View Single Post
Old 04-11-2013, 02:26 PM   #493  
Winged Wheeler
Libertarian Resistance
 
Winged Wheeler's Avatar
 
Joined APC: Feb 2007
Position: 757 FO
Posts: 992
Default

Quote:
Originally Posted by JamesNoBrakes View Post
No offense, but that's just as bad as the science you are trying to discredit. Being within 95% is the accepted gold standard, and you have no idea if it will stay there or not, but so far they are correct. There are much more advanced quantitative methods that can be used to quantify and explain data, which I assume is what the scientist do, but you're not really helping your cause.
JamesNoBrakes,

I thought a bit more about this issue that you've raised.

Gallup, and other political pollsters, construct polls to a 95% confidence interval (CI) and they claim accuracy to +/- 3 or 4 or 5%.

On the image about which we've had this discussion the model projected temperature rise for 2015 is about .7C. If the model was as accurate as a Gallup poll (+/-5%) the 95% CI would range from about .66 to .74; instead the range is, roughly, about .4 to .8. That seems very large--no one would be impressed by a political poll if the range was, by percentage, that big.

This may comply with the gold standard for climate science but I don't think we should be making consequential economic decisions based on predictions as imprecise as these.

The width of the confidence interval gives us some idea about how uncertain we are about the unknown parameter. A very wide interval may indicate that more data should be collected before anything very definite can be said about the parameter.

Statistics Glossary - confidence intervals
Winged Wheeler is offline