Originally Posted by B Butler

I am more concerned with communicating ideas than precise use of words in a technical sense when the individuals attempting to communicate aren't professionals.Q

Because the two are mutually exclusive or because you are just a pretender?

Among professionals familiar with technical linguistics, clarity is improved by technical language. Attempts to impose proper technical language on those not intimate to the field does not always result in clarity. How many years did it take for you to acquire a professional vocabulary? Do you think you can force feed it in an internet forum?

Are you ever going to address the fact that the terms quantitative and qualitative are not technical language nor professional vocabulary? Or are you just upset at me for calling you a pretender and are trying to "win" by changing the field of "battle"? You are not as clever as you think you are.

The understanding of when technical language is important doesn't seem to have a place in your values system. This is not the first time a threat has been filibustered to death when you go into word obsession. I don't know that I can fix bad habits, but you should at least be aware of them.

Instead of concern trolling, how about you try to parse the actual quote in question. Just do it. Tell me what it actually says.

Is it? OK. I'm an engineer, rather than a scientist, but I haven't seen the distinction become important.

Address the actual line as it was used and you tell me if the distinction is important.

Even now, it seems important only in the sense of 'winning' an internet flame spat rather than illuminating the nature of climate science.

Says the guy who is repeatedly trying to chide me without addressing the actual line in question.

***

You did make a comment that we all don't jump on Eric when he says something outlandish. For example, from time to time he will predict election results based on astrology.

He can try to predict the results by flipping a coin for all I care. But, when he tries to claim that real science supports his delusional astrological crap, I care.

Then again, JPT will make values judgements based on Bible quotes,

He makes subjective values statements using his subjective interpretation of biblical prose? How is that in any way related to my concern here?

you obsess around word meanings,

No. I address people pretending that science supports what they wish was true. One way to differentiate between pretenders and simple slip ups is to investigate how well they really know the meaning of the terms they use.

and most everyone has a particular perspective which provides unusual insights at the cost of losing one's audience the further the writer's world view conflicts with the audience's.

Or the audience is only pretending to be an audience but is really engaging in the "fight" themselves. Hint, hint.

If I wrote a disclaimer that I don't acknowledge astrology as a means for learning objective truth each time Eric evokes astrology, if I confronted JPT each time he quotes the Bible, etc... I'd be very busy.

You choose to spend your time how you wish and I'll choose to spend my time how I wish. Just because you don't like how I spend mine does not mean I'm "wrong", "wasting my time", or even "out of line".

There are too many people who look at things from too many angles.

In your completely subjective, personal opinion you mean.

I've bumped heads with Eric once upon a time. I think we each picked up a few illuminating tidbits from each other, but we didn't shift each other's core world views significantly. Not surprised. Humans work that way.

Shifting "core world views"? That's this forum's purpose? Says who?

Are you trying to improve our understanding of the role of computer models in climate science?

No. I'm pointing out that the person who was attempting to do that either messed up or doesn't really know what he is talking about. What you choose to take away from the exchange is between you and your "core world view".

Are you trying to change our basic world views?

No. As I've already stated, I think your focus on such a thing (despite your inability to demonstrate its real existence) is boring.

Are you concerned that we use a few specific words in the proper technical sense?

Only if you are trying to "teach" others about the science while only pretending to understand it.

Is it emotional thing, that your status has been questioned and that you must 'win' the flame war or your status will be decreased?

How exactly does one "win" a flame war? What is my status? Does your core world view give you insight into the status of posters to the forum?

What is your goal here?

Oh, do I need to state it for the eleventybillionth time?

I look for attempts by people to claim science supports their opinions and attempt to determine if they really do understand the concepts or are just pretending. What others do with the information is up to them.

This is wonderful. I get to watch Butler, PRB, and Vandal pick the fly shit out of the pepper, to avoid starvation.

Why, yes, my meds ARE working...

ROFLMAO!

Originally Posted by Vandal-72

Here is the quote in question, in context: The problem is that we just don't fully know the consequences. We can't simply say "Increase motor-fuel consumption by 10% and temperatures will rise 5C" or give a precise timetable on the rising of the seas. Our attempted predictions, even if based on quantitative data, can give only qualitative results.

You tell me what he said.

I interpret it to say. Out attempted predictions, even if based on quantitative data, can only give imprecise results. He was sloppy in his use of qualitative and you were sloppy in your use of decry.

Originally Posted by Vandal-72

That's nice. Now address what I actually wrote.

BTW: Did you even read what your link actually said?

Quantitative research, scientific investigation of quantitative properties
Quantitative verse, a metrical system in poetry
Statistics, also known as quantitative analysis
Numerical data, also known as quantitative data




Not true. Nominal scales are still quantitative data.



Nope.

Edited to add: Efficiency ratios do not have any units, are they not quantitative data?



Not once did I make that claim, but you go ahead and just keep that vicious strawman at bay.



You might want to go inform all those statisticians out there that they don't have a clue what they are doing.

No you're right. I stand corrected. My initial reaction was quantitative means numerical. But then when I looked up that article and saw "scientific investigation of quantitative properties" I immediately added in my mind "as opposed to qualitative properties", thought of the movie 10 and the nominal scale used for rating women's "hotness" which is clearly a qualitative property and surmised that "scientific investigation of qualitative properties" was not quantitative in a scientific sense although it is still quantitative in a mathematical or statistical sense. I believe you were speaking about science and not math. After reading through I now see it is a philosophical idea and so transcends the difference between science and math."scientific investigation of quantitative properties""scientifscc"scientific investigation of quantitative properties"

Originally Posted by Vandal-72

Civil cases and policy purposes are now how we express scientific knowledge? When was that shift announced and why wasn't I CC'd?
.

The global warming debate is informed by science (and a lot of other things). But it is not just a scientific question. When asked a yes/no question science reserves the right to give one of three answers: yes, no, and I don't know--more study is needed. In other areas of life like the law or business the third option is not always available. A go/no go decision on what, if anything, to do about global warming will be made long before science has moved out of "I don't know, more study is needed". So, if a category 1 or higher tropical storm with wind speed that was 4 sigmas above the mean wind speed for category 1 or higher tropical storms historically occurred, to explain the event as a simply a random outlier out of the normal population such storms could be ruled out as beyond a shadow of doubt (a scientist would not make that call, a policymaker or judge would). Far more likely is that a special cause was operative (here the scientist would agree). Since there is no reason to believe that higher surface temperatures cannot lead to stronger storms this would be a very likely candidate for the special cause. Further investigation would be triggered by this event and its reasonable to believe that a theory of how warmer temperatures can produce super storms like this one would emerge. But until such a super storm happened there is no reason why such a theory would be developed. Its like when the Voyager probe discovered braided rings around Saturn. Previously nobody had thought braided rings were possible under the theory of gravity. Now that it had been shown that they are indeed possible, scientists looked into it and it turned out such rings were theoretically possible, nobody had worked it out before because nobody had ever seen such a thing before. I submit if a 4 sigma storm happened it would be unprecedented like the braided ring and after the fact the possibility of such things would become part of meteorologists understanding of weather and climate.

Originally Posted by Mikebert

I interpret it to say. Out attempted predictions, even if based on quantitative data, can only give imprecise results. He was sloppy in his use of qualitative

I would agree with your assessment if he hadn't doubled down on the mistake when he responded to my initial critique.

and you were sloppy in your use of decry.

My "decry" happened after reading his response. It turns out he wasn't being sloppy, he really didn't understand the mistake he was making.

Originally Posted by Mikebert

No you're right. I stand corrected.

Fair enough.

My initial reaction was quantitative means numerical. But then when I looked up that article and saw "scientific investigation of quantitative properties" I immediately added in my mind "as opposed to qualitative properties", thought of the movie 10 and the nominal scale used for rating women's "hotness" which is clearly a qualitative property and surmised that "scientific investigation of qualitative properties" was not quantitative in a scientific sense although it is still quantitative in a mathematical or statistical sense.

Yeah, I can see where your thinking went. Quantitative versus qualitative got intermixed with subjective versus objective. My students struggle with differentiating the ideas because in our ordinary usage of language we accept many of the terms as synonymous.

I believe you were speaking about science and not math. After reading through I now see it is a philosophical idea and so transcends the difference between science and math."scientific investigation of quantitative properties""scientifscc"scientific investigation of quantitative properties"

Not sure that I get what your are on about here.

Originally Posted by Mikebert

The global warming debate is informed by science (and a lot of other things). But it is not just a scientific question.

Pbrower was making a direct comment about the science itself (climate models).

When asked a yes/no question science reserves the right to give one of three answers: yes, no, and I don't know--more study is needed. In other areas of life like the law or business the third option is not always available. A go/no go decision on what, if anything, to do about global warming will be made long before science has moved out of "I don't know, more study is needed". So, if a category 1 or higher tropical storm with wind speed that was 4 sigmas above the mean wind speed for category 1 or higher tropical storms historically occurred, to explain the event as a simply a random outlier out of the normal population such storms could be ruled out as beyond a shadow of doubt (a scientist would not make that call, a policymaker or judge would). Far more likely is that a special cause was operative (here the scientist would agree). Since there is no reason to believe that higher surface temperatures cannot lead to stronger storms this would be a very likely candidate for the special cause. Further investigation would be triggered by this event and its reasonable to believe that a theory of how warmer temperatures can produce super storms like this one would emerge. But until such a super storm happened there is no reason why such a theory would be developed. Its like when the Voyager probe discovered braided rings around Saturn. Previously nobody had thought braided rings were possible under the theory of gravity. Now that it had been shown that they are indeed possible, scientists looked into it and it turned out such rings were theoretically possible, nobody had worked it out before because nobody had ever seen such a thing before. I submit if a 4 sigma storm happened it would be unprecedented like the braided ring and after the fact the possibility of such things would become part of meteorologists understanding of weather and climate.

You still didn't explain what metric you are measuring your four sigma deviation on. Once you explain your metric we can then look at the historical data to see what different storms were like within the distribution. It may very well turn out that a four sigma storm's physics would be impossible. Think about the already great deal of variation between a puff of wind on my face in the morning versus a hurricane. Both of those events are "typical" and therefor would fall well within one or two sigmas. What exactly would a four sigma wind speed be?

in short, your hypothetical is nothing but that, hypothetical.

Originally Posted by Vandal-72

So now it's to be ad hominem? Too bad your barb is poorly aimed. BS in Zoology, minor in chemistry. Secondary education certifications in biology, physics, earth science, chemistry and anthropology. Seven years experience as a research technician in natural resource management and eleven as an educator. But, you keep on believing whatever you want.

If it is ad hominem it is at someone other than you -- at people with thorough contempt for education beyond the Three R's. It's at people who can't see FoX "News" for what it is -- propaganda and for people who think that organizations with omnibus appeals (Freedom Works, Club for Growth, Americans for Prosperity) have innocuous agendas.

I make a distinction between the intelligentsia (people whose jobs require above-average education as a bona fide occupational qualification) and intellectuals, people who think much and do little aside from communicating their ideas. Good K-12 teachers do real work if they are to do any real teaching because they must see who is getting the lesson and who isn't. Intellectuals are a rarefied elite -- and at times they can be very wrong and get away with much. I named names. We need good education just to avoid being bamboozled.

What may have happened is that working-class white people have been able to strike back at the English teacher who corrects their grammar in the presence of other students or the biology teacher who pooh-poohed creationism in biology by aligning themselves with politicians who share a contempt for the middle class. It's OK to be ignorant and proud of it, as those politicians imply. That such does not apply to non-whites in America creates a huge political divide. I can say this: if we Americans are to prosper, then many of us are going to need to need to change our ways and respect learning again. I do not refer to you.

When I hear someone like Sarah Palin mangling the English language I get much the same effect as finger nails scratching a chalkboard. That said, she excites certain segments of the populace.

Revenge *ON* the Grammar Nazis. You may have something there.

Originally Posted by Vandal-72

Pbrower was making a direct comment about the science itself (climate models).



You still didn't explain what metric you are measuring your four sigma deviation on. Once you explain your metric we can then look at the historical data to see what different storms were like within the distribution. It may very well turn out that a four sigma storm's physics would be impossible. Think about the already great deal of variation between a puff of wind on my face in the morning versus a hurricane. Both of those events are "typical" and therefor would fall well within one or two sigmas. What exactly would a four sigma wind speed be?

in short, your hypothetical is nothing but that, hypothetical.

"Sigma" is a reference to the standard deviation (also abbreviated "s.d" but usually shown with the lower-case Greek letter sigma [σ]).



For a normal distribution 4 standard deviations from the median is highly unlikely and suggests either an outlier as a mistake (as in, "somebody misplaced a decimal point" or "dirty glassware"), fabrication, or something faulty about the assumptions behind the theory.

Range Population in range Expected frequency outside range Approx. frequency for daily event
μ ± 1σ 0.682689492137086 1 in 3 Twice a week
μ ± 1.5σ 0.866385597462284 1 in 7 Weekly
μ ± 2σ 0.954499736103642 1 in 22 Every three weeks
μ ± 2.5σ 0.987580669348448 1 in 81 Quarterly
μ ± 3σ 0.997300203936740 1 in 370 Yearly
μ ± 3.5σ 0.999534741841929 1 in 2149 Every six years
μ ± 4σ 0.999936657516334 1 in 15,787 Every 43 years (twice in a lifetime)
μ ± 4.5σ 0.999993204653751 1 in 147,160 Every 403 years
μ ± 5σ 0.999999426696856 1 in 1,744,278 Every 4,776 years (once in recorded history)
μ ± 5.5σ 0.999999962020875 1 in 26,330,254 Every 72,090 years
μ ± 6σ 0.999999998026825 1 in 506,797,346 Every 1.38 million years (history of humankind)
μ ± 6.5σ 0.999999999919680 1 in 12,450,197,393 Every 34 million years
μ ± 7σ 0.999999999997440 1 in 390,682,215,445 Every 1.07 billion years

https://en.wikipedia.org/wiki/68%E2%...80%9399.7_rule

Originally Posted by Bad Dog

Revenge *ON* the Grammar Nazis. You may have something there.

I know a nice mailing list for said parties. <----- www.intp.org

The ultimate locale for nitpicking, grammar parsing, and of course geeky science threads.

Originally Posted by pbrower2a

If it is ad hominem it is at someone other than you -- at people with thorough contempt for education beyond the Three R's. It's at people who can't see FoX "News" for what it is -- propaganda and for people who think that organizations with omnibus appeals (Freedom Works, Club for Growth, Americans for Prosperity) have innocuous agendas.

No it wasn't directed at someone else. You included a very specific jab at a particular degree implying that someone with such a degree was not really qualified to discuss the topic. The degree you chose as an example was the degree you assumed I possessed myself.

I make a distinction between the intelligentsia (people whose jobs require above-average education as a bona fide occupational qualification) and intellectuals, people who think much and do little aside from communicating their ideas. Good K-12 teachers do real work if they are to do any real teaching because they must see who is getting the lesson and who isn't. Intellectuals are a rarefied elite -- and at times they can be very wrong and get away with much. I named names. We need good education just to avoid being bamboozled.

But the stereotype you used was not generated by people who are aware of let alone agree to your distinction.

What may have happened is that working-class white people have been able to strike back at the English teacher who corrects their grammar in the presence of other students or the biology teacher who pooh-poohed creationism in biology by aligning themselves with politicians who share a contempt for the middle class.

I don't think the attitude is restricted to just white working class. I also don't agree that the political champions of the anti-intellectual are striking out because their targets are middle class. That may have been true of earlier generations of conservative politicians but say what you will about Xers, our politicians almost certainly believe in what they say.

If a younger Boomer or Xer politician expresses anti-intellectual views, it's a safe bet that they really do hold educated people in contempt. The hypocrisy of their own educational achievements gets buried in a wave of special pleading and no true Scotsman fallacies. I don't think that they deep down know that they are pandering.

It's OK to be ignorant and proud of it, as those politicians imply. That such does not apply to non-whites in America creates a huge political divide. I can say this: if we Americans are to prosper, then many of us are going to need to need to change our ways and respect learning again. I do not refer to you.

Maybe you did and maybe you didn't. Either way, it's no skin off my nose.

When I hear someone like Sarah Palin mangling the English language I get much the same effect as finger nails scratching a chalkboard. That said, she excites certain segments of the populace.

And as much as a political persona is image manipulation, I don't think her reading list snafu was anything but an honest assessment of her intellectual proclivities.

Originally Posted by Bad Dog

This is wonderful. I get to watch Butler, PRB, and Vandal pick the fly shit out of the pepper, to avoid starvation.

You forgot Mikebert. Now, if you wander off to some other threads, you can of course toss in Eric as well.

It's all good man. I feel like I'm in the peanut gallery at a B rated movie.

Originally Posted by Ragnarök_62

You forgot Mikebert. Now, if you wander off to some other threads, you can of course toss in Eric as well.

It's all good man. I feel like I'm in the peanut gallery at a B rated movie.

or, on MST3K.

Originally Posted by Vandal-72

Pbrower was making a direct comment about the science itself (climate models).

Yes he did, And he used a word incorrectly as you did with decry. So what?

You still didn't explain what metric you are measuring your four sigma deviation on. Once you explain your metric we can then look at the historical data to see what different storms were like within the distribution.

I suggested wind speed for category 1 & higher tropical storms. What is so complicated about this?

Originally Posted by Mikebert

Yes he did, And he used a word incorrectly as you did with decry. So what?

Was his incorrect usage a simple mistake or a sign of not understanding what he was saying?

Decry means to publicly denounce. Not sure I see my misuse.

I suggested wind speed for category 1 & higher tropical storms. What is so complicated about this?

OK, what would the wind speed of a four sigma storm be? The problem is that wind speed is not necessarily capable of any theoretical speed but the statistical analysis may not take that into account. Another problem is that when you calculate your distribution curve based on a pre-warming history, the statistics say that eventually given enough time you will get an occasional four sigma storm.

To demonstrate actual warming you would need a large sample size that you could statistically compare to the pre-warming data. It is still impossible to attribute a single storm to warming. That is simply how statistics work.

In short, a single four sigma storm can never, ever be attributed to warming.

Originally Posted by Ragnarök_62

I know a nice mailing list for said parties. <----- www.intp.org

The ultimate locale for nitpicking, grammar parsing, and of course geeky science threads.

Try a Mensa gathering, as well. Blech!

Originally Posted by Vandal-72

Decry means to publicly denounce. Not sure I see my misuse.

See my post.

OK, what would the wind speed of a four sigma storm be? The problem is that wind speed is not necessarily capable of any theoretical speed but the statistical analysis may not take that into account.

This is irrelevant. If it is not possible then it won’t happen. The question is if it DID happen, what could be concluded from this single event?

To demonstrate actual warming you would need a large sample size that you could statistically compare to the pre-warming data.

And if you find a low probability (e.g. p <0.01) you would conclude that you have demonstrated statistical significance.

As you point out, p <0.01 means that, eventually given enough time this will happen. The fact that the analysis can never give certainty, only probability, does not mean that such an investigation can never be used to establish an effect of warming.

It is still impossible to attribute a single storm to warming. That is simply how statistics work.

Statistics provide ways to analyze outcomes in order to assign probabilities to them. Since the world is normally “smooth” one does not usually see very extreme events in the absence of a large number of less extreme events. That is, if warming gives stronger storms the most likely outcome would be more strong storms or slightly stronger storms, either of which would show up as higher average strength as so be amenable to statistical analysis of the two populations using an appropriate distribution model. You probably wouldn’t use a Gaussian-type distribution (e.g. a t-test) on untransformed data in this case.

But the question is can one very unlikely event (1/30000) be statistically significant all by itself?

The idea here is there exists a category of things called strong storms (say cat 4 & higher). Suppose the historical record holds 300 such storms. This dataset is an external reference set. It is the gold standard to use for assigning probabilities.

Now you watch what happens when the treatment (post-2013 global warming) is applied to see what happens. Suppose 1 of next 50 storms is higher than any previous storm. This storm is an outlier and comparison to the reference set would assign a 1/300 probability to this storm happening. Of course there were 50 storms over the period during which the evaluation was done. The relevant probability is the probability that any one of them would be a record holder. This is 50 x 1/300 or 1/6. That is, it is not very unlikely that when adding 50 more storms to the 300-strong historical record that one of them would set a new record.

Now suppose you had 30000 storms over the past 4000 years in the historical record, and one of the next 50 established a new record. This would be a statistically-significant event with a 1/600 probability. Now it could be a one-off (just as a conventional t-test done on a larger data set might be). And if after several millennia no additional outlier happens it would become more and more likely that it was just luck.

But if would not be incorrect to conclude after the initial record holder occurred that the global warming had done something to storm strength. A larger number of less extreme events that statistical analysis would show has a low probability would make one feel better about making such conclusions because it supports our intuition that the world is smooth. But the probabilities are the same.

Originally Posted by Bad Dog

Revenge *ON* the Grammar Nazis. You may have something there.

Because the typical grade-school classroom allows little privacy and teachers dislike bad grammar as in

"I seen them horses"

the conflict is predictable. Thirty years later it strikes back.

Originally Posted by Mikebert

See my post.


This is irrelevant. If it is not possible then it won’t happen. The question is if it DID happen, what could be concluded from this single event?

A single event? Nothing. That's the point. Statistics doesn't work that way. To call an event a four sigma means that the event might be rare but is completely likely given a large enough sample.

And if you find a low probability (e.g. p <0.01) you would conclude that you have demonstrated statistical significance.

You can't run a statistical test for significance on a single event.

As you point out, p <0.01 means that, eventually given enough time this will happen. The fact that the analysis can never give certainty, only probability, does not mean that such an investigation can never be used to establish an effect of warming.

Of course not. I never said otherwise. What you can do is compare the distributions of storm energies before and after certain dates. That requires sample sizes greater than one however. A single storm is not adequate.

Statistics provide ways to analyze outcomes in order to assign probabilities to them. Since the world is normally “smooth” one does not usually see very extreme events in the absence of a large number of less extreme events. That is, if warming gives stronger storms the most likely outcome would be more strong storms or slightly stronger storms, either of which would show up as higher average strength as so be amenable to statistical analysis of the two populations using an appropriate distribution model. You probably wouldn’t use a Gaussian-type distribution (e.g. a t-test) on untransformed data in this case.

But the question is can one very unlikely event (1/30000) be statistically significant all by itself?

No. All you can give is the probability of such an event but you can not claim statistical significance with a sample size of one.

The idea here is there exists a category of things called strong storms (say cat 4 & higher). Suppose the historical record holds 300 such storms. This dataset is an external reference set. It is the gold standard to use for assigning probabilities.

Now you watch what happens when the treatment (post-2013 global warming) is applied to see what happens. Suppose 1 of next 50 storms is higher than any previous storm. This storm is an outlier and comparison to the reference set would assign a 1/300 probability to this storm happening. Of course there were 50 storms over the period during which the evaluation was done. The relevant probability is the probability that any one of them would be a record holder. This is 50 x 1/300 or 1/6. That is, it is not very unlikely that when adding 50 more storms to the 300-strong historical record that one of them would set a new record.

Now suppose you had 30000 storms over the past 4000 years in the historical record, and one of the next 50 established a new record. This would be a statistically-significant event with a 1/600 probability. Now it could be a one-off (just as a conventional t-test done on a larger data set might be). And if after several millennia no additional outlier happens it would become more and more likely that it was just luck.

Exactly. You can never ever draw a conclusion of statistical significance for a single event.

But if would not be incorrect to conclude after the initial record holder occurred that the global warming had done something to storm strength.

Only if you have a post-warming dataset. A dataset requires more than a single storm. You could then conclude that a large storm has an increased probability of happening but if a large storm occurs that doesn't confirm that warming is responsible.

A larger number of less extreme events that statistical analysis would show has a low probability would make one feel better about making such conclusions because it supports our intuition that the world is smooth. But the probabilities are the same.

But, to declare the single storm as significant by itself is not a valid claim. That is simply not how stochastic events work.

Let's say warming will increase the likelihood of large storms. The average per year rises from 7 to 9 for example. Can you claim that the 8th storm of the season is due to warming? Why isn't it the 5th storm or the 1st? While the increased storm activity can be attributed to warming, no individual storm alone can be.

Originally Posted by Vandal-72

Let's say warming will increase the likelihood of large storms. The average per year rises from 7 to 9 for example. Can you claim that the 8th storm of the season is due to warming? Why isn't it the 5th storm or the 1st? While the increased storm activity can be attributed to warming, no individual storm alone can be.

I was not talking about changes in frequency, but level. Here's an example of what I had in mind. Suppose you have a historical record of measured levels that average about 50, and then the most recent level measured (the 51st) was 94.0, see below.

94.0 32.6 42.7 57.7 46.3 53.9 45.9 56.3 46.0 59.3 64.6 62.2 57.5 57.4 45.2 33.9 38.3 54.0 51.1 59.9 53.8 42.7 59.3 61.6 75.9 58.8 47.7 31.2 40.9 50.4 33.0 41.5 41.0 55.0 38.9 58.6 51.0 47.7 47.0 46.9 73.5 48.5 54.9 38.8 38.4 53.1 45.1 67.5 59.3 65.3 48.6

Is this value signficantly different from the others? Is it a true outlier that does not belong with the rest of the data because of some special cause? It seems to me Grubb's test would be suitable. I pasted these values into the calculator and it gave a Z value of 3.56 which is significant at the 0.01 level.

Originally Posted by Mikebert

I was not talking about changes in frequency, but level. Here's an example of what I had in mind. Suppose you have a historical record of measured levels that average about 50, and then the most recent level measured (the 51st) was 94.0, see below.

94.0 32.6 42.7 57.7 46.3 53.9 45.9 56.3 46.0 59.3 64.6 62.2 57.5 57.4 45.2 33.9 38.3 54.0 51.1 59.9 53.8 42.7 59.3 61.6 75.9 58.8 47.7 31.2 40.9 50.4 33.0 41.5 41.0 55.0 38.9 58.6 51.0 47.7 47.0 46.9 73.5 48.5 54.9 38.8 38.4 53.1 45.1 67.5 59.3 65.3 48.6

Is this value signficantly different from the others? Is it a true outlier that does not belong with the rest of the data because of some special cause? It seems to me Grubb's test would be suitable. I pasted these values into the calculator and it gave a Z value of 3.56 which is significant at the 0.01 level.

Grubb's test is used to rid data sets of outliers to make other statistical tests more reliable. The larger your dataset, the less likely that Grubbs will show it as an outlier even though it's absolute value remains the same. You are still stuck in the same boat. You can not attribute the outlier to warming if its outlier status is dependent on sample size.

if your example dataset is truly normally distributed then you would expect a 94.0 to pop up eventually. But, you can not claim that just because it popped up as the 51st rather than as the 551st entry there is anything significantly special about it.

You are using Grubb's test for a purpose for which it is not appropriate.

Let my try an analogous problem. Take a truly random coin. If you flip it 10,000 times, then there is a good chance that somewhere in the sequence of flips there will be a sequence of ten heads in a row. Do you conclude that the coin is not truly random because of the ten head sequence? No. You take into account all the flips and see that the average is incredibly close to 50% heads.

Now, what if the very first ten flips were all heads? This is in essence what you are claiming as a four sigma storm! But, when we continue to monitor the flips the average rapidly returns to the expected 50%. We can only conclude something is "funky" with our coin if we have a complete set of flips that significantly differs from the predicted 50% and ten is nowhere near enough. We can only detect the effects of warming on storm strengths by comparing large sets of data, not individual storms.

Because you are declaring the data collection complete as soon as you have a single four sigma outlier storm, you are cherry picking and/or engaging in confirmation bias. You are purposely ignoring some data, subsequent storms. It is the same invalid statistical procedure that most parapsychology advocates engage in. They watch the data as it is coming in, calculating their t-tests as they go and as soon as they have a significant score (just due to random chance) they stop their collecting and declare ESP is real. It's statistical fraud.

It's like testing our coin's randomness by flipping it, keeping a running total and stopping the instant our heads percentage is 60 or 65%. There is a reason that researchers are required to calculate minimum sample sizes BEFORE they begin their data collection.

Originally Posted by Vandal-72

Grubb's test is used to rid data sets of outliers to make other statistical tests more reliable. The larger your dataset, the less likely that Grubbs will show it as an outlier even though it's absolute value remains the same. You are still stuck in the same boat. You can not attribute the outlier to warming if its outlier status is dependent on sample size.

if your example dataset is truly normally distributed then you would expect a 94.0 to pop up eventually. But, you can not claim that just because it popped up as the 51st rather than as the 551st entry there is anything significantly special about it.

You are using Grubb's test for a purpose for which it is not appropriate.

Let my try an analogous problem. Take a truly random coin. If you flip it 10,000 times, then there is a good chance that somewhere in the sequence of flips there will be a sequence of ten heads in a row. Do you conclude that the coin is not truly random because of the ten head sequence? No. You take into account all the flips and see that the average is incredibly close to 50% heads.

Now, what if the very first ten flips were all heads? This is in essence what you are claiming as a four sigma storm! But, when we continue to monitor the flips the average rapidly returns to the expected 50%. We can only conclude something is "funky" with our coin if we have a complete set of flips that significantly differs from the predicted 50% and ten is nowhere near enough. We can only detect the effects of warming on storm strengths by comparing large sets of data, not individual storms.

Because you are declaring the data collection complete as soon as you have a single four sigma outlier storm, you are cherry picking and/or engaging in confirmation bias. You are purposely ignoring some data, subsequent storms. It is the same invalid statistical procedure that most parapsychology advocates engage in. They watch the data as it is coming in, calculating their t-tests as they go and as soon as they have a significant score (just due to random chance) they stop their collecting and declare ESP is real. It's statistical fraud.

It's like testing our coin's randomness by flipping it, keeping a running total and stopping the instant our heads percentage is 60 or 65%. There is a reason that researchers are required to calculate minimum sample sizes BEFORE they begin their data collection.

So boiling all this down to the essence, we can't know anything for sure until we know everything for sure. This reminds me of the argument that Keynes made about classical economics, that long after the storm has passed and seas are once again calm, we can learn the solution to the problem. This is a reactionary mindset. Science is a discipline, but one of limited value if it can't be used to prevent disasters. That means making decision based on incomplete data and risking potential error.

It can' t be used proactively and be any other way.