these, I think working with a good lawyer, we could create a discussion that shows people (a) what we are talking about, (b) why it matters. To optimize the discussion we would need to details of what Mann told the NAS (which I will admit to not knowing.) But I think this could be done.
To start In words: R2 scores matter in Mann’s work because Mann uses things like “tree ring width” to attempt to tease out the earth’s temperature. The R2 score between “tree ring width” and the earths’ temperature “how much” of earths’ temperature might be detectable by looking at a tree relative to “how much” some other unknown factor (e.g. precipitation, sunlight, wind) might be affecting the width of tree rings. If R2 is low, then other factors have a large effect. If it is high, then temperature has a large effect.
Because the meaning of R2 is well known to scientists, engineers and statisticians, they often look at this value when evaluating data. Of course, they tend to resort to math, and start saying things like “R2 is the square of the correlation coefficient, R”. (Likely lay people who read articles with titles like “chocolate found to be good for you!”, will have heard the word “correlation coefficient”, though perhaps not.)
However, lay people don’t often use jargon terms like “R2″, and frequently, they are presented with algebra. But in fact, it’s entirely possible to do this visually. So, let’s do that.
Let’s start by showing a figure that illustrates a relation between “ring width” and “temperature” if it turned out that this ‘R2′ thing has R2=0. That is the “correlation” is exactly zero.
For this primer with big pictures then please read this page.source