The Holocene Deniers

[grmorton]

Interesting article on the low number of sunspots and the fact that it should cool the earth--contra the claims of IPCC

http://www.cjonline.com/news/local/2009-09-20/earth_approaching_sunspot_records

The interesting thing I notice in the sun as a possible cause debate is that the solar scientists in general think the sun plays a larger role in warming than do the climatologists.

On the JPL Acrim site most of the articles referenced are critical of the standard view of global warming--

Graphics Gallery

And here is a video of a climatologist critical of the normal view of global warming--not that this will influence the believers

Al Fin: Global Warming Lecture Video John Christy UAH

 

[grmorton]

Had dinner with a friend so I just got home. I will try to get to the posts tomorrow. I thought I would post a picture I just put up on my blog. It shows a seasonal bias 2000-2008 for the temperature difference between Montevideo, MN and Milan MN, two towns only 16 miles apart. This is clearly a problem. If the thermometer record were actually working, we shouldn't see this kind of pattern.

A bigger version of this can be found at weatherMNMontevideoMilan180dave2000-2008.jpg (image)

More than a degree difference in the delta-temperature between two towns 16 miles apart is more than should be the case. And you all beleive that the temperature record is OK. What a laugh.
http://3.bp.blogspot.com/_Lxqre8hMG.../weatherMNMontevideoMilan180dave2000-2008.jpg

 

[AV1611VET]

More than a degree difference in the delta-temperature between two towns 16 miles apart is more than should be the case.

The closest I come to global warming, bro, is thinking maybe these holes in the Ozone layer everyone is taught to be concerned about is actually the 'windows of heaven' mentioned in Genesis 7.

I can't prove it, of course --- just a thought.

God bless!

 

[grmorton]

grmorton,

On how you can determine whether a drought is caused by global warming, the BOM are in the process of working this out right now for the drought in south-eastern Australia.

Study links drought with rising emissions

Note that there is still disagreement on this issue.

Well, the models presented by Latif at the WCC showed that they can't get the rainfall correct with these models. So, why are you trusting that the model showing what you want it to be, a drougt/warming connection, is correct???

And I didn't ask you about a study, I asked you what criterion you would use to tell that a particular drought is from global warming. Australia has had very bad droughts in the past, Drought in Australia a natural phenomenon

However, the models are matching what is happening in the real world - lower rainfall in south-eastern Australia, higher rainfall in northern Australia. I have spoken via email to Dr Timbal about some of this - it is very interesting stuff.

And models elsewhere don't work and don't predict correctly. How do you know this isn't just a blind pig finds an acorn, especially since there were huge droughts in the past when there was no global warming?

Are you seriously saying that the model is the only way to do it?

On causation of events by global warming, it is always going to be tricky. For example, even in a period of reduced average rainfall, there are still going to be periods of high rainfall. If we examine the recent Victorian bushfires, the claim that global warming caused them is not correct. After all, many of them were caused by arsonists. Their intensity was caused by the dryness of the bush, the very high ambient temperatures on the day and high winds. The dryness of the bush was caused by a long period with below average rainfall - part of which was likely caused by climate change. So: were the Victorian fires caused by climate change? No. But it was one causal factor that influenced the way the event panned out.

The real problem that many advocates of global warming have (maybe not you) is that everything is caused by global warming- Cooling is caused by global warming, a bad hurricane is caused by global warming, heck, athletes foot is probably caused by global warming.

When something is the universal cause it is no cause at all.

 

[grmorton]

Note that Hansen is talking about urban and periurban (urban fringe - I had to look that up!) areas, while the page I linked to is talking about the average over the entire dataset. It is likely that there was no adjustment required for rural areas due to the UHI effect, after all. Thus, when averaged over the entire dataset, .1 degrees farenheit is not an inconsistent result with the numbers that Hansen is talking about.

I laughed as you said you looked up peri-urban. I took Greek in college and about the only benefit I have seen to my life is that I remembered that peri in Greek means around. A whole semester of Greek and that is all I have to show for it. Otherwise I too would have been at the dictionary.

As to you not finding the explanations of the adjustments reasonable, fair enough. But I do. Hence, the basis for our disagreement over the data record becomes a disagreement over this particular page on the internet. Interesting.

Ok, then answer this. If the procedures and thermometers of 1900 require NO correction, why don't we go back to those methodologies and standards so that we can have an error free record and save some of our tax dollars by laying off some of the data correctors that wouldn't be needed if we didn't have such bad thermometers today?

What that chart says is that 1900 had better technology than we have today. I don't beleive that.

 

[grmorton]

grmorton,

I have graphed the keeling curve versus temperature increase since 1958 using multiple datasets. All of the graphs show that the temperature increase due to a doubling of CO2 falls well inside the predicted range. You still have not told me what numbers that you got for a doubling of CO2. Can you please do so?

You can find my math here The Migrant Mind: A Backward Look at IPCC predictions of Temperature Rise

 

[grmorton]

The guy who commented on your blog about finding that the curve you plotted gives a temperature sensitivity of around 2.9 degrees per doubling is pretty much spot on.

I disagree and I beleive I replied. It seems to me that you have a criterion that any support for global warming is automatically correct and any criticism of global warming is automatically wrong. That is what young-earth creationists do.

I have asked this of others and get no reply. If you want to measure the temperature in your house for the air conditioner thermostat, would you put the thermometer next to your stove?

If you wanted to measure your child's fever temperature, would you put the thermometer next to a match before reading it for the doctor?

And do you think the picture I posted last night shows that the temperature is being measured properly?

Here it is again

Maybe tonight I will post 1968-1975 for the same two towns only 16 miles apart.

The danger for science that I see from crap like the above is that people will stop believing science. There is no way one can say with a straight face (scientifically speaking) that two towns only 16 miles apart should have temperature differences like this. Yet scientists want to defend this crap, which causes them to lose their credibility. Credibility is the only thing any of us have.

 

[thaumaturgy]

A bigger version of this can be found at weatherMNMontevideoMilan180dave2000-2008.jpg (image)

More than a degree difference in the delta-temperature between two towns 16 miles apart is more than should be the case. And you all beleive that the temperature record is OK. What a laugh.

So how long did it take of going over the data to find a consistent average bias of -1.5[sup]o[/sup]F?

Now the thing I don't see here (and I'd hope you'd look up, I tried and those STATION HISTORY files are very complicated to parse through). A couple questions:

Is there a TOB (time of observation) difference in play here? Has it been homogenized to account for known differences that can come from time of observation? (You can find the USHCN station histories on line).

But just going off the raw data here I, too, plotted the 2000-2008 differences. But I went one step further and plotted daily differences in the MIN and MAX temperatures as well as AVERAGE:

You should be happy, Glenn, these are reasonably or pretty close to normally distributed! We can work with the means rather than just the medians.

Now, yes, there is a mean difference in T[sub]ave[/sub] of -1.5degF for these 8 years. But it is reasonably normally distributed around that, which means that sometimes there's a positive bias!

Same thing for the MINs and MAXs.

So, does it mean that maybe one station is sited "cooler" than the other or the other one is "hotter"? Well, there's definitely something. But -1.5[sup]o[/sup]F? Really?

Do you think this error will propogate significantly in a gridded average?

I would be very interested in seeing the station history for these two stations. But I'll have to wait to download them.

(By the way: so now we assume that there must be less than 1.5degF difference between station readings in your assessment in order for the data to be "acceptable"?)

 

[thaumaturgy]

Quick question: Glenn, why do you prefer the 180day running average to the raw daily data?

The following is a comparison between the data presented as "raw" data and as a 180 day running "average" data (Calculated as the following 180 days averaged from any given point).

The average bias is still about the same at -1.53, but when you look at the raw (unfiltered) data you can see the fact that there are data points on both sides of zero.


Just curious.

 

[grmorton]

So how long did it take of going over the data to find a consistent average bias of -1.5[sup]o[/sup]F?

Now the thing I don't see here (and I'd hope you'd look up, I tried and those STATION HISTORY files are very complicated to parse through). A couple questions:

Is there a TOB (time of observation) difference in play here? Has it been homogenized to account for known differences that can come from time of observation? (You can find the USHCN station histories on line).

Yeah those files are hard to work with. I use the raw observation. The reason I don't like the tobs is that it is a midnight correction.

"Next, the temperature data are adjusted for the time-of-observation bias (Karl, et al. 1986) which occurs when observing times are changed from midnight to some time earlier in the day. The TOB is the first of several adjustments. The ending time of the 24 hour climatological day varies from station to station and/or over a period of years at a given station. The TOB introduces a non climatic bias into the monthly means. The TOB software is an empirical model used to estimate the time of observation biases associated with different observation schedules and the routine computes the TOB with respect to daily readings taken at midnight." GHCN Global Gridded Data

Why do I not believe it? Because as the picture below shows, once again there is little Tobs correction to the values in 1900.

http://cdiac.ornl.gov/epubs/ndp/ushcn/ts.ushcn_anom25_diffs_pg.gif

Wow, does this mean that the people in 1900 got up at midnight to read the stations, or read them at 10 pm so the Tobs correction is small? I doubt that very seriously. But, the Tobs correction is selectively applied in a positive fashion only to stations after 1970--and guess what? That is about the time that the temperature rise took off.

But just going off the raw data here I, too, plotted the 2000-2008 differences. But I went one step further and plotted daily differences in the MIN and MAX temperatures as well as AVERAGE:

You should be happy, Glenn, these are reasonably or pretty close to normally distributed! We can work with the means rather than just the medians.

Now, yes, there is a mean difference in T[sub]ave[/sub] of -1.5degF for these 8 years. But it is reasonably normally distributed around that, which means that sometimes there's a positive bias!

Same thing for the MINs and MAXs.

So, does it mean that maybe one station is sited "cooler" than the other or the other one is "hotter"? Well, there's definitely something. But -1.5[sup]o[/sup]F? Really?

Do you think this error will propogate significantly in a gridded average?

I would be very interested in seeing the station history for these two stations. But I'll have to wait to download them.

We still might quibble a bit about how gaussian the distribution is, but lets set that aside. Once again you ignore PHYSICS! We shouldn't see this kind of seasonal bias if we are measuring the temperature of both towns correctly.

Now, the overall bias also is variable.For 2000-2008 Montevideo is hotter than Milan, but for 1968-75 it is the opposite. It is exactly the reverse of the 2000-2008 trend. The stations reverse, but not by the same amount.

(By the way: so now we assume that there must be less than 1.5degF difference between station readings in your assessment in order for the data to be "acceptable"?)

No, the actual spread is from +1.5 to around - 1.5 deg F--over the history of the towns we have had quite a variation in the temperature differences. And remember it isn't the distribution of the temperature difference that is important. The temperatures should be the same, or so dang close to it that it is not often different. Yet we find 2.5 deg variation throughout a year, if we look over a long enough period.

Tomorrow night I will post the average of the daily temperature difference for each day between these two towns. The seasonality is there and that means that we are measuring something other than the earth's temperature. We are measuring whatever that heat source is.

Yet that doesn't bother you. You only think that distribution is important. It isn't. We shouldn't HAVE a variation of this size if we are actually measuring what we say we are measuring.

By the way Thau, you have consistently avoided my questions.

Would you take your child's temperature and put the thermometer next to a match or the stove before you read it for the doctor?

Would you, if you wanted to know the temperature inside your house, put the thermometer next to the stove?

If so, why? If not so, why? Answer the darn questions.

 

[grmorton]

Quick question: Glenn, why do you prefer the 180day running average to the raw daily data?

Because as you should know from your statistics, an average removes the noise--it averages it out. We use this technique when we have noisy seismic data to enhance the signal and make it stand out.

The following is a comparison between the data presented as "raw" data and as a 180 day running "average" data (Calculated as the following 180 days averaged from any given point).

The average bias is still about the same at -1.53, but when you look at the raw (unfiltered) data you can see the fact that there are data points on both sides of zero.


Just curious.

It is physics that led me to use the 180 day running average. Since heaters /air conditioners are running at different times, the 180 day average separtes the winter from the summer. That is why it is important to pay attention to the physics of the situation. Using a 365 day will average out the seasonal difference.

You can see the periodicity in the raw data that you post and I saw it there as well but the 180 day average will separate the 6 warm months from the 6 cold months and smooth away the high frequency noise. I first saw the variation in the raw data of Brookhaven City MS and Monticello MS, and it was there that I realized how stupid I had been not to think of the 180 day periodicity in the temperature differences before.

 

[grmorton]

I was able to do the average daily temperature difference for each day from 1894-2008 tonight so I thought I would post it. This is for Montevideo and Milan MN. Clearly there is a stronger temperature difference in the winters than in the summers for the entire period of the record. The picture below is from Jan 1 to Dec 31 and each day's temperature difference is included in the total average from 1894-2008.

The red curve is a 30 day running average. One might think that this bias could be used to correct the temperature record--subtract out the bias, but it won't do it because it will leave long periods of time where the two towns don't give anything like a similar temperature. See my blog tomorrow night.

 

[thaumaturgy]

We still might quibble a bit about how gaussian the distribution is,

The normal quantile plot looked like it was within the bounds, but I just ran a shapiro-wilkes test here at home and it showed a rather low p-value. I'll double check on JMP tomorrow.

but lets set that aside. Once again you ignore PHYSICS! We shouldn't see this kind of seasonal bias if we are measuring the temperature of both towns correctly.

Measuring correctly? An average bias of less than 2 degrees. And that is sufficient to call the whole enterprise off?

Yet that doesn't bother you. You only think that distribution is important. It isn't.

The distribution gives me both an idea of the relative error and how the error is distributed. That way I'm not swayed by gut feelings based on appearances.

By the way Thau, you have consistently avoided my questions.

Would you take your child's temperature and put the thermometer next to a match or the stove before you read it for the doctor?

I've dealt with the air conditioner temperature. I'm not going to slavishly answer every variant of this question you ask.

Go back and re-read the posts. Post #585.

Would you, if you wanted to know the temperature inside your house, put the thermometer next to the stove?

Isn't it interesting that you use a 180day running average to "reduce noise" in the data when it suits you but you won't allow the climatologists to grid-average the data?

I find that fascinating.

So you can average and "reduce noise" to your heart's content but no one else can? How does that work?

If so, why? If not so, why? Answer the darn questions.

I answered the 8% heat source question using mathematics several pages ago. (Post #585)

 

[grmorton]

Maybe this analogy can make people understand why I think Thau's treatment of the statistical distribution is irrelevant.

Let's say you need to move a large chair through a narrow doorway. The chair, you see is about 3 feet wide but so is the door, this from visual observation. You send your stupid brother-in-law to measure the width of the chair. He has an 8 ft tape-measure. You tell him to make 100 measurements of the chair's width.

He goes off, you do whatever you are doing and then he comes back with a perfectly gaussian set of width measurements. But when you look at the measurements, you see that he has some 6 ft widths, some 1 ft widths as measures of the chair's width. You look some more. Then you see that on a few outlier measurements he has said that the chair is 20 feet wide and others that say the chair is 2 inches wide.

You are stunned by this. You would expect that he could do a more precise job. You expected that all the measurements should be within 1 inch of each other. But your silly brother-in-law has made measurements which diverge by feet--several times the width of the chair. Clearly he was doing something wrong in the measurement.

The measurements which 1 ft or a couple of inches equally are unbelievable measurements. How in the heck can an approximately 3 ft wide chair measure 2 inches wide??? It can't.

You challenge the brother-in-law and say that the measurements are wrong. He goes away mad, saying that all is OK and that you are stupid to doubt the data. The brother-in-law comes back with a chart showing how his measurements are perfectly, and I mean PERFECTLY gaussian. Eveything is OK, he proclaims.

You object that gaussianicity doesn't matter. It is irrelevant because the measurements are not correct. You shouldn't have such a spread in measured widths of the chair. Your brother-in-law tells you how stupid you are and he has lots of his friends tell you the same thing. They also accuse you of being mean and unreasonable when you doubt what the brother-in-law says.

What do you think? Is the brother-in-law right?

Two towns 16 miles apart should hardly ever have temperatures greater than 10 deg F.

 

[Thistlethorn]

Maybe this analogy can make people understand why I think Thau's treatment of the statistical distribution is irrelevant.

Let's say you need to move a large chair through a narrow doorway. The chair, you see is about 3 feet wide but so is the door, this from visual observation. You send your stupid brother-in-law to measure the width of the chair. He has an 8 ft tape-measure. You tell him to make 100 measurements of the chair's width.

He goes off, you do whatever you are doing and then he comes back with a perfectly gaussian set of width measurements. But when you look at the measurements, you see that he has some 6 ft widths, some 1 ft widths as measures of the chair's width. You look some more. Then you see that on a few outlier measurements he has said that the chair is 20 feet wide and others that say the chair is 2 inches wide.

You are stunned by this. You would expect that he could do a more precise job. You expected that all the measurements should be within 1 inch of each other. But your silly brother-in-law has made measurements which diverge by feet--several times the width of the chair. Clearly he was doing something wrong in the measurement.

The measurements which 1 ft or a couple of inches equally are unbelievable measurements. How in the heck can an approximately 3 ft wide chair measure 2 inches wide??? It can't.

You challenge the brother-in-law and say that the measurements are wrong. He goes away mad, saying that all is OK and that you are stupid to doubt the data. The brother-in-law comes back with a chart showing how his measurements are perfectly, and I mean PERFECTLY gaussian. Eveything is OK, he proclaims.

You object that gaussianicity doesn't matter. It is irrelevant because the measurements are not correct. You shouldn't have such a spread in measured widths of the chair. Your brother-in-law tells you how stupid you are and he has lots of his friends tell you the same thing. They also accuse you of being mean and unreasonable when you doubt what the brother-in-law says.

What do you think? Is the brother-in-law right?

Two towns 16 miles apart should hardly ever have temperatures greater than 10 deg F.

Ah, the data is incorrect because you can see that it is obviously incorrect, thus we should mistrust statistical analysis of the data and instead trust that you know what's best. Sorry, no deal. You aren't smarter than all the climatologists, only more arrogant. The climate isn't a chair, nor is the situation in any way comparable to the one in your analogy.

Statistical treatment, like the one climatologists use, is imperative to achieving a proper understanding of what's going on. Your problem is that said statistical treatment shows that you and your politically driven ideas are wrong, so you'd rather not be confronted with it. Well, sucks to be you.

About the two towns, please get a clue. You have been told why it's irrelevant to the bigger picture. Drop it or expect to be taken a lot less seriously (if possible).

 

[grmorton]

Isn't it interesting that you use a 180day running average to "reduce noise" in the data when it suits you but you won't allow the climatologists to grid-average the data?

I don't recall saying that the climatologists can't grid the data. Can you document where I said that? I might be wrong--I have slept several times during my two periods on this list. But I demand evidence, not your bare assertion.


And you still haven't answered the questions I asked you about the placement of a thermometer next to the stove which thermometer was supposed to measure the temperature inside your house. Let's make it your thermostat. Would you put your thermostat for your heater or air conditioning next to the stove? If not why not. If so why so?

You dare not answer this. You are a coward when it comes to these questions.

Would you measure your child's fever temperature and then put the thermometer on the stove next to the flame before reading it for your doctor? if not why not? If so why so?

Coward. You won't answer these questions. Coward.

As to your supposed mathematical answer to my question about 8% of the stations being heated, you really didn't use math there. You merely asserted that it wasn't a problem.

 

[thaumaturgy]

Maybe this analogy can make people understand why I think Thau's treatment of the statistical distribution is irrelevant.

You spend waaay too much time critiquing stats without directly using mathematics. (You could have done a nice calculation on this example, but you didnt'.)
I find that interesting.

Let's say you need to move a large chair through a narrow doorway. The chair, you see is about 3 feet wide but so is the door, this from visual observation. You send your stupid brother-in-law to measure the width of the chair. He has an 8 ft tape-measure. You tell him to make 100 measurements of the chair's width.

He goes off, you do whatever you are doing and then he comes back with a perfectly gaussian set of width measurements. But when you look at the measurements, you see that he has some 6 ft widths, some 1 ft widths as measures of the chair's width. You look some more. Then you see that on a few outlier measurements he has said that the chair is 20 feet wide and others that say the chair is 2 inches wide.

You are stunned by this.

So you calculate the mean width and the 95% confidence interval. which since he measured it 100 times will be about 1/5th the size of one standard deviation.

Now of course, as you know, it would be relatively impossible to have a perfectly gaussian distribution centered around 3 with a tail at 20, unless, of course you allow my idiot brother-in-law to measure negative distances.

So I'll cut you some slack and allow you to have a non-poisson distribution in the present case but throw in an outlier at 20.

Here's what the distribution looks like WITH and WITHOUT the 20 outlier:

The 20' outlier really skewed the results but didn't shift the mean too much. WIth that outlier and a significant skew it calls into question the 95% confidence interval at the top. BUT, I can easily remove the outlier and I get the bottom distribution, which as you can see form the normal quantile plot (and a Shapiro Wilkes test p-value = 0.1964) that it is nice and normal.

In this case even though the standard deviation of this data set was 1' the 95% confidence interval on the mean width of this chair is less than 3 INCHES.

I can say with about 95% confidence that as long as I am using the normally distributed data the chair is probably 2.99feet + 3 inches

You would expect that he could do a more precise job. You expected that all the measurements should be within 1 inch of each other.

Well, in the end since he measured it so many times and you told me it was a gaussian distribution, I can get within 3".

But your silly brother-in-law has made measurements which diverge by feet--several times the width of the chair. Clearly he was doing something wrong in the measurement.

Clearly. But you said it was a gaussian distribution, not a poisson or skewed distribution.

The measurements which 1 ft or a couple of inches equally are unbelievable measurements. How in the heck can an approximately 3 ft wide chair measure 2 inches wide??? It can't.

But look how close the mean + 95% confidence interval came after I removed your 20' outlier!

You challenge the brother-in-law and say that the measurements are wrong. He goes away mad, saying that all is OK and that you are stupid to doubt the data. The brother-in-law comes back with a chart showing how his measurements are perfectly, and I mean PERFECTLY gaussian. Eveything is OK, he proclaims.

I have got to see you generate a perfectly gaussian distribution centered at 3' but with a positive tail at 20' unless you are allowing him to measure negative distance.

You object that gaussianicity doesn't matter.

Actually in this case it DOES matter. It aids in narrowing in the true value much more precisely.

It is irrelevant because the measurements are not correct. You shouldn't have such a spread in measured widths of the chair. Your brother-in-law tells you how stupid you are and he has lots of his friends tell you the same thing. They also accuse you of being mean and unreasonable when you doubt what the brother-in-law says.

Here we go with the "persecution complex".

What do you think? Is the brother-in-law right?

Remember, even an anosmic pig can sometimes find a truffle. If the dist'n was gaussian and centered on 3' I can know within a pretty narrow range what the true mean was.

Now, all this fun aside: IF my brother-in-law provided me with chair widths ranging from 2" to 20' I would immediately suspect the measurements and probably couldn't guarantee he knew what he was doing.

I would throw out the data.

But let's make it more real-world, shall we? Let's assume that I am running the chair factory and I have data for a random sampling of all chairs made and they have a tolerance spec that must be met.

If I measure 100 chairs and I have a standard deviation of one foot in the measurements I can still calculate the 95% confidence interval down to less than 3" if the data are normally distributed.

Two towns 16 miles apart should hardly ever have temperatures greater than 10 deg F.

Assuming you live in the same world as we do where robots are not doing all the work and machines sometimes break or give bad measurements, I think you need to determine exactly how bad the anecdotal data you consistently deal in actually is.

That is where statistics comes in. That's why I'm always so impressed that these stations you dig up have median differences of less than 2 degrees over the course of 60 to 100 years and have rather narrow distributions at that, albeit with heavy tails and outliers.

 

[grmorton]

You spend waaay too much time critiquing stats without directly using mathematics. (You could have done a nice calculation on this example, but you didnt'.)
I find that interesting.

I find it fascinating that you don't seem to understand that if you say a chair is 2" wide when it is really 3 feet wide that one can't use statistics to cover up the crappy measurements. I think you don't have a clue about science outside of geology.

So you calculate the mean width and the 95% confidence interval. which since he measured it 100 times will be about 1/5th the size of one standard deviation.

Now of course, as you know, it would be relatively impossible to have a perfectly gaussian distribution centered around 3 with a tail at 20, unless, of course you allow my idiot brother-in-law to measure negative distances.

One bad measurement at 20 can't rule out the Gaussian distribution. And once again, you utterly miss the point. I think intentionally miss the point. IF the measurements are no good, statistics don't matter.

So I'll cut you some slack and allow you to have a non-poisson distribution in the present case but throw in an outlier at 20.

And I cut you some slack yesterday when I told you that your distribution of the temperature differences weren't gaussian as you erroneously claimed.

The 20' outlier really skewed the results but didn't shift the mean too much. WIth that outlier and a significant skew it calls into question the 95% confidence interval at the top. BUT, I can easily remove the outlier and I get the bottom distribution, which as you can see form the normal quantile plot (and a Shapiro Wilkes test p-value = 0.1964) that it is nice and normal.

In this case even though the standard deviation of this data set was 1' the 95% confidence interval on the mean width of this chair is less than 3 INCHES.

And you just proved my point. Just because the idiot brings in a set of measurements that can't match reality doesn't mean that a statistical approach will be valid.

Here we go with the "persecution complex".

No, Just a case of where you have your friends say stupid things.

Hey Gracchus. ever want to answer the question about the station at Happy Camp?

Nah, he doesn't want to comment on that. He would prefer to say I am unreasonable.

Remember, even an anosmic pig can sometimes find a truffle. If the dist'n was gaussian and centered on 3' I can know within a pretty narrow range what the true mean was.

And a no-nothing statistician can use measurements to prove nothing as well.

Now, all this fun aside: IF my brother-in-law provided me with chair widths ranging from 2" to 20' I would immediately suspect the measurements and probably couldn't guarantee he knew what he was doing

.

Exactly MY point. YOu win the kewpie doll. The measurements of two closely spaced towns should not differ by 10 deg based upon physical grounds. Of course you won't care. YOu are a religious believer in global warming. Your religion is global warming.

If I measure 100 chairs and I have a standard deviation of one foot in the measurements I can still calculate the 95% confidence interval down to less than 3" if the data are normally distributed.

That is where statistics comes in. That's why I'm always so impressed that these stations you dig up have median differences of less than 2 degrees over the course of 60 to 100 years and have rather narrow distributions at that, albeit with heavy tails and outliers.

You are too easily impressed. Indeed, you are gullible.

I might add that you can't rule out gaussianicity based upon one 20' width. That has a very small probability.

 

[grmorton]

Ah, the data is incorrect because you can see that it is obviously incorrect, thus we should mistrust statistical analysis of the data and instead trust that you know what's best. Sorry, no deal. You aren't smarter than all the climatologists, only more arrogant. The climate isn't a chair, nor is the situation in any way comparable to the one in your analogy.

And Thistle, I contend that if you have more than 2 10+ deg F differences in temperature between two towns less than 16 miles apart, that we should mistrust the measurement. YOu on the other hand ignore it.

Let me ask you the questions. would you put your thermostat for your heater or air conditioner next to the stove?

Statistical treatment, like the one climatologists use, is imperative to achieving a proper understanding of what's going on. Your problem is that said statistical treatment shows that you and your politically driven ideas are wrong, so you'd rather not be confronted with it. Well, sucks to be you.

MY political ideas? I am not the one who wants to tax everyone to death by putting taxes on energy. Maybe it is you who have the political agenda.

Would you take your child's temperature and put the thermometer on the stove before reading it and telling the doctor what the temperature is? Are you THAT stupid?

About the two towns, please get a clue. You have been told why it's irrelevant to the bigger picture. Drop it or expect to be taken a lot less seriously (if possible).

I don't really give a flip what you think about me. You are the guy who illogically claimed that if one isn't likeable one can't tell the truth. Would you care to explain your logic on this? Why does likeableness indicate truthfulness?

 

[grmorton]

Isn't it interesting that you use a 180day running average to "reduce noise" in the data when it suits you but you won't allow the climatologists to grid-average the data?

I thought I would take another whack at this lunacy. You can see the 180 day periodicity in the raw data. I don't need to 'reduce the noise to see it.

Your problem is that no data will be able to make you doubt. That is a sign of fanatic.

Fanatics also don't answer scientific questions put to them, like why should we believe the unbelievable temperature differences between closely spaced towns? And you wont answer why we should see a seasonal periodicity. You said you were fascinated by that but you haven't actually addressed it.

A real scientist would.