He think Stuart is factually wrong and the global warming question isn’t a good predictor. Fortunately that’s something we can test.
Before we run the numbers, what’s your confidence interval for the IQ difference in the LessWrong poll of 2012 between on the people who believe that p(global warming)>0.9 versus people p(global warming)<0.5?
If you just correlate p values with IQ, what’s your confidence interval for the resulting correlation coefficient?
As IQ might not be rationality, how do you think the global warming answer will predict whether someone gives rational answers to the CFAR questions?
My 90% confidence interval for the correlation between IQ and p(global warming) is orgjrra ebhtuyl artngvir mreb cbvag bar naq cbfvgvir mreb cbvag gjb, jvgu n crnx pybfr gb mreb. V’q or yrff fhecevfrq gb frr n pbeeryngvba orgjrra c(tybony jnezvat) naq gur PSNE dhrfgvbaf (gubhtu V’q whfg hfr 5-7, nf gur bguref frrz gb unir zber cbgragvny pbasbhaqref), ohg V’q fgvyy rkcrpg dhvgr n ybj bar.
You need to specify the “global warming” part better. “The global climate has warmed since the beginning of the XX century” is a different claim from “Human emissions of CO2 caused the warming of the global climate” which is a different claim from “The current warming is unprecedented in known history” which is a different claim from “We need to reduce the CO2 emissions”.
In this post I intend to reference the Lesswrong census. In it the question was worded:
P(Warming) What is the probability that significant global warming is occurring or will soon occur, and is primarily caused by human actions?
Hopefully we will have another census this year. If you think that there is better question to get at the hard core of the global warming issue, I also invite you to make prediction about how such a question would correlate.
The question could be added to the next poll and we could then see how the results of the question correlate.
The way the question was worded it asked two different questions (maybe even three) and I’m not sure the respondents treated it as a logical expression along the lines of is.true((A OR B) AND C)...
I don’t know what do you mean by the “hard core of the global warming issue”.
The way the question was worded it asked two different questions (maybe even three) and I’m not sure the respondents treated it as a logical expression along the lines of is.true((A OR B) AND C)...
That would probably correlate with rationality too.
I’m not responsible for the question being worded the way it is. I don’t think the wording is optimal.
If you think the question gets interpreted by different people in a different way, propose a better question to measure global warming beliefs for the next census.
Whether you are responsible or not is distinct from whether it will do a good job measuring what you want it to measure.
Responsibility changes the meaning of the word ‘good’. If I design something to measure Y I have a higher standard for ‘good’ than when I search for an already existing measure of Y.
If people who read the post say: “I don’t think IQ correlates with the answer of that question” that an answer that moves the discussion forward.
If they say: “I think IQ correlates with the answer to a differently worded question about global warming” that also moves the discussion forward. We can test that hypothesis in the next census.
If you don’t like IQ as proxy than we had the CFAR questions in the last census to measure rationality. They are also not perfect and we can think up a better metric for the next census.
For that matter, “I estimate human emissions of CO2 caused 49% of the warming of the global climate” is a different question from “I estimate human emissions of CO2 caused 51% of the warming of the global climate”. Is it really a fantastic expression of rationality to say that people making the first claim are basically creationists, but people making the second claim are upstanding rationalists whose numbers help to demonstrate how much popular support I have?
If you try to lump people into discrete categories over a continuously varying question then you are inherently introducing ambiguity; the first step toward setting up a Worst Argument in the World is the creation of overly-broad categories, after all. If you demand that Turquoise people self-identify as Blues or Greens, you shouldn’t be surprised when you get suspected of having motives other than the pure refinement of rational thought.
Scientists have already found p(null hypothesis) < 0.05 on AGW. It’s time we stopped variations of probability estimates over nuanced versions of possible positions and accepted the proposition supported by statistically significant evidence and a consensus of experts behind that evidence.
(Side note: Yes, I know I just blasphemed against the Great God Bayes by invoking frequentist statistics. Too bad.)
He think Stuart is factually wrong and the global warming question isn’t a good predictor. Fortunately that’s something we can test.
Before we run the numbers, what’s your confidence interval for the IQ difference in the LessWrong poll of 2012 between on the people who believe that p(global warming)>0.9 versus people p(global warming)<0.5?
If you just correlate p values with IQ, what’s your confidence interval for the resulting correlation coefficient?
As IQ might not be rationality, how do you think the global warming answer will predict whether someone gives rational answers to the CFAR questions?
I’ll bite.
My 90% confidence interval for the correlation between IQ and p(global warming) is orgjrra ebhtuyl artngvir mreb cbvag bar naq cbfvgvir mreb cbvag gjb, jvgu n crnx pybfr gb mreb. V’q or yrff fhecevfrq gb frr n pbeeryngvba orgjrra c(tybony jnezvat) naq gur PSNE dhrfgvbaf (gubhtu V’q whfg hfr 5-7, nf gur bguref frrz gb unir zber cbgragvny pbasbhaqref), ohg V’q fgvyy rkcrpg dhvgr n ybj bar.
(ROT13ed to avoid anchoring future readers.)
You need to specify the “global warming” part better. “The global climate has warmed since the beginning of the XX century” is a different claim from “Human emissions of CO2 caused the warming of the global climate” which is a different claim from “The current warming is unprecedented in known history” which is a different claim from “We need to reduce the CO2 emissions”.
In this post I intend to reference the Lesswrong census. In it the question was worded:
Hopefully we will have another census this year. If you think that there is better question to get at the hard core of the global warming issue, I also invite you to make prediction about how such a question would correlate. The question could be added to the next poll and we could then see how the results of the question correlate.
The way the question was worded it asked two different questions (maybe even three) and I’m not sure the respondents treated it as a logical expression along the lines of is.true((A OR B) AND C)...
I don’t know what do you mean by the “hard core of the global warming issue”.
That would probably correlate with rationality too.
I’m not responsible for the question being worded the way it is. I don’t think the wording is optimal.
If you think the question gets interpreted by different people in a different way, propose a better question to measure global warming beliefs for the next census.
The first question is what is it that you want to measure.
Whether you are responsible or not is distinct from whether it will do a good job measuring what you want it to measure.
Responsibility changes the meaning of the word ‘good’. If I design something to measure Y I have a higher standard for ‘good’ than when I search for an already existing measure of Y.
If people who read the post say: “I don’t think IQ correlates with the answer of that question” that an answer that moves the discussion forward.
If they say: “I think IQ correlates with the answer to a differently worded question about global warming” that also moves the discussion forward. We can test that hypothesis in the next census.
If you don’t like IQ as proxy than we had the CFAR questions in the last census to measure rationality. They are also not perfect and we can think up a better metric for the next census.
For that matter, “I estimate human emissions of CO2 caused 49% of the warming of the global climate” is a different question from “I estimate human emissions of CO2 caused 51% of the warming of the global climate”. Is it really a fantastic expression of rationality to say that people making the first claim are basically creationists, but people making the second claim are upstanding rationalists whose numbers help to demonstrate how much popular support I have?
If you try to lump people into discrete categories over a continuously varying question then you are inherently introducing ambiguity; the first step toward setting up a Worst Argument in the World is the creation of overly-broad categories, after all. If you demand that Turquoise people self-identify as Blues or Greens, you shouldn’t be surprised when you get suspected of having motives other than the pure refinement of rational thought.
Well you can probably say that anyone who thinks humans are entirely responsible, or not responsible at all is irrational on that question.
Scientists have already found p(null hypothesis) < 0.05 on AGW. It’s time we stopped variations of probability estimates over nuanced versions of possible positions and accepted the proposition supported by statistically significant evidence and a consensus of experts behind that evidence.
(Side note: Yes, I know I just blasphemed against the Great God Bayes by invoking frequentist statistics. Too bad.)