When we think of science, we think of educated men in high tech, state-of-the-art labs conducting experiments that will bring about the discoveries of tomorrow. These scientists spend their days with beakers in hand, carefully pouring various chemicals, and spending countless hours in front of a chalk board searching for the key that will complete a complex mathematical formula. Unfortunately, this picture is inaccurate to describe the science behind global warming. The science behind global warming is not in a lab with chemicals or at a chalk board with formulas, but at a computer conducting regression models based on statistics. Our culture is used to the firm formulas of Einstein and Newton, what does regression modeling mean and what are its limits?
I am not a scientist, although I’d like to think of myself as an amateur one. However, the same modeling used in global warming “science” is the same I’ve used on many occasions to make economic models. In fact, I created one to illustrate how modeling works. My model is how voting for Obama causes unemployment. Below is the mathematical output of my model. The model was created by using the April unemployment rates from the 50 states and comparing that to whether or not that state swung for Obama. For those of you who know something of modeling and want to know a little about my methodology; I set up dummy variables to establish the mathematics needed for the “did the state swing for Obama in the election” variable. States that voted for Obama were assigned the number 1 and states that voted for McCain were assigned the number 0. All of this is statistically sound. However, please note this is an example only. I’m not actually going to try and make the argument that this model actually proves voting for Obama causes unemployment.
First thing is first, how does modeling work? In the case of global warming, variables (average global temperatures and measurements of carbon dioxide in the atmosphere) are placed side by side in a mathematical comparison to determine if there is a similarity between the increases and decreases in these variables. The strength of this relationship is determined by a statistical measurement called R square. It must be noted that the R square for global warming models that I’ve seen tend to be over 60%, which is a good strong relationship in the statistical sense and is part of the reason why the models have gained a lot of support. My model has an R square of 66% (pretty strong relationship). There are other statistical tests that need to be reviewed such as P-Value, F significance, and assumption testing (residual analysis). I won’t go into these because this is a political blog and not a science blog and it would probably put you to sleep. The P-Value and F significance for my model look very good. I did not do a residual analysis.
Regression modeling is not in and of itself science, it is statistics. Modeling is a tool used to help determine relationships between variables and how strong those relationships are. Modeling does not spit out scientific law, but requires scientific analysis to interpret. It is not all that different from one of global warming’s most favorite tool, the thermometer. The thermometer can measure the temperature of a single location, at a particular altitude, at a particular point in time. As a tool, the thermometer has certain limitations and operating requirements that must be followed for accuracy. For example, don’t place your thermometer in direct sunlight if you want an accurate measurement. Like the thermometer, modeling has its own limitations.
As stated above and I reitterate, statistical modeling in and of itself does not establish a cause and effect relationship. You cannot say, “My global warming model has over a 50% R square value, therefore carbon dioxide causes global warming.” I cannot say based on my model that voting for Obama causes unemployment. Instead, it shows that there is some kind of relationship. It is then left to scientists to interpret and reason what that relationship is. The cause and effect relationship must be established from this reasoning. It is in this interpretation and reasoning of the results that muddy the waters of objective science. I’m sure I could think up a couple of theories for explaining how voting for Obama could in fact lead to unemployment. For example, there were many people who voted and supported Obama who thought, “They would no longer have to worry about putting gas in their cars or pay for their mortgage (a direct quote from a zealous supporter).” States that swung for Obama probably had large populations of these types of supporters. Once Obama was elected, they followed through with their misguided beliefs and decided to stop working, since they believed Obama would take care of their every need.
For global warming, we know that carbon dioxide is a green house gas and could be a factor in global temperatures, but what about the variables that affect carbon dioxide? Could the dependant and independent variables be inverted? This would mean that increasing temperatures may be affecting and causing an increase in carbon dioxide. Perhaps, there is a third variable (like the sun), that may be affecting both global temperatures and the amount of carbon dioxide in the atmosphere at the same time. These are just two of thousands of possibilities that would cause problems for the global warming models.
My model argument that explains my model could be threatened by the fact that unemployment was increasing before Obama was elected. Perhaps, the relationship is coincidental. Perhaps, there is another variable that is making it look like there is a direct relationship. For example, Dick Cheney got together with all the corporations that he secretly runs and they conspired to fire millions of people to make Obama look bad. Maybe high levels of unemployment at the time of the election became a leading reason why people voted for Obama?
Another issue is the accuracy of measurement? Most models prefer taking data from a conglomeration of 2,100 temperature stations around the world. The data at these stations are constantly being found to be overstated. For example, last October Russia’s temperatures were the exact same temperatures recorded in September. Is the methodology for determining average global temperature sound? There is a great deal of variation in temperature from one location to the next in both distance and altitude and these could have a large affect on the accuracy of the data.
The model I chose above is a sample of states and does not actually hold water in the real world. If I took all 50 states my R square would be 16%. I decided to try a population and roll the dice to see if I could get a better result with a sample model instead. Since global warming models are using sample data, the methodology for calculating average global temperature needs to be sound or else you will come up with an inaccurate model like I did.
The issues above are problems with the science that should be reported to the public and openly discussed. Doomsday predictions, on the other hand, are entirely dishonest. Using modeling to make predictions 100 years away is not only wrong, but completely outside of the acceptable use and methodology of mathematical modeling. Modeling is useful for making predictions in the short term. However, global warming alarmists are predicting temperatures for nearly more years in total than the number of years used in their data set. Scientists seem to get a pass on this rule, because they are dealing in “science.” I do not agree, there is absolutely no reason or logic that could defend these predictions as being mathematically accurate and in making these predictions, the credibility of their models must be questioned.
There are other possible issues with the science of global warming, but I’ll deal with those in my next two posts on global warming. As I stated in my last post on Science, I’m not trying to debunk or promote global warming. Instead, I’m trying to highlight on the issues that is not being discussed honestly in the media and in some cases being explained by scientists.
