The Decade, 1998 to 2007 - No Global Warming

[intention]

@Tuckerfan

Do you think that if Canada decided that they were going to spend $3 billion to clean up air and water this would come without a bureacratic increase? Even if Canda decides to simply pass new regulations to force businesses to spend $3 billion to do it, there’s going to be a bureacratic increase. Got to hire more inspectors to make sure that companies are in compliance with the new regulations. Got to test and certify new equipment to make sure it does what the manufacturer claims. Of course, a country could just forego all that and take it on the “honor system” that everybody’s doing what they’re supposed to, but if you think that’s a good idea, then I’ve got some Chinese made toothpaste for you.

Tuckerfan, thanks for your reply. Sure, no matter what a government does, it will increase the bureaucracy. The difference is that the Canadian carbon bureaucracy did nothing, the CO2 emissions continued to rise. This is why it is a harm to society, because nothing of value was produced.

@Tuckerfan

It’s a harm to society if the Germans simply roll over and take it (like US automakers did when gas prices spiked in the 1970s). If the Germans react positively, and look at ways to be more efficient, they’ll soon see an economic boom, like the Japanese did in the 1970s, when they shelled out billions of dollars to switch their industry over to more energy efficient equipment. If the Germans successfully switch over to more efficient methods before their competitors do, they’ll have an economic advantage over their competitors as energy costs continue their inevitable rise.

Say what? So the Germans get more energy efficient, and eventually, they finally get back to where their total energy cost is the same as it is today … how does that give them a competitive advantage? Plus, they will still need to recoup the total of the losses they will incur until they finally get back to energy costs equal to today … sounds like harm to me.

In addition, you say that this potential benefit will only happen “(i)f the Germans successfully switch over to more efficient methods before their competitors do” … but many of their competitors (China, Japan, US, Australia) are not suffering the same harm that the Germans are suffering. It’s not like the 1970s when the whole world was hit by higher energy prices and the winners were the ones that could adapt most quickly. It is a problem that is unique to Germany (and to a lesser extent to the other Kyoto signatories), which is a very different situation.

@Tuckerfan

Name one thing that doesn’t have a negative to go along with it’s positive.

Hey, my friend, you were the one who claimed that there was no harm being done by assuming CO2 was a problem. Now you’re changing your tune, and saying there is harm … which is true, but it’s not what you said before.

w.

[Tuckerfan]

@intention

Tuckerfan, thanks for your reply. Sure, no matter what a government does, it will increase the bureaucracy. The difference is that the Canadian carbon bureaucracy did nothing, the CO2 emissions continued to rise. This is why it is a harm to society, because nothing of value was produced.

That, I will submit, is a fault of the Canadian government in how they set up the bureaucracy, properly set up, this situation wouldn’t have happened. (Of course, if the Canadian government is anything like the current US government, it’s a foregone conclusion, they’ll FUBAR it.)

Say what? So the Germans get more energy efficient, and eventually, they finally get back to where their total energy cost is the same as it is today … how does that give them a competitive advantage? Plus, they will still need to recoup the total of the losses they will incur until they finally get back to energy costs equal to today … sounds like harm to me.

I don’t know about German corporations, but I can tell you that I’ve worked for a number of companies here in the US that used positively ancient equipment. How old? Try built in the 1920s! Simply pulling the electric motors off of those things and switching them out with more modern units would have yielded significant energy savings, as well as massive productivity increases. Dumping the machines for something that had been built within the past decade would have yielded even more. IIRC, the energy savings for replacing equipment that’s five years old is enough to pay for the cost of the new equipment within a year or two.

In addition, you say that this potential benefit will only happen “(i)f the Germans successfully switch over to more efficient methods before their competitors do” … but many of their competitors (China, Japan, US, Australia) are not suffering the same harm that the Germans are suffering. It’s not like the 1970s when the whole world was hit by higher energy prices and the winners were the ones that could adapt most quickly. It is a problem that is unique to Germany (and to a lesser extent to the other Kyoto signatories), which is a very different situation.

Only in the short term, however. Both the economies of China and India are growing at a furious pace and their demands for energy are slowly pushing global energy costs higher. (I’ve not seen any estimates which include the demands of the other rising economies as well.) In the fairly near future, each one of them will require the total daily global oil production, simply to maintain their economies, forget about expanding their economies. That alone should be enough reason to start thinking about switching over to more efficient methods and alternative energy sources. Mind you, I’m not a fan of Kyoto, as I think that too many governments will take a “screw the little guy” attitude to meeting the goals of it. I’d be willing to bet that if the Germans had gone a different route towards meeting their targets, they could have done it with less pain.

Hey, my friend, you were the one who claimed that there was no harm being done by assuming CO2 was a problem. Now you’re changing your tune, and saying there is harm … which is true, but it’s not what you said before.

w.

And I should have been more clear. I can’t see any way in which the harm from trying to prevent global warming when there is none can be as great as doing nothing when there is global warming. If we do nothing and there is global warming, we end up boiled alive, if there’s no global warming and we do something, the worst that can happen is economic distruption for a short period of time.

[wevets]

I don’t know how you got that. I should have remembered my warmly enthusiastic elementary school teacher “show all your work!” Let me know what I did differently from you.

Here’s how I arrived at 0.00044 for 1998 to 2007:

(my equations won’t paste from Word, so pardon the text)

b = (sum of products xy) divided by (sum of x[sup]2[/sup])

sum of products (xy) = 73

sum of x[sup]2[/sup] = 1645

73 divided by 1645 = 0.00044

The regression line is

y (hat) = b*x + a

so the slope of the trend is 0.00044

What did you use to get 0.009?

@intention

I note, however, that you agree that you were wrong about how starting from a temperature peak will necessarily result in a decreasing trend …

Yes, I should have done the math first, then posted – my fiery old stats professor would have given me a dollar’s worth of nickel-sized lumps for not doing that! You have to go back to the data set 1998-2004 before you find a negative slope for the trend line of –0.00022. But that’s why it’s great that we’ve done the math – we can see that although the trend for 1998-2007 is not negative, we can demonstrate that the year 1998 did indeed have a dramatic effect on the slope of the trend line:

Slope of the trend line for 1990 to 2006: 0.0058

Slope of the trend line for 1998 to 2007: 0.00044

Slope of the trend line for 1999 to 2005: 0.0013

Slope of the trend line for 1989 to 1996: 0.0047

It’s very difficult for me to look at this list and understand how people in the heat of the debate upthread tried to say that the El Niño of 1997-1998 didn’t have a huge effect on the temperature anomaly trend. It looks to me like the slope of the trend line drops dramatically closer to zero for small data sets that start at 1997-1998 compared to the trend lines for the eight years before and after 1997-1998.

The real point of all this is not that I think these numbers are significant (I’m doing the math just to keep myself in practice.) The real point is that the following argument has been presented:

Argument: Based on a small data set and a statistically insignificant regression, global warming does not exist.

Response: Nonsense, not only was that not the evidence in favor of global warming in the first place, you got to pick a time period that would look favorable.
Is it even still in dispute that 1998-2007 is a time period favorable to finding a minimized trend line in temperature anomalies?

Perhaps the best tag line for this thread would be: Just because you can make a calculation doesn’t mean that calculation means what you think it means – or anything at all! And I’m guilty of that as well, but it keeps me in practice at math and punnery.

[wevets]

I missed my edit window, but the above sum of products (xy) should be:

0.73, not 73

As Gerald Ford once said “It’s OK, nobody’s human.”

[intention]

wevets, thanks for showing your work. You have made an error in part of the calculations. You say:

b = (sum of products xy) divided by (sum of x2)

I assume this is shorthand for the actual formula, which is:

b = (sum of (x - mean x)*(y-mean y) divided by (sum of (x - mean x) ^2)

Following your commendable suggestion to show all of the work, here’s the full calculation:

 Mean x	       2002.5					
 Mean y	         0.24					
						
x	   x - mean x	   (x - mean x)^2	      y	   y - mean y	   (x-mean x)(y-mean y)	
1998   	         -4.5	            20.25	    0.5	         0.26	                  -1.17	
1999   	         -3.5	            12.25	   0.03	        -0.21	                  0.735	
2000   	         -2.5	             6.25	   0.02	        -0.22	                   0.55	
2001   	         -1.5	             2.25	   0.19	        -0.05	                  0.075	
2002   	         -0.5	             0.25	    0.3	         0.06	                  -0.03	
2003   	          0.5	             0.25	   0.26	         0.02	                   0.01	
2004   	          1.5	             2.25	   0.19	        -0.05	                 -0.075	
2005   	          2.5	             6.25	   0.33	         0.09	                  0.225	
2006   	          3.5	            12.25	   0.28	         0.04	                   0.14	
2007   	          4.5	            20.25	    0.3	         0.06	                   0.27	
Sums   	             	             82.5	       	             	                   0.73	

The slope of the line is then 0.73/82.5, or .009.

You are entirely correct that this decadal record does not show that the globe is not warming, or that it is warming, since the trend is not statistically significant.

w.

[Half_Man_Half_Wit]

ClimateGuy, you seem to be subscribed to the idea that there isn’t any man-made global warming; so maybe you could answer a question I’ve posed in this thread, namely: how is the increase in the atmospheric CO2 concentration supposed to not warm the earth?

[wevets]

@intention

wevets, thanks for showing your work. You have made an error in part of the calculations. You say:

b = (sum of products xy) divided by (sum of x2)

I assume this is shorthand for the actual formula, which is:

b = (sum of (x - mean x)*(y-mean y) divided by (sum of (x - mean x) ^2)

My warmest regards, intention, it’s no problem - I can see now what’s going on. That’s not shorthand. You’re doing a Model II regression, while I’m doing a Model I regression. If I’d screwed up thinking (sum of products xy = sum of (x - mean x)*(y-mean y)) I’d be in hotter water than in a teapot inside a deep-fried turducken.

The key difference can be seen when you include the term

(sum of (x - mean x)

You’re calculating the variance of x (years in this data set) around its mean to construct a best-fit line that minimizes variation in both x and y. However, this is not necessary since there is no measurement variability in x - we know exactly which year each measurement belongs in, and there is no random variability in x: 1997 always follows 1996, which always follows 1995.

Check here for some more on Model I and Model II regressions:

@http://udel.edu/~mcdonald/statregression.html

Both of the examples above are Model I regressions. In a Model I regression, the assumption is made that the X variable is controlled by the investigator, or “fixed,” and is measured without error. Only the Y variables have natural variation or are measured with error. Of course, no continuous variable is ever known without any error, but as long as the error in the X variable is small relative to the range of the X variables, a Model I regression can be used.
In a Model II regression, both the X and Y variables have natural variation or are measured with error. For most such variables, correlation seems more appropriate to me than a Model II regression, so we will not discuss Model II regressions any further; when I say “regression,” you can assume I mean a Model I regression.

So where does this leave us with respect to 1998-2007? Would you agree that it is favorable for finding a minimized trend line?

[intention]

@wevets

My warmest regards, intention, it’s no problem - I can see now what’s going on. That’s not shorthand. You’re doing a Model II regression, while I’m doing a Model I regression. If I’d screwed up thinking (sum of products xy = sum of (x - mean x)*(y-mean y)) I’d be in hotter water than in a teapot inside a deep-fried turducken.

Busted out laughing at that one, very good.

@wevets

… You’re calculating the variance of x (years in this data set) around its mean to construct a best-fit line that minimizes variation in both x and y. However, this is not necessary since there is no measurement variability in x - we know exactly which year each measurement belongs in, and there is no random variability in x: 1997 always follows 1996, which always follows 1995.

Check here for some more on Model I and Model II regressions:

Thanks, wevets, I hadn’t thought about that. However, having done so, I respectfully disagree that there is no error in the X variable. Since it is an average, rather than an instantaneous measurement, it seems to me that there is an error of up to half a year, and thus Model II regression is the appropriate model. I’d appreciate your thoughts on this.

@wevets

So where does this leave us with respect to 1998-2007? Would you agree that it is favorable for finding a minimized trend line?

Not sure what you mean by a “minimized trend line” …

All the best,

w.

[wevets]

@intention

Thanks, wevets, I hadn’t thought about that. However, having done so, I respectfully disagree that there is no error in the X variable. Since it is an average, rather than an instantaneous measurement, it seems to me that there is an error of up to half a year, and thus Model II regression is the appropriate model. I’d appreciate your thoughts on this.

I hadn’t thought about that perspective. In this case it’s a little complex - we’re working with a data set that is the average of the monthly data the OP linked to, which itself is likely the average of daily data.

From The Probable Error of a Mean

@Student (a.k.a. William Gosset)

If the number of experiments be very large, we may have precise information as to
the value of the mean, but if our sample be small, we have two sources of uncertainty: (1) owing to the “error of random sampling” the mean of our series of experiments deviates more or less widely from the mean of the population, and (2) the sample is not sufficiently large to determine what is the law of distribution of individuals. It is usual, however, to assume a normal distribution, because, in a very large number of cases, this gives an approximation so close that a small sample will give no real information as to the manner in which the population deviates from normality: since some law of distribution must he assumed it is better to work with a curve whose area and ordinates are tabled, and whose properties are well known.

In this case it seems to me that the random sampling (1) isn’t too much of an issue - the sampling is regular and evenly distributed in time (x), and the law of distribution of individuals (2) in X is also known. We can’t assume normality, however, because we know the X-axis is evenly distributed.

But I can see that we might not be in agreement on this, and rather than engage in fiery debate over the assumptions of a test, let’s do both tests. After all, if you build a man a fire, he’ll be warm for a night, while if you set him on fire, he’s warm for the rest of his life.

We can do Model II regressions on the periods 1989 to 1996, and 1999 to 2007, to check and see if the slopes have that rewarmed leftovers feeling - a similar relationship to the Model I regressions. I don’t have time to do them today, but if you want I could do them on Friday, or you could do them.

Not sure what you mean by a “minimized trend line” …

Sorry I wasn’t clear - I just mean that if you have an interest in finding a trend line whose slope is close to zero (minimize the slope of the trend line), then 1998 is an excellent starting point, since it will give you smaller slopes than starting in other recent years.

[intention]

wevets, many thanks for your thoughts. The problem with comparing the slopes from 1989-1996 and 1999-2007 is that they are not statistically different from each other, and neither of them is statistically different from zero … doesn’t leave much that we can say about one, the other, or both.

However, your point is well taken, that if you are looking for a flat trend you should start from a high spot, and vice versa if you are looking for a steep trend. Me, I don’t like trend lines much, nature doesn’t move in straight lines. I find Gaussian averages much more informative.

My best to you,

w.

[wevets]

@intention

wevets, many thanks for your thoughts.

And thank you for encouraging me to check my assumptions and math.

The problem with comparing the slopes from 1989-1996 and 1999-2007 is that they are not statistically different from each other, and neither of them is statistically different from zero … doesn’t leave much that we can say about one, the other, or both.

I disagree - the essence of what we can say does not relate to global warming, but it does relate to the methodology of evidence used by the OP. Succinctly put, that the method’s like a toad offered a job by the three sisters in MacBeth - cooked!

As far as I can tell the slopes of regression (Model II) lines on the temperature anomalies:

1989-1996: 0.01 …small but we can do better
1998-2007: 0.0088 … ooh, nice and tiny, within spitting distance of zero
1999-2007: 0.034 …several times bigger, let’s coast down that El Niño hill instead

As you and I both know, none of these are statistically significant, and none were offered as evidence of global warming by climate scientists anyway. But if one is confident enough, it might sound as though it were, especially if none of us had been versed in statistics.

However, your point is well taken, that if you are looking for a flat trend you should start from a high spot, and vice versa if you are looking for a steep trend.

I’m glad we’re in agreement on this - it’s the essence of what I was trying to say earlier before I checked the math. We’re thrown a lot of barrels on the internet, and I’m just trying to figure out if they’re the happy monkey-filled kind or tossed by Donkey Kong.